{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999954144006163,
  "eval_steps": 500,
  "global_step": 13629,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "lm_loss": 1.6875,
      "step": 0,
      "vm_loss": 0.154
    },
    {
      "epoch": 0,
      "lm_loss": 1.6455,
      "step": 0,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0,
      "lm_loss": 1.4582,
      "step": 0,
      "vm_loss": 0.1799
    },
    {
      "epoch": 0,
      "lm_loss": 1.6385,
      "step": 0,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0,
      "lm_loss": 1.464,
      "step": 0,
      "vm_loss": 0.1526
    },
    {
      "epoch": 0,
      "lm_loss": 1.2552,
      "step": 0,
      "vm_loss": 0.188
    },
    {
      "epoch": 0,
      "lm_loss": 1.5966,
      "step": 0,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0,
      "lm_loss": 0.8822,
      "step": 0,
      "vm_loss": 0.1513
    },
    {
      "epoch": 7.336959013912708e-05,
      "grad_norm": 13.949854872865103,
      "learning_rate": 2.4449877750611247e-08,
      "loss": 1.5556,
      "step": 1
    },
    {
      "epoch": 0.00014673918027825416,
      "grad_norm": 17.14922543255026,
      "learning_rate": 4.8899755501222494e-08,
      "loss": 1.5745,
      "step": 2
    },
    {
      "epoch": 0.00022010877041738126,
      "grad_norm": 14.390987790533607,
      "learning_rate": 7.334963325183375e-08,
      "loss": 1.5279,
      "step": 3
    },
    {
      "epoch": 0.0002934783605565083,
      "grad_norm": 13.162819208526459,
      "learning_rate": 9.779951100244499e-08,
      "loss": 1.65,
      "step": 4
    },
    {
      "epoch": 0.00036684795069563545,
      "grad_norm": 4.0230523397743205,
      "learning_rate": 1.2224938875305626e-07,
      "loss": 0.8156,
      "step": 5
    },
    {
      "epoch": 0.0004402175408347625,
      "grad_norm": 15.328325187166582,
      "learning_rate": 1.466992665036675e-07,
      "loss": 1.6078,
      "step": 6
    },
    {
      "epoch": 0.0005135871309738896,
      "grad_norm": 11.58796003605192,
      "learning_rate": 1.7114914425427872e-07,
      "loss": 1.5566,
      "step": 7
    },
    {
      "epoch": 0.0005869567211130167,
      "grad_norm": 4.189568033449158,
      "learning_rate": 1.9559902200488998e-07,
      "loss": 0.806,
      "step": 8
    },
    {
      "epoch": 0.0005869567211130167,
      "lm_loss": 0.7712,
      "step": 8,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0005869567211130167,
      "lm_loss": 0.8643,
      "step": 8,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0005869567211130167,
      "lm_loss": 0.8557,
      "step": 8,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0005869567211130167,
      "lm_loss": 0.9784,
      "step": 8,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0005869567211130167,
      "lm_loss": 0.7999,
      "step": 8,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0005869567211130167,
      "lm_loss": 0.7979,
      "step": 8,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0005869567211130167,
      "lm_loss": 0.7152,
      "step": 8,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0005869567211130167,
      "lm_loss": 0.8631,
      "step": 8,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0006603263112521438,
      "grad_norm": 4.552868938472152,
      "learning_rate": 2.2004889975550124e-07,
      "loss": 0.84,
      "step": 9
    },
    {
      "epoch": 0.0007336959013912709,
      "grad_norm": 10.234276768522259,
      "learning_rate": 2.444987775061125e-07,
      "loss": 1.501,
      "step": 10
    },
    {
      "epoch": 0.0008070654915303979,
      "grad_norm": 13.611724362843264,
      "learning_rate": 2.6894865525672375e-07,
      "loss": 1.5488,
      "step": 11
    },
    {
      "epoch": 0.000880435081669525,
      "grad_norm": 11.1204921647466,
      "learning_rate": 2.93398533007335e-07,
      "loss": 1.5619,
      "step": 12
    },
    {
      "epoch": 0.0009538046718086522,
      "grad_norm": 12.166821877654275,
      "learning_rate": 3.1784841075794626e-07,
      "loss": 1.5521,
      "step": 13
    },
    {
      "epoch": 0.0010271742619477793,
      "grad_norm": 4.0140138164128265,
      "learning_rate": 3.4229828850855744e-07,
      "loss": 0.7891,
      "step": 14
    },
    {
      "epoch": 0.0011005438520869063,
      "grad_norm": 12.236030761629417,
      "learning_rate": 3.667481662591688e-07,
      "loss": 1.5515,
      "step": 15
    },
    {
      "epoch": 0.0011739134422260333,
      "grad_norm": 10.636596588174305,
      "learning_rate": 3.9119804400977996e-07,
      "loss": 1.539,
      "step": 16
    },
    {
      "epoch": 0.0011739134422260333,
      "lm_loss": 1.3472,
      "step": 16,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.0011739134422260333,
      "lm_loss": 1.6928,
      "step": 16,
      "vm_loss": 0.1241
    },
    {
      "epoch": 0.0011739134422260333,
      "lm_loss": 2.3721,
      "step": 16,
      "vm_loss": 0.2433
    },
    {
      "epoch": 0.0011739134422260333,
      "lm_loss": 1.3716,
      "step": 16,
      "vm_loss": 0.0684
    },
    {
      "epoch": 0.0011739134422260333,
      "lm_loss": 1.7764,
      "step": 16,
      "vm_loss": 0.0964
    },
    {
      "epoch": 0.0011739134422260333,
      "lm_loss": 1.3334,
      "step": 16,
      "vm_loss": 0.1185
    },
    {
      "epoch": 0.0011739134422260333,
      "lm_loss": 1.5259,
      "step": 16,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.0011739134422260333,
      "lm_loss": 1.6235,
      "step": 16,
      "vm_loss": 0.1882
    },
    {
      "epoch": 0.0012472830323651605,
      "grad_norm": 9.287192347868773,
      "learning_rate": 4.156479217603912e-07,
      "loss": 1.4982,
      "step": 17
    },
    {
      "epoch": 0.0013206526225042876,
      "grad_norm": 9.801182956407159,
      "learning_rate": 4.4009779951100247e-07,
      "loss": 1.5781,
      "step": 18
    },
    {
      "epoch": 0.0013940222126434146,
      "grad_norm": 4.129356896089923,
      "learning_rate": 4.645476772616137e-07,
      "loss": 0.7965,
      "step": 19
    },
    {
      "epoch": 0.0014673918027825418,
      "grad_norm": 9.235234492241526,
      "learning_rate": 4.88997555012225e-07,
      "loss": 1.5022,
      "step": 20
    },
    {
      "epoch": 0.0015407613929216688,
      "grad_norm": 8.387494212552857,
      "learning_rate": 5.134474327628362e-07,
      "loss": 1.4545,
      "step": 21
    },
    {
      "epoch": 0.0016141309830607958,
      "grad_norm": 5.508057800316507,
      "learning_rate": 5.378973105134475e-07,
      "loss": 1.4283,
      "step": 22
    },
    {
      "epoch": 0.001687500573199923,
      "grad_norm": 5.977795069098969,
      "learning_rate": 5.623471882640588e-07,
      "loss": 1.5002,
      "step": 23
    },
    {
      "epoch": 0.00176087016333905,
      "grad_norm": 4.8016063256436885,
      "learning_rate": 5.8679706601467e-07,
      "loss": 1.4782,
      "step": 24
    },
    {
      "epoch": 0.00176087016333905,
      "lm_loss": 0.8298,
      "step": 24,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.00176087016333905,
      "lm_loss": 0.342,
      "step": 24,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.00176087016333905,
      "lm_loss": 0.7871,
      "step": 24,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.00176087016333905,
      "lm_loss": 0.8038,
      "step": 24,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.00176087016333905,
      "lm_loss": 0.8576,
      "step": 24,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.00176087016333905,
      "lm_loss": 0.8232,
      "step": 24,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.00176087016333905,
      "lm_loss": 0.7447,
      "step": 24,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.00176087016333905,
      "lm_loss": 0.7541,
      "step": 24,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.001834239753478177,
      "grad_norm": 3.7394154118928977,
      "learning_rate": 6.112469437652812e-07,
      "loss": 0.7937,
      "step": 25
    },
    {
      "epoch": 0.0019076093436173043,
      "grad_norm": 4.525377937700574,
      "learning_rate": 6.356968215158925e-07,
      "loss": 1.4509,
      "step": 26
    },
    {
      "epoch": 0.001980978933756431,
      "grad_norm": 4.825048440564797,
      "learning_rate": 6.601466992665037e-07,
      "loss": 1.3508,
      "step": 27
    },
    {
      "epoch": 0.0020543485238955586,
      "grad_norm": 4.036842637878663,
      "learning_rate": 6.845965770171149e-07,
      "loss": 1.4851,
      "step": 28
    },
    {
      "epoch": 0.0021277181140346856,
      "grad_norm": 3.731422544482464,
      "learning_rate": 7.090464547677262e-07,
      "loss": 1.3807,
      "step": 29
    },
    {
      "epoch": 0.0022010877041738126,
      "grad_norm": 3.631944259295039,
      "learning_rate": 7.334963325183376e-07,
      "loss": 1.3594,
      "step": 30
    },
    {
      "epoch": 0.0022744572943129396,
      "grad_norm": 3.1332239511377775,
      "learning_rate": 7.579462102689487e-07,
      "loss": 0.7788,
      "step": 31
    },
    {
      "epoch": 0.0023478268844520666,
      "grad_norm": 3.7859258734597865,
      "learning_rate": 7.823960880195599e-07,
      "loss": 1.4042,
      "step": 32
    },
    {
      "epoch": 0.0023478268844520666,
      "lm_loss": 1.5194,
      "step": 32,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.0023478268844520666,
      "lm_loss": 0.8634,
      "step": 32,
      "vm_loss": 0.2292
    },
    {
      "epoch": 0.0023478268844520666,
      "lm_loss": 0.7213,
      "step": 32,
      "vm_loss": 0.1236
    },
    {
      "epoch": 0.0023478268844520666,
      "lm_loss": 1.2514,
      "step": 32,
      "vm_loss": 0.1868
    },
    {
      "epoch": 0.0023478268844520666,
      "lm_loss": 1.1351,
      "step": 32,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.0023478268844520666,
      "lm_loss": 1.4916,
      "step": 32,
      "vm_loss": 0.0808
    },
    {
      "epoch": 0.0023478268844520666,
      "lm_loss": 1.0053,
      "step": 32,
      "vm_loss": 0.2629
    },
    {
      "epoch": 0.0023478268844520666,
      "lm_loss": 1.3288,
      "step": 32,
      "vm_loss": 0.1398
    },
    {
      "epoch": 0.0024211964745911936,
      "grad_norm": 3.4784566683192044,
      "learning_rate": 8.068459657701712e-07,
      "loss": 1.4243,
      "step": 33
    },
    {
      "epoch": 0.002494566064730321,
      "grad_norm": 2.900037396406231,
      "learning_rate": 8.312958435207824e-07,
      "loss": 0.7908,
      "step": 34
    },
    {
      "epoch": 0.002567935654869448,
      "grad_norm": 3.2114517546339747,
      "learning_rate": 8.557457212713938e-07,
      "loss": 1.4085,
      "step": 35
    },
    {
      "epoch": 0.002641305245008575,
      "grad_norm": 3.3839625748942446,
      "learning_rate": 8.801955990220049e-07,
      "loss": 1.3516,
      "step": 36
    },
    {
      "epoch": 0.002714674835147702,
      "grad_norm": 2.742489987035647,
      "learning_rate": 9.046454767726162e-07,
      "loss": 0.7762,
      "step": 37
    },
    {
      "epoch": 0.002788044425286829,
      "grad_norm": 3.299751527225773,
      "learning_rate": 9.290953545232274e-07,
      "loss": 1.2746,
      "step": 38
    },
    {
      "epoch": 0.002861414015425956,
      "grad_norm": 2.9906059974428185,
      "learning_rate": 9.535452322738387e-07,
      "loss": 1.3459,
      "step": 39
    },
    {
      "epoch": 0.0029347836055650836,
      "grad_norm": 2.5865831383159557,
      "learning_rate": 9.7799511002445e-07,
      "loss": 0.7333,
      "step": 40
    },
    {
      "epoch": 0.0029347836055650836,
      "lm_loss": 0.5919,
      "step": 40,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0029347836055650836,
      "lm_loss": 0.8558,
      "step": 40,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0029347836055650836,
      "lm_loss": 0.7728,
      "step": 40,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0029347836055650836,
      "lm_loss": 0.9148,
      "step": 40,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0029347836055650836,
      "lm_loss": 0.7538,
      "step": 40,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0029347836055650836,
      "lm_loss": 0.8268,
      "step": 40,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0029347836055650836,
      "lm_loss": 0.6927,
      "step": 40,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0029347836055650836,
      "lm_loss": 0.7916,
      "step": 40,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0030081531957042106,
      "grad_norm": 2.410836162561251,
      "learning_rate": 1.0024449877750611e-06,
      "loss": 0.7512,
      "step": 41
    },
    {
      "epoch": 0.0030815227858433376,
      "grad_norm": 2.9966638534550794,
      "learning_rate": 1.0268948655256724e-06,
      "loss": 1.3902,
      "step": 42
    },
    {
      "epoch": 0.0031548923759824646,
      "grad_norm": 3.114048939330943,
      "learning_rate": 1.0513447432762837e-06,
      "loss": 1.2913,
      "step": 43
    },
    {
      "epoch": 0.0032282619661215916,
      "grad_norm": 2.890024445414154,
      "learning_rate": 1.075794621026895e-06,
      "loss": 1.3479,
      "step": 44
    },
    {
      "epoch": 0.0033016315562607187,
      "grad_norm": 3.1017800981850483,
      "learning_rate": 1.100244498777506e-06,
      "loss": 1.3555,
      "step": 45
    },
    {
      "epoch": 0.003375001146399846,
      "grad_norm": 3.1945547322172403,
      "learning_rate": 1.1246943765281176e-06,
      "loss": 1.244,
      "step": 46
    },
    {
      "epoch": 0.003448370736538973,
      "grad_norm": 3.1483334596867163,
      "learning_rate": 1.1491442542787286e-06,
      "loss": 1.3511,
      "step": 47
    },
    {
      "epoch": 0.0035217403266781,
      "grad_norm": 3.256064512220961,
      "learning_rate": 1.17359413202934e-06,
      "loss": 1.3614,
      "step": 48
    },
    {
      "epoch": 0.0035217403266781,
      "lm_loss": 0.7562,
      "step": 48,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0035217403266781,
      "lm_loss": 0.6246,
      "step": 48,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0035217403266781,
      "lm_loss": 0.7986,
      "step": 48,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0035217403266781,
      "lm_loss": 0.7713,
      "step": 48,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0035217403266781,
      "lm_loss": 0.8223,
      "step": 48,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0035217403266781,
      "lm_loss": 0.9056,
      "step": 48,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0035217403266781,
      "lm_loss": 0.9257,
      "step": 48,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0035217403266781,
      "lm_loss": 0.6978,
      "step": 48,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.003595109916817227,
      "grad_norm": 1.6828825606370188,
      "learning_rate": 1.1980440097799512e-06,
      "loss": 0.7565,
      "step": 49
    },
    {
      "epoch": 0.003668479506956354,
      "grad_norm": 3.1125133414258808,
      "learning_rate": 1.2224938875305625e-06,
      "loss": 1.3328,
      "step": 50
    },
    {
      "epoch": 0.003741849097095481,
      "grad_norm": 2.846159744197563,
      "learning_rate": 1.2469437652811738e-06,
      "loss": 1.2479,
      "step": 51
    },
    {
      "epoch": 0.0038152186872346086,
      "grad_norm": 2.787602410304938,
      "learning_rate": 1.271393643031785e-06,
      "loss": 1.2238,
      "step": 52
    },
    {
      "epoch": 0.0038885882773737356,
      "grad_norm": 2.7932844773964147,
      "learning_rate": 1.2958435207823961e-06,
      "loss": 1.2956,
      "step": 53
    },
    {
      "epoch": 0.003961957867512862,
      "grad_norm": 2.716133686508251,
      "learning_rate": 1.3202933985330074e-06,
      "loss": 1.2743,
      "step": 54
    },
    {
      "epoch": 0.00403532745765199,
      "grad_norm": 3.0892505902351797,
      "learning_rate": 1.3447432762836187e-06,
      "loss": 1.4112,
      "step": 55
    },
    {
      "epoch": 0.004108697047791117,
      "grad_norm": 1.4943321640476432,
      "learning_rate": 1.3691931540342298e-06,
      "loss": 0.7122,
      "step": 56
    },
    {
      "epoch": 0.004108697047791117,
      "lm_loss": 0.4557,
      "step": 56,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.004108697047791117,
      "lm_loss": 0.5827,
      "step": 56,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.004108697047791117,
      "lm_loss": 0.8618,
      "step": 56,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.004108697047791117,
      "lm_loss": 0.4597,
      "step": 56,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.004108697047791117,
      "lm_loss": 0.4735,
      "step": 56,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.004108697047791117,
      "lm_loss": 0.4007,
      "step": 56,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.004108697047791117,
      "lm_loss": 0.6592,
      "step": 56,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.004108697047791117,
      "lm_loss": 0.8286,
      "step": 56,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.004182066637930244,
      "grad_norm": 1.50430829424951,
      "learning_rate": 1.3936430317848413e-06,
      "loss": 0.7203,
      "step": 57
    },
    {
      "epoch": 0.004255436228069371,
      "grad_norm": 2.3711360482353245,
      "learning_rate": 1.4180929095354523e-06,
      "loss": 1.2618,
      "step": 58
    },
    {
      "epoch": 0.004328805818208498,
      "grad_norm": 2.4500379299704185,
      "learning_rate": 1.4425427872860636e-06,
      "loss": 1.3028,
      "step": 59
    },
    {
      "epoch": 0.004402175408347625,
      "grad_norm": 2.501116793226835,
      "learning_rate": 1.4669926650366751e-06,
      "loss": 1.2049,
      "step": 60
    },
    {
      "epoch": 0.004475544998486752,
      "grad_norm": 2.479200463200409,
      "learning_rate": 1.4914425427872862e-06,
      "loss": 1.225,
      "step": 61
    },
    {
      "epoch": 0.004548914588625879,
      "grad_norm": 1.4455095129808373,
      "learning_rate": 1.5158924205378975e-06,
      "loss": 0.7506,
      "step": 62
    },
    {
      "epoch": 0.004622284178765006,
      "grad_norm": 2.6519272583420204,
      "learning_rate": 1.5403422982885085e-06,
      "loss": 1.2466,
      "step": 63
    },
    {
      "epoch": 0.004695653768904133,
      "grad_norm": 2.5136259961277845,
      "learning_rate": 1.5647921760391198e-06,
      "loss": 1.3202,
      "step": 64
    },
    {
      "epoch": 0.004695653768904133,
      "lm_loss": 2.0462,
      "step": 64,
      "vm_loss": 0.0627
    },
    {
      "epoch": 0.004695653768904133,
      "lm_loss": 1.2755,
      "step": 64,
      "vm_loss": 0.1818
    },
    {
      "epoch": 0.004695653768904133,
      "lm_loss": 1.1287,
      "step": 64,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.004695653768904133,
      "lm_loss": 1.1339,
      "step": 64,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.004695653768904133,
      "lm_loss": 1.7322,
      "step": 64,
      "vm_loss": 0.1704
    },
    {
      "epoch": 0.004695653768904133,
      "lm_loss": 1.3073,
      "step": 64,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.004695653768904133,
      "lm_loss": 1.5756,
      "step": 64,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.004695653768904133,
      "lm_loss": 1.0006,
      "step": 64,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.00476902335904326,
      "grad_norm": 2.445380504474271,
      "learning_rate": 1.5892420537897313e-06,
      "loss": 1.3184,
      "step": 65
    },
    {
      "epoch": 0.004842392949182387,
      "grad_norm": 2.0850666157990605,
      "learning_rate": 1.6136919315403424e-06,
      "loss": 1.252,
      "step": 66
    },
    {
      "epoch": 0.004915762539321515,
      "grad_norm": 1.259893853385679,
      "learning_rate": 1.6381418092909537e-06,
      "loss": 0.7107,
      "step": 67
    },
    {
      "epoch": 0.004989132129460642,
      "grad_norm": 1.3692638959478625,
      "learning_rate": 1.6625916870415647e-06,
      "loss": 0.7246,
      "step": 68
    },
    {
      "epoch": 0.005062501719599769,
      "grad_norm": 1.280936120346811,
      "learning_rate": 1.6870415647921762e-06,
      "loss": 0.7148,
      "step": 69
    },
    {
      "epoch": 0.005135871309738896,
      "grad_norm": 2.418667178148446,
      "learning_rate": 1.7114914425427875e-06,
      "loss": 1.1576,
      "step": 70
    },
    {
      "epoch": 0.005209240899878023,
      "grad_norm": 2.411420090131142,
      "learning_rate": 1.7359413202933986e-06,
      "loss": 1.2053,
      "step": 71
    },
    {
      "epoch": 0.00528261049001715,
      "grad_norm": 2.531332806021802,
      "learning_rate": 1.7603911980440099e-06,
      "loss": 1.211,
      "step": 72
    },
    {
      "epoch": 0.00528261049001715,
      "lm_loss": 1.5899,
      "step": 72,
      "vm_loss": 0.0556
    },
    {
      "epoch": 0.00528261049001715,
      "lm_loss": 0.5588,
      "step": 72,
      "vm_loss": 0.0651
    },
    {
      "epoch": 0.00528261049001715,
      "lm_loss": 1.4613,
      "step": 72,
      "vm_loss": 0.2208
    },
    {
      "epoch": 0.00528261049001715,
      "lm_loss": 0.3118,
      "step": 72,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.00528261049001715,
      "lm_loss": 1.0938,
      "step": 72,
      "vm_loss": 0.0927
    },
    {
      "epoch": 0.00528261049001715,
      "lm_loss": 1.0177,
      "step": 72,
      "vm_loss": 0.204
    },
    {
      "epoch": 0.00528261049001715,
      "lm_loss": 0.9003,
      "step": 72,
      "vm_loss": 0.1123
    },
    {
      "epoch": 0.00528261049001715,
      "lm_loss": 1.0174,
      "step": 72,
      "vm_loss": 0.1604
    },
    {
      "epoch": 0.005355980080156277,
      "grad_norm": 2.504514845129083,
      "learning_rate": 1.7848410757946212e-06,
      "loss": 1.2332,
      "step": 73
    },
    {
      "epoch": 0.005429349670295404,
      "grad_norm": 1.0309942952855724,
      "learning_rate": 1.8092909535452324e-06,
      "loss": 0.7024,
      "step": 74
    },
    {
      "epoch": 0.005502719260434531,
      "grad_norm": 1.0999912330111044,
      "learning_rate": 1.8337408312958437e-06,
      "loss": 0.7169,
      "step": 75
    },
    {
      "epoch": 0.005576088850573658,
      "grad_norm": 1.0449442595993288,
      "learning_rate": 1.8581907090464548e-06,
      "loss": 0.7038,
      "step": 76
    },
    {
      "epoch": 0.005649458440712785,
      "grad_norm": 0.9454896127376247,
      "learning_rate": 1.8826405867970663e-06,
      "loss": 0.6755,
      "step": 77
    },
    {
      "epoch": 0.005722828030851912,
      "grad_norm": 2.586667567975974,
      "learning_rate": 1.9070904645476774e-06,
      "loss": 1.243,
      "step": 78
    },
    {
      "epoch": 0.00579619762099104,
      "grad_norm": 2.450368993771399,
      "learning_rate": 1.9315403422982887e-06,
      "loss": 1.3116,
      "step": 79
    },
    {
      "epoch": 0.005869567211130167,
      "grad_norm": 0.9149494783124169,
      "learning_rate": 1.9559902200489e-06,
      "loss": 0.7244,
      "step": 80
    },
    {
      "epoch": 0.005869567211130167,
      "lm_loss": 0.7109,
      "step": 80,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.005869567211130167,
      "lm_loss": 0.6934,
      "step": 80,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.005869567211130167,
      "lm_loss": 0.6761,
      "step": 80,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.005869567211130167,
      "lm_loss": 0.7947,
      "step": 80,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.005869567211130167,
      "lm_loss": 0.5915,
      "step": 80,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.005869567211130167,
      "lm_loss": 0.6986,
      "step": 80,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.005869567211130167,
      "lm_loss": 0.6776,
      "step": 80,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.005869567211130167,
      "lm_loss": 0.6858,
      "step": 80,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.005942936801269294,
      "grad_norm": 0.8518673023292882,
      "learning_rate": 1.9804400977995112e-06,
      "loss": 0.6806,
      "step": 81
    },
    {
      "epoch": 0.006016306391408421,
      "grad_norm": 0.8116677387669391,
      "learning_rate": 2.0048899755501223e-06,
      "loss": 0.6698,
      "step": 82
    },
    {
      "epoch": 0.006089675981547548,
      "grad_norm": 2.474076543797611,
      "learning_rate": 2.0293398533007334e-06,
      "loss": 1.2439,
      "step": 83
    },
    {
      "epoch": 0.006163045571686675,
      "grad_norm": 2.3349397783512273,
      "learning_rate": 2.053789731051345e-06,
      "loss": 1.1578,
      "step": 84
    },
    {
      "epoch": 0.006236415161825802,
      "grad_norm": 2.3058385282552813,
      "learning_rate": 2.0782396088019564e-06,
      "loss": 1.1995,
      "step": 85
    },
    {
      "epoch": 0.006309784751964929,
      "grad_norm": 0.7930275848226958,
      "learning_rate": 2.1026894865525674e-06,
      "loss": 0.6959,
      "step": 86
    },
    {
      "epoch": 0.006383154342104056,
      "grad_norm": 2.5055856243398247,
      "learning_rate": 2.127139364303179e-06,
      "loss": 1.1752,
      "step": 87
    },
    {
      "epoch": 0.006456523932243183,
      "grad_norm": 2.2943346200752766,
      "learning_rate": 2.15158924205379e-06,
      "loss": 1.1676,
      "step": 88
    },
    {
      "epoch": 0.006456523932243183,
      "lm_loss": 0.5793,
      "step": 88,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.006456523932243183,
      "lm_loss": 0.7291,
      "step": 88,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.006456523932243183,
      "lm_loss": 0.6091,
      "step": 88,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.006456523932243183,
      "lm_loss": 0.6872,
      "step": 88,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.006456523932243183,
      "lm_loss": 0.5887,
      "step": 88,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.006456523932243183,
      "lm_loss": 0.5634,
      "step": 88,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.006456523932243183,
      "lm_loss": 0.7188,
      "step": 88,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.006456523932243183,
      "lm_loss": 0.5098,
      "step": 88,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.00652989352238231,
      "grad_norm": 0.7772748545540918,
      "learning_rate": 2.176039119804401e-06,
      "loss": 0.69,
      "step": 89
    },
    {
      "epoch": 0.006603263112521437,
      "grad_norm": 0.7702799637947174,
      "learning_rate": 2.200488997555012e-06,
      "loss": 0.7088,
      "step": 90
    },
    {
      "epoch": 0.006676632702660564,
      "grad_norm": 2.824608357233149,
      "learning_rate": 2.2249388753056236e-06,
      "loss": 1.1738,
      "step": 91
    },
    {
      "epoch": 0.006750002292799692,
      "grad_norm": 2.324472388053963,
      "learning_rate": 2.249388753056235e-06,
      "loss": 1.2328,
      "step": 92
    },
    {
      "epoch": 0.006823371882938819,
      "grad_norm": 2.4540563557766712,
      "learning_rate": 2.273838630806846e-06,
      "loss": 1.2761,
      "step": 93
    },
    {
      "epoch": 0.006896741473077946,
      "grad_norm": 2.2376279559399186,
      "learning_rate": 2.2982885085574573e-06,
      "loss": 1.2368,
      "step": 94
    },
    {
      "epoch": 0.006970111063217073,
      "grad_norm": 0.7533772757659127,
      "learning_rate": 2.3227383863080683e-06,
      "loss": 0.6691,
      "step": 95
    },
    {
      "epoch": 0.0070434806533562,
      "grad_norm": 2.535573877624857,
      "learning_rate": 2.34718826405868e-06,
      "loss": 1.2755,
      "step": 96
    },
    {
      "epoch": 0.0070434806533562,
      "lm_loss": 1.4647,
      "step": 96,
      "vm_loss": 0.1108
    },
    {
      "epoch": 0.0070434806533562,
      "lm_loss": 1.3072,
      "step": 96,
      "vm_loss": 0.0669
    },
    {
      "epoch": 0.0070434806533562,
      "lm_loss": 1.2875,
      "step": 96,
      "vm_loss": 0.1474
    },
    {
      "epoch": 0.0070434806533562,
      "lm_loss": 0.9735,
      "step": 96,
      "vm_loss": 0.1663
    },
    {
      "epoch": 0.0070434806533562,
      "lm_loss": 0.551,
      "step": 96,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.0070434806533562,
      "lm_loss": 1.7839,
      "step": 96,
      "vm_loss": 0.1963
    },
    {
      "epoch": 0.0070434806533562,
      "lm_loss": 1.3147,
      "step": 96,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.0070434806533562,
      "lm_loss": 0.8116,
      "step": 96,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.007116850243495327,
      "grad_norm": 2.20110299901957,
      "learning_rate": 2.3716381418092913e-06,
      "loss": 1.246,
      "step": 97
    },
    {
      "epoch": 0.007190219833634454,
      "grad_norm": 0.7378645304932016,
      "learning_rate": 2.3960880195599024e-06,
      "loss": 0.705,
      "step": 98
    },
    {
      "epoch": 0.007263589423773581,
      "grad_norm": 2.292481511779046,
      "learning_rate": 2.420537897310514e-06,
      "loss": 1.2791,
      "step": 99
    },
    {
      "epoch": 0.007336959013912708,
      "grad_norm": 2.4890300348281325,
      "learning_rate": 2.444987775061125e-06,
      "loss": 1.1703,
      "step": 100
    },
    {
      "epoch": 0.007410328604051835,
      "grad_norm": 0.7254211965479692,
      "learning_rate": 2.469437652811736e-06,
      "loss": 0.6903,
      "step": 101
    },
    {
      "epoch": 0.007483698194190962,
      "grad_norm": 2.3659155031853527,
      "learning_rate": 2.4938875305623475e-06,
      "loss": 1.1553,
      "step": 102
    },
    {
      "epoch": 0.007557067784330089,
      "grad_norm": 2.4036039607657287,
      "learning_rate": 2.5183374083129586e-06,
      "loss": 1.1965,
      "step": 103
    },
    {
      "epoch": 0.007630437374469217,
      "grad_norm": 2.0081802103686597,
      "learning_rate": 2.54278728606357e-06,
      "loss": 1.1673,
      "step": 104
    },
    {
      "epoch": 0.007630437374469217,
      "lm_loss": 2.1072,
      "step": 104,
      "vm_loss": 0.0505
    },
    {
      "epoch": 0.007630437374469217,
      "lm_loss": 1.3338,
      "step": 104,
      "vm_loss": 0.1284
    },
    {
      "epoch": 0.007630437374469217,
      "lm_loss": 0.8886,
      "step": 104,
      "vm_loss": 0.2404
    },
    {
      "epoch": 0.007630437374469217,
      "lm_loss": 1.4987,
      "step": 104,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.007630437374469217,
      "lm_loss": 0.4289,
      "step": 104,
      "vm_loss": 0.0931
    },
    {
      "epoch": 0.007630437374469217,
      "lm_loss": 1.06,
      "step": 104,
      "vm_loss": 0.1321
    },
    {
      "epoch": 0.007630437374469217,
      "lm_loss": 1.0352,
      "step": 104,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.007630437374469217,
      "lm_loss": 1.024,
      "step": 104,
      "vm_loss": 0.1925
    },
    {
      "epoch": 0.007703806964608344,
      "grad_norm": 2.1351285456694846,
      "learning_rate": 2.567237163814181e-06,
      "loss": 1.1562,
      "step": 105
    },
    {
      "epoch": 0.007777176554747471,
      "grad_norm": 0.7096822818536741,
      "learning_rate": 2.5916870415647923e-06,
      "loss": 0.6928,
      "step": 106
    },
    {
      "epoch": 0.007850546144886597,
      "grad_norm": 2.281305927084199,
      "learning_rate": 2.6161369193154033e-06,
      "loss": 1.1652,
      "step": 107
    },
    {
      "epoch": 0.007923915735025724,
      "grad_norm": 0.6948248623639308,
      "learning_rate": 2.640586797066015e-06,
      "loss": 0.6816,
      "step": 108
    },
    {
      "epoch": 0.007997285325164851,
      "grad_norm": 0.6780092677008625,
      "learning_rate": 2.6650366748166263e-06,
      "loss": 0.6977,
      "step": 109
    },
    {
      "epoch": 0.00807065491530398,
      "grad_norm": 0.7419299452650174,
      "learning_rate": 2.6894865525672374e-06,
      "loss": 0.6718,
      "step": 110
    },
    {
      "epoch": 0.008144024505443107,
      "grad_norm": 2.3260082929792003,
      "learning_rate": 2.7139364303178485e-06,
      "loss": 1.2063,
      "step": 111
    },
    {
      "epoch": 0.008217394095582234,
      "grad_norm": 0.6794087034442512,
      "learning_rate": 2.7383863080684595e-06,
      "loss": 0.679,
      "step": 112
    },
    {
      "epoch": 0.008217394095582234,
      "lm_loss": 0.5922,
      "step": 112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.008217394095582234,
      "lm_loss": 0.7316,
      "step": 112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.008217394095582234,
      "lm_loss": 0.693,
      "step": 112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.008217394095582234,
      "lm_loss": 0.6791,
      "step": 112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.008217394095582234,
      "lm_loss": 0.751,
      "step": 112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.008217394095582234,
      "lm_loss": 0.5449,
      "step": 112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.008217394095582234,
      "lm_loss": 0.7829,
      "step": 112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.008217394095582234,
      "lm_loss": 0.6126,
      "step": 112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.008290763685721361,
      "grad_norm": 0.6909597499732625,
      "learning_rate": 2.762836185819071e-06,
      "loss": 0.7018,
      "step": 113
    },
    {
      "epoch": 0.008364133275860488,
      "grad_norm": 2.290016319363524,
      "learning_rate": 2.7872860635696825e-06,
      "loss": 1.04,
      "step": 114
    },
    {
      "epoch": 0.008437502865999615,
      "grad_norm": 0.6688291200909907,
      "learning_rate": 2.811735941320294e-06,
      "loss": 0.6403,
      "step": 115
    },
    {
      "epoch": 0.008510872456138742,
      "grad_norm": 2.2053366009837427,
      "learning_rate": 2.8361858190709047e-06,
      "loss": 1.1473,
      "step": 116
    },
    {
      "epoch": 0.00858424204627787,
      "grad_norm": 2.3073001301395095,
      "learning_rate": 2.860635696821516e-06,
      "loss": 1.0588,
      "step": 117
    },
    {
      "epoch": 0.008657611636416996,
      "grad_norm": 2.5596283094145313,
      "learning_rate": 2.8850855745721272e-06,
      "loss": 1.23,
      "step": 118
    },
    {
      "epoch": 0.008730981226556123,
      "grad_norm": 0.6634372968949911,
      "learning_rate": 2.9095354523227387e-06,
      "loss": 0.6771,
      "step": 119
    },
    {
      "epoch": 0.00880435081669525,
      "grad_norm": 2.2954747419994526,
      "learning_rate": 2.9339853300733502e-06,
      "loss": 1.1467,
      "step": 120
    },
    {
      "epoch": 0.00880435081669525,
      "lm_loss": 1.1295,
      "step": 120,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.00880435081669525,
      "lm_loss": 0.2776,
      "step": 120,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.00880435081669525,
      "lm_loss": 0.6046,
      "step": 120,
      "vm_loss": 0.1839
    },
    {
      "epoch": 0.00880435081669525,
      "lm_loss": 1.0377,
      "step": 120,
      "vm_loss": 0.0983
    },
    {
      "epoch": 0.00880435081669525,
      "lm_loss": 1.2643,
      "step": 120,
      "vm_loss": 0.1287
    },
    {
      "epoch": 0.00880435081669525,
      "lm_loss": 1.243,
      "step": 120,
      "vm_loss": 0.163
    },
    {
      "epoch": 0.00880435081669525,
      "lm_loss": 0.651,
      "step": 120,
      "vm_loss": 0.1662
    },
    {
      "epoch": 0.00880435081669525,
      "lm_loss": 0.9808,
      "step": 120,
      "vm_loss": 0.1913
    },
    {
      "epoch": 0.008877720406834377,
      "grad_norm": 2.2898981311591884,
      "learning_rate": 2.958435207823961e-06,
      "loss": 1.1693,
      "step": 121
    },
    {
      "epoch": 0.008951089996973504,
      "grad_norm": 0.6614849366356456,
      "learning_rate": 2.9828850855745724e-06,
      "loss": 0.6742,
      "step": 122
    },
    {
      "epoch": 0.009024459587112631,
      "grad_norm": 2.376702725243052,
      "learning_rate": 3.0073349633251834e-06,
      "loss": 1.0909,
      "step": 123
    },
    {
      "epoch": 0.009097829177251758,
      "grad_norm": 2.163284431930976,
      "learning_rate": 3.031784841075795e-06,
      "loss": 1.1785,
      "step": 124
    },
    {
      "epoch": 0.009171198767390885,
      "grad_norm": 2.1646094979626294,
      "learning_rate": 3.0562347188264064e-06,
      "loss": 1.2918,
      "step": 125
    },
    {
      "epoch": 0.009244568357530012,
      "grad_norm": 2.052767109097211,
      "learning_rate": 3.080684596577017e-06,
      "loss": 1.1845,
      "step": 126
    },
    {
      "epoch": 0.00931793794766914,
      "grad_norm": 2.174425541597895,
      "learning_rate": 3.1051344743276286e-06,
      "loss": 1.1337,
      "step": 127
    },
    {
      "epoch": 0.009391307537808266,
      "grad_norm": 0.6719381058168739,
      "learning_rate": 3.1295843520782396e-06,
      "loss": 0.6874,
      "step": 128
    },
    {
      "epoch": 0.009391307537808266,
      "lm_loss": 1.2218,
      "step": 128,
      "vm_loss": 0.048
    },
    {
      "epoch": 0.009391307537808266,
      "lm_loss": 1.2292,
      "step": 128,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.009391307537808266,
      "lm_loss": 0.4679,
      "step": 128,
      "vm_loss": 0.1793
    },
    {
      "epoch": 0.009391307537808266,
      "lm_loss": 0.5791,
      "step": 128,
      "vm_loss": 0.1737
    },
    {
      "epoch": 0.009391307537808266,
      "lm_loss": 1.1001,
      "step": 128,
      "vm_loss": 0.164
    },
    {
      "epoch": 0.009391307537808266,
      "lm_loss": 0.6597,
      "step": 128,
      "vm_loss": 0.0688
    },
    {
      "epoch": 0.009391307537808266,
      "lm_loss": 1.1929,
      "step": 128,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.009391307537808266,
      "lm_loss": 1.0523,
      "step": 128,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.009464677127947393,
      "grad_norm": 2.05640199707348,
      "learning_rate": 3.154034229828851e-06,
      "loss": 1.1522,
      "step": 129
    },
    {
      "epoch": 0.00953804671808652,
      "grad_norm": 2.2302341743590297,
      "learning_rate": 3.1784841075794626e-06,
      "loss": 1.2048,
      "step": 130
    },
    {
      "epoch": 0.009611416308225647,
      "grad_norm": 2.1918760771787063,
      "learning_rate": 3.2029339853300733e-06,
      "loss": 1.1024,
      "step": 131
    },
    {
      "epoch": 0.009684785898364775,
      "grad_norm": 2.384932451484826,
      "learning_rate": 3.2273838630806848e-06,
      "loss": 1.1809,
      "step": 132
    },
    {
      "epoch": 0.009758155488503902,
      "grad_norm": 2.125387777814191,
      "learning_rate": 3.2518337408312963e-06,
      "loss": 1.1125,
      "step": 133
    },
    {
      "epoch": 0.00983152507864303,
      "grad_norm": 0.654702406065746,
      "learning_rate": 3.2762836185819073e-06,
      "loss": 0.6705,
      "step": 134
    },
    {
      "epoch": 0.009904894668782157,
      "grad_norm": 2.358449633244695,
      "learning_rate": 3.300733496332519e-06,
      "loss": 1.1025,
      "step": 135
    },
    {
      "epoch": 0.009978264258921284,
      "grad_norm": 2.352745068451857,
      "learning_rate": 3.3251833740831295e-06,
      "loss": 1.101,
      "step": 136
    },
    {
      "epoch": 0.009978264258921284,
      "lm_loss": 1.6416,
      "step": 136,
      "vm_loss": 0.0955
    },
    {
      "epoch": 0.009978264258921284,
      "lm_loss": 0.9477,
      "step": 136,
      "vm_loss": 0.1519
    },
    {
      "epoch": 0.009978264258921284,
      "lm_loss": 1.2947,
      "step": 136,
      "vm_loss": 0.0791
    },
    {
      "epoch": 0.009978264258921284,
      "lm_loss": 1.059,
      "step": 136,
      "vm_loss": 0.0929
    },
    {
      "epoch": 0.009978264258921284,
      "lm_loss": 0.8822,
      "step": 136,
      "vm_loss": 0.2338
    },
    {
      "epoch": 0.009978264258921284,
      "lm_loss": 0.6428,
      "step": 136,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.009978264258921284,
      "lm_loss": 1.175,
      "step": 136,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.009978264258921284,
      "lm_loss": 0.9888,
      "step": 136,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.010051633849060411,
      "grad_norm": 2.12540003748675,
      "learning_rate": 3.349633251833741e-06,
      "loss": 1.083,
      "step": 137
    },
    {
      "epoch": 0.010125003439199538,
      "grad_norm": 2.3808481133036743,
      "learning_rate": 3.3740831295843525e-06,
      "loss": 1.1475,
      "step": 138
    },
    {
      "epoch": 0.010198373029338665,
      "grad_norm": 2.3301638342148543,
      "learning_rate": 3.3985330073349636e-06,
      "loss": 1.0124,
      "step": 139
    },
    {
      "epoch": 0.010271742619477792,
      "grad_norm": 0.6791708017709376,
      "learning_rate": 3.422982885085575e-06,
      "loss": 0.66,
      "step": 140
    },
    {
      "epoch": 0.01034511220961692,
      "grad_norm": 2.1440223626132147,
      "learning_rate": 3.4474327628361857e-06,
      "loss": 1.1869,
      "step": 141
    },
    {
      "epoch": 0.010418481799756046,
      "grad_norm": 2.2671969872143882,
      "learning_rate": 3.471882640586797e-06,
      "loss": 1.0709,
      "step": 142
    },
    {
      "epoch": 0.010491851389895173,
      "grad_norm": 2.3051810152669883,
      "learning_rate": 3.4963325183374087e-06,
      "loss": 1.1002,
      "step": 143
    },
    {
      "epoch": 0.0105652209800343,
      "grad_norm": 2.1515028663093845,
      "learning_rate": 3.5207823960880198e-06,
      "loss": 1.1689,
      "step": 144
    },
    {
      "epoch": 0.0105652209800343,
      "lm_loss": 1.3149,
      "step": 144,
      "vm_loss": 0.0488
    },
    {
      "epoch": 0.0105652209800343,
      "lm_loss": 0.6765,
      "step": 144,
      "vm_loss": 0.1328
    },
    {
      "epoch": 0.0105652209800343,
      "lm_loss": 0.5857,
      "step": 144,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.0105652209800343,
      "lm_loss": 1.1284,
      "step": 144,
      "vm_loss": 0.1327
    },
    {
      "epoch": 0.0105652209800343,
      "lm_loss": 0.9275,
      "step": 144,
      "vm_loss": 0.1309
    },
    {
      "epoch": 0.0105652209800343,
      "lm_loss": 0.9452,
      "step": 144,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.0105652209800343,
      "lm_loss": 0.8569,
      "step": 144,
      "vm_loss": 0.1085
    },
    {
      "epoch": 0.0105652209800343,
      "lm_loss": 0.9022,
      "step": 144,
      "vm_loss": 0.0635
    },
    {
      "epoch": 0.010638590570173427,
      "grad_norm": 2.160815841180979,
      "learning_rate": 3.5452322738386313e-06,
      "loss": 1.1314,
      "step": 145
    },
    {
      "epoch": 0.010711960160312554,
      "grad_norm": 2.205391711068033,
      "learning_rate": 3.5696821515892423e-06,
      "loss": 1.0413,
      "step": 146
    },
    {
      "epoch": 0.010785329750451681,
      "grad_norm": 0.654020343057737,
      "learning_rate": 3.5941320293398534e-06,
      "loss": 0.6658,
      "step": 147
    },
    {
      "epoch": 0.010858699340590808,
      "grad_norm": 2.2666450186172313,
      "learning_rate": 3.618581907090465e-06,
      "loss": 1.134,
      "step": 148
    },
    {
      "epoch": 0.010932068930729935,
      "grad_norm": 0.6836730620500845,
      "learning_rate": 3.6430317848410764e-06,
      "loss": 0.6494,
      "step": 149
    },
    {
      "epoch": 0.011005438520869062,
      "grad_norm": 0.6511813297023793,
      "learning_rate": 3.6674816625916875e-06,
      "loss": 0.6643,
      "step": 150
    },
    {
      "epoch": 0.01107880811100819,
      "grad_norm": 2.1765745926886084,
      "learning_rate": 3.6919315403422985e-06,
      "loss": 1.0278,
      "step": 151
    },
    {
      "epoch": 0.011152177701147317,
      "grad_norm": 2.366012576274979,
      "learning_rate": 3.7163814180929096e-06,
      "loss": 1.0544,
      "step": 152
    },
    {
      "epoch": 0.011152177701147317,
      "lm_loss": 1.4586,
      "step": 152,
      "vm_loss": 0.0512
    },
    {
      "epoch": 0.011152177701147317,
      "lm_loss": 0.4576,
      "step": 152,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.011152177701147317,
      "lm_loss": 1.0942,
      "step": 152,
      "vm_loss": 0.1157
    },
    {
      "epoch": 0.011152177701147317,
      "lm_loss": 1.0894,
      "step": 152,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.011152177701147317,
      "lm_loss": 1.847,
      "step": 152,
      "vm_loss": 0.1141
    },
    {
      "epoch": 0.011152177701147317,
      "lm_loss": 1.3238,
      "step": 152,
      "vm_loss": 0.18
    },
    {
      "epoch": 0.011152177701147317,
      "lm_loss": 0.3859,
      "step": 152,
      "vm_loss": 0.0984
    },
    {
      "epoch": 0.011152177701147317,
      "lm_loss": 0.6787,
      "step": 152,
      "vm_loss": 0.1524
    },
    {
      "epoch": 0.011225547291286444,
      "grad_norm": 2.1752358103680836,
      "learning_rate": 3.740831295843521e-06,
      "loss": 1.0854,
      "step": 153
    },
    {
      "epoch": 0.01129891688142557,
      "grad_norm": 2.2041304451567183,
      "learning_rate": 3.7652811735941326e-06,
      "loss": 1.1217,
      "step": 154
    },
    {
      "epoch": 0.011372286471564698,
      "grad_norm": 0.6585038821344238,
      "learning_rate": 3.7897310513447437e-06,
      "loss": 0.6602,
      "step": 155
    },
    {
      "epoch": 0.011445656061703825,
      "grad_norm": 2.2272350361182296,
      "learning_rate": 3.8141809290953547e-06,
      "loss": 1.0361,
      "step": 156
    },
    {
      "epoch": 0.011519025651842952,
      "grad_norm": 0.6370390632363525,
      "learning_rate": 3.838630806845966e-06,
      "loss": 0.6681,
      "step": 157
    },
    {
      "epoch": 0.01159239524198208,
      "grad_norm": 0.6662237296394309,
      "learning_rate": 3.863080684596577e-06,
      "loss": 0.6279,
      "step": 158
    },
    {
      "epoch": 0.011665764832121207,
      "grad_norm": 2.3193734611698313,
      "learning_rate": 3.887530562347189e-06,
      "loss": 1.1677,
      "step": 159
    },
    {
      "epoch": 0.011739134422260334,
      "grad_norm": 2.631698387251152,
      "learning_rate": 3.9119804400978e-06,
      "loss": 1.0949,
      "step": 160
    },
    {
      "epoch": 0.011739134422260334,
      "lm_loss": 1.4439,
      "step": 160,
      "vm_loss": 0.0805
    },
    {
      "epoch": 0.011739134422260334,
      "lm_loss": 1.8046,
      "step": 160,
      "vm_loss": 0.2292
    },
    {
      "epoch": 0.011739134422260334,
      "lm_loss": 1.248,
      "step": 160,
      "vm_loss": 0.1812
    },
    {
      "epoch": 0.011739134422260334,
      "lm_loss": 0.3998,
      "step": 160,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.011739134422260334,
      "lm_loss": 0.6201,
      "step": 160,
      "vm_loss": 0.0577
    },
    {
      "epoch": 0.011739134422260334,
      "lm_loss": 0.9236,
      "step": 160,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.011739134422260334,
      "lm_loss": 1.1917,
      "step": 160,
      "vm_loss": 0.1711
    },
    {
      "epoch": 0.011739134422260334,
      "lm_loss": 0.8849,
      "step": 160,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.011812504012399461,
      "grad_norm": 2.288005500293563,
      "learning_rate": 3.936430317848411e-06,
      "loss": 1.1423,
      "step": 161
    },
    {
      "epoch": 0.011885873602538588,
      "grad_norm": 2.096020452339059,
      "learning_rate": 3.9608801955990224e-06,
      "loss": 1.0802,
      "step": 162
    },
    {
      "epoch": 0.011959243192677715,
      "grad_norm": 2.4635184640578314,
      "learning_rate": 3.985330073349634e-06,
      "loss": 1.1182,
      "step": 163
    },
    {
      "epoch": 0.012032612782816842,
      "grad_norm": 0.6702255933390099,
      "learning_rate": 4.009779951100245e-06,
      "loss": 0.6896,
      "step": 164
    },
    {
      "epoch": 0.01210598237295597,
      "grad_norm": 2.4031757068588853,
      "learning_rate": 4.034229828850856e-06,
      "loss": 1.1818,
      "step": 165
    },
    {
      "epoch": 0.012179351963095096,
      "grad_norm": 0.631293031063405,
      "learning_rate": 4.058679706601467e-06,
      "loss": 0.6589,
      "step": 166
    },
    {
      "epoch": 0.012252721553234223,
      "grad_norm": 0.6322817022906407,
      "learning_rate": 4.083129584352078e-06,
      "loss": 0.671,
      "step": 167
    },
    {
      "epoch": 0.01232609114337335,
      "grad_norm": 2.295689666292653,
      "learning_rate": 4.10757946210269e-06,
      "loss": 1.2219,
      "step": 168
    },
    {
      "epoch": 0.01232609114337335,
      "lm_loss": 1.7736,
      "step": 168,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.01232609114337335,
      "lm_loss": 0.1343,
      "step": 168,
      "vm_loss": 0.0913
    },
    {
      "epoch": 0.01232609114337335,
      "lm_loss": 0.8716,
      "step": 168,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.01232609114337335,
      "lm_loss": 1.0951,
      "step": 168,
      "vm_loss": 0.1483
    },
    {
      "epoch": 0.01232609114337335,
      "lm_loss": 0.6951,
      "step": 168,
      "vm_loss": 0.139
    },
    {
      "epoch": 0.01232609114337335,
      "lm_loss": 0.5909,
      "step": 168,
      "vm_loss": 0.174
    },
    {
      "epoch": 0.01232609114337335,
      "lm_loss": 0.6455,
      "step": 168,
      "vm_loss": 0.1984
    },
    {
      "epoch": 0.01232609114337335,
      "lm_loss": 0.9684,
      "step": 168,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.012399460733512477,
      "grad_norm": 2.0959651854025187,
      "learning_rate": 4.132029339853301e-06,
      "loss": 1.0602,
      "step": 169
    },
    {
      "epoch": 0.012472830323651605,
      "grad_norm": 0.6304708227987803,
      "learning_rate": 4.156479217603913e-06,
      "loss": 0.6506,
      "step": 170
    },
    {
      "epoch": 0.012546199913790732,
      "grad_norm": 0.6515859698172024,
      "learning_rate": 4.180929095354523e-06,
      "loss": 0.6538,
      "step": 171
    },
    {
      "epoch": 0.012619569503929859,
      "grad_norm": 0.637671905988297,
      "learning_rate": 4.205378973105135e-06,
      "loss": 0.6594,
      "step": 172
    },
    {
      "epoch": 0.012692939094068986,
      "grad_norm": 0.6268765731435346,
      "learning_rate": 4.229828850855746e-06,
      "loss": 0.6499,
      "step": 173
    },
    {
      "epoch": 0.012766308684208113,
      "grad_norm": 0.6177726295586489,
      "learning_rate": 4.254278728606358e-06,
      "loss": 0.6607,
      "step": 174
    },
    {
      "epoch": 0.01283967827434724,
      "grad_norm": 0.6552730112733693,
      "learning_rate": 4.2787286063569685e-06,
      "loss": 0.6525,
      "step": 175
    },
    {
      "epoch": 0.012913047864486367,
      "grad_norm": 0.626687565172208,
      "learning_rate": 4.30317848410758e-06,
      "loss": 0.6589,
      "step": 176
    },
    {
      "epoch": 0.012913047864486367,
      "lm_loss": 0.5388,
      "step": 176,
      "vm_loss": 0.1239
    },
    {
      "epoch": 0.012913047864486367,
      "lm_loss": 1.4383,
      "step": 176,
      "vm_loss": 0.2174
    },
    {
      "epoch": 0.012913047864486367,
      "lm_loss": 0.2625,
      "step": 176,
      "vm_loss": 0.2199
    },
    {
      "epoch": 0.012913047864486367,
      "lm_loss": 1.1498,
      "step": 176,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.012913047864486367,
      "lm_loss": 0.2942,
      "step": 176,
      "vm_loss": 0.15
    },
    {
      "epoch": 0.012913047864486367,
      "lm_loss": 1.3695,
      "step": 176,
      "vm_loss": 0.2332
    },
    {
      "epoch": 0.012913047864486367,
      "lm_loss": 0.8206,
      "step": 176,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.012913047864486367,
      "lm_loss": 1.6608,
      "step": 176,
      "vm_loss": 0.199
    },
    {
      "epoch": 0.012986417454625494,
      "grad_norm": 2.67714639613813,
      "learning_rate": 4.327628361858191e-06,
      "loss": 1.1314,
      "step": 177
    },
    {
      "epoch": 0.01305978704476462,
      "grad_norm": 2.2015267488823453,
      "learning_rate": 4.352078239608802e-06,
      "loss": 1.1314,
      "step": 178
    },
    {
      "epoch": 0.013133156634903748,
      "grad_norm": 2.182926141536045,
      "learning_rate": 4.376528117359414e-06,
      "loss": 1.0627,
      "step": 179
    },
    {
      "epoch": 0.013206526225042875,
      "grad_norm": 2.499968037152802,
      "learning_rate": 4.400977995110024e-06,
      "loss": 1.1188,
      "step": 180
    },
    {
      "epoch": 0.013279895815182002,
      "grad_norm": 0.6338899652712139,
      "learning_rate": 4.425427872860636e-06,
      "loss": 0.6429,
      "step": 181
    },
    {
      "epoch": 0.013353265405321129,
      "grad_norm": 2.249134614028811,
      "learning_rate": 4.449877750611247e-06,
      "loss": 0.9954,
      "step": 182
    },
    {
      "epoch": 0.013426634995460257,
      "grad_norm": 2.6816911312388463,
      "learning_rate": 4.474327628361859e-06,
      "loss": 1.0666,
      "step": 183
    },
    {
      "epoch": 0.013500004585599384,
      "grad_norm": 2.0902552475694347,
      "learning_rate": 4.49877750611247e-06,
      "loss": 1.0634,
      "step": 184
    },
    {
      "epoch": 0.013500004585599384,
      "lm_loss": 1.4479,
      "step": 184,
      "vm_loss": 0.1027
    },
    {
      "epoch": 0.013500004585599384,
      "lm_loss": 0.7756,
      "step": 184,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.013500004585599384,
      "lm_loss": 0.3487,
      "step": 184,
      "vm_loss": 0.1657
    },
    {
      "epoch": 0.013500004585599384,
      "lm_loss": 0.4932,
      "step": 184,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.013500004585599384,
      "lm_loss": 1.0242,
      "step": 184,
      "vm_loss": 0.1899
    },
    {
      "epoch": 0.013500004585599384,
      "lm_loss": 0.2451,
      "step": 184,
      "vm_loss": 0.1101
    },
    {
      "epoch": 0.013500004585599384,
      "lm_loss": 1.1646,
      "step": 184,
      "vm_loss": 0.1868
    },
    {
      "epoch": 0.013500004585599384,
      "lm_loss": 1.1015,
      "step": 184,
      "vm_loss": 0.2003
    },
    {
      "epoch": 0.013573374175738511,
      "grad_norm": 2.2539564899009923,
      "learning_rate": 4.523227383863081e-06,
      "loss": 1.0868,
      "step": 185
    },
    {
      "epoch": 0.013646743765877638,
      "grad_norm": 2.1604746622965423,
      "learning_rate": 4.547677261613692e-06,
      "loss": 1.078,
      "step": 186
    },
    {
      "epoch": 0.013720113356016765,
      "grad_norm": 0.6694566572539817,
      "learning_rate": 4.572127139364304e-06,
      "loss": 0.6533,
      "step": 187
    },
    {
      "epoch": 0.013793482946155892,
      "grad_norm": 2.2460821242085593,
      "learning_rate": 4.5965770171149145e-06,
      "loss": 1.0789,
      "step": 188
    },
    {
      "epoch": 0.01386685253629502,
      "grad_norm": 0.6371383551642719,
      "learning_rate": 4.621026894865526e-06,
      "loss": 0.6492,
      "step": 189
    },
    {
      "epoch": 0.013940222126434147,
      "grad_norm": 0.6767965514266882,
      "learning_rate": 4.645476772616137e-06,
      "loss": 0.6657,
      "step": 190
    },
    {
      "epoch": 0.014013591716573274,
      "grad_norm": 2.353424544462374,
      "learning_rate": 4.669926650366748e-06,
      "loss": 1.0733,
      "step": 191
    },
    {
      "epoch": 0.0140869613067124,
      "grad_norm": 0.6335495139546046,
      "learning_rate": 4.69437652811736e-06,
      "loss": 0.6534,
      "step": 192
    },
    {
      "epoch": 0.0140869613067124,
      "lm_loss": 3.1451,
      "step": 192,
      "vm_loss": 0.0598
    },
    {
      "epoch": 0.0140869613067124,
      "lm_loss": 1.5743,
      "step": 192,
      "vm_loss": 0.1611
    },
    {
      "epoch": 0.0140869613067124,
      "lm_loss": 0.8929,
      "step": 192,
      "vm_loss": 0.1584
    },
    {
      "epoch": 0.0140869613067124,
      "lm_loss": 0.8848,
      "step": 192,
      "vm_loss": 0.1231
    },
    {
      "epoch": 0.0140869613067124,
      "lm_loss": 0.9927,
      "step": 192,
      "vm_loss": 0.2099
    },
    {
      "epoch": 0.0140869613067124,
      "lm_loss": 1.5859,
      "step": 192,
      "vm_loss": 0.0984
    },
    {
      "epoch": 0.0140869613067124,
      "lm_loss": 0.3592,
      "step": 192,
      "vm_loss": 0.1649
    },
    {
      "epoch": 0.0140869613067124,
      "lm_loss": 0.715,
      "step": 192,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.014160330896851528,
      "grad_norm": 2.332576159639579,
      "learning_rate": 4.718826405867971e-06,
      "loss": 1.1655,
      "step": 193
    },
    {
      "epoch": 0.014233700486990655,
      "grad_norm": 0.6480283762253788,
      "learning_rate": 4.743276283618583e-06,
      "loss": 0.6517,
      "step": 194
    },
    {
      "epoch": 0.014307070077129782,
      "grad_norm": 2.43192092016456,
      "learning_rate": 4.767726161369193e-06,
      "loss": 1.0801,
      "step": 195
    },
    {
      "epoch": 0.014380439667268909,
      "grad_norm": 2.416183823842868,
      "learning_rate": 4.792176039119805e-06,
      "loss": 1.0972,
      "step": 196
    },
    {
      "epoch": 0.014453809257408036,
      "grad_norm": 2.1236609204211616,
      "learning_rate": 4.816625916870416e-06,
      "loss": 0.9564,
      "step": 197
    },
    {
      "epoch": 0.014527178847547163,
      "grad_norm": 0.6544296197009957,
      "learning_rate": 4.841075794621028e-06,
      "loss": 0.66,
      "step": 198
    },
    {
      "epoch": 0.01460054843768629,
      "grad_norm": 0.6838799057434243,
      "learning_rate": 4.8655256723716385e-06,
      "loss": 0.6597,
      "step": 199
    },
    {
      "epoch": 0.014673918027825417,
      "grad_norm": 2.134885539574403,
      "learning_rate": 4.88997555012225e-06,
      "loss": 0.9768,
      "step": 200
    },
    {
      "epoch": 0.014673918027825417,
      "lm_loss": 1.2469,
      "step": 200,
      "vm_loss": 0.1015
    },
    {
      "epoch": 0.014673918027825417,
      "lm_loss": 1.7447,
      "step": 200,
      "vm_loss": 0.1341
    },
    {
      "epoch": 0.014673918027825417,
      "lm_loss": 0.2925,
      "step": 200,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.014673918027825417,
      "lm_loss": 0.6049,
      "step": 200,
      "vm_loss": 0.0565
    },
    {
      "epoch": 0.014673918027825417,
      "lm_loss": 1.2549,
      "step": 200,
      "vm_loss": 0.2229
    },
    {
      "epoch": 0.014673918027825417,
      "lm_loss": 0.503,
      "step": 200,
      "vm_loss": 0.1354
    },
    {
      "epoch": 0.014673918027825417,
      "lm_loss": 1.2802,
      "step": 200,
      "vm_loss": 0.1452
    },
    {
      "epoch": 0.014673918027825417,
      "lm_loss": 0.2495,
      "step": 200,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.014747287617964544,
      "grad_norm": 1.9709940911501016,
      "learning_rate": 4.914425427872861e-06,
      "loss": 1.0333,
      "step": 201
    },
    {
      "epoch": 0.01482065720810367,
      "grad_norm": 2.2670699669845704,
      "learning_rate": 4.938875305623472e-06,
      "loss": 1.0817,
      "step": 202
    },
    {
      "epoch": 0.014894026798242798,
      "grad_norm": 2.3687349482738456,
      "learning_rate": 4.963325183374084e-06,
      "loss": 1.088,
      "step": 203
    },
    {
      "epoch": 0.014967396388381925,
      "grad_norm": 0.6359316279080445,
      "learning_rate": 4.987775061124695e-06,
      "loss": 0.6401,
      "step": 204
    },
    {
      "epoch": 0.015040765978521052,
      "grad_norm": 2.2173704346283998,
      "learning_rate": 5.012224938875307e-06,
      "loss": 1.0535,
      "step": 205
    },
    {
      "epoch": 0.015114135568660179,
      "grad_norm": 2.164930830543929,
      "learning_rate": 5.036674816625917e-06,
      "loss": 1.0078,
      "step": 206
    },
    {
      "epoch": 0.015187505158799307,
      "grad_norm": 0.6301582801316657,
      "learning_rate": 5.061124694376528e-06,
      "loss": 0.6442,
      "step": 207
    },
    {
      "epoch": 0.015260874748938435,
      "grad_norm": 0.6097509679359269,
      "learning_rate": 5.08557457212714e-06,
      "loss": 0.6247,
      "step": 208
    },
    {
      "epoch": 0.015260874748938435,
      "lm_loss": 1.7298,
      "step": 208,
      "vm_loss": 0.0717
    },
    {
      "epoch": 0.015260874748938435,
      "lm_loss": 0.3234,
      "step": 208,
      "vm_loss": 0.1848
    },
    {
      "epoch": 0.015260874748938435,
      "lm_loss": 1.0461,
      "step": 208,
      "vm_loss": 0.1476
    },
    {
      "epoch": 0.015260874748938435,
      "lm_loss": 1.0021,
      "step": 208,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.015260874748938435,
      "lm_loss": 1.6362,
      "step": 208,
      "vm_loss": 0.172
    },
    {
      "epoch": 0.015260874748938435,
      "lm_loss": 1.4818,
      "step": 208,
      "vm_loss": 0.1204
    },
    {
      "epoch": 0.015260874748938435,
      "lm_loss": 0.4847,
      "step": 208,
      "vm_loss": 0.2034
    },
    {
      "epoch": 0.015260874748938435,
      "lm_loss": 0.2832,
      "step": 208,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.015334244339077562,
      "grad_norm": 2.316164671288134,
      "learning_rate": 5.110024449877751e-06,
      "loss": 1.076,
      "step": 209
    },
    {
      "epoch": 0.015407613929216689,
      "grad_norm": 2.3209923641789003,
      "learning_rate": 5.134474327628362e-06,
      "loss": 0.9523,
      "step": 210
    },
    {
      "epoch": 0.015480983519355816,
      "grad_norm": 2.209912744214072,
      "learning_rate": 5.158924205378973e-06,
      "loss": 1.0293,
      "step": 211
    },
    {
      "epoch": 0.015554353109494943,
      "grad_norm": 0.6385288160480319,
      "learning_rate": 5.1833740831295845e-06,
      "loss": 0.6195,
      "step": 212
    },
    {
      "epoch": 0.01562772269963407,
      "grad_norm": 2.226355851993297,
      "learning_rate": 5.207823960880196e-06,
      "loss": 1.0926,
      "step": 213
    },
    {
      "epoch": 0.015701092289773195,
      "grad_norm": 0.6612561563937556,
      "learning_rate": 5.232273838630807e-06,
      "loss": 0.6212,
      "step": 214
    },
    {
      "epoch": 0.015774461879912324,
      "grad_norm": 0.6460068636466283,
      "learning_rate": 5.256723716381419e-06,
      "loss": 0.6487,
      "step": 215
    },
    {
      "epoch": 0.01584783147005145,
      "grad_norm": 2.2123201189302066,
      "learning_rate": 5.28117359413203e-06,
      "loss": 1.0386,
      "step": 216
    },
    {
      "epoch": 0.01584783147005145,
      "lm_loss": 1.6225,
      "step": 216,
      "vm_loss": 0.0639
    },
    {
      "epoch": 0.01584783147005145,
      "lm_loss": 0.9407,
      "step": 216,
      "vm_loss": 0.1792
    },
    {
      "epoch": 0.01584783147005145,
      "lm_loss": 0.1998,
      "step": 216,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.01584783147005145,
      "lm_loss": 1.452,
      "step": 216,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.01584783147005145,
      "lm_loss": 1.2193,
      "step": 216,
      "vm_loss": 0.2675
    },
    {
      "epoch": 0.01584783147005145,
      "lm_loss": 1.1363,
      "step": 216,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.01584783147005145,
      "lm_loss": 1.1275,
      "step": 216,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.01584783147005145,
      "lm_loss": 1.2548,
      "step": 216,
      "vm_loss": 0.1458
    },
    {
      "epoch": 0.015921201060190578,
      "grad_norm": 1.969309074081657,
      "learning_rate": 5.30562347188264e-06,
      "loss": 1.0636,
      "step": 217
    },
    {
      "epoch": 0.015994570650329703,
      "grad_norm": 2.150531609851516,
      "learning_rate": 5.330073349633253e-06,
      "loss": 0.9531,
      "step": 218
    },
    {
      "epoch": 0.01606794024046883,
      "grad_norm": 2.1136919802934035,
      "learning_rate": 5.354523227383863e-06,
      "loss": 1.0569,
      "step": 219
    },
    {
      "epoch": 0.01614130983060796,
      "grad_norm": 2.2885029739208442,
      "learning_rate": 5.378973105134475e-06,
      "loss": 1.0785,
      "step": 220
    },
    {
      "epoch": 0.016214679420747086,
      "grad_norm": 2.023084706810908,
      "learning_rate": 5.403422982885086e-06,
      "loss": 0.98,
      "step": 221
    },
    {
      "epoch": 0.016288049010886214,
      "grad_norm": 2.300821723493127,
      "learning_rate": 5.427872860635697e-06,
      "loss": 1.0264,
      "step": 222
    },
    {
      "epoch": 0.01636141860102534,
      "grad_norm": 2.0203278504593776,
      "learning_rate": 5.452322738386308e-06,
      "loss": 0.9938,
      "step": 223
    },
    {
      "epoch": 0.01643478819116447,
      "grad_norm": 2.1487445227241224,
      "learning_rate": 5.476772616136919e-06,
      "loss": 1.0587,
      "step": 224
    },
    {
      "epoch": 0.01643478819116447,
      "lm_loss": 1.1844,
      "step": 224,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.01643478819116447,
      "lm_loss": 0.5384,
      "step": 224,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.01643478819116447,
      "lm_loss": 0.4413,
      "step": 224,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.01643478819116447,
      "lm_loss": 0.736,
      "step": 224,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.01643478819116447,
      "lm_loss": 1.1513,
      "step": 224,
      "vm_loss": 0.0893
    },
    {
      "epoch": 0.01643478819116447,
      "lm_loss": 0.3003,
      "step": 224,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.01643478819116447,
      "lm_loss": 0.8195,
      "step": 224,
      "vm_loss": 0.1256
    },
    {
      "epoch": 0.01643478819116447,
      "lm_loss": 0.5503,
      "step": 224,
      "vm_loss": 0.1115
    },
    {
      "epoch": 0.016508157781303594,
      "grad_norm": 1.9309904553283577,
      "learning_rate": 5.501222493887531e-06,
      "loss": 0.9642,
      "step": 225
    },
    {
      "epoch": 0.016581527371442722,
      "grad_norm": 2.3259113347994593,
      "learning_rate": 5.525672371638142e-06,
      "loss": 1.0238,
      "step": 226
    },
    {
      "epoch": 0.016654896961581848,
      "grad_norm": 2.4872313462951374,
      "learning_rate": 5.550122249388753e-06,
      "loss": 1.0021,
      "step": 227
    },
    {
      "epoch": 0.016728266551720977,
      "grad_norm": 2.0845832397958,
      "learning_rate": 5.574572127139365e-06,
      "loss": 0.9794,
      "step": 228
    },
    {
      "epoch": 0.016801636141860102,
      "grad_norm": 0.739139045783906,
      "learning_rate": 5.599022004889976e-06,
      "loss": 0.6314,
      "step": 229
    },
    {
      "epoch": 0.01687500573199923,
      "grad_norm": 2.2576151852800495,
      "learning_rate": 5.623471882640588e-06,
      "loss": 1.084,
      "step": 230
    },
    {
      "epoch": 0.016948375322138356,
      "grad_norm": 2.1757053713045855,
      "learning_rate": 5.647921760391199e-06,
      "loss": 1.0296,
      "step": 231
    },
    {
      "epoch": 0.017021744912277485,
      "grad_norm": 2.1461691138139383,
      "learning_rate": 5.672371638141809e-06,
      "loss": 1.0451,
      "step": 232
    },
    {
      "epoch": 0.017021744912277485,
      "lm_loss": 1.5605,
      "step": 232,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.017021744912277485,
      "lm_loss": 1.1024,
      "step": 232,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.017021744912277485,
      "lm_loss": 0.4997,
      "step": 232,
      "vm_loss": 0.1254
    },
    {
      "epoch": 0.017021744912277485,
      "lm_loss": 0.4376,
      "step": 232,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.017021744912277485,
      "lm_loss": 0.8053,
      "step": 232,
      "vm_loss": 0.0924
    },
    {
      "epoch": 0.017021744912277485,
      "lm_loss": 1.0854,
      "step": 232,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.017021744912277485,
      "lm_loss": 0.8764,
      "step": 232,
      "vm_loss": 0.1401
    },
    {
      "epoch": 0.017021744912277485,
      "lm_loss": 1.2507,
      "step": 232,
      "vm_loss": 0.0853
    },
    {
      "epoch": 0.01709511450241661,
      "grad_norm": 2.0537784556706593,
      "learning_rate": 5.696821515892421e-06,
      "loss": 0.9914,
      "step": 233
    },
    {
      "epoch": 0.01716848409255574,
      "grad_norm": 2.054649065275807,
      "learning_rate": 5.721271393643032e-06,
      "loss": 1.0343,
      "step": 234
    },
    {
      "epoch": 0.017241853682694864,
      "grad_norm": 1.8973266943757794,
      "learning_rate": 5.745721271393644e-06,
      "loss": 1.0315,
      "step": 235
    },
    {
      "epoch": 0.017315223272833993,
      "grad_norm": 2.4142327768966942,
      "learning_rate": 5.7701711491442545e-06,
      "loss": 1.0834,
      "step": 236
    },
    {
      "epoch": 0.017388592862973118,
      "grad_norm": 2.199751118455896,
      "learning_rate": 5.794621026894865e-06,
      "loss": 1.0952,
      "step": 237
    },
    {
      "epoch": 0.017461962453112247,
      "grad_norm": 2.1012757031702747,
      "learning_rate": 5.8190709046454775e-06,
      "loss": 0.9222,
      "step": 238
    },
    {
      "epoch": 0.017535332043251372,
      "grad_norm": 2.5514180383813887,
      "learning_rate": 5.843520782396088e-06,
      "loss": 1.0308,
      "step": 239
    },
    {
      "epoch": 0.0176087016333905,
      "grad_norm": 2.4840951721374074,
      "learning_rate": 5.8679706601467004e-06,
      "loss": 1.0224,
      "step": 240
    },
    {
      "epoch": 0.0176087016333905,
      "lm_loss": 0.6979,
      "step": 240,
      "vm_loss": 0.0503
    },
    {
      "epoch": 0.0176087016333905,
      "lm_loss": 0.3103,
      "step": 240,
      "vm_loss": 0.1539
    },
    {
      "epoch": 0.0176087016333905,
      "lm_loss": 0.3698,
      "step": 240,
      "vm_loss": 0.2043
    },
    {
      "epoch": 0.0176087016333905,
      "lm_loss": 0.9774,
      "step": 240,
      "vm_loss": 0.1897
    },
    {
      "epoch": 0.0176087016333905,
      "lm_loss": 0.3663,
      "step": 240,
      "vm_loss": 0.1331
    },
    {
      "epoch": 0.0176087016333905,
      "lm_loss": 0.4835,
      "step": 240,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.0176087016333905,
      "lm_loss": 0.1012,
      "step": 240,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.0176087016333905,
      "lm_loss": 1.1856,
      "step": 240,
      "vm_loss": 0.1032
    },
    {
      "epoch": 0.017682071223529626,
      "grad_norm": 2.562352547337534,
      "learning_rate": 5.892420537897311e-06,
      "loss": 0.97,
      "step": 241
    },
    {
      "epoch": 0.017755440813668755,
      "grad_norm": 2.3154566900934976,
      "learning_rate": 5.916870415647922e-06,
      "loss": 1.0562,
      "step": 242
    },
    {
      "epoch": 0.017828810403807883,
      "grad_norm": 1.9029830996582382,
      "learning_rate": 5.941320293398534e-06,
      "loss": 1.0253,
      "step": 243
    },
    {
      "epoch": 0.01790217999394701,
      "grad_norm": 2.4769905331050723,
      "learning_rate": 5.965770171149145e-06,
      "loss": 1.0372,
      "step": 244
    },
    {
      "epoch": 0.017975549584086137,
      "grad_norm": 2.1766081739248504,
      "learning_rate": 5.990220048899756e-06,
      "loss": 1.0347,
      "step": 245
    },
    {
      "epoch": 0.018048919174225263,
      "grad_norm": 2.0469748614503436,
      "learning_rate": 6.014669926650367e-06,
      "loss": 0.8519,
      "step": 246
    },
    {
      "epoch": 0.01812228876436439,
      "grad_norm": 2.3032831547308623,
      "learning_rate": 6.039119804400978e-06,
      "loss": 1.0421,
      "step": 247
    },
    {
      "epoch": 0.018195658354503517,
      "grad_norm": 2.058350954335164,
      "learning_rate": 6.06356968215159e-06,
      "loss": 1.0014,
      "step": 248
    },
    {
      "epoch": 0.018195658354503517,
      "lm_loss": 1.5075,
      "step": 248,
      "vm_loss": 0.0654
    },
    {
      "epoch": 0.018195658354503517,
      "lm_loss": 0.8211,
      "step": 248,
      "vm_loss": 0.2056
    },
    {
      "epoch": 0.018195658354503517,
      "lm_loss": 0.2395,
      "step": 248,
      "vm_loss": 0.0687
    },
    {
      "epoch": 0.018195658354503517,
      "lm_loss": 1.359,
      "step": 248,
      "vm_loss": 0.1452
    },
    {
      "epoch": 0.018195658354503517,
      "lm_loss": 0.9168,
      "step": 248,
      "vm_loss": 0.151
    },
    {
      "epoch": 0.018195658354503517,
      "lm_loss": 1.5074,
      "step": 248,
      "vm_loss": 0.1876
    },
    {
      "epoch": 0.018195658354503517,
      "lm_loss": 0.8279,
      "step": 248,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.018195658354503517,
      "lm_loss": 1.0609,
      "step": 248,
      "vm_loss": 0.1135
    },
    {
      "epoch": 0.018269027944642646,
      "grad_norm": 2.040122173016839,
      "learning_rate": 6.0880195599022005e-06,
      "loss": 1.0789,
      "step": 249
    },
    {
      "epoch": 0.01834239753478177,
      "grad_norm": 2.090086400161521,
      "learning_rate": 6.112469437652813e-06,
      "loss": 1.0848,
      "step": 250
    },
    {
      "epoch": 0.0184157671249209,
      "grad_norm": 2.278617203624695,
      "learning_rate": 6.1369193154034235e-06,
      "loss": 1.0289,
      "step": 251
    },
    {
      "epoch": 0.018489136715060025,
      "grad_norm": 2.195464655189527,
      "learning_rate": 6.161369193154034e-06,
      "loss": 0.9564,
      "step": 252
    },
    {
      "epoch": 0.018562506305199154,
      "grad_norm": 0.886941830877389,
      "learning_rate": 6.1858190709046465e-06,
      "loss": 0.6672,
      "step": 253
    },
    {
      "epoch": 0.01863587589533828,
      "grad_norm": 0.7852281836767879,
      "learning_rate": 6.210268948655257e-06,
      "loss": 0.6524,
      "step": 254
    },
    {
      "epoch": 0.018709245485477408,
      "grad_norm": 0.6932945813408519,
      "learning_rate": 6.234718826405869e-06,
      "loss": 0.6665,
      "step": 255
    },
    {
      "epoch": 0.018782615075616533,
      "grad_norm": 2.291706441599183,
      "learning_rate": 6.259168704156479e-06,
      "loss": 0.9815,
      "step": 256
    },
    {
      "epoch": 0.018782615075616533,
      "lm_loss": 1.8391,
      "step": 256,
      "vm_loss": 0.0995
    },
    {
      "epoch": 0.018782615075616533,
      "lm_loss": 0.69,
      "step": 256,
      "vm_loss": 0.0732
    },
    {
      "epoch": 0.018782615075616533,
      "lm_loss": 0.7772,
      "step": 256,
      "vm_loss": 0.1243
    },
    {
      "epoch": 0.018782615075616533,
      "lm_loss": 0.5833,
      "step": 256,
      "vm_loss": 0.1136
    },
    {
      "epoch": 0.018782615075616533,
      "lm_loss": 0.2715,
      "step": 256,
      "vm_loss": 0.1819
    },
    {
      "epoch": 0.018782615075616533,
      "lm_loss": 0.9493,
      "step": 256,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.018782615075616533,
      "lm_loss": 1.2078,
      "step": 256,
      "vm_loss": 0.1462
    },
    {
      "epoch": 0.018782615075616533,
      "lm_loss": 0.337,
      "step": 256,
      "vm_loss": 0.1469
    },
    {
      "epoch": 0.01885598466575566,
      "grad_norm": 2.026361375355289,
      "learning_rate": 6.283618581907091e-06,
      "loss": 0.9806,
      "step": 257
    },
    {
      "epoch": 0.018929354255894787,
      "grad_norm": 0.9914502475248819,
      "learning_rate": 6.308068459657702e-06,
      "loss": 0.6349,
      "step": 258
    },
    {
      "epoch": 0.019002723846033916,
      "grad_norm": 2.224055268878649,
      "learning_rate": 6.332518337408313e-06,
      "loss": 1.0647,
      "step": 259
    },
    {
      "epoch": 0.01907609343617304,
      "grad_norm": 2.2199538266315755,
      "learning_rate": 6.356968215158925e-06,
      "loss": 1.0305,
      "step": 260
    },
    {
      "epoch": 0.01914946302631217,
      "grad_norm": 2.034544304631688,
      "learning_rate": 6.381418092909536e-06,
      "loss": 1.0649,
      "step": 261
    },
    {
      "epoch": 0.019222832616451295,
      "grad_norm": 2.1796235436765916,
      "learning_rate": 6.4058679706601466e-06,
      "loss": 1.0582,
      "step": 262
    },
    {
      "epoch": 0.019296202206590424,
      "grad_norm": 2.265806690655013,
      "learning_rate": 6.430317848410759e-06,
      "loss": 1.0369,
      "step": 263
    },
    {
      "epoch": 0.01936957179672955,
      "grad_norm": 2.03257176546188,
      "learning_rate": 6.4547677261613696e-06,
      "loss": 0.9193,
      "step": 264
    },
    {
      "epoch": 0.01936957179672955,
      "lm_loss": 0.7305,
      "step": 264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.01936957179672955,
      "lm_loss": 0.5136,
      "step": 264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.01936957179672955,
      "lm_loss": 0.6793,
      "step": 264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.01936957179672955,
      "lm_loss": 0.6267,
      "step": 264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.01936957179672955,
      "lm_loss": 0.9346,
      "step": 264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.01936957179672955,
      "lm_loss": 0.6644,
      "step": 264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.01936957179672955,
      "lm_loss": 0.6339,
      "step": 264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.01936957179672955,
      "lm_loss": 0.7604,
      "step": 264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.019442941386868678,
      "grad_norm": 0.8849607669417269,
      "learning_rate": 6.479217603911981e-06,
      "loss": 0.6472,
      "step": 265
    },
    {
      "epoch": 0.019516310977007803,
      "grad_norm": 2.746234945454752,
      "learning_rate": 6.5036674816625926e-06,
      "loss": 1.0239,
      "step": 266
    },
    {
      "epoch": 0.019589680567146932,
      "grad_norm": 2.255877150680161,
      "learning_rate": 6.528117359413203e-06,
      "loss": 1.128,
      "step": 267
    },
    {
      "epoch": 0.01966305015728606,
      "grad_norm": 2.158620449108679,
      "learning_rate": 6.552567237163815e-06,
      "loss": 1.0882,
      "step": 268
    },
    {
      "epoch": 0.019736419747425186,
      "grad_norm": 2.250812740405228,
      "learning_rate": 6.577017114914425e-06,
      "loss": 1.0931,
      "step": 269
    },
    {
      "epoch": 0.019809789337564315,
      "grad_norm": 2.1034854739273054,
      "learning_rate": 6.601466992665038e-06,
      "loss": 1.0415,
      "step": 270
    },
    {
      "epoch": 0.01988315892770344,
      "grad_norm": 2.1696672545102333,
      "learning_rate": 6.625916870415648e-06,
      "loss": 1.0207,
      "step": 271
    },
    {
      "epoch": 0.01995652851784257,
      "grad_norm": 2.2129381140366973,
      "learning_rate": 6.650366748166259e-06,
      "loss": 1.0805,
      "step": 272
    },
    {
      "epoch": 0.01995652851784257,
      "lm_loss": 1.3689,
      "step": 272,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.01995652851784257,
      "lm_loss": 0.3798,
      "step": 272,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.01995652851784257,
      "lm_loss": 1.2705,
      "step": 272,
      "vm_loss": 0.2185
    },
    {
      "epoch": 0.01995652851784257,
      "lm_loss": 1.0246,
      "step": 272,
      "vm_loss": 0.2398
    },
    {
      "epoch": 0.01995652851784257,
      "lm_loss": 0.5951,
      "step": 272,
      "vm_loss": 0.1707
    },
    {
      "epoch": 0.01995652851784257,
      "lm_loss": 0.3632,
      "step": 272,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.01995652851784257,
      "lm_loss": 0.5376,
      "step": 272,
      "vm_loss": 0.0764
    },
    {
      "epoch": 0.01995652851784257,
      "lm_loss": 0.2568,
      "step": 272,
      "vm_loss": 0.2332
    },
    {
      "epoch": 0.020029898107981694,
      "grad_norm": 2.4576892142524365,
      "learning_rate": 6.674816625916871e-06,
      "loss": 1.0766,
      "step": 273
    },
    {
      "epoch": 0.020103267698120823,
      "grad_norm": 2.077157111283198,
      "learning_rate": 6.699266503667482e-06,
      "loss": 1.0361,
      "step": 274
    },
    {
      "epoch": 0.020176637288259948,
      "grad_norm": 2.867788522754872,
      "learning_rate": 6.723716381418094e-06,
      "loss": 1.0842,
      "step": 275
    },
    {
      "epoch": 0.020250006878399077,
      "grad_norm": 2.2323342351728845,
      "learning_rate": 6.748166259168705e-06,
      "loss": 1.097,
      "step": 276
    },
    {
      "epoch": 0.020323376468538202,
      "grad_norm": 0.9806967566430834,
      "learning_rate": 6.772616136919316e-06,
      "loss": 0.6475,
      "step": 277
    },
    {
      "epoch": 0.02039674605867733,
      "grad_norm": 0.8897438350732189,
      "learning_rate": 6.797066014669927e-06,
      "loss": 0.6331,
      "step": 278
    },
    {
      "epoch": 0.020470115648816456,
      "grad_norm": 2.463073666472682,
      "learning_rate": 6.821515892420539e-06,
      "loss": 1.2184,
      "step": 279
    },
    {
      "epoch": 0.020543485238955585,
      "grad_norm": 2.427337528154975,
      "learning_rate": 6.84596577017115e-06,
      "loss": 0.9492,
      "step": 280
    },
    {
      "epoch": 0.020543485238955585,
      "lm_loss": 1.6845,
      "step": 280,
      "vm_loss": 0.1888
    },
    {
      "epoch": 0.020543485238955585,
      "lm_loss": 0.8357,
      "step": 280,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.020543485238955585,
      "lm_loss": 1.4928,
      "step": 280,
      "vm_loss": 0.1916
    },
    {
      "epoch": 0.020543485238955585,
      "lm_loss": 1.5484,
      "step": 280,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.020543485238955585,
      "lm_loss": 0.3154,
      "step": 280,
      "vm_loss": 0.0872
    },
    {
      "epoch": 0.020543485238955585,
      "lm_loss": 0.7552,
      "step": 280,
      "vm_loss": 0.142
    },
    {
      "epoch": 0.020543485238955585,
      "lm_loss": 0.2559,
      "step": 280,
      "vm_loss": 0.1986
    },
    {
      "epoch": 0.020543485238955585,
      "lm_loss": 1.2739,
      "step": 280,
      "vm_loss": 0.2169
    },
    {
      "epoch": 0.02061685482909471,
      "grad_norm": 1.9724368345402177,
      "learning_rate": 6.870415647921761e-06,
      "loss": 0.9062,
      "step": 281
    },
    {
      "epoch": 0.02069022441923384,
      "grad_norm": 2.096648433257418,
      "learning_rate": 6.894865525672371e-06,
      "loss": 0.9907,
      "step": 282
    },
    {
      "epoch": 0.020763594009372964,
      "grad_norm": 2.233869599599319,
      "learning_rate": 6.919315403422984e-06,
      "loss": 1.0224,
      "step": 283
    },
    {
      "epoch": 0.020836963599512093,
      "grad_norm": 2.503118312211794,
      "learning_rate": 6.943765281173594e-06,
      "loss": 0.9939,
      "step": 284
    },
    {
      "epoch": 0.020910333189651218,
      "grad_norm": 1.2227689977838248,
      "learning_rate": 6.968215158924207e-06,
      "loss": 0.6517,
      "step": 285
    },
    {
      "epoch": 0.020983702779790347,
      "grad_norm": 2.285504286693405,
      "learning_rate": 6.992665036674817e-06,
      "loss": 1.0361,
      "step": 286
    },
    {
      "epoch": 0.021057072369929472,
      "grad_norm": 2.3450492959579012,
      "learning_rate": 7.017114914425428e-06,
      "loss": 0.9937,
      "step": 287
    },
    {
      "epoch": 0.0211304419600686,
      "grad_norm": 2.1684425991684355,
      "learning_rate": 7.0415647921760395e-06,
      "loss": 0.917,
      "step": 288
    },
    {
      "epoch": 0.0211304419600686,
      "lm_loss": 1.3784,
      "step": 288,
      "vm_loss": 0.0626
    },
    {
      "epoch": 0.0211304419600686,
      "lm_loss": 1.3058,
      "step": 288,
      "vm_loss": 0.1936
    },
    {
      "epoch": 0.0211304419600686,
      "lm_loss": 0.8727,
      "step": 288,
      "vm_loss": 0.2001
    },
    {
      "epoch": 0.0211304419600686,
      "lm_loss": 1.1614,
      "step": 288,
      "vm_loss": 0.1748
    },
    {
      "epoch": 0.0211304419600686,
      "lm_loss": 1.3904,
      "step": 288,
      "vm_loss": 0.2
    },
    {
      "epoch": 0.0211304419600686,
      "lm_loss": 1.2482,
      "step": 288,
      "vm_loss": 0.186
    },
    {
      "epoch": 0.0211304419600686,
      "lm_loss": 0.9536,
      "step": 288,
      "vm_loss": 0.1638
    },
    {
      "epoch": 0.0211304419600686,
      "lm_loss": 0.7573,
      "step": 288,
      "vm_loss": 0.0779
    },
    {
      "epoch": 0.021203811550207726,
      "grad_norm": 2.2615054316673926,
      "learning_rate": 7.066014669926651e-06,
      "loss": 1.0806,
      "step": 289
    },
    {
      "epoch": 0.021277181140346855,
      "grad_norm": 2.093534456418319,
      "learning_rate": 7.0904645476772625e-06,
      "loss": 1.0788,
      "step": 290
    },
    {
      "epoch": 0.02135055073048598,
      "grad_norm": 2.303108031977538,
      "learning_rate": 7.114914425427873e-06,
      "loss": 1.0385,
      "step": 291
    },
    {
      "epoch": 0.02142392032062511,
      "grad_norm": 0.7783826009431124,
      "learning_rate": 7.139364303178485e-06,
      "loss": 0.6409,
      "step": 292
    },
    {
      "epoch": 0.021497289910764238,
      "grad_norm": 2.281280731717503,
      "learning_rate": 7.163814180929096e-06,
      "loss": 1.0042,
      "step": 293
    },
    {
      "epoch": 0.021570659500903363,
      "grad_norm": 2.4224660052153464,
      "learning_rate": 7.188264058679707e-06,
      "loss": 0.9372,
      "step": 294
    },
    {
      "epoch": 0.02164402909104249,
      "grad_norm": 2.3110981125077887,
      "learning_rate": 7.212713936430319e-06,
      "loss": 0.975,
      "step": 295
    },
    {
      "epoch": 0.021717398681181617,
      "grad_norm": 2.1514932796139603,
      "learning_rate": 7.23716381418093e-06,
      "loss": 1.013,
      "step": 296
    },
    {
      "epoch": 0.021717398681181617,
      "lm_loss": 1.3092,
      "step": 296,
      "vm_loss": 0.0548
    },
    {
      "epoch": 0.021717398681181617,
      "lm_loss": 1.0106,
      "step": 296,
      "vm_loss": 0.1204
    },
    {
      "epoch": 0.021717398681181617,
      "lm_loss": 1.0317,
      "step": 296,
      "vm_loss": 0.1302
    },
    {
      "epoch": 0.021717398681181617,
      "lm_loss": 0.2433,
      "step": 296,
      "vm_loss": 0.1843
    },
    {
      "epoch": 0.021717398681181617,
      "lm_loss": 0.7507,
      "step": 296,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.021717398681181617,
      "lm_loss": 0.1863,
      "step": 296,
      "vm_loss": 0.1269
    },
    {
      "epoch": 0.021717398681181617,
      "lm_loss": 1.0444,
      "step": 296,
      "vm_loss": 0.1634
    },
    {
      "epoch": 0.021717398681181617,
      "lm_loss": 0.7214,
      "step": 296,
      "vm_loss": 0.1027
    },
    {
      "epoch": 0.021790768271320746,
      "grad_norm": 2.088131063821777,
      "learning_rate": 7.2616136919315404e-06,
      "loss": 0.9299,
      "step": 297
    },
    {
      "epoch": 0.02186413786145987,
      "grad_norm": 2.107195902931495,
      "learning_rate": 7.286063569682153e-06,
      "loss": 1.0484,
      "step": 298
    },
    {
      "epoch": 0.021937507451599,
      "grad_norm": 2.1603513623157315,
      "learning_rate": 7.3105134474327634e-06,
      "loss": 0.9921,
      "step": 299
    },
    {
      "epoch": 0.022010877041738125,
      "grad_norm": 2.0758796252139455,
      "learning_rate": 7.334963325183375e-06,
      "loss": 1.0381,
      "step": 300
    },
    {
      "epoch": 0.022084246631877254,
      "grad_norm": 0.747144050207587,
      "learning_rate": 7.3594132029339856e-06,
      "loss": 0.6388,
      "step": 301
    },
    {
      "epoch": 0.02215761622201638,
      "grad_norm": 2.342845545125978,
      "learning_rate": 7.383863080684597e-06,
      "loss": 0.9561,
      "step": 302
    },
    {
      "epoch": 0.022230985812155508,
      "grad_norm": 2.185912317870724,
      "learning_rate": 7.4083129584352086e-06,
      "loss": 1.0771,
      "step": 303
    },
    {
      "epoch": 0.022304355402294633,
      "grad_norm": 2.1231526412450465,
      "learning_rate": 7.432762836185819e-06,
      "loss": 1.0457,
      "step": 304
    },
    {
      "epoch": 0.022304355402294633,
      "lm_loss": 0.4494,
      "step": 304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.022304355402294633,
      "lm_loss": 0.7274,
      "step": 304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.022304355402294633,
      "lm_loss": 0.6186,
      "step": 304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.022304355402294633,
      "lm_loss": 0.6125,
      "step": 304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.022304355402294633,
      "lm_loss": 0.4734,
      "step": 304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.022304355402294633,
      "lm_loss": 0.4207,
      "step": 304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.022304355402294633,
      "lm_loss": 0.6521,
      "step": 304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.022304355402294633,
      "lm_loss": 0.4953,
      "step": 304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.022377724992433762,
      "grad_norm": 0.644593037322542,
      "learning_rate": 7.4572127139364316e-06,
      "loss": 0.6141,
      "step": 305
    },
    {
      "epoch": 0.022451094582572887,
      "grad_norm": 0.6571987683216124,
      "learning_rate": 7.481662591687042e-06,
      "loss": 0.6623,
      "step": 306
    },
    {
      "epoch": 0.022524464172712016,
      "grad_norm": 2.10751573713854,
      "learning_rate": 7.506112469437653e-06,
      "loss": 1.0809,
      "step": 307
    },
    {
      "epoch": 0.02259783376285114,
      "grad_norm": 2.1274452048231103,
      "learning_rate": 7.530562347188265e-06,
      "loss": 1.0241,
      "step": 308
    },
    {
      "epoch": 0.02267120335299027,
      "grad_norm": 2.1436546108327907,
      "learning_rate": 7.555012224938876e-06,
      "loss": 1.0363,
      "step": 309
    },
    {
      "epoch": 0.022744572943129395,
      "grad_norm": 1.8051408202859525,
      "learning_rate": 7.579462102689487e-06,
      "loss": 0.8307,
      "step": 310
    },
    {
      "epoch": 0.022817942533268524,
      "grad_norm": 0.7939601643439287,
      "learning_rate": 7.603911980440099e-06,
      "loss": 0.6354,
      "step": 311
    },
    {
      "epoch": 0.02289131212340765,
      "grad_norm": 2.1437056914871824,
      "learning_rate": 7.6283618581907095e-06,
      "loss": 1.0347,
      "step": 312
    },
    {
      "epoch": 0.02289131212340765,
      "lm_loss": 1.8539,
      "step": 312,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.02289131212340765,
      "lm_loss": 0.0681,
      "step": 312,
      "vm_loss": 0.0897
    },
    {
      "epoch": 0.02289131212340765,
      "lm_loss": 1.2408,
      "step": 312,
      "vm_loss": 0.0505
    },
    {
      "epoch": 0.02289131212340765,
      "lm_loss": 0.6984,
      "step": 312,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.02289131212340765,
      "lm_loss": 1.4347,
      "step": 312,
      "vm_loss": 0.1223
    },
    {
      "epoch": 0.02289131212340765,
      "lm_loss": 0.3617,
      "step": 312,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.02289131212340765,
      "lm_loss": 0.2852,
      "step": 312,
      "vm_loss": 0.1464
    },
    {
      "epoch": 0.02289131212340765,
      "lm_loss": 0.3704,
      "step": 312,
      "vm_loss": 0.1339
    },
    {
      "epoch": 0.022964681713546778,
      "grad_norm": 2.2161631381543647,
      "learning_rate": 7.652811735941321e-06,
      "loss": 1.0584,
      "step": 313
    },
    {
      "epoch": 0.023038051303685903,
      "grad_norm": 1.9234278198251875,
      "learning_rate": 7.677261613691932e-06,
      "loss": 0.9429,
      "step": 314
    },
    {
      "epoch": 0.023111420893825032,
      "grad_norm": 2.0404315427358637,
      "learning_rate": 7.701711491442544e-06,
      "loss": 1.0562,
      "step": 315
    },
    {
      "epoch": 0.02318479048396416,
      "grad_norm": 2.03523051107716,
      "learning_rate": 7.726161369193155e-06,
      "loss": 0.9228,
      "step": 316
    },
    {
      "epoch": 0.023258160074103286,
      "grad_norm": 0.6437728153905306,
      "learning_rate": 7.750611246943765e-06,
      "loss": 0.6312,
      "step": 317
    },
    {
      "epoch": 0.023331529664242415,
      "grad_norm": 2.1292403420110175,
      "learning_rate": 7.775061124694378e-06,
      "loss": 0.9757,
      "step": 318
    },
    {
      "epoch": 0.02340489925438154,
      "grad_norm": 2.463997064807774,
      "learning_rate": 7.799511002444988e-06,
      "loss": 1.0761,
      "step": 319
    },
    {
      "epoch": 0.02347826884452067,
      "grad_norm": 2.038492312779783,
      "learning_rate": 7.8239608801956e-06,
      "loss": 1.0121,
      "step": 320
    },
    {
      "epoch": 0.02347826884452067,
      "lm_loss": 1.6328,
      "step": 320,
      "vm_loss": 0.046
    },
    {
      "epoch": 0.02347826884452067,
      "lm_loss": 1.3534,
      "step": 320,
      "vm_loss": 0.1357
    },
    {
      "epoch": 0.02347826884452067,
      "lm_loss": 1.0977,
      "step": 320,
      "vm_loss": 0.1408
    },
    {
      "epoch": 0.02347826884452067,
      "lm_loss": 0.5394,
      "step": 320,
      "vm_loss": 0.1152
    },
    {
      "epoch": 0.02347826884452067,
      "lm_loss": 0.4178,
      "step": 320,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.02347826884452067,
      "lm_loss": 1.2127,
      "step": 320,
      "vm_loss": 0.215
    },
    {
      "epoch": 0.02347826884452067,
      "lm_loss": 0.8281,
      "step": 320,
      "vm_loss": 0.1926
    },
    {
      "epoch": 0.02347826884452067,
      "lm_loss": 1.1731,
      "step": 320,
      "vm_loss": 0.1642
    },
    {
      "epoch": 0.023551638434659794,
      "grad_norm": 2.1786498993611634,
      "learning_rate": 7.848410757946211e-06,
      "loss": 1.0054,
      "step": 321
    },
    {
      "epoch": 0.023625008024798923,
      "grad_norm": 2.2498461280952404,
      "learning_rate": 7.872860635696822e-06,
      "loss": 1.0501,
      "step": 322
    },
    {
      "epoch": 0.023698377614938048,
      "grad_norm": 2.1674378684646034,
      "learning_rate": 7.897310513447434e-06,
      "loss": 1.0314,
      "step": 323
    },
    {
      "epoch": 0.023771747205077177,
      "grad_norm": 2.2575505431590193,
      "learning_rate": 7.921760391198045e-06,
      "loss": 0.9876,
      "step": 324
    },
    {
      "epoch": 0.023845116795216302,
      "grad_norm": 0.6681009082590931,
      "learning_rate": 7.946210268948657e-06,
      "loss": 0.6227,
      "step": 325
    },
    {
      "epoch": 0.02391848638535543,
      "grad_norm": 2.3017173002170765,
      "learning_rate": 7.970660146699268e-06,
      "loss": 0.8817,
      "step": 326
    },
    {
      "epoch": 0.023991855975494556,
      "grad_norm": 2.1934507157289036,
      "learning_rate": 7.995110024449879e-06,
      "loss": 0.9203,
      "step": 327
    },
    {
      "epoch": 0.024065225565633685,
      "grad_norm": 0.6293419782654525,
      "learning_rate": 8.01955990220049e-06,
      "loss": 0.6141,
      "step": 328
    },
    {
      "epoch": 0.024065225565633685,
      "lm_loss": 1.6318,
      "step": 328,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.024065225565633685,
      "lm_loss": 1.1556,
      "step": 328,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.024065225565633685,
      "lm_loss": 1.2825,
      "step": 328,
      "vm_loss": 0.0846
    },
    {
      "epoch": 0.024065225565633685,
      "lm_loss": 0.0627,
      "step": 328,
      "vm_loss": 0.0994
    },
    {
      "epoch": 0.024065225565633685,
      "lm_loss": 1.0455,
      "step": 328,
      "vm_loss": 0.1963
    },
    {
      "epoch": 0.024065225565633685,
      "lm_loss": 0.5009,
      "step": 328,
      "vm_loss": 0.0731
    },
    {
      "epoch": 0.024065225565633685,
      "lm_loss": 0.9166,
      "step": 328,
      "vm_loss": 0.0721
    },
    {
      "epoch": 0.024065225565633685,
      "lm_loss": 0.3222,
      "step": 328,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.02413859515577281,
      "grad_norm": 2.0880164046952783,
      "learning_rate": 8.044009779951102e-06,
      "loss": 0.9364,
      "step": 329
    },
    {
      "epoch": 0.02421196474591194,
      "grad_norm": 2.308294957992222,
      "learning_rate": 8.068459657701712e-06,
      "loss": 0.9619,
      "step": 330
    },
    {
      "epoch": 0.024285334336051064,
      "grad_norm": 1.9754583119210045,
      "learning_rate": 8.092909535452323e-06,
      "loss": 1.0199,
      "step": 331
    },
    {
      "epoch": 0.024358703926190193,
      "grad_norm": 1.9994276550617325,
      "learning_rate": 8.117359413202933e-06,
      "loss": 0.8971,
      "step": 332
    },
    {
      "epoch": 0.024432073516329318,
      "grad_norm": 0.6740196860897879,
      "learning_rate": 8.141809290953546e-06,
      "loss": 0.6298,
      "step": 333
    },
    {
      "epoch": 0.024505443106468447,
      "grad_norm": 0.6358931762655585,
      "learning_rate": 8.166259168704156e-06,
      "loss": 0.6368,
      "step": 334
    },
    {
      "epoch": 0.024578812696607572,
      "grad_norm": 2.278019026721889,
      "learning_rate": 8.190709046454769e-06,
      "loss": 1.0071,
      "step": 335
    },
    {
      "epoch": 0.0246521822867467,
      "grad_norm": 2.024743711634147,
      "learning_rate": 8.21515892420538e-06,
      "loss": 0.8973,
      "step": 336
    },
    {
      "epoch": 0.0246521822867467,
      "lm_loss": 1.6486,
      "step": 336,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.0246521822867467,
      "lm_loss": 0.9315,
      "step": 336,
      "vm_loss": 0.0917
    },
    {
      "epoch": 0.0246521822867467,
      "lm_loss": 0.8846,
      "step": 336,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.0246521822867467,
      "lm_loss": 0.4201,
      "step": 336,
      "vm_loss": 0.0628
    },
    {
      "epoch": 0.0246521822867467,
      "lm_loss": 1.1813,
      "step": 336,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.0246521822867467,
      "lm_loss": 0.3129,
      "step": 336,
      "vm_loss": 0.1981
    },
    {
      "epoch": 0.0246521822867467,
      "lm_loss": 1.1579,
      "step": 336,
      "vm_loss": 0.149
    },
    {
      "epoch": 0.0246521822867467,
      "lm_loss": 0.7145,
      "step": 336,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.024725551876885826,
      "grad_norm": 1.8217891985254313,
      "learning_rate": 8.23960880195599e-06,
      "loss": 0.907,
      "step": 337
    },
    {
      "epoch": 0.024798921467024955,
      "grad_norm": 2.13296039523179,
      "learning_rate": 8.264058679706602e-06,
      "loss": 1.1064,
      "step": 338
    },
    {
      "epoch": 0.02487229105716408,
      "grad_norm": 0.6366622845091627,
      "learning_rate": 8.288508557457213e-06,
      "loss": 0.6523,
      "step": 339
    },
    {
      "epoch": 0.02494566064730321,
      "grad_norm": 0.644233328934219,
      "learning_rate": 8.312958435207825e-06,
      "loss": 0.626,
      "step": 340
    },
    {
      "epoch": 0.025019030237442338,
      "grad_norm": 0.6506924868204661,
      "learning_rate": 8.337408312958436e-06,
      "loss": 0.6498,
      "step": 341
    },
    {
      "epoch": 0.025092399827581463,
      "grad_norm": 2.216980911327621,
      "learning_rate": 8.361858190709047e-06,
      "loss": 1.0072,
      "step": 342
    },
    {
      "epoch": 0.025165769417720592,
      "grad_norm": 0.6406786942462592,
      "learning_rate": 8.386308068459659e-06,
      "loss": 0.6307,
      "step": 343
    },
    {
      "epoch": 0.025239139007859717,
      "grad_norm": 2.115200555760761,
      "learning_rate": 8.41075794621027e-06,
      "loss": 1.0663,
      "step": 344
    },
    {
      "epoch": 0.025239139007859717,
      "lm_loss": 1.4528,
      "step": 344,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.025239139007859717,
      "lm_loss": 0.8244,
      "step": 344,
      "vm_loss": 0.0774
    },
    {
      "epoch": 0.025239139007859717,
      "lm_loss": 0.5442,
      "step": 344,
      "vm_loss": 0.1816
    },
    {
      "epoch": 0.025239139007859717,
      "lm_loss": 0.5551,
      "step": 344,
      "vm_loss": 0.1012
    },
    {
      "epoch": 0.025239139007859717,
      "lm_loss": 0.7513,
      "step": 344,
      "vm_loss": 0.1481
    },
    {
      "epoch": 0.025239139007859717,
      "lm_loss": 0.5342,
      "step": 344,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.025239139007859717,
      "lm_loss": 1.0299,
      "step": 344,
      "vm_loss": 0.1737
    },
    {
      "epoch": 0.025239139007859717,
      "lm_loss": 0.6067,
      "step": 344,
      "vm_loss": 0.0734
    },
    {
      "epoch": 0.025312508597998846,
      "grad_norm": 1.8760666240825679,
      "learning_rate": 8.43520782396088e-06,
      "loss": 1.1201,
      "step": 345
    },
    {
      "epoch": 0.02538587818813797,
      "grad_norm": 1.9430307540343792,
      "learning_rate": 8.459657701711493e-06,
      "loss": 0.9835,
      "step": 346
    },
    {
      "epoch": 0.0254592477782771,
      "grad_norm": 2.0619570764540613,
      "learning_rate": 8.484107579462103e-06,
      "loss": 1.0369,
      "step": 347
    },
    {
      "epoch": 0.025532617368416225,
      "grad_norm": 1.9930681889567075,
      "learning_rate": 8.508557457212716e-06,
      "loss": 1.0093,
      "step": 348
    },
    {
      "epoch": 0.025605986958555354,
      "grad_norm": 0.8519193349549523,
      "learning_rate": 8.533007334963326e-06,
      "loss": 0.652,
      "step": 349
    },
    {
      "epoch": 0.02567935654869448,
      "grad_norm": 1.9728813536588286,
      "learning_rate": 8.557457212713937e-06,
      "loss": 0.9345,
      "step": 350
    },
    {
      "epoch": 0.025752726138833608,
      "grad_norm": 1.9142804563150249,
      "learning_rate": 8.58190709046455e-06,
      "loss": 1.0348,
      "step": 351
    },
    {
      "epoch": 0.025826095728972733,
      "grad_norm": 2.2321889768543635,
      "learning_rate": 8.60635696821516e-06,
      "loss": 1.0287,
      "step": 352
    },
    {
      "epoch": 0.025826095728972733,
      "lm_loss": 0.6993,
      "step": 352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.025826095728972733,
      "lm_loss": 0.8687,
      "step": 352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.025826095728972733,
      "lm_loss": 0.6739,
      "step": 352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.025826095728972733,
      "lm_loss": 0.4604,
      "step": 352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.025826095728972733,
      "lm_loss": 0.6658,
      "step": 352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.025826095728972733,
      "lm_loss": 0.5991,
      "step": 352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.025826095728972733,
      "lm_loss": 0.7152,
      "step": 352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.025826095728972733,
      "lm_loss": 0.6908,
      "step": 352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.025899465319111862,
      "grad_norm": 0.6793782810787005,
      "learning_rate": 8.63080684596577e-06,
      "loss": 0.662,
      "step": 353
    },
    {
      "epoch": 0.025972834909250987,
      "grad_norm": 2.2633225228350535,
      "learning_rate": 8.655256723716381e-06,
      "loss": 1.0448,
      "step": 354
    },
    {
      "epoch": 0.026046204499390116,
      "grad_norm": 1.9547851665730718,
      "learning_rate": 8.679706601466994e-06,
      "loss": 0.9252,
      "step": 355
    },
    {
      "epoch": 0.02611957408952924,
      "grad_norm": 1.9277979156732348,
      "learning_rate": 8.704156479217604e-06,
      "loss": 1.0096,
      "step": 356
    },
    {
      "epoch": 0.02619294367966837,
      "grad_norm": 1.9203354538351038,
      "learning_rate": 8.728606356968215e-06,
      "loss": 0.9129,
      "step": 357
    },
    {
      "epoch": 0.026266313269807495,
      "grad_norm": 1.960737237888475,
      "learning_rate": 8.753056234718827e-06,
      "loss": 0.9454,
      "step": 358
    },
    {
      "epoch": 0.026339682859946624,
      "grad_norm": 0.682861922634135,
      "learning_rate": 8.777506112469438e-06,
      "loss": 0.6073,
      "step": 359
    },
    {
      "epoch": 0.02641305245008575,
      "grad_norm": 2.293867271793944,
      "learning_rate": 8.801955990220049e-06,
      "loss": 1.1104,
      "step": 360
    },
    {
      "epoch": 0.02641305245008575,
      "lm_loss": 1.3395,
      "step": 360,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.02641305245008575,
      "lm_loss": 1.7026,
      "step": 360,
      "vm_loss": 0.1448
    },
    {
      "epoch": 0.02641305245008575,
      "lm_loss": 1.3527,
      "step": 360,
      "vm_loss": 0.1676
    },
    {
      "epoch": 0.02641305245008575,
      "lm_loss": 1.0158,
      "step": 360,
      "vm_loss": 0.1788
    },
    {
      "epoch": 0.02641305245008575,
      "lm_loss": 0.8999,
      "step": 360,
      "vm_loss": 0.1442
    },
    {
      "epoch": 0.02641305245008575,
      "lm_loss": 1.324,
      "step": 360,
      "vm_loss": 0.1434
    },
    {
      "epoch": 0.02641305245008575,
      "lm_loss": 1.4598,
      "step": 360,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.02641305245008575,
      "lm_loss": 0.7926,
      "step": 360,
      "vm_loss": 0.1002
    },
    {
      "epoch": 0.026486422040224878,
      "grad_norm": 1.9091231740438406,
      "learning_rate": 8.826405867970661e-06,
      "loss": 1.036,
      "step": 361
    },
    {
      "epoch": 0.026559791630364003,
      "grad_norm": 0.6855497712165789,
      "learning_rate": 8.850855745721272e-06,
      "loss": 0.6501,
      "step": 362
    },
    {
      "epoch": 0.026633161220503132,
      "grad_norm": 2.2856474336105443,
      "learning_rate": 8.875305623471884e-06,
      "loss": 0.9278,
      "step": 363
    },
    {
      "epoch": 0.026706530810642257,
      "grad_norm": 2.468400472678478,
      "learning_rate": 8.899755501222495e-06,
      "loss": 0.9341,
      "step": 364
    },
    {
      "epoch": 0.026779900400781386,
      "grad_norm": 0.6359099961759844,
      "learning_rate": 8.924205378973105e-06,
      "loss": 0.6398,
      "step": 365
    },
    {
      "epoch": 0.026853269990920515,
      "grad_norm": 2.562305283772661,
      "learning_rate": 8.948655256723718e-06,
      "loss": 1.0935,
      "step": 366
    },
    {
      "epoch": 0.02692663958105964,
      "grad_norm": 2.1742455189250567,
      "learning_rate": 8.973105134474328e-06,
      "loss": 0.9445,
      "step": 367
    },
    {
      "epoch": 0.02700000917119877,
      "grad_norm": 1.8944497998532708,
      "learning_rate": 8.99755501222494e-06,
      "loss": 0.9249,
      "step": 368
    },
    {
      "epoch": 0.02700000917119877,
      "lm_loss": 1.6922,
      "step": 368,
      "vm_loss": 0.0541
    },
    {
      "epoch": 0.02700000917119877,
      "lm_loss": 0.2366,
      "step": 368,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.02700000917119877,
      "lm_loss": 0.3707,
      "step": 368,
      "vm_loss": 0.2072
    },
    {
      "epoch": 0.02700000917119877,
      "lm_loss": 1.13,
      "step": 368,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.02700000917119877,
      "lm_loss": 0.8483,
      "step": 368,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.02700000917119877,
      "lm_loss": 0.7675,
      "step": 368,
      "vm_loss": 0.1845
    },
    {
      "epoch": 0.02700000917119877,
      "lm_loss": 0.8184,
      "step": 368,
      "vm_loss": 0.1274
    },
    {
      "epoch": 0.02700000917119877,
      "lm_loss": 0.4992,
      "step": 368,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.027073378761337894,
      "grad_norm": 1.920495811461622,
      "learning_rate": 9.022004889975551e-06,
      "loss": 0.9723,
      "step": 369
    },
    {
      "epoch": 0.027146748351477023,
      "grad_norm": 0.7305733509758611,
      "learning_rate": 9.046454767726162e-06,
      "loss": 0.6188,
      "step": 370
    },
    {
      "epoch": 0.027220117941616148,
      "grad_norm": 1.9771875753694346,
      "learning_rate": 9.070904645476774e-06,
      "loss": 0.9419,
      "step": 371
    },
    {
      "epoch": 0.027293487531755277,
      "grad_norm": 1.9207288391813506,
      "learning_rate": 9.095354523227385e-06,
      "loss": 1.0535,
      "step": 372
    },
    {
      "epoch": 0.027366857121894402,
      "grad_norm": 2.0793347257404453,
      "learning_rate": 9.119804400977995e-06,
      "loss": 1.0514,
      "step": 373
    },
    {
      "epoch": 0.02744022671203353,
      "grad_norm": 1.8885999677434873,
      "learning_rate": 9.144254278728608e-06,
      "loss": 0.969,
      "step": 374
    },
    {
      "epoch": 0.027513596302172656,
      "grad_norm": 2.0214614699344002,
      "learning_rate": 9.168704156479218e-06,
      "loss": 0.9185,
      "step": 375
    },
    {
      "epoch": 0.027586965892311785,
      "grad_norm": 2.102250353988439,
      "learning_rate": 9.193154034229829e-06,
      "loss": 1.0095,
      "step": 376
    },
    {
      "epoch": 0.027586965892311785,
      "lm_loss": 1.3468,
      "step": 376,
      "vm_loss": 0.0483
    },
    {
      "epoch": 0.027586965892311785,
      "lm_loss": 0.2013,
      "step": 376,
      "vm_loss": 0.0342
    },
    {
      "epoch": 0.027586965892311785,
      "lm_loss": 0.9019,
      "step": 376,
      "vm_loss": 0.231
    },
    {
      "epoch": 0.027586965892311785,
      "lm_loss": 0.5913,
      "step": 376,
      "vm_loss": 0.1943
    },
    {
      "epoch": 0.027586965892311785,
      "lm_loss": 0.7055,
      "step": 376,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.027586965892311785,
      "lm_loss": 1.1418,
      "step": 376,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.027586965892311785,
      "lm_loss": 1.2745,
      "step": 376,
      "vm_loss": 0.1482
    },
    {
      "epoch": 0.027586965892311785,
      "lm_loss": 1.2071,
      "step": 376,
      "vm_loss": 0.1416
    },
    {
      "epoch": 0.02766033548245091,
      "grad_norm": 2.141731601017551,
      "learning_rate": 9.21760391198044e-06,
      "loss": 0.9622,
      "step": 377
    },
    {
      "epoch": 0.02773370507259004,
      "grad_norm": 2.092495498688828,
      "learning_rate": 9.242053789731052e-06,
      "loss": 0.9705,
      "step": 378
    },
    {
      "epoch": 0.027807074662729164,
      "grad_norm": 1.8997720636640805,
      "learning_rate": 9.266503667481663e-06,
      "loss": 0.9196,
      "step": 379
    },
    {
      "epoch": 0.027880444252868293,
      "grad_norm": 2.061792294005077,
      "learning_rate": 9.290953545232273e-06,
      "loss": 0.987,
      "step": 380
    },
    {
      "epoch": 0.02795381384300742,
      "grad_norm": 1.828938680444968,
      "learning_rate": 9.315403422982886e-06,
      "loss": 1.0431,
      "step": 381
    },
    {
      "epoch": 0.028027183433146547,
      "grad_norm": 0.8065614503311613,
      "learning_rate": 9.339853300733496e-06,
      "loss": 0.6324,
      "step": 382
    },
    {
      "epoch": 0.028100553023285672,
      "grad_norm": 0.6827510448666071,
      "learning_rate": 9.364303178484109e-06,
      "loss": 0.6401,
      "step": 383
    },
    {
      "epoch": 0.0281739226134248,
      "grad_norm": 2.2832643089041422,
      "learning_rate": 9.38875305623472e-06,
      "loss": 0.9705,
      "step": 384
    },
    {
      "epoch": 0.0281739226134248,
      "lm_loss": 0.9044,
      "step": 384,
      "vm_loss": 0.0785
    },
    {
      "epoch": 0.0281739226134248,
      "lm_loss": 0.3104,
      "step": 384,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.0281739226134248,
      "lm_loss": 1.0725,
      "step": 384,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.0281739226134248,
      "lm_loss": 0.7925,
      "step": 384,
      "vm_loss": 0.2512
    },
    {
      "epoch": 0.0281739226134248,
      "lm_loss": 1.2549,
      "step": 384,
      "vm_loss": 0.1344
    },
    {
      "epoch": 0.0281739226134248,
      "lm_loss": 0.7789,
      "step": 384,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.0281739226134248,
      "lm_loss": 1.4147,
      "step": 384,
      "vm_loss": 0.1237
    },
    {
      "epoch": 0.0281739226134248,
      "lm_loss": 1.339,
      "step": 384,
      "vm_loss": 0.0584
    },
    {
      "epoch": 0.028247292203563926,
      "grad_norm": 2.117619268809174,
      "learning_rate": 9.41320293398533e-06,
      "loss": 0.8936,
      "step": 385
    },
    {
      "epoch": 0.028320661793703055,
      "grad_norm": 2.047934625749102,
      "learning_rate": 9.437652811735942e-06,
      "loss": 1.0144,
      "step": 386
    },
    {
      "epoch": 0.02839403138384218,
      "grad_norm": 2.1215542100729214,
      "learning_rate": 9.462102689486553e-06,
      "loss": 1.0054,
      "step": 387
    },
    {
      "epoch": 0.02846740097398131,
      "grad_norm": 1.8680401579257064,
      "learning_rate": 9.486552567237165e-06,
      "loss": 0.9722,
      "step": 388
    },
    {
      "epoch": 0.028540770564120438,
      "grad_norm": 0.9155906604653334,
      "learning_rate": 9.511002444987776e-06,
      "loss": 0.6406,
      "step": 389
    },
    {
      "epoch": 0.028614140154259563,
      "grad_norm": 2.151548585222165,
      "learning_rate": 9.535452322738387e-06,
      "loss": 1.0867,
      "step": 390
    },
    {
      "epoch": 0.028687509744398692,
      "grad_norm": 2.1285405744407733,
      "learning_rate": 9.559902200488999e-06,
      "loss": 1.0317,
      "step": 391
    },
    {
      "epoch": 0.028760879334537817,
      "grad_norm": 0.709616300533557,
      "learning_rate": 9.58435207823961e-06,
      "loss": 0.6289,
      "step": 392
    },
    {
      "epoch": 0.028760879334537817,
      "lm_loss": 1.4176,
      "step": 392,
      "vm_loss": 0.1275
    },
    {
      "epoch": 0.028760879334537817,
      "lm_loss": 2.3937,
      "step": 392,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.028760879334537817,
      "lm_loss": 1.3515,
      "step": 392,
      "vm_loss": 0.0825
    },
    {
      "epoch": 0.028760879334537817,
      "lm_loss": 1.4589,
      "step": 392,
      "vm_loss": 0.2239
    },
    {
      "epoch": 0.028760879334537817,
      "lm_loss": 1.1073,
      "step": 392,
      "vm_loss": 0.1608
    },
    {
      "epoch": 0.028760879334537817,
      "lm_loss": 0.5704,
      "step": 392,
      "vm_loss": 0.1814
    },
    {
      "epoch": 0.028760879334537817,
      "lm_loss": 0.239,
      "step": 392,
      "vm_loss": 0.0622
    },
    {
      "epoch": 0.028760879334537817,
      "lm_loss": 0.8684,
      "step": 392,
      "vm_loss": 0.1261
    },
    {
      "epoch": 0.028834248924676946,
      "grad_norm": 2.233742509727273,
      "learning_rate": 9.608801955990222e-06,
      "loss": 0.9695,
      "step": 393
    },
    {
      "epoch": 0.02890761851481607,
      "grad_norm": 0.5975848075208674,
      "learning_rate": 9.633251833740833e-06,
      "loss": 0.6207,
      "step": 394
    },
    {
      "epoch": 0.0289809881049552,
      "grad_norm": 2.124863387799213,
      "learning_rate": 9.657701711491443e-06,
      "loss": 1.0156,
      "step": 395
    },
    {
      "epoch": 0.029054357695094325,
      "grad_norm": 2.1674242452706274,
      "learning_rate": 9.682151589242056e-06,
      "loss": 0.9796,
      "step": 396
    },
    {
      "epoch": 0.029127727285233454,
      "grad_norm": 0.7398004054962883,
      "learning_rate": 9.706601466992666e-06,
      "loss": 0.6394,
      "step": 397
    },
    {
      "epoch": 0.02920109687537258,
      "grad_norm": 2.8495111824348736,
      "learning_rate": 9.731051344743277e-06,
      "loss": 1.1458,
      "step": 398
    },
    {
      "epoch": 0.029274466465511708,
      "grad_norm": 1.8959984847851596,
      "learning_rate": 9.755501222493888e-06,
      "loss": 0.9441,
      "step": 399
    },
    {
      "epoch": 0.029347836055650833,
      "grad_norm": 0.629773962208602,
      "learning_rate": 9.7799511002445e-06,
      "loss": 0.5802,
      "step": 400
    },
    {
      "epoch": 0.029347836055650833,
      "lm_loss": 1.2852,
      "step": 400,
      "vm_loss": 0.1472
    },
    {
      "epoch": 0.029347836055650833,
      "lm_loss": 1.0404,
      "step": 400,
      "vm_loss": 0.0681
    },
    {
      "epoch": 0.029347836055650833,
      "lm_loss": 0.1754,
      "step": 400,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.029347836055650833,
      "lm_loss": 1.4408,
      "step": 400,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.029347836055650833,
      "lm_loss": 1.2924,
      "step": 400,
      "vm_loss": 0.1742
    },
    {
      "epoch": 0.029347836055650833,
      "lm_loss": 0.5244,
      "step": 400,
      "vm_loss": 0.1146
    },
    {
      "epoch": 0.029347836055650833,
      "lm_loss": 0.2983,
      "step": 400,
      "vm_loss": 0.1727
    },
    {
      "epoch": 0.029347836055650833,
      "lm_loss": 0.3837,
      "step": 400,
      "vm_loss": 0.1423
    },
    {
      "epoch": 0.029421205645789962,
      "grad_norm": 2.2970270827820527,
      "learning_rate": 9.80440097799511e-06,
      "loss": 0.9383,
      "step": 401
    },
    {
      "epoch": 0.029494575235929087,
      "grad_norm": 2.1890153453696586,
      "learning_rate": 9.828850855745721e-06,
      "loss": 1.0467,
      "step": 402
    },
    {
      "epoch": 0.029567944826068216,
      "grad_norm": 1.879791189911424,
      "learning_rate": 9.853300733496334e-06,
      "loss": 0.9424,
      "step": 403
    },
    {
      "epoch": 0.02964131441620734,
      "grad_norm": 2.432264599324967,
      "learning_rate": 9.877750611246944e-06,
      "loss": 1.066,
      "step": 404
    },
    {
      "epoch": 0.02971468400634647,
      "grad_norm": 1.9748499850108714,
      "learning_rate": 9.902200488997555e-06,
      "loss": 0.8356,
      "step": 405
    },
    {
      "epoch": 0.029788053596485595,
      "grad_norm": 1.9791179981209308,
      "learning_rate": 9.926650366748167e-06,
      "loss": 1.0388,
      "step": 406
    },
    {
      "epoch": 0.029861423186624724,
      "grad_norm": 2.3733254760513285,
      "learning_rate": 9.951100244498778e-06,
      "loss": 0.9967,
      "step": 407
    },
    {
      "epoch": 0.02993479277676385,
      "grad_norm": 2.1448774302388784,
      "learning_rate": 9.97555012224939e-06,
      "loss": 1.104,
      "step": 408
    },
    {
      "epoch": 0.02993479277676385,
      "lm_loss": 0.7425,
      "step": 408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.02993479277676385,
      "lm_loss": 0.5803,
      "step": 408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.02993479277676385,
      "lm_loss": 0.8277,
      "step": 408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.02993479277676385,
      "lm_loss": 0.7863,
      "step": 408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.02993479277676385,
      "lm_loss": 0.8118,
      "step": 408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.02993479277676385,
      "lm_loss": 0.6251,
      "step": 408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.02993479277676385,
      "lm_loss": 0.6536,
      "step": 408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.02993479277676385,
      "lm_loss": 0.6781,
      "step": 408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.030008162366902978,
      "grad_norm": 0.8955579934680322,
      "learning_rate": 1e-05,
      "loss": 0.6518,
      "step": 409
    },
    {
      "epoch": 0.030081531957042103,
      "grad_norm": 2.1880955531167823,
      "learning_rate": 9.999999858818809e-06,
      "loss": 1.0065,
      "step": 410
    },
    {
      "epoch": 0.030154901547181232,
      "grad_norm": 1.8423406714897113,
      "learning_rate": 9.999999435275246e-06,
      "loss": 1.014,
      "step": 411
    },
    {
      "epoch": 0.030228271137320357,
      "grad_norm": 2.17116397400663,
      "learning_rate": 9.999998729369329e-06,
      "loss": 0.9557,
      "step": 412
    },
    {
      "epoch": 0.030301640727459486,
      "grad_norm": 1.829415669483918,
      "learning_rate": 9.999997741101102e-06,
      "loss": 0.8953,
      "step": 413
    },
    {
      "epoch": 0.030375010317598615,
      "grad_norm": 0.7513681338202698,
      "learning_rate": 9.999996470470622e-06,
      "loss": 0.6304,
      "step": 414
    },
    {
      "epoch": 0.03044837990773774,
      "grad_norm": 2.0585895867834103,
      "learning_rate": 9.999994917477956e-06,
      "loss": 0.979,
      "step": 415
    },
    {
      "epoch": 0.03052174949787687,
      "grad_norm": 2.2254125697308114,
      "learning_rate": 9.999993082123197e-06,
      "loss": 1.1043,
      "step": 416
    },
    {
      "epoch": 0.03052174949787687,
      "lm_loss": 0.5849,
      "step": 416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03052174949787687,
      "lm_loss": 0.6722,
      "step": 416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03052174949787687,
      "lm_loss": 0.4544,
      "step": 416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03052174949787687,
      "lm_loss": 0.6361,
      "step": 416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03052174949787687,
      "lm_loss": 0.5838,
      "step": 416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03052174949787687,
      "lm_loss": 0.4895,
      "step": 416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03052174949787687,
      "lm_loss": 0.6852,
      "step": 416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03052174949787687,
      "lm_loss": 0.6973,
      "step": 416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.030595119088015994,
      "grad_norm": 0.7238716711337465,
      "learning_rate": 9.999990964406447e-06,
      "loss": 0.6214,
      "step": 417
    },
    {
      "epoch": 0.030668488678155123,
      "grad_norm": 2.095406129036888,
      "learning_rate": 9.999988564327823e-06,
      "loss": 0.982,
      "step": 418
    },
    {
      "epoch": 0.03074185826829425,
      "grad_norm": 2.145246172967302,
      "learning_rate": 9.999985881887463e-06,
      "loss": 1.0043,
      "step": 419
    },
    {
      "epoch": 0.030815227858433377,
      "grad_norm": 0.7105021796395189,
      "learning_rate": 9.99998291708552e-06,
      "loss": 0.6205,
      "step": 420
    },
    {
      "epoch": 0.030888597448572502,
      "grad_norm": 2.1874102073829134,
      "learning_rate": 9.999979669922158e-06,
      "loss": 1.0146,
      "step": 421
    },
    {
      "epoch": 0.03096196703871163,
      "grad_norm": 2.004428033962235,
      "learning_rate": 9.999976140397561e-06,
      "loss": 1.024,
      "step": 422
    },
    {
      "epoch": 0.031035336628850756,
      "grad_norm": 2.2944449784696843,
      "learning_rate": 9.99997232851193e-06,
      "loss": 0.9288,
      "step": 423
    },
    {
      "epoch": 0.031108706218989885,
      "grad_norm": 1.9562303837488997,
      "learning_rate": 9.99996823426548e-06,
      "loss": 1.0337,
      "step": 424
    },
    {
      "epoch": 0.031108706218989885,
      "lm_loss": 1.4681,
      "step": 424,
      "vm_loss": 0.0474
    },
    {
      "epoch": 0.031108706218989885,
      "lm_loss": 1.3877,
      "step": 424,
      "vm_loss": 0.1941
    },
    {
      "epoch": 0.031108706218989885,
      "lm_loss": 0.276,
      "step": 424,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.031108706218989885,
      "lm_loss": 0.621,
      "step": 424,
      "vm_loss": 0.2521
    },
    {
      "epoch": 0.031108706218989885,
      "lm_loss": 0.7302,
      "step": 424,
      "vm_loss": 0.1543
    },
    {
      "epoch": 0.031108706218989885,
      "lm_loss": 0.3042,
      "step": 424,
      "vm_loss": 0.1681
    },
    {
      "epoch": 0.031108706218989885,
      "lm_loss": 0.3341,
      "step": 424,
      "vm_loss": 0.1589
    },
    {
      "epoch": 0.031108706218989885,
      "lm_loss": 0.401,
      "step": 424,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.03118207580912901,
      "grad_norm": 2.153476515150368,
      "learning_rate": 9.99996385765844e-06,
      "loss": 1.0286,
      "step": 425
    },
    {
      "epoch": 0.03125544539926814,
      "grad_norm": 2.029865115867883,
      "learning_rate": 9.99995919869106e-06,
      "loss": 1.0545,
      "step": 426
    },
    {
      "epoch": 0.03132881498940727,
      "grad_norm": 1.9871636457857083,
      "learning_rate": 9.9999542573636e-06,
      "loss": 0.8887,
      "step": 427
    },
    {
      "epoch": 0.03140218457954639,
      "grad_norm": 1.8947582626762238,
      "learning_rate": 9.999949033676344e-06,
      "loss": 0.9619,
      "step": 428
    },
    {
      "epoch": 0.03147555416968552,
      "grad_norm": 2.026741770342214,
      "learning_rate": 9.99994352762958e-06,
      "loss": 1.1044,
      "step": 429
    },
    {
      "epoch": 0.03154892375982465,
      "grad_norm": 0.8999403613414854,
      "learning_rate": 9.999937739223626e-06,
      "loss": 0.6214,
      "step": 430
    },
    {
      "epoch": 0.031622293349963776,
      "grad_norm": 2.0780043572539917,
      "learning_rate": 9.999931668458805e-06,
      "loss": 1.101,
      "step": 431
    },
    {
      "epoch": 0.0316956629401029,
      "grad_norm": 2.216866988929998,
      "learning_rate": 9.99992531533546e-06,
      "loss": 1.0984,
      "step": 432
    },
    {
      "epoch": 0.0316956629401029,
      "lm_loss": 1.4107,
      "step": 432,
      "vm_loss": 0.0518
    },
    {
      "epoch": 0.0316956629401029,
      "lm_loss": 0.5524,
      "step": 432,
      "vm_loss": 0.0967
    },
    {
      "epoch": 0.0316956629401029,
      "lm_loss": 0.5437,
      "step": 432,
      "vm_loss": 0.1775
    },
    {
      "epoch": 0.0316956629401029,
      "lm_loss": 0.8118,
      "step": 432,
      "vm_loss": 0.1757
    },
    {
      "epoch": 0.0316956629401029,
      "lm_loss": 0.6173,
      "step": 432,
      "vm_loss": 0.1914
    },
    {
      "epoch": 0.0316956629401029,
      "lm_loss": 1.4628,
      "step": 432,
      "vm_loss": 0.2977
    },
    {
      "epoch": 0.0316956629401029,
      "lm_loss": 0.508,
      "step": 432,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.0316956629401029,
      "lm_loss": 0.6291,
      "step": 432,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.031769032530242027,
      "grad_norm": 2.1825030219836483,
      "learning_rate": 9.999918679853951e-06,
      "loss": 1.0207,
      "step": 433
    },
    {
      "epoch": 0.031842402120381155,
      "grad_norm": 1.9490273888367935,
      "learning_rate": 9.999911762014652e-06,
      "loss": 1.0554,
      "step": 434
    },
    {
      "epoch": 0.031915771710520284,
      "grad_norm": 1.934778415006373,
      "learning_rate": 9.999904561817953e-06,
      "loss": 0.9844,
      "step": 435
    },
    {
      "epoch": 0.031989141300659406,
      "grad_norm": 0.8411213847947785,
      "learning_rate": 9.999897079264263e-06,
      "loss": 0.6245,
      "step": 436
    },
    {
      "epoch": 0.032062510890798535,
      "grad_norm": 0.771067520008896,
      "learning_rate": 9.999889314354e-06,
      "loss": 0.6229,
      "step": 437
    },
    {
      "epoch": 0.03213588048093766,
      "grad_norm": 2.1238563100258405,
      "learning_rate": 9.999881267087608e-06,
      "loss": 1.0235,
      "step": 438
    },
    {
      "epoch": 0.03220925007107679,
      "grad_norm": 0.6631517676897026,
      "learning_rate": 9.999872937465538e-06,
      "loss": 0.615,
      "step": 439
    },
    {
      "epoch": 0.03228261966121592,
      "grad_norm": 2.2046099272554316,
      "learning_rate": 9.999864325488262e-06,
      "loss": 1.0106,
      "step": 440
    },
    {
      "epoch": 0.03228261966121592,
      "lm_loss": 1.4915,
      "step": 440,
      "vm_loss": 0.0408
    },
    {
      "epoch": 0.03228261966121592,
      "lm_loss": 1.386,
      "step": 440,
      "vm_loss": 0.1189
    },
    {
      "epoch": 0.03228261966121592,
      "lm_loss": 1.348,
      "step": 440,
      "vm_loss": 0.0724
    },
    {
      "epoch": 0.03228261966121592,
      "lm_loss": 0.7771,
      "step": 440,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.03228261966121592,
      "lm_loss": 0.5267,
      "step": 440,
      "vm_loss": 0.0659
    },
    {
      "epoch": 0.03228261966121592,
      "lm_loss": 1.2141,
      "step": 440,
      "vm_loss": 0.1666
    },
    {
      "epoch": 0.03228261966121592,
      "lm_loss": 1.3644,
      "step": 440,
      "vm_loss": 0.2558
    },
    {
      "epoch": 0.03228261966121592,
      "lm_loss": 1.0088,
      "step": 440,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.03235598925135504,
      "grad_norm": 1.938581754068523,
      "learning_rate": 9.999855431156264e-06,
      "loss": 0.9257,
      "step": 441
    },
    {
      "epoch": 0.03242935884149417,
      "grad_norm": 1.9765908687499723,
      "learning_rate": 9.999846254470049e-06,
      "loss": 0.8473,
      "step": 442
    },
    {
      "epoch": 0.0325027284316333,
      "grad_norm": 1.0204405024037668,
      "learning_rate": 9.999836795430134e-06,
      "loss": 0.6029,
      "step": 443
    },
    {
      "epoch": 0.03257609802177243,
      "grad_norm": 2.124255364077735,
      "learning_rate": 9.999827054037052e-06,
      "loss": 1.0345,
      "step": 444
    },
    {
      "epoch": 0.03264946761191155,
      "grad_norm": 0.6980189862837639,
      "learning_rate": 9.999817030291357e-06,
      "loss": 0.6141,
      "step": 445
    },
    {
      "epoch": 0.03272283720205068,
      "grad_norm": 2.275971551788646,
      "learning_rate": 9.99980672419361e-06,
      "loss": 1.0777,
      "step": 446
    },
    {
      "epoch": 0.03279620679218981,
      "grad_norm": 2.2949049672695865,
      "learning_rate": 9.999796135744397e-06,
      "loss": 1.111,
      "step": 447
    },
    {
      "epoch": 0.03286957638232894,
      "grad_norm": 2.368677521043113,
      "learning_rate": 9.999785264944312e-06,
      "loss": 1.0434,
      "step": 448
    },
    {
      "epoch": 0.03286957638232894,
      "lm_loss": 1.4751,
      "step": 448,
      "vm_loss": 0.0536
    },
    {
      "epoch": 0.03286957638232894,
      "lm_loss": 0.6563,
      "step": 448,
      "vm_loss": 0.1217
    },
    {
      "epoch": 0.03286957638232894,
      "lm_loss": 0.7742,
      "step": 448,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.03286957638232894,
      "lm_loss": 0.6835,
      "step": 448,
      "vm_loss": 0.1769
    },
    {
      "epoch": 0.03286957638232894,
      "lm_loss": 1.3018,
      "step": 448,
      "vm_loss": 0.07
    },
    {
      "epoch": 0.03286957638232894,
      "lm_loss": 0.8694,
      "step": 448,
      "vm_loss": 0.1716
    },
    {
      "epoch": 0.03286957638232894,
      "lm_loss": 1.472,
      "step": 448,
      "vm_loss": 0.1747
    },
    {
      "epoch": 0.03286957638232894,
      "lm_loss": 0.7924,
      "step": 448,
      "vm_loss": 0.2051
    },
    {
      "epoch": 0.03294294597246806,
      "grad_norm": 1.9808401896308585,
      "learning_rate": 9.999774111793974e-06,
      "loss": 0.9749,
      "step": 449
    },
    {
      "epoch": 0.03301631556260719,
      "grad_norm": 1.9909640014559906,
      "learning_rate": 9.99976267629401e-06,
      "loss": 0.9662,
      "step": 450
    },
    {
      "epoch": 0.033089685152746316,
      "grad_norm": 2.036755472318875,
      "learning_rate": 9.999750958445065e-06,
      "loss": 0.8933,
      "step": 451
    },
    {
      "epoch": 0.033163054742885445,
      "grad_norm": 2.0850172458951834,
      "learning_rate": 9.999738958247801e-06,
      "loss": 0.9696,
      "step": 452
    },
    {
      "epoch": 0.03323642433302457,
      "grad_norm": 2.121250897832326,
      "learning_rate": 9.999726675702898e-06,
      "loss": 0.9467,
      "step": 453
    },
    {
      "epoch": 0.033309793923163696,
      "grad_norm": 2.335876853302419,
      "learning_rate": 9.999714110811047e-06,
      "loss": 1.0202,
      "step": 454
    },
    {
      "epoch": 0.033383163513302824,
      "grad_norm": 1.1993025024710906,
      "learning_rate": 9.999701263572961e-06,
      "loss": 0.6303,
      "step": 455
    },
    {
      "epoch": 0.03345653310344195,
      "grad_norm": 0.9765916355703964,
      "learning_rate": 9.999688133989361e-06,
      "loss": 0.6095,
      "step": 456
    },
    {
      "epoch": 0.03345653310344195,
      "lm_loss": 0.3524,
      "step": 456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03345653310344195,
      "lm_loss": 0.6915,
      "step": 456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03345653310344195,
      "lm_loss": 0.5353,
      "step": 456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03345653310344195,
      "lm_loss": 0.5333,
      "step": 456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03345653310344195,
      "lm_loss": 0.5538,
      "step": 456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03345653310344195,
      "lm_loss": 0.6746,
      "step": 456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03345653310344195,
      "lm_loss": 0.5887,
      "step": 456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03345653310344195,
      "lm_loss": 0.7877,
      "step": 456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.033529902693581075,
      "grad_norm": 0.6716362045887199,
      "learning_rate": 9.99967472206099e-06,
      "loss": 0.6158,
      "step": 457
    },
    {
      "epoch": 0.033603272283720204,
      "grad_norm": 0.6801565559851107,
      "learning_rate": 9.999661027788607e-06,
      "loss": 0.617,
      "step": 458
    },
    {
      "epoch": 0.03367664187385933,
      "grad_norm": 2.789656603933121,
      "learning_rate": 9.999647051172982e-06,
      "loss": 0.9949,
      "step": 459
    },
    {
      "epoch": 0.03375001146399846,
      "grad_norm": 2.3399936889897797,
      "learning_rate": 9.999632792214909e-06,
      "loss": 1.0081,
      "step": 460
    },
    {
      "epoch": 0.03382338105413758,
      "grad_norm": 1.4516623467796068,
      "learning_rate": 9.99961825091519e-06,
      "loss": 0.5941,
      "step": 461
    },
    {
      "epoch": 0.03389675064427671,
      "grad_norm": 2.3280342310760527,
      "learning_rate": 9.999603427274649e-06,
      "loss": 1.162,
      "step": 462
    },
    {
      "epoch": 0.03397012023441584,
      "grad_norm": 2.016372320279188,
      "learning_rate": 9.999588321294119e-06,
      "loss": 0.9353,
      "step": 463
    },
    {
      "epoch": 0.03404348982455497,
      "grad_norm": 0.7943835566418151,
      "learning_rate": 9.999572932974455e-06,
      "loss": 0.6105,
      "step": 464
    },
    {
      "epoch": 0.03404348982455497,
      "lm_loss": 1.809,
      "step": 464,
      "vm_loss": 0.0478
    },
    {
      "epoch": 0.03404348982455497,
      "lm_loss": 0.9794,
      "step": 464,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.03404348982455497,
      "lm_loss": 0.7107,
      "step": 464,
      "vm_loss": 0.2196
    },
    {
      "epoch": 0.03404348982455497,
      "lm_loss": 1.0313,
      "step": 464,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.03404348982455497,
      "lm_loss": 1.2544,
      "step": 464,
      "vm_loss": 0.0851
    },
    {
      "epoch": 0.03404348982455497,
      "lm_loss": 0.7293,
      "step": 464,
      "vm_loss": 0.1115
    },
    {
      "epoch": 0.03404348982455497,
      "lm_loss": 0.9392,
      "step": 464,
      "vm_loss": 0.1086
    },
    {
      "epoch": 0.03404348982455497,
      "lm_loss": 0.6826,
      "step": 464,
      "vm_loss": 0.1337
    },
    {
      "epoch": 0.0341168594146941,
      "grad_norm": 2.0036443739313077,
      "learning_rate": 9.999557262316528e-06,
      "loss": 0.9721,
      "step": 465
    },
    {
      "epoch": 0.03419022900483322,
      "grad_norm": 2.0706077089283705,
      "learning_rate": 9.99954130932122e-06,
      "loss": 1.0402,
      "step": 466
    },
    {
      "epoch": 0.03426359859497235,
      "grad_norm": 0.7412039030414711,
      "learning_rate": 9.999525073989434e-06,
      "loss": 0.6277,
      "step": 467
    },
    {
      "epoch": 0.03433696818511148,
      "grad_norm": 1.9018966903598549,
      "learning_rate": 9.999508556322083e-06,
      "loss": 1.0642,
      "step": 468
    },
    {
      "epoch": 0.034410337775250606,
      "grad_norm": 2.0843037487576836,
      "learning_rate": 9.999491756320105e-06,
      "loss": 1.0087,
      "step": 469
    },
    {
      "epoch": 0.03448370736538973,
      "grad_norm": 0.8331463499618759,
      "learning_rate": 9.999474673984448e-06,
      "loss": 0.6156,
      "step": 470
    },
    {
      "epoch": 0.034557076955528856,
      "grad_norm": 0.7515567805830877,
      "learning_rate": 9.999457309316073e-06,
      "loss": 0.6322,
      "step": 471
    },
    {
      "epoch": 0.034630446545667985,
      "grad_norm": 0.6432958825059372,
      "learning_rate": 9.999439662315964e-06,
      "loss": 0.6348,
      "step": 472
    },
    {
      "epoch": 0.034630446545667985,
      "lm_loss": 0.9857,
      "step": 472,
      "vm_loss": 0.053
    },
    {
      "epoch": 0.034630446545667985,
      "lm_loss": 1.1315,
      "step": 472,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.034630446545667985,
      "lm_loss": 0.5786,
      "step": 472,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.034630446545667985,
      "lm_loss": 0.6855,
      "step": 472,
      "vm_loss": 0.1209
    },
    {
      "epoch": 0.034630446545667985,
      "lm_loss": 0.1916,
      "step": 472,
      "vm_loss": 0.0534
    },
    {
      "epoch": 0.034630446545667985,
      "lm_loss": 0.9805,
      "step": 472,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.034630446545667985,
      "lm_loss": 1.4202,
      "step": 472,
      "vm_loss": 0.2618
    },
    {
      "epoch": 0.034630446545667985,
      "lm_loss": 0.1549,
      "step": 472,
      "vm_loss": 0.1483
    },
    {
      "epoch": 0.034703816135807114,
      "grad_norm": 2.239998841419035,
      "learning_rate": 9.999421732985115e-06,
      "loss": 0.976,
      "step": 473
    },
    {
      "epoch": 0.034777185725946236,
      "grad_norm": 0.7361118078877915,
      "learning_rate": 9.999403521324541e-06,
      "loss": 0.6297,
      "step": 474
    },
    {
      "epoch": 0.034850555316085365,
      "grad_norm": 1.868898446131533,
      "learning_rate": 9.99938502733527e-06,
      "loss": 0.9563,
      "step": 475
    },
    {
      "epoch": 0.03492392490622449,
      "grad_norm": 1.8039258321083982,
      "learning_rate": 9.999366251018344e-06,
      "loss": 0.9882,
      "step": 476
    },
    {
      "epoch": 0.03499729449636362,
      "grad_norm": 0.8828386893411717,
      "learning_rate": 9.999347192374826e-06,
      "loss": 0.6432,
      "step": 477
    },
    {
      "epoch": 0.035070664086502744,
      "grad_norm": 0.7876899544716072,
      "learning_rate": 9.999327851405792e-06,
      "loss": 0.6183,
      "step": 478
    },
    {
      "epoch": 0.03514403367664187,
      "grad_norm": 2.1730544489539265,
      "learning_rate": 9.999308228112332e-06,
      "loss": 1.0445,
      "step": 479
    },
    {
      "epoch": 0.035217403266781,
      "grad_norm": 2.2174619227860166,
      "learning_rate": 9.999288322495557e-06,
      "loss": 1.0161,
      "step": 480
    },
    {
      "epoch": 0.035217403266781,
      "lm_loss": 1.6507,
      "step": 480,
      "vm_loss": 0.0731
    },
    {
      "epoch": 0.035217403266781,
      "lm_loss": 0.3199,
      "step": 480,
      "vm_loss": 0.1258
    },
    {
      "epoch": 0.035217403266781,
      "lm_loss": 0.2888,
      "step": 480,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.035217403266781,
      "lm_loss": 1.0893,
      "step": 480,
      "vm_loss": 0.1882
    },
    {
      "epoch": 0.035217403266781,
      "lm_loss": 0.7379,
      "step": 480,
      "vm_loss": 0.1262
    },
    {
      "epoch": 0.035217403266781,
      "lm_loss": 0.2576,
      "step": 480,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.035217403266781,
      "lm_loss": 0.6671,
      "step": 480,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.035217403266781,
      "lm_loss": 0.7868,
      "step": 480,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.03529077285692013,
      "grad_norm": 1.8499954880517784,
      "learning_rate": 9.99926813455659e-06,
      "loss": 0.9136,
      "step": 481
    },
    {
      "epoch": 0.03536414244705925,
      "grad_norm": 2.103986917083345,
      "learning_rate": 9.999247664296569e-06,
      "loss": 0.9679,
      "step": 482
    },
    {
      "epoch": 0.03543751203719838,
      "grad_norm": 2.0505881264971033,
      "learning_rate": 9.999226911716654e-06,
      "loss": 1.0082,
      "step": 483
    },
    {
      "epoch": 0.03551088162733751,
      "grad_norm": 1.9937248300895734,
      "learning_rate": 9.999205876818015e-06,
      "loss": 0.9689,
      "step": 484
    },
    {
      "epoch": 0.03558425121747664,
      "grad_norm": 2.0729960267352667,
      "learning_rate": 9.999184559601838e-06,
      "loss": 0.9007,
      "step": 485
    },
    {
      "epoch": 0.03565762080761577,
      "grad_norm": 1.8670558873112009,
      "learning_rate": 9.999162960069332e-06,
      "loss": 0.9815,
      "step": 486
    },
    {
      "epoch": 0.03573099039775489,
      "grad_norm": 2.04302420665162,
      "learning_rate": 9.99914107822171e-06,
      "loss": 1.0278,
      "step": 487
    },
    {
      "epoch": 0.03580435998789402,
      "grad_norm": 1.9223372666615715,
      "learning_rate": 9.999118914060213e-06,
      "loss": 1.0112,
      "step": 488
    },
    {
      "epoch": 0.03580435998789402,
      "lm_loss": 1.1529,
      "step": 488,
      "vm_loss": 0.1292
    },
    {
      "epoch": 0.03580435998789402,
      "lm_loss": 0.247,
      "step": 488,
      "vm_loss": 0.154
    },
    {
      "epoch": 0.03580435998789402,
      "lm_loss": 0.8172,
      "step": 488,
      "vm_loss": 0.155
    },
    {
      "epoch": 0.03580435998789402,
      "lm_loss": 0.7368,
      "step": 488,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.03580435998789402,
      "lm_loss": 1.1675,
      "step": 488,
      "vm_loss": 0.0434
    },
    {
      "epoch": 0.03580435998789402,
      "lm_loss": 1.3699,
      "step": 488,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.03580435998789402,
      "lm_loss": 0.9173,
      "step": 488,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.03580435998789402,
      "lm_loss": 1.3918,
      "step": 488,
      "vm_loss": 0.0414
    },
    {
      "epoch": 0.035877729578033146,
      "grad_norm": 1.993911487151053,
      "learning_rate": 9.99909646758609e-06,
      "loss": 0.9313,
      "step": 489
    },
    {
      "epoch": 0.035951099168172275,
      "grad_norm": 1.8521138310877296,
      "learning_rate": 9.999073738800609e-06,
      "loss": 0.9269,
      "step": 490
    },
    {
      "epoch": 0.0360244687583114,
      "grad_norm": 1.761899909361011,
      "learning_rate": 9.999050727705053e-06,
      "loss": 0.9697,
      "step": 491
    },
    {
      "epoch": 0.036097838348450526,
      "grad_norm": 1.323237067189758,
      "learning_rate": 9.999027434300725e-06,
      "loss": 0.6234,
      "step": 492
    },
    {
      "epoch": 0.036171207938589654,
      "grad_norm": 1.8654245169398438,
      "learning_rate": 9.999003858588937e-06,
      "loss": 1.032,
      "step": 493
    },
    {
      "epoch": 0.03624457752872878,
      "grad_norm": 2.063996451659018,
      "learning_rate": 9.998980000571021e-06,
      "loss": 0.9793,
      "step": 494
    },
    {
      "epoch": 0.036317947118867905,
      "grad_norm": 2.0544720110580283,
      "learning_rate": 9.998955860248325e-06,
      "loss": 0.9764,
      "step": 495
    },
    {
      "epoch": 0.036391316709007034,
      "grad_norm": 1.805215214504421,
      "learning_rate": 9.998931437622213e-06,
      "loss": 0.9381,
      "step": 496
    },
    {
      "epoch": 0.036391316709007034,
      "lm_loss": 1.3345,
      "step": 496,
      "vm_loss": 0.0355
    },
    {
      "epoch": 0.036391316709007034,
      "lm_loss": 0.2352,
      "step": 496,
      "vm_loss": 0.1216
    },
    {
      "epoch": 0.036391316709007034,
      "lm_loss": 1.3783,
      "step": 496,
      "vm_loss": 0.1757
    },
    {
      "epoch": 0.036391316709007034,
      "lm_loss": 0.3593,
      "step": 496,
      "vm_loss": 0.0838
    },
    {
      "epoch": 0.036391316709007034,
      "lm_loss": 0.3635,
      "step": 496,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.036391316709007034,
      "lm_loss": 1.1138,
      "step": 496,
      "vm_loss": 0.1488
    },
    {
      "epoch": 0.036391316709007034,
      "lm_loss": 1.3568,
      "step": 496,
      "vm_loss": 0.166
    },
    {
      "epoch": 0.036391316709007034,
      "lm_loss": 0.7823,
      "step": 496,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.03646468629914616,
      "grad_norm": 1.9112149041699629,
      "learning_rate": 9.99890673269406e-06,
      "loss": 0.9874,
      "step": 497
    },
    {
      "epoch": 0.03653805588928529,
      "grad_norm": 0.7496342299662401,
      "learning_rate": 9.998881745465267e-06,
      "loss": 0.6039,
      "step": 498
    },
    {
      "epoch": 0.03661142547942441,
      "grad_norm": 1.869859292857326,
      "learning_rate": 9.998856475937242e-06,
      "loss": 1.0344,
      "step": 499
    },
    {
      "epoch": 0.03668479506956354,
      "grad_norm": 2.0666561468687448,
      "learning_rate": 9.998830924111411e-06,
      "loss": 1.0551,
      "step": 500
    },
    {
      "epoch": 0.03675816465970267,
      "grad_norm": 1.6608633467024032,
      "learning_rate": 9.99880508998922e-06,
      "loss": 0.996,
      "step": 501
    },
    {
      "epoch": 0.0368315342498418,
      "grad_norm": 2.043507480919614,
      "learning_rate": 9.998778973572124e-06,
      "loss": 1.0203,
      "step": 502
    },
    {
      "epoch": 0.03690490383998092,
      "grad_norm": 0.6717807733344361,
      "learning_rate": 9.9987525748616e-06,
      "loss": 0.6097,
      "step": 503
    },
    {
      "epoch": 0.03697827343012005,
      "grad_norm": 1.9109423382063562,
      "learning_rate": 9.99872589385914e-06,
      "loss": 0.8814,
      "step": 504
    },
    {
      "epoch": 0.03697827343012005,
      "lm_loss": 0.1513,
      "step": 504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03697827343012005,
      "lm_loss": 0.5252,
      "step": 504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03697827343012005,
      "lm_loss": 0.6241,
      "step": 504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03697827343012005,
      "lm_loss": 0.6791,
      "step": 504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03697827343012005,
      "lm_loss": 0.6603,
      "step": 504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03697827343012005,
      "lm_loss": 0.5248,
      "step": 504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03697827343012005,
      "lm_loss": 0.5021,
      "step": 504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03697827343012005,
      "lm_loss": 0.7502,
      "step": 504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.03705164302025918,
      "grad_norm": 0.6544071569693122,
      "learning_rate": 9.998698930566251e-06,
      "loss": 0.6065,
      "step": 505
    },
    {
      "epoch": 0.03712501261039831,
      "grad_norm": 2.006877239489741,
      "learning_rate": 9.998671684984454e-06,
      "loss": 0.9457,
      "step": 506
    },
    {
      "epoch": 0.03719838220053743,
      "grad_norm": 0.6096468980018867,
      "learning_rate": 9.998644157115287e-06,
      "loss": 0.6362,
      "step": 507
    },
    {
      "epoch": 0.03727175179067656,
      "grad_norm": 1.6536709918213397,
      "learning_rate": 9.998616346960306e-06,
      "loss": 0.8768,
      "step": 508
    },
    {
      "epoch": 0.037345121380815686,
      "grad_norm": 1.7758522008087108,
      "learning_rate": 9.99858825452108e-06,
      "loss": 1.0,
      "step": 509
    },
    {
      "epoch": 0.037418490970954815,
      "grad_norm": 1.9282046366987,
      "learning_rate": 9.998559879799196e-06,
      "loss": 0.9544,
      "step": 510
    },
    {
      "epoch": 0.037491860561093944,
      "grad_norm": 2.11036001936599,
      "learning_rate": 9.99853122279626e-06,
      "loss": 0.9511,
      "step": 511
    },
    {
      "epoch": 0.037565230151233066,
      "grad_norm": 1.6927330058179761,
      "learning_rate": 9.998502283513884e-06,
      "loss": 0.9095,
      "step": 512
    },
    {
      "epoch": 0.037565230151233066,
      "lm_loss": 0.8155,
      "step": 512,
      "vm_loss": 0.0728
    },
    {
      "epoch": 0.037565230151233066,
      "lm_loss": 1.8347,
      "step": 512,
      "vm_loss": 0.082
    },
    {
      "epoch": 0.037565230151233066,
      "lm_loss": 1.3079,
      "step": 512,
      "vm_loss": 0.1923
    },
    {
      "epoch": 0.037565230151233066,
      "lm_loss": 1.2053,
      "step": 512,
      "vm_loss": 0.1738
    },
    {
      "epoch": 0.037565230151233066,
      "lm_loss": 0.6791,
      "step": 512,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.037565230151233066,
      "lm_loss": 1.107,
      "step": 512,
      "vm_loss": 0.1426
    },
    {
      "epoch": 0.037565230151233066,
      "lm_loss": 0.4078,
      "step": 512,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.037565230151233066,
      "lm_loss": 0.1795,
      "step": 512,
      "vm_loss": 0.0863
    },
    {
      "epoch": 0.037638599741372195,
      "grad_norm": 1.7643234118162612,
      "learning_rate": 9.998473061953706e-06,
      "loss": 0.9723,
      "step": 513
    },
    {
      "epoch": 0.03771196933151132,
      "grad_norm": 1.7650179336812752,
      "learning_rate": 9.998443558117376e-06,
      "loss": 0.9824,
      "step": 514
    },
    {
      "epoch": 0.03778533892165045,
      "grad_norm": 1.9761772823220307,
      "learning_rate": 9.998413772006561e-06,
      "loss": 1.0747,
      "step": 515
    },
    {
      "epoch": 0.037858708511789574,
      "grad_norm": 2.005283548948104,
      "learning_rate": 9.998383703622942e-06,
      "loss": 0.9561,
      "step": 516
    },
    {
      "epoch": 0.0379320781019287,
      "grad_norm": 1.9992391920233525,
      "learning_rate": 9.998353352968215e-06,
      "loss": 1.0064,
      "step": 517
    },
    {
      "epoch": 0.03800544769206783,
      "grad_norm": 1.869266556840006,
      "learning_rate": 9.998322720044098e-06,
      "loss": 0.8989,
      "step": 518
    },
    {
      "epoch": 0.03807881728220696,
      "grad_norm": 1.6715549867067792,
      "learning_rate": 9.998291804852318e-06,
      "loss": 0.9425,
      "step": 519
    },
    {
      "epoch": 0.03815218687234608,
      "grad_norm": 1.9933132921805428,
      "learning_rate": 9.99826060739462e-06,
      "loss": 0.9113,
      "step": 520
    },
    {
      "epoch": 0.03815218687234608,
      "lm_loss": 1.2836,
      "step": 520,
      "vm_loss": 0.0947
    },
    {
      "epoch": 0.03815218687234608,
      "lm_loss": 1.3969,
      "step": 520,
      "vm_loss": 0.0445
    },
    {
      "epoch": 0.03815218687234608,
      "lm_loss": 0.8619,
      "step": 520,
      "vm_loss": 0.149
    },
    {
      "epoch": 0.03815218687234608,
      "lm_loss": 1.012,
      "step": 520,
      "vm_loss": 0.094
    },
    {
      "epoch": 0.03815218687234608,
      "lm_loss": 0.7243,
      "step": 520,
      "vm_loss": 0.1702
    },
    {
      "epoch": 0.03815218687234608,
      "lm_loss": 0.1972,
      "step": 520,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.03815218687234608,
      "lm_loss": 0.2665,
      "step": 520,
      "vm_loss": 0.1492
    },
    {
      "epoch": 0.03815218687234608,
      "lm_loss": 1.3389,
      "step": 520,
      "vm_loss": 0.1806
    },
    {
      "epoch": 0.03822555646248521,
      "grad_norm": 2.080047265141777,
      "learning_rate": 9.998229127672771e-06,
      "loss": 0.968,
      "step": 521
    },
    {
      "epoch": 0.03829892605262434,
      "grad_norm": 1.7462989649942078,
      "learning_rate": 9.998197365688542e-06,
      "loss": 0.9837,
      "step": 522
    },
    {
      "epoch": 0.03837229564276347,
      "grad_norm": 1.7867400101631747,
      "learning_rate": 9.998165321443731e-06,
      "loss": 0.9155,
      "step": 523
    },
    {
      "epoch": 0.03844566523290259,
      "grad_norm": 1.8342081119388745,
      "learning_rate": 9.998132994940149e-06,
      "loss": 0.9291,
      "step": 524
    },
    {
      "epoch": 0.03851903482304172,
      "grad_norm": 2.1357803387472947,
      "learning_rate": 9.998100386179615e-06,
      "loss": 0.9218,
      "step": 525
    },
    {
      "epoch": 0.03859240441318085,
      "grad_norm": 1.9029198671635024,
      "learning_rate": 9.998067495163977e-06,
      "loss": 0.9603,
      "step": 526
    },
    {
      "epoch": 0.038665774003319976,
      "grad_norm": 1.9081634370226306,
      "learning_rate": 9.99803432189509e-06,
      "loss": 0.957,
      "step": 527
    },
    {
      "epoch": 0.0387391435934591,
      "grad_norm": 0.8666680137340933,
      "learning_rate": 9.998000866374826e-06,
      "loss": 0.6489,
      "step": 528
    },
    {
      "epoch": 0.0387391435934591,
      "lm_loss": 1.4015,
      "step": 528,
      "vm_loss": 0.1767
    },
    {
      "epoch": 0.0387391435934591,
      "lm_loss": 0.8851,
      "step": 528,
      "vm_loss": 0.1059
    },
    {
      "epoch": 0.0387391435934591,
      "lm_loss": 0.4855,
      "step": 528,
      "vm_loss": 0.2022
    },
    {
      "epoch": 0.0387391435934591,
      "lm_loss": 0.2672,
      "step": 528,
      "vm_loss": 0.1196
    },
    {
      "epoch": 0.0387391435934591,
      "lm_loss": 0.2456,
      "step": 528,
      "vm_loss": 0.0995
    },
    {
      "epoch": 0.0387391435934591,
      "lm_loss": 0.7539,
      "step": 528,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.0387391435934591,
      "lm_loss": 0.4944,
      "step": 528,
      "vm_loss": 0.1326
    },
    {
      "epoch": 0.0387391435934591,
      "lm_loss": 1.4614,
      "step": 528,
      "vm_loss": 0.2327
    },
    {
      "epoch": 0.03881251318359823,
      "grad_norm": 1.839036213930504,
      "learning_rate": 9.997967128605078e-06,
      "loss": 0.9388,
      "step": 529
    },
    {
      "epoch": 0.038885882773737356,
      "grad_norm": 0.6886881773578881,
      "learning_rate": 9.997933108587747e-06,
      "loss": 0.636,
      "step": 530
    },
    {
      "epoch": 0.038959252363876484,
      "grad_norm": 1.8827140739396369,
      "learning_rate": 9.997898806324758e-06,
      "loss": 0.9697,
      "step": 531
    },
    {
      "epoch": 0.039032621954015606,
      "grad_norm": 1.800697593244807,
      "learning_rate": 9.997864221818045e-06,
      "loss": 0.9419,
      "step": 532
    },
    {
      "epoch": 0.039105991544154735,
      "grad_norm": 0.7006711922529779,
      "learning_rate": 9.997829355069561e-06,
      "loss": 0.597,
      "step": 533
    },
    {
      "epoch": 0.039179361134293864,
      "grad_norm": 1.9404991847856405,
      "learning_rate": 9.99779420608128e-06,
      "loss": 1.0659,
      "step": 534
    },
    {
      "epoch": 0.03925273072443299,
      "grad_norm": 0.7203800123363285,
      "learning_rate": 9.99775877485518e-06,
      "loss": 0.6312,
      "step": 535
    },
    {
      "epoch": 0.03932610031457212,
      "grad_norm": 1.9694916531101312,
      "learning_rate": 9.997723061393267e-06,
      "loss": 0.8501,
      "step": 536
    },
    {
      "epoch": 0.03932610031457212,
      "lm_loss": 2.0626,
      "step": 536,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.03932610031457212,
      "lm_loss": 0.539,
      "step": 536,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.03932610031457212,
      "lm_loss": 0.3248,
      "step": 536,
      "vm_loss": 0.0802
    },
    {
      "epoch": 0.03932610031457212,
      "lm_loss": 1.1541,
      "step": 536,
      "vm_loss": 0.1697
    },
    {
      "epoch": 0.03932610031457212,
      "lm_loss": 1.1389,
      "step": 536,
      "vm_loss": 0.0831
    },
    {
      "epoch": 0.03932610031457212,
      "lm_loss": 0.1669,
      "step": 536,
      "vm_loss": 0.1451
    },
    {
      "epoch": 0.03932610031457212,
      "lm_loss": 0.8388,
      "step": 536,
      "vm_loss": 0.1854
    },
    {
      "epoch": 0.03932610031457212,
      "lm_loss": 1.0703,
      "step": 536,
      "vm_loss": 0.0431
    },
    {
      "epoch": 0.03939946990471124,
      "grad_norm": 2.0243000467903007,
      "learning_rate": 9.997687065697556e-06,
      "loss": 1.0037,
      "step": 537
    },
    {
      "epoch": 0.03947283949485037,
      "grad_norm": 1.7806116102824112,
      "learning_rate": 9.99765078777008e-06,
      "loss": 0.9987,
      "step": 538
    },
    {
      "epoch": 0.0395462090849895,
      "grad_norm": 1.8511437131102848,
      "learning_rate": 9.997614227612887e-06,
      "loss": 0.983,
      "step": 539
    },
    {
      "epoch": 0.03961957867512863,
      "grad_norm": 2.031558285561765,
      "learning_rate": 9.997577385228043e-06,
      "loss": 0.9383,
      "step": 540
    },
    {
      "epoch": 0.03969294826526775,
      "grad_norm": 1.7711036481670819,
      "learning_rate": 9.997540260617626e-06,
      "loss": 1.0345,
      "step": 541
    },
    {
      "epoch": 0.03976631785540688,
      "grad_norm": 0.8003465631404526,
      "learning_rate": 9.997502853783736e-06,
      "loss": 0.6012,
      "step": 542
    },
    {
      "epoch": 0.03983968744554601,
      "grad_norm": 1.9929937013507577,
      "learning_rate": 9.997465164728482e-06,
      "loss": 0.9801,
      "step": 543
    },
    {
      "epoch": 0.03991305703568514,
      "grad_norm": 0.7990288013009333,
      "learning_rate": 9.997427193453996e-06,
      "loss": 0.6085,
      "step": 544
    },
    {
      "epoch": 0.03991305703568514,
      "lm_loss": 0.7963,
      "step": 544,
      "vm_loss": 0.1284
    },
    {
      "epoch": 0.03991305703568514,
      "lm_loss": 1.3268,
      "step": 544,
      "vm_loss": 0.1569
    },
    {
      "epoch": 0.03991305703568514,
      "lm_loss": 0.2034,
      "step": 544,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.03991305703568514,
      "lm_loss": 1.085,
      "step": 544,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.03991305703568514,
      "lm_loss": 0.4435,
      "step": 544,
      "vm_loss": 0.1271
    },
    {
      "epoch": 0.03991305703568514,
      "lm_loss": 1.1828,
      "step": 544,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.03991305703568514,
      "lm_loss": 1.2214,
      "step": 544,
      "vm_loss": 0.1099
    },
    {
      "epoch": 0.03991305703568514,
      "lm_loss": 1.0699,
      "step": 544,
      "vm_loss": 0.1599
    },
    {
      "epoch": 0.03998642662582426,
      "grad_norm": 1.684691474192487,
      "learning_rate": 9.99738893996242e-06,
      "loss": 0.9507,
      "step": 545
    },
    {
      "epoch": 0.04005979621596339,
      "grad_norm": 1.8397458100431405,
      "learning_rate": 9.997350404255915e-06,
      "loss": 0.9753,
      "step": 546
    },
    {
      "epoch": 0.040133165806102516,
      "grad_norm": 1.9108201098234971,
      "learning_rate": 9.997311586336655e-06,
      "loss": 0.9033,
      "step": 547
    },
    {
      "epoch": 0.040206535396241645,
      "grad_norm": 1.961957924282491,
      "learning_rate": 9.997272486206836e-06,
      "loss": 1.094,
      "step": 548
    },
    {
      "epoch": 0.04027990498638077,
      "grad_norm": 1.947860911564198,
      "learning_rate": 9.997233103868664e-06,
      "loss": 1.0501,
      "step": 549
    },
    {
      "epoch": 0.040353274576519896,
      "grad_norm": 0.776854474877715,
      "learning_rate": 9.997193439324362e-06,
      "loss": 0.6176,
      "step": 550
    },
    {
      "epoch": 0.040426644166659025,
      "grad_norm": 2.030430875301233,
      "learning_rate": 9.997153492576172e-06,
      "loss": 0.9869,
      "step": 551
    },
    {
      "epoch": 0.04050001375679815,
      "grad_norm": 1.7734825074932918,
      "learning_rate": 9.997113263626349e-06,
      "loss": 1.0535,
      "step": 552
    },
    {
      "epoch": 0.04050001375679815,
      "lm_loss": 0.1977,
      "step": 552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04050001375679815,
      "lm_loss": 0.6575,
      "step": 552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04050001375679815,
      "lm_loss": 0.4668,
      "step": 552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04050001375679815,
      "lm_loss": 0.7678,
      "step": 552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04050001375679815,
      "lm_loss": 0.4688,
      "step": 552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04050001375679815,
      "lm_loss": 0.4829,
      "step": 552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04050001375679815,
      "lm_loss": 0.7895,
      "step": 552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04050001375679815,
      "lm_loss": 0.6647,
      "step": 552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.040573383346937275,
      "grad_norm": 0.712045064109056,
      "learning_rate": 9.997072752477165e-06,
      "loss": 0.6323,
      "step": 553
    },
    {
      "epoch": 0.040646752937076404,
      "grad_norm": 1.7630652931456015,
      "learning_rate": 9.997031959130908e-06,
      "loss": 0.9941,
      "step": 554
    },
    {
      "epoch": 0.04072012252721553,
      "grad_norm": 1.7989494637952235,
      "learning_rate": 9.99699088358988e-06,
      "loss": 0.8184,
      "step": 555
    },
    {
      "epoch": 0.04079349211735466,
      "grad_norm": 0.6447905899312101,
      "learning_rate": 9.996949525856404e-06,
      "loss": 0.6005,
      "step": 556
    },
    {
      "epoch": 0.04086686170749378,
      "grad_norm": 0.6284217477362696,
      "learning_rate": 9.996907885932812e-06,
      "loss": 0.609,
      "step": 557
    },
    {
      "epoch": 0.04094023129763291,
      "grad_norm": 0.6140459387926713,
      "learning_rate": 9.996865963821455e-06,
      "loss": 0.6038,
      "step": 558
    },
    {
      "epoch": 0.04101360088777204,
      "grad_norm": 2.300437687769957,
      "learning_rate": 9.996823759524705e-06,
      "loss": 0.9802,
      "step": 559
    },
    {
      "epoch": 0.04108697047791117,
      "grad_norm": 2.1736821690428676,
      "learning_rate": 9.996781273044943e-06,
      "loss": 0.9704,
      "step": 560
    },
    {
      "epoch": 0.04108697047791117,
      "lm_loss": 1.5166,
      "step": 560,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.04108697047791117,
      "lm_loss": 0.7098,
      "step": 560,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.04108697047791117,
      "lm_loss": 1.3164,
      "step": 560,
      "vm_loss": 0.0974
    },
    {
      "epoch": 0.04108697047791117,
      "lm_loss": 1.1239,
      "step": 560,
      "vm_loss": 0.1808
    },
    {
      "epoch": 0.04108697047791117,
      "lm_loss": 0.5981,
      "step": 560,
      "vm_loss": 0.1789
    },
    {
      "epoch": 0.04108697047791117,
      "lm_loss": 1.004,
      "step": 560,
      "vm_loss": 0.1786
    },
    {
      "epoch": 0.04108697047791117,
      "lm_loss": 0.7925,
      "step": 560,
      "vm_loss": 0.0863
    },
    {
      "epoch": 0.04108697047791117,
      "lm_loss": 0.8559,
      "step": 560,
      "vm_loss": 0.1537
    },
    {
      "epoch": 0.0411603400680503,
      "grad_norm": 1.9444462175030675,
      "learning_rate": 9.996738504384568e-06,
      "loss": 0.997,
      "step": 561
    },
    {
      "epoch": 0.04123370965818942,
      "grad_norm": 1.9124010687952355,
      "learning_rate": 9.996695453545996e-06,
      "loss": 0.8978,
      "step": 562
    },
    {
      "epoch": 0.04130707924832855,
      "grad_norm": 1.6862482221853283,
      "learning_rate": 9.996652120531657e-06,
      "loss": 0.9412,
      "step": 563
    },
    {
      "epoch": 0.04138044883846768,
      "grad_norm": 0.8058812095028838,
      "learning_rate": 9.996608505343999e-06,
      "loss": 0.6137,
      "step": 564
    },
    {
      "epoch": 0.041453818428606806,
      "grad_norm": 2.2733845618463544,
      "learning_rate": 9.996564607985484e-06,
      "loss": 0.9558,
      "step": 565
    },
    {
      "epoch": 0.04152718801874593,
      "grad_norm": 0.6887232772215448,
      "learning_rate": 9.996520428458592e-06,
      "loss": 0.6418,
      "step": 566
    },
    {
      "epoch": 0.04160055760888506,
      "grad_norm": 2.1725852778182904,
      "learning_rate": 9.996475966765818e-06,
      "loss": 1.0063,
      "step": 567
    },
    {
      "epoch": 0.041673927199024186,
      "grad_norm": 1.908388131923212,
      "learning_rate": 9.996431222909672e-06,
      "loss": 1.022,
      "step": 568
    },
    {
      "epoch": 0.041673927199024186,
      "lm_loss": 0.669,
      "step": 568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.041673927199024186,
      "lm_loss": 0.6784,
      "step": 568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.041673927199024186,
      "lm_loss": 0.7293,
      "step": 568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.041673927199024186,
      "lm_loss": 0.6378,
      "step": 568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.041673927199024186,
      "lm_loss": 0.6383,
      "step": 568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.041673927199024186,
      "lm_loss": 0.7523,
      "step": 568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.041673927199024186,
      "lm_loss": 0.6128,
      "step": 568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.041673927199024186,
      "lm_loss": 0.467,
      "step": 568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.041747296789163314,
      "grad_norm": 0.608238830338338,
      "learning_rate": 9.996386196892683e-06,
      "loss": 0.617,
      "step": 569
    },
    {
      "epoch": 0.041820666379302436,
      "grad_norm": 2.06171307876913,
      "learning_rate": 9.99634088871739e-06,
      "loss": 0.9452,
      "step": 570
    },
    {
      "epoch": 0.041894035969441565,
      "grad_norm": 0.6593646778587742,
      "learning_rate": 9.996295298386355e-06,
      "loss": 0.6067,
      "step": 571
    },
    {
      "epoch": 0.041967405559580694,
      "grad_norm": 0.6516602514412855,
      "learning_rate": 9.996249425902152e-06,
      "loss": 0.6021,
      "step": 572
    },
    {
      "epoch": 0.04204077514971982,
      "grad_norm": 1.8807581631214125,
      "learning_rate": 9.99620327126737e-06,
      "loss": 0.9591,
      "step": 573
    },
    {
      "epoch": 0.042114144739858944,
      "grad_norm": 0.6318677312310618,
      "learning_rate": 9.996156834484619e-06,
      "loss": 0.6318,
      "step": 574
    },
    {
      "epoch": 0.04218751432999807,
      "grad_norm": 1.8366083838860436,
      "learning_rate": 9.996110115556514e-06,
      "loss": 0.9835,
      "step": 575
    },
    {
      "epoch": 0.0422608839201372,
      "grad_norm": 1.6618538372800622,
      "learning_rate": 9.996063114485701e-06,
      "loss": 0.8865,
      "step": 576
    },
    {
      "epoch": 0.0422608839201372,
      "lm_loss": 0.3969,
      "step": 576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0422608839201372,
      "lm_loss": 0.5575,
      "step": 576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0422608839201372,
      "lm_loss": 0.6956,
      "step": 576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0422608839201372,
      "lm_loss": 0.5858,
      "step": 576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0422608839201372,
      "lm_loss": 0.6307,
      "step": 576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0422608839201372,
      "lm_loss": 0.5655,
      "step": 576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0422608839201372,
      "lm_loss": 0.2901,
      "step": 576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0422608839201372,
      "lm_loss": 0.6919,
      "step": 576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04233425351027633,
      "grad_norm": 0.7088917050947213,
      "learning_rate": 9.996015831274831e-06,
      "loss": 0.6302,
      "step": 577
    },
    {
      "epoch": 0.04240762310041545,
      "grad_norm": 1.8902262242609202,
      "learning_rate": 9.995968265926574e-06,
      "loss": 0.9453,
      "step": 578
    },
    {
      "epoch": 0.04248099269055458,
      "grad_norm": 2.121196031750506,
      "learning_rate": 9.995920418443619e-06,
      "loss": 1.0064,
      "step": 579
    },
    {
      "epoch": 0.04255436228069371,
      "grad_norm": 1.8567427035306525,
      "learning_rate": 9.995872288828662e-06,
      "loss": 0.8824,
      "step": 580
    },
    {
      "epoch": 0.04262773187083284,
      "grad_norm": 0.6539189302652387,
      "learning_rate": 9.995823877084427e-06,
      "loss": 0.6052,
      "step": 581
    },
    {
      "epoch": 0.04270110146097196,
      "grad_norm": 1.9778946248910092,
      "learning_rate": 9.995775183213647e-06,
      "loss": 0.9372,
      "step": 582
    },
    {
      "epoch": 0.04277447105111109,
      "grad_norm": 0.6118402101346287,
      "learning_rate": 9.995726207219068e-06,
      "loss": 0.5987,
      "step": 583
    },
    {
      "epoch": 0.04284784064125022,
      "grad_norm": 1.6888316352509767,
      "learning_rate": 9.995676949103459e-06,
      "loss": 0.8672,
      "step": 584
    },
    {
      "epoch": 0.04284784064125022,
      "lm_loss": 1.9472,
      "step": 584,
      "vm_loss": 0.0582
    },
    {
      "epoch": 0.04284784064125022,
      "lm_loss": 1.1216,
      "step": 584,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.04284784064125022,
      "lm_loss": 0.6984,
      "step": 584,
      "vm_loss": 0.2327
    },
    {
      "epoch": 0.04284784064125022,
      "lm_loss": 0.9035,
      "step": 584,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.04284784064125022,
      "lm_loss": 0.8254,
      "step": 584,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.04284784064125022,
      "lm_loss": 0.9635,
      "step": 584,
      "vm_loss": 0.1425
    },
    {
      "epoch": 0.04284784064125022,
      "lm_loss": 0.3729,
      "step": 584,
      "vm_loss": 0.1959
    },
    {
      "epoch": 0.04284784064125022,
      "lm_loss": 0.1129,
      "step": 584,
      "vm_loss": 0.1503
    },
    {
      "epoch": 0.042921210231389346,
      "grad_norm": 1.8954530112627053,
      "learning_rate": 9.995627408869602e-06,
      "loss": 1.101,
      "step": 585
    },
    {
      "epoch": 0.042994579821528475,
      "grad_norm": 0.6146171570998747,
      "learning_rate": 9.995577586520294e-06,
      "loss": 0.6149,
      "step": 586
    },
    {
      "epoch": 0.0430679494116676,
      "grad_norm": 1.9980250686577419,
      "learning_rate": 9.995527482058348e-06,
      "loss": 0.9406,
      "step": 587
    },
    {
      "epoch": 0.043141319001806726,
      "grad_norm": 1.9592984357145213,
      "learning_rate": 9.995477095486592e-06,
      "loss": 0.9971,
      "step": 588
    },
    {
      "epoch": 0.043214688591945855,
      "grad_norm": 1.6760990903840063,
      "learning_rate": 9.995426426807875e-06,
      "loss": 0.9985,
      "step": 589
    },
    {
      "epoch": 0.04328805818208498,
      "grad_norm": 1.747695475880315,
      "learning_rate": 9.995375476025058e-06,
      "loss": 0.9033,
      "step": 590
    },
    {
      "epoch": 0.043361427772224105,
      "grad_norm": 2.045500234400235,
      "learning_rate": 9.995324243141016e-06,
      "loss": 0.9425,
      "step": 591
    },
    {
      "epoch": 0.043434797362363234,
      "grad_norm": 1.8285019495034234,
      "learning_rate": 9.995272728158642e-06,
      "loss": 0.9036,
      "step": 592
    },
    {
      "epoch": 0.043434797362363234,
      "lm_loss": 1.1545,
      "step": 592,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.043434797362363234,
      "lm_loss": 1.0945,
      "step": 592,
      "vm_loss": 0.1833
    },
    {
      "epoch": 0.043434797362363234,
      "lm_loss": 0.9702,
      "step": 592,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.043434797362363234,
      "lm_loss": 1.3806,
      "step": 592,
      "vm_loss": 0.1634
    },
    {
      "epoch": 0.043434797362363234,
      "lm_loss": 0.809,
      "step": 592,
      "vm_loss": 0.1813
    },
    {
      "epoch": 0.043434797362363234,
      "lm_loss": 1.2173,
      "step": 592,
      "vm_loss": 0.1375
    },
    {
      "epoch": 0.043434797362363234,
      "lm_loss": 1.2768,
      "step": 592,
      "vm_loss": 0.1871
    },
    {
      "epoch": 0.043434797362363234,
      "lm_loss": 1.0006,
      "step": 592,
      "vm_loss": 0.1359
    },
    {
      "epoch": 0.04350816695250236,
      "grad_norm": 1.8575939534539943,
      "learning_rate": 9.995220931080847e-06,
      "loss": 0.9963,
      "step": 593
    },
    {
      "epoch": 0.04358153654264149,
      "grad_norm": 1.9469047454100148,
      "learning_rate": 9.995168851910556e-06,
      "loss": 0.8941,
      "step": 594
    },
    {
      "epoch": 0.04365490613278061,
      "grad_norm": 2.0159076755719103,
      "learning_rate": 9.99511649065071e-06,
      "loss": 0.9806,
      "step": 595
    },
    {
      "epoch": 0.04372827572291974,
      "grad_norm": 0.7279362857192273,
      "learning_rate": 9.995063847304264e-06,
      "loss": 0.613,
      "step": 596
    },
    {
      "epoch": 0.04380164531305887,
      "grad_norm": 2.025131395003544,
      "learning_rate": 9.995010921874192e-06,
      "loss": 1.0016,
      "step": 597
    },
    {
      "epoch": 0.043875014903198,
      "grad_norm": 0.6183206765115573,
      "learning_rate": 9.994957714363487e-06,
      "loss": 0.5991,
      "step": 598
    },
    {
      "epoch": 0.04394838449333712,
      "grad_norm": 1.9509076096688434,
      "learning_rate": 9.994904224775149e-06,
      "loss": 0.8529,
      "step": 599
    },
    {
      "epoch": 0.04402175408347625,
      "grad_norm": 2.26143884523317,
      "learning_rate": 9.994850453112198e-06,
      "loss": 0.9957,
      "step": 600
    },
    {
      "epoch": 0.04402175408347625,
      "lm_loss": 1.2893,
      "step": 600,
      "vm_loss": 0.0831
    },
    {
      "epoch": 0.04402175408347625,
      "lm_loss": 0.4159,
      "step": 600,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.04402175408347625,
      "lm_loss": 0.6624,
      "step": 600,
      "vm_loss": 0.0987
    },
    {
      "epoch": 0.04402175408347625,
      "lm_loss": 0.0067,
      "step": 600,
      "vm_loss": 0.1286
    },
    {
      "epoch": 0.04402175408347625,
      "lm_loss": 0.5599,
      "step": 600,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.04402175408347625,
      "lm_loss": 0.8116,
      "step": 600,
      "vm_loss": 0.082
    },
    {
      "epoch": 0.04402175408347625,
      "lm_loss": 0.5589,
      "step": 600,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.04402175408347625,
      "lm_loss": 1.3878,
      "step": 600,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.04409512367361538,
      "grad_norm": 1.9199228993422977,
      "learning_rate": 9.994796399377671e-06,
      "loss": 0.8592,
      "step": 601
    },
    {
      "epoch": 0.04416849326375451,
      "grad_norm": 2.0330567090121483,
      "learning_rate": 9.994742063574625e-06,
      "loss": 0.9821,
      "step": 602
    },
    {
      "epoch": 0.04424186285389363,
      "grad_norm": 1.960448815670139,
      "learning_rate": 9.994687445706123e-06,
      "loss": 0.898,
      "step": 603
    },
    {
      "epoch": 0.04431523244403276,
      "grad_norm": 1.8053295865849717,
      "learning_rate": 9.994632545775254e-06,
      "loss": 0.9966,
      "step": 604
    },
    {
      "epoch": 0.04438860203417189,
      "grad_norm": 2.25351761718805,
      "learning_rate": 9.994577363785116e-06,
      "loss": 0.9279,
      "step": 605
    },
    {
      "epoch": 0.044461971624311016,
      "grad_norm": 1.819010904679616,
      "learning_rate": 9.994521899738825e-06,
      "loss": 0.9021,
      "step": 606
    },
    {
      "epoch": 0.04453534121445014,
      "grad_norm": 2.224047254892503,
      "learning_rate": 9.994466153639513e-06,
      "loss": 0.9814,
      "step": 607
    },
    {
      "epoch": 0.044608710804589266,
      "grad_norm": 0.8119874257068708,
      "learning_rate": 9.99441012549033e-06,
      "loss": 0.6265,
      "step": 608
    },
    {
      "epoch": 0.044608710804589266,
      "lm_loss": 1.5086,
      "step": 608,
      "vm_loss": 0.0301
    },
    {
      "epoch": 0.044608710804589266,
      "lm_loss": 0.5329,
      "step": 608,
      "vm_loss": 0.0696
    },
    {
      "epoch": 0.044608710804589266,
      "lm_loss": 1.0117,
      "step": 608,
      "vm_loss": 0.2457
    },
    {
      "epoch": 0.044608710804589266,
      "lm_loss": 0.4644,
      "step": 608,
      "vm_loss": 0.1638
    },
    {
      "epoch": 0.044608710804589266,
      "lm_loss": 1.2803,
      "step": 608,
      "vm_loss": 0.235
    },
    {
      "epoch": 0.044608710804589266,
      "lm_loss": 0.6798,
      "step": 608,
      "vm_loss": 0.1338
    },
    {
      "epoch": 0.044608710804589266,
      "lm_loss": 0.3502,
      "step": 608,
      "vm_loss": 0.167
    },
    {
      "epoch": 0.044608710804589266,
      "lm_loss": 1.1415,
      "step": 608,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.044682080394728395,
      "grad_norm": 1.8498170927149784,
      "learning_rate": 9.994353815294438e-06,
      "loss": 0.9242,
      "step": 609
    },
    {
      "epoch": 0.044755449984867524,
      "grad_norm": 1.9906346230260012,
      "learning_rate": 9.994297223055016e-06,
      "loss": 1.0787,
      "step": 610
    },
    {
      "epoch": 0.04482881957500665,
      "grad_norm": 0.6273594092099986,
      "learning_rate": 9.994240348775263e-06,
      "loss": 0.6006,
      "step": 611
    },
    {
      "epoch": 0.044902189165145774,
      "grad_norm": 1.8172115450675883,
      "learning_rate": 9.99418319245839e-06,
      "loss": 1.0385,
      "step": 612
    },
    {
      "epoch": 0.0449755587552849,
      "grad_norm": 1.7544125439333749,
      "learning_rate": 9.994125754107624e-06,
      "loss": 0.9937,
      "step": 613
    },
    {
      "epoch": 0.04504892834542403,
      "grad_norm": 0.6697774992036789,
      "learning_rate": 9.994068033726208e-06,
      "loss": 0.594,
      "step": 614
    },
    {
      "epoch": 0.04512229793556316,
      "grad_norm": 1.9830712587001489,
      "learning_rate": 9.994010031317405e-06,
      "loss": 0.9759,
      "step": 615
    },
    {
      "epoch": 0.04519566752570228,
      "grad_norm": 0.6619602772548572,
      "learning_rate": 9.993951746884485e-06,
      "loss": 0.6119,
      "step": 616
    },
    {
      "epoch": 0.04519566752570228,
      "lm_loss": 1.7158,
      "step": 616,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.04519566752570228,
      "lm_loss": 0.0139,
      "step": 616,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.04519566752570228,
      "lm_loss": 0.9887,
      "step": 616,
      "vm_loss": 0.1926
    },
    {
      "epoch": 0.04519566752570228,
      "lm_loss": 0.9493,
      "step": 616,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.04519566752570228,
      "lm_loss": 0.901,
      "step": 616,
      "vm_loss": 0.1298
    },
    {
      "epoch": 0.04519566752570228,
      "lm_loss": 0.9541,
      "step": 616,
      "vm_loss": 0.14
    },
    {
      "epoch": 0.04519566752570228,
      "lm_loss": 0.5524,
      "step": 616,
      "vm_loss": 0.2798
    },
    {
      "epoch": 0.04519566752570228,
      "lm_loss": 0.1054,
      "step": 616,
      "vm_loss": 0.0399
    },
    {
      "epoch": 0.04526903711584141,
      "grad_norm": 2.1095588312640934,
      "learning_rate": 9.993893180430745e-06,
      "loss": 1.0082,
      "step": 617
    },
    {
      "epoch": 0.04534240670598054,
      "grad_norm": 0.6035921201628658,
      "learning_rate": 9.993834331959488e-06,
      "loss": 0.6054,
      "step": 618
    },
    {
      "epoch": 0.04541577629611967,
      "grad_norm": 0.6081368927062502,
      "learning_rate": 9.993775201474041e-06,
      "loss": 0.5912,
      "step": 619
    },
    {
      "epoch": 0.04548914588625879,
      "grad_norm": 1.912503719436441,
      "learning_rate": 9.993715788977741e-06,
      "loss": 0.9231,
      "step": 620
    },
    {
      "epoch": 0.04556251547639792,
      "grad_norm": 1.9458192442936875,
      "learning_rate": 9.993656094473942e-06,
      "loss": 1.027,
      "step": 621
    },
    {
      "epoch": 0.04563588506653705,
      "grad_norm": 0.6445756149591452,
      "learning_rate": 9.993596117966019e-06,
      "loss": 0.5979,
      "step": 622
    },
    {
      "epoch": 0.045709254656676176,
      "grad_norm": 0.6332865714097382,
      "learning_rate": 9.993535859457354e-06,
      "loss": 0.6002,
      "step": 623
    },
    {
      "epoch": 0.0457826242468153,
      "grad_norm": 1.8835615989909587,
      "learning_rate": 9.993475318951356e-06,
      "loss": 0.9253,
      "step": 624
    },
    {
      "epoch": 0.0457826242468153,
      "lm_loss": 1.2888,
      "step": 624,
      "vm_loss": 0.0537
    },
    {
      "epoch": 0.0457826242468153,
      "lm_loss": 0.8795,
      "step": 624,
      "vm_loss": 0.0718
    },
    {
      "epoch": 0.0457826242468153,
      "lm_loss": 0.3419,
      "step": 624,
      "vm_loss": 0.145
    },
    {
      "epoch": 0.0457826242468153,
      "lm_loss": 1.0076,
      "step": 624,
      "vm_loss": 0.1321
    },
    {
      "epoch": 0.0457826242468153,
      "lm_loss": 0.3542,
      "step": 624,
      "vm_loss": 0.1461
    },
    {
      "epoch": 0.0457826242468153,
      "lm_loss": 1.5453,
      "step": 624,
      "vm_loss": 0.1867
    },
    {
      "epoch": 0.0457826242468153,
      "lm_loss": 0.5421,
      "step": 624,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.0457826242468153,
      "lm_loss": 0.9963,
      "step": 624,
      "vm_loss": 0.2183
    },
    {
      "epoch": 0.04585599383695443,
      "grad_norm": 1.8409458857140177,
      "learning_rate": 9.99341449645144e-06,
      "loss": 0.9293,
      "step": 625
    },
    {
      "epoch": 0.045929363427093556,
      "grad_norm": 2.0181464451479316,
      "learning_rate": 9.99335339196104e-06,
      "loss": 0.906,
      "step": 626
    },
    {
      "epoch": 0.046002733017232685,
      "grad_norm": 2.1613016130112106,
      "learning_rate": 9.993292005483608e-06,
      "loss": 0.9622,
      "step": 627
    },
    {
      "epoch": 0.046076102607371806,
      "grad_norm": 1.8454312994687347,
      "learning_rate": 9.993230337022612e-06,
      "loss": 1.0719,
      "step": 628
    },
    {
      "epoch": 0.046149472197510935,
      "grad_norm": 1.9286561888407943,
      "learning_rate": 9.993168386581533e-06,
      "loss": 0.8649,
      "step": 629
    },
    {
      "epoch": 0.046222841787650064,
      "grad_norm": 0.6844266617006667,
      "learning_rate": 9.99310615416387e-06,
      "loss": 0.6026,
      "step": 630
    },
    {
      "epoch": 0.04629621137778919,
      "grad_norm": 2.100940677331969,
      "learning_rate": 9.993043639773136e-06,
      "loss": 1.0137,
      "step": 631
    },
    {
      "epoch": 0.04636958096792832,
      "grad_norm": 1.9193001322937273,
      "learning_rate": 9.992980843412863e-06,
      "loss": 0.955,
      "step": 632
    },
    {
      "epoch": 0.04636958096792832,
      "lm_loss": 0.8249,
      "step": 632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04636958096792832,
      "lm_loss": 0.5513,
      "step": 632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04636958096792832,
      "lm_loss": 0.5396,
      "step": 632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04636958096792832,
      "lm_loss": 0.5315,
      "step": 632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04636958096792832,
      "lm_loss": 0.4432,
      "step": 632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04636958096792832,
      "lm_loss": 0.5868,
      "step": 632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04636958096792832,
      "lm_loss": 0.539,
      "step": 632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04636958096792832,
      "lm_loss": 0.4806,
      "step": 632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04644295055806744,
      "grad_norm": 0.5991638871254488,
      "learning_rate": 9.992917765086598e-06,
      "loss": 0.5938,
      "step": 633
    },
    {
      "epoch": 0.04651632014820657,
      "grad_norm": 2.2975069880343244,
      "learning_rate": 9.992854404797902e-06,
      "loss": 1.062,
      "step": 634
    },
    {
      "epoch": 0.0465896897383457,
      "grad_norm": 1.8732958357061937,
      "learning_rate": 9.992790762550353e-06,
      "loss": 1.0117,
      "step": 635
    },
    {
      "epoch": 0.04666305932848483,
      "grad_norm": 1.9037318342969878,
      "learning_rate": 9.992726838347544e-06,
      "loss": 1.0254,
      "step": 636
    },
    {
      "epoch": 0.04673642891862395,
      "grad_norm": 0.643047929384566,
      "learning_rate": 9.992662632193087e-06,
      "loss": 0.5873,
      "step": 637
    },
    {
      "epoch": 0.04680979850876308,
      "grad_norm": 2.1985904074618015,
      "learning_rate": 9.992598144090607e-06,
      "loss": 1.0163,
      "step": 638
    },
    {
      "epoch": 0.04688316809890221,
      "grad_norm": 1.7848935915208597,
      "learning_rate": 9.992533374043747e-06,
      "loss": 0.956,
      "step": 639
    },
    {
      "epoch": 0.04695653768904134,
      "grad_norm": 1.8479803854315193,
      "learning_rate": 9.99246832205616e-06,
      "loss": 0.9168,
      "step": 640
    },
    {
      "epoch": 0.04695653768904134,
      "lm_loss": 0.8742,
      "step": 640,
      "vm_loss": 0.0452
    },
    {
      "epoch": 0.04695653768904134,
      "lm_loss": 1.6324,
      "step": 640,
      "vm_loss": 0.1843
    },
    {
      "epoch": 0.04695653768904134,
      "lm_loss": 1.1803,
      "step": 640,
      "vm_loss": 0.1787
    },
    {
      "epoch": 0.04695653768904134,
      "lm_loss": 1.9275,
      "step": 640,
      "vm_loss": 0.1387
    },
    {
      "epoch": 0.04695653768904134,
      "lm_loss": 0.67,
      "step": 640,
      "vm_loss": 0.1852
    },
    {
      "epoch": 0.04695653768904134,
      "lm_loss": 0.984,
      "step": 640,
      "vm_loss": 0.1698
    },
    {
      "epoch": 0.04695653768904134,
      "lm_loss": 0.8562,
      "step": 640,
      "vm_loss": 0.0852
    },
    {
      "epoch": 0.04695653768904134,
      "lm_loss": 0.3244,
      "step": 640,
      "vm_loss": 0.0569
    },
    {
      "epoch": 0.04702990727918046,
      "grad_norm": 1.6618957142415947,
      "learning_rate": 9.992402988131526e-06,
      "loss": 1.0142,
      "step": 641
    },
    {
      "epoch": 0.04710327686931959,
      "grad_norm": 1.80999623129112,
      "learning_rate": 9.992337372273534e-06,
      "loss": 0.9028,
      "step": 642
    },
    {
      "epoch": 0.04717664645945872,
      "grad_norm": 1.954886667198129,
      "learning_rate": 9.992271474485883e-06,
      "loss": 0.8955,
      "step": 643
    },
    {
      "epoch": 0.047250016049597846,
      "grad_norm": 1.7998211993117619,
      "learning_rate": 9.992205294772302e-06,
      "loss": 0.9776,
      "step": 644
    },
    {
      "epoch": 0.04732338563973697,
      "grad_norm": 1.9507313349099824,
      "learning_rate": 9.992138833136522e-06,
      "loss": 1.0038,
      "step": 645
    },
    {
      "epoch": 0.047396755229876096,
      "grad_norm": 0.7384728056593868,
      "learning_rate": 9.992072089582302e-06,
      "loss": 0.6171,
      "step": 646
    },
    {
      "epoch": 0.047470124820015225,
      "grad_norm": 2.3345488553781166,
      "learning_rate": 9.992005064113407e-06,
      "loss": 1.0548,
      "step": 647
    },
    {
      "epoch": 0.047543494410154354,
      "grad_norm": 1.9619744622660236,
      "learning_rate": 9.991937756733622e-06,
      "loss": 0.8905,
      "step": 648
    },
    {
      "epoch": 0.047543494410154354,
      "lm_loss": 1.559,
      "step": 648,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.047543494410154354,
      "lm_loss": 0.8013,
      "step": 648,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.047543494410154354,
      "lm_loss": 0.4428,
      "step": 648,
      "vm_loss": 0.2346
    },
    {
      "epoch": 0.047543494410154354,
      "lm_loss": 1.0791,
      "step": 648,
      "vm_loss": 0.1673
    },
    {
      "epoch": 0.047543494410154354,
      "lm_loss": 0.4056,
      "step": 648,
      "vm_loss": 0.1212
    },
    {
      "epoch": 0.047543494410154354,
      "lm_loss": 1.0494,
      "step": 648,
      "vm_loss": 0.2052
    },
    {
      "epoch": 0.047543494410154354,
      "lm_loss": 0.4001,
      "step": 648,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.047543494410154354,
      "lm_loss": 0.8198,
      "step": 648,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.047616864000293475,
      "grad_norm": 1.7699215576120286,
      "learning_rate": 9.991870167446751e-06,
      "loss": 0.9073,
      "step": 649
    },
    {
      "epoch": 0.047690233590432604,
      "grad_norm": 1.918724363783821,
      "learning_rate": 9.991802296256609e-06,
      "loss": 0.9329,
      "step": 650
    },
    {
      "epoch": 0.04776360318057173,
      "grad_norm": 2.2181671411117487,
      "learning_rate": 9.991734143167031e-06,
      "loss": 0.9312,
      "step": 651
    },
    {
      "epoch": 0.04783697277071086,
      "grad_norm": 1.7505411879881305,
      "learning_rate": 9.991665708181864e-06,
      "loss": 0.9887,
      "step": 652
    },
    {
      "epoch": 0.04791034236084998,
      "grad_norm": 0.7221160785940747,
      "learning_rate": 9.99159699130497e-06,
      "loss": 0.6117,
      "step": 653
    },
    {
      "epoch": 0.04798371195098911,
      "grad_norm": 0.6751612780342345,
      "learning_rate": 9.991527992540236e-06,
      "loss": 0.613,
      "step": 654
    },
    {
      "epoch": 0.04805708154112824,
      "grad_norm": 1.7608988589839554,
      "learning_rate": 9.991458711891554e-06,
      "loss": 1.0812,
      "step": 655
    },
    {
      "epoch": 0.04813045113126737,
      "grad_norm": 0.6017475925062626,
      "learning_rate": 9.991389149362836e-06,
      "loss": 0.6181,
      "step": 656
    },
    {
      "epoch": 0.04813045113126737,
      "lm_loss": 0.4809,
      "step": 656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04813045113126737,
      "lm_loss": 0.5259,
      "step": 656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04813045113126737,
      "lm_loss": 0.5672,
      "step": 656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04813045113126737,
      "lm_loss": 0.8195,
      "step": 656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04813045113126737,
      "lm_loss": 0.5707,
      "step": 656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04813045113126737,
      "lm_loss": 0.6957,
      "step": 656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04813045113126737,
      "lm_loss": 0.5168,
      "step": 656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.04813045113126737,
      "lm_loss": 0.5486,
      "step": 656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0482038207214065,
      "grad_norm": 0.671035921665103,
      "learning_rate": 9.991319304958013e-06,
      "loss": 0.6312,
      "step": 657
    },
    {
      "epoch": 0.04827719031154562,
      "grad_norm": 2.058413994787591,
      "learning_rate": 9.991249178681027e-06,
      "loss": 0.8899,
      "step": 658
    },
    {
      "epoch": 0.04835055990168475,
      "grad_norm": 1.715149111254405,
      "learning_rate": 9.99117877053584e-06,
      "loss": 0.9204,
      "step": 659
    },
    {
      "epoch": 0.04842392949182388,
      "grad_norm": 2.084525995519766,
      "learning_rate": 9.991108080526426e-06,
      "loss": 0.9343,
      "step": 660
    },
    {
      "epoch": 0.048497299081963006,
      "grad_norm": 0.7813610559020299,
      "learning_rate": 9.991037108656781e-06,
      "loss": 0.624,
      "step": 661
    },
    {
      "epoch": 0.04857066867210213,
      "grad_norm": 0.7144056219326685,
      "learning_rate": 9.990965854930908e-06,
      "loss": 0.5915,
      "step": 662
    },
    {
      "epoch": 0.04864403826224126,
      "grad_norm": 2.1358275014847528,
      "learning_rate": 9.990894319352836e-06,
      "loss": 0.9595,
      "step": 663
    },
    {
      "epoch": 0.048717407852380386,
      "grad_norm": 2.2835725471652224,
      "learning_rate": 9.9908225019266e-06,
      "loss": 0.9424,
      "step": 664
    },
    {
      "epoch": 0.048717407852380386,
      "lm_loss": 2.118,
      "step": 664,
      "vm_loss": 0.0834
    },
    {
      "epoch": 0.048717407852380386,
      "lm_loss": 1.51,
      "step": 664,
      "vm_loss": 0.1338
    },
    {
      "epoch": 0.048717407852380386,
      "lm_loss": 1.1531,
      "step": 664,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.048717407852380386,
      "lm_loss": 0.8127,
      "step": 664,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.048717407852380386,
      "lm_loss": 1.1054,
      "step": 664,
      "vm_loss": 0.1581
    },
    {
      "epoch": 0.048717407852380386,
      "lm_loss": 1.8857,
      "step": 664,
      "vm_loss": 0.0883
    },
    {
      "epoch": 0.048717407852380386,
      "lm_loss": 0.1416,
      "step": 664,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.048717407852380386,
      "lm_loss": 0.5389,
      "step": 664,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.048790777442519515,
      "grad_norm": 1.82016339361743,
      "learning_rate": 9.990750402656257e-06,
      "loss": 0.8944,
      "step": 665
    },
    {
      "epoch": 0.048864147032658636,
      "grad_norm": 1.7626327252866936,
      "learning_rate": 9.990678021545881e-06,
      "loss": 0.9374,
      "step": 666
    },
    {
      "epoch": 0.048937516622797765,
      "grad_norm": 1.932413445191641,
      "learning_rate": 9.990605358599556e-06,
      "loss": 0.8872,
      "step": 667
    },
    {
      "epoch": 0.049010886212936894,
      "grad_norm": 1.8292344742134974,
      "learning_rate": 9.99053241382139e-06,
      "loss": 1.0725,
      "step": 668
    },
    {
      "epoch": 0.04908425580307602,
      "grad_norm": 1.9167642950359012,
      "learning_rate": 9.990459187215498e-06,
      "loss": 1.1015,
      "step": 669
    },
    {
      "epoch": 0.049157625393215144,
      "grad_norm": 1.9495556129891778,
      "learning_rate": 9.990385678786018e-06,
      "loss": 0.9287,
      "step": 670
    },
    {
      "epoch": 0.04923099498335427,
      "grad_norm": 1.8821817886869963,
      "learning_rate": 9.9903118885371e-06,
      "loss": 1.0343,
      "step": 671
    },
    {
      "epoch": 0.0493043645734934,
      "grad_norm": 1.9749074973562308,
      "learning_rate": 9.990237816472909e-06,
      "loss": 0.9945,
      "step": 672
    },
    {
      "epoch": 0.0493043645734934,
      "lm_loss": 0.8587,
      "step": 672,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.0493043645734934,
      "lm_loss": 1.4108,
      "step": 672,
      "vm_loss": 0.062
    },
    {
      "epoch": 0.0493043645734934,
      "lm_loss": 1.1808,
      "step": 672,
      "vm_loss": 0.2136
    },
    {
      "epoch": 0.0493043645734934,
      "lm_loss": 1.0012,
      "step": 672,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.0493043645734934,
      "lm_loss": 0.9626,
      "step": 672,
      "vm_loss": 0.1593
    },
    {
      "epoch": 0.0493043645734934,
      "lm_loss": 1.165,
      "step": 672,
      "vm_loss": 0.1161
    },
    {
      "epoch": 0.0493043645734934,
      "lm_loss": 0.7723,
      "step": 672,
      "vm_loss": 0.2446
    },
    {
      "epoch": 0.0493043645734934,
      "lm_loss": 0.7804,
      "step": 672,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.04937773416363253,
      "grad_norm": 2.0067205608463903,
      "learning_rate": 9.990163462597634e-06,
      "loss": 1.0829,
      "step": 673
    },
    {
      "epoch": 0.04945110375377165,
      "grad_norm": 1.7923397749997962,
      "learning_rate": 9.99008882691547e-06,
      "loss": 0.9431,
      "step": 674
    },
    {
      "epoch": 0.04952447334391078,
      "grad_norm": 1.983394943952618,
      "learning_rate": 9.990013909430629e-06,
      "loss": 0.9029,
      "step": 675
    },
    {
      "epoch": 0.04959784293404991,
      "grad_norm": 1.3547339806710121,
      "learning_rate": 9.989938710147347e-06,
      "loss": 0.6043,
      "step": 676
    },
    {
      "epoch": 0.04967121252418904,
      "grad_norm": 1.7920036863558986,
      "learning_rate": 9.989863229069869e-06,
      "loss": 0.9786,
      "step": 677
    },
    {
      "epoch": 0.04974458211432816,
      "grad_norm": 1.7297513147139585,
      "learning_rate": 9.989787466202457e-06,
      "loss": 0.8284,
      "step": 678
    },
    {
      "epoch": 0.04981795170446729,
      "grad_norm": 0.6900432194750661,
      "learning_rate": 9.989711421549389e-06,
      "loss": 0.6231,
      "step": 679
    },
    {
      "epoch": 0.04989132129460642,
      "grad_norm": 1.9490684139791097,
      "learning_rate": 9.989635095114962e-06,
      "loss": 0.969,
      "step": 680
    },
    {
      "epoch": 0.04989132129460642,
      "lm_loss": 0.9076,
      "step": 680,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.04989132129460642,
      "lm_loss": 0.5384,
      "step": 680,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.04989132129460642,
      "lm_loss": 0.4717,
      "step": 680,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.04989132129460642,
      "lm_loss": 0.391,
      "step": 680,
      "vm_loss": 0.0588
    },
    {
      "epoch": 0.04989132129460642,
      "lm_loss": 0.6717,
      "step": 680,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.04989132129460642,
      "lm_loss": 0.7593,
      "step": 680,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.04989132129460642,
      "lm_loss": 0.6175,
      "step": 680,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.04989132129460642,
      "lm_loss": 0.905,
      "step": 680,
      "vm_loss": 0.0806
    },
    {
      "epoch": 0.04996469088474555,
      "grad_norm": 1.8964931613401106,
      "learning_rate": 9.989558486903483e-06,
      "loss": 0.9105,
      "step": 681
    },
    {
      "epoch": 0.050038060474884676,
      "grad_norm": 2.0064464506871675,
      "learning_rate": 9.989481596919281e-06,
      "loss": 1.0481,
      "step": 682
    },
    {
      "epoch": 0.0501114300650238,
      "grad_norm": 1.0046013520158539,
      "learning_rate": 9.989404425166696e-06,
      "loss": 0.626,
      "step": 683
    },
    {
      "epoch": 0.050184799655162926,
      "grad_norm": 1.5645284854976054,
      "learning_rate": 9.989326971650087e-06,
      "loss": 0.893,
      "step": 684
    },
    {
      "epoch": 0.050258169245302055,
      "grad_norm": 0.7872469511829603,
      "learning_rate": 9.989249236373827e-06,
      "loss": 0.6036,
      "step": 685
    },
    {
      "epoch": 0.050331538835441184,
      "grad_norm": 1.931145817029639,
      "learning_rate": 9.989171219342309e-06,
      "loss": 0.9802,
      "step": 686
    },
    {
      "epoch": 0.050404908425580305,
      "grad_norm": 1.9356923399729142,
      "learning_rate": 9.989092920559937e-06,
      "loss": 0.9726,
      "step": 687
    },
    {
      "epoch": 0.050478278015719434,
      "grad_norm": 1.8805478189840936,
      "learning_rate": 9.989014340031131e-06,
      "loss": 0.9281,
      "step": 688
    },
    {
      "epoch": 0.050478278015719434,
      "lm_loss": 0.1835,
      "step": 688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.050478278015719434,
      "lm_loss": 0.3745,
      "step": 688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.050478278015719434,
      "lm_loss": 0.7165,
      "step": 688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.050478278015719434,
      "lm_loss": 0.5519,
      "step": 688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.050478278015719434,
      "lm_loss": 0.6157,
      "step": 688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.050478278015719434,
      "lm_loss": 0.8087,
      "step": 688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.050478278015719434,
      "lm_loss": 0.5955,
      "step": 688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.050478278015719434,
      "lm_loss": 0.6293,
      "step": 688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05055164760585856,
      "grad_norm": 0.6487917064845105,
      "learning_rate": 9.98893547776033e-06,
      "loss": 0.5809,
      "step": 689
    },
    {
      "epoch": 0.05062501719599769,
      "grad_norm": 1.8501629444764134,
      "learning_rate": 9.98885633375199e-06,
      "loss": 1.0213,
      "step": 690
    },
    {
      "epoch": 0.05069838678613681,
      "grad_norm": 1.9397939901525192,
      "learning_rate": 9.988776908010576e-06,
      "loss": 1.0564,
      "step": 691
    },
    {
      "epoch": 0.05077175637627594,
      "grad_norm": 1.944545541976914,
      "learning_rate": 9.988697200540576e-06,
      "loss": 0.9301,
      "step": 692
    },
    {
      "epoch": 0.05084512596641507,
      "grad_norm": 1.6985636648816254,
      "learning_rate": 9.98861721134649e-06,
      "loss": 0.9763,
      "step": 693
    },
    {
      "epoch": 0.0509184955565542,
      "grad_norm": 1.9889786310238453,
      "learning_rate": 9.988536940432839e-06,
      "loss": 0.9449,
      "step": 694
    },
    {
      "epoch": 0.05099186514669332,
      "grad_norm": 1.7757697262927605,
      "learning_rate": 9.988456387804149e-06,
      "loss": 0.9586,
      "step": 695
    },
    {
      "epoch": 0.05106523473683245,
      "grad_norm": 0.7240792577906359,
      "learning_rate": 9.988375553464976e-06,
      "loss": 0.5888,
      "step": 696
    },
    {
      "epoch": 0.05106523473683245,
      "lm_loss": 0.394,
      "step": 696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05106523473683245,
      "lm_loss": 0.7411,
      "step": 696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05106523473683245,
      "lm_loss": 0.6462,
      "step": 696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05106523473683245,
      "lm_loss": 0.6884,
      "step": 696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05106523473683245,
      "lm_loss": 0.4419,
      "step": 696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05106523473683245,
      "lm_loss": 0.7632,
      "step": 696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05106523473683245,
      "lm_loss": 0.6391,
      "step": 696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05106523473683245,
      "lm_loss": 0.6303,
      "step": 696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05113860432697158,
      "grad_norm": 0.7218851156186707,
      "learning_rate": 9.988294437419882e-06,
      "loss": 0.616,
      "step": 697
    },
    {
      "epoch": 0.05121197391711071,
      "grad_norm": 0.6338532857108443,
      "learning_rate": 9.988213039673445e-06,
      "loss": 0.6032,
      "step": 698
    },
    {
      "epoch": 0.05128534350724983,
      "grad_norm": 0.5832986023843331,
      "learning_rate": 9.988131360230266e-06,
      "loss": 0.5933,
      "step": 699
    },
    {
      "epoch": 0.05135871309738896,
      "grad_norm": 2.6577947644405464,
      "learning_rate": 9.988049399094958e-06,
      "loss": 1.0784,
      "step": 700
    },
    {
      "epoch": 0.05143208268752809,
      "grad_norm": 0.7102257533638603,
      "learning_rate": 9.987967156272146e-06,
      "loss": 0.6279,
      "step": 701
    },
    {
      "epoch": 0.051505452277667216,
      "grad_norm": 2.0317608884626157,
      "learning_rate": 9.987884631766475e-06,
      "loss": 1.0811,
      "step": 702
    },
    {
      "epoch": 0.05157882186780634,
      "grad_norm": 1.9562023490810976,
      "learning_rate": 9.987801825582608e-06,
      "loss": 0.8943,
      "step": 703
    },
    {
      "epoch": 0.051652191457945466,
      "grad_norm": 1.9924984959338945,
      "learning_rate": 9.987718737725221e-06,
      "loss": 0.9376,
      "step": 704
    },
    {
      "epoch": 0.051652191457945466,
      "lm_loss": 0.8141,
      "step": 704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.051652191457945466,
      "lm_loss": 0.6782,
      "step": 704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.051652191457945466,
      "lm_loss": 0.5915,
      "step": 704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.051652191457945466,
      "lm_loss": 0.5282,
      "step": 704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.051652191457945466,
      "lm_loss": 0.4926,
      "step": 704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.051652191457945466,
      "lm_loss": 0.6912,
      "step": 704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.051652191457945466,
      "lm_loss": 0.6109,
      "step": 704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.051652191457945466,
      "lm_loss": 0.5733,
      "step": 704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.051725561048084595,
      "grad_norm": 0.7799303956987139,
      "learning_rate": 9.987635368199003e-06,
      "loss": 0.6152,
      "step": 705
    },
    {
      "epoch": 0.051798930638223724,
      "grad_norm": 0.6915051250743525,
      "learning_rate": 9.987551717008665e-06,
      "loss": 0.6043,
      "step": 706
    },
    {
      "epoch": 0.05187230022836285,
      "grad_norm": 0.65550592000624,
      "learning_rate": 9.98746778415893e-06,
      "loss": 0.6075,
      "step": 707
    },
    {
      "epoch": 0.051945669818501974,
      "grad_norm": 0.6057233558932524,
      "learning_rate": 9.987383569654538e-06,
      "loss": 0.5762,
      "step": 708
    },
    {
      "epoch": 0.0520190394086411,
      "grad_norm": 0.6550562716564069,
      "learning_rate": 9.987299073500245e-06,
      "loss": 0.607,
      "step": 709
    },
    {
      "epoch": 0.05209240899878023,
      "grad_norm": 3.0614741184248953,
      "learning_rate": 9.987214295700823e-06,
      "loss": 0.9929,
      "step": 710
    },
    {
      "epoch": 0.05216577858891936,
      "grad_norm": 0.7008675343485786,
      "learning_rate": 9.987129236261057e-06,
      "loss": 0.5973,
      "step": 711
    },
    {
      "epoch": 0.05223914817905848,
      "grad_norm": 1.920615301470746,
      "learning_rate": 9.987043895185754e-06,
      "loss": 0.8877,
      "step": 712
    },
    {
      "epoch": 0.05223914817905848,
      "lm_loss": 0.4556,
      "step": 712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05223914817905848,
      "lm_loss": 0.5889,
      "step": 712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05223914817905848,
      "lm_loss": 0.6762,
      "step": 712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05223914817905848,
      "lm_loss": 0.2248,
      "step": 712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05223914817905848,
      "lm_loss": 0.639,
      "step": 712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05223914817905848,
      "lm_loss": 0.4541,
      "step": 712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05223914817905848,
      "lm_loss": 0.4771,
      "step": 712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05223914817905848,
      "lm_loss": 0.7312,
      "step": 712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05231251776919761,
      "grad_norm": 0.6893342474493583,
      "learning_rate": 9.986958272479733e-06,
      "loss": 0.5958,
      "step": 713
    },
    {
      "epoch": 0.05238588735933674,
      "grad_norm": 1.8684091025104173,
      "learning_rate": 9.986872368147826e-06,
      "loss": 0.9733,
      "step": 714
    },
    {
      "epoch": 0.05245925694947587,
      "grad_norm": 2.348388452612711,
      "learning_rate": 9.986786182194887e-06,
      "loss": 1.0001,
      "step": 715
    },
    {
      "epoch": 0.05253262653961499,
      "grad_norm": 1.827986797597365,
      "learning_rate": 9.986699714625785e-06,
      "loss": 0.9426,
      "step": 716
    },
    {
      "epoch": 0.05260599612975412,
      "grad_norm": 1.8293358577114713,
      "learning_rate": 9.986612965445398e-06,
      "loss": 0.8965,
      "step": 717
    },
    {
      "epoch": 0.05267936571989325,
      "grad_norm": 0.6647345272331046,
      "learning_rate": 9.986525934658631e-06,
      "loss": 0.5902,
      "step": 718
    },
    {
      "epoch": 0.05275273531003238,
      "grad_norm": 1.8075186273002066,
      "learning_rate": 9.986438622270392e-06,
      "loss": 0.9543,
      "step": 719
    },
    {
      "epoch": 0.0528261049001715,
      "grad_norm": 2.1611739634005582,
      "learning_rate": 9.986351028285617e-06,
      "loss": 0.9293,
      "step": 720
    },
    {
      "epoch": 0.0528261049001715,
      "lm_loss": 0.7369,
      "step": 720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0528261049001715,
      "lm_loss": 0.6917,
      "step": 720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0528261049001715,
      "lm_loss": 0.5986,
      "step": 720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0528261049001715,
      "lm_loss": 0.7388,
      "step": 720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0528261049001715,
      "lm_loss": 0.5729,
      "step": 720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0528261049001715,
      "lm_loss": 0.6526,
      "step": 720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0528261049001715,
      "lm_loss": 0.6996,
      "step": 720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0528261049001715,
      "lm_loss": 0.5357,
      "step": 720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05289947449031063,
      "grad_norm": 0.7294368315247728,
      "learning_rate": 9.986263152709251e-06,
      "loss": 0.6131,
      "step": 721
    },
    {
      "epoch": 0.052972844080449756,
      "grad_norm": 0.7390297310087122,
      "learning_rate": 9.986174995546256e-06,
      "loss": 0.5977,
      "step": 722
    },
    {
      "epoch": 0.053046213670588885,
      "grad_norm": 0.6826838198894006,
      "learning_rate": 9.986086556801612e-06,
      "loss": 0.6036,
      "step": 723
    },
    {
      "epoch": 0.05311958326072801,
      "grad_norm": 2.2621802185362783,
      "learning_rate": 9.985997836480312e-06,
      "loss": 0.9596,
      "step": 724
    },
    {
      "epoch": 0.053192952850867135,
      "grad_norm": 2.2115051396419325,
      "learning_rate": 9.985908834587366e-06,
      "loss": 0.976,
      "step": 725
    },
    {
      "epoch": 0.053266322441006264,
      "grad_norm": 1.7854471732373545,
      "learning_rate": 9.9858195511278e-06,
      "loss": 1.0309,
      "step": 726
    },
    {
      "epoch": 0.05333969203114539,
      "grad_norm": 2.1611336048694656,
      "learning_rate": 9.985729986106658e-06,
      "loss": 0.9389,
      "step": 727
    },
    {
      "epoch": 0.053413061621284515,
      "grad_norm": 0.8925036392363183,
      "learning_rate": 9.985640139528996e-06,
      "loss": 0.6037,
      "step": 728
    },
    {
      "epoch": 0.053413061621284515,
      "lm_loss": 0.6281,
      "step": 728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.053413061621284515,
      "lm_loss": 0.5917,
      "step": 728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.053413061621284515,
      "lm_loss": 0.645,
      "step": 728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.053413061621284515,
      "lm_loss": 0.8934,
      "step": 728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.053413061621284515,
      "lm_loss": 0.6142,
      "step": 728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.053413061621284515,
      "lm_loss": 0.4205,
      "step": 728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.053413061621284515,
      "lm_loss": 0.7944,
      "step": 728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.053413061621284515,
      "lm_loss": 0.6718,
      "step": 728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05348643121142364,
      "grad_norm": 0.8311197434188924,
      "learning_rate": 9.985550011399889e-06,
      "loss": 0.5967,
      "step": 729
    },
    {
      "epoch": 0.05355980080156277,
      "grad_norm": 0.6428004453363372,
      "learning_rate": 9.985459601724427e-06,
      "loss": 0.6207,
      "step": 730
    },
    {
      "epoch": 0.0536331703917019,
      "grad_norm": 0.6182715290777449,
      "learning_rate": 9.985368910507713e-06,
      "loss": 0.6084,
      "step": 731
    },
    {
      "epoch": 0.05370653998184103,
      "grad_norm": 2.308991524964361,
      "learning_rate": 9.985277937754873e-06,
      "loss": 0.9491,
      "step": 732
    },
    {
      "epoch": 0.05377990957198015,
      "grad_norm": 2.1619588305465824,
      "learning_rate": 9.985186683471042e-06,
      "loss": 0.9166,
      "step": 733
    },
    {
      "epoch": 0.05385327916211928,
      "grad_norm": 0.8967314849926603,
      "learning_rate": 9.985095147661372e-06,
      "loss": 0.6042,
      "step": 734
    },
    {
      "epoch": 0.05392664875225841,
      "grad_norm": 2.150461003079083,
      "learning_rate": 9.985003330331033e-06,
      "loss": 0.98,
      "step": 735
    },
    {
      "epoch": 0.05400001834239754,
      "grad_norm": 1.7438157944451322,
      "learning_rate": 9.984911231485212e-06,
      "loss": 0.8971,
      "step": 736
    },
    {
      "epoch": 0.05400001834239754,
      "lm_loss": 1.652,
      "step": 736,
      "vm_loss": 0.0608
    },
    {
      "epoch": 0.05400001834239754,
      "lm_loss": 0.524,
      "step": 736,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.05400001834239754,
      "lm_loss": 0.9075,
      "step": 736,
      "vm_loss": 0.1352
    },
    {
      "epoch": 0.05400001834239754,
      "lm_loss": 0.3029,
      "step": 736,
      "vm_loss": 0.1204
    },
    {
      "epoch": 0.05400001834239754,
      "lm_loss": 0.2505,
      "step": 736,
      "vm_loss": 0.2241
    },
    {
      "epoch": 0.05400001834239754,
      "lm_loss": 0.5719,
      "step": 736,
      "vm_loss": 0.2119
    },
    {
      "epoch": 0.05400001834239754,
      "lm_loss": 1.0944,
      "step": 736,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.05400001834239754,
      "lm_loss": 0.4612,
      "step": 736,
      "vm_loss": 0.24
    },
    {
      "epoch": 0.05407338793253666,
      "grad_norm": 1.8361201763105623,
      "learning_rate": 9.984818851129108e-06,
      "loss": 0.9261,
      "step": 737
    },
    {
      "epoch": 0.05414675752267579,
      "grad_norm": 0.7756636955172441,
      "learning_rate": 9.984726189267939e-06,
      "loss": 0.6016,
      "step": 738
    },
    {
      "epoch": 0.05422012711281492,
      "grad_norm": 2.1183910878429044,
      "learning_rate": 9.984633245906939e-06,
      "loss": 0.9735,
      "step": 739
    },
    {
      "epoch": 0.054293496702954046,
      "grad_norm": 1.967004539827843,
      "learning_rate": 9.984540021051354e-06,
      "loss": 1.0382,
      "step": 740
    },
    {
      "epoch": 0.05436686629309317,
      "grad_norm": 0.642576624004984,
      "learning_rate": 9.984446514706451e-06,
      "loss": 0.6199,
      "step": 741
    },
    {
      "epoch": 0.054440235883232296,
      "grad_norm": 2.085420542304351,
      "learning_rate": 9.98435272687751e-06,
      "loss": 0.9324,
      "step": 742
    },
    {
      "epoch": 0.054513605473371425,
      "grad_norm": 0.639904273178802,
      "learning_rate": 9.984258657569824e-06,
      "loss": 0.6095,
      "step": 743
    },
    {
      "epoch": 0.054586975063510554,
      "grad_norm": 2.174258654243695,
      "learning_rate": 9.984164306788711e-06,
      "loss": 0.9092,
      "step": 744
    },
    {
      "epoch": 0.054586975063510554,
      "lm_loss": 1.5452,
      "step": 744,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.054586975063510554,
      "lm_loss": 1.1198,
      "step": 744,
      "vm_loss": 0.1763
    },
    {
      "epoch": 0.054586975063510554,
      "lm_loss": 0.6429,
      "step": 744,
      "vm_loss": 0.1905
    },
    {
      "epoch": 0.054586975063510554,
      "lm_loss": 0.6281,
      "step": 744,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.054586975063510554,
      "lm_loss": 0.0506,
      "step": 744,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.054586975063510554,
      "lm_loss": 1.4778,
      "step": 744,
      "vm_loss": 0.1496
    },
    {
      "epoch": 0.054586975063510554,
      "lm_loss": 0.761,
      "step": 744,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.054586975063510554,
      "lm_loss": 1.0025,
      "step": 744,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.054660344653649676,
      "grad_norm": 2.0609026983421286,
      "learning_rate": 9.984069674539495e-06,
      "loss": 1.0032,
      "step": 745
    },
    {
      "epoch": 0.054733714243788804,
      "grad_norm": 1.780025516332889,
      "learning_rate": 9.983974760827521e-06,
      "loss": 1.0208,
      "step": 746
    },
    {
      "epoch": 0.05480708383392793,
      "grad_norm": 1.7985817360055605,
      "learning_rate": 9.983879565658152e-06,
      "loss": 0.9307,
      "step": 747
    },
    {
      "epoch": 0.05488045342406706,
      "grad_norm": 2.0197862990710846,
      "learning_rate": 9.983784089036761e-06,
      "loss": 0.9426,
      "step": 748
    },
    {
      "epoch": 0.054953823014206184,
      "grad_norm": 0.7044383062272871,
      "learning_rate": 9.98368833096874e-06,
      "loss": 0.5918,
      "step": 749
    },
    {
      "epoch": 0.05502719260434531,
      "grad_norm": 1.8823175044648863,
      "learning_rate": 9.983592291459496e-06,
      "loss": 0.8876,
      "step": 750
    },
    {
      "epoch": 0.05510056219448444,
      "grad_norm": 1.998073879923454,
      "learning_rate": 9.983495970514455e-06,
      "loss": 0.8922,
      "step": 751
    },
    {
      "epoch": 0.05517393178462357,
      "grad_norm": 1.6961573028698784,
      "learning_rate": 9.983399368139054e-06,
      "loss": 0.8702,
      "step": 752
    },
    {
      "epoch": 0.05517393178462357,
      "lm_loss": 1.4148,
      "step": 752,
      "vm_loss": 0.0711
    },
    {
      "epoch": 0.05517393178462357,
      "lm_loss": 0.9824,
      "step": 752,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.05517393178462357,
      "lm_loss": 0.1031,
      "step": 752,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.05517393178462357,
      "lm_loss": 1.045,
      "step": 752,
      "vm_loss": 0.1149
    },
    {
      "epoch": 0.05517393178462357,
      "lm_loss": 1.0801,
      "step": 752,
      "vm_loss": 0.2329
    },
    {
      "epoch": 0.05517393178462357,
      "lm_loss": 0.8774,
      "step": 752,
      "vm_loss": 0.1392
    },
    {
      "epoch": 0.05517393178462357,
      "lm_loss": 1.0694,
      "step": 752,
      "vm_loss": 0.1799
    },
    {
      "epoch": 0.05517393178462357,
      "lm_loss": 1.1321,
      "step": 752,
      "vm_loss": 0.1636
    },
    {
      "epoch": 0.0552473013747627,
      "grad_norm": 1.861505594135888,
      "learning_rate": 9.98330248433875e-06,
      "loss": 1.0654,
      "step": 753
    },
    {
      "epoch": 0.05532067096490182,
      "grad_norm": 2.0820874354969066,
      "learning_rate": 9.983205319119016e-06,
      "loss": 1.018,
      "step": 754
    },
    {
      "epoch": 0.05539404055504095,
      "grad_norm": 1.7525142829452,
      "learning_rate": 9.983107872485334e-06,
      "loss": 0.8348,
      "step": 755
    },
    {
      "epoch": 0.05546741014518008,
      "grad_norm": 1.8590598881028342,
      "learning_rate": 9.98301014444321e-06,
      "loss": 0.919,
      "step": 756
    },
    {
      "epoch": 0.05554077973531921,
      "grad_norm": 0.7089438214353508,
      "learning_rate": 9.982912134998165e-06,
      "loss": 0.6161,
      "step": 757
    },
    {
      "epoch": 0.05561414932545833,
      "grad_norm": 1.7535579100143772,
      "learning_rate": 9.98281384415573e-06,
      "loss": 0.8705,
      "step": 758
    },
    {
      "epoch": 0.05568751891559746,
      "grad_norm": 0.6197275502299507,
      "learning_rate": 9.982715271921457e-06,
      "loss": 0.5922,
      "step": 759
    },
    {
      "epoch": 0.055760888505736586,
      "grad_norm": 1.8549609363619988,
      "learning_rate": 9.982616418300914e-06,
      "loss": 0.8686,
      "step": 760
    },
    {
      "epoch": 0.055760888505736586,
      "lm_loss": 1.4265,
      "step": 760,
      "vm_loss": 0.0619
    },
    {
      "epoch": 0.055760888505736586,
      "lm_loss": 0.9821,
      "step": 760,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.055760888505736586,
      "lm_loss": 0.8254,
      "step": 760,
      "vm_loss": 0.0805
    },
    {
      "epoch": 0.055760888505736586,
      "lm_loss": 0.2939,
      "step": 760,
      "vm_loss": 0.1771
    },
    {
      "epoch": 0.055760888505736586,
      "lm_loss": 0.0785,
      "step": 760,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.055760888505736586,
      "lm_loss": 1.0091,
      "step": 760,
      "vm_loss": 0.0692
    },
    {
      "epoch": 0.055760888505736586,
      "lm_loss": 0.1274,
      "step": 760,
      "vm_loss": 0.1235
    },
    {
      "epoch": 0.055760888505736586,
      "lm_loss": 1.251,
      "step": 760,
      "vm_loss": 0.1396
    },
    {
      "epoch": 0.055834258095875715,
      "grad_norm": 1.9213873111774205,
      "learning_rate": 9.982517283299684e-06,
      "loss": 0.8882,
      "step": 761
    },
    {
      "epoch": 0.05590762768601484,
      "grad_norm": 0.6348933690563834,
      "learning_rate": 9.982417866923362e-06,
      "loss": 0.6103,
      "step": 762
    },
    {
      "epoch": 0.055980997276153965,
      "grad_norm": 1.7631369307272982,
      "learning_rate": 9.982318169177565e-06,
      "loss": 0.9745,
      "step": 763
    },
    {
      "epoch": 0.056054366866293094,
      "grad_norm": 1.640168861795445,
      "learning_rate": 9.982218190067922e-06,
      "loss": 0.9617,
      "step": 764
    },
    {
      "epoch": 0.05612773645643222,
      "grad_norm": 1.763075379081011,
      "learning_rate": 9.982117929600081e-06,
      "loss": 1.0963,
      "step": 765
    },
    {
      "epoch": 0.056201106046571345,
      "grad_norm": 1.8639134957917338,
      "learning_rate": 9.982017387779701e-06,
      "loss": 0.978,
      "step": 766
    },
    {
      "epoch": 0.05627447563671047,
      "grad_norm": 1.6569898718497134,
      "learning_rate": 9.981916564612463e-06,
      "loss": 0.9899,
      "step": 767
    },
    {
      "epoch": 0.0563478452268496,
      "grad_norm": 1.6925748517786097,
      "learning_rate": 9.981815460104058e-06,
      "loss": 0.93,
      "step": 768
    },
    {
      "epoch": 0.0563478452268496,
      "lm_loss": 1.5622,
      "step": 768,
      "vm_loss": 0.0476
    },
    {
      "epoch": 0.0563478452268496,
      "lm_loss": 1.5181,
      "step": 768,
      "vm_loss": 0.0789
    },
    {
      "epoch": 0.0563478452268496,
      "lm_loss": 0.844,
      "step": 768,
      "vm_loss": 0.0497
    },
    {
      "epoch": 0.0563478452268496,
      "lm_loss": 1.0815,
      "step": 768,
      "vm_loss": 0.1349
    },
    {
      "epoch": 0.0563478452268496,
      "lm_loss": 0.725,
      "step": 768,
      "vm_loss": 0.1875
    },
    {
      "epoch": 0.0563478452268496,
      "lm_loss": 0.8901,
      "step": 768,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.0563478452268496,
      "lm_loss": 0.3862,
      "step": 768,
      "vm_loss": 0.1964
    },
    {
      "epoch": 0.0563478452268496,
      "lm_loss": 1.2152,
      "step": 768,
      "vm_loss": 0.1378
    },
    {
      "epoch": 0.05642121481698873,
      "grad_norm": 2.0049732986748108,
      "learning_rate": 9.981714074260196e-06,
      "loss": 1.0066,
      "step": 769
    },
    {
      "epoch": 0.05649458440712785,
      "grad_norm": 1.8719760528745701,
      "learning_rate": 9.981612407086605e-06,
      "loss": 0.8665,
      "step": 770
    },
    {
      "epoch": 0.05656795399726698,
      "grad_norm": 1.8208302927455258,
      "learning_rate": 9.981510458589023e-06,
      "loss": 0.951,
      "step": 771
    },
    {
      "epoch": 0.05664132358740611,
      "grad_norm": 1.7048730101398488,
      "learning_rate": 9.98140822877321e-06,
      "loss": 0.903,
      "step": 772
    },
    {
      "epoch": 0.05671469317754524,
      "grad_norm": 1.820347936344509,
      "learning_rate": 9.981305717644938e-06,
      "loss": 1.0584,
      "step": 773
    },
    {
      "epoch": 0.05678806276768436,
      "grad_norm": 1.523092045195476,
      "learning_rate": 9.981202925209994e-06,
      "loss": 0.906,
      "step": 774
    },
    {
      "epoch": 0.05686143235782349,
      "grad_norm": 1.8122155968774931,
      "learning_rate": 9.981099851474187e-06,
      "loss": 0.9668,
      "step": 775
    },
    {
      "epoch": 0.05693480194796262,
      "grad_norm": 1.9086132636725412,
      "learning_rate": 9.980996496443338e-06,
      "loss": 0.9143,
      "step": 776
    },
    {
      "epoch": 0.05693480194796262,
      "lm_loss": 1.4553,
      "step": 776,
      "vm_loss": 0.0551
    },
    {
      "epoch": 0.05693480194796262,
      "lm_loss": 0.1365,
      "step": 776,
      "vm_loss": 0.0605
    },
    {
      "epoch": 0.05693480194796262,
      "lm_loss": 0.4117,
      "step": 776,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.05693480194796262,
      "lm_loss": 0.1408,
      "step": 776,
      "vm_loss": 0.1331
    },
    {
      "epoch": 0.05693480194796262,
      "lm_loss": 0.8136,
      "step": 776,
      "vm_loss": 0.1857
    },
    {
      "epoch": 0.05693480194796262,
      "lm_loss": 1.2384,
      "step": 776,
      "vm_loss": 0.224
    },
    {
      "epoch": 0.05693480194796262,
      "lm_loss": 0.6301,
      "step": 776,
      "vm_loss": 0.107
    },
    {
      "epoch": 0.05693480194796262,
      "lm_loss": 1.1251,
      "step": 776,
      "vm_loss": 0.2068
    },
    {
      "epoch": 0.05700817153810175,
      "grad_norm": 1.8367897022538382,
      "learning_rate": 9.980892860123278e-06,
      "loss": 0.9959,
      "step": 777
    },
    {
      "epoch": 0.057081541128240876,
      "grad_norm": 1.5669435332252284,
      "learning_rate": 9.980788942519864e-06,
      "loss": 0.8032,
      "step": 778
    },
    {
      "epoch": 0.05715491071838,
      "grad_norm": 1.7279643238003093,
      "learning_rate": 9.980684743638965e-06,
      "loss": 0.991,
      "step": 779
    },
    {
      "epoch": 0.057228280308519126,
      "grad_norm": 2.0446338734388303,
      "learning_rate": 9.980580263486464e-06,
      "loss": 1.0037,
      "step": 780
    },
    {
      "epoch": 0.057301649898658255,
      "grad_norm": 1.9019726041937002,
      "learning_rate": 9.98047550206826e-06,
      "loss": 0.9671,
      "step": 781
    },
    {
      "epoch": 0.057375019488797384,
      "grad_norm": 0.7690753278412942,
      "learning_rate": 9.980370459390272e-06,
      "loss": 0.622,
      "step": 782
    },
    {
      "epoch": 0.057448389078936506,
      "grad_norm": 1.8256409078830256,
      "learning_rate": 9.98026513545843e-06,
      "loss": 0.9984,
      "step": 783
    },
    {
      "epoch": 0.057521758669075634,
      "grad_norm": 0.6271503427953384,
      "learning_rate": 9.98015953027868e-06,
      "loss": 0.6008,
      "step": 784
    },
    {
      "epoch": 0.057521758669075634,
      "lm_loss": 0.526,
      "step": 784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.057521758669075634,
      "lm_loss": 0.6987,
      "step": 784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.057521758669075634,
      "lm_loss": 0.6258,
      "step": 784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.057521758669075634,
      "lm_loss": 0.5213,
      "step": 784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.057521758669075634,
      "lm_loss": 0.7211,
      "step": 784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.057521758669075634,
      "lm_loss": 0.6531,
      "step": 784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.057521758669075634,
      "lm_loss": 0.6212,
      "step": 784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.057521758669075634,
      "lm_loss": 0.4861,
      "step": 784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05759512825921476,
      "grad_norm": 0.627402809511005,
      "learning_rate": 9.98005364385699e-06,
      "loss": 0.6237,
      "step": 785
    },
    {
      "epoch": 0.05766849784935389,
      "grad_norm": 1.840187232620948,
      "learning_rate": 9.979947476199339e-06,
      "loss": 0.8367,
      "step": 786
    },
    {
      "epoch": 0.057741867439493014,
      "grad_norm": 2.0291710660555466,
      "learning_rate": 9.97984102731172e-06,
      "loss": 0.9881,
      "step": 787
    },
    {
      "epoch": 0.05781523702963214,
      "grad_norm": 2.010358602465697,
      "learning_rate": 9.979734297200146e-06,
      "loss": 0.9817,
      "step": 788
    },
    {
      "epoch": 0.05788860661977127,
      "grad_norm": 1.8346307939894853,
      "learning_rate": 9.979627285870644e-06,
      "loss": 0.8867,
      "step": 789
    },
    {
      "epoch": 0.0579619762099104,
      "grad_norm": 0.6879343112141907,
      "learning_rate": 9.97951999332926e-06,
      "loss": 0.5979,
      "step": 790
    },
    {
      "epoch": 0.05803534580004952,
      "grad_norm": 1.6443847577817696,
      "learning_rate": 9.979412419582047e-06,
      "loss": 0.9368,
      "step": 791
    },
    {
      "epoch": 0.05810871539018865,
      "grad_norm": 1.8270732648482708,
      "learning_rate": 9.979304564635084e-06,
      "loss": 0.8926,
      "step": 792
    },
    {
      "epoch": 0.05810871539018865,
      "lm_loss": 1.0849,
      "step": 792,
      "vm_loss": 0.0332
    },
    {
      "epoch": 0.05810871539018865,
      "lm_loss": 1.4074,
      "step": 792,
      "vm_loss": 0.1395
    },
    {
      "epoch": 0.05810871539018865,
      "lm_loss": 0.527,
      "step": 792,
      "vm_loss": 0.1621
    },
    {
      "epoch": 0.05810871539018865,
      "lm_loss": 0.616,
      "step": 792,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.05810871539018865,
      "lm_loss": 0.7021,
      "step": 792,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.05810871539018865,
      "lm_loss": 1.064,
      "step": 792,
      "vm_loss": 0.2174
    },
    {
      "epoch": 0.05810871539018865,
      "lm_loss": 0.1427,
      "step": 792,
      "vm_loss": 0.0944
    },
    {
      "epoch": 0.05810871539018865,
      "lm_loss": 0.4751,
      "step": 792,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.05818208498032778,
      "grad_norm": 1.6232702002278177,
      "learning_rate": 9.97919642849446e-06,
      "loss": 0.9005,
      "step": 793
    },
    {
      "epoch": 0.05825545457046691,
      "grad_norm": 1.8853347252448556,
      "learning_rate": 9.979088011166284e-06,
      "loss": 0.9315,
      "step": 794
    },
    {
      "epoch": 0.05832882416060603,
      "grad_norm": 1.773001290807959,
      "learning_rate": 9.97897931265668e-06,
      "loss": 0.9344,
      "step": 795
    },
    {
      "epoch": 0.05840219375074516,
      "grad_norm": 1.7433795504211278,
      "learning_rate": 9.978870332971782e-06,
      "loss": 0.9466,
      "step": 796
    },
    {
      "epoch": 0.05847556334088429,
      "grad_norm": 1.8586010206360104,
      "learning_rate": 9.978761072117746e-06,
      "loss": 1.0184,
      "step": 797
    },
    {
      "epoch": 0.058548932931023416,
      "grad_norm": 1.652285526806935,
      "learning_rate": 9.978651530100743e-06,
      "loss": 0.8913,
      "step": 798
    },
    {
      "epoch": 0.05862230252116254,
      "grad_norm": 1.9584742042671432,
      "learning_rate": 9.978541706926959e-06,
      "loss": 0.9402,
      "step": 799
    },
    {
      "epoch": 0.05869567211130167,
      "grad_norm": 0.7073356768343528,
      "learning_rate": 9.978431602602594e-06,
      "loss": 0.5988,
      "step": 800
    },
    {
      "epoch": 0.05869567211130167,
      "lm_loss": 1.5953,
      "step": 800,
      "vm_loss": 0.0518
    },
    {
      "epoch": 0.05869567211130167,
      "lm_loss": 0.6497,
      "step": 800,
      "vm_loss": 0.0624
    },
    {
      "epoch": 0.05869567211130167,
      "lm_loss": 0.8865,
      "step": 800,
      "vm_loss": 0.1627
    },
    {
      "epoch": 0.05869567211130167,
      "lm_loss": 0.6678,
      "step": 800,
      "vm_loss": 0.149
    },
    {
      "epoch": 0.05869567211130167,
      "lm_loss": 1.1246,
      "step": 800,
      "vm_loss": 0.151
    },
    {
      "epoch": 0.05869567211130167,
      "lm_loss": 0.8015,
      "step": 800,
      "vm_loss": 0.1405
    },
    {
      "epoch": 0.05869567211130167,
      "lm_loss": 1.0685,
      "step": 800,
      "vm_loss": 0.1635
    },
    {
      "epoch": 0.05869567211130167,
      "lm_loss": 0.7171,
      "step": 800,
      "vm_loss": 0.128
    },
    {
      "epoch": 0.058769041701440795,
      "grad_norm": 1.6778650373065493,
      "learning_rate": 9.978321217133869e-06,
      "loss": 0.8135,
      "step": 801
    },
    {
      "epoch": 0.058842411291579924,
      "grad_norm": 1.8500727492057418,
      "learning_rate": 9.978210550527017e-06,
      "loss": 0.9694,
      "step": 802
    },
    {
      "epoch": 0.05891578088171905,
      "grad_norm": 0.6132928224805714,
      "learning_rate": 9.978099602788286e-06,
      "loss": 0.5687,
      "step": 803
    },
    {
      "epoch": 0.058989150471858175,
      "grad_norm": 2.0774923379744554,
      "learning_rate": 9.977988373923942e-06,
      "loss": 0.9499,
      "step": 804
    },
    {
      "epoch": 0.0590625200619973,
      "grad_norm": 1.597599715153716,
      "learning_rate": 9.977876863940266e-06,
      "loss": 0.8421,
      "step": 805
    },
    {
      "epoch": 0.05913588965213643,
      "grad_norm": 0.5922272935141611,
      "learning_rate": 9.977765072843557e-06,
      "loss": 0.6002,
      "step": 806
    },
    {
      "epoch": 0.05920925924227556,
      "grad_norm": 1.8513943245951283,
      "learning_rate": 9.977653000640126e-06,
      "loss": 0.9475,
      "step": 807
    },
    {
      "epoch": 0.05928262883241468,
      "grad_norm": 1.6843201255581155,
      "learning_rate": 9.977540647336305e-06,
      "loss": 0.872,
      "step": 808
    },
    {
      "epoch": 0.05928262883241468,
      "lm_loss": 0.6829,
      "step": 808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05928262883241468,
      "lm_loss": 0.7306,
      "step": 808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05928262883241468,
      "lm_loss": 0.6817,
      "step": 808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05928262883241468,
      "lm_loss": 0.6907,
      "step": 808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05928262883241468,
      "lm_loss": 0.4597,
      "step": 808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05928262883241468,
      "lm_loss": 0.6697,
      "step": 808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05928262883241468,
      "lm_loss": 0.6397,
      "step": 808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05928262883241468,
      "lm_loss": 0.6124,
      "step": 808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.05935599842255381,
      "grad_norm": 0.5983793073240639,
      "learning_rate": 9.977428012938437e-06,
      "loss": 0.5974,
      "step": 809
    },
    {
      "epoch": 0.05942936801269294,
      "grad_norm": 0.5902764967722154,
      "learning_rate": 9.977315097452882e-06,
      "loss": 0.5734,
      "step": 810
    },
    {
      "epoch": 0.05950273760283207,
      "grad_norm": 0.6152870691370211,
      "learning_rate": 9.977201900886017e-06,
      "loss": 0.5983,
      "step": 811
    },
    {
      "epoch": 0.05957610719297119,
      "grad_norm": 1.951752334750438,
      "learning_rate": 9.977088423244234e-06,
      "loss": 0.9302,
      "step": 812
    },
    {
      "epoch": 0.05964947678311032,
      "grad_norm": 1.860358120498438,
      "learning_rate": 9.976974664533944e-06,
      "loss": 0.9493,
      "step": 813
    },
    {
      "epoch": 0.05972284637324945,
      "grad_norm": 1.6554356410260023,
      "learning_rate": 9.976860624761568e-06,
      "loss": 0.9408,
      "step": 814
    },
    {
      "epoch": 0.05979621596338858,
      "grad_norm": 1.741632285964485,
      "learning_rate": 9.97674630393355e-06,
      "loss": 0.9673,
      "step": 815
    },
    {
      "epoch": 0.0598695855535277,
      "grad_norm": 0.6662444779607162,
      "learning_rate": 9.976631702056342e-06,
      "loss": 0.6162,
      "step": 816
    },
    {
      "epoch": 0.0598695855535277,
      "lm_loss": 1.463,
      "step": 816,
      "vm_loss": 0.0888
    },
    {
      "epoch": 0.0598695855535277,
      "lm_loss": 1.382,
      "step": 816,
      "vm_loss": 0.176
    },
    {
      "epoch": 0.0598695855535277,
      "lm_loss": 0.302,
      "step": 816,
      "vm_loss": 0.0456
    },
    {
      "epoch": 0.0598695855535277,
      "lm_loss": 1.2595,
      "step": 816,
      "vm_loss": 0.189
    },
    {
      "epoch": 0.0598695855535277,
      "lm_loss": 1.0025,
      "step": 816,
      "vm_loss": 0.1893
    },
    {
      "epoch": 0.0598695855535277,
      "lm_loss": 0.2585,
      "step": 816,
      "vm_loss": 0.191
    },
    {
      "epoch": 0.0598695855535277,
      "lm_loss": 1.0098,
      "step": 816,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.0598695855535277,
      "lm_loss": 1.2438,
      "step": 816,
      "vm_loss": 0.1688
    },
    {
      "epoch": 0.05994295514366683,
      "grad_norm": 1.7330650024908554,
      "learning_rate": 9.976516819136414e-06,
      "loss": 1.0175,
      "step": 817
    },
    {
      "epoch": 0.060016324733805956,
      "grad_norm": 0.6388645799816154,
      "learning_rate": 9.976401655180262e-06,
      "loss": 0.6263,
      "step": 818
    },
    {
      "epoch": 0.060089694323945085,
      "grad_norm": 1.7329632564012456,
      "learning_rate": 9.976286210194382e-06,
      "loss": 0.829,
      "step": 819
    },
    {
      "epoch": 0.06016306391408421,
      "grad_norm": 0.6017523523237349,
      "learning_rate": 9.976170484185297e-06,
      "loss": 0.5901,
      "step": 820
    },
    {
      "epoch": 0.060236433504223336,
      "grad_norm": 1.8712600683938911,
      "learning_rate": 9.976054477159541e-06,
      "loss": 0.9672,
      "step": 821
    },
    {
      "epoch": 0.060309803094362464,
      "grad_norm": 2.0880950291659874,
      "learning_rate": 9.975938189123666e-06,
      "loss": 0.9691,
      "step": 822
    },
    {
      "epoch": 0.06038317268450159,
      "grad_norm": 1.6721228347217612,
      "learning_rate": 9.97582162008424e-06,
      "loss": 0.9575,
      "step": 823
    },
    {
      "epoch": 0.060456542274640715,
      "grad_norm": 1.74631741628863,
      "learning_rate": 9.975704770047843e-06,
      "loss": 0.9559,
      "step": 824
    },
    {
      "epoch": 0.060456542274640715,
      "lm_loss": 0.5778,
      "step": 824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.060456542274640715,
      "lm_loss": 0.5469,
      "step": 824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.060456542274640715,
      "lm_loss": 0.5353,
      "step": 824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.060456542274640715,
      "lm_loss": 0.5543,
      "step": 824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.060456542274640715,
      "lm_loss": 0.5914,
      "step": 824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.060456542274640715,
      "lm_loss": 0.6534,
      "step": 824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.060456542274640715,
      "lm_loss": 0.6434,
      "step": 824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.060456542274640715,
      "lm_loss": 0.7539,
      "step": 824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.060529911864779844,
      "grad_norm": 0.6966602529987008,
      "learning_rate": 9.975587639021075e-06,
      "loss": 0.6083,
      "step": 825
    },
    {
      "epoch": 0.06060328145491897,
      "grad_norm": 0.6429462647477228,
      "learning_rate": 9.975470227010553e-06,
      "loss": 0.5935,
      "step": 826
    },
    {
      "epoch": 0.0606766510450581,
      "grad_norm": 0.6333858323459024,
      "learning_rate": 9.975352534022906e-06,
      "loss": 0.607,
      "step": 827
    },
    {
      "epoch": 0.06075002063519723,
      "grad_norm": 2.012836418315821,
      "learning_rate": 9.97523456006478e-06,
      "loss": 0.9999,
      "step": 828
    },
    {
      "epoch": 0.06082339022533635,
      "grad_norm": 1.9988109348943286,
      "learning_rate": 9.975116305142836e-06,
      "loss": 0.956,
      "step": 829
    },
    {
      "epoch": 0.06089675981547548,
      "grad_norm": 1.8314225860095203,
      "learning_rate": 9.974997769263753e-06,
      "loss": 0.9715,
      "step": 830
    },
    {
      "epoch": 0.06097012940561461,
      "grad_norm": 1.9943729277530078,
      "learning_rate": 9.974878952434227e-06,
      "loss": 0.8534,
      "step": 831
    },
    {
      "epoch": 0.06104349899575374,
      "grad_norm": 1.8105200514387143,
      "learning_rate": 9.974759854660968e-06,
      "loss": 0.9356,
      "step": 832
    },
    {
      "epoch": 0.06104349899575374,
      "lm_loss": 0.5173,
      "step": 832,
      "vm_loss": 0.035
    },
    {
      "epoch": 0.06104349899575374,
      "lm_loss": 1.2116,
      "step": 832,
      "vm_loss": 0.1485
    },
    {
      "epoch": 0.06104349899575374,
      "lm_loss": 0.8373,
      "step": 832,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.06104349899575374,
      "lm_loss": 0.2637,
      "step": 832,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.06104349899575374,
      "lm_loss": 0.3521,
      "step": 832,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.06104349899575374,
      "lm_loss": 1.2136,
      "step": 832,
      "vm_loss": 0.2255
    },
    {
      "epoch": 0.06104349899575374,
      "lm_loss": 0.894,
      "step": 832,
      "vm_loss": 0.1015
    },
    {
      "epoch": 0.06104349899575374,
      "lm_loss": 1.2498,
      "step": 832,
      "vm_loss": 0.2501
    },
    {
      "epoch": 0.06111686858589286,
      "grad_norm": 1.596139193438852,
      "learning_rate": 9.974640475950697e-06,
      "loss": 0.9817,
      "step": 833
    },
    {
      "epoch": 0.06119023817603199,
      "grad_norm": 1.8528894165559373,
      "learning_rate": 9.97452081631016e-06,
      "loss": 0.8899,
      "step": 834
    },
    {
      "epoch": 0.06126360776617112,
      "grad_norm": 1.7717876762401406,
      "learning_rate": 9.974400875746115e-06,
      "loss": 0.9416,
      "step": 835
    },
    {
      "epoch": 0.061336977356310246,
      "grad_norm": 1.8429333837679382,
      "learning_rate": 9.974280654265332e-06,
      "loss": 0.9044,
      "step": 836
    },
    {
      "epoch": 0.06141034694644937,
      "grad_norm": 1.6361245388253702,
      "learning_rate": 9.9741601518746e-06,
      "loss": 0.9465,
      "step": 837
    },
    {
      "epoch": 0.0614837165365885,
      "grad_norm": 1.1708346915481493,
      "learning_rate": 9.97403936858073e-06,
      "loss": 0.6133,
      "step": 838
    },
    {
      "epoch": 0.061557086126727625,
      "grad_norm": 0.9017282483818224,
      "learning_rate": 9.973918304390536e-06,
      "loss": 0.599,
      "step": 839
    },
    {
      "epoch": 0.061630455716866754,
      "grad_norm": 1.7539951256986999,
      "learning_rate": 9.97379695931086e-06,
      "loss": 0.9459,
      "step": 840
    },
    {
      "epoch": 0.061630455716866754,
      "lm_loss": 0.5971,
      "step": 840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.061630455716866754,
      "lm_loss": 0.5818,
      "step": 840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.061630455716866754,
      "lm_loss": 0.5509,
      "step": 840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.061630455716866754,
      "lm_loss": 0.6472,
      "step": 840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.061630455716866754,
      "lm_loss": 0.4804,
      "step": 840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.061630455716866754,
      "lm_loss": 0.5908,
      "step": 840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.061630455716866754,
      "lm_loss": 0.5411,
      "step": 840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.061630455716866754,
      "lm_loss": 0.7823,
      "step": 840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.061703825307005876,
      "grad_norm": 0.6657201983515136,
      "learning_rate": 9.97367533334855e-06,
      "loss": 0.5982,
      "step": 841
    },
    {
      "epoch": 0.061777194897145005,
      "grad_norm": 1.8426942687286285,
      "learning_rate": 9.973553426510478e-06,
      "loss": 0.953,
      "step": 842
    },
    {
      "epoch": 0.06185056448728413,
      "grad_norm": 1.7338103081526104,
      "learning_rate": 9.973431238803526e-06,
      "loss": 0.9503,
      "step": 843
    },
    {
      "epoch": 0.06192393407742326,
      "grad_norm": 1.7029734378521635,
      "learning_rate": 9.973308770234598e-06,
      "loss": 0.8443,
      "step": 844
    },
    {
      "epoch": 0.061997303667562384,
      "grad_norm": 1.781440225320916,
      "learning_rate": 9.973186020810605e-06,
      "loss": 1.0049,
      "step": 845
    },
    {
      "epoch": 0.06207067325770151,
      "grad_norm": 2.0437541885964117,
      "learning_rate": 9.973062990538483e-06,
      "loss": 0.8821,
      "step": 846
    },
    {
      "epoch": 0.06214404284784064,
      "grad_norm": 1.9013998477720473,
      "learning_rate": 9.972939679425177e-06,
      "loss": 0.9354,
      "step": 847
    },
    {
      "epoch": 0.06221741243797977,
      "grad_norm": 1.6079896171857067,
      "learning_rate": 9.972816087477652e-06,
      "loss": 0.8204,
      "step": 848
    },
    {
      "epoch": 0.06221741243797977,
      "lm_loss": 1.4532,
      "step": 848,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.06221741243797977,
      "lm_loss": 1.1749,
      "step": 848,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.06221741243797977,
      "lm_loss": 1.1799,
      "step": 848,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.06221741243797977,
      "lm_loss": 0.1404,
      "step": 848,
      "vm_loss": 0.112
    },
    {
      "epoch": 0.06221741243797977,
      "lm_loss": 0.1845,
      "step": 848,
      "vm_loss": 0.1032
    },
    {
      "epoch": 0.06221741243797977,
      "lm_loss": 0.2796,
      "step": 848,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.06221741243797977,
      "lm_loss": 1.1097,
      "step": 848,
      "vm_loss": 0.1152
    },
    {
      "epoch": 0.06221741243797977,
      "lm_loss": 0.7347,
      "step": 848,
      "vm_loss": 0.1741
    },
    {
      "epoch": 0.06229078202811889,
      "grad_norm": 2.0591276064786355,
      "learning_rate": 9.97269221470289e-06,
      "loss": 0.8711,
      "step": 849
    },
    {
      "epoch": 0.06236415161825802,
      "grad_norm": 1.6164531823722064,
      "learning_rate": 9.972568061107882e-06,
      "loss": 0.6231,
      "step": 850
    },
    {
      "epoch": 0.06243752120839715,
      "grad_norm": 2.1529934537856787,
      "learning_rate": 9.97244362669964e-06,
      "loss": 0.9705,
      "step": 851
    },
    {
      "epoch": 0.06251089079853628,
      "grad_norm": 2.0934099815934775,
      "learning_rate": 9.972318911485196e-06,
      "loss": 1.012,
      "step": 852
    },
    {
      "epoch": 0.0625842603886754,
      "grad_norm": 1.8710418967206381,
      "learning_rate": 9.972193915471586e-06,
      "loss": 0.9672,
      "step": 853
    },
    {
      "epoch": 0.06265762997881454,
      "grad_norm": 0.7537702497222224,
      "learning_rate": 9.972068638665873e-06,
      "loss": 0.6224,
      "step": 854
    },
    {
      "epoch": 0.06273099956895366,
      "grad_norm": 1.9376856686193826,
      "learning_rate": 9.971943081075132e-06,
      "loss": 0.8903,
      "step": 855
    },
    {
      "epoch": 0.06280436915909278,
      "grad_norm": 1.8282797264198611,
      "learning_rate": 9.971817242706452e-06,
      "loss": 1.0506,
      "step": 856
    },
    {
      "epoch": 0.06280436915909278,
      "lm_loss": 0.6717,
      "step": 856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06280436915909278,
      "lm_loss": 0.5611,
      "step": 856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06280436915909278,
      "lm_loss": 0.6025,
      "step": 856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06280436915909278,
      "lm_loss": 0.6843,
      "step": 856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06280436915909278,
      "lm_loss": 0.6849,
      "step": 856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06280436915909278,
      "lm_loss": 0.6287,
      "step": 856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06280436915909278,
      "lm_loss": 0.5729,
      "step": 856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06280436915909278,
      "lm_loss": 0.7923,
      "step": 856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06287773874923192,
      "grad_norm": 0.7609511066129937,
      "learning_rate": 9.971691123566938e-06,
      "loss": 0.6116,
      "step": 857
    },
    {
      "epoch": 0.06295110833937104,
      "grad_norm": 1.9491934032377138,
      "learning_rate": 9.971564723663716e-06,
      "loss": 0.9717,
      "step": 858
    },
    {
      "epoch": 0.06302447792951017,
      "grad_norm": 0.8093742131639665,
      "learning_rate": 9.971438043003923e-06,
      "loss": 0.6048,
      "step": 859
    },
    {
      "epoch": 0.0630978475196493,
      "grad_norm": 1.7854691097516489,
      "learning_rate": 9.97131108159471e-06,
      "loss": 0.8676,
      "step": 860
    },
    {
      "epoch": 0.06317121710978842,
      "grad_norm": 2.247944755208738,
      "learning_rate": 9.97118383944325e-06,
      "loss": 0.9867,
      "step": 861
    },
    {
      "epoch": 0.06324458669992755,
      "grad_norm": 1.61209873782341,
      "learning_rate": 9.971056316556727e-06,
      "loss": 0.862,
      "step": 862
    },
    {
      "epoch": 0.06331795629006667,
      "grad_norm": 1.6524688584298823,
      "learning_rate": 9.970928512942344e-06,
      "loss": 0.9136,
      "step": 863
    },
    {
      "epoch": 0.0633913258802058,
      "grad_norm": 1.702594189252745,
      "learning_rate": 9.970800428607317e-06,
      "loss": 0.8549,
      "step": 864
    },
    {
      "epoch": 0.0633913258802058,
      "lm_loss": 0.6484,
      "step": 864,
      "vm_loss": 0.1624
    },
    {
      "epoch": 0.0633913258802058,
      "lm_loss": 1.3481,
      "step": 864,
      "vm_loss": 0.1572
    },
    {
      "epoch": 0.0633913258802058,
      "lm_loss": 0.5514,
      "step": 864,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.0633913258802058,
      "lm_loss": 0.7349,
      "step": 864,
      "vm_loss": 0.2506
    },
    {
      "epoch": 0.0633913258802058,
      "lm_loss": 0.1405,
      "step": 864,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.0633913258802058,
      "lm_loss": 0.289,
      "step": 864,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.0633913258802058,
      "lm_loss": 0.0446,
      "step": 864,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.0633913258802058,
      "lm_loss": 0.1103,
      "step": 864,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.06346469547034493,
      "grad_norm": 1.7827865051954817,
      "learning_rate": 9.97067206355888e-06,
      "loss": 0.9189,
      "step": 865
    },
    {
      "epoch": 0.06353806506048405,
      "grad_norm": 2.000635017519496,
      "learning_rate": 9.970543417804283e-06,
      "loss": 0.9031,
      "step": 866
    },
    {
      "epoch": 0.06361143465062319,
      "grad_norm": 1.7491420346985969,
      "learning_rate": 9.970414491350789e-06,
      "loss": 0.9872,
      "step": 867
    },
    {
      "epoch": 0.06368480424076231,
      "grad_norm": 0.9548955661957542,
      "learning_rate": 9.970285284205679e-06,
      "loss": 0.615,
      "step": 868
    },
    {
      "epoch": 0.06375817383090143,
      "grad_norm": 0.7931241346359832,
      "learning_rate": 9.97015579637625e-06,
      "loss": 0.6105,
      "step": 869
    },
    {
      "epoch": 0.06383154342104057,
      "grad_norm": 1.981651457289292,
      "learning_rate": 9.970026027869817e-06,
      "loss": 0.8827,
      "step": 870
    },
    {
      "epoch": 0.06390491301117969,
      "grad_norm": 1.944795144675373,
      "learning_rate": 9.969895978693704e-06,
      "loss": 0.9725,
      "step": 871
    },
    {
      "epoch": 0.06397828260131881,
      "grad_norm": 0.763674982058434,
      "learning_rate": 9.969765648855259e-06,
      "loss": 0.5941,
      "step": 872
    },
    {
      "epoch": 0.06397828260131881,
      "lm_loss": 1.3837,
      "step": 872,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.06397828260131881,
      "lm_loss": 0.8539,
      "step": 872,
      "vm_loss": 0.1715
    },
    {
      "epoch": 0.06397828260131881,
      "lm_loss": 0.8922,
      "step": 872,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.06397828260131881,
      "lm_loss": 0.2084,
      "step": 872,
      "vm_loss": 0.1019
    },
    {
      "epoch": 0.06397828260131881,
      "lm_loss": 1.1687,
      "step": 872,
      "vm_loss": 0.1609
    },
    {
      "epoch": 0.06397828260131881,
      "lm_loss": 0.2423,
      "step": 872,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.06397828260131881,
      "lm_loss": 1.1492,
      "step": 872,
      "vm_loss": 0.2143
    },
    {
      "epoch": 0.06397828260131881,
      "lm_loss": 0.7512,
      "step": 872,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.06405165219145795,
      "grad_norm": 1.742287223622716,
      "learning_rate": 9.969635038361839e-06,
      "loss": 0.9263,
      "step": 873
    },
    {
      "epoch": 0.06412502178159707,
      "grad_norm": 0.8616641527045595,
      "learning_rate": 9.96950414722082e-06,
      "loss": 0.6076,
      "step": 874
    },
    {
      "epoch": 0.0641983913717362,
      "grad_norm": 0.7403204395125433,
      "learning_rate": 9.969372975439598e-06,
      "loss": 0.6096,
      "step": 875
    },
    {
      "epoch": 0.06427176096187533,
      "grad_norm": 1.8766510426832363,
      "learning_rate": 9.969241523025577e-06,
      "loss": 0.92,
      "step": 876
    },
    {
      "epoch": 0.06434513055201445,
      "grad_norm": 0.6527775493510531,
      "learning_rate": 9.96910978998618e-06,
      "loss": 0.6047,
      "step": 877
    },
    {
      "epoch": 0.06441850014215358,
      "grad_norm": 1.8764845271520576,
      "learning_rate": 9.968977776328848e-06,
      "loss": 0.9162,
      "step": 878
    },
    {
      "epoch": 0.0644918697322927,
      "grad_norm": 1.7554392751026824,
      "learning_rate": 9.968845482061036e-06,
      "loss": 0.841,
      "step": 879
    },
    {
      "epoch": 0.06456523932243184,
      "grad_norm": 1.8032223345674232,
      "learning_rate": 9.968712907190213e-06,
      "loss": 0.9001,
      "step": 880
    },
    {
      "epoch": 0.06456523932243184,
      "lm_loss": 1.4938,
      "step": 880,
      "vm_loss": 0.0423
    },
    {
      "epoch": 0.06456523932243184,
      "lm_loss": 1.4279,
      "step": 880,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.06456523932243184,
      "lm_loss": 0.4855,
      "step": 880,
      "vm_loss": 0.1601
    },
    {
      "epoch": 0.06456523932243184,
      "lm_loss": 1.2839,
      "step": 880,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.06456523932243184,
      "lm_loss": 0.9796,
      "step": 880,
      "vm_loss": 0.1949
    },
    {
      "epoch": 0.06456523932243184,
      "lm_loss": 0.5087,
      "step": 880,
      "vm_loss": 0.1772
    },
    {
      "epoch": 0.06456523932243184,
      "lm_loss": 1.0207,
      "step": 880,
      "vm_loss": 0.2042
    },
    {
      "epoch": 0.06456523932243184,
      "lm_loss": 0.7507,
      "step": 880,
      "vm_loss": 0.1603
    },
    {
      "epoch": 0.06463860891257096,
      "grad_norm": 1.7704210463295869,
      "learning_rate": 9.968580051723869e-06,
      "loss": 0.9979,
      "step": 881
    },
    {
      "epoch": 0.06471197850271009,
      "grad_norm": 1.033831369932537,
      "learning_rate": 9.968446915669503e-06,
      "loss": 0.5801,
      "step": 882
    },
    {
      "epoch": 0.06478534809284922,
      "grad_norm": 1.9200938574136253,
      "learning_rate": 9.968313499034638e-06,
      "loss": 0.8593,
      "step": 883
    },
    {
      "epoch": 0.06485871768298834,
      "grad_norm": 0.8394556865578653,
      "learning_rate": 9.968179801826804e-06,
      "loss": 0.6113,
      "step": 884
    },
    {
      "epoch": 0.06493208727312746,
      "grad_norm": 1.834256243246114,
      "learning_rate": 9.968045824053553e-06,
      "loss": 0.9047,
      "step": 885
    },
    {
      "epoch": 0.0650054568632666,
      "grad_norm": 1.8477698744535729,
      "learning_rate": 9.967911565722452e-06,
      "loss": 0.9151,
      "step": 886
    },
    {
      "epoch": 0.06507882645340572,
      "grad_norm": 1.9588111465482432,
      "learning_rate": 9.967777026841082e-06,
      "loss": 0.9546,
      "step": 887
    },
    {
      "epoch": 0.06515219604354486,
      "grad_norm": 2.289315964712761,
      "learning_rate": 9.967642207417041e-06,
      "loss": 0.9479,
      "step": 888
    },
    {
      "epoch": 0.06515219604354486,
      "lm_loss": 0.6034,
      "step": 888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06515219604354486,
      "lm_loss": 0.5744,
      "step": 888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06515219604354486,
      "lm_loss": 0.4921,
      "step": 888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06515219604354486,
      "lm_loss": 0.5796,
      "step": 888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06515219604354486,
      "lm_loss": 0.6659,
      "step": 888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06515219604354486,
      "lm_loss": 0.6284,
      "step": 888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06515219604354486,
      "lm_loss": 0.6527,
      "step": 888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06515219604354486,
      "lm_loss": 0.6323,
      "step": 888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06522556563368398,
      "grad_norm": 0.9894981558685353,
      "learning_rate": 9.967507107457942e-06,
      "loss": 0.6175,
      "step": 889
    },
    {
      "epoch": 0.0652989352238231,
      "grad_norm": 1.6679247613227974,
      "learning_rate": 9.967371726971415e-06,
      "loss": 0.8388,
      "step": 890
    },
    {
      "epoch": 0.06537230481396224,
      "grad_norm": 2.30158772654806,
      "learning_rate": 9.967236065965105e-06,
      "loss": 0.9966,
      "step": 891
    },
    {
      "epoch": 0.06544567440410136,
      "grad_norm": 0.7058797980793953,
      "learning_rate": 9.967100124446673e-06,
      "loss": 0.575,
      "step": 892
    },
    {
      "epoch": 0.06551904399424048,
      "grad_norm": 1.8936053161383228,
      "learning_rate": 9.966963902423796e-06,
      "loss": 0.8668,
      "step": 893
    },
    {
      "epoch": 0.06559241358437962,
      "grad_norm": 1.9024834502342498,
      "learning_rate": 9.96682739990417e-06,
      "loss": 0.9279,
      "step": 894
    },
    {
      "epoch": 0.06566578317451874,
      "grad_norm": 0.6324338858221323,
      "learning_rate": 9.966690616895496e-06,
      "loss": 0.5851,
      "step": 895
    },
    {
      "epoch": 0.06573915276465787,
      "grad_norm": 0.6402131945494025,
      "learning_rate": 9.966553553405505e-06,
      "loss": 0.5801,
      "step": 896
    },
    {
      "epoch": 0.06573915276465787,
      "lm_loss": 0.6929,
      "step": 896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06573915276465787,
      "lm_loss": 0.5356,
      "step": 896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06573915276465787,
      "lm_loss": 0.6055,
      "step": 896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06573915276465787,
      "lm_loss": 0.4655,
      "step": 896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06573915276465787,
      "lm_loss": 0.6264,
      "step": 896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06573915276465787,
      "lm_loss": 0.5456,
      "step": 896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06573915276465787,
      "lm_loss": 0.6161,
      "step": 896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06573915276465787,
      "lm_loss": 0.4498,
      "step": 896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.065812522354797,
      "grad_norm": 0.6221909258209969,
      "learning_rate": 9.966416209441936e-06,
      "loss": 0.5701,
      "step": 897
    },
    {
      "epoch": 0.06588589194493612,
      "grad_norm": 1.9667187438394285,
      "learning_rate": 9.966278585012544e-06,
      "loss": 0.8995,
      "step": 898
    },
    {
      "epoch": 0.06595926153507525,
      "grad_norm": 1.7814792167992302,
      "learning_rate": 9.9661406801251e-06,
      "loss": 0.8741,
      "step": 899
    },
    {
      "epoch": 0.06603263112521437,
      "grad_norm": 2.0107570662660494,
      "learning_rate": 9.966002494787394e-06,
      "loss": 0.9161,
      "step": 900
    },
    {
      "epoch": 0.06610600071535351,
      "grad_norm": 0.7034202255949259,
      "learning_rate": 9.96586402900723e-06,
      "loss": 0.5613,
      "step": 901
    },
    {
      "epoch": 0.06617937030549263,
      "grad_norm": 0.6962968753014263,
      "learning_rate": 9.965725282792427e-06,
      "loss": 0.6139,
      "step": 902
    },
    {
      "epoch": 0.06625273989563175,
      "grad_norm": 1.8337521685531892,
      "learning_rate": 9.96558625615082e-06,
      "loss": 0.9498,
      "step": 903
    },
    {
      "epoch": 0.06632610948577089,
      "grad_norm": 0.6352233528247525,
      "learning_rate": 9.965446949090258e-06,
      "loss": 0.6097,
      "step": 904
    },
    {
      "epoch": 0.06632610948577089,
      "lm_loss": 1.5018,
      "step": 904,
      "vm_loss": 0.0608
    },
    {
      "epoch": 0.06632610948577089,
      "lm_loss": 0.3662,
      "step": 904,
      "vm_loss": 0.2
    },
    {
      "epoch": 0.06632610948577089,
      "lm_loss": 0.4766,
      "step": 904,
      "vm_loss": 0.2476
    },
    {
      "epoch": 0.06632610948577089,
      "lm_loss": 0.6558,
      "step": 904,
      "vm_loss": 0.1297
    },
    {
      "epoch": 0.06632610948577089,
      "lm_loss": 1.1621,
      "step": 904,
      "vm_loss": 0.1225
    },
    {
      "epoch": 0.06632610948577089,
      "lm_loss": 0.9292,
      "step": 904,
      "vm_loss": 0.1228
    },
    {
      "epoch": 0.06632610948577089,
      "lm_loss": 1.1835,
      "step": 904,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.06632610948577089,
      "lm_loss": 0.7208,
      "step": 904,
      "vm_loss": 0.1571
    },
    {
      "epoch": 0.06639947907591001,
      "grad_norm": 1.8064809904073567,
      "learning_rate": 9.965307361618613e-06,
      "loss": 0.9367,
      "step": 905
    },
    {
      "epoch": 0.06647284866604913,
      "grad_norm": 1.8304691978153167,
      "learning_rate": 9.965167493743762e-06,
      "loss": 0.8073,
      "step": 906
    },
    {
      "epoch": 0.06654621825618827,
      "grad_norm": 1.8290706625518915,
      "learning_rate": 9.965027345473608e-06,
      "loss": 0.898,
      "step": 907
    },
    {
      "epoch": 0.06661958784632739,
      "grad_norm": 2.2079811069679605,
      "learning_rate": 9.964886916816063e-06,
      "loss": 0.9755,
      "step": 908
    },
    {
      "epoch": 0.06669295743646653,
      "grad_norm": 1.6865905081141377,
      "learning_rate": 9.96474620777906e-06,
      "loss": 1.016,
      "step": 909
    },
    {
      "epoch": 0.06676632702660565,
      "grad_norm": 1.7878472269471672,
      "learning_rate": 9.964605218370543e-06,
      "loss": 0.9463,
      "step": 910
    },
    {
      "epoch": 0.06683969661674477,
      "grad_norm": 1.7327248059828455,
      "learning_rate": 9.964463948598476e-06,
      "loss": 0.8919,
      "step": 911
    },
    {
      "epoch": 0.0669130662068839,
      "grad_norm": 0.8778597228902181,
      "learning_rate": 9.964322398470833e-06,
      "loss": 0.5851,
      "step": 912
    },
    {
      "epoch": 0.0669130662068839,
      "lm_loss": 1.2271,
      "step": 912,
      "vm_loss": 0.0573
    },
    {
      "epoch": 0.0669130662068839,
      "lm_loss": 1.0619,
      "step": 912,
      "vm_loss": 0.1237
    },
    {
      "epoch": 0.0669130662068839,
      "lm_loss": 1.3374,
      "step": 912,
      "vm_loss": 0.1483
    },
    {
      "epoch": 0.0669130662068839,
      "lm_loss": 0.2245,
      "step": 912,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.0669130662068839,
      "lm_loss": 0.8574,
      "step": 912,
      "vm_loss": 0.1402
    },
    {
      "epoch": 0.0669130662068839,
      "lm_loss": 0.9691,
      "step": 912,
      "vm_loss": 0.0955
    },
    {
      "epoch": 0.0669130662068839,
      "lm_loss": 0.259,
      "step": 912,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.0669130662068839,
      "lm_loss": 0.2043,
      "step": 912,
      "vm_loss": 0.0597
    },
    {
      "epoch": 0.06698643579702303,
      "grad_norm": 2.0511035390228605,
      "learning_rate": 9.964180567995612e-06,
      "loss": 0.9289,
      "step": 913
    },
    {
      "epoch": 0.06705980538716215,
      "grad_norm": 1.7842260866506,
      "learning_rate": 9.96403845718082e-06,
      "loss": 0.9263,
      "step": 914
    },
    {
      "epoch": 0.06713317497730129,
      "grad_norm": 1.827756774328866,
      "learning_rate": 9.963896066034484e-06,
      "loss": 0.9392,
      "step": 915
    },
    {
      "epoch": 0.06720654456744041,
      "grad_norm": 0.6337242687466003,
      "learning_rate": 9.963753394564644e-06,
      "loss": 0.5868,
      "step": 916
    },
    {
      "epoch": 0.06727991415757954,
      "grad_norm": 0.6141787382999422,
      "learning_rate": 9.963610442779356e-06,
      "loss": 0.5682,
      "step": 917
    },
    {
      "epoch": 0.06735328374771866,
      "grad_norm": 2.033605392905421,
      "learning_rate": 9.963467210686695e-06,
      "loss": 0.9823,
      "step": 918
    },
    {
      "epoch": 0.06742665333785779,
      "grad_norm": 0.6291680041512766,
      "learning_rate": 9.963323698294749e-06,
      "loss": 0.5759,
      "step": 919
    },
    {
      "epoch": 0.06750002292799692,
      "grad_norm": 1.5252097241287195,
      "learning_rate": 9.963179905611622e-06,
      "loss": 0.873,
      "step": 920
    },
    {
      "epoch": 0.06750002292799692,
      "lm_loss": 0.2357,
      "step": 920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06750002292799692,
      "lm_loss": 0.6361,
      "step": 920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06750002292799692,
      "lm_loss": 0.5597,
      "step": 920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06750002292799692,
      "lm_loss": 0.6967,
      "step": 920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06750002292799692,
      "lm_loss": 0.7739,
      "step": 920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06750002292799692,
      "lm_loss": 0.6535,
      "step": 920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06750002292799692,
      "lm_loss": 0.5407,
      "step": 920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06750002292799692,
      "lm_loss": 0.4024,
      "step": 920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06757339251813604,
      "grad_norm": 0.6115012111439864,
      "learning_rate": 9.963035832645435e-06,
      "loss": 0.5838,
      "step": 921
    },
    {
      "epoch": 0.06764676210827517,
      "grad_norm": 0.6267633162573453,
      "learning_rate": 9.962891479404324e-06,
      "loss": 0.59,
      "step": 922
    },
    {
      "epoch": 0.0677201316984143,
      "grad_norm": 0.5888431844417953,
      "learning_rate": 9.962746845896439e-06,
      "loss": 0.5855,
      "step": 923
    },
    {
      "epoch": 0.06779350128855342,
      "grad_norm": 1.8046108797673335,
      "learning_rate": 9.962601932129951e-06,
      "loss": 0.9049,
      "step": 924
    },
    {
      "epoch": 0.06786687087869256,
      "grad_norm": 1.7437200901702365,
      "learning_rate": 9.962456738113043e-06,
      "loss": 0.935,
      "step": 925
    },
    {
      "epoch": 0.06794024046883168,
      "grad_norm": 2.0604950738840198,
      "learning_rate": 9.962311263853912e-06,
      "loss": 0.9207,
      "step": 926
    },
    {
      "epoch": 0.0680136100589708,
      "grad_norm": 1.8548186686951977,
      "learning_rate": 9.962165509360774e-06,
      "loss": 0.9887,
      "step": 927
    },
    {
      "epoch": 0.06808697964910994,
      "grad_norm": 1.7992005212862032,
      "learning_rate": 9.962019474641864e-06,
      "loss": 0.908,
      "step": 928
    },
    {
      "epoch": 0.06808697964910994,
      "lm_loss": 0.9142,
      "step": 928,
      "vm_loss": 0.1495
    },
    {
      "epoch": 0.06808697964910994,
      "lm_loss": 0.949,
      "step": 928,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.06808697964910994,
      "lm_loss": 0.9256,
      "step": 928,
      "vm_loss": 0.0654
    },
    {
      "epoch": 0.06808697964910994,
      "lm_loss": 1.2225,
      "step": 928,
      "vm_loss": 0.166
    },
    {
      "epoch": 0.06808697964910994,
      "lm_loss": 0.8619,
      "step": 928,
      "vm_loss": 0.0954
    },
    {
      "epoch": 0.06808697964910994,
      "lm_loss": 0.3016,
      "step": 928,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.06808697964910994,
      "lm_loss": 0.271,
      "step": 928,
      "vm_loss": 0.0635
    },
    {
      "epoch": 0.06808697964910994,
      "lm_loss": 1.1147,
      "step": 928,
      "vm_loss": 0.2108
    },
    {
      "epoch": 0.06816034923924906,
      "grad_norm": 1.6116739682281327,
      "learning_rate": 9.961873159705426e-06,
      "loss": 0.9597,
      "step": 929
    },
    {
      "epoch": 0.0682337188293882,
      "grad_norm": 1.8587466238874317,
      "learning_rate": 9.96172656455972e-06,
      "loss": 1.0347,
      "step": 930
    },
    {
      "epoch": 0.06830708841952732,
      "grad_norm": 0.8162115799351732,
      "learning_rate": 9.96157968921303e-06,
      "loss": 0.5893,
      "step": 931
    },
    {
      "epoch": 0.06838045800966644,
      "grad_norm": 1.8597907519953445,
      "learning_rate": 9.961432533673647e-06,
      "loss": 0.8388,
      "step": 932
    },
    {
      "epoch": 0.06845382759980558,
      "grad_norm": 1.7821637145846219,
      "learning_rate": 9.961285097949882e-06,
      "loss": 0.8103,
      "step": 933
    },
    {
      "epoch": 0.0685271971899447,
      "grad_norm": 0.6476346180317599,
      "learning_rate": 9.96113738205006e-06,
      "loss": 0.5899,
      "step": 934
    },
    {
      "epoch": 0.06860056678008382,
      "grad_norm": 1.7393602383846802,
      "learning_rate": 9.960989385982524e-06,
      "loss": 0.9102,
      "step": 935
    },
    {
      "epoch": 0.06867393637022295,
      "grad_norm": 1.8060350480057816,
      "learning_rate": 9.960841109755634e-06,
      "loss": 1.0433,
      "step": 936
    },
    {
      "epoch": 0.06867393637022295,
      "lm_loss": 1.9742,
      "step": 936,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.06867393637022295,
      "lm_loss": 0.0655,
      "step": 936,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.06867393637022295,
      "lm_loss": 0.3703,
      "step": 936,
      "vm_loss": 0.2016
    },
    {
      "epoch": 0.06867393637022295,
      "lm_loss": 0.6477,
      "step": 936,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.06867393637022295,
      "lm_loss": 1.046,
      "step": 936,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.06867393637022295,
      "lm_loss": 0.6214,
      "step": 936,
      "vm_loss": 0.0891
    },
    {
      "epoch": 0.06867393637022295,
      "lm_loss": 0.3812,
      "step": 936,
      "vm_loss": 0.1109
    },
    {
      "epoch": 0.06867393637022295,
      "lm_loss": 0.2108,
      "step": 936,
      "vm_loss": 0.1947
    },
    {
      "epoch": 0.06874730596036208,
      "grad_norm": 1.7392409175389654,
      "learning_rate": 9.96069255337776e-06,
      "loss": 0.9584,
      "step": 937
    },
    {
      "epoch": 0.06882067555050121,
      "grad_norm": 1.908670562609859,
      "learning_rate": 9.960543716857292e-06,
      "loss": 0.8974,
      "step": 938
    },
    {
      "epoch": 0.06889404514064033,
      "grad_norm": 1.7317646979951946,
      "learning_rate": 9.960394600202636e-06,
      "loss": 0.8257,
      "step": 939
    },
    {
      "epoch": 0.06896741473077946,
      "grad_norm": 0.6518053752953978,
      "learning_rate": 9.960245203422213e-06,
      "loss": 0.595,
      "step": 940
    },
    {
      "epoch": 0.06904078432091859,
      "grad_norm": 2.1983434647433433,
      "learning_rate": 9.96009552652446e-06,
      "loss": 0.9503,
      "step": 941
    },
    {
      "epoch": 0.06911415391105771,
      "grad_norm": 0.6288210274903535,
      "learning_rate": 9.959945569517827e-06,
      "loss": 0.6067,
      "step": 942
    },
    {
      "epoch": 0.06918752350119683,
      "grad_norm": 2.315600446287086,
      "learning_rate": 9.959795332410786e-06,
      "loss": 1.0035,
      "step": 943
    },
    {
      "epoch": 0.06926089309133597,
      "grad_norm": 0.5923084203706249,
      "learning_rate": 9.959644815211823e-06,
      "loss": 0.5778,
      "step": 944
    },
    {
      "epoch": 0.06926089309133597,
      "lm_loss": 0.487,
      "step": 944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06926089309133597,
      "lm_loss": 0.5468,
      "step": 944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06926089309133597,
      "lm_loss": 0.5258,
      "step": 944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06926089309133597,
      "lm_loss": 0.4992,
      "step": 944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06926089309133597,
      "lm_loss": 0.8059,
      "step": 944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06926089309133597,
      "lm_loss": 0.5911,
      "step": 944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06926089309133597,
      "lm_loss": 0.6813,
      "step": 944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06926089309133597,
      "lm_loss": 0.6884,
      "step": 944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.06933426268147509,
      "grad_norm": 0.6196983874799713,
      "learning_rate": 9.95949401792943e-06,
      "loss": 0.6181,
      "step": 945
    },
    {
      "epoch": 0.06940763227161423,
      "grad_norm": 1.9086098297124325,
      "learning_rate": 9.95934294057213e-06,
      "loss": 0.9092,
      "step": 946
    },
    {
      "epoch": 0.06948100186175335,
      "grad_norm": 1.946625539736542,
      "learning_rate": 9.959191583148452e-06,
      "loss": 1.0759,
      "step": 947
    },
    {
      "epoch": 0.06955437145189247,
      "grad_norm": 0.6122672435785633,
      "learning_rate": 9.959039945666945e-06,
      "loss": 0.5897,
      "step": 948
    },
    {
      "epoch": 0.06962774104203161,
      "grad_norm": 1.971221888845571,
      "learning_rate": 9.95888802813617e-06,
      "loss": 0.9369,
      "step": 949
    },
    {
      "epoch": 0.06970111063217073,
      "grad_norm": 1.7742893285381485,
      "learning_rate": 9.958735830564708e-06,
      "loss": 0.9171,
      "step": 950
    },
    {
      "epoch": 0.06977448022230986,
      "grad_norm": 1.8398477159896123,
      "learning_rate": 9.958583352961155e-06,
      "loss": 1.1195,
      "step": 951
    },
    {
      "epoch": 0.06984784981244899,
      "grad_norm": 1.9457955088116285,
      "learning_rate": 9.958430595334118e-06,
      "loss": 0.975,
      "step": 952
    },
    {
      "epoch": 0.06984784981244899,
      "lm_loss": 1.52,
      "step": 952,
      "vm_loss": 0.0403
    },
    {
      "epoch": 0.06984784981244899,
      "lm_loss": 0.9895,
      "step": 952,
      "vm_loss": 0.1341
    },
    {
      "epoch": 0.06984784981244899,
      "lm_loss": 1.0396,
      "step": 952,
      "vm_loss": 0.221
    },
    {
      "epoch": 0.06984784981244899,
      "lm_loss": 0.7938,
      "step": 952,
      "vm_loss": 0.1223
    },
    {
      "epoch": 0.06984784981244899,
      "lm_loss": 1.0145,
      "step": 952,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.06984784981244899,
      "lm_loss": 0.1779,
      "step": 952,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.06984784981244899,
      "lm_loss": 0.7207,
      "step": 952,
      "vm_loss": 0.1244
    },
    {
      "epoch": 0.06984784981244899,
      "lm_loss": 1.4643,
      "step": 952,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0.06992121940258811,
      "grad_norm": 1.8320005519398896,
      "learning_rate": 9.958277557692228e-06,
      "loss": 0.9406,
      "step": 953
    },
    {
      "epoch": 0.06999458899272724,
      "grad_norm": 1.8216385610194203,
      "learning_rate": 9.958124240044124e-06,
      "loss": 0.9238,
      "step": 954
    },
    {
      "epoch": 0.07006795858286637,
      "grad_norm": 2.31221343922835,
      "learning_rate": 9.957970642398467e-06,
      "loss": 0.9358,
      "step": 955
    },
    {
      "epoch": 0.07014132817300549,
      "grad_norm": 1.7626350328192177,
      "learning_rate": 9.957816764763927e-06,
      "loss": 0.9008,
      "step": 956
    },
    {
      "epoch": 0.07021469776314462,
      "grad_norm": 1.8520500481793496,
      "learning_rate": 9.957662607149199e-06,
      "loss": 0.9315,
      "step": 957
    },
    {
      "epoch": 0.07028806735328375,
      "grad_norm": 1.7491505336222084,
      "learning_rate": 9.957508169562987e-06,
      "loss": 0.9777,
      "step": 958
    },
    {
      "epoch": 0.07036143694342288,
      "grad_norm": 0.7130714623064229,
      "learning_rate": 9.95735345201401e-06,
      "loss": 0.6022,
      "step": 959
    },
    {
      "epoch": 0.070434806533562,
      "grad_norm": 0.6500676336016623,
      "learning_rate": 9.957198454511006e-06,
      "loss": 0.5949,
      "step": 960
    },
    {
      "epoch": 0.070434806533562,
      "lm_loss": 1.6875,
      "step": 960,
      "vm_loss": 0.0419
    },
    {
      "epoch": 0.070434806533562,
      "lm_loss": 0.9423,
      "step": 960,
      "vm_loss": 0.0448
    },
    {
      "epoch": 0.070434806533562,
      "lm_loss": 0.3753,
      "step": 960,
      "vm_loss": 0.1725
    },
    {
      "epoch": 0.070434806533562,
      "lm_loss": 0.442,
      "step": 960,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.070434806533562,
      "lm_loss": 0.5854,
      "step": 960,
      "vm_loss": 0.0888
    },
    {
      "epoch": 0.070434806533562,
      "lm_loss": 0.0951,
      "step": 960,
      "vm_loss": 0.0886
    },
    {
      "epoch": 0.070434806533562,
      "lm_loss": 0.4951,
      "step": 960,
      "vm_loss": 0.1665
    },
    {
      "epoch": 0.070434806533562,
      "lm_loss": 1.016,
      "step": 960,
      "vm_loss": 0.1673
    },
    {
      "epoch": 0.07050817612370112,
      "grad_norm": 2.001406583192009,
      "learning_rate": 9.957043177062728e-06,
      "loss": 0.8323,
      "step": 961
    },
    {
      "epoch": 0.07058154571384026,
      "grad_norm": 2.0868549324551657,
      "learning_rate": 9.956887619677948e-06,
      "loss": 1.0228,
      "step": 962
    },
    {
      "epoch": 0.07065491530397938,
      "grad_norm": 0.6631318521363654,
      "learning_rate": 9.956731782365448e-06,
      "loss": 0.6085,
      "step": 963
    },
    {
      "epoch": 0.0707282848941185,
      "grad_norm": 1.8323511268495059,
      "learning_rate": 9.95657566513403e-06,
      "loss": 0.9733,
      "step": 964
    },
    {
      "epoch": 0.07080165448425764,
      "grad_norm": 1.7148326869491752,
      "learning_rate": 9.956419267992509e-06,
      "loss": 0.9397,
      "step": 965
    },
    {
      "epoch": 0.07087502407439676,
      "grad_norm": 1.5454994198853562,
      "learning_rate": 9.956262590949716e-06,
      "loss": 0.8751,
      "step": 966
    },
    {
      "epoch": 0.0709483936645359,
      "grad_norm": 1.7006188513613416,
      "learning_rate": 9.956105634014503e-06,
      "loss": 1.0251,
      "step": 967
    },
    {
      "epoch": 0.07102176325467502,
      "grad_norm": 0.7140210816033276,
      "learning_rate": 9.95594839719573e-06,
      "loss": 0.5668,
      "step": 968
    },
    {
      "epoch": 0.07102176325467502,
      "lm_loss": 1.2336,
      "step": 968,
      "vm_loss": 0.0421
    },
    {
      "epoch": 0.07102176325467502,
      "lm_loss": 0.0206,
      "step": 968,
      "vm_loss": 0.1081
    },
    {
      "epoch": 0.07102176325467502,
      "lm_loss": 0.3985,
      "step": 968,
      "vm_loss": 0.156
    },
    {
      "epoch": 0.07102176325467502,
      "lm_loss": 0.1537,
      "step": 968,
      "vm_loss": 0.07
    },
    {
      "epoch": 0.07102176325467502,
      "lm_loss": 0.8539,
      "step": 968,
      "vm_loss": 0.1262
    },
    {
      "epoch": 0.07102176325467502,
      "lm_loss": 0.7645,
      "step": 968,
      "vm_loss": 0.17
    },
    {
      "epoch": 0.07102176325467502,
      "lm_loss": 0.3269,
      "step": 968,
      "vm_loss": 0.0785
    },
    {
      "epoch": 0.07102176325467502,
      "lm_loss": 1.1574,
      "step": 968,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.07109513284481414,
      "grad_norm": 1.7980897569043044,
      "learning_rate": 9.955790880502278e-06,
      "loss": 0.8383,
      "step": 969
    },
    {
      "epoch": 0.07116850243495328,
      "grad_norm": 1.70598857776902,
      "learning_rate": 9.955633083943042e-06,
      "loss": 0.8346,
      "step": 970
    },
    {
      "epoch": 0.0712418720250924,
      "grad_norm": 1.7847840037551352,
      "learning_rate": 9.955475007526933e-06,
      "loss": 0.9656,
      "step": 971
    },
    {
      "epoch": 0.07131524161523153,
      "grad_norm": 1.8302663309241471,
      "learning_rate": 9.955316651262879e-06,
      "loss": 0.9759,
      "step": 972
    },
    {
      "epoch": 0.07138861120537066,
      "grad_norm": 1.8596213002099187,
      "learning_rate": 9.955158015159822e-06,
      "loss": 0.9584,
      "step": 973
    },
    {
      "epoch": 0.07146198079550978,
      "grad_norm": 1.9317084098314794,
      "learning_rate": 9.954999099226721e-06,
      "loss": 0.9168,
      "step": 974
    },
    {
      "epoch": 0.07153535038564891,
      "grad_norm": 1.7633944172359266,
      "learning_rate": 9.954839903472548e-06,
      "loss": 1.0131,
      "step": 975
    },
    {
      "epoch": 0.07160871997578803,
      "grad_norm": 0.6458027361749221,
      "learning_rate": 9.954680427906298e-06,
      "loss": 0.5885,
      "step": 976
    },
    {
      "epoch": 0.07160871997578803,
      "lm_loss": 1.8319,
      "step": 976,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.07160871997578803,
      "lm_loss": 0.0648,
      "step": 976,
      "vm_loss": 0.1553
    },
    {
      "epoch": 0.07160871997578803,
      "lm_loss": 0.5095,
      "step": 976,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.07160871997578803,
      "lm_loss": 1.1562,
      "step": 976,
      "vm_loss": 0.1611
    },
    {
      "epoch": 0.07160871997578803,
      "lm_loss": 0.9845,
      "step": 976,
      "vm_loss": 0.1
    },
    {
      "epoch": 0.07160871997578803,
      "lm_loss": 0.8509,
      "step": 976,
      "vm_loss": 0.2488
    },
    {
      "epoch": 0.07160871997578803,
      "lm_loss": 1.2449,
      "step": 976,
      "vm_loss": 0.1701
    },
    {
      "epoch": 0.07160871997578803,
      "lm_loss": 0.4336,
      "step": 976,
      "vm_loss": 0.1909
    },
    {
      "epoch": 0.07168208956592716,
      "grad_norm": 2.242417016044528,
      "learning_rate": 9.954520672536973e-06,
      "loss": 0.9405,
      "step": 977
    },
    {
      "epoch": 0.07175545915606629,
      "grad_norm": 0.62607949633818,
      "learning_rate": 9.954360637373595e-06,
      "loss": 0.5876,
      "step": 978
    },
    {
      "epoch": 0.07182882874620541,
      "grad_norm": 1.8130558938671333,
      "learning_rate": 9.954200322425204e-06,
      "loss": 1.0304,
      "step": 979
    },
    {
      "epoch": 0.07190219833634455,
      "grad_norm": 1.722133626794696,
      "learning_rate": 9.95403972770085e-06,
      "loss": 0.8718,
      "step": 980
    },
    {
      "epoch": 0.07197556792648367,
      "grad_norm": 1.7278096127717368,
      "learning_rate": 9.953878853209606e-06,
      "loss": 0.9034,
      "step": 981
    },
    {
      "epoch": 0.0720489375166228,
      "grad_norm": 1.6987665839321267,
      "learning_rate": 9.953717698960554e-06,
      "loss": 0.9381,
      "step": 982
    },
    {
      "epoch": 0.07212230710676193,
      "grad_norm": 1.8874828215266282,
      "learning_rate": 9.953556264962797e-06,
      "loss": 0.9978,
      "step": 983
    },
    {
      "epoch": 0.07219567669690105,
      "grad_norm": 0.707855949118773,
      "learning_rate": 9.953394551225451e-06,
      "loss": 0.605,
      "step": 984
    },
    {
      "epoch": 0.07219567669690105,
      "lm_loss": 1.5676,
      "step": 984,
      "vm_loss": 0.0323
    },
    {
      "epoch": 0.07219567669690105,
      "lm_loss": 1.159,
      "step": 984,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.07219567669690105,
      "lm_loss": 0.1378,
      "step": 984,
      "vm_loss": 0.1578
    },
    {
      "epoch": 0.07219567669690105,
      "lm_loss": 0.7639,
      "step": 984,
      "vm_loss": 0.1597
    },
    {
      "epoch": 0.07219567669690105,
      "lm_loss": 1.0239,
      "step": 984,
      "vm_loss": 0.1667
    },
    {
      "epoch": 0.07219567669690105,
      "lm_loss": 0.3896,
      "step": 984,
      "vm_loss": 0.1074
    },
    {
      "epoch": 0.07219567669690105,
      "lm_loss": 0.4729,
      "step": 984,
      "vm_loss": 0.1123
    },
    {
      "epoch": 0.07219567669690105,
      "lm_loss": 1.1641,
      "step": 984,
      "vm_loss": 0.1365
    },
    {
      "epoch": 0.07226904628704017,
      "grad_norm": 1.7515160223974187,
      "learning_rate": 9.953232557757646e-06,
      "loss": 1.0011,
      "step": 985
    },
    {
      "epoch": 0.07234241587717931,
      "grad_norm": 1.9203869105095128,
      "learning_rate": 9.953070284568531e-06,
      "loss": 1.0831,
      "step": 986
    },
    {
      "epoch": 0.07241578546731843,
      "grad_norm": 1.7497429793326442,
      "learning_rate": 9.952907731667273e-06,
      "loss": 0.9089,
      "step": 987
    },
    {
      "epoch": 0.07248915505745757,
      "grad_norm": 0.6357711769609136,
      "learning_rate": 9.95274489906305e-06,
      "loss": 0.5627,
      "step": 988
    },
    {
      "epoch": 0.07256252464759669,
      "grad_norm": 1.7522104889252363,
      "learning_rate": 9.952581786765057e-06,
      "loss": 0.8637,
      "step": 989
    },
    {
      "epoch": 0.07263589423773581,
      "grad_norm": 1.6504523356143934,
      "learning_rate": 9.952418394782505e-06,
      "loss": 0.8855,
      "step": 990
    },
    {
      "epoch": 0.07270926382787495,
      "grad_norm": 1.807925915908064,
      "learning_rate": 9.952254723124621e-06,
      "loss": 1.0028,
      "step": 991
    },
    {
      "epoch": 0.07278263341801407,
      "grad_norm": 1.7521704551194102,
      "learning_rate": 9.95209077180065e-06,
      "loss": 0.945,
      "step": 992
    },
    {
      "epoch": 0.07278263341801407,
      "lm_loss": 1.0854,
      "step": 992,
      "vm_loss": 0.0779
    },
    {
      "epoch": 0.07278263341801407,
      "lm_loss": 0.3892,
      "step": 992,
      "vm_loss": 0.1236
    },
    {
      "epoch": 0.07278263341801407,
      "lm_loss": 0.5232,
      "step": 992,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.07278263341801407,
      "lm_loss": 0.6359,
      "step": 992,
      "vm_loss": 0.1601
    },
    {
      "epoch": 0.07278263341801407,
      "lm_loss": 0.6856,
      "step": 992,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.07278263341801407,
      "lm_loss": 0.4723,
      "step": 992,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.07278263341801407,
      "lm_loss": 0.9534,
      "step": 992,
      "vm_loss": 0.1938
    },
    {
      "epoch": 0.07278263341801407,
      "lm_loss": 0.8018,
      "step": 992,
      "vm_loss": 0.1668
    },
    {
      "epoch": 0.07285600300815319,
      "grad_norm": 1.9621348154374505,
      "learning_rate": 9.951926540819849e-06,
      "loss": 0.9373,
      "step": 993
    },
    {
      "epoch": 0.07292937259829232,
      "grad_norm": 1.7749513492572566,
      "learning_rate": 9.951762030191493e-06,
      "loss": 0.8694,
      "step": 994
    },
    {
      "epoch": 0.07300274218843145,
      "grad_norm": 1.730699510658165,
      "learning_rate": 9.951597239924872e-06,
      "loss": 0.9385,
      "step": 995
    },
    {
      "epoch": 0.07307611177857058,
      "grad_norm": 1.940984878995435,
      "learning_rate": 9.951432170029294e-06,
      "loss": 0.8921,
      "step": 996
    },
    {
      "epoch": 0.0731494813687097,
      "grad_norm": 0.6862359009451199,
      "learning_rate": 9.951266820514078e-06,
      "loss": 0.599,
      "step": 997
    },
    {
      "epoch": 0.07322285095884883,
      "grad_norm": 1.7826776729697094,
      "learning_rate": 9.951101191388563e-06,
      "loss": 0.8052,
      "step": 998
    },
    {
      "epoch": 0.07329622054898796,
      "grad_norm": 1.7888207800288045,
      "learning_rate": 9.950935282662103e-06,
      "loss": 0.9611,
      "step": 999
    },
    {
      "epoch": 0.07336959013912708,
      "grad_norm": 0.6044617586746314,
      "learning_rate": 9.950769094344066e-06,
      "loss": 0.5717,
      "step": 1000
    },
    {
      "epoch": 0.07336959013912708,
      "lm_loss": 0.8958,
      "step": 1000,
      "vm_loss": 0.0394
    },
    {
      "epoch": 0.07336959013912708,
      "lm_loss": 0.2628,
      "step": 1000,
      "vm_loss": 0.1158
    },
    {
      "epoch": 0.07336959013912708,
      "lm_loss": 0.3391,
      "step": 1000,
      "vm_loss": 0.1212
    },
    {
      "epoch": 0.07336959013912708,
      "lm_loss": 0.1191,
      "step": 1000,
      "vm_loss": 0.2228
    },
    {
      "epoch": 0.07336959013912708,
      "lm_loss": 1.1478,
      "step": 1000,
      "vm_loss": 0.141
    },
    {
      "epoch": 0.07336959013912708,
      "lm_loss": 0.4994,
      "step": 1000,
      "vm_loss": 0.0422
    },
    {
      "epoch": 0.07336959013912708,
      "lm_loss": 0.9386,
      "step": 1000,
      "vm_loss": 0.198
    },
    {
      "epoch": 0.07336959013912708,
      "lm_loss": 0.7258,
      "step": 1000,
      "vm_loss": 0.1717
    },
    {
      "epoch": 0.07344295972926622,
      "grad_norm": 1.7343059196342503,
      "learning_rate": 9.95060262644384e-06,
      "loss": 0.8826,
      "step": 1001
    },
    {
      "epoch": 0.07351632931940534,
      "grad_norm": 1.8257317638772645,
      "learning_rate": 9.95043587897082e-06,
      "loss": 0.9093,
      "step": 1002
    },
    {
      "epoch": 0.07358969890954446,
      "grad_norm": 0.5989039799526195,
      "learning_rate": 9.95026885193443e-06,
      "loss": 0.5673,
      "step": 1003
    },
    {
      "epoch": 0.0736630684996836,
      "grad_norm": 0.6289178190528953,
      "learning_rate": 9.950101545344097e-06,
      "loss": 0.5947,
      "step": 1004
    },
    {
      "epoch": 0.07373643808982272,
      "grad_norm": 1.891965601712267,
      "learning_rate": 9.94993395920927e-06,
      "loss": 0.9433,
      "step": 1005
    },
    {
      "epoch": 0.07380980767996184,
      "grad_norm": 1.8609618879576653,
      "learning_rate": 9.949766093539417e-06,
      "loss": 0.9543,
      "step": 1006
    },
    {
      "epoch": 0.07388317727010098,
      "grad_norm": 1.7887214315021571,
      "learning_rate": 9.949597948344012e-06,
      "loss": 0.9729,
      "step": 1007
    },
    {
      "epoch": 0.0739565468602401,
      "grad_norm": 1.755231223457212,
      "learning_rate": 9.949429523632557e-06,
      "loss": 0.9411,
      "step": 1008
    },
    {
      "epoch": 0.0739565468602401,
      "lm_loss": 2.7494,
      "step": 1008,
      "vm_loss": 0.1214
    },
    {
      "epoch": 0.0739565468602401,
      "lm_loss": 0.7608,
      "step": 1008,
      "vm_loss": 0.0967
    },
    {
      "epoch": 0.0739565468602401,
      "lm_loss": 0.6223,
      "step": 1008,
      "vm_loss": 0.0628
    },
    {
      "epoch": 0.0739565468602401,
      "lm_loss": 0.7509,
      "step": 1008,
      "vm_loss": 0.0945
    },
    {
      "epoch": 0.0739565468602401,
      "lm_loss": 1.2093,
      "step": 1008,
      "vm_loss": 0.0641
    },
    {
      "epoch": 0.0739565468602401,
      "lm_loss": 1.2196,
      "step": 1008,
      "vm_loss": 0.197
    },
    {
      "epoch": 0.0739565468602401,
      "lm_loss": 0.9145,
      "step": 1008,
      "vm_loss": 0.1377
    },
    {
      "epoch": 0.0739565468602401,
      "lm_loss": 0.9809,
      "step": 1008,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.07402991645037924,
      "grad_norm": 1.61588223738359,
      "learning_rate": 9.949260819414557e-06,
      "loss": 0.8702,
      "step": 1009
    },
    {
      "epoch": 0.07410328604051836,
      "grad_norm": 1.8664153988465337,
      "learning_rate": 9.949091835699541e-06,
      "loss": 0.8661,
      "step": 1010
    },
    {
      "epoch": 0.07417665563065748,
      "grad_norm": 1.7475262575172723,
      "learning_rate": 9.948922572497056e-06,
      "loss": 0.9006,
      "step": 1011
    },
    {
      "epoch": 0.07425002522079661,
      "grad_norm": 0.7141631610908543,
      "learning_rate": 9.948753029816654e-06,
      "loss": 0.6042,
      "step": 1012
    },
    {
      "epoch": 0.07432339481093574,
      "grad_norm": 1.7793577444917437,
      "learning_rate": 9.948583207667915e-06,
      "loss": 0.8953,
      "step": 1013
    },
    {
      "epoch": 0.07439676440107486,
      "grad_norm": 0.5984492678060748,
      "learning_rate": 9.948413106060426e-06,
      "loss": 0.5811,
      "step": 1014
    },
    {
      "epoch": 0.074470133991214,
      "grad_norm": 1.7001563889778584,
      "learning_rate": 9.948242725003796e-06,
      "loss": 0.843,
      "step": 1015
    },
    {
      "epoch": 0.07454350358135312,
      "grad_norm": 1.5425746052956193,
      "learning_rate": 9.948072064507645e-06,
      "loss": 0.8609,
      "step": 1016
    },
    {
      "epoch": 0.07454350358135312,
      "lm_loss": 1.4395,
      "step": 1016,
      "vm_loss": 0.0538
    },
    {
      "epoch": 0.07454350358135312,
      "lm_loss": 1.2676,
      "step": 1016,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.07454350358135312,
      "lm_loss": 0.8668,
      "step": 1016,
      "vm_loss": 0.1283
    },
    {
      "epoch": 0.07454350358135312,
      "lm_loss": 0.2324,
      "step": 1016,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.07454350358135312,
      "lm_loss": 1.0285,
      "step": 1016,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.07454350358135312,
      "lm_loss": 0.8957,
      "step": 1016,
      "vm_loss": 0.1491
    },
    {
      "epoch": 0.07454350358135312,
      "lm_loss": 1.1128,
      "step": 1016,
      "vm_loss": 0.2155
    },
    {
      "epoch": 0.07454350358135312,
      "lm_loss": 0.9703,
      "step": 1016,
      "vm_loss": 0.1311
    },
    {
      "epoch": 0.07461687317149225,
      "grad_norm": 1.778786768014517,
      "learning_rate": 9.947901124581609e-06,
      "loss": 0.9312,
      "step": 1017
    },
    {
      "epoch": 0.07469024276163137,
      "grad_norm": 0.7374930703890689,
      "learning_rate": 9.947729905235346e-06,
      "loss": 0.5835,
      "step": 1018
    },
    {
      "epoch": 0.0747636123517705,
      "grad_norm": 0.6686508610802854,
      "learning_rate": 9.94755840647852e-06,
      "loss": 0.5889,
      "step": 1019
    },
    {
      "epoch": 0.07483698194190963,
      "grad_norm": 1.6875851229125194,
      "learning_rate": 9.947386628320818e-06,
      "loss": 0.9469,
      "step": 1020
    },
    {
      "epoch": 0.07491035153204875,
      "grad_norm": 1.8660122445687537,
      "learning_rate": 9.947214570771944e-06,
      "loss": 0.8153,
      "step": 1021
    },
    {
      "epoch": 0.07498372112218789,
      "grad_norm": 1.6583238630703212,
      "learning_rate": 9.94704223384161e-06,
      "loss": 0.9321,
      "step": 1022
    },
    {
      "epoch": 0.07505709071232701,
      "grad_norm": 1.6666649868410544,
      "learning_rate": 9.94686961753955e-06,
      "loss": 0.9471,
      "step": 1023
    },
    {
      "epoch": 0.07513046030246613,
      "grad_norm": 1.8054311860868082,
      "learning_rate": 9.946696721875511e-06,
      "loss": 1.0521,
      "step": 1024
    },
    {
      "epoch": 0.07513046030246613,
      "lm_loss": 1.481,
      "step": 1024,
      "vm_loss": 0.1079
    },
    {
      "epoch": 0.07513046030246613,
      "lm_loss": 0.0784,
      "step": 1024,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.07513046030246613,
      "lm_loss": 0.3364,
      "step": 1024,
      "vm_loss": 0.1603
    },
    {
      "epoch": 0.07513046030246613,
      "lm_loss": 0.9241,
      "step": 1024,
      "vm_loss": 0.1672
    },
    {
      "epoch": 0.07513046030246613,
      "lm_loss": 0.6641,
      "step": 1024,
      "vm_loss": 0.1245
    },
    {
      "epoch": 0.07513046030246613,
      "lm_loss": 0.8227,
      "step": 1024,
      "vm_loss": 0.0784
    },
    {
      "epoch": 0.07513046030246613,
      "lm_loss": 0.5346,
      "step": 1024,
      "vm_loss": 0.2595
    },
    {
      "epoch": 0.07513046030246613,
      "lm_loss": 0.5795,
      "step": 1024,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.07520382989260527,
      "grad_norm": 1.714547247838713,
      "learning_rate": 9.94652354685926e-06,
      "loss": 0.9435,
      "step": 1025
    },
    {
      "epoch": 0.07527719948274439,
      "grad_norm": 1.7852731172261838,
      "learning_rate": 9.946350092500574e-06,
      "loss": 0.9066,
      "step": 1026
    },
    {
      "epoch": 0.07535056907288351,
      "grad_norm": 0.8647884342500941,
      "learning_rate": 9.946176358809247e-06,
      "loss": 0.588,
      "step": 1027
    },
    {
      "epoch": 0.07542393866302265,
      "grad_norm": 1.9293866644551962,
      "learning_rate": 9.946002345795093e-06,
      "loss": 0.8718,
      "step": 1028
    },
    {
      "epoch": 0.07549730825316177,
      "grad_norm": 0.6570264246428822,
      "learning_rate": 9.945828053467939e-06,
      "loss": 0.5884,
      "step": 1029
    },
    {
      "epoch": 0.0755706778433009,
      "grad_norm": 2.1199101273867296,
      "learning_rate": 9.945653481837625e-06,
      "loss": 0.9425,
      "step": 1030
    },
    {
      "epoch": 0.07564404743344003,
      "grad_norm": 2.0962740335435917,
      "learning_rate": 9.945478630914012e-06,
      "loss": 0.9792,
      "step": 1031
    },
    {
      "epoch": 0.07571741702357915,
      "grad_norm": 1.535427366195211,
      "learning_rate": 9.945303500706973e-06,
      "loss": 0.8798,
      "step": 1032
    },
    {
      "epoch": 0.07571741702357915,
      "lm_loss": 0.9089,
      "step": 1032,
      "vm_loss": 0.0371
    },
    {
      "epoch": 0.07571741702357915,
      "lm_loss": 1.1438,
      "step": 1032,
      "vm_loss": 0.1252
    },
    {
      "epoch": 0.07571741702357915,
      "lm_loss": 1.1872,
      "step": 1032,
      "vm_loss": 0.121
    },
    {
      "epoch": 0.07571741702357915,
      "lm_loss": 0.3834,
      "step": 1032,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.07571741702357915,
      "lm_loss": 0.0788,
      "step": 1032,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.07571741702357915,
      "lm_loss": 0.0806,
      "step": 1032,
      "vm_loss": 0.1811
    },
    {
      "epoch": 0.07571741702357915,
      "lm_loss": 0.962,
      "step": 1032,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.07571741702357915,
      "lm_loss": 0.9867,
      "step": 1032,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.07579078661371828,
      "grad_norm": 1.7143184697384057,
      "learning_rate": 9.9451280912264e-06,
      "loss": 0.8729,
      "step": 1033
    },
    {
      "epoch": 0.0758641562038574,
      "grad_norm": 2.2365808566818344,
      "learning_rate": 9.944952402482195e-06,
      "loss": 0.8743,
      "step": 1034
    },
    {
      "epoch": 0.07593752579399653,
      "grad_norm": 1.001289623894492,
      "learning_rate": 9.944776434484283e-06,
      "loss": 0.566,
      "step": 1035
    },
    {
      "epoch": 0.07601089538413566,
      "grad_norm": 1.6375741303450349,
      "learning_rate": 9.9446001872426e-06,
      "loss": 0.8306,
      "step": 1036
    },
    {
      "epoch": 0.07608426497427478,
      "grad_norm": 0.7276487661718131,
      "learning_rate": 9.9444236607671e-06,
      "loss": 0.5807,
      "step": 1037
    },
    {
      "epoch": 0.07615763456441392,
      "grad_norm": 0.613227337395032,
      "learning_rate": 9.944246855067749e-06,
      "loss": 0.5777,
      "step": 1038
    },
    {
      "epoch": 0.07623100415455304,
      "grad_norm": 0.6578320249194508,
      "learning_rate": 9.944069770154537e-06,
      "loss": 0.6008,
      "step": 1039
    },
    {
      "epoch": 0.07630437374469216,
      "grad_norm": 2.1494143430486763,
      "learning_rate": 9.94389240603746e-06,
      "loss": 0.9838,
      "step": 1040
    },
    {
      "epoch": 0.07630437374469216,
      "lm_loss": 1.5673,
      "step": 1040,
      "vm_loss": 0.0699
    },
    {
      "epoch": 0.07630437374469216,
      "lm_loss": 0.3636,
      "step": 1040,
      "vm_loss": 0.1157
    },
    {
      "epoch": 0.07630437374469216,
      "lm_loss": 1.2042,
      "step": 1040,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.07630437374469216,
      "lm_loss": 0.73,
      "step": 1040,
      "vm_loss": 0.1455
    },
    {
      "epoch": 0.07630437374469216,
      "lm_loss": 0.9458,
      "step": 1040,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.07630437374469216,
      "lm_loss": 0.2298,
      "step": 1040,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.07630437374469216,
      "lm_loss": 0.9567,
      "step": 1040,
      "vm_loss": 0.1717
    },
    {
      "epoch": 0.07630437374469216,
      "lm_loss": 0.323,
      "step": 1040,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.0763777433348313,
      "grad_norm": 1.859968020161217,
      "learning_rate": 9.943714762726536e-06,
      "loss": 0.9614,
      "step": 1041
    },
    {
      "epoch": 0.07645111292497042,
      "grad_norm": 0.8773661096867845,
      "learning_rate": 9.943536840231795e-06,
      "loss": 0.5861,
      "step": 1042
    },
    {
      "epoch": 0.07652448251510954,
      "grad_norm": 1.739050225992769,
      "learning_rate": 9.943358638563288e-06,
      "loss": 1.0411,
      "step": 1043
    },
    {
      "epoch": 0.07659785210524868,
      "grad_norm": 1.8261359515133089,
      "learning_rate": 9.943180157731074e-06,
      "loss": 0.9008,
      "step": 1044
    },
    {
      "epoch": 0.0766712216953878,
      "grad_norm": 1.7424081347121263,
      "learning_rate": 9.943001397745238e-06,
      "loss": 0.902,
      "step": 1045
    },
    {
      "epoch": 0.07674459128552694,
      "grad_norm": 1.6536217084848797,
      "learning_rate": 9.942822358615868e-06,
      "loss": 0.9928,
      "step": 1046
    },
    {
      "epoch": 0.07681796087566606,
      "grad_norm": 1.876960900951353,
      "learning_rate": 9.942643040353083e-06,
      "loss": 0.9627,
      "step": 1047
    },
    {
      "epoch": 0.07689133046580518,
      "grad_norm": 1.8304232737827666,
      "learning_rate": 9.942463442967004e-06,
      "loss": 0.9739,
      "step": 1048
    },
    {
      "epoch": 0.07689133046580518,
      "lm_loss": 0.8944,
      "step": 1048,
      "vm_loss": 0.0611
    },
    {
      "epoch": 0.07689133046580518,
      "lm_loss": 1.2356,
      "step": 1048,
      "vm_loss": 0.171
    },
    {
      "epoch": 0.07689133046580518,
      "lm_loss": 0.6715,
      "step": 1048,
      "vm_loss": 0.1298
    },
    {
      "epoch": 0.07689133046580518,
      "lm_loss": 0.2457,
      "step": 1048,
      "vm_loss": 0.0883
    },
    {
      "epoch": 0.07689133046580518,
      "lm_loss": 0.0833,
      "step": 1048,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.07689133046580518,
      "lm_loss": 0.191,
      "step": 1048,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.07689133046580518,
      "lm_loss": 0.4165,
      "step": 1048,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.07689133046580518,
      "lm_loss": 1.0714,
      "step": 1048,
      "vm_loss": 0.1498
    },
    {
      "epoch": 0.07696470005594432,
      "grad_norm": 1.7695300521628412,
      "learning_rate": 9.942283566467773e-06,
      "loss": 0.8972,
      "step": 1049
    },
    {
      "epoch": 0.07703806964608344,
      "grad_norm": 0.7833105567952089,
      "learning_rate": 9.942103410865551e-06,
      "loss": 0.5885,
      "step": 1050
    },
    {
      "epoch": 0.07711143923622257,
      "grad_norm": 1.8819787268309371,
      "learning_rate": 9.94192297617051e-06,
      "loss": 0.9157,
      "step": 1051
    },
    {
      "epoch": 0.0771848088263617,
      "grad_norm": 1.8320262175013102,
      "learning_rate": 9.94174226239284e-06,
      "loss": 0.8772,
      "step": 1052
    },
    {
      "epoch": 0.07725817841650082,
      "grad_norm": 0.6574758591296698,
      "learning_rate": 9.941561269542748e-06,
      "loss": 0.5952,
      "step": 1053
    },
    {
      "epoch": 0.07733154800663995,
      "grad_norm": 0.6267541369334355,
      "learning_rate": 9.941379997630452e-06,
      "loss": 0.5829,
      "step": 1054
    },
    {
      "epoch": 0.07740491759677907,
      "grad_norm": 1.7855567665013272,
      "learning_rate": 9.94119844666619e-06,
      "loss": 0.8319,
      "step": 1055
    },
    {
      "epoch": 0.0774782871869182,
      "grad_norm": 1.6957046650883003,
      "learning_rate": 9.941016616660215e-06,
      "loss": 0.8296,
      "step": 1056
    },
    {
      "epoch": 0.0774782871869182,
      "lm_loss": 1.1217,
      "step": 1056,
      "vm_loss": 0.0473
    },
    {
      "epoch": 0.0774782871869182,
      "lm_loss": 1.061,
      "step": 1056,
      "vm_loss": 0.1673
    },
    {
      "epoch": 0.0774782871869182,
      "lm_loss": 0.3996,
      "step": 1056,
      "vm_loss": 0.1541
    },
    {
      "epoch": 0.0774782871869182,
      "lm_loss": 0.0878,
      "step": 1056,
      "vm_loss": 0.1762
    },
    {
      "epoch": 0.0774782871869182,
      "lm_loss": 1.009,
      "step": 1056,
      "vm_loss": 0.1724
    },
    {
      "epoch": 0.0774782871869182,
      "lm_loss": 1.2619,
      "step": 1056,
      "vm_loss": 0.1266
    },
    {
      "epoch": 0.0774782871869182,
      "lm_loss": 0.5359,
      "step": 1056,
      "vm_loss": 0.0722
    },
    {
      "epoch": 0.0774782871869182,
      "lm_loss": 0.7402,
      "step": 1056,
      "vm_loss": 0.174
    },
    {
      "epoch": 0.07755165677705733,
      "grad_norm": 1.8514155525607165,
      "learning_rate": 9.940834507622796e-06,
      "loss": 1.0241,
      "step": 1057
    },
    {
      "epoch": 0.07762502636719645,
      "grad_norm": 1.7123923873243556,
      "learning_rate": 9.940652119564216e-06,
      "loss": 0.9211,
      "step": 1058
    },
    {
      "epoch": 0.07769839595733559,
      "grad_norm": 0.7059099692032161,
      "learning_rate": 9.940469452494778e-06,
      "loss": 0.5934,
      "step": 1059
    },
    {
      "epoch": 0.07777176554747471,
      "grad_norm": 1.8856644555924766,
      "learning_rate": 9.940286506424792e-06,
      "loss": 1.037,
      "step": 1060
    },
    {
      "epoch": 0.07784513513761383,
      "grad_norm": 1.7558863080323384,
      "learning_rate": 9.940103281364595e-06,
      "loss": 0.8405,
      "step": 1061
    },
    {
      "epoch": 0.07791850472775297,
      "grad_norm": 1.5293651384781013,
      "learning_rate": 9.939919777324531e-06,
      "loss": 0.8415,
      "step": 1062
    },
    {
      "epoch": 0.07799187431789209,
      "grad_norm": 0.6402387890573057,
      "learning_rate": 9.939735994314963e-06,
      "loss": 0.5668,
      "step": 1063
    },
    {
      "epoch": 0.07806524390803121,
      "grad_norm": 1.9704751822297093,
      "learning_rate": 9.93955193234627e-06,
      "loss": 0.8046,
      "step": 1064
    },
    {
      "epoch": 0.07806524390803121,
      "lm_loss": 1.9051,
      "step": 1064,
      "vm_loss": 0.0374
    },
    {
      "epoch": 0.07806524390803121,
      "lm_loss": 1.4886,
      "step": 1064,
      "vm_loss": 0.1501
    },
    {
      "epoch": 0.07806524390803121,
      "lm_loss": 0.3601,
      "step": 1064,
      "vm_loss": 0.1404
    },
    {
      "epoch": 0.07806524390803121,
      "lm_loss": 0.5437,
      "step": 1064,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.07806524390803121,
      "lm_loss": 0.8135,
      "step": 1064,
      "vm_loss": 0.092
    },
    {
      "epoch": 0.07806524390803121,
      "lm_loss": 0.7164,
      "step": 1064,
      "vm_loss": 0.0952
    },
    {
      "epoch": 0.07806524390803121,
      "lm_loss": 0.8263,
      "step": 1064,
      "vm_loss": 0.2073
    },
    {
      "epoch": 0.07806524390803121,
      "lm_loss": 1.0113,
      "step": 1064,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.07813861349817035,
      "grad_norm": 1.8812744970845894,
      "learning_rate": 9.939367591428848e-06,
      "loss": 0.944,
      "step": 1065
    },
    {
      "epoch": 0.07821198308830947,
      "grad_norm": 1.6888723527718021,
      "learning_rate": 9.939182971573105e-06,
      "loss": 0.8729,
      "step": 1066
    },
    {
      "epoch": 0.0782853526784486,
      "grad_norm": 1.8453709093344695,
      "learning_rate": 9.938998072789467e-06,
      "loss": 0.8635,
      "step": 1067
    },
    {
      "epoch": 0.07835872226858773,
      "grad_norm": 1.7311297556712841,
      "learning_rate": 9.938812895088379e-06,
      "loss": 0.8781,
      "step": 1068
    },
    {
      "epoch": 0.07843209185872685,
      "grad_norm": 1.8024405270748012,
      "learning_rate": 9.938627438480295e-06,
      "loss": 0.8881,
      "step": 1069
    },
    {
      "epoch": 0.07850546144886598,
      "grad_norm": 1.7662469066938085,
      "learning_rate": 9.938441702975689e-06,
      "loss": 0.9055,
      "step": 1070
    },
    {
      "epoch": 0.0785788310390051,
      "grad_norm": 1.6846284966168574,
      "learning_rate": 9.93825568858505e-06,
      "loss": 0.8683,
      "step": 1071
    },
    {
      "epoch": 0.07865220062914424,
      "grad_norm": 1.7178678025601202,
      "learning_rate": 9.938069395318883e-06,
      "loss": 0.8838,
      "step": 1072
    },
    {
      "epoch": 0.07865220062914424,
      "lm_loss": 0.4329,
      "step": 1072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.07865220062914424,
      "lm_loss": 0.5188,
      "step": 1072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.07865220062914424,
      "lm_loss": 0.6122,
      "step": 1072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.07865220062914424,
      "lm_loss": 0.6012,
      "step": 1072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.07865220062914424,
      "lm_loss": 0.6303,
      "step": 1072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.07865220062914424,
      "lm_loss": 0.6557,
      "step": 1072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.07865220062914424,
      "lm_loss": 0.6607,
      "step": 1072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.07865220062914424,
      "lm_loss": 0.6067,
      "step": 1072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.07872557021928336,
      "grad_norm": 0.7230211673194481,
      "learning_rate": 9.93788282318771e-06,
      "loss": 0.5807,
      "step": 1073
    },
    {
      "epoch": 0.07879893980942249,
      "grad_norm": 1.8529631636683834,
      "learning_rate": 9.937695972202063e-06,
      "loss": 0.8317,
      "step": 1074
    },
    {
      "epoch": 0.07887230939956162,
      "grad_norm": 0.646549995063564,
      "learning_rate": 9.937508842372499e-06,
      "loss": 0.5728,
      "step": 1075
    },
    {
      "epoch": 0.07894567898970074,
      "grad_norm": 1.7828489340246887,
      "learning_rate": 9.937321433709581e-06,
      "loss": 0.9514,
      "step": 1076
    },
    {
      "epoch": 0.07901904857983987,
      "grad_norm": 0.6410402959785697,
      "learning_rate": 9.937133746223894e-06,
      "loss": 0.5762,
      "step": 1077
    },
    {
      "epoch": 0.079092418169979,
      "grad_norm": 1.6989024717209553,
      "learning_rate": 9.93694577992604e-06,
      "loss": 0.9362,
      "step": 1078
    },
    {
      "epoch": 0.07916578776011812,
      "grad_norm": 0.6775256247539853,
      "learning_rate": 9.93675753482663e-06,
      "loss": 0.601,
      "step": 1079
    },
    {
      "epoch": 0.07923915735025726,
      "grad_norm": 1.932438969649081,
      "learning_rate": 9.936569010936299e-06,
      "loss": 0.9269,
      "step": 1080
    },
    {
      "epoch": 0.07923915735025726,
      "lm_loss": 1.5436,
      "step": 1080,
      "vm_loss": 0.0737
    },
    {
      "epoch": 0.07923915735025726,
      "lm_loss": 1.186,
      "step": 1080,
      "vm_loss": 0.2012
    },
    {
      "epoch": 0.07923915735025726,
      "lm_loss": 0.3272,
      "step": 1080,
      "vm_loss": 0.1314
    },
    {
      "epoch": 0.07923915735025726,
      "lm_loss": 0.7128,
      "step": 1080,
      "vm_loss": 0.089
    },
    {
      "epoch": 0.07923915735025726,
      "lm_loss": 0.0489,
      "step": 1080,
      "vm_loss": 0.1556
    },
    {
      "epoch": 0.07923915735025726,
      "lm_loss": 0.9851,
      "step": 1080,
      "vm_loss": 0.0748
    },
    {
      "epoch": 0.07923915735025726,
      "lm_loss": 1.7313,
      "step": 1080,
      "vm_loss": 0.1741
    },
    {
      "epoch": 0.07923915735025726,
      "lm_loss": 0.1524,
      "step": 1080,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.07931252694039638,
      "grad_norm": 1.7585029322567842,
      "learning_rate": 9.936380208265688e-06,
      "loss": 0.8705,
      "step": 1081
    },
    {
      "epoch": 0.0793858965305355,
      "grad_norm": 1.922094613709717,
      "learning_rate": 9.936191126825462e-06,
      "loss": 0.8146,
      "step": 1082
    },
    {
      "epoch": 0.07945926612067464,
      "grad_norm": 1.7299947106558793,
      "learning_rate": 9.9360017666263e-06,
      "loss": 0.8982,
      "step": 1083
    },
    {
      "epoch": 0.07953263571081376,
      "grad_norm": 0.6616548419607162,
      "learning_rate": 9.935812127678894e-06,
      "loss": 0.576,
      "step": 1084
    },
    {
      "epoch": 0.07960600530095288,
      "grad_norm": 0.6712974935962605,
      "learning_rate": 9.935622209993954e-06,
      "loss": 0.5786,
      "step": 1085
    },
    {
      "epoch": 0.07967937489109202,
      "grad_norm": 1.836726785036467,
      "learning_rate": 9.935432013582204e-06,
      "loss": 0.8762,
      "step": 1086
    },
    {
      "epoch": 0.07975274448123114,
      "grad_norm": 1.8282014725322193,
      "learning_rate": 9.935241538454387e-06,
      "loss": 0.9382,
      "step": 1087
    },
    {
      "epoch": 0.07982611407137027,
      "grad_norm": 1.6152446849266007,
      "learning_rate": 9.935050784621258e-06,
      "loss": 0.9826,
      "step": 1088
    },
    {
      "epoch": 0.07982611407137027,
      "lm_loss": 1.5281,
      "step": 1088,
      "vm_loss": 0.2187
    },
    {
      "epoch": 0.07982611407137027,
      "lm_loss": 1.0682,
      "step": 1088,
      "vm_loss": 0.0819
    },
    {
      "epoch": 0.07982611407137027,
      "lm_loss": 0.531,
      "step": 1088,
      "vm_loss": 0.0707
    },
    {
      "epoch": 0.07982611407137027,
      "lm_loss": 0.7411,
      "step": 1088,
      "vm_loss": 0.1862
    },
    {
      "epoch": 0.07982611407137027,
      "lm_loss": 0.0782,
      "step": 1088,
      "vm_loss": 0.2459
    },
    {
      "epoch": 0.07982611407137027,
      "lm_loss": 1.0897,
      "step": 1088,
      "vm_loss": 0.162
    },
    {
      "epoch": 0.07982611407137027,
      "lm_loss": 1.0456,
      "step": 1088,
      "vm_loss": 0.1733
    },
    {
      "epoch": 0.07982611407137027,
      "lm_loss": 0.4311,
      "step": 1088,
      "vm_loss": 0.056
    },
    {
      "epoch": 0.0798994836615094,
      "grad_norm": 1.7758837047246265,
      "learning_rate": 9.93485975209359e-06,
      "loss": 0.9709,
      "step": 1089
    },
    {
      "epoch": 0.07997285325164852,
      "grad_norm": 1.7565799993071227,
      "learning_rate": 9.93466844088217e-06,
      "loss": 0.9621,
      "step": 1090
    },
    {
      "epoch": 0.08004622284178765,
      "grad_norm": 1.8902571141511544,
      "learning_rate": 9.934476850997802e-06,
      "loss": 0.8523,
      "step": 1091
    },
    {
      "epoch": 0.08011959243192678,
      "grad_norm": 0.915174367026828,
      "learning_rate": 9.934284982451308e-06,
      "loss": 0.6004,
      "step": 1092
    },
    {
      "epoch": 0.0801929620220659,
      "grad_norm": 1.7642838910827348,
      "learning_rate": 9.934092835253521e-06,
      "loss": 0.9294,
      "step": 1093
    },
    {
      "epoch": 0.08026633161220503,
      "grad_norm": 2.0009511173256778,
      "learning_rate": 9.933900409415293e-06,
      "loss": 0.8562,
      "step": 1094
    },
    {
      "epoch": 0.08033970120234415,
      "grad_norm": 0.6898073092570608,
      "learning_rate": 9.933707704947489e-06,
      "loss": 0.5818,
      "step": 1095
    },
    {
      "epoch": 0.08041307079248329,
      "grad_norm": 2.057389215358504,
      "learning_rate": 9.933514721860992e-06,
      "loss": 0.8988,
      "step": 1096
    },
    {
      "epoch": 0.08041307079248329,
      "lm_loss": 1.0945,
      "step": 1096,
      "vm_loss": 0.0572
    },
    {
      "epoch": 0.08041307079248329,
      "lm_loss": 1.4691,
      "step": 1096,
      "vm_loss": 0.1515
    },
    {
      "epoch": 0.08041307079248329,
      "lm_loss": 0.2021,
      "step": 1096,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.08041307079248329,
      "lm_loss": 0.2145,
      "step": 1096,
      "vm_loss": 0.1648
    },
    {
      "epoch": 0.08041307079248329,
      "lm_loss": 1.0507,
      "step": 1096,
      "vm_loss": 0.2648
    },
    {
      "epoch": 0.08041307079248329,
      "lm_loss": 0.3681,
      "step": 1096,
      "vm_loss": 0.129
    },
    {
      "epoch": 0.08041307079248329,
      "lm_loss": 0.6399,
      "step": 1096,
      "vm_loss": 0.1224
    },
    {
      "epoch": 0.08041307079248329,
      "lm_loss": 1.2091,
      "step": 1096,
      "vm_loss": 0.1974
    },
    {
      "epoch": 0.08048644038262241,
      "grad_norm": 1.6343669595705752,
      "learning_rate": 9.933321460166704e-06,
      "loss": 0.9366,
      "step": 1097
    },
    {
      "epoch": 0.08055980997276153,
      "grad_norm": 1.9556688691653314,
      "learning_rate": 9.933127919875533e-06,
      "loss": 0.8934,
      "step": 1098
    },
    {
      "epoch": 0.08063317956290067,
      "grad_norm": 0.9095465724321892,
      "learning_rate": 9.932934100998415e-06,
      "loss": 0.5659,
      "step": 1099
    },
    {
      "epoch": 0.08070654915303979,
      "grad_norm": 1.9457997224311718,
      "learning_rate": 9.93274000354629e-06,
      "loss": 0.8195,
      "step": 1100
    },
    {
      "epoch": 0.08077991874317893,
      "grad_norm": 1.602295677140399,
      "learning_rate": 9.932545627530122e-06,
      "loss": 0.84,
      "step": 1101
    },
    {
      "epoch": 0.08085328833331805,
      "grad_norm": 1.8232410204093423,
      "learning_rate": 9.932350972960885e-06,
      "loss": 0.9769,
      "step": 1102
    },
    {
      "epoch": 0.08092665792345717,
      "grad_norm": 0.6625786115907959,
      "learning_rate": 9.932156039849577e-06,
      "loss": 0.5777,
      "step": 1103
    },
    {
      "epoch": 0.0810000275135963,
      "grad_norm": 1.831346333336728,
      "learning_rate": 9.931960828207201e-06,
      "loss": 0.9338,
      "step": 1104
    },
    {
      "epoch": 0.0810000275135963,
      "lm_loss": 1.5636,
      "step": 1104,
      "vm_loss": 0.0669
    },
    {
      "epoch": 0.0810000275135963,
      "lm_loss": 1.2431,
      "step": 1104,
      "vm_loss": 0.1786
    },
    {
      "epoch": 0.0810000275135963,
      "lm_loss": 1.0125,
      "step": 1104,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.0810000275135963,
      "lm_loss": 0.5909,
      "step": 1104,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.0810000275135963,
      "lm_loss": 0.4951,
      "step": 1104,
      "vm_loss": 0.1991
    },
    {
      "epoch": 0.0810000275135963,
      "lm_loss": 0.9098,
      "step": 1104,
      "vm_loss": 0.1502
    },
    {
      "epoch": 0.0810000275135963,
      "lm_loss": 0.209,
      "step": 1104,
      "vm_loss": 0.0536
    },
    {
      "epoch": 0.0810000275135963,
      "lm_loss": 0.97,
      "step": 1104,
      "vm_loss": 0.0896
    },
    {
      "epoch": 0.08107339710373543,
      "grad_norm": 2.0437888811032128,
      "learning_rate": 9.931765338044782e-06,
      "loss": 0.9698,
      "step": 1105
    },
    {
      "epoch": 0.08114676669387455,
      "grad_norm": 1.652007232950619,
      "learning_rate": 9.931569569373364e-06,
      "loss": 0.9182,
      "step": 1106
    },
    {
      "epoch": 0.08122013628401369,
      "grad_norm": 2.0500439284613567,
      "learning_rate": 9.931373522203998e-06,
      "loss": 0.851,
      "step": 1107
    },
    {
      "epoch": 0.08129350587415281,
      "grad_norm": 1.8592201708662492,
      "learning_rate": 9.931177196547757e-06,
      "loss": 0.9177,
      "step": 1108
    },
    {
      "epoch": 0.08136687546429194,
      "grad_norm": 1.7144052978699085,
      "learning_rate": 9.930980592415728e-06,
      "loss": 0.8936,
      "step": 1109
    },
    {
      "epoch": 0.08144024505443107,
      "grad_norm": 1.831201807449431,
      "learning_rate": 9.930783709819014e-06,
      "loss": 0.9332,
      "step": 1110
    },
    {
      "epoch": 0.08151361464457019,
      "grad_norm": 0.8261696093125925,
      "learning_rate": 9.930586548768733e-06,
      "loss": 0.5838,
      "step": 1111
    },
    {
      "epoch": 0.08158698423470932,
      "grad_norm": 1.7322400424438886,
      "learning_rate": 9.93038910927602e-06,
      "loss": 0.9469,
      "step": 1112
    },
    {
      "epoch": 0.08158698423470932,
      "lm_loss": 0.3933,
      "step": 1112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08158698423470932,
      "lm_loss": 0.6843,
      "step": 1112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08158698423470932,
      "lm_loss": 0.6214,
      "step": 1112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08158698423470932,
      "lm_loss": 0.6497,
      "step": 1112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08158698423470932,
      "lm_loss": 0.5087,
      "step": 1112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08158698423470932,
      "lm_loss": 0.8293,
      "step": 1112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08158698423470932,
      "lm_loss": 0.6553,
      "step": 1112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08158698423470932,
      "lm_loss": 0.5824,
      "step": 1112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08166035382484844,
      "grad_norm": 0.6178862419948691,
      "learning_rate": 9.930191391352023e-06,
      "loss": 0.585,
      "step": 1113
    },
    {
      "epoch": 0.08173372341498757,
      "grad_norm": 1.8322824646996465,
      "learning_rate": 9.929993395007908e-06,
      "loss": 0.9347,
      "step": 1114
    },
    {
      "epoch": 0.0818070930051267,
      "grad_norm": 1.5294863314776457,
      "learning_rate": 9.929795120254857e-06,
      "loss": 0.9524,
      "step": 1115
    },
    {
      "epoch": 0.08188046259526582,
      "grad_norm": 1.7325486876419043,
      "learning_rate": 9.929596567104069e-06,
      "loss": 0.8152,
      "step": 1116
    },
    {
      "epoch": 0.08195383218540496,
      "grad_norm": 1.7142518426733286,
      "learning_rate": 9.929397735566754e-06,
      "loss": 0.844,
      "step": 1117
    },
    {
      "epoch": 0.08202720177554408,
      "grad_norm": 2.0383919076363783,
      "learning_rate": 9.92919862565414e-06,
      "loss": 0.8952,
      "step": 1118
    },
    {
      "epoch": 0.0821005713656832,
      "grad_norm": 1.720359506995676,
      "learning_rate": 9.928999237377475e-06,
      "loss": 0.8927,
      "step": 1119
    },
    {
      "epoch": 0.08217394095582234,
      "grad_norm": 1.8088932526717139,
      "learning_rate": 9.928799570748016e-06,
      "loss": 0.962,
      "step": 1120
    },
    {
      "epoch": 0.08217394095582234,
      "lm_loss": 1.5611,
      "step": 1120,
      "vm_loss": 0.0254
    },
    {
      "epoch": 0.08217394095582234,
      "lm_loss": 1.0432,
      "step": 1120,
      "vm_loss": 0.0894
    },
    {
      "epoch": 0.08217394095582234,
      "lm_loss": 0.3329,
      "step": 1120,
      "vm_loss": 0.1385
    },
    {
      "epoch": 0.08217394095582234,
      "lm_loss": 0.8175,
      "step": 1120,
      "vm_loss": 0.1341
    },
    {
      "epoch": 0.08217394095582234,
      "lm_loss": 0.3488,
      "step": 1120,
      "vm_loss": 0.1328
    },
    {
      "epoch": 0.08217394095582234,
      "lm_loss": 0.7964,
      "step": 1120,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.08217394095582234,
      "lm_loss": 1.0074,
      "step": 1120,
      "vm_loss": 0.1402
    },
    {
      "epoch": 0.08217394095582234,
      "lm_loss": 0.2593,
      "step": 1120,
      "vm_loss": 0.1597
    },
    {
      "epoch": 0.08224731054596146,
      "grad_norm": 1.849467508323586,
      "learning_rate": 9.928599625777037e-06,
      "loss": 0.9145,
      "step": 1121
    },
    {
      "epoch": 0.0823206801361006,
      "grad_norm": 1.849158249543982,
      "learning_rate": 9.928399402475835e-06,
      "loss": 1.034,
      "step": 1122
    },
    {
      "epoch": 0.08239404972623972,
      "grad_norm": 1.8668563925287764,
      "learning_rate": 9.928198900855712e-06,
      "loss": 0.9627,
      "step": 1123
    },
    {
      "epoch": 0.08246741931637884,
      "grad_norm": 1.6901666426945292,
      "learning_rate": 9.927998120927994e-06,
      "loss": 0.9563,
      "step": 1124
    },
    {
      "epoch": 0.08254078890651798,
      "grad_norm": 1.0002442158751372,
      "learning_rate": 9.927797062704016e-06,
      "loss": 0.6025,
      "step": 1125
    },
    {
      "epoch": 0.0826141584966571,
      "grad_norm": 1.9305374036408065,
      "learning_rate": 9.927595726195135e-06,
      "loss": 0.9985,
      "step": 1126
    },
    {
      "epoch": 0.08268752808679622,
      "grad_norm": 1.7122060860471497,
      "learning_rate": 9.92739411141272e-06,
      "loss": 0.958,
      "step": 1127
    },
    {
      "epoch": 0.08276089767693535,
      "grad_norm": 1.6924888940192777,
      "learning_rate": 9.927192218368159e-06,
      "loss": 1.039,
      "step": 1128
    },
    {
      "epoch": 0.08276089767693535,
      "lm_loss": 1.3641,
      "step": 1128,
      "vm_loss": 0.0898
    },
    {
      "epoch": 0.08276089767693535,
      "lm_loss": 2.8225,
      "step": 1128,
      "vm_loss": 0.1585
    },
    {
      "epoch": 0.08276089767693535,
      "lm_loss": 1.2381,
      "step": 1128,
      "vm_loss": 0.1376
    },
    {
      "epoch": 0.08276089767693535,
      "lm_loss": 1.0527,
      "step": 1128,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.08276089767693535,
      "lm_loss": 0.5705,
      "step": 1128,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.08276089767693535,
      "lm_loss": 0.2775,
      "step": 1128,
      "vm_loss": 0.1833
    },
    {
      "epoch": 0.08276089767693535,
      "lm_loss": 1.2079,
      "step": 1128,
      "vm_loss": 0.2587
    },
    {
      "epoch": 0.08276089767693535,
      "lm_loss": 0.0593,
      "step": 1128,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.08283426726707448,
      "grad_norm": 1.747120521767331,
      "learning_rate": 9.926990047072849e-06,
      "loss": 1.019,
      "step": 1129
    },
    {
      "epoch": 0.08290763685721361,
      "grad_norm": 1.91997221529123,
      "learning_rate": 9.926787597538211e-06,
      "loss": 0.9765,
      "step": 1130
    },
    {
      "epoch": 0.08298100644735273,
      "grad_norm": 2.0254098719121054,
      "learning_rate": 9.926584869775677e-06,
      "loss": 0.9274,
      "step": 1131
    },
    {
      "epoch": 0.08305437603749186,
      "grad_norm": 1.592453192088265,
      "learning_rate": 9.926381863796693e-06,
      "loss": 0.8806,
      "step": 1132
    },
    {
      "epoch": 0.08312774562763099,
      "grad_norm": 1.9678970652317118,
      "learning_rate": 9.926178579612726e-06,
      "loss": 0.9844,
      "step": 1133
    },
    {
      "epoch": 0.08320111521777011,
      "grad_norm": 1.8642814649506376,
      "learning_rate": 9.925975017235256e-06,
      "loss": 0.8584,
      "step": 1134
    },
    {
      "epoch": 0.08327448480790924,
      "grad_norm": 0.9085310724961425,
      "learning_rate": 9.925771176675779e-06,
      "loss": 0.5804,
      "step": 1135
    },
    {
      "epoch": 0.08334785439804837,
      "grad_norm": 1.66293600139992,
      "learning_rate": 9.925567057945802e-06,
      "loss": 0.8041,
      "step": 1136
    },
    {
      "epoch": 0.08334785439804837,
      "lm_loss": 0.6231,
      "step": 1136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08334785439804837,
      "lm_loss": 0.3787,
      "step": 1136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08334785439804837,
      "lm_loss": 0.6206,
      "step": 1136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08334785439804837,
      "lm_loss": 0.7321,
      "step": 1136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08334785439804837,
      "lm_loss": 0.3984,
      "step": 1136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08334785439804837,
      "lm_loss": 0.691,
      "step": 1136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08334785439804837,
      "lm_loss": 0.6599,
      "step": 1136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08334785439804837,
      "lm_loss": 0.8149,
      "step": 1136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08342122398818749,
      "grad_norm": 0.6200633562094147,
      "learning_rate": 9.925362661056856e-06,
      "loss": 0.5901,
      "step": 1137
    },
    {
      "epoch": 0.08349459357832663,
      "grad_norm": 0.6191993345573936,
      "learning_rate": 9.925157986020483e-06,
      "loss": 0.5886,
      "step": 1138
    },
    {
      "epoch": 0.08356796316846575,
      "grad_norm": 1.9825236653693548,
      "learning_rate": 9.924953032848243e-06,
      "loss": 0.7828,
      "step": 1139
    },
    {
      "epoch": 0.08364133275860487,
      "grad_norm": 0.7394946441659106,
      "learning_rate": 9.924747801551708e-06,
      "loss": 0.5827,
      "step": 1140
    },
    {
      "epoch": 0.08371470234874401,
      "grad_norm": 1.7874760580790403,
      "learning_rate": 9.92454229214247e-06,
      "loss": 0.9071,
      "step": 1141
    },
    {
      "epoch": 0.08378807193888313,
      "grad_norm": 1.7834732857259148,
      "learning_rate": 9.92433650463213e-06,
      "loss": 0.8963,
      "step": 1142
    },
    {
      "epoch": 0.08386144152902227,
      "grad_norm": 1.5693351466958338,
      "learning_rate": 9.924130439032316e-06,
      "loss": 0.8624,
      "step": 1143
    },
    {
      "epoch": 0.08393481111916139,
      "grad_norm": 0.6410864941849718,
      "learning_rate": 9.92392409535466e-06,
      "loss": 0.5688,
      "step": 1144
    },
    {
      "epoch": 0.08393481111916139,
      "lm_loss": 0.5399,
      "step": 1144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08393481111916139,
      "lm_loss": 0.6614,
      "step": 1144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08393481111916139,
      "lm_loss": 0.6252,
      "step": 1144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08393481111916139,
      "lm_loss": 0.6639,
      "step": 1144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08393481111916139,
      "lm_loss": 0.4764,
      "step": 1144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08393481111916139,
      "lm_loss": 0.7809,
      "step": 1144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08393481111916139,
      "lm_loss": 0.6634,
      "step": 1144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08393481111916139,
      "lm_loss": 0.4379,
      "step": 1144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08400818070930051,
      "grad_norm": 0.6191795397618424,
      "learning_rate": 9.923717473610818e-06,
      "loss": 0.5785,
      "step": 1145
    },
    {
      "epoch": 0.08408155029943964,
      "grad_norm": 1.6985106002878791,
      "learning_rate": 9.923510573812456e-06,
      "loss": 0.8574,
      "step": 1146
    },
    {
      "epoch": 0.08415491988957877,
      "grad_norm": 1.6330716745615241,
      "learning_rate": 9.923303395971257e-06,
      "loss": 0.8607,
      "step": 1147
    },
    {
      "epoch": 0.08422828947971789,
      "grad_norm": 1.6068868728376273,
      "learning_rate": 9.923095940098925e-06,
      "loss": 0.9022,
      "step": 1148
    },
    {
      "epoch": 0.08430165906985702,
      "grad_norm": 1.9591091619121426,
      "learning_rate": 9.922888206207174e-06,
      "loss": 0.936,
      "step": 1149
    },
    {
      "epoch": 0.08437502865999615,
      "grad_norm": 1.6762593671557486,
      "learning_rate": 9.922680194307735e-06,
      "loss": 0.829,
      "step": 1150
    },
    {
      "epoch": 0.08444839825013528,
      "grad_norm": 1.6906391419245004,
      "learning_rate": 9.922471904412353e-06,
      "loss": 0.9114,
      "step": 1151
    },
    {
      "epoch": 0.0845217678402744,
      "grad_norm": 1.7695345369251814,
      "learning_rate": 9.922263336532792e-06,
      "loss": 0.9991,
      "step": 1152
    },
    {
      "epoch": 0.0845217678402744,
      "lm_loss": 0.6412,
      "step": 1152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0845217678402744,
      "lm_loss": 0.6013,
      "step": 1152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0845217678402744,
      "lm_loss": 0.57,
      "step": 1152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0845217678402744,
      "lm_loss": 0.6359,
      "step": 1152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0845217678402744,
      "lm_loss": 0.6097,
      "step": 1152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0845217678402744,
      "lm_loss": 0.7464,
      "step": 1152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0845217678402744,
      "lm_loss": 0.571,
      "step": 1152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0845217678402744,
      "lm_loss": 0.5684,
      "step": 1152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08459513743041353,
      "grad_norm": 0.8272573195893036,
      "learning_rate": 9.922054490680832e-06,
      "loss": 0.5886,
      "step": 1153
    },
    {
      "epoch": 0.08466850702055266,
      "grad_norm": 0.7088217727571873,
      "learning_rate": 9.921845366868264e-06,
      "loss": 0.5888,
      "step": 1154
    },
    {
      "epoch": 0.08474187661069178,
      "grad_norm": 1.8380089621260574,
      "learning_rate": 9.9216359651069e-06,
      "loss": 0.8829,
      "step": 1155
    },
    {
      "epoch": 0.0848152462008309,
      "grad_norm": 1.952807741779425,
      "learning_rate": 9.921426285408567e-06,
      "loss": 0.9249,
      "step": 1156
    },
    {
      "epoch": 0.08488861579097004,
      "grad_norm": 0.7039326359273967,
      "learning_rate": 9.921216327785101e-06,
      "loss": 0.5783,
      "step": 1157
    },
    {
      "epoch": 0.08496198538110916,
      "grad_norm": 0.7374321218703311,
      "learning_rate": 9.921006092248363e-06,
      "loss": 0.5992,
      "step": 1158
    },
    {
      "epoch": 0.0850353549712483,
      "grad_norm": 2.188961798937895,
      "learning_rate": 9.920795578810223e-06,
      "loss": 1.023,
      "step": 1159
    },
    {
      "epoch": 0.08510872456138742,
      "grad_norm": 1.8322221693844551,
      "learning_rate": 9.920584787482573e-06,
      "loss": 0.8852,
      "step": 1160
    },
    {
      "epoch": 0.08510872456138742,
      "lm_loss": 1.3229,
      "step": 1160,
      "vm_loss": 0.069
    },
    {
      "epoch": 0.08510872456138742,
      "lm_loss": 0.8111,
      "step": 1160,
      "vm_loss": 0.1766
    },
    {
      "epoch": 0.08510872456138742,
      "lm_loss": 0.1544,
      "step": 1160,
      "vm_loss": 0.1545
    },
    {
      "epoch": 0.08510872456138742,
      "lm_loss": 0.349,
      "step": 1160,
      "vm_loss": 0.1422
    },
    {
      "epoch": 0.08510872456138742,
      "lm_loss": 0.7382,
      "step": 1160,
      "vm_loss": 0.1685
    },
    {
      "epoch": 0.08510872456138742,
      "lm_loss": 0.4185,
      "step": 1160,
      "vm_loss": 0.1305
    },
    {
      "epoch": 0.08510872456138742,
      "lm_loss": 0.198,
      "step": 1160,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.08510872456138742,
      "lm_loss": 0.5802,
      "step": 1160,
      "vm_loss": 0.1028
    },
    {
      "epoch": 0.08518209415152654,
      "grad_norm": 1.9414508740742136,
      "learning_rate": 9.920373718277312e-06,
      "loss": 0.9225,
      "step": 1161
    },
    {
      "epoch": 0.08525546374166568,
      "grad_norm": 0.7331894500973415,
      "learning_rate": 9.920162371206365e-06,
      "loss": 0.5762,
      "step": 1162
    },
    {
      "epoch": 0.0853288333318048,
      "grad_norm": 0.6255471126604872,
      "learning_rate": 9.919950746281661e-06,
      "loss": 0.5646,
      "step": 1163
    },
    {
      "epoch": 0.08540220292194392,
      "grad_norm": 0.6017088525003954,
      "learning_rate": 9.919738843515157e-06,
      "loss": 0.5753,
      "step": 1164
    },
    {
      "epoch": 0.08547557251208306,
      "grad_norm": 1.9708873645628102,
      "learning_rate": 9.919526662918815e-06,
      "loss": 0.9978,
      "step": 1165
    },
    {
      "epoch": 0.08554894210222218,
      "grad_norm": 1.7471287668676778,
      "learning_rate": 9.91931420450462e-06,
      "loss": 0.9457,
      "step": 1166
    },
    {
      "epoch": 0.08562231169236131,
      "grad_norm": 1.6440461726129125,
      "learning_rate": 9.91910146828457e-06,
      "loss": 0.8926,
      "step": 1167
    },
    {
      "epoch": 0.08569568128250044,
      "grad_norm": 1.7901259249382742,
      "learning_rate": 9.918888454270678e-06,
      "loss": 0.9044,
      "step": 1168
    },
    {
      "epoch": 0.08569568128250044,
      "lm_loss": 2.1009,
      "step": 1168,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.08569568128250044,
      "lm_loss": 0.301,
      "step": 1168,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.08569568128250044,
      "lm_loss": 1.7193,
      "step": 1168,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.08569568128250044,
      "lm_loss": 0.8498,
      "step": 1168,
      "vm_loss": 0.1877
    },
    {
      "epoch": 0.08569568128250044,
      "lm_loss": 0.2044,
      "step": 1168,
      "vm_loss": 0.0869
    },
    {
      "epoch": 0.08569568128250044,
      "lm_loss": 0.4137,
      "step": 1168,
      "vm_loss": 0.1126
    },
    {
      "epoch": 0.08569568128250044,
      "lm_loss": 0.201,
      "step": 1168,
      "vm_loss": 0.1869
    },
    {
      "epoch": 0.08569568128250044,
      "lm_loss": 0.8514,
      "step": 1168,
      "vm_loss": 0.1153
    },
    {
      "epoch": 0.08576905087263956,
      "grad_norm": 1.6897974220442913,
      "learning_rate": 9.918675162474974e-06,
      "loss": 0.7933,
      "step": 1169
    },
    {
      "epoch": 0.08584242046277869,
      "grad_norm": 1.7686909551462133,
      "learning_rate": 9.9184615929095e-06,
      "loss": 0.8751,
      "step": 1170
    },
    {
      "epoch": 0.08591579005291781,
      "grad_norm": 0.9586696595524921,
      "learning_rate": 9.918247745586323e-06,
      "loss": 0.5946,
      "step": 1171
    },
    {
      "epoch": 0.08598915964305695,
      "grad_norm": 1.965985116904017,
      "learning_rate": 9.918033620517514e-06,
      "loss": 0.9304,
      "step": 1172
    },
    {
      "epoch": 0.08606252923319607,
      "grad_norm": 0.703094398984644,
      "learning_rate": 9.917819217715168e-06,
      "loss": 0.58,
      "step": 1173
    },
    {
      "epoch": 0.0861358988233352,
      "grad_norm": 1.9894484542193456,
      "learning_rate": 9.917604537191391e-06,
      "loss": 0.9152,
      "step": 1174
    },
    {
      "epoch": 0.08620926841347433,
      "grad_norm": 1.548594613904752,
      "learning_rate": 9.917389578958308e-06,
      "loss": 0.9293,
      "step": 1175
    },
    {
      "epoch": 0.08628263800361345,
      "grad_norm": 0.8055359887070436,
      "learning_rate": 9.917174343028058e-06,
      "loss": 0.5931,
      "step": 1176
    },
    {
      "epoch": 0.08628263800361345,
      "lm_loss": 0.3574,
      "step": 1176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08628263800361345,
      "lm_loss": 0.7002,
      "step": 1176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08628263800361345,
      "lm_loss": 0.6095,
      "step": 1176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08628263800361345,
      "lm_loss": 0.468,
      "step": 1176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08628263800361345,
      "lm_loss": 0.5049,
      "step": 1176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08628263800361345,
      "lm_loss": 0.6745,
      "step": 1176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08628263800361345,
      "lm_loss": 0.4688,
      "step": 1176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08628263800361345,
      "lm_loss": 0.6419,
      "step": 1176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08635600759375257,
      "grad_norm": 0.7569455883730685,
      "learning_rate": 9.916958829412795e-06,
      "loss": 0.5877,
      "step": 1177
    },
    {
      "epoch": 0.08642937718389171,
      "grad_norm": 1.7868147980041909,
      "learning_rate": 9.91674303812469e-06,
      "loss": 0.9118,
      "step": 1178
    },
    {
      "epoch": 0.08650274677403083,
      "grad_norm": 1.925618669888244,
      "learning_rate": 9.916526969175932e-06,
      "loss": 0.8988,
      "step": 1179
    },
    {
      "epoch": 0.08657611636416997,
      "grad_norm": 0.6754116308473458,
      "learning_rate": 9.916310622578718e-06,
      "loss": 0.5914,
      "step": 1180
    },
    {
      "epoch": 0.08664948595430909,
      "grad_norm": 1.6916754918795396,
      "learning_rate": 9.916093998345269e-06,
      "loss": 0.9206,
      "step": 1181
    },
    {
      "epoch": 0.08672285554444821,
      "grad_norm": 1.8311320985457453,
      "learning_rate": 9.915877096487817e-06,
      "loss": 0.941,
      "step": 1182
    },
    {
      "epoch": 0.08679622513458735,
      "grad_norm": 1.7319531611376364,
      "learning_rate": 9.915659917018613e-06,
      "loss": 0.9184,
      "step": 1183
    },
    {
      "epoch": 0.08686959472472647,
      "grad_norm": 1.7713046731475655,
      "learning_rate": 9.915442459949917e-06,
      "loss": 0.8981,
      "step": 1184
    },
    {
      "epoch": 0.08686959472472647,
      "lm_loss": 1.3461,
      "step": 1184,
      "vm_loss": 0.0626
    },
    {
      "epoch": 0.08686959472472647,
      "lm_loss": 1.196,
      "step": 1184,
      "vm_loss": 0.1885
    },
    {
      "epoch": 0.08686959472472647,
      "lm_loss": 0.8226,
      "step": 1184,
      "vm_loss": 0.2006
    },
    {
      "epoch": 0.08686959472472647,
      "lm_loss": 0.87,
      "step": 1184,
      "vm_loss": 0.1232
    },
    {
      "epoch": 0.08686959472472647,
      "lm_loss": 1.2067,
      "step": 1184,
      "vm_loss": 0.1153
    },
    {
      "epoch": 0.08686959472472647,
      "lm_loss": 0.897,
      "step": 1184,
      "vm_loss": 0.1317
    },
    {
      "epoch": 0.08686959472472647,
      "lm_loss": 1.185,
      "step": 1184,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.08686959472472647,
      "lm_loss": 0.4508,
      "step": 1184,
      "vm_loss": 0.1611
    },
    {
      "epoch": 0.08694296431486559,
      "grad_norm": 1.4985719081587288,
      "learning_rate": 9.915224725294015e-06,
      "loss": 0.9414,
      "step": 1185
    },
    {
      "epoch": 0.08701633390500473,
      "grad_norm": 1.8317397661106158,
      "learning_rate": 9.9150067130632e-06,
      "loss": 0.8977,
      "step": 1186
    },
    {
      "epoch": 0.08708970349514385,
      "grad_norm": 1.697194585543144,
      "learning_rate": 9.914788423269783e-06,
      "loss": 0.9571,
      "step": 1187
    },
    {
      "epoch": 0.08716307308528298,
      "grad_norm": 1.8161788989754055,
      "learning_rate": 9.914569855926092e-06,
      "loss": 0.9335,
      "step": 1188
    },
    {
      "epoch": 0.0872364426754221,
      "grad_norm": 0.7748509757247201,
      "learning_rate": 9.914351011044472e-06,
      "loss": 0.574,
      "step": 1189
    },
    {
      "epoch": 0.08730981226556123,
      "grad_norm": 2.1582708661154495,
      "learning_rate": 9.91413188863728e-06,
      "loss": 0.9319,
      "step": 1190
    },
    {
      "epoch": 0.08738318185570036,
      "grad_norm": 1.8152731156459667,
      "learning_rate": 9.91391248871689e-06,
      "loss": 0.878,
      "step": 1191
    },
    {
      "epoch": 0.08745655144583948,
      "grad_norm": 1.7483031652852203,
      "learning_rate": 9.913692811295695e-06,
      "loss": 0.9354,
      "step": 1192
    },
    {
      "epoch": 0.08745655144583948,
      "lm_loss": 1.5366,
      "step": 1192,
      "vm_loss": 0.173
    },
    {
      "epoch": 0.08745655144583948,
      "lm_loss": 0.868,
      "step": 1192,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.08745655144583948,
      "lm_loss": 0.1437,
      "step": 1192,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.08745655144583948,
      "lm_loss": 0.9378,
      "step": 1192,
      "vm_loss": 0.0574
    },
    {
      "epoch": 0.08745655144583948,
      "lm_loss": 0.2855,
      "step": 1192,
      "vm_loss": 0.1553
    },
    {
      "epoch": 0.08745655144583948,
      "lm_loss": 1.4268,
      "step": 1192,
      "vm_loss": 0.106
    },
    {
      "epoch": 0.08745655144583948,
      "lm_loss": 0.4005,
      "step": 1192,
      "vm_loss": 0.1419
    },
    {
      "epoch": 0.08745655144583948,
      "lm_loss": 0.5774,
      "step": 1192,
      "vm_loss": 0.1321
    },
    {
      "epoch": 0.08752992103597862,
      "grad_norm": 1.8302580851638028,
      "learning_rate": 9.913472856386094e-06,
      "loss": 0.9041,
      "step": 1193
    },
    {
      "epoch": 0.08760329062611774,
      "grad_norm": 1.7020394421522602,
      "learning_rate": 9.913252624000516e-06,
      "loss": 0.8495,
      "step": 1194
    },
    {
      "epoch": 0.08767666021625686,
      "grad_norm": 1.5273042984259382,
      "learning_rate": 9.913032114151395e-06,
      "loss": 0.8812,
      "step": 1195
    },
    {
      "epoch": 0.087750029806396,
      "grad_norm": 1.7877468261847973,
      "learning_rate": 9.912811326851183e-06,
      "loss": 0.8321,
      "step": 1196
    },
    {
      "epoch": 0.08782339939653512,
      "grad_norm": 1.7994305088644043,
      "learning_rate": 9.91259026211235e-06,
      "loss": 0.8581,
      "step": 1197
    },
    {
      "epoch": 0.08789676898667424,
      "grad_norm": 1.5824769683821502,
      "learning_rate": 9.912368919947377e-06,
      "loss": 0.8214,
      "step": 1198
    },
    {
      "epoch": 0.08797013857681338,
      "grad_norm": 0.731465417283119,
      "learning_rate": 9.912147300368766e-06,
      "loss": 0.5683,
      "step": 1199
    },
    {
      "epoch": 0.0880435081669525,
      "grad_norm": 1.9983654725398423,
      "learning_rate": 9.911925403389034e-06,
      "loss": 0.879,
      "step": 1200
    },
    {
      "epoch": 0.0880435081669525,
      "lm_loss": 0.5622,
      "step": 1200,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.0880435081669525,
      "lm_loss": 1.5334,
      "step": 1200,
      "vm_loss": 0.043
    },
    {
      "epoch": 0.0880435081669525,
      "lm_loss": 0.17,
      "step": 1200,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.0880435081669525,
      "lm_loss": 0.294,
      "step": 1200,
      "vm_loss": 0.2065
    },
    {
      "epoch": 0.0880435081669525,
      "lm_loss": 0.1025,
      "step": 1200,
      "vm_loss": 0.135
    },
    {
      "epoch": 0.0880435081669525,
      "lm_loss": 0.1501,
      "step": 1200,
      "vm_loss": 0.0959
    },
    {
      "epoch": 0.0880435081669525,
      "lm_loss": 0.7008,
      "step": 1200,
      "vm_loss": 0.1274
    },
    {
      "epoch": 0.0880435081669525,
      "lm_loss": 0.8797,
      "step": 1200,
      "vm_loss": 0.3355
    },
    {
      "epoch": 0.08811687775709164,
      "grad_norm": 1.7885248660289086,
      "learning_rate": 9.911703229020707e-06,
      "loss": 0.8883,
      "step": 1201
    },
    {
      "epoch": 0.08819024734723076,
      "grad_norm": 1.7164975765780872,
      "learning_rate": 9.911480777276338e-06,
      "loss": 0.8781,
      "step": 1202
    },
    {
      "epoch": 0.08826361693736988,
      "grad_norm": 1.822889361228979,
      "learning_rate": 9.911258048168486e-06,
      "loss": 0.8678,
      "step": 1203
    },
    {
      "epoch": 0.08833698652750901,
      "grad_norm": 0.6152897877755019,
      "learning_rate": 9.91103504170973e-06,
      "loss": 0.5803,
      "step": 1204
    },
    {
      "epoch": 0.08841035611764814,
      "grad_norm": 1.6509089735701414,
      "learning_rate": 9.910811757912662e-06,
      "loss": 0.9037,
      "step": 1205
    },
    {
      "epoch": 0.08848372570778726,
      "grad_norm": 1.857765956048378,
      "learning_rate": 9.910588196789893e-06,
      "loss": 0.8986,
      "step": 1206
    },
    {
      "epoch": 0.0885570952979264,
      "grad_norm": 1.6307479550638542,
      "learning_rate": 9.910364358354048e-06,
      "loss": 0.8167,
      "step": 1207
    },
    {
      "epoch": 0.08863046488806552,
      "grad_norm": 1.7008182304547805,
      "learning_rate": 9.910140242617766e-06,
      "loss": 0.9078,
      "step": 1208
    },
    {
      "epoch": 0.08863046488806552,
      "lm_loss": 1.3053,
      "step": 1208,
      "vm_loss": 0.1126
    },
    {
      "epoch": 0.08863046488806552,
      "lm_loss": 0.5435,
      "step": 1208,
      "vm_loss": 0.0416
    },
    {
      "epoch": 0.08863046488806552,
      "lm_loss": 0.9694,
      "step": 1208,
      "vm_loss": 0.1554
    },
    {
      "epoch": 0.08863046488806552,
      "lm_loss": 0.4264,
      "step": 1208,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.08863046488806552,
      "lm_loss": 0.9286,
      "step": 1208,
      "vm_loss": 0.2009
    },
    {
      "epoch": 0.08863046488806552,
      "lm_loss": 0.9953,
      "step": 1208,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.08863046488806552,
      "lm_loss": 1.1793,
      "step": 1208,
      "vm_loss": 0.1477
    },
    {
      "epoch": 0.08863046488806552,
      "lm_loss": 0.526,
      "step": 1208,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.08870383447820465,
      "grad_norm": 1.7142587333777974,
      "learning_rate": 9.909915849593705e-06,
      "loss": 0.9173,
      "step": 1209
    },
    {
      "epoch": 0.08877720406834377,
      "grad_norm": 1.9842226998901717,
      "learning_rate": 9.909691179294538e-06,
      "loss": 0.8228,
      "step": 1210
    },
    {
      "epoch": 0.0888505736584829,
      "grad_norm": 1.8275391181113125,
      "learning_rate": 9.909466231732952e-06,
      "loss": 0.9185,
      "step": 1211
    },
    {
      "epoch": 0.08892394324862203,
      "grad_norm": 1.8832627579355987,
      "learning_rate": 9.909241006921649e-06,
      "loss": 0.93,
      "step": 1212
    },
    {
      "epoch": 0.08899731283876115,
      "grad_norm": 1.7984331498616564,
      "learning_rate": 9.909015504873347e-06,
      "loss": 0.8835,
      "step": 1213
    },
    {
      "epoch": 0.08907068242890027,
      "grad_norm": 1.7310527526989905,
      "learning_rate": 9.908789725600785e-06,
      "loss": 0.9118,
      "step": 1214
    },
    {
      "epoch": 0.08914405201903941,
      "grad_norm": 0.6732377082752309,
      "learning_rate": 9.908563669116708e-06,
      "loss": 0.5789,
      "step": 1215
    },
    {
      "epoch": 0.08921742160917853,
      "grad_norm": 1.7325935869697184,
      "learning_rate": 9.90833733543389e-06,
      "loss": 0.8288,
      "step": 1216
    },
    {
      "epoch": 0.08921742160917853,
      "lm_loss": 1.2239,
      "step": 1216,
      "vm_loss": 0.0441
    },
    {
      "epoch": 0.08921742160917853,
      "lm_loss": 1.8744,
      "step": 1216,
      "vm_loss": 0.1036
    },
    {
      "epoch": 0.08921742160917853,
      "lm_loss": 0.7809,
      "step": 1216,
      "vm_loss": 0.1796
    },
    {
      "epoch": 0.08921742160917853,
      "lm_loss": 0.9564,
      "step": 1216,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.08921742160917853,
      "lm_loss": 1.2543,
      "step": 1216,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.08921742160917853,
      "lm_loss": 0.641,
      "step": 1216,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.08921742160917853,
      "lm_loss": 0.9085,
      "step": 1216,
      "vm_loss": 0.1066
    },
    {
      "epoch": 0.08921742160917853,
      "lm_loss": 0.3264,
      "step": 1216,
      "vm_loss": 0.1602
    },
    {
      "epoch": 0.08929079119931767,
      "grad_norm": 1.8954312587464208,
      "learning_rate": 9.908110724565102e-06,
      "loss": 0.9758,
      "step": 1217
    },
    {
      "epoch": 0.08936416078945679,
      "grad_norm": 1.8334210543331106,
      "learning_rate": 9.90788383652315e-06,
      "loss": 0.9318,
      "step": 1218
    },
    {
      "epoch": 0.08943753037959591,
      "grad_norm": 1.7092210593209451,
      "learning_rate": 9.90765667132084e-06,
      "loss": 0.9108,
      "step": 1219
    },
    {
      "epoch": 0.08951089996973505,
      "grad_norm": 0.5906226466337423,
      "learning_rate": 9.907429228971007e-06,
      "loss": 0.5798,
      "step": 1220
    },
    {
      "epoch": 0.08958426955987417,
      "grad_norm": 1.8759419942086586,
      "learning_rate": 9.907201509486491e-06,
      "loss": 0.932,
      "step": 1221
    },
    {
      "epoch": 0.0896576391500133,
      "grad_norm": 0.6055430010515298,
      "learning_rate": 9.906973512880155e-06,
      "loss": 0.5909,
      "step": 1222
    },
    {
      "epoch": 0.08973100874015243,
      "grad_norm": 0.5805371497522855,
      "learning_rate": 9.906745239164871e-06,
      "loss": 0.5985,
      "step": 1223
    },
    {
      "epoch": 0.08980437833029155,
      "grad_norm": 1.6889594644633124,
      "learning_rate": 9.906516688353533e-06,
      "loss": 0.9719,
      "step": 1224
    },
    {
      "epoch": 0.08980437833029155,
      "lm_loss": 0.6323,
      "step": 1224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08980437833029155,
      "lm_loss": 0.5767,
      "step": 1224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08980437833029155,
      "lm_loss": 0.6196,
      "step": 1224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08980437833029155,
      "lm_loss": 0.565,
      "step": 1224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08980437833029155,
      "lm_loss": 0.6498,
      "step": 1224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08980437833029155,
      "lm_loss": 0.5944,
      "step": 1224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08980437833029155,
      "lm_loss": 0.4953,
      "step": 1224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08980437833029155,
      "lm_loss": 0.692,
      "step": 1224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.08987774792043068,
      "grad_norm": 0.5705691802910875,
      "learning_rate": 9.906287860459048e-06,
      "loss": 0.5921,
      "step": 1225
    },
    {
      "epoch": 0.0899511175105698,
      "grad_norm": 1.7513962585814071,
      "learning_rate": 9.906058755494334e-06,
      "loss": 0.9554,
      "step": 1226
    },
    {
      "epoch": 0.09002448710070893,
      "grad_norm": 0.6180940524318562,
      "learning_rate": 9.905829373472336e-06,
      "loss": 0.5762,
      "step": 1227
    },
    {
      "epoch": 0.09009785669084806,
      "grad_norm": 1.691726703128518,
      "learning_rate": 9.905599714406002e-06,
      "loss": 0.8892,
      "step": 1228
    },
    {
      "epoch": 0.09017122628098719,
      "grad_norm": 1.6966838661936432,
      "learning_rate": 9.905369778308304e-06,
      "loss": 0.8598,
      "step": 1229
    },
    {
      "epoch": 0.09024459587112632,
      "grad_norm": 1.8009873823826599,
      "learning_rate": 9.905139565192227e-06,
      "loss": 0.9884,
      "step": 1230
    },
    {
      "epoch": 0.09031796546126544,
      "grad_norm": 0.6555891403540605,
      "learning_rate": 9.904909075070772e-06,
      "loss": 0.5874,
      "step": 1231
    },
    {
      "epoch": 0.09039133505140456,
      "grad_norm": 1.8590016810875414,
      "learning_rate": 9.904678307956955e-06,
      "loss": 0.8765,
      "step": 1232
    },
    {
      "epoch": 0.09039133505140456,
      "lm_loss": 1.2783,
      "step": 1232,
      "vm_loss": 0.081
    },
    {
      "epoch": 0.09039133505140456,
      "lm_loss": 1.2802,
      "step": 1232,
      "vm_loss": 0.1323
    },
    {
      "epoch": 0.09039133505140456,
      "lm_loss": 0.5076,
      "step": 1232,
      "vm_loss": 0.1884
    },
    {
      "epoch": 0.09039133505140456,
      "lm_loss": 0.7312,
      "step": 1232,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.09039133505140456,
      "lm_loss": 0.9102,
      "step": 1232,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.09039133505140456,
      "lm_loss": 1.1346,
      "step": 1232,
      "vm_loss": 0.0733
    },
    {
      "epoch": 0.09039133505140456,
      "lm_loss": 0.7873,
      "step": 1232,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.09039133505140456,
      "lm_loss": 0.5552,
      "step": 1232,
      "vm_loss": 0.0833
    },
    {
      "epoch": 0.0904647046415437,
      "grad_norm": 1.715980070447813,
      "learning_rate": 9.904447263863805e-06,
      "loss": 0.9606,
      "step": 1233
    },
    {
      "epoch": 0.09053807423168282,
      "grad_norm": 1.6207530364368432,
      "learning_rate": 9.904215942804374e-06,
      "loss": 0.8544,
      "step": 1234
    },
    {
      "epoch": 0.09061144382182194,
      "grad_norm": 1.8183379243833486,
      "learning_rate": 9.903984344791724e-06,
      "loss": 0.8763,
      "step": 1235
    },
    {
      "epoch": 0.09068481341196108,
      "grad_norm": 1.8985932809407842,
      "learning_rate": 9.903752469838934e-06,
      "loss": 0.9104,
      "step": 1236
    },
    {
      "epoch": 0.0907581830021002,
      "grad_norm": 1.6830038096946678,
      "learning_rate": 9.903520317959096e-06,
      "loss": 0.9064,
      "step": 1237
    },
    {
      "epoch": 0.09083155259223934,
      "grad_norm": 1.9078422552845704,
      "learning_rate": 9.903287889165323e-06,
      "loss": 0.8553,
      "step": 1238
    },
    {
      "epoch": 0.09090492218237846,
      "grad_norm": 0.6336071884295325,
      "learning_rate": 9.903055183470742e-06,
      "loss": 0.5723,
      "step": 1239
    },
    {
      "epoch": 0.09097829177251758,
      "grad_norm": 1.9638567675142462,
      "learning_rate": 9.902822200888491e-06,
      "loss": 0.9984,
      "step": 1240
    },
    {
      "epoch": 0.09097829177251758,
      "lm_loss": 0.6962,
      "step": 1240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09097829177251758,
      "lm_loss": 0.9029,
      "step": 1240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09097829177251758,
      "lm_loss": 0.6592,
      "step": 1240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09097829177251758,
      "lm_loss": 0.5133,
      "step": 1240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09097829177251758,
      "lm_loss": 0.6195,
      "step": 1240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09097829177251758,
      "lm_loss": 0.4507,
      "step": 1240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09097829177251758,
      "lm_loss": 0.5364,
      "step": 1240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09097829177251758,
      "lm_loss": 0.5573,
      "step": 1240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09105166136265672,
      "grad_norm": 0.6140089659707674,
      "learning_rate": 9.90258894143173e-06,
      "loss": 0.5983,
      "step": 1241
    },
    {
      "epoch": 0.09112503095279584,
      "grad_norm": 1.7825951932985273,
      "learning_rate": 9.902355405113628e-06,
      "loss": 0.8809,
      "step": 1242
    },
    {
      "epoch": 0.09119840054293497,
      "grad_norm": 2.420170302320807,
      "learning_rate": 9.902121591947378e-06,
      "loss": 0.7994,
      "step": 1243
    },
    {
      "epoch": 0.0912717701330741,
      "grad_norm": 1.763427627820706,
      "learning_rate": 9.90188750194618e-06,
      "loss": 0.9559,
      "step": 1244
    },
    {
      "epoch": 0.09134513972321322,
      "grad_norm": 1.7113766941469075,
      "learning_rate": 9.901653135123256e-06,
      "loss": 0.858,
      "step": 1245
    },
    {
      "epoch": 0.09141850931335235,
      "grad_norm": 2.115997269623762,
      "learning_rate": 9.901418491491842e-06,
      "loss": 0.8482,
      "step": 1246
    },
    {
      "epoch": 0.09149187890349147,
      "grad_norm": 0.7067354762350991,
      "learning_rate": 9.901183571065188e-06,
      "loss": 0.5744,
      "step": 1247
    },
    {
      "epoch": 0.0915652484936306,
      "grad_norm": 0.6560485972438481,
      "learning_rate": 9.900948373856558e-06,
      "loss": 0.5529,
      "step": 1248
    },
    {
      "epoch": 0.0915652484936306,
      "lm_loss": 1.4279,
      "step": 1248,
      "vm_loss": 0.0616
    },
    {
      "epoch": 0.0915652484936306,
      "lm_loss": 0.1818,
      "step": 1248,
      "vm_loss": 0.0895
    },
    {
      "epoch": 0.0915652484936306,
      "lm_loss": 0.6177,
      "step": 1248,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.0915652484936306,
      "lm_loss": 1.2487,
      "step": 1248,
      "vm_loss": 0.1907
    },
    {
      "epoch": 0.0915652484936306,
      "lm_loss": 0.4039,
      "step": 1248,
      "vm_loss": 0.152
    },
    {
      "epoch": 0.0915652484936306,
      "lm_loss": 0.1789,
      "step": 1248,
      "vm_loss": 0.0745
    },
    {
      "epoch": 0.0915652484936306,
      "lm_loss": 0.849,
      "step": 1248,
      "vm_loss": 0.1649
    },
    {
      "epoch": 0.0915652484936306,
      "lm_loss": 0.3502,
      "step": 1248,
      "vm_loss": 0.1886
    },
    {
      "epoch": 0.09163861808376973,
      "grad_norm": 2.4282418223173825,
      "learning_rate": 9.900712899879237e-06,
      "loss": 0.8931,
      "step": 1249
    },
    {
      "epoch": 0.09171198767390885,
      "grad_norm": 1.6915681431916303,
      "learning_rate": 9.900477149146522e-06,
      "loss": 0.8167,
      "step": 1250
    },
    {
      "epoch": 0.09178535726404799,
      "grad_norm": 0.6476928823892584,
      "learning_rate": 9.900241121671727e-06,
      "loss": 0.5722,
      "step": 1251
    },
    {
      "epoch": 0.09185872685418711,
      "grad_norm": 0.6548680795314009,
      "learning_rate": 9.900004817468181e-06,
      "loss": 0.5809,
      "step": 1252
    },
    {
      "epoch": 0.09193209644432623,
      "grad_norm": 2.6909292369442315,
      "learning_rate": 9.899768236549228e-06,
      "loss": 0.9681,
      "step": 1253
    },
    {
      "epoch": 0.09200546603446537,
      "grad_norm": 2.1608494895790287,
      "learning_rate": 9.89953137892823e-06,
      "loss": 0.9112,
      "step": 1254
    },
    {
      "epoch": 0.09207883562460449,
      "grad_norm": 1.7439820471125507,
      "learning_rate": 9.899294244618559e-06,
      "loss": 0.8542,
      "step": 1255
    },
    {
      "epoch": 0.09215220521474361,
      "grad_norm": 1.6951798301195111,
      "learning_rate": 9.899056833633609e-06,
      "loss": 0.9519,
      "step": 1256
    },
    {
      "epoch": 0.09215220521474361,
      "lm_loss": 1.6413,
      "step": 1256,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.09215220521474361,
      "lm_loss": 0.7591,
      "step": 1256,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.09215220521474361,
      "lm_loss": 0.3041,
      "step": 1256,
      "vm_loss": 0.2005
    },
    {
      "epoch": 0.09215220521474361,
      "lm_loss": 0.5344,
      "step": 1256,
      "vm_loss": 0.1693
    },
    {
      "epoch": 0.09215220521474361,
      "lm_loss": 0.4748,
      "step": 1256,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.09215220521474361,
      "lm_loss": 0.9116,
      "step": 1256,
      "vm_loss": 0.1529
    },
    {
      "epoch": 0.09215220521474361,
      "lm_loss": 0.2257,
      "step": 1256,
      "vm_loss": 0.0772
    },
    {
      "epoch": 0.09215220521474361,
      "lm_loss": 0.8368,
      "step": 1256,
      "vm_loss": 0.1283
    },
    {
      "epoch": 0.09222557480488275,
      "grad_norm": 1.8138291920074208,
      "learning_rate": 9.898819145986789e-06,
      "loss": 0.8405,
      "step": 1257
    },
    {
      "epoch": 0.09229894439502187,
      "grad_norm": 1.7451395548141668,
      "learning_rate": 9.89858118169152e-06,
      "loss": 0.8351,
      "step": 1258
    },
    {
      "epoch": 0.092372313985161,
      "grad_norm": 1.7220873123936185,
      "learning_rate": 9.89834294076124e-06,
      "loss": 0.8878,
      "step": 1259
    },
    {
      "epoch": 0.09244568357530013,
      "grad_norm": 0.7235296701815871,
      "learning_rate": 9.898104423209402e-06,
      "loss": 0.5527,
      "step": 1260
    },
    {
      "epoch": 0.09251905316543925,
      "grad_norm": 0.6551268264496081,
      "learning_rate": 9.89786562904948e-06,
      "loss": 0.5806,
      "step": 1261
    },
    {
      "epoch": 0.09259242275557839,
      "grad_norm": 2.011419415280142,
      "learning_rate": 9.897626558294955e-06,
      "loss": 0.9325,
      "step": 1262
    },
    {
      "epoch": 0.09266579234571751,
      "grad_norm": 1.8416113621104209,
      "learning_rate": 9.897387210959331e-06,
      "loss": 0.9022,
      "step": 1263
    },
    {
      "epoch": 0.09273916193585664,
      "grad_norm": 1.9490303417864039,
      "learning_rate": 9.897147587056122e-06,
      "loss": 0.9503,
      "step": 1264
    },
    {
      "epoch": 0.09273916193585664,
      "lm_loss": 2.316,
      "step": 1264,
      "vm_loss": 0.1077
    },
    {
      "epoch": 0.09273916193585664,
      "lm_loss": 0.7733,
      "step": 1264,
      "vm_loss": 0.1355
    },
    {
      "epoch": 0.09273916193585664,
      "lm_loss": 1.0623,
      "step": 1264,
      "vm_loss": 0.2086
    },
    {
      "epoch": 0.09273916193585664,
      "lm_loss": 1.0239,
      "step": 1264,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.09273916193585664,
      "lm_loss": 1.348,
      "step": 1264,
      "vm_loss": 0.2083
    },
    {
      "epoch": 0.09273916193585664,
      "lm_loss": 0.5922,
      "step": 1264,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.09273916193585664,
      "lm_loss": 0.7386,
      "step": 1264,
      "vm_loss": 0.2002
    },
    {
      "epoch": 0.09273916193585664,
      "lm_loss": 0.6361,
      "step": 1264,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.09281253152599576,
      "grad_norm": 1.692923853917379,
      "learning_rate": 9.896907686598862e-06,
      "loss": 0.9279,
      "step": 1265
    },
    {
      "epoch": 0.09288590111613489,
      "grad_norm": 1.7634939808845034,
      "learning_rate": 9.896667509601097e-06,
      "loss": 0.8854,
      "step": 1266
    },
    {
      "epoch": 0.09295927070627402,
      "grad_norm": 1.6621208229335138,
      "learning_rate": 9.89642705607639e-06,
      "loss": 0.8832,
      "step": 1267
    },
    {
      "epoch": 0.09303264029641314,
      "grad_norm": 2.063574713643318,
      "learning_rate": 9.896186326038325e-06,
      "loss": 0.9044,
      "step": 1268
    },
    {
      "epoch": 0.09310600988655227,
      "grad_norm": 1.6685834074573465,
      "learning_rate": 9.895945319500488e-06,
      "loss": 0.9059,
      "step": 1269
    },
    {
      "epoch": 0.0931793794766914,
      "grad_norm": 1.7305833543820026,
      "learning_rate": 9.895704036476498e-06,
      "loss": 0.957,
      "step": 1270
    },
    {
      "epoch": 0.09325274906683052,
      "grad_norm": 0.9615849525094259,
      "learning_rate": 9.895462476979975e-06,
      "loss": 0.5837,
      "step": 1271
    },
    {
      "epoch": 0.09332611865696966,
      "grad_norm": 1.7853884289092044,
      "learning_rate": 9.895220641024566e-06,
      "loss": 0.8434,
      "step": 1272
    },
    {
      "epoch": 0.09332611865696966,
      "lm_loss": 0.546,
      "step": 1272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09332611865696966,
      "lm_loss": 0.5363,
      "step": 1272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09332611865696966,
      "lm_loss": 0.5124,
      "step": 1272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09332611865696966,
      "lm_loss": 0.4846,
      "step": 1272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09332611865696966,
      "lm_loss": 0.5045,
      "step": 1272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09332611865696966,
      "lm_loss": 0.4935,
      "step": 1272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09332611865696966,
      "lm_loss": 0.5314,
      "step": 1272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09332611865696966,
      "lm_loss": 0.5649,
      "step": 1272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09339948824710878,
      "grad_norm": 0.6920962073621446,
      "learning_rate": 9.89497852862392e-06,
      "loss": 0.5695,
      "step": 1273
    },
    {
      "epoch": 0.0934728578372479,
      "grad_norm": 1.8134693539590638,
      "learning_rate": 9.894736139791718e-06,
      "loss": 0.8579,
      "step": 1274
    },
    {
      "epoch": 0.09354622742738704,
      "grad_norm": 1.980236127357552,
      "learning_rate": 9.894493474541643e-06,
      "loss": 0.8792,
      "step": 1275
    },
    {
      "epoch": 0.09361959701752616,
      "grad_norm": 0.7511776073583721,
      "learning_rate": 9.8942505328874e-06,
      "loss": 0.5824,
      "step": 1276
    },
    {
      "epoch": 0.09369296660766528,
      "grad_norm": 1.9366958334911757,
      "learning_rate": 9.89400731484271e-06,
      "loss": 0.8759,
      "step": 1277
    },
    {
      "epoch": 0.09376633619780442,
      "grad_norm": 1.8633999217832822,
      "learning_rate": 9.893763820421305e-06,
      "loss": 0.937,
      "step": 1278
    },
    {
      "epoch": 0.09383970578794354,
      "grad_norm": 1.9302904435338304,
      "learning_rate": 9.89352004963694e-06,
      "loss": 0.8609,
      "step": 1279
    },
    {
      "epoch": 0.09391307537808267,
      "grad_norm": 1.7297307604263892,
      "learning_rate": 9.893276002503379e-06,
      "loss": 0.8427,
      "step": 1280
    },
    {
      "epoch": 0.09391307537808267,
      "lm_loss": 0.7383,
      "step": 1280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09391307537808267,
      "lm_loss": 0.4595,
      "step": 1280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09391307537808267,
      "lm_loss": 0.5996,
      "step": 1280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09391307537808267,
      "lm_loss": 0.4413,
      "step": 1280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09391307537808267,
      "lm_loss": 0.44,
      "step": 1280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09391307537808267,
      "lm_loss": 0.7012,
      "step": 1280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09391307537808267,
      "lm_loss": 0.578,
      "step": 1280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09391307537808267,
      "lm_loss": 0.5898,
      "step": 1280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0939864449682218,
      "grad_norm": 0.7039583314934562,
      "learning_rate": 9.893031679034404e-06,
      "loss": 0.5595,
      "step": 1281
    },
    {
      "epoch": 0.09405981455836092,
      "grad_norm": 1.8281510001608878,
      "learning_rate": 9.892787079243812e-06,
      "loss": 0.976,
      "step": 1282
    },
    {
      "epoch": 0.09413318414850005,
      "grad_norm": 1.652891480429621,
      "learning_rate": 9.892542203145418e-06,
      "loss": 0.8806,
      "step": 1283
    },
    {
      "epoch": 0.09420655373863918,
      "grad_norm": 0.6527871148032165,
      "learning_rate": 9.892297050753049e-06,
      "loss": 0.5703,
      "step": 1284
    },
    {
      "epoch": 0.0942799233287783,
      "grad_norm": 1.7655638244733949,
      "learning_rate": 9.89205162208055e-06,
      "loss": 0.8903,
      "step": 1285
    },
    {
      "epoch": 0.09435329291891743,
      "grad_norm": 1.8050195188036737,
      "learning_rate": 9.89180591714178e-06,
      "loss": 0.9053,
      "step": 1286
    },
    {
      "epoch": 0.09442666250905656,
      "grad_norm": 1.7952294393983135,
      "learning_rate": 9.891559935950617e-06,
      "loss": 0.9226,
      "step": 1287
    },
    {
      "epoch": 0.09450003209919569,
      "grad_norm": 1.7324129789411342,
      "learning_rate": 9.89131367852095e-06,
      "loss": 0.91,
      "step": 1288
    },
    {
      "epoch": 0.09450003209919569,
      "lm_loss": 0.5373,
      "step": 1288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09450003209919569,
      "lm_loss": 0.4906,
      "step": 1288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09450003209919569,
      "lm_loss": 0.5558,
      "step": 1288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09450003209919569,
      "lm_loss": 0.6451,
      "step": 1288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09450003209919569,
      "lm_loss": 0.6374,
      "step": 1288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09450003209919569,
      "lm_loss": 0.5594,
      "step": 1288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09450003209919569,
      "lm_loss": 0.533,
      "step": 1288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09450003209919569,
      "lm_loss": 0.5273,
      "step": 1288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09457340168933481,
      "grad_norm": 0.7070645636071138,
      "learning_rate": 9.891067144866687e-06,
      "loss": 0.5772,
      "step": 1289
    },
    {
      "epoch": 0.09464677127947393,
      "grad_norm": 1.8001649336137702,
      "learning_rate": 9.890820335001749e-06,
      "loss": 0.9907,
      "step": 1290
    },
    {
      "epoch": 0.09472014086961307,
      "grad_norm": 1.8006186994957254,
      "learning_rate": 9.890573248940077e-06,
      "loss": 0.781,
      "step": 1291
    },
    {
      "epoch": 0.09479351045975219,
      "grad_norm": 1.7636089672938193,
      "learning_rate": 9.89032588669562e-06,
      "loss": 0.9421,
      "step": 1292
    },
    {
      "epoch": 0.09486688004989133,
      "grad_norm": 0.6511024627972826,
      "learning_rate": 9.890078248282349e-06,
      "loss": 0.5801,
      "step": 1293
    },
    {
      "epoch": 0.09494024964003045,
      "grad_norm": 1.839550439124969,
      "learning_rate": 9.889830333714249e-06,
      "loss": 0.9356,
      "step": 1294
    },
    {
      "epoch": 0.09501361923016957,
      "grad_norm": 1.940014160876079,
      "learning_rate": 9.889582143005323e-06,
      "loss": 1.0132,
      "step": 1295
    },
    {
      "epoch": 0.09508698882030871,
      "grad_norm": 0.6449766706065484,
      "learning_rate": 9.889333676169584e-06,
      "loss": 0.5821,
      "step": 1296
    },
    {
      "epoch": 0.09508698882030871,
      "lm_loss": 1.7679,
      "step": 1296,
      "vm_loss": 0.0649
    },
    {
      "epoch": 0.09508698882030871,
      "lm_loss": 0.8825,
      "step": 1296,
      "vm_loss": 0.2178
    },
    {
      "epoch": 0.09508698882030871,
      "lm_loss": 0.7306,
      "step": 1296,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.09508698882030871,
      "lm_loss": 0.7764,
      "step": 1296,
      "vm_loss": 0.1332
    },
    {
      "epoch": 0.09508698882030871,
      "lm_loss": 1.0529,
      "step": 1296,
      "vm_loss": 0.2492
    },
    {
      "epoch": 0.09508698882030871,
      "lm_loss": 0.3928,
      "step": 1296,
      "vm_loss": 0.1536
    },
    {
      "epoch": 0.09508698882030871,
      "lm_loss": 0.9358,
      "step": 1296,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.09508698882030871,
      "lm_loss": 1.2734,
      "step": 1296,
      "vm_loss": 0.1914
    },
    {
      "epoch": 0.09516035841044783,
      "grad_norm": 1.8925749086532477,
      "learning_rate": 9.889084933221063e-06,
      "loss": 0.9479,
      "step": 1297
    },
    {
      "epoch": 0.09523372800058695,
      "grad_norm": 1.7009423107982116,
      "learning_rate": 9.888835914173808e-06,
      "loss": 1.0195,
      "step": 1298
    },
    {
      "epoch": 0.09530709759072609,
      "grad_norm": 0.6729118099286865,
      "learning_rate": 9.888586619041882e-06,
      "loss": 0.5917,
      "step": 1299
    },
    {
      "epoch": 0.09538046718086521,
      "grad_norm": 1.7764842164687902,
      "learning_rate": 9.888337047839365e-06,
      "loss": 0.9727,
      "step": 1300
    },
    {
      "epoch": 0.09545383677100434,
      "grad_norm": 1.6071838099744717,
      "learning_rate": 9.888087200580347e-06,
      "loss": 0.8881,
      "step": 1301
    },
    {
      "epoch": 0.09552720636114347,
      "grad_norm": 1.6946268123479418,
      "learning_rate": 9.88783707727894e-06,
      "loss": 0.846,
      "step": 1302
    },
    {
      "epoch": 0.09560057595128259,
      "grad_norm": 1.716765751168012,
      "learning_rate": 9.88758667794927e-06,
      "loss": 0.868,
      "step": 1303
    },
    {
      "epoch": 0.09567394554142172,
      "grad_norm": 1.5952637627043693,
      "learning_rate": 9.887336002605476e-06,
      "loss": 1.0242,
      "step": 1304
    },
    {
      "epoch": 0.09567394554142172,
      "lm_loss": 1.8397,
      "step": 1304,
      "vm_loss": 0.0345
    },
    {
      "epoch": 0.09567394554142172,
      "lm_loss": 0.363,
      "step": 1304,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.09567394554142172,
      "lm_loss": 0.8009,
      "step": 1304,
      "vm_loss": 0.0705
    },
    {
      "epoch": 0.09567394554142172,
      "lm_loss": 0.4767,
      "step": 1304,
      "vm_loss": 0.0856
    },
    {
      "epoch": 0.09567394554142172,
      "lm_loss": 0.374,
      "step": 1304,
      "vm_loss": 0.0744
    },
    {
      "epoch": 0.09567394554142172,
      "lm_loss": 0.8763,
      "step": 1304,
      "vm_loss": 0.2171
    },
    {
      "epoch": 0.09567394554142172,
      "lm_loss": 1.8891,
      "step": 1304,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.09567394554142172,
      "lm_loss": 0.4148,
      "step": 1304,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.09574731513156085,
      "grad_norm": 1.799574674829333,
      "learning_rate": 9.887085051261715e-06,
      "loss": 0.8921,
      "step": 1305
    },
    {
      "epoch": 0.09582068472169997,
      "grad_norm": 1.7871065092313845,
      "learning_rate": 9.886833823932157e-06,
      "loss": 0.9671,
      "step": 1306
    },
    {
      "epoch": 0.0958940543118391,
      "grad_norm": 1.7062156624013602,
      "learning_rate": 9.886582320630992e-06,
      "loss": 0.9476,
      "step": 1307
    },
    {
      "epoch": 0.09596742390197822,
      "grad_norm": 0.7446822870609241,
      "learning_rate": 9.886330541372422e-06,
      "loss": 0.5664,
      "step": 1308
    },
    {
      "epoch": 0.09604079349211736,
      "grad_norm": 0.691162743631335,
      "learning_rate": 9.886078486170665e-06,
      "loss": 0.5786,
      "step": 1309
    },
    {
      "epoch": 0.09611416308225648,
      "grad_norm": 1.751193517868003,
      "learning_rate": 9.885826155039954e-06,
      "loss": 0.9323,
      "step": 1310
    },
    {
      "epoch": 0.0961875326723956,
      "grad_norm": 0.6217134613907823,
      "learning_rate": 9.885573547994545e-06,
      "loss": 0.5779,
      "step": 1311
    },
    {
      "epoch": 0.09626090226253474,
      "grad_norm": 1.6580795167918425,
      "learning_rate": 9.885320665048694e-06,
      "loss": 0.9207,
      "step": 1312
    },
    {
      "epoch": 0.09626090226253474,
      "lm_loss": 1.515,
      "step": 1312,
      "vm_loss": 0.0596
    },
    {
      "epoch": 0.09626090226253474,
      "lm_loss": 1.5678,
      "step": 1312,
      "vm_loss": 0.1431
    },
    {
      "epoch": 0.09626090226253474,
      "lm_loss": 0.3193,
      "step": 1312,
      "vm_loss": 0.1683
    },
    {
      "epoch": 0.09626090226253474,
      "lm_loss": 0.8119,
      "step": 1312,
      "vm_loss": 0.1721
    },
    {
      "epoch": 0.09626090226253474,
      "lm_loss": 0.3439,
      "step": 1312,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.09626090226253474,
      "lm_loss": 1.1942,
      "step": 1312,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.09626090226253474,
      "lm_loss": 0.4645,
      "step": 1312,
      "vm_loss": 0.1271
    },
    {
      "epoch": 0.09626090226253474,
      "lm_loss": 0.7342,
      "step": 1312,
      "vm_loss": 0.101
    },
    {
      "epoch": 0.09633427185267386,
      "grad_norm": 1.686823232081418,
      "learning_rate": 9.88506750621669e-06,
      "loss": 0.9166,
      "step": 1313
    },
    {
      "epoch": 0.096407641442813,
      "grad_norm": 1.8645289010245838,
      "learning_rate": 9.884814071512826e-06,
      "loss": 1.0117,
      "step": 1314
    },
    {
      "epoch": 0.09648101103295212,
      "grad_norm": 0.8749262243360675,
      "learning_rate": 9.884560360951412e-06,
      "loss": 0.576,
      "step": 1315
    },
    {
      "epoch": 0.09655438062309124,
      "grad_norm": 2.0087509350379724,
      "learning_rate": 9.884306374546781e-06,
      "loss": 0.9264,
      "step": 1316
    },
    {
      "epoch": 0.09662775021323038,
      "grad_norm": 1.572417443726623,
      "learning_rate": 9.884052112313272e-06,
      "loss": 0.7844,
      "step": 1317
    },
    {
      "epoch": 0.0967011198033695,
      "grad_norm": 1.6257645858554772,
      "learning_rate": 9.883797574265243e-06,
      "loss": 0.8937,
      "step": 1318
    },
    {
      "epoch": 0.09677448939350862,
      "grad_norm": 1.9013167211536877,
      "learning_rate": 9.883542760417074e-06,
      "loss": 0.9741,
      "step": 1319
    },
    {
      "epoch": 0.09684785898364776,
      "grad_norm": 1.806208464352724,
      "learning_rate": 9.88328767078315e-06,
      "loss": 0.8357,
      "step": 1320
    },
    {
      "epoch": 0.09684785898364776,
      "lm_loss": 1.4511,
      "step": 1320,
      "vm_loss": 0.0455
    },
    {
      "epoch": 0.09684785898364776,
      "lm_loss": 1.37,
      "step": 1320,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.09684785898364776,
      "lm_loss": 0.2522,
      "step": 1320,
      "vm_loss": 0.0554
    },
    {
      "epoch": 0.09684785898364776,
      "lm_loss": 0.9136,
      "step": 1320,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.09684785898364776,
      "lm_loss": 0.2466,
      "step": 1320,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.09684785898364776,
      "lm_loss": 0.2622,
      "step": 1320,
      "vm_loss": 0.1704
    },
    {
      "epoch": 0.09684785898364776,
      "lm_loss": 0.5795,
      "step": 1320,
      "vm_loss": 0.1545
    },
    {
      "epoch": 0.09684785898364776,
      "lm_loss": 0.2316,
      "step": 1320,
      "vm_loss": 0.1469
    },
    {
      "epoch": 0.09692122857378688,
      "grad_norm": 1.4430623606080097,
      "learning_rate": 9.883032305377877e-06,
      "loss": 0.8355,
      "step": 1321
    },
    {
      "epoch": 0.09699459816392601,
      "grad_norm": 0.7384258417932348,
      "learning_rate": 9.882776664215679e-06,
      "loss": 0.5687,
      "step": 1322
    },
    {
      "epoch": 0.09706796775406513,
      "grad_norm": 1.651471406266585,
      "learning_rate": 9.882520747310988e-06,
      "loss": 0.8164,
      "step": 1323
    },
    {
      "epoch": 0.09714133734420426,
      "grad_norm": 1.668596304934749,
      "learning_rate": 9.882264554678262e-06,
      "loss": 0.9214,
      "step": 1324
    },
    {
      "epoch": 0.09721470693434339,
      "grad_norm": 1.995372077578569,
      "learning_rate": 9.882008086331964e-06,
      "loss": 0.8566,
      "step": 1325
    },
    {
      "epoch": 0.09728807652448251,
      "grad_norm": 0.6076783273135657,
      "learning_rate": 9.88175134228658e-06,
      "loss": 0.5679,
      "step": 1326
    },
    {
      "epoch": 0.09736144611462164,
      "grad_norm": 1.7715541591707364,
      "learning_rate": 9.881494322556608e-06,
      "loss": 0.935,
      "step": 1327
    },
    {
      "epoch": 0.09743481570476077,
      "grad_norm": 1.7288463647098589,
      "learning_rate": 9.881237027156563e-06,
      "loss": 0.9699,
      "step": 1328
    },
    {
      "epoch": 0.09743481570476077,
      "lm_loss": 0.6601,
      "step": 1328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09743481570476077,
      "lm_loss": 0.4751,
      "step": 1328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09743481570476077,
      "lm_loss": 0.5414,
      "step": 1328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09743481570476077,
      "lm_loss": 0.4607,
      "step": 1328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09743481570476077,
      "lm_loss": 0.527,
      "step": 1328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09743481570476077,
      "lm_loss": 0.6005,
      "step": 1328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09743481570476077,
      "lm_loss": 0.5737,
      "step": 1328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.09743481570476077,
      "lm_loss": 0.6162,
      "step": 1328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.0975081852948999,
      "grad_norm": 0.6228591394896582,
      "learning_rate": 9.880979456100974e-06,
      "loss": 0.5931,
      "step": 1329
    },
    {
      "epoch": 0.09758155488503903,
      "grad_norm": 0.5910893661802733,
      "learning_rate": 9.880721609404389e-06,
      "loss": 0.5878,
      "step": 1330
    },
    {
      "epoch": 0.09765492447517815,
      "grad_norm": 1.8270093524770745,
      "learning_rate": 9.880463487081367e-06,
      "loss": 0.9544,
      "step": 1331
    },
    {
      "epoch": 0.09772829406531727,
      "grad_norm": 1.6999915294125927,
      "learning_rate": 9.880205089146487e-06,
      "loss": 0.8254,
      "step": 1332
    },
    {
      "epoch": 0.09780166365545641,
      "grad_norm": 1.7863767675538524,
      "learning_rate": 9.879946415614338e-06,
      "loss": 0.8904,
      "step": 1333
    },
    {
      "epoch": 0.09787503324559553,
      "grad_norm": 1.6493522702463703,
      "learning_rate": 9.87968746649953e-06,
      "loss": 1.0332,
      "step": 1334
    },
    {
      "epoch": 0.09794840283573465,
      "grad_norm": 0.7041429456153145,
      "learning_rate": 9.879428241816687e-06,
      "loss": 0.5986,
      "step": 1335
    },
    {
      "epoch": 0.09802177242587379,
      "grad_norm": 1.8177832806938292,
      "learning_rate": 9.879168741580446e-06,
      "loss": 0.9654,
      "step": 1336
    },
    {
      "epoch": 0.09802177242587379,
      "lm_loss": 1.5772,
      "step": 1336,
      "vm_loss": 0.0782
    },
    {
      "epoch": 0.09802177242587379,
      "lm_loss": 1.1523,
      "step": 1336,
      "vm_loss": 0.076
    },
    {
      "epoch": 0.09802177242587379,
      "lm_loss": 1.1847,
      "step": 1336,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.09802177242587379,
      "lm_loss": 0.3766,
      "step": 1336,
      "vm_loss": 0.1011
    },
    {
      "epoch": 0.09802177242587379,
      "lm_loss": 0.4678,
      "step": 1336,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.09802177242587379,
      "lm_loss": 0.8857,
      "step": 1336,
      "vm_loss": 0.0994
    },
    {
      "epoch": 0.09802177242587379,
      "lm_loss": 0.4413,
      "step": 1336,
      "vm_loss": 0.2067
    },
    {
      "epoch": 0.09802177242587379,
      "lm_loss": 0.3935,
      "step": 1336,
      "vm_loss": 0.1618
    },
    {
      "epoch": 0.09809514201601291,
      "grad_norm": 1.8046819571035975,
      "learning_rate": 9.878908965805467e-06,
      "loss": 0.8588,
      "step": 1337
    },
    {
      "epoch": 0.09816851160615205,
      "grad_norm": 0.5934769976448946,
      "learning_rate": 9.878648914506413e-06,
      "loss": 0.5747,
      "step": 1338
    },
    {
      "epoch": 0.09824188119629117,
      "grad_norm": 1.6172220290399393,
      "learning_rate": 9.878388587697975e-06,
      "loss": 0.8623,
      "step": 1339
    },
    {
      "epoch": 0.09831525078643029,
      "grad_norm": 1.6870546431745757,
      "learning_rate": 9.878127985394851e-06,
      "loss": 0.9254,
      "step": 1340
    },
    {
      "epoch": 0.09838862037656942,
      "grad_norm": 1.49184914038526,
      "learning_rate": 9.877867107611759e-06,
      "loss": 0.9818,
      "step": 1341
    },
    {
      "epoch": 0.09846198996670855,
      "grad_norm": 1.7380058735031931,
      "learning_rate": 9.877605954363432e-06,
      "loss": 0.9593,
      "step": 1342
    },
    {
      "epoch": 0.09853535955684768,
      "grad_norm": 1.567549297091087,
      "learning_rate": 9.877344525664618e-06,
      "loss": 0.9519,
      "step": 1343
    },
    {
      "epoch": 0.0986087291469868,
      "grad_norm": 1.7904572410731212,
      "learning_rate": 9.87708282153008e-06,
      "loss": 0.9251,
      "step": 1344
    },
    {
      "epoch": 0.0986087291469868,
      "lm_loss": 1.3619,
      "step": 1344,
      "vm_loss": 0.0539
    },
    {
      "epoch": 0.0986087291469868,
      "lm_loss": 0.7799,
      "step": 1344,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.0986087291469868,
      "lm_loss": 0.7982,
      "step": 1344,
      "vm_loss": 0.0925
    },
    {
      "epoch": 0.0986087291469868,
      "lm_loss": 1.2463,
      "step": 1344,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.0986087291469868,
      "lm_loss": 1.3258,
      "step": 1344,
      "vm_loss": 0.2238
    },
    {
      "epoch": 0.0986087291469868,
      "lm_loss": 0.9831,
      "step": 1344,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.0986087291469868,
      "lm_loss": 0.5625,
      "step": 1344,
      "vm_loss": 0.1322
    },
    {
      "epoch": 0.0986087291469868,
      "lm_loss": 0.2552,
      "step": 1344,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.09868209873712593,
      "grad_norm": 1.703614782545958,
      "learning_rate": 9.876820841974597e-06,
      "loss": 0.8893,
      "step": 1345
    },
    {
      "epoch": 0.09875546832726506,
      "grad_norm": 0.8076659245706674,
      "learning_rate": 9.876558587012963e-06,
      "loss": 0.5811,
      "step": 1346
    },
    {
      "epoch": 0.09882883791740418,
      "grad_norm": 1.5072782982745283,
      "learning_rate": 9.87629605665999e-06,
      "loss": 0.8819,
      "step": 1347
    },
    {
      "epoch": 0.0989022075075433,
      "grad_norm": 0.619975676970031,
      "learning_rate": 9.876033250930504e-06,
      "loss": 0.5559,
      "step": 1348
    },
    {
      "epoch": 0.09897557709768244,
      "grad_norm": 1.7028014117575803,
      "learning_rate": 9.875770169839343e-06,
      "loss": 0.9946,
      "step": 1349
    },
    {
      "epoch": 0.09904894668782156,
      "grad_norm": 1.5569491375597013,
      "learning_rate": 9.875506813401365e-06,
      "loss": 0.8868,
      "step": 1350
    },
    {
      "epoch": 0.0991223162779607,
      "grad_norm": 0.727361912349477,
      "learning_rate": 9.875243181631446e-06,
      "loss": 0.5804,
      "step": 1351
    },
    {
      "epoch": 0.09919568586809982,
      "grad_norm": 0.6566008589426826,
      "learning_rate": 9.87497927454447e-06,
      "loss": 0.5898,
      "step": 1352
    },
    {
      "epoch": 0.09919568586809982,
      "lm_loss": 1.3324,
      "step": 1352,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.09919568586809982,
      "lm_loss": 0.0617,
      "step": 1352,
      "vm_loss": 0.0683
    },
    {
      "epoch": 0.09919568586809982,
      "lm_loss": 1.3267,
      "step": 1352,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.09919568586809982,
      "lm_loss": 0.2051,
      "step": 1352,
      "vm_loss": 0.0737
    },
    {
      "epoch": 0.09919568586809982,
      "lm_loss": 0.5764,
      "step": 1352,
      "vm_loss": 0.1305
    },
    {
      "epoch": 0.09919568586809982,
      "lm_loss": 0.3856,
      "step": 1352,
      "vm_loss": 0.088
    },
    {
      "epoch": 0.09919568586809982,
      "lm_loss": 0.5808,
      "step": 1352,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.09919568586809982,
      "lm_loss": 0.8881,
      "step": 1352,
      "vm_loss": 0.1215
    },
    {
      "epoch": 0.09926905545823894,
      "grad_norm": 1.5921705865452644,
      "learning_rate": 9.87471509215534e-06,
      "loss": 0.8173,
      "step": 1353
    },
    {
      "epoch": 0.09934242504837808,
      "grad_norm": 0.5966775032367321,
      "learning_rate": 9.874450634478979e-06,
      "loss": 0.5803,
      "step": 1354
    },
    {
      "epoch": 0.0994157946385172,
      "grad_norm": 1.845120834385049,
      "learning_rate": 9.874185901530319e-06,
      "loss": 0.9511,
      "step": 1355
    },
    {
      "epoch": 0.09948916422865632,
      "grad_norm": 1.5792061445238605,
      "learning_rate": 9.87392089332431e-06,
      "loss": 0.9064,
      "step": 1356
    },
    {
      "epoch": 0.09956253381879546,
      "grad_norm": 0.7066035175869031,
      "learning_rate": 9.873655609875918e-06,
      "loss": 0.5536,
      "step": 1357
    },
    {
      "epoch": 0.09963590340893458,
      "grad_norm": 0.7065824234295393,
      "learning_rate": 9.873390051200125e-06,
      "loss": 0.5711,
      "step": 1358
    },
    {
      "epoch": 0.09970927299907371,
      "grad_norm": 0.6277446982564467,
      "learning_rate": 9.873124217311926e-06,
      "loss": 0.5774,
      "step": 1359
    },
    {
      "epoch": 0.09978264258921284,
      "grad_norm": 1.7177485570411237,
      "learning_rate": 9.872858108226335e-06,
      "loss": 0.9557,
      "step": 1360
    },
    {
      "epoch": 0.09978264258921284,
      "lm_loss": 1.3192,
      "step": 1360,
      "vm_loss": 0.1033
    },
    {
      "epoch": 0.09978264258921284,
      "lm_loss": 1.446,
      "step": 1360,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.09978264258921284,
      "lm_loss": 0.8643,
      "step": 1360,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.09978264258921284,
      "lm_loss": 0.7015,
      "step": 1360,
      "vm_loss": 0.2242
    },
    {
      "epoch": 0.09978264258921284,
      "lm_loss": 1.1812,
      "step": 1360,
      "vm_loss": 0.208
    },
    {
      "epoch": 0.09978264258921284,
      "lm_loss": 0.669,
      "step": 1360,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.09978264258921284,
      "lm_loss": 0.551,
      "step": 1360,
      "vm_loss": 0.1728
    },
    {
      "epoch": 0.09978264258921284,
      "lm_loss": 0.3238,
      "step": 1360,
      "vm_loss": 0.144
    },
    {
      "epoch": 0.09985601217935196,
      "grad_norm": 2.2957526601470315,
      "learning_rate": 9.872591723958379e-06,
      "loss": 1.0165,
      "step": 1361
    },
    {
      "epoch": 0.0999293817694911,
      "grad_norm": 0.8265070702046895,
      "learning_rate": 9.872325064523101e-06,
      "loss": 0.5621,
      "step": 1362
    },
    {
      "epoch": 0.10000275135963022,
      "grad_norm": 0.7158812449176881,
      "learning_rate": 9.87205812993556e-06,
      "loss": 0.57,
      "step": 1363
    },
    {
      "epoch": 0.10007612094976935,
      "grad_norm": 1.7397601536231628,
      "learning_rate": 9.871790920210833e-06,
      "loss": 0.9515,
      "step": 1364
    },
    {
      "epoch": 0.10014949053990847,
      "grad_norm": 2.637017709665078,
      "learning_rate": 9.871523435364008e-06,
      "loss": 0.9076,
      "step": 1365
    },
    {
      "epoch": 0.1002228601300476,
      "grad_norm": 0.7105246531441785,
      "learning_rate": 9.871255675410188e-06,
      "loss": 0.5662,
      "step": 1366
    },
    {
      "epoch": 0.10029622972018673,
      "grad_norm": 1.7433502397691563,
      "learning_rate": 9.8709876403645e-06,
      "loss": 0.8419,
      "step": 1367
    },
    {
      "epoch": 0.10036959931032585,
      "grad_norm": 1.8593613823982387,
      "learning_rate": 9.870719330242073e-06,
      "loss": 0.919,
      "step": 1368
    },
    {
      "epoch": 0.10036959931032585,
      "lm_loss": 1.542,
      "step": 1368,
      "vm_loss": 0.0611
    },
    {
      "epoch": 0.10036959931032585,
      "lm_loss": 0.809,
      "step": 1368,
      "vm_loss": 0.1108
    },
    {
      "epoch": 0.10036959931032585,
      "lm_loss": 0.5696,
      "step": 1368,
      "vm_loss": 0.1442
    },
    {
      "epoch": 0.10036959931032585,
      "lm_loss": 0.6387,
      "step": 1368,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.10036959931032585,
      "lm_loss": 0.8415,
      "step": 1368,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0.10036959931032585,
      "lm_loss": 0.7228,
      "step": 1368,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.10036959931032585,
      "lm_loss": 0.958,
      "step": 1368,
      "vm_loss": 0.162
    },
    {
      "epoch": 0.10036959931032585,
      "lm_loss": 0.8352,
      "step": 1368,
      "vm_loss": 0.1528
    },
    {
      "epoch": 0.10044296890046497,
      "grad_norm": 2.032371219968466,
      "learning_rate": 9.870450745058066e-06,
      "loss": 0.9413,
      "step": 1369
    },
    {
      "epoch": 0.10051633849060411,
      "grad_norm": 0.7152372581095653,
      "learning_rate": 9.870181884827643e-06,
      "loss": 0.5643,
      "step": 1370
    },
    {
      "epoch": 0.10058970808074323,
      "grad_norm": 1.7862620313295994,
      "learning_rate": 9.869912749565989e-06,
      "loss": 0.897,
      "step": 1371
    },
    {
      "epoch": 0.10066307767088237,
      "grad_norm": 1.9423727970557514,
      "learning_rate": 9.869643339288302e-06,
      "loss": 0.7609,
      "step": 1372
    },
    {
      "epoch": 0.10073644726102149,
      "grad_norm": 1.8747644972636122,
      "learning_rate": 9.869373654009795e-06,
      "loss": 0.939,
      "step": 1373
    },
    {
      "epoch": 0.10080981685116061,
      "grad_norm": 1.7215029935720958,
      "learning_rate": 9.8691036937457e-06,
      "loss": 1.022,
      "step": 1374
    },
    {
      "epoch": 0.10088318644129975,
      "grad_norm": 1.5901514167704616,
      "learning_rate": 9.86883345851126e-06,
      "loss": 0.9088,
      "step": 1375
    },
    {
      "epoch": 0.10095655603143887,
      "grad_norm": 1.7331104734973701,
      "learning_rate": 9.868562948321738e-06,
      "loss": 0.928,
      "step": 1376
    },
    {
      "epoch": 0.10095655603143887,
      "lm_loss": 1.4786,
      "step": 1376,
      "vm_loss": 0.0343
    },
    {
      "epoch": 0.10095655603143887,
      "lm_loss": 0.8744,
      "step": 1376,
      "vm_loss": 0.2108
    },
    {
      "epoch": 0.10095655603143887,
      "lm_loss": 0.9258,
      "step": 1376,
      "vm_loss": 0.2056
    },
    {
      "epoch": 0.10095655603143887,
      "lm_loss": 1.1936,
      "step": 1376,
      "vm_loss": 0.1787
    },
    {
      "epoch": 0.10095655603143887,
      "lm_loss": 0.0515,
      "step": 1376,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.10095655603143887,
      "lm_loss": 0.3088,
      "step": 1376,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.10095655603143887,
      "lm_loss": 1.1343,
      "step": 1376,
      "vm_loss": 0.2242
    },
    {
      "epoch": 0.10095655603143887,
      "lm_loss": 0.0239,
      "step": 1376,
      "vm_loss": 0.0927
    },
    {
      "epoch": 0.10102992562157799,
      "grad_norm": 1.8462149576996316,
      "learning_rate": 9.86829216319241e-06,
      "loss": 0.9628,
      "step": 1377
    },
    {
      "epoch": 0.10110329521171713,
      "grad_norm": 0.6790249374035983,
      "learning_rate": 9.868021103138567e-06,
      "loss": 0.5663,
      "step": 1378
    },
    {
      "epoch": 0.10117666480185625,
      "grad_norm": 0.6469379139648153,
      "learning_rate": 9.867749768175516e-06,
      "loss": 0.5863,
      "step": 1379
    },
    {
      "epoch": 0.10125003439199538,
      "grad_norm": 0.5801965605110686,
      "learning_rate": 9.867478158318582e-06,
      "loss": 0.56,
      "step": 1380
    },
    {
      "epoch": 0.1013234039821345,
      "grad_norm": 1.9892909726715997,
      "learning_rate": 9.867206273583103e-06,
      "loss": 0.819,
      "step": 1381
    },
    {
      "epoch": 0.10139677357227363,
      "grad_norm": 1.800152447222365,
      "learning_rate": 9.866934113984431e-06,
      "loss": 0.8603,
      "step": 1382
    },
    {
      "epoch": 0.10147014316241276,
      "grad_norm": 0.7520920625491339,
      "learning_rate": 9.866661679537937e-06,
      "loss": 0.5754,
      "step": 1383
    },
    {
      "epoch": 0.10154351275255188,
      "grad_norm": 1.6886405890295924,
      "learning_rate": 9.866388970259005e-06,
      "loss": 0.842,
      "step": 1384
    },
    {
      "epoch": 0.10154351275255188,
      "lm_loss": 0.6438,
      "step": 1384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10154351275255188,
      "lm_loss": 0.5261,
      "step": 1384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10154351275255188,
      "lm_loss": 0.6885,
      "step": 1384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10154351275255188,
      "lm_loss": 0.5559,
      "step": 1384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10154351275255188,
      "lm_loss": 0.5188,
      "step": 1384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10154351275255188,
      "lm_loss": 0.5874,
      "step": 1384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10154351275255188,
      "lm_loss": 0.4325,
      "step": 1384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10154351275255188,
      "lm_loss": 0.5826,
      "step": 1384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10161688234269102,
      "grad_norm": 0.628614051511754,
      "learning_rate": 9.866115986163038e-06,
      "loss": 0.5536,
      "step": 1385
    },
    {
      "epoch": 0.10169025193283014,
      "grad_norm": 0.6073393604724562,
      "learning_rate": 9.86584272726545e-06,
      "loss": 0.5706,
      "step": 1386
    },
    {
      "epoch": 0.10176362152296926,
      "grad_norm": 1.9119977103069865,
      "learning_rate": 9.865569193581671e-06,
      "loss": 0.9247,
      "step": 1387
    },
    {
      "epoch": 0.1018369911131084,
      "grad_norm": 0.6395117780954267,
      "learning_rate": 9.865295385127153e-06,
      "loss": 0.5896,
      "step": 1388
    },
    {
      "epoch": 0.10191036070324752,
      "grad_norm": 1.9100444508681802,
      "learning_rate": 9.865021301917358e-06,
      "loss": 0.9862,
      "step": 1389
    },
    {
      "epoch": 0.10198373029338664,
      "grad_norm": 0.6732459195233548,
      "learning_rate": 9.864746943967758e-06,
      "loss": 0.579,
      "step": 1390
    },
    {
      "epoch": 0.10205709988352578,
      "grad_norm": 1.690952139270686,
      "learning_rate": 9.864472311293851e-06,
      "loss": 0.892,
      "step": 1391
    },
    {
      "epoch": 0.1021304694736649,
      "grad_norm": 0.6130748857506111,
      "learning_rate": 9.864197403911147e-06,
      "loss": 0.5782,
      "step": 1392
    },
    {
      "epoch": 0.1021304694736649,
      "lm_loss": 1.6219,
      "step": 1392,
      "vm_loss": 0.0848
    },
    {
      "epoch": 0.1021304694736649,
      "lm_loss": 0.497,
      "step": 1392,
      "vm_loss": 0.0533
    },
    {
      "epoch": 0.1021304694736649,
      "lm_loss": 0.0732,
      "step": 1392,
      "vm_loss": 0.0964
    },
    {
      "epoch": 0.1021304694736649,
      "lm_loss": 0.4226,
      "step": 1392,
      "vm_loss": 0.158
    },
    {
      "epoch": 0.1021304694736649,
      "lm_loss": 0.4463,
      "step": 1392,
      "vm_loss": 0.1018
    },
    {
      "epoch": 0.1021304694736649,
      "lm_loss": 0.8354,
      "step": 1392,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.1021304694736649,
      "lm_loss": 1.0393,
      "step": 1392,
      "vm_loss": 0.177
    },
    {
      "epoch": 0.1021304694736649,
      "lm_loss": 0.9075,
      "step": 1392,
      "vm_loss": 0.1755
    },
    {
      "epoch": 0.10220383906380404,
      "grad_norm": 1.6931921586651588,
      "learning_rate": 9.86392222183517e-06,
      "loss": 0.938,
      "step": 1393
    },
    {
      "epoch": 0.10227720865394316,
      "grad_norm": 1.7873412248735925,
      "learning_rate": 9.863646765081458e-06,
      "loss": 0.9717,
      "step": 1394
    },
    {
      "epoch": 0.10235057824408228,
      "grad_norm": 0.7779634886882094,
      "learning_rate": 9.86337103366557e-06,
      "loss": 0.5719,
      "step": 1395
    },
    {
      "epoch": 0.10242394783422142,
      "grad_norm": 1.568699291148138,
      "learning_rate": 9.863095027603078e-06,
      "loss": 0.739,
      "step": 1396
    },
    {
      "epoch": 0.10249731742436054,
      "grad_norm": 0.6114169122854237,
      "learning_rate": 9.862818746909562e-06,
      "loss": 0.5683,
      "step": 1397
    },
    {
      "epoch": 0.10257068701449966,
      "grad_norm": 1.6741654746543668,
      "learning_rate": 9.862542191600632e-06,
      "loss": 0.7942,
      "step": 1398
    },
    {
      "epoch": 0.1026440566046388,
      "grad_norm": 0.600576499123685,
      "learning_rate": 9.862265361691901e-06,
      "loss": 0.5762,
      "step": 1399
    },
    {
      "epoch": 0.10271742619477792,
      "grad_norm": 1.6881521980023375,
      "learning_rate": 9.861988257199005e-06,
      "loss": 0.8817,
      "step": 1400
    },
    {
      "epoch": 0.10271742619477792,
      "lm_loss": 0.6232,
      "step": 1400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10271742619477792,
      "lm_loss": 0.672,
      "step": 1400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10271742619477792,
      "lm_loss": 0.5691,
      "step": 1400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10271742619477792,
      "lm_loss": 0.4107,
      "step": 1400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10271742619477792,
      "lm_loss": 0.6344,
      "step": 1400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10271742619477792,
      "lm_loss": 0.5562,
      "step": 1400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10271742619477792,
      "lm_loss": 0.2996,
      "step": 1400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10271742619477792,
      "lm_loss": 0.5543,
      "step": 1400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10279079578491705,
      "grad_norm": 0.676868663461447,
      "learning_rate": 9.861710878137591e-06,
      "loss": 0.5705,
      "step": 1401
    },
    {
      "epoch": 0.10286416537505617,
      "grad_norm": 0.6526405546535923,
      "learning_rate": 9.861433224523324e-06,
      "loss": 0.5798,
      "step": 1402
    },
    {
      "epoch": 0.1029375349651953,
      "grad_norm": 0.5960937786589253,
      "learning_rate": 9.861155296371883e-06,
      "loss": 0.5692,
      "step": 1403
    },
    {
      "epoch": 0.10301090455533443,
      "grad_norm": 0.5590468804990016,
      "learning_rate": 9.860877093698964e-06,
      "loss": 0.561,
      "step": 1404
    },
    {
      "epoch": 0.10308427414547355,
      "grad_norm": 0.5997546172311803,
      "learning_rate": 9.860598616520279e-06,
      "loss": 0.5745,
      "step": 1405
    },
    {
      "epoch": 0.10315764373561268,
      "grad_norm": 2.010683212341099,
      "learning_rate": 9.860319864851552e-06,
      "loss": 0.9414,
      "step": 1406
    },
    {
      "epoch": 0.10323101332575181,
      "grad_norm": 1.779464114218687,
      "learning_rate": 9.860040838708526e-06,
      "loss": 0.8839,
      "step": 1407
    },
    {
      "epoch": 0.10330438291589093,
      "grad_norm": 1.671115501444992,
      "learning_rate": 9.859761538106958e-06,
      "loss": 0.9947,
      "step": 1408
    },
    {
      "epoch": 0.10330438291589093,
      "lm_loss": 1.5797,
      "step": 1408,
      "vm_loss": 0.0661
    },
    {
      "epoch": 0.10330438291589093,
      "lm_loss": 0.3257,
      "step": 1408,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.10330438291589093,
      "lm_loss": 0.4717,
      "step": 1408,
      "vm_loss": 0.1535
    },
    {
      "epoch": 0.10330438291589093,
      "lm_loss": 0.7798,
      "step": 1408,
      "vm_loss": 0.1631
    },
    {
      "epoch": 0.10330438291589093,
      "lm_loss": 1.1173,
      "step": 1408,
      "vm_loss": 0.1314
    },
    {
      "epoch": 0.10330438291589093,
      "lm_loss": 1.367,
      "step": 1408,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.10330438291589093,
      "lm_loss": 0.8985,
      "step": 1408,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.10330438291589093,
      "lm_loss": 0.4633,
      "step": 1408,
      "vm_loss": 0.1423
    },
    {
      "epoch": 0.10337775250603007,
      "grad_norm": 1.845050257979234,
      "learning_rate": 9.859481963062623e-06,
      "loss": 0.9113,
      "step": 1409
    },
    {
      "epoch": 0.10345112209616919,
      "grad_norm": 1.8848995304913598,
      "learning_rate": 9.859202113591303e-06,
      "loss": 0.8438,
      "step": 1410
    },
    {
      "epoch": 0.10352449168630831,
      "grad_norm": 1.7423025679921798,
      "learning_rate": 9.858921989708809e-06,
      "loss": 0.9152,
      "step": 1411
    },
    {
      "epoch": 0.10359786127644745,
      "grad_norm": 0.9890003842272728,
      "learning_rate": 9.858641591430958e-06,
      "loss": 0.596,
      "step": 1412
    },
    {
      "epoch": 0.10367123086658657,
      "grad_norm": 1.7717614260954464,
      "learning_rate": 9.858360918773583e-06,
      "loss": 0.8569,
      "step": 1413
    },
    {
      "epoch": 0.1037446004567257,
      "grad_norm": 1.7971363931381916,
      "learning_rate": 9.858079971752535e-06,
      "loss": 0.8253,
      "step": 1414
    },
    {
      "epoch": 0.10381797004686483,
      "grad_norm": 1.872122347229277,
      "learning_rate": 9.857798750383681e-06,
      "loss": 0.9045,
      "step": 1415
    },
    {
      "epoch": 0.10389133963700395,
      "grad_norm": 1.4659110533342883,
      "learning_rate": 9.857517254682902e-06,
      "loss": 0.8412,
      "step": 1416
    },
    {
      "epoch": 0.10389133963700395,
      "lm_loss": 1.3048,
      "step": 1416,
      "vm_loss": 0.091
    },
    {
      "epoch": 0.10389133963700395,
      "lm_loss": 0.3436,
      "step": 1416,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.10389133963700395,
      "lm_loss": 0.1338,
      "step": 1416,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.10389133963700395,
      "lm_loss": 0.3444,
      "step": 1416,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.10389133963700395,
      "lm_loss": 0.0516,
      "step": 1416,
      "vm_loss": 0.1271
    },
    {
      "epoch": 0.10389133963700395,
      "lm_loss": 0.6788,
      "step": 1416,
      "vm_loss": 0.2089
    },
    {
      "epoch": 0.10389133963700395,
      "lm_loss": 0.8219,
      "step": 1416,
      "vm_loss": 0.0772
    },
    {
      "epoch": 0.10389133963700395,
      "lm_loss": 0.7891,
      "step": 1416,
      "vm_loss": 0.1632
    },
    {
      "epoch": 0.10396470922714308,
      "grad_norm": 1.620481687388103,
      "learning_rate": 9.857235484666093e-06,
      "loss": 0.8426,
      "step": 1417
    },
    {
      "epoch": 0.1040380788172822,
      "grad_norm": 0.7535934152751089,
      "learning_rate": 9.856953440349167e-06,
      "loss": 0.5483,
      "step": 1418
    },
    {
      "epoch": 0.10411144840742133,
      "grad_norm": 2.281000699670842,
      "learning_rate": 9.856671121748053e-06,
      "loss": 0.9637,
      "step": 1419
    },
    {
      "epoch": 0.10418481799756046,
      "grad_norm": 2.126645377056296,
      "learning_rate": 9.856388528878693e-06,
      "loss": 0.9671,
      "step": 1420
    },
    {
      "epoch": 0.10425818758769959,
      "grad_norm": 0.6219212623295823,
      "learning_rate": 9.856105661757045e-06,
      "loss": 0.5629,
      "step": 1421
    },
    {
      "epoch": 0.10433155717783872,
      "grad_norm": 0.610580958684777,
      "learning_rate": 9.855822520399087e-06,
      "loss": 0.5727,
      "step": 1422
    },
    {
      "epoch": 0.10440492676797784,
      "grad_norm": 1.8512049507904889,
      "learning_rate": 9.855539104820804e-06,
      "loss": 0.8378,
      "step": 1423
    },
    {
      "epoch": 0.10447829635811696,
      "grad_norm": 1.9867432855495752,
      "learning_rate": 9.855255415038205e-06,
      "loss": 0.8528,
      "step": 1424
    },
    {
      "epoch": 0.10447829635811696,
      "lm_loss": 0.585,
      "step": 1424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10447829635811696,
      "lm_loss": 0.6253,
      "step": 1424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10447829635811696,
      "lm_loss": 0.5565,
      "step": 1424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10447829635811696,
      "lm_loss": 0.6525,
      "step": 1424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10447829635811696,
      "lm_loss": 0.5453,
      "step": 1424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10447829635811696,
      "lm_loss": 0.7037,
      "step": 1424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10447829635811696,
      "lm_loss": 0.6241,
      "step": 1424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10447829635811696,
      "lm_loss": 0.4194,
      "step": 1424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1045516659482561,
      "grad_norm": 0.7378296087668671,
      "learning_rate": 9.854971451067308e-06,
      "loss": 0.5535,
      "step": 1425
    },
    {
      "epoch": 0.10462503553839522,
      "grad_norm": 1.9433167838395045,
      "learning_rate": 9.854687212924149e-06,
      "loss": 0.8924,
      "step": 1426
    },
    {
      "epoch": 0.10469840512853434,
      "grad_norm": 1.7689303657685758,
      "learning_rate": 9.854402700624782e-06,
      "loss": 0.8561,
      "step": 1427
    },
    {
      "epoch": 0.10477177471867348,
      "grad_norm": 1.7549775935622847,
      "learning_rate": 9.854117914185271e-06,
      "loss": 0.9313,
      "step": 1428
    },
    {
      "epoch": 0.1048451443088126,
      "grad_norm": 0.6520192388972211,
      "learning_rate": 9.853832853621703e-06,
      "loss": 0.5844,
      "step": 1429
    },
    {
      "epoch": 0.10491851389895174,
      "grad_norm": 0.6412226503466825,
      "learning_rate": 9.853547518950171e-06,
      "loss": 0.5867,
      "step": 1430
    },
    {
      "epoch": 0.10499188348909086,
      "grad_norm": 0.6314149092556169,
      "learning_rate": 9.853261910186793e-06,
      "loss": 0.571,
      "step": 1431
    },
    {
      "epoch": 0.10506525307922998,
      "grad_norm": 2.269635984658962,
      "learning_rate": 9.852976027347693e-06,
      "loss": 0.9404,
      "step": 1432
    },
    {
      "epoch": 0.10506525307922998,
      "lm_loss": 0.556,
      "step": 1432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10506525307922998,
      "lm_loss": 0.7014,
      "step": 1432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10506525307922998,
      "lm_loss": 0.4382,
      "step": 1432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10506525307922998,
      "lm_loss": 0.4494,
      "step": 1432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10506525307922998,
      "lm_loss": 0.5601,
      "step": 1432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10506525307922998,
      "lm_loss": 0.7455,
      "step": 1432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10506525307922998,
      "lm_loss": 0.4531,
      "step": 1432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10506525307922998,
      "lm_loss": 0.8314,
      "step": 1432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10513862266936912,
      "grad_norm": 0.6560221647899559,
      "learning_rate": 9.852689870449022e-06,
      "loss": 0.5737,
      "step": 1433
    },
    {
      "epoch": 0.10521199225950824,
      "grad_norm": 0.6007352904536093,
      "learning_rate": 9.852403439506935e-06,
      "loss": 0.5651,
      "step": 1434
    },
    {
      "epoch": 0.10528536184964737,
      "grad_norm": 1.9139082805084886,
      "learning_rate": 9.85211673453761e-06,
      "loss": 0.8741,
      "step": 1435
    },
    {
      "epoch": 0.1053587314397865,
      "grad_norm": 1.8216815994539919,
      "learning_rate": 9.851829755557237e-06,
      "loss": 0.8813,
      "step": 1436
    },
    {
      "epoch": 0.10543210102992562,
      "grad_norm": 0.6543987058397496,
      "learning_rate": 9.85154250258202e-06,
      "loss": 0.5751,
      "step": 1437
    },
    {
      "epoch": 0.10550547062006475,
      "grad_norm": 1.8739489591220848,
      "learning_rate": 9.851254975628184e-06,
      "loss": 0.9134,
      "step": 1438
    },
    {
      "epoch": 0.10557884021020388,
      "grad_norm": 1.7240923061688584,
      "learning_rate": 9.850967174711968e-06,
      "loss": 0.8941,
      "step": 1439
    },
    {
      "epoch": 0.105652209800343,
      "grad_norm": 1.7072941356955431,
      "learning_rate": 9.85067909984962e-06,
      "loss": 0.9111,
      "step": 1440
    },
    {
      "epoch": 0.105652209800343,
      "lm_loss": 1.4818,
      "step": 1440,
      "vm_loss": 0.0453
    },
    {
      "epoch": 0.105652209800343,
      "lm_loss": 0.313,
      "step": 1440,
      "vm_loss": 0.1914
    },
    {
      "epoch": 0.105652209800343,
      "lm_loss": 1.0782,
      "step": 1440,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.105652209800343,
      "lm_loss": 0.7717,
      "step": 1440,
      "vm_loss": 0.0935
    },
    {
      "epoch": 0.105652209800343,
      "lm_loss": 1.2319,
      "step": 1440,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.105652209800343,
      "lm_loss": 0.1387,
      "step": 1440,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.105652209800343,
      "lm_loss": 0.0688,
      "step": 1440,
      "vm_loss": 0.1491
    },
    {
      "epoch": 0.105652209800343,
      "lm_loss": 0.2702,
      "step": 1440,
      "vm_loss": 0.1428
    },
    {
      "epoch": 0.10572557939048213,
      "grad_norm": 1.7686776111183715,
      "learning_rate": 9.85039075105741e-06,
      "loss": 0.847,
      "step": 1441
    },
    {
      "epoch": 0.10579894898062125,
      "grad_norm": 0.6906833549441087,
      "learning_rate": 9.850102128351625e-06,
      "loss": 0.5606,
      "step": 1442
    },
    {
      "epoch": 0.10587231857076039,
      "grad_norm": 1.6313154069471827,
      "learning_rate": 9.84981323174856e-06,
      "loss": 0.8478,
      "step": 1443
    },
    {
      "epoch": 0.10594568816089951,
      "grad_norm": 1.7664532221473856,
      "learning_rate": 9.849524061264533e-06,
      "loss": 0.9656,
      "step": 1444
    },
    {
      "epoch": 0.10601905775103863,
      "grad_norm": 1.7667649727076942,
      "learning_rate": 9.849234616915871e-06,
      "loss": 0.884,
      "step": 1445
    },
    {
      "epoch": 0.10609242734117777,
      "grad_norm": 1.6902385067373165,
      "learning_rate": 9.848944898718922e-06,
      "loss": 0.9221,
      "step": 1446
    },
    {
      "epoch": 0.10616579693131689,
      "grad_norm": 1.6311237626478734,
      "learning_rate": 9.848654906690047e-06,
      "loss": 0.7943,
      "step": 1447
    },
    {
      "epoch": 0.10623916652145601,
      "grad_norm": 0.6345626048343568,
      "learning_rate": 9.848364640845621e-06,
      "loss": 0.5628,
      "step": 1448
    },
    {
      "epoch": 0.10623916652145601,
      "lm_loss": 0.6216,
      "step": 1448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10623916652145601,
      "lm_loss": 0.4703,
      "step": 1448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10623916652145601,
      "lm_loss": 0.6252,
      "step": 1448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10623916652145601,
      "lm_loss": 0.4677,
      "step": 1448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10623916652145601,
      "lm_loss": 0.4548,
      "step": 1448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10623916652145601,
      "lm_loss": 0.3423,
      "step": 1448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10623916652145601,
      "lm_loss": 0.6758,
      "step": 1448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10623916652145601,
      "lm_loss": 0.5804,
      "step": 1448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.10631253611159515,
      "grad_norm": 0.6402002532535139,
      "learning_rate": 9.848074101202037e-06,
      "loss": 0.5586,
      "step": 1449
    },
    {
      "epoch": 0.10638590570173427,
      "grad_norm": 1.8995962538106548,
      "learning_rate": 9.847783287775703e-06,
      "loss": 0.9008,
      "step": 1450
    },
    {
      "epoch": 0.1064592752918734,
      "grad_norm": 1.8565544719193368,
      "learning_rate": 9.847492200583041e-06,
      "loss": 0.8947,
      "step": 1451
    },
    {
      "epoch": 0.10653264488201253,
      "grad_norm": 1.7923729547877998,
      "learning_rate": 9.847200839640489e-06,
      "loss": 0.8916,
      "step": 1452
    },
    {
      "epoch": 0.10660601447215165,
      "grad_norm": 0.6757050723419348,
      "learning_rate": 9.846909204964502e-06,
      "loss": 0.5705,
      "step": 1453
    },
    {
      "epoch": 0.10667938406229079,
      "grad_norm": 0.6246073616523808,
      "learning_rate": 9.84661729657155e-06,
      "loss": 0.5508,
      "step": 1454
    },
    {
      "epoch": 0.10675275365242991,
      "grad_norm": 1.7217126649497962,
      "learning_rate": 9.846325114478117e-06,
      "loss": 0.827,
      "step": 1455
    },
    {
      "epoch": 0.10682612324256903,
      "grad_norm": 1.7947367527721843,
      "learning_rate": 9.846032658700704e-06,
      "loss": 0.8827,
      "step": 1456
    },
    {
      "epoch": 0.10682612324256903,
      "lm_loss": 1.8615,
      "step": 1456,
      "vm_loss": 0.1248
    },
    {
      "epoch": 0.10682612324256903,
      "lm_loss": 1.0304,
      "step": 1456,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.10682612324256903,
      "lm_loss": 0.8684,
      "step": 1456,
      "vm_loss": 0.1643
    },
    {
      "epoch": 0.10682612324256903,
      "lm_loss": 0.8393,
      "step": 1456,
      "vm_loss": 0.1463
    },
    {
      "epoch": 0.10682612324256903,
      "lm_loss": 0.6435,
      "step": 1456,
      "vm_loss": 0.1861
    },
    {
      "epoch": 0.10682612324256903,
      "lm_loss": 0.1235,
      "step": 1456,
      "vm_loss": 0.1065
    },
    {
      "epoch": 0.10682612324256903,
      "lm_loss": 0.0844,
      "step": 1456,
      "vm_loss": 0.1581
    },
    {
      "epoch": 0.10682612324256903,
      "lm_loss": 0.8308,
      "step": 1456,
      "vm_loss": 0.1495
    },
    {
      "epoch": 0.10689949283270817,
      "grad_norm": 1.847543945298392,
      "learning_rate": 9.845739929255823e-06,
      "loss": 0.9498,
      "step": 1457
    },
    {
      "epoch": 0.10697286242284729,
      "grad_norm": 1.681118220973734,
      "learning_rate": 9.845446926160009e-06,
      "loss": 0.9107,
      "step": 1458
    },
    {
      "epoch": 0.10704623201298642,
      "grad_norm": 2.196749466320404,
      "learning_rate": 9.845153649429808e-06,
      "loss": 1.0104,
      "step": 1459
    },
    {
      "epoch": 0.10711960160312554,
      "grad_norm": 1.54041954142902,
      "learning_rate": 9.84486009908178e-06,
      "loss": 0.8582,
      "step": 1460
    },
    {
      "epoch": 0.10719297119326467,
      "grad_norm": 1.7587879493017735,
      "learning_rate": 9.844566275132508e-06,
      "loss": 0.9333,
      "step": 1461
    },
    {
      "epoch": 0.1072663407834038,
      "grad_norm": 1.7260628682677701,
      "learning_rate": 9.844272177598577e-06,
      "loss": 0.8197,
      "step": 1462
    },
    {
      "epoch": 0.10733971037354292,
      "grad_norm": 1.6672025779512505,
      "learning_rate": 9.843977806496601e-06,
      "loss": 0.7806,
      "step": 1463
    },
    {
      "epoch": 0.10741307996368206,
      "grad_norm": 2.109737158143538,
      "learning_rate": 9.843683161843202e-06,
      "loss": 1.03,
      "step": 1464
    },
    {
      "epoch": 0.10741307996368206,
      "lm_loss": 1.2836,
      "step": 1464,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.10741307996368206,
      "lm_loss": 0.2536,
      "step": 1464,
      "vm_loss": 0.1401
    },
    {
      "epoch": 0.10741307996368206,
      "lm_loss": 0.4671,
      "step": 1464,
      "vm_loss": 0.1352
    },
    {
      "epoch": 0.10741307996368206,
      "lm_loss": 0.8483,
      "step": 1464,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.10741307996368206,
      "lm_loss": 0.4073,
      "step": 1464,
      "vm_loss": 0.1944
    },
    {
      "epoch": 0.10741307996368206,
      "lm_loss": 0.6118,
      "step": 1464,
      "vm_loss": 0.1344
    },
    {
      "epoch": 0.10741307996368206,
      "lm_loss": 1.143,
      "step": 1464,
      "vm_loss": 0.1553
    },
    {
      "epoch": 0.10741307996368206,
      "lm_loss": 1.0416,
      "step": 1464,
      "vm_loss": 0.1584
    },
    {
      "epoch": 0.10748644955382118,
      "grad_norm": 1.880870944907211,
      "learning_rate": 9.843388243655021e-06,
      "loss": 0.9371,
      "step": 1465
    },
    {
      "epoch": 0.1075598191439603,
      "grad_norm": 1.663595663983944,
      "learning_rate": 9.84309305194871e-06,
      "loss": 0.8337,
      "step": 1466
    },
    {
      "epoch": 0.10763318873409944,
      "grad_norm": 1.851774642720001,
      "learning_rate": 9.842797586740942e-06,
      "loss": 0.9308,
      "step": 1467
    },
    {
      "epoch": 0.10770655832423856,
      "grad_norm": 1.7300900381293285,
      "learning_rate": 9.8425018480484e-06,
      "loss": 0.9525,
      "step": 1468
    },
    {
      "epoch": 0.10777992791437768,
      "grad_norm": 1.5876372779126284,
      "learning_rate": 9.842205835887785e-06,
      "loss": 0.8584,
      "step": 1469
    },
    {
      "epoch": 0.10785329750451682,
      "grad_norm": 1.9275174062315237,
      "learning_rate": 9.84190955027582e-06,
      "loss": 0.8744,
      "step": 1470
    },
    {
      "epoch": 0.10792666709465594,
      "grad_norm": 1.7398297995147354,
      "learning_rate": 9.841612991229228e-06,
      "loss": 1.0024,
      "step": 1471
    },
    {
      "epoch": 0.10800003668479508,
      "grad_norm": 1.3463660004101927,
      "learning_rate": 9.84131615876476e-06,
      "loss": 0.5896,
      "step": 1472
    },
    {
      "epoch": 0.10800003668479508,
      "lm_loss": 1.4923,
      "step": 1472,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.10800003668479508,
      "lm_loss": 0.6014,
      "step": 1472,
      "vm_loss": 0.2232
    },
    {
      "epoch": 0.10800003668479508,
      "lm_loss": 0.8949,
      "step": 1472,
      "vm_loss": 0.1268
    },
    {
      "epoch": 0.10800003668479508,
      "lm_loss": 1.2735,
      "step": 1472,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.10800003668479508,
      "lm_loss": 1.3649,
      "step": 1472,
      "vm_loss": 0.174
    },
    {
      "epoch": 0.10800003668479508,
      "lm_loss": 0.9748,
      "step": 1472,
      "vm_loss": 0.1435
    },
    {
      "epoch": 0.10800003668479508,
      "lm_loss": 0.6593,
      "step": 1472,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.10800003668479508,
      "lm_loss": 0.0432,
      "step": 1472,
      "vm_loss": 0.0919
    },
    {
      "epoch": 0.1080734062749342,
      "grad_norm": 1.978233592872041,
      "learning_rate": 9.84101905289918e-06,
      "loss": 0.8875,
      "step": 1473
    },
    {
      "epoch": 0.10814677586507332,
      "grad_norm": 1.747066660284599,
      "learning_rate": 9.840721673649265e-06,
      "loss": 0.8368,
      "step": 1474
    },
    {
      "epoch": 0.10822014545521245,
      "grad_norm": 1.899392738842545,
      "learning_rate": 9.84042402103181e-06,
      "loss": 0.8739,
      "step": 1475
    },
    {
      "epoch": 0.10829351504535158,
      "grad_norm": 1.829587789146768,
      "learning_rate": 9.840126095063624e-06,
      "loss": 0.9918,
      "step": 1476
    },
    {
      "epoch": 0.1083668846354907,
      "grad_norm": 1.2326670839311813,
      "learning_rate": 9.83982789576153e-06,
      "loss": 0.5592,
      "step": 1477
    },
    {
      "epoch": 0.10844025422562983,
      "grad_norm": 1.0065805922407343,
      "learning_rate": 9.839529423142369e-06,
      "loss": 0.563,
      "step": 1478
    },
    {
      "epoch": 0.10851362381576896,
      "grad_norm": 1.851352101625139,
      "learning_rate": 9.839230677222997e-06,
      "loss": 0.8262,
      "step": 1479
    },
    {
      "epoch": 0.10858699340590809,
      "grad_norm": 1.978404973164708,
      "learning_rate": 9.838931658020286e-06,
      "loss": 0.9305,
      "step": 1480
    },
    {
      "epoch": 0.10858699340590809,
      "lm_loss": 0.7778,
      "step": 1480,
      "vm_loss": 0.0266
    },
    {
      "epoch": 0.10858699340590809,
      "lm_loss": 0.9917,
      "step": 1480,
      "vm_loss": 0.1215
    },
    {
      "epoch": 0.10858699340590809,
      "lm_loss": 0.1477,
      "step": 1480,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.10858699340590809,
      "lm_loss": 1.465,
      "step": 1480,
      "vm_loss": 0.1085
    },
    {
      "epoch": 0.10858699340590809,
      "lm_loss": 1.0431,
      "step": 1480,
      "vm_loss": 0.1851
    },
    {
      "epoch": 0.10858699340590809,
      "lm_loss": 1.0384,
      "step": 1480,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.10858699340590809,
      "lm_loss": 0.3533,
      "step": 1480,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.10858699340590809,
      "lm_loss": 0.8591,
      "step": 1480,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.10866036299604721,
      "grad_norm": 1.8142689548550615,
      "learning_rate": 9.838632365551119e-06,
      "loss": 0.826,
      "step": 1481
    },
    {
      "epoch": 0.10873373258618634,
      "grad_norm": 0.9451536686038512,
      "learning_rate": 9.8383327998324e-06,
      "loss": 0.584,
      "step": 1482
    },
    {
      "epoch": 0.10880710217632547,
      "grad_norm": 1.767837120829489,
      "learning_rate": 9.838032960881045e-06,
      "loss": 0.8664,
      "step": 1483
    },
    {
      "epoch": 0.10888047176646459,
      "grad_norm": 1.7455430523710393,
      "learning_rate": 9.837732848713989e-06,
      "loss": 0.868,
      "step": 1484
    },
    {
      "epoch": 0.10895384135660373,
      "grad_norm": 0.8072809287023187,
      "learning_rate": 9.837432463348177e-06,
      "loss": 0.5696,
      "step": 1485
    },
    {
      "epoch": 0.10902721094674285,
      "grad_norm": 1.6241511784409497,
      "learning_rate": 9.837131804800577e-06,
      "loss": 0.8761,
      "step": 1486
    },
    {
      "epoch": 0.10910058053688197,
      "grad_norm": 1.8429476751923277,
      "learning_rate": 9.836830873088163e-06,
      "loss": 0.893,
      "step": 1487
    },
    {
      "epoch": 0.10917395012702111,
      "grad_norm": 1.9637790828575221,
      "learning_rate": 9.836529668227932e-06,
      "loss": 0.8283,
      "step": 1488
    },
    {
      "epoch": 0.10917395012702111,
      "lm_loss": 1.4848,
      "step": 1488,
      "vm_loss": 0.047
    },
    {
      "epoch": 0.10917395012702111,
      "lm_loss": 0.7704,
      "step": 1488,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.10917395012702111,
      "lm_loss": 0.5928,
      "step": 1488,
      "vm_loss": 0.1249
    },
    {
      "epoch": 0.10917395012702111,
      "lm_loss": 1.2633,
      "step": 1488,
      "vm_loss": 0.1292
    },
    {
      "epoch": 0.10917395012702111,
      "lm_loss": 0.138,
      "step": 1488,
      "vm_loss": 0.0723
    },
    {
      "epoch": 0.10917395012702111,
      "lm_loss": 0.6134,
      "step": 1488,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.10917395012702111,
      "lm_loss": 0.5849,
      "step": 1488,
      "vm_loss": 0.1231
    },
    {
      "epoch": 0.10917395012702111,
      "lm_loss": 0.0789,
      "step": 1488,
      "vm_loss": 0.1149
    },
    {
      "epoch": 0.10924731971716023,
      "grad_norm": 1.8741429753941765,
      "learning_rate": 9.836228190236892e-06,
      "loss": 0.8994,
      "step": 1489
    },
    {
      "epoch": 0.10932068930729935,
      "grad_norm": 0.6864961514306323,
      "learning_rate": 9.83592643913207e-06,
      "loss": 0.5753,
      "step": 1490
    },
    {
      "epoch": 0.10939405889743849,
      "grad_norm": 1.623834338381151,
      "learning_rate": 9.835624414930506e-06,
      "loss": 0.8758,
      "step": 1491
    },
    {
      "epoch": 0.10946742848757761,
      "grad_norm": 1.5290351440241232,
      "learning_rate": 9.835322117649255e-06,
      "loss": 0.7437,
      "step": 1492
    },
    {
      "epoch": 0.10954079807771674,
      "grad_norm": 1.8281406516146084,
      "learning_rate": 9.83501954730539e-06,
      "loss": 1.0453,
      "step": 1493
    },
    {
      "epoch": 0.10961416766785587,
      "grad_norm": 1.5736194224250435,
      "learning_rate": 9.834716703916e-06,
      "loss": 0.8352,
      "step": 1494
    },
    {
      "epoch": 0.10968753725799499,
      "grad_norm": 2.0104956282358892,
      "learning_rate": 9.834413587498182e-06,
      "loss": 0.9524,
      "step": 1495
    },
    {
      "epoch": 0.10976090684813412,
      "grad_norm": 0.7370978007666739,
      "learning_rate": 9.834110198069058e-06,
      "loss": 0.5958,
      "step": 1496
    },
    {
      "epoch": 0.10976090684813412,
      "lm_loss": 1.3192,
      "step": 1496,
      "vm_loss": 0.0736
    },
    {
      "epoch": 0.10976090684813412,
      "lm_loss": 1.3903,
      "step": 1496,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.10976090684813412,
      "lm_loss": 0.9202,
      "step": 1496,
      "vm_loss": 0.1269
    },
    {
      "epoch": 0.10976090684813412,
      "lm_loss": 0.5273,
      "step": 1496,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.10976090684813412,
      "lm_loss": 0.739,
      "step": 1496,
      "vm_loss": 0.1077
    },
    {
      "epoch": 0.10976090684813412,
      "lm_loss": 0.4932,
      "step": 1496,
      "vm_loss": 0.2887
    },
    {
      "epoch": 0.10976090684813412,
      "lm_loss": 0.0672,
      "step": 1496,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.10976090684813412,
      "lm_loss": 0.1206,
      "step": 1496,
      "vm_loss": 0.0731
    },
    {
      "epoch": 0.10983427643827325,
      "grad_norm": 1.5635336413701897,
      "learning_rate": 9.833806535645759e-06,
      "loss": 0.9464,
      "step": 1497
    },
    {
      "epoch": 0.10990764602841237,
      "grad_norm": 1.9070364704881755,
      "learning_rate": 9.833502600245435e-06,
      "loss": 0.9547,
      "step": 1498
    },
    {
      "epoch": 0.1099810156185515,
      "grad_norm": 1.6374035360238175,
      "learning_rate": 9.833198391885248e-06,
      "loss": 0.9072,
      "step": 1499
    },
    {
      "epoch": 0.11005438520869062,
      "grad_norm": 0.6575946259720862,
      "learning_rate": 9.832893910582378e-06,
      "loss": 0.5926,
      "step": 1500
    },
    {
      "epoch": 0.11012775479882976,
      "grad_norm": 1.74531542140111,
      "learning_rate": 9.832589156354021e-06,
      "loss": 0.8845,
      "step": 1501
    },
    {
      "epoch": 0.11020112438896888,
      "grad_norm": 0.6120553171570788,
      "learning_rate": 9.832284129217389e-06,
      "loss": 0.5572,
      "step": 1502
    },
    {
      "epoch": 0.110274493979108,
      "grad_norm": 0.587864494475871,
      "learning_rate": 9.831978829189702e-06,
      "loss": 0.5807,
      "step": 1503
    },
    {
      "epoch": 0.11034786356924714,
      "grad_norm": 1.7302314979408453,
      "learning_rate": 9.831673256288206e-06,
      "loss": 0.8125,
      "step": 1504
    },
    {
      "epoch": 0.11034786356924714,
      "lm_loss": 1.3222,
      "step": 1504,
      "vm_loss": 0.1848
    },
    {
      "epoch": 0.11034786356924714,
      "lm_loss": 0.2632,
      "step": 1504,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.11034786356924714,
      "lm_loss": 0.9136,
      "step": 1504,
      "vm_loss": 0.1635
    },
    {
      "epoch": 0.11034786356924714,
      "lm_loss": 0.193,
      "step": 1504,
      "vm_loss": 0.1357
    },
    {
      "epoch": 0.11034786356924714,
      "lm_loss": 0.5162,
      "step": 1504,
      "vm_loss": 0.1344
    },
    {
      "epoch": 0.11034786356924714,
      "lm_loss": 0.7856,
      "step": 1504,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.11034786356924714,
      "lm_loss": 1.3433,
      "step": 1504,
      "vm_loss": 0.0772
    },
    {
      "epoch": 0.11034786356924714,
      "lm_loss": 0.0907,
      "step": 1504,
      "vm_loss": 0.1844
    },
    {
      "epoch": 0.11042123315938626,
      "grad_norm": 1.9042174814357178,
      "learning_rate": 9.831367410530154e-06,
      "loss": 0.9277,
      "step": 1505
    },
    {
      "epoch": 0.1104946027495254,
      "grad_norm": 1.7160213757724712,
      "learning_rate": 9.831061291932822e-06,
      "loss": 0.9303,
      "step": 1506
    },
    {
      "epoch": 0.11056797233966452,
      "grad_norm": 0.6382247659952183,
      "learning_rate": 9.830754900513495e-06,
      "loss": 0.5732,
      "step": 1507
    },
    {
      "epoch": 0.11064134192980364,
      "grad_norm": 1.7816075049925302,
      "learning_rate": 9.830448236289474e-06,
      "loss": 0.9804,
      "step": 1508
    },
    {
      "epoch": 0.11071471151994278,
      "grad_norm": 1.5962657230176052,
      "learning_rate": 9.83014129927808e-06,
      "loss": 0.8098,
      "step": 1509
    },
    {
      "epoch": 0.1107880811100819,
      "grad_norm": 1.5587247720925013,
      "learning_rate": 9.829834089496644e-06,
      "loss": 1.0189,
      "step": 1510
    },
    {
      "epoch": 0.11086145070022102,
      "grad_norm": 1.6204594859552752,
      "learning_rate": 9.829526606962516e-06,
      "loss": 0.9155,
      "step": 1511
    },
    {
      "epoch": 0.11093482029036016,
      "grad_norm": 1.531888800530989,
      "learning_rate": 9.829218851693062e-06,
      "loss": 0.8573,
      "step": 1512
    },
    {
      "epoch": 0.11093482029036016,
      "lm_loss": 1.3766,
      "step": 1512,
      "vm_loss": 0.0717
    },
    {
      "epoch": 0.11093482029036016,
      "lm_loss": 1.045,
      "step": 1512,
      "vm_loss": 0.2838
    },
    {
      "epoch": 0.11093482029036016,
      "lm_loss": 1.0638,
      "step": 1512,
      "vm_loss": 0.1738
    },
    {
      "epoch": 0.11093482029036016,
      "lm_loss": 0.0675,
      "step": 1512,
      "vm_loss": 0.1027
    },
    {
      "epoch": 0.11093482029036016,
      "lm_loss": 2.1304,
      "step": 1512,
      "vm_loss": 0.1589
    },
    {
      "epoch": 0.11093482029036016,
      "lm_loss": 0.4189,
      "step": 1512,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.11093482029036016,
      "lm_loss": 1.4387,
      "step": 1512,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.11093482029036016,
      "lm_loss": 0.9361,
      "step": 1512,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.11100818988049928,
      "grad_norm": 1.8753464480823976,
      "learning_rate": 9.828910823705659e-06,
      "loss": 1.0345,
      "step": 1513
    },
    {
      "epoch": 0.11108155947063841,
      "grad_norm": 0.6378965230660448,
      "learning_rate": 9.828602523017703e-06,
      "loss": 0.5698,
      "step": 1514
    },
    {
      "epoch": 0.11115492906077754,
      "grad_norm": 1.5598366248854794,
      "learning_rate": 9.828293949646606e-06,
      "loss": 0.9218,
      "step": 1515
    },
    {
      "epoch": 0.11122829865091666,
      "grad_norm": 1.5616827369076085,
      "learning_rate": 9.82798510360979e-06,
      "loss": 0.8746,
      "step": 1516
    },
    {
      "epoch": 0.11130166824105579,
      "grad_norm": 1.7823665989831337,
      "learning_rate": 9.827675984924701e-06,
      "loss": 0.9805,
      "step": 1517
    },
    {
      "epoch": 0.11137503783119491,
      "grad_norm": 1.6287071497390901,
      "learning_rate": 9.827366593608794e-06,
      "loss": 0.791,
      "step": 1518
    },
    {
      "epoch": 0.11144840742133404,
      "grad_norm": 0.5810990518511732,
      "learning_rate": 9.82705692967954e-06,
      "loss": 0.5874,
      "step": 1519
    },
    {
      "epoch": 0.11152177701147317,
      "grad_norm": 1.770074426364236,
      "learning_rate": 9.826746993154426e-06,
      "loss": 0.9034,
      "step": 1520
    },
    {
      "epoch": 0.11152177701147317,
      "lm_loss": 1.8935,
      "step": 1520,
      "vm_loss": 0.1252
    },
    {
      "epoch": 0.11152177701147317,
      "lm_loss": 0.2979,
      "step": 1520,
      "vm_loss": 0.1131
    },
    {
      "epoch": 0.11152177701147317,
      "lm_loss": 0.4676,
      "step": 1520,
      "vm_loss": 0.1521
    },
    {
      "epoch": 0.11152177701147317,
      "lm_loss": 0.4013,
      "step": 1520,
      "vm_loss": 0.2
    },
    {
      "epoch": 0.11152177701147317,
      "lm_loss": 0.7854,
      "step": 1520,
      "vm_loss": 0.1647
    },
    {
      "epoch": 0.11152177701147317,
      "lm_loss": 1.0727,
      "step": 1520,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.11152177701147317,
      "lm_loss": 0.2341,
      "step": 1520,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.11152177701147317,
      "lm_loss": 0.8337,
      "step": 1520,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.1115951466016123,
      "grad_norm": 1.7232454702360747,
      "learning_rate": 9.826436784050957e-06,
      "loss": 0.885,
      "step": 1521
    },
    {
      "epoch": 0.11166851619175143,
      "grad_norm": 1.823027664719375,
      "learning_rate": 9.826126302386649e-06,
      "loss": 0.9401,
      "step": 1522
    },
    {
      "epoch": 0.11174188578189055,
      "grad_norm": 0.6164954522978003,
      "learning_rate": 9.825815548179038e-06,
      "loss": 0.5715,
      "step": 1523
    },
    {
      "epoch": 0.11181525537202967,
      "grad_norm": 1.8561788262541892,
      "learning_rate": 9.825504521445673e-06,
      "loss": 0.9942,
      "step": 1524
    },
    {
      "epoch": 0.11188862496216881,
      "grad_norm": 1.7614218156983434,
      "learning_rate": 9.825193222204116e-06,
      "loss": 0.8937,
      "step": 1525
    },
    {
      "epoch": 0.11196199455230793,
      "grad_norm": 1.68833207650294,
      "learning_rate": 9.82488165047195e-06,
      "loss": 0.9186,
      "step": 1526
    },
    {
      "epoch": 0.11203536414244705,
      "grad_norm": 1.4984021710123991,
      "learning_rate": 9.824569806266769e-06,
      "loss": 0.8455,
      "step": 1527
    },
    {
      "epoch": 0.11210873373258619,
      "grad_norm": 1.633720688575249,
      "learning_rate": 9.824257689606181e-06,
      "loss": 0.8459,
      "step": 1528
    },
    {
      "epoch": 0.11210873373258619,
      "lm_loss": 0.1255,
      "step": 1528,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.11210873373258619,
      "lm_loss": 1.208,
      "step": 1528,
      "vm_loss": 0.1276
    },
    {
      "epoch": 0.11210873373258619,
      "lm_loss": 1.3743,
      "step": 1528,
      "vm_loss": 0.1318
    },
    {
      "epoch": 0.11210873373258619,
      "lm_loss": 0.7355,
      "step": 1528,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.11210873373258619,
      "lm_loss": 0.8337,
      "step": 1528,
      "vm_loss": 0.173
    },
    {
      "epoch": 0.11210873373258619,
      "lm_loss": 1.2589,
      "step": 1528,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.11210873373258619,
      "lm_loss": 1.0518,
      "step": 1528,
      "vm_loss": 0.1598
    },
    {
      "epoch": 0.11210873373258619,
      "lm_loss": 0.076,
      "step": 1528,
      "vm_loss": 0.1135
    },
    {
      "epoch": 0.11218210332272531,
      "grad_norm": 1.6446627254556128,
      "learning_rate": 9.823945300507815e-06,
      "loss": 0.9162,
      "step": 1529
    },
    {
      "epoch": 0.11225547291286445,
      "grad_norm": 1.6367090144337635,
      "learning_rate": 9.823632638989312e-06,
      "loss": 0.9554,
      "step": 1530
    },
    {
      "epoch": 0.11232884250300357,
      "grad_norm": 1.7306774725897187,
      "learning_rate": 9.823319705068328e-06,
      "loss": 0.9325,
      "step": 1531
    },
    {
      "epoch": 0.11240221209314269,
      "grad_norm": 1.5500944105455894,
      "learning_rate": 9.823006498762536e-06,
      "loss": 0.928,
      "step": 1532
    },
    {
      "epoch": 0.11247558168328183,
      "grad_norm": 0.627318582102332,
      "learning_rate": 9.822693020089622e-06,
      "loss": 0.5526,
      "step": 1533
    },
    {
      "epoch": 0.11254895127342095,
      "grad_norm": 1.7051604419947775,
      "learning_rate": 9.822379269067292e-06,
      "loss": 0.9027,
      "step": 1534
    },
    {
      "epoch": 0.11262232086356008,
      "grad_norm": 1.947788646082013,
      "learning_rate": 9.82206524571326e-06,
      "loss": 0.8411,
      "step": 1535
    },
    {
      "epoch": 0.1126956904536992,
      "grad_norm": 1.57005313271,
      "learning_rate": 9.821750950045264e-06,
      "loss": 0.8175,
      "step": 1536
    },
    {
      "epoch": 0.1126956904536992,
      "lm_loss": 0.7205,
      "step": 1536,
      "vm_loss": 0.1404
    },
    {
      "epoch": 0.1126956904536992,
      "lm_loss": 0.7439,
      "step": 1536,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.1126956904536992,
      "lm_loss": 0.6112,
      "step": 1536,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.1126956904536992,
      "lm_loss": 1.8735,
      "step": 1536,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.1126956904536992,
      "lm_loss": 0.6478,
      "step": 1536,
      "vm_loss": 0.0884
    },
    {
      "epoch": 0.1126956904536992,
      "lm_loss": 0.6223,
      "step": 1536,
      "vm_loss": 0.1015
    },
    {
      "epoch": 0.1126956904536992,
      "lm_loss": 0.9137,
      "step": 1536,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.1126956904536992,
      "lm_loss": 0.0984,
      "step": 1536,
      "vm_loss": 0.1431
    },
    {
      "epoch": 0.11276906004383833,
      "grad_norm": 2.2938768587747735,
      "learning_rate": 9.821436382081049e-06,
      "loss": 0.8959,
      "step": 1537
    },
    {
      "epoch": 0.11284242963397746,
      "grad_norm": 1.7030593198782584,
      "learning_rate": 9.821121541838381e-06,
      "loss": 0.8804,
      "step": 1538
    },
    {
      "epoch": 0.11291579922411658,
      "grad_norm": 1.9384914923052194,
      "learning_rate": 9.820806429335042e-06,
      "loss": 0.9679,
      "step": 1539
    },
    {
      "epoch": 0.1129891688142557,
      "grad_norm": 1.933837606553221,
      "learning_rate": 9.820491044588825e-06,
      "loss": 1.0075,
      "step": 1540
    },
    {
      "epoch": 0.11306253840439484,
      "grad_norm": 1.5447587516961518,
      "learning_rate": 9.820175387617541e-06,
      "loss": 0.8884,
      "step": 1541
    },
    {
      "epoch": 0.11313590799453396,
      "grad_norm": 2.055327885966402,
      "learning_rate": 9.819859458439015e-06,
      "loss": 1.0241,
      "step": 1542
    },
    {
      "epoch": 0.1132092775846731,
      "grad_norm": 0.6001513172685823,
      "learning_rate": 9.819543257071092e-06,
      "loss": 0.5565,
      "step": 1543
    },
    {
      "epoch": 0.11328264717481222,
      "grad_norm": 1.6332254683475522,
      "learning_rate": 9.819226783531623e-06,
      "loss": 0.8939,
      "step": 1544
    },
    {
      "epoch": 0.11328264717481222,
      "lm_loss": 0.1555,
      "step": 1544,
      "vm_loss": 0.0471
    },
    {
      "epoch": 0.11328264717481222,
      "lm_loss": 4.4631,
      "step": 1544,
      "vm_loss": 0.052
    },
    {
      "epoch": 0.11328264717481222,
      "lm_loss": 0.4594,
      "step": 1544,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.11328264717481222,
      "lm_loss": 0.8477,
      "step": 1544,
      "vm_loss": 0.1404
    },
    {
      "epoch": 0.11328264717481222,
      "lm_loss": 0.8453,
      "step": 1544,
      "vm_loss": 0.0988
    },
    {
      "epoch": 0.11328264717481222,
      "lm_loss": 1.0196,
      "step": 1544,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.11328264717481222,
      "lm_loss": 0.2858,
      "step": 1544,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.11328264717481222,
      "lm_loss": 1.1989,
      "step": 1544,
      "vm_loss": 0.1973
    },
    {
      "epoch": 0.11335601676495134,
      "grad_norm": 1.5295789044405674,
      "learning_rate": 9.818910037838482e-06,
      "loss": 1.019,
      "step": 1545
    },
    {
      "epoch": 0.11342938635509048,
      "grad_norm": 1.8612331126437185,
      "learning_rate": 9.81859302000956e-06,
      "loss": 0.8759,
      "step": 1546
    },
    {
      "epoch": 0.1135027559452296,
      "grad_norm": 1.9650968991162188,
      "learning_rate": 9.818275730062756e-06,
      "loss": 0.9029,
      "step": 1547
    },
    {
      "epoch": 0.11357612553536872,
      "grad_norm": 1.7052131371773491,
      "learning_rate": 9.81795816801599e-06,
      "loss": 0.8489,
      "step": 1548
    },
    {
      "epoch": 0.11364949512550786,
      "grad_norm": 1.6647367243885123,
      "learning_rate": 9.817640333887194e-06,
      "loss": 0.8183,
      "step": 1549
    },
    {
      "epoch": 0.11372286471564698,
      "grad_norm": 0.6196260557156067,
      "learning_rate": 9.817322227694318e-06,
      "loss": 0.5732,
      "step": 1550
    },
    {
      "epoch": 0.11379623430578611,
      "grad_norm": 0.6110708242270934,
      "learning_rate": 9.817003849455325e-06,
      "loss": 0.5717,
      "step": 1551
    },
    {
      "epoch": 0.11386960389592524,
      "grad_norm": 1.8512896134093835,
      "learning_rate": 9.816685199188196e-06,
      "loss": 0.9834,
      "step": 1552
    },
    {
      "epoch": 0.11386960389592524,
      "lm_loss": 1.6132,
      "step": 1552,
      "vm_loss": 0.0352
    },
    {
      "epoch": 0.11386960389592524,
      "lm_loss": 0.8145,
      "step": 1552,
      "vm_loss": 0.1069
    },
    {
      "epoch": 0.11386960389592524,
      "lm_loss": 1.4148,
      "step": 1552,
      "vm_loss": 0.1699
    },
    {
      "epoch": 0.11386960389592524,
      "lm_loss": 0.5472,
      "step": 1552,
      "vm_loss": 0.0609
    },
    {
      "epoch": 0.11386960389592524,
      "lm_loss": 0.919,
      "step": 1552,
      "vm_loss": 0.1781
    },
    {
      "epoch": 0.11386960389592524,
      "lm_loss": 0.6163,
      "step": 1552,
      "vm_loss": 0.2126
    },
    {
      "epoch": 0.11386960389592524,
      "lm_loss": 0.3991,
      "step": 1552,
      "vm_loss": 0.1333
    },
    {
      "epoch": 0.11386960389592524,
      "lm_loss": 0.8805,
      "step": 1552,
      "vm_loss": 0.0806
    },
    {
      "epoch": 0.11394297348606436,
      "grad_norm": 2.01217500256061,
      "learning_rate": 9.816366276910927e-06,
      "loss": 0.9171,
      "step": 1553
    },
    {
      "epoch": 0.1140163430762035,
      "grad_norm": 0.5913180226500299,
      "learning_rate": 9.816047082641526e-06,
      "loss": 0.5682,
      "step": 1554
    },
    {
      "epoch": 0.11408971266634262,
      "grad_norm": 1.6874992516194496,
      "learning_rate": 9.81572761639802e-06,
      "loss": 0.8474,
      "step": 1555
    },
    {
      "epoch": 0.11416308225648175,
      "grad_norm": 1.7234216656166288,
      "learning_rate": 9.815407878198445e-06,
      "loss": 0.8645,
      "step": 1556
    },
    {
      "epoch": 0.11423645184662087,
      "grad_norm": 1.7771425211145397,
      "learning_rate": 9.815087868060867e-06,
      "loss": 0.8631,
      "step": 1557
    },
    {
      "epoch": 0.11430982143676,
      "grad_norm": 0.6391250512941272,
      "learning_rate": 9.814767586003352e-06,
      "loss": 0.5799,
      "step": 1558
    },
    {
      "epoch": 0.11438319102689913,
      "grad_norm": 1.8080424918936127,
      "learning_rate": 9.814447032043987e-06,
      "loss": 0.8888,
      "step": 1559
    },
    {
      "epoch": 0.11445656061703825,
      "grad_norm": 1.8129655548156376,
      "learning_rate": 9.814126206200877e-06,
      "loss": 0.8384,
      "step": 1560
    },
    {
      "epoch": 0.11445656061703825,
      "lm_loss": 1.7363,
      "step": 1560,
      "vm_loss": 0.1177
    },
    {
      "epoch": 0.11445656061703825,
      "lm_loss": 0.4373,
      "step": 1560,
      "vm_loss": 0.1463
    },
    {
      "epoch": 0.11445656061703825,
      "lm_loss": 1.0085,
      "step": 1560,
      "vm_loss": 0.158
    },
    {
      "epoch": 0.11445656061703825,
      "lm_loss": 1.145,
      "step": 1560,
      "vm_loss": 0.178
    },
    {
      "epoch": 0.11445656061703825,
      "lm_loss": 0.5341,
      "step": 1560,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.11445656061703825,
      "lm_loss": 1.1095,
      "step": 1560,
      "vm_loss": 0.153
    },
    {
      "epoch": 0.11445656061703825,
      "lm_loss": 0.6295,
      "step": 1560,
      "vm_loss": 0.1416
    },
    {
      "epoch": 0.11445656061703825,
      "lm_loss": 0.5539,
      "step": 1560,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.11452993020717737,
      "grad_norm": 1.7461876951760347,
      "learning_rate": 9.813805108492137e-06,
      "loss": 0.934,
      "step": 1561
    },
    {
      "epoch": 0.11460329979731651,
      "grad_norm": 1.6954919065631007,
      "learning_rate": 9.813483738935899e-06,
      "loss": 0.9587,
      "step": 1562
    },
    {
      "epoch": 0.11467666938745563,
      "grad_norm": 0.625564900425852,
      "learning_rate": 9.813162097550317e-06,
      "loss": 0.5422,
      "step": 1563
    },
    {
      "epoch": 0.11475003897759477,
      "grad_norm": 1.792961218710056,
      "learning_rate": 9.81284018435355e-06,
      "loss": 0.8641,
      "step": 1564
    },
    {
      "epoch": 0.11482340856773389,
      "grad_norm": 1.517001343135339,
      "learning_rate": 9.81251799936378e-06,
      "loss": 0.9182,
      "step": 1565
    },
    {
      "epoch": 0.11489677815787301,
      "grad_norm": 1.8453332132263531,
      "learning_rate": 9.8121955425992e-06,
      "loss": 0.9719,
      "step": 1566
    },
    {
      "epoch": 0.11497014774801215,
      "grad_norm": 1.5696471129746357,
      "learning_rate": 9.811872814078021e-06,
      "loss": 0.9522,
      "step": 1567
    },
    {
      "epoch": 0.11504351733815127,
      "grad_norm": 1.596922146216201,
      "learning_rate": 9.811549813818466e-06,
      "loss": 0.8558,
      "step": 1568
    },
    {
      "epoch": 0.11504351733815127,
      "lm_loss": 0.4937,
      "step": 1568,
      "vm_loss": 0.0468
    },
    {
      "epoch": 0.11504351733815127,
      "lm_loss": 0.6918,
      "step": 1568,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.11504351733815127,
      "lm_loss": 0.8111,
      "step": 1568,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.11504351733815127,
      "lm_loss": 1.0299,
      "step": 1568,
      "vm_loss": 0.2251
    },
    {
      "epoch": 0.11504351733815127,
      "lm_loss": 0.6139,
      "step": 1568,
      "vm_loss": 0.164
    },
    {
      "epoch": 0.11504351733815127,
      "lm_loss": 0.6932,
      "step": 1568,
      "vm_loss": 0.0677
    },
    {
      "epoch": 0.11504351733815127,
      "lm_loss": 0.7063,
      "step": 1568,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.11504351733815127,
      "lm_loss": 1.4344,
      "step": 1568,
      "vm_loss": 0.1234
    },
    {
      "epoch": 0.11511688692829039,
      "grad_norm": 1.6557849186045015,
      "learning_rate": 9.81122654183878e-06,
      "loss": 0.8671,
      "step": 1569
    },
    {
      "epoch": 0.11519025651842953,
      "grad_norm": 1.9456748239140487,
      "learning_rate": 9.810902998157215e-06,
      "loss": 0.8139,
      "step": 1570
    },
    {
      "epoch": 0.11526362610856865,
      "grad_norm": 1.4402533896846292,
      "learning_rate": 9.810579182792044e-06,
      "loss": 0.8897,
      "step": 1571
    },
    {
      "epoch": 0.11533699569870778,
      "grad_norm": 1.6575674653711752,
      "learning_rate": 9.810255095761554e-06,
      "loss": 0.8049,
      "step": 1572
    },
    {
      "epoch": 0.1154103652888469,
      "grad_norm": 1.6381220323743686,
      "learning_rate": 9.809930737084045e-06,
      "loss": 0.9097,
      "step": 1573
    },
    {
      "epoch": 0.11548373487898603,
      "grad_norm": 1.641845523911982,
      "learning_rate": 9.809606106777835e-06,
      "loss": 0.8565,
      "step": 1574
    },
    {
      "epoch": 0.11555710446912516,
      "grad_norm": 1.6007386411308229,
      "learning_rate": 9.80928120486126e-06,
      "loss": 0.8134,
      "step": 1575
    },
    {
      "epoch": 0.11563047405926428,
      "grad_norm": 1.6167432468943963,
      "learning_rate": 9.808956031352664e-06,
      "loss": 0.9321,
      "step": 1576
    },
    {
      "epoch": 0.11563047405926428,
      "lm_loss": 1.3729,
      "step": 1576,
      "vm_loss": 0.0619
    },
    {
      "epoch": 0.11563047405926428,
      "lm_loss": 1.2344,
      "step": 1576,
      "vm_loss": 0.1072
    },
    {
      "epoch": 0.11563047405926428,
      "lm_loss": 0.9999,
      "step": 1576,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.11563047405926428,
      "lm_loss": 0.8604,
      "step": 1576,
      "vm_loss": 0.105
    },
    {
      "epoch": 0.11563047405926428,
      "lm_loss": 1.211,
      "step": 1576,
      "vm_loss": 0.1143
    },
    {
      "epoch": 0.11563047405926428,
      "lm_loss": 1.1677,
      "step": 1576,
      "vm_loss": 0.1772
    },
    {
      "epoch": 0.11563047405926428,
      "lm_loss": 0.8918,
      "step": 1576,
      "vm_loss": 0.1412
    },
    {
      "epoch": 0.11563047405926428,
      "lm_loss": 0.6794,
      "step": 1576,
      "vm_loss": 0.1033
    },
    {
      "epoch": 0.1157038436494034,
      "grad_norm": 1.8860807904319985,
      "learning_rate": 9.80863058627041e-06,
      "loss": 1.0538,
      "step": 1577
    },
    {
      "epoch": 0.11577721323954254,
      "grad_norm": 1.6769806720339717,
      "learning_rate": 9.808304869632882e-06,
      "loss": 0.8802,
      "step": 1578
    },
    {
      "epoch": 0.11585058282968166,
      "grad_norm": 1.772082078936176,
      "learning_rate": 9.807978881458468e-06,
      "loss": 0.9272,
      "step": 1579
    },
    {
      "epoch": 0.1159239524198208,
      "grad_norm": 1.583974806961792,
      "learning_rate": 9.807652621765582e-06,
      "loss": 0.9019,
      "step": 1580
    },
    {
      "epoch": 0.11599732200995992,
      "grad_norm": 1.9307482848641335,
      "learning_rate": 9.807326090572645e-06,
      "loss": 0.8689,
      "step": 1581
    },
    {
      "epoch": 0.11607069160009904,
      "grad_norm": 1.5746939497503802,
      "learning_rate": 9.806999287898098e-06,
      "loss": 0.9255,
      "step": 1582
    },
    {
      "epoch": 0.11614406119023818,
      "grad_norm": 0.6895305166309826,
      "learning_rate": 9.806672213760398e-06,
      "loss": 0.5702,
      "step": 1583
    },
    {
      "epoch": 0.1162174307803773,
      "grad_norm": 1.5150155603083784,
      "learning_rate": 9.806344868178013e-06,
      "loss": 0.825,
      "step": 1584
    },
    {
      "epoch": 0.1162174307803773,
      "lm_loss": 0.7862,
      "step": 1584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1162174307803773,
      "lm_loss": 0.4985,
      "step": 1584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1162174307803773,
      "lm_loss": 0.5992,
      "step": 1584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1162174307803773,
      "lm_loss": 0.6545,
      "step": 1584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1162174307803773,
      "lm_loss": 0.3753,
      "step": 1584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1162174307803773,
      "lm_loss": 0.6624,
      "step": 1584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1162174307803773,
      "lm_loss": 0.7116,
      "step": 1584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1162174307803773,
      "lm_loss": 0.4729,
      "step": 1584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11629080037051644,
      "grad_norm": 0.6326192562551638,
      "learning_rate": 9.80601725116943e-06,
      "loss": 0.5573,
      "step": 1585
    },
    {
      "epoch": 0.11636416996065556,
      "grad_norm": 0.5856914492444462,
      "learning_rate": 9.805689362753154e-06,
      "loss": 0.5664,
      "step": 1586
    },
    {
      "epoch": 0.11643753955079468,
      "grad_norm": 1.6770990372114025,
      "learning_rate": 9.805361202947696e-06,
      "loss": 0.9375,
      "step": 1587
    },
    {
      "epoch": 0.11651090914093382,
      "grad_norm": 0.6031621220358179,
      "learning_rate": 9.805032771771592e-06,
      "loss": 0.5526,
      "step": 1588
    },
    {
      "epoch": 0.11658427873107294,
      "grad_norm": 0.6299524614751034,
      "learning_rate": 9.804704069243389e-06,
      "loss": 0.5791,
      "step": 1589
    },
    {
      "epoch": 0.11665764832121206,
      "grad_norm": 1.6061544624010484,
      "learning_rate": 9.804375095381646e-06,
      "loss": 1.0292,
      "step": 1590
    },
    {
      "epoch": 0.1167310179113512,
      "grad_norm": 1.6744074254688168,
      "learning_rate": 9.804045850204944e-06,
      "loss": 0.942,
      "step": 1591
    },
    {
      "epoch": 0.11680438750149032,
      "grad_norm": 1.749879618272436,
      "learning_rate": 9.803716333731877e-06,
      "loss": 0.9159,
      "step": 1592
    },
    {
      "epoch": 0.11680438750149032,
      "lm_loss": 0.5781,
      "step": 1592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11680438750149032,
      "lm_loss": 0.497,
      "step": 1592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11680438750149032,
      "lm_loss": 0.4593,
      "step": 1592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11680438750149032,
      "lm_loss": 0.6644,
      "step": 1592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11680438750149032,
      "lm_loss": 0.5261,
      "step": 1592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11680438750149032,
      "lm_loss": 0.5464,
      "step": 1592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11680438750149032,
      "lm_loss": 0.4691,
      "step": 1592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11680438750149032,
      "lm_loss": 0.6005,
      "step": 1592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11687775709162945,
      "grad_norm": 0.6333200660115301,
      "learning_rate": 9.803386545981053e-06,
      "loss": 0.5721,
      "step": 1593
    },
    {
      "epoch": 0.11695112668176857,
      "grad_norm": 1.649693545655846,
      "learning_rate": 9.803056486971094e-06,
      "loss": 0.8782,
      "step": 1594
    },
    {
      "epoch": 0.1170244962719077,
      "grad_norm": 1.8660886266085719,
      "learning_rate": 9.802726156720641e-06,
      "loss": 0.7265,
      "step": 1595
    },
    {
      "epoch": 0.11709786586204683,
      "grad_norm": 0.6015323487269691,
      "learning_rate": 9.802395555248348e-06,
      "loss": 0.5538,
      "step": 1596
    },
    {
      "epoch": 0.11717123545218595,
      "grad_norm": 1.6556418021808235,
      "learning_rate": 9.802064682572885e-06,
      "loss": 0.8426,
      "step": 1597
    },
    {
      "epoch": 0.11724460504232508,
      "grad_norm": 0.6248805247636476,
      "learning_rate": 9.801733538712936e-06,
      "loss": 0.5882,
      "step": 1598
    },
    {
      "epoch": 0.11731797463246421,
      "grad_norm": 1.8033953972725334,
      "learning_rate": 9.801402123687205e-06,
      "loss": 0.9244,
      "step": 1599
    },
    {
      "epoch": 0.11739134422260333,
      "grad_norm": 0.5961131315561417,
      "learning_rate": 9.801070437514404e-06,
      "loss": 0.5823,
      "step": 1600
    },
    {
      "epoch": 0.11739134422260333,
      "lm_loss": 0.7581,
      "step": 1600,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.11739134422260333,
      "lm_loss": 1.1149,
      "step": 1600,
      "vm_loss": 0.0834
    },
    {
      "epoch": 0.11739134422260333,
      "lm_loss": 1.0518,
      "step": 1600,
      "vm_loss": 0.1849
    },
    {
      "epoch": 0.11739134422260333,
      "lm_loss": 0.7669,
      "step": 1600,
      "vm_loss": 0.1658
    },
    {
      "epoch": 0.11739134422260333,
      "lm_loss": 0.6432,
      "step": 1600,
      "vm_loss": 0.1435
    },
    {
      "epoch": 0.11739134422260333,
      "lm_loss": 0.3306,
      "step": 1600,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.11739134422260333,
      "lm_loss": 0.1399,
      "step": 1600,
      "vm_loss": 0.1219
    },
    {
      "epoch": 0.11739134422260333,
      "lm_loss": 0.6731,
      "step": 1600,
      "vm_loss": 0.0825
    },
    {
      "epoch": 0.11746471381274247,
      "grad_norm": 1.737409188212783,
      "learning_rate": 9.800738480213268e-06,
      "loss": 0.8403,
      "step": 1601
    },
    {
      "epoch": 0.11753808340288159,
      "grad_norm": 1.6693276615026276,
      "learning_rate": 9.80040625180254e-06,
      "loss": 0.8714,
      "step": 1602
    },
    {
      "epoch": 0.11761145299302071,
      "grad_norm": 1.6480637200063073,
      "learning_rate": 9.800073752300983e-06,
      "loss": 0.8881,
      "step": 1603
    },
    {
      "epoch": 0.11768482258315985,
      "grad_norm": 0.6264687645865935,
      "learning_rate": 9.799740981727372e-06,
      "loss": 0.5563,
      "step": 1604
    },
    {
      "epoch": 0.11775819217329897,
      "grad_norm": 1.587790475864367,
      "learning_rate": 9.799407940100503e-06,
      "loss": 0.8398,
      "step": 1605
    },
    {
      "epoch": 0.1178315617634381,
      "grad_norm": 1.573369695058842,
      "learning_rate": 9.799074627439183e-06,
      "loss": 1.0054,
      "step": 1606
    },
    {
      "epoch": 0.11790493135357723,
      "grad_norm": 0.5911368611117717,
      "learning_rate": 9.798741043762234e-06,
      "loss": 0.5715,
      "step": 1607
    },
    {
      "epoch": 0.11797830094371635,
      "grad_norm": 1.732790278846043,
      "learning_rate": 9.798407189088493e-06,
      "loss": 0.875,
      "step": 1608
    },
    {
      "epoch": 0.11797830094371635,
      "lm_loss": 1.7685,
      "step": 1608,
      "vm_loss": 0.0784
    },
    {
      "epoch": 0.11797830094371635,
      "lm_loss": 0.1004,
      "step": 1608,
      "vm_loss": 0.0512
    },
    {
      "epoch": 0.11797830094371635,
      "lm_loss": 0.8054,
      "step": 1608,
      "vm_loss": 0.2689
    },
    {
      "epoch": 0.11797830094371635,
      "lm_loss": 0.7686,
      "step": 1608,
      "vm_loss": 0.0776
    },
    {
      "epoch": 0.11797830094371635,
      "lm_loss": 0.632,
      "step": 1608,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.11797830094371635,
      "lm_loss": 0.6644,
      "step": 1608,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.11797830094371635,
      "lm_loss": 1.0303,
      "step": 1608,
      "vm_loss": 0.2673
    },
    {
      "epoch": 0.11797830094371635,
      "lm_loss": 0.087,
      "step": 1608,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.11805167053385549,
      "grad_norm": 1.6624936825573464,
      "learning_rate": 9.798073063436815e-06,
      "loss": 0.8462,
      "step": 1609
    },
    {
      "epoch": 0.1181250401239946,
      "grad_norm": 1.6027323423758124,
      "learning_rate": 9.79773866682607e-06,
      "loss": 0.9216,
      "step": 1610
    },
    {
      "epoch": 0.11819840971413373,
      "grad_norm": 1.6267132786396088,
      "learning_rate": 9.79740399927514e-06,
      "loss": 0.8583,
      "step": 1611
    },
    {
      "epoch": 0.11827177930427286,
      "grad_norm": 1.7142168393179555,
      "learning_rate": 9.797069060802927e-06,
      "loss": 0.9189,
      "step": 1612
    },
    {
      "epoch": 0.11834514889441199,
      "grad_norm": 1.687718110349504,
      "learning_rate": 9.796733851428344e-06,
      "loss": 0.8163,
      "step": 1613
    },
    {
      "epoch": 0.11841851848455112,
      "grad_norm": 0.6743365673937296,
      "learning_rate": 9.79639837117032e-06,
      "loss": 0.5681,
      "step": 1614
    },
    {
      "epoch": 0.11849188807469024,
      "grad_norm": 1.8488935579078918,
      "learning_rate": 9.796062620047804e-06,
      "loss": 1.0495,
      "step": 1615
    },
    {
      "epoch": 0.11856525766482937,
      "grad_norm": 0.6333252216127492,
      "learning_rate": 9.795726598079753e-06,
      "loss": 0.5875,
      "step": 1616
    },
    {
      "epoch": 0.11856525766482937,
      "lm_loss": 0.6359,
      "step": 1616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11856525766482937,
      "lm_loss": 0.5405,
      "step": 1616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11856525766482937,
      "lm_loss": 0.5034,
      "step": 1616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11856525766482937,
      "lm_loss": 0.5548,
      "step": 1616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11856525766482937,
      "lm_loss": 0.5904,
      "step": 1616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11856525766482937,
      "lm_loss": 0.5299,
      "step": 1616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11856525766482937,
      "lm_loss": 0.4019,
      "step": 1616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11856525766482937,
      "lm_loss": 0.4643,
      "step": 1616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1186386272549685,
      "grad_norm": 0.5867478564005751,
      "learning_rate": 9.795390305285144e-06,
      "loss": 0.5664,
      "step": 1617
    },
    {
      "epoch": 0.11871199684510762,
      "grad_norm": 0.5794046844359997,
      "learning_rate": 9.79505374168297e-06,
      "loss": 0.5604,
      "step": 1618
    },
    {
      "epoch": 0.11878536643524674,
      "grad_norm": 1.8907945269404771,
      "learning_rate": 9.794716907292237e-06,
      "loss": 0.8738,
      "step": 1619
    },
    {
      "epoch": 0.11885873602538588,
      "grad_norm": 1.8441762584176757,
      "learning_rate": 9.794379802131966e-06,
      "loss": 0.9921,
      "step": 1620
    },
    {
      "epoch": 0.118932105615525,
      "grad_norm": 0.6089430411420252,
      "learning_rate": 9.794042426221193e-06,
      "loss": 0.5562,
      "step": 1621
    },
    {
      "epoch": 0.11900547520566414,
      "grad_norm": 0.6258679940915681,
      "learning_rate": 9.793704779578974e-06,
      "loss": 0.5758,
      "step": 1622
    },
    {
      "epoch": 0.11907884479580326,
      "grad_norm": 2.076083671921554,
      "learning_rate": 9.793366862224374e-06,
      "loss": 0.9487,
      "step": 1623
    },
    {
      "epoch": 0.11915221438594238,
      "grad_norm": 1.635730929900204,
      "learning_rate": 9.793028674176476e-06,
      "loss": 0.8695,
      "step": 1624
    },
    {
      "epoch": 0.11915221438594238,
      "lm_loss": 1.1393,
      "step": 1624,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.11915221438594238,
      "lm_loss": 1.6705,
      "step": 1624,
      "vm_loss": 0.0946
    },
    {
      "epoch": 0.11915221438594238,
      "lm_loss": 0.2638,
      "step": 1624,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.11915221438594238,
      "lm_loss": 0.1452,
      "step": 1624,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.11915221438594238,
      "lm_loss": 0.1431,
      "step": 1624,
      "vm_loss": 0.1307
    },
    {
      "epoch": 0.11915221438594238,
      "lm_loss": 1.1218,
      "step": 1624,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.11915221438594238,
      "lm_loss": 0.5643,
      "step": 1624,
      "vm_loss": 0.129
    },
    {
      "epoch": 0.11915221438594238,
      "lm_loss": 1.2634,
      "step": 1624,
      "vm_loss": 0.2157
    },
    {
      "epoch": 0.11922558397608152,
      "grad_norm": 1.6713185676654763,
      "learning_rate": 9.792690215454382e-06,
      "loss": 0.9817,
      "step": 1625
    },
    {
      "epoch": 0.11929895356622064,
      "grad_norm": 0.6076327056192982,
      "learning_rate": 9.792351486077198e-06,
      "loss": 0.5772,
      "step": 1626
    },
    {
      "epoch": 0.11937232315635976,
      "grad_norm": 1.9268587848454588,
      "learning_rate": 9.792012486064061e-06,
      "loss": 0.9831,
      "step": 1627
    },
    {
      "epoch": 0.1194456927464989,
      "grad_norm": 0.630654765948731,
      "learning_rate": 9.79167321543411e-06,
      "loss": 0.5598,
      "step": 1628
    },
    {
      "epoch": 0.11951906233663802,
      "grad_norm": 0.6141957982485899,
      "learning_rate": 9.791333674206507e-06,
      "loss": 0.5904,
      "step": 1629
    },
    {
      "epoch": 0.11959243192677715,
      "grad_norm": 1.6645271860375006,
      "learning_rate": 9.790993862400426e-06,
      "loss": 0.797,
      "step": 1630
    },
    {
      "epoch": 0.11966580151691628,
      "grad_norm": 1.8326203120942028,
      "learning_rate": 9.790653780035056e-06,
      "loss": 0.9208,
      "step": 1631
    },
    {
      "epoch": 0.1197391711070554,
      "grad_norm": 0.7661551153831075,
      "learning_rate": 9.790313427129603e-06,
      "loss": 0.5593,
      "step": 1632
    },
    {
      "epoch": 0.1197391711070554,
      "lm_loss": 0.8267,
      "step": 1632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1197391711070554,
      "lm_loss": 0.5904,
      "step": 1632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1197391711070554,
      "lm_loss": 0.5716,
      "step": 1632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1197391711070554,
      "lm_loss": 0.5107,
      "step": 1632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1197391711070554,
      "lm_loss": 0.6563,
      "step": 1632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1197391711070554,
      "lm_loss": 0.4751,
      "step": 1632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1197391711070554,
      "lm_loss": 0.5964,
      "step": 1632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1197391711070554,
      "lm_loss": 0.4967,
      "step": 1632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.11981254069719453,
      "grad_norm": 0.7107102518068635,
      "learning_rate": 9.78997280370329e-06,
      "loss": 0.551,
      "step": 1633
    },
    {
      "epoch": 0.11988591028733366,
      "grad_norm": 1.8254657005662867,
      "learning_rate": 9.789631909775348e-06,
      "loss": 0.8312,
      "step": 1634
    },
    {
      "epoch": 0.11995927987747279,
      "grad_norm": 1.7659933593157773,
      "learning_rate": 9.789290745365032e-06,
      "loss": 0.8345,
      "step": 1635
    },
    {
      "epoch": 0.12003264946761191,
      "grad_norm": 0.5999418676597955,
      "learning_rate": 9.788949310491607e-06,
      "loss": 0.5605,
      "step": 1636
    },
    {
      "epoch": 0.12010601905775103,
      "grad_norm": 1.5710407118879097,
      "learning_rate": 9.788607605174353e-06,
      "loss": 0.9053,
      "step": 1637
    },
    {
      "epoch": 0.12017938864789017,
      "grad_norm": 1.6702635832980748,
      "learning_rate": 9.788265629432572e-06,
      "loss": 0.8797,
      "step": 1638
    },
    {
      "epoch": 0.12025275823802929,
      "grad_norm": 1.717981820602497,
      "learning_rate": 9.787923383285571e-06,
      "loss": 0.8472,
      "step": 1639
    },
    {
      "epoch": 0.12032612782816841,
      "grad_norm": 0.7585392271561795,
      "learning_rate": 9.78758086675268e-06,
      "loss": 0.557,
      "step": 1640
    },
    {
      "epoch": 0.12032612782816841,
      "lm_loss": 1.4062,
      "step": 1640,
      "vm_loss": 0.0692
    },
    {
      "epoch": 0.12032612782816841,
      "lm_loss": 0.813,
      "step": 1640,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.12032612782816841,
      "lm_loss": 0.7579,
      "step": 1640,
      "vm_loss": 0.0772
    },
    {
      "epoch": 0.12032612782816841,
      "lm_loss": 0.3117,
      "step": 1640,
      "vm_loss": 0.1483
    },
    {
      "epoch": 0.12032612782816841,
      "lm_loss": 1.0159,
      "step": 1640,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.12032612782816841,
      "lm_loss": 0.365,
      "step": 1640,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.12032612782816841,
      "lm_loss": 0.7378,
      "step": 1640,
      "vm_loss": 0.1939
    },
    {
      "epoch": 0.12032612782816841,
      "lm_loss": 0.2848,
      "step": 1640,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.12039949741830755,
      "grad_norm": 1.9209913501959328,
      "learning_rate": 9.78723807985324e-06,
      "loss": 0.8957,
      "step": 1641
    },
    {
      "epoch": 0.12047286700844667,
      "grad_norm": 0.6375618068802232,
      "learning_rate": 9.786895022606612e-06,
      "loss": 0.5694,
      "step": 1642
    },
    {
      "epoch": 0.12054623659858581,
      "grad_norm": 1.623512945774169,
      "learning_rate": 9.786551695032166e-06,
      "loss": 0.8188,
      "step": 1643
    },
    {
      "epoch": 0.12061960618872493,
      "grad_norm": 1.8540383080462688,
      "learning_rate": 9.786208097149295e-06,
      "loss": 0.9358,
      "step": 1644
    },
    {
      "epoch": 0.12069297577886405,
      "grad_norm": 1.7224097853470326,
      "learning_rate": 9.785864228977397e-06,
      "loss": 0.9598,
      "step": 1645
    },
    {
      "epoch": 0.12076634536900319,
      "grad_norm": 1.9276128380648587,
      "learning_rate": 9.785520090535895e-06,
      "loss": 0.8969,
      "step": 1646
    },
    {
      "epoch": 0.12083971495914231,
      "grad_norm": 1.8121850289014185,
      "learning_rate": 9.785175681844222e-06,
      "loss": 1.0101,
      "step": 1647
    },
    {
      "epoch": 0.12091308454928143,
      "grad_norm": 1.6001734310176638,
      "learning_rate": 9.784831002921828e-06,
      "loss": 0.8221,
      "step": 1648
    },
    {
      "epoch": 0.12091308454928143,
      "lm_loss": 0.8435,
      "step": 1648,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.12091308454928143,
      "lm_loss": 2.5496,
      "step": 1648,
      "vm_loss": 0.0699
    },
    {
      "epoch": 0.12091308454928143,
      "lm_loss": 0.026,
      "step": 1648,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.12091308454928143,
      "lm_loss": 1.0936,
      "step": 1648,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.12091308454928143,
      "lm_loss": 1.0361,
      "step": 1648,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.12091308454928143,
      "lm_loss": 1.2678,
      "step": 1648,
      "vm_loss": 0.2204
    },
    {
      "epoch": 0.12091308454928143,
      "lm_loss": 0.0659,
      "step": 1648,
      "vm_loss": 0.1161
    },
    {
      "epoch": 0.12091308454928143,
      "lm_loss": 1.0457,
      "step": 1648,
      "vm_loss": 0.161
    },
    {
      "epoch": 0.12098645413942057,
      "grad_norm": 1.6790012806624321,
      "learning_rate": 9.784486053788179e-06,
      "loss": 0.894,
      "step": 1649
    },
    {
      "epoch": 0.12105982372955969,
      "grad_norm": 0.7985142754663297,
      "learning_rate": 9.784140834462752e-06,
      "loss": 0.5885,
      "step": 1650
    },
    {
      "epoch": 0.12113319331969882,
      "grad_norm": 1.6596058907135294,
      "learning_rate": 9.783795344965046e-06,
      "loss": 0.874,
      "step": 1651
    },
    {
      "epoch": 0.12120656290983794,
      "grad_norm": 0.6737457634441546,
      "learning_rate": 9.783449585314569e-06,
      "loss": 0.5709,
      "step": 1652
    },
    {
      "epoch": 0.12127993249997707,
      "grad_norm": 1.7277528762475118,
      "learning_rate": 9.783103555530849e-06,
      "loss": 0.9473,
      "step": 1653
    },
    {
      "epoch": 0.1213533020901162,
      "grad_norm": 1.6829111609452103,
      "learning_rate": 9.782757255633423e-06,
      "loss": 0.8657,
      "step": 1654
    },
    {
      "epoch": 0.12142667168025532,
      "grad_norm": 0.6578296723156034,
      "learning_rate": 9.782410685641855e-06,
      "loss": 0.5718,
      "step": 1655
    },
    {
      "epoch": 0.12150004127039446,
      "grad_norm": 0.6291935119862381,
      "learning_rate": 9.78206384557571e-06,
      "loss": 0.5699,
      "step": 1656
    },
    {
      "epoch": 0.12150004127039446,
      "lm_loss": 1.6973,
      "step": 1656,
      "vm_loss": 0.0652
    },
    {
      "epoch": 0.12150004127039446,
      "lm_loss": 0.2574,
      "step": 1656,
      "vm_loss": 0.106
    },
    {
      "epoch": 0.12150004127039446,
      "lm_loss": 1.2441,
      "step": 1656,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.12150004127039446,
      "lm_loss": 1.0724,
      "step": 1656,
      "vm_loss": 0.2009
    },
    {
      "epoch": 0.12150004127039446,
      "lm_loss": 0.0655,
      "step": 1656,
      "vm_loss": 0.1443
    },
    {
      "epoch": 0.12150004127039446,
      "lm_loss": 0.2872,
      "step": 1656,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.12150004127039446,
      "lm_loss": 1.1324,
      "step": 1656,
      "vm_loss": 0.2139
    },
    {
      "epoch": 0.12150004127039446,
      "lm_loss": 0.2268,
      "step": 1656,
      "vm_loss": 0.0525
    },
    {
      "epoch": 0.12157341086053358,
      "grad_norm": 1.8242550942267697,
      "learning_rate": 9.781716735454576e-06,
      "loss": 0.8994,
      "step": 1657
    },
    {
      "epoch": 0.1216467804506727,
      "grad_norm": 0.6351404082571144,
      "learning_rate": 9.781369355298056e-06,
      "loss": 0.5696,
      "step": 1658
    },
    {
      "epoch": 0.12172015004081184,
      "grad_norm": 1.5699933748529773,
      "learning_rate": 9.78102170512577e-06,
      "loss": 0.7768,
      "step": 1659
    },
    {
      "epoch": 0.12179351963095096,
      "grad_norm": 1.887258585580494,
      "learning_rate": 9.780673784957347e-06,
      "loss": 0.9524,
      "step": 1660
    },
    {
      "epoch": 0.12186688922109008,
      "grad_norm": 0.667111032110631,
      "learning_rate": 9.780325594812436e-06,
      "loss": 0.5757,
      "step": 1661
    },
    {
      "epoch": 0.12194025881122922,
      "grad_norm": 1.597449663581071,
      "learning_rate": 9.7799771347107e-06,
      "loss": 0.9226,
      "step": 1662
    },
    {
      "epoch": 0.12201362840136834,
      "grad_norm": 1.536443631477155,
      "learning_rate": 9.779628404671819e-06,
      "loss": 0.9143,
      "step": 1663
    },
    {
      "epoch": 0.12208699799150748,
      "grad_norm": 1.6559484610671131,
      "learning_rate": 9.779279404715485e-06,
      "loss": 0.9222,
      "step": 1664
    },
    {
      "epoch": 0.12208699799150748,
      "lm_loss": 1.7508,
      "step": 1664,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.12208699799150748,
      "lm_loss": 3.3065,
      "step": 1664,
      "vm_loss": 0.185
    },
    {
      "epoch": 0.12208699799150748,
      "lm_loss": 0.3553,
      "step": 1664,
      "vm_loss": 0.11
    },
    {
      "epoch": 0.12208699799150748,
      "lm_loss": 0.0476,
      "step": 1664,
      "vm_loss": 0.086
    },
    {
      "epoch": 0.12208699799150748,
      "lm_loss": 0.6726,
      "step": 1664,
      "vm_loss": 0.1435
    },
    {
      "epoch": 0.12208699799150748,
      "lm_loss": 0.4693,
      "step": 1664,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.12208699799150748,
      "lm_loss": 0.2747,
      "step": 1664,
      "vm_loss": 0.1093
    },
    {
      "epoch": 0.12208699799150748,
      "lm_loss": 0.5255,
      "step": 1664,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.1221603675816466,
      "grad_norm": 1.4436705641438528,
      "learning_rate": 9.778930134861407e-06,
      "loss": 0.8202,
      "step": 1665
    },
    {
      "epoch": 0.12223373717178572,
      "grad_norm": 0.6475949419868169,
      "learning_rate": 9.77858059512931e-06,
      "loss": 0.5634,
      "step": 1666
    },
    {
      "epoch": 0.12230710676192486,
      "grad_norm": 1.7705715984718347,
      "learning_rate": 9.778230785538932e-06,
      "loss": 0.8285,
      "step": 1667
    },
    {
      "epoch": 0.12238047635206398,
      "grad_norm": 1.4960094694080737,
      "learning_rate": 9.77788070611003e-06,
      "loss": 0.8678,
      "step": 1668
    },
    {
      "epoch": 0.1224538459422031,
      "grad_norm": 1.6103712999147926,
      "learning_rate": 9.77753035686237e-06,
      "loss": 0.7897,
      "step": 1669
    },
    {
      "epoch": 0.12252721553234223,
      "grad_norm": 1.5059988758540934,
      "learning_rate": 9.777179737815743e-06,
      "loss": 0.8481,
      "step": 1670
    },
    {
      "epoch": 0.12260058512248136,
      "grad_norm": 1.8713719054534983,
      "learning_rate": 9.776828848989942e-06,
      "loss": 0.9145,
      "step": 1671
    },
    {
      "epoch": 0.12267395471262049,
      "grad_norm": 1.6999365001378108,
      "learning_rate": 9.776477690404787e-06,
      "loss": 0.9557,
      "step": 1672
    },
    {
      "epoch": 0.12267395471262049,
      "lm_loss": 0.5979,
      "step": 1672,
      "vm_loss": 0.048
    },
    {
      "epoch": 0.12267395471262049,
      "lm_loss": 0.5853,
      "step": 1672,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.12267395471262049,
      "lm_loss": 0.4429,
      "step": 1672,
      "vm_loss": 0.1769
    },
    {
      "epoch": 0.12267395471262049,
      "lm_loss": 0.7137,
      "step": 1672,
      "vm_loss": 0.1692
    },
    {
      "epoch": 0.12267395471262049,
      "lm_loss": 0.1888,
      "step": 1672,
      "vm_loss": 0.2179
    },
    {
      "epoch": 0.12267395471262049,
      "lm_loss": 1.3328,
      "step": 1672,
      "vm_loss": 0.2434
    },
    {
      "epoch": 0.12267395471262049,
      "lm_loss": 0.0847,
      "step": 1672,
      "vm_loss": 0.1499
    },
    {
      "epoch": 0.12267395471262049,
      "lm_loss": 1.0512,
      "step": 1672,
      "vm_loss": 0.1339
    },
    {
      "epoch": 0.12274732430275961,
      "grad_norm": 1.6488760958680837,
      "learning_rate": 9.77612626208011e-06,
      "loss": 0.8593,
      "step": 1673
    },
    {
      "epoch": 0.12282069389289874,
      "grad_norm": 1.7021494196974842,
      "learning_rate": 9.775774564035754e-06,
      "loss": 0.9415,
      "step": 1674
    },
    {
      "epoch": 0.12289406348303787,
      "grad_norm": 0.6842028168589762,
      "learning_rate": 9.77542259629158e-06,
      "loss": 0.5618,
      "step": 1675
    },
    {
      "epoch": 0.122967433073177,
      "grad_norm": 1.8192866156715797,
      "learning_rate": 9.775070358867469e-06,
      "loss": 0.9903,
      "step": 1676
    },
    {
      "epoch": 0.12304080266331613,
      "grad_norm": 0.6172593479126706,
      "learning_rate": 9.774717851783306e-06,
      "loss": 0.5648,
      "step": 1677
    },
    {
      "epoch": 0.12311417225345525,
      "grad_norm": 1.8413361598014766,
      "learning_rate": 9.774365075059003e-06,
      "loss": 0.9797,
      "step": 1678
    },
    {
      "epoch": 0.12318754184359437,
      "grad_norm": 1.6868148417387911,
      "learning_rate": 9.77401202871448e-06,
      "loss": 0.9073,
      "step": 1679
    },
    {
      "epoch": 0.12326091143373351,
      "grad_norm": 1.771632707294874,
      "learning_rate": 9.773658712769678e-06,
      "loss": 0.8667,
      "step": 1680
    },
    {
      "epoch": 0.12326091143373351,
      "lm_loss": 0.5734,
      "step": 1680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12326091143373351,
      "lm_loss": 0.3755,
      "step": 1680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12326091143373351,
      "lm_loss": 0.4581,
      "step": 1680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12326091143373351,
      "lm_loss": 0.497,
      "step": 1680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12326091143373351,
      "lm_loss": 0.4946,
      "step": 1680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12326091143373351,
      "lm_loss": 0.5921,
      "step": 1680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12326091143373351,
      "lm_loss": 0.5585,
      "step": 1680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12326091143373351,
      "lm_loss": 0.582,
      "step": 1680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12333428102387263,
      "grad_norm": 0.6586972645976671,
      "learning_rate": 9.773305127244544e-06,
      "loss": 0.5627,
      "step": 1681
    },
    {
      "epoch": 0.12340765061401175,
      "grad_norm": 1.730201219851751,
      "learning_rate": 9.77295127215905e-06,
      "loss": 0.7701,
      "step": 1682
    },
    {
      "epoch": 0.12348102020415089,
      "grad_norm": 1.5600575404018278,
      "learning_rate": 9.772597147533175e-06,
      "loss": 0.8251,
      "step": 1683
    },
    {
      "epoch": 0.12355438979429001,
      "grad_norm": 1.6817585648088442,
      "learning_rate": 9.772242753386924e-06,
      "loss": 0.9058,
      "step": 1684
    },
    {
      "epoch": 0.12362775938442915,
      "grad_norm": 0.6222087806581881,
      "learning_rate": 9.771888089740304e-06,
      "loss": 0.545,
      "step": 1685
    },
    {
      "epoch": 0.12370112897456827,
      "grad_norm": 1.8502226209088393,
      "learning_rate": 9.771533156613346e-06,
      "loss": 0.9014,
      "step": 1686
    },
    {
      "epoch": 0.12377449856470739,
      "grad_norm": 1.5964784449358178,
      "learning_rate": 9.771177954026095e-06,
      "loss": 0.8597,
      "step": 1687
    },
    {
      "epoch": 0.12384786815484652,
      "grad_norm": 1.7262244595301173,
      "learning_rate": 9.770822481998608e-06,
      "loss": 0.9689,
      "step": 1688
    },
    {
      "epoch": 0.12384786815484652,
      "lm_loss": 0.9419,
      "step": 1688,
      "vm_loss": 0.0597
    },
    {
      "epoch": 0.12384786815484652,
      "lm_loss": 0.0241,
      "step": 1688,
      "vm_loss": 0.0808
    },
    {
      "epoch": 0.12384786815484652,
      "lm_loss": 1.1392,
      "step": 1688,
      "vm_loss": 0.1256
    },
    {
      "epoch": 0.12384786815484652,
      "lm_loss": 0.325,
      "step": 1688,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.12384786815484652,
      "lm_loss": 1.0608,
      "step": 1688,
      "vm_loss": 0.1583
    },
    {
      "epoch": 0.12384786815484652,
      "lm_loss": 1.5174,
      "step": 1688,
      "vm_loss": 0.0747
    },
    {
      "epoch": 0.12384786815484652,
      "lm_loss": 0.5848,
      "step": 1688,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.12384786815484652,
      "lm_loss": 0.8758,
      "step": 1688,
      "vm_loss": 0.1866
    },
    {
      "epoch": 0.12392123774498565,
      "grad_norm": 1.674851026209492,
      "learning_rate": 9.770466740550963e-06,
      "loss": 0.8119,
      "step": 1689
    },
    {
      "epoch": 0.12399460733512477,
      "grad_norm": 1.5385496924131263,
      "learning_rate": 9.770110729703247e-06,
      "loss": 0.7806,
      "step": 1690
    },
    {
      "epoch": 0.1240679769252639,
      "grad_norm": 1.6676725277478923,
      "learning_rate": 9.769754449475562e-06,
      "loss": 0.8956,
      "step": 1691
    },
    {
      "epoch": 0.12414134651540303,
      "grad_norm": 1.5572426666843235,
      "learning_rate": 9.769397899888034e-06,
      "loss": 0.9679,
      "step": 1692
    },
    {
      "epoch": 0.12421471610554216,
      "grad_norm": 0.648167402416836,
      "learning_rate": 9.769041080960794e-06,
      "loss": 0.5618,
      "step": 1693
    },
    {
      "epoch": 0.12428808569568128,
      "grad_norm": 1.7316616723774048,
      "learning_rate": 9.768683992713993e-06,
      "loss": 0.886,
      "step": 1694
    },
    {
      "epoch": 0.1243614552858204,
      "grad_norm": 1.398492406843519,
      "learning_rate": 9.7683266351678e-06,
      "loss": 0.8291,
      "step": 1695
    },
    {
      "epoch": 0.12443482487595954,
      "grad_norm": 1.7371964852057067,
      "learning_rate": 9.767969008342392e-06,
      "loss": 0.8923,
      "step": 1696
    },
    {
      "epoch": 0.12443482487595954,
      "lm_loss": 1.3382,
      "step": 1696,
      "vm_loss": 0.1422
    },
    {
      "epoch": 0.12443482487595954,
      "lm_loss": 0.4507,
      "step": 1696,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.12443482487595954,
      "lm_loss": 1.1391,
      "step": 1696,
      "vm_loss": 0.0976
    },
    {
      "epoch": 0.12443482487595954,
      "lm_loss": 0.1372,
      "step": 1696,
      "vm_loss": 0.0448
    },
    {
      "epoch": 0.12443482487595954,
      "lm_loss": 0.8294,
      "step": 1696,
      "vm_loss": 0.1749
    },
    {
      "epoch": 0.12443482487595954,
      "lm_loss": 0.4789,
      "step": 1696,
      "vm_loss": 0.098
    },
    {
      "epoch": 0.12443482487595954,
      "lm_loss": 0.2342,
      "step": 1696,
      "vm_loss": 0.1964
    },
    {
      "epoch": 0.12443482487595954,
      "lm_loss": 0.1697,
      "step": 1696,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.12450819446609866,
      "grad_norm": 1.678193344784059,
      "learning_rate": 9.767611112257967e-06,
      "loss": 0.834,
      "step": 1697
    },
    {
      "epoch": 0.12458156405623778,
      "grad_norm": 1.547138286162831,
      "learning_rate": 9.767252946934736e-06,
      "loss": 0.8743,
      "step": 1698
    },
    {
      "epoch": 0.12465493364637692,
      "grad_norm": 1.7227216694567953,
      "learning_rate": 9.766894512392923e-06,
      "loss": 0.8644,
      "step": 1699
    },
    {
      "epoch": 0.12472830323651604,
      "grad_norm": 1.5040207749751417,
      "learning_rate": 9.766535808652774e-06,
      "loss": 0.8532,
      "step": 1700
    },
    {
      "epoch": 0.12480167282665518,
      "grad_norm": 1.6279668443129336,
      "learning_rate": 9.766176835734545e-06,
      "loss": 0.9494,
      "step": 1701
    },
    {
      "epoch": 0.1248750424167943,
      "grad_norm": 0.7235564799004175,
      "learning_rate": 9.765817593658505e-06,
      "loss": 0.5691,
      "step": 1702
    },
    {
      "epoch": 0.12494841200693342,
      "grad_norm": 1.7132202571376498,
      "learning_rate": 9.765458082444942e-06,
      "loss": 0.9311,
      "step": 1703
    },
    {
      "epoch": 0.12502178159707256,
      "grad_norm": 1.6561389374857238,
      "learning_rate": 9.765098302114162e-06,
      "loss": 0.9007,
      "step": 1704
    },
    {
      "epoch": 0.12502178159707256,
      "lm_loss": 1.0194,
      "step": 1704,
      "vm_loss": 0.0762
    },
    {
      "epoch": 0.12502178159707256,
      "lm_loss": 0.0862,
      "step": 1704,
      "vm_loss": 0.123
    },
    {
      "epoch": 0.12502178159707256,
      "lm_loss": 0.3864,
      "step": 1704,
      "vm_loss": 0.1192
    },
    {
      "epoch": 0.12502178159707256,
      "lm_loss": 1.0542,
      "step": 1704,
      "vm_loss": 0.1913
    },
    {
      "epoch": 0.12502178159707256,
      "lm_loss": 0.8933,
      "step": 1704,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.12502178159707256,
      "lm_loss": 1.3113,
      "step": 1704,
      "vm_loss": 0.1588
    },
    {
      "epoch": 0.12502178159707256,
      "lm_loss": 0.4962,
      "step": 1704,
      "vm_loss": 0.1367
    },
    {
      "epoch": 0.12502178159707256,
      "lm_loss": 1.1945,
      "step": 1704,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.12509515118721168,
      "grad_norm": 1.6536870260100063,
      "learning_rate": 9.764738252686478e-06,
      "loss": 0.8499,
      "step": 1705
    },
    {
      "epoch": 0.1251685207773508,
      "grad_norm": 1.8560743746020147,
      "learning_rate": 9.764377934182227e-06,
      "loss": 0.9271,
      "step": 1706
    },
    {
      "epoch": 0.12524189036748992,
      "grad_norm": 1.6084072885154033,
      "learning_rate": 9.764017346621755e-06,
      "loss": 0.8973,
      "step": 1707
    },
    {
      "epoch": 0.12531525995762907,
      "grad_norm": 0.6413778777039348,
      "learning_rate": 9.763656490025423e-06,
      "loss": 0.5609,
      "step": 1708
    },
    {
      "epoch": 0.1253886295477682,
      "grad_norm": 1.7854354620173385,
      "learning_rate": 9.763295364413616e-06,
      "loss": 0.8236,
      "step": 1709
    },
    {
      "epoch": 0.12546199913790732,
      "grad_norm": 1.7774826893666986,
      "learning_rate": 9.762933969806722e-06,
      "loss": 0.8848,
      "step": 1710
    },
    {
      "epoch": 0.12553536872804644,
      "grad_norm": 1.5366767321396702,
      "learning_rate": 9.76257230622515e-06,
      "loss": 0.8736,
      "step": 1711
    },
    {
      "epoch": 0.12560873831818556,
      "grad_norm": 0.6105163378525098,
      "learning_rate": 9.762210373689327e-06,
      "loss": 0.5616,
      "step": 1712
    },
    {
      "epoch": 0.12560873831818556,
      "lm_loss": 1.3186,
      "step": 1712,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.12560873831818556,
      "lm_loss": 1.1882,
      "step": 1712,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.12560873831818556,
      "lm_loss": 0.3499,
      "step": 1712,
      "vm_loss": 0.0704
    },
    {
      "epoch": 0.12560873831818556,
      "lm_loss": 1.0804,
      "step": 1712,
      "vm_loss": 0.2982
    },
    {
      "epoch": 0.12560873831818556,
      "lm_loss": 1.4335,
      "step": 1712,
      "vm_loss": 0.2274
    },
    {
      "epoch": 0.12560873831818556,
      "lm_loss": 0.2582,
      "step": 1712,
      "vm_loss": 0.0958
    },
    {
      "epoch": 0.12560873831818556,
      "lm_loss": 0.3322,
      "step": 1712,
      "vm_loss": 0.2013
    },
    {
      "epoch": 0.12560873831818556,
      "lm_loss": 1.1027,
      "step": 1712,
      "vm_loss": 0.1672
    },
    {
      "epoch": 0.1256821079083247,
      "grad_norm": 1.8236188315429167,
      "learning_rate": 9.76184817221969e-06,
      "loss": 0.9282,
      "step": 1713
    },
    {
      "epoch": 0.12575547749846383,
      "grad_norm": 1.390232667179379,
      "learning_rate": 9.761485701836696e-06,
      "loss": 0.8287,
      "step": 1714
    },
    {
      "epoch": 0.12582884708860295,
      "grad_norm": 0.5970525988682917,
      "learning_rate": 9.761122962560812e-06,
      "loss": 0.5703,
      "step": 1715
    },
    {
      "epoch": 0.12590221667874207,
      "grad_norm": 1.6456100683184702,
      "learning_rate": 9.760759954412523e-06,
      "loss": 0.8881,
      "step": 1716
    },
    {
      "epoch": 0.1259755862688812,
      "grad_norm": 1.789674045150771,
      "learning_rate": 9.76039667741233e-06,
      "loss": 0.9072,
      "step": 1717
    },
    {
      "epoch": 0.12604895585902035,
      "grad_norm": 1.5866158487338449,
      "learning_rate": 9.760033131580748e-06,
      "loss": 0.9048,
      "step": 1718
    },
    {
      "epoch": 0.12612232544915947,
      "grad_norm": 1.6313985750968072,
      "learning_rate": 9.759669316938307e-06,
      "loss": 0.9559,
      "step": 1719
    },
    {
      "epoch": 0.1261956950392986,
      "grad_norm": 0.6540034864190021,
      "learning_rate": 9.759305233505552e-06,
      "loss": 0.5533,
      "step": 1720
    },
    {
      "epoch": 0.1261956950392986,
      "lm_loss": 1.6972,
      "step": 1720,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.1261956950392986,
      "lm_loss": 0.2379,
      "step": 1720,
      "vm_loss": 0.0659
    },
    {
      "epoch": 0.1261956950392986,
      "lm_loss": 1.016,
      "step": 1720,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.1261956950392986,
      "lm_loss": 1.3617,
      "step": 1720,
      "vm_loss": 0.1086
    },
    {
      "epoch": 0.1261956950392986,
      "lm_loss": 0.8842,
      "step": 1720,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.1261956950392986,
      "lm_loss": 0.2504,
      "step": 1720,
      "vm_loss": 0.0638
    },
    {
      "epoch": 0.1261956950392986,
      "lm_loss": 0.5623,
      "step": 1720,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.1261956950392986,
      "lm_loss": 1.2041,
      "step": 1720,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.1262690646294377,
      "grad_norm": 1.677236115205347,
      "learning_rate": 9.758940881303043e-06,
      "loss": 0.837,
      "step": 1721
    },
    {
      "epoch": 0.12634243421957683,
      "grad_norm": 0.6066402317630766,
      "learning_rate": 9.75857626035136e-06,
      "loss": 0.551,
      "step": 1722
    },
    {
      "epoch": 0.12641580380971595,
      "grad_norm": 0.5639510560046026,
      "learning_rate": 9.758211370671089e-06,
      "loss": 0.5488,
      "step": 1723
    },
    {
      "epoch": 0.1264891733998551,
      "grad_norm": 1.7454428671897575,
      "learning_rate": 9.757846212282839e-06,
      "loss": 0.9281,
      "step": 1724
    },
    {
      "epoch": 0.12656254298999423,
      "grad_norm": 1.8627478401383708,
      "learning_rate": 9.757480785207231e-06,
      "loss": 0.9577,
      "step": 1725
    },
    {
      "epoch": 0.12663591258013335,
      "grad_norm": 1.5798124188889304,
      "learning_rate": 9.757115089464902e-06,
      "loss": 0.9006,
      "step": 1726
    },
    {
      "epoch": 0.12670928217027247,
      "grad_norm": 1.6111145637818522,
      "learning_rate": 9.756749125076502e-06,
      "loss": 0.9089,
      "step": 1727
    },
    {
      "epoch": 0.1267826517604116,
      "grad_norm": 1.6119015078509606,
      "learning_rate": 9.756382892062699e-06,
      "loss": 0.9923,
      "step": 1728
    },
    {
      "epoch": 0.1267826517604116,
      "lm_loss": 1.567,
      "step": 1728,
      "vm_loss": 0.0284
    },
    {
      "epoch": 0.1267826517604116,
      "lm_loss": 1.7913,
      "step": 1728,
      "vm_loss": 0.197
    },
    {
      "epoch": 0.1267826517604116,
      "lm_loss": 0.9292,
      "step": 1728,
      "vm_loss": 0.145
    },
    {
      "epoch": 0.1267826517604116,
      "lm_loss": 0.1142,
      "step": 1728,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.1267826517604116,
      "lm_loss": 0.9194,
      "step": 1728,
      "vm_loss": 0.1268
    },
    {
      "epoch": 0.1267826517604116,
      "lm_loss": 0.9901,
      "step": 1728,
      "vm_loss": 0.1156
    },
    {
      "epoch": 0.1267826517604116,
      "lm_loss": 0.4883,
      "step": 1728,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.1267826517604116,
      "lm_loss": 0.4468,
      "step": 1728,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.12685602135055074,
      "grad_norm": 1.6027324823848461,
      "learning_rate": 9.756016390444174e-06,
      "loss": 0.9203,
      "step": 1729
    },
    {
      "epoch": 0.12692939094068986,
      "grad_norm": 1.5682433076348306,
      "learning_rate": 9.755649620241627e-06,
      "loss": 0.8626,
      "step": 1730
    },
    {
      "epoch": 0.12700276053082898,
      "grad_norm": 1.698235362526084,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.8607,
      "step": 1731
    },
    {
      "epoch": 0.1270761301209681,
      "grad_norm": 1.6942785573276706,
      "learning_rate": 9.754915274167328e-06,
      "loss": 0.8279,
      "step": 1732
    },
    {
      "epoch": 0.12714949971110723,
      "grad_norm": 1.607534908157935,
      "learning_rate": 9.754547698337043e-06,
      "loss": 0.8325,
      "step": 1733
    },
    {
      "epoch": 0.12722286930124638,
      "grad_norm": 1.6520768481639554,
      "learning_rate": 9.754179854005678e-06,
      "loss": 0.9815,
      "step": 1734
    },
    {
      "epoch": 0.1272962388913855,
      "grad_norm": 1.7669669641297976,
      "learning_rate": 9.753811741194002e-06,
      "loss": 0.893,
      "step": 1735
    },
    {
      "epoch": 0.12736960848152462,
      "grad_norm": 1.6848194111043482,
      "learning_rate": 9.753443359922803e-06,
      "loss": 0.8474,
      "step": 1736
    },
    {
      "epoch": 0.12736960848152462,
      "lm_loss": 1.17,
      "step": 1736,
      "vm_loss": 0.0962
    },
    {
      "epoch": 0.12736960848152462,
      "lm_loss": 0.5646,
      "step": 1736,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.12736960848152462,
      "lm_loss": 0.2072,
      "step": 1736,
      "vm_loss": 0.1283
    },
    {
      "epoch": 0.12736960848152462,
      "lm_loss": 1.127,
      "step": 1736,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.12736960848152462,
      "lm_loss": 0.0683,
      "step": 1736,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.12736960848152462,
      "lm_loss": 0.2861,
      "step": 1736,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.12736960848152462,
      "lm_loss": 0.8059,
      "step": 1736,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.12736960848152462,
      "lm_loss": 0.5299,
      "step": 1736,
      "vm_loss": 0.2034
    },
    {
      "epoch": 0.12744297807166374,
      "grad_norm": 1.715943239176004,
      "learning_rate": 9.753074710212888e-06,
      "loss": 0.8522,
      "step": 1737
    },
    {
      "epoch": 0.12751634766180286,
      "grad_norm": 1.0955919837054924,
      "learning_rate": 9.752705792085074e-06,
      "loss": 0.6,
      "step": 1738
    },
    {
      "epoch": 0.12758971725194201,
      "grad_norm": 1.722396149262254,
      "learning_rate": 9.752336605560191e-06,
      "loss": 0.9774,
      "step": 1739
    },
    {
      "epoch": 0.12766308684208114,
      "grad_norm": 1.7945361560146653,
      "learning_rate": 9.751967150659093e-06,
      "loss": 0.8214,
      "step": 1740
    },
    {
      "epoch": 0.12773645643222026,
      "grad_norm": 1.6938664705410313,
      "learning_rate": 9.751597427402642e-06,
      "loss": 0.9464,
      "step": 1741
    },
    {
      "epoch": 0.12780982602235938,
      "grad_norm": 1.7403474206508112,
      "learning_rate": 9.751227435811717e-06,
      "loss": 0.8705,
      "step": 1742
    },
    {
      "epoch": 0.1278831956124985,
      "grad_norm": 0.8337628531939011,
      "learning_rate": 9.75085717590721e-06,
      "loss": 0.5624,
      "step": 1743
    },
    {
      "epoch": 0.12795656520263762,
      "grad_norm": 1.7858676471930146,
      "learning_rate": 9.750486647710036e-06,
      "loss": 0.8838,
      "step": 1744
    },
    {
      "epoch": 0.12795656520263762,
      "lm_loss": 0.501,
      "step": 1744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12795656520263762,
      "lm_loss": 0.4529,
      "step": 1744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12795656520263762,
      "lm_loss": 0.5282,
      "step": 1744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12795656520263762,
      "lm_loss": 0.5974,
      "step": 1744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12795656520263762,
      "lm_loss": 0.498,
      "step": 1744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12795656520263762,
      "lm_loss": 0.6425,
      "step": 1744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12795656520263762,
      "lm_loss": 0.4269,
      "step": 1744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12795656520263762,
      "lm_loss": 0.5824,
      "step": 1744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.12802993479277677,
      "grad_norm": 0.7424343226907866,
      "learning_rate": 9.750115851241116e-06,
      "loss": 0.5488,
      "step": 1745
    },
    {
      "epoch": 0.1281033043829159,
      "grad_norm": 0.6545428459932701,
      "learning_rate": 9.74974478652139e-06,
      "loss": 0.5832,
      "step": 1746
    },
    {
      "epoch": 0.12817667397305502,
      "grad_norm": 0.621476205355688,
      "learning_rate": 9.749373453571813e-06,
      "loss": 0.5442,
      "step": 1747
    },
    {
      "epoch": 0.12825004356319414,
      "grad_norm": 1.6667718598230796,
      "learning_rate": 9.749001852413357e-06,
      "loss": 0.9122,
      "step": 1748
    },
    {
      "epoch": 0.12832341315333326,
      "grad_norm": 1.9208572494437806,
      "learning_rate": 9.748629983067004e-06,
      "loss": 0.8084,
      "step": 1749
    },
    {
      "epoch": 0.1283967827434724,
      "grad_norm": 1.6009626555638437,
      "learning_rate": 9.748257845553757e-06,
      "loss": 0.899,
      "step": 1750
    },
    {
      "epoch": 0.12847015233361153,
      "grad_norm": 1.6658135150344704,
      "learning_rate": 9.74788543989463e-06,
      "loss": 0.992,
      "step": 1751
    },
    {
      "epoch": 0.12854352192375065,
      "grad_norm": 1.6823970089085931,
      "learning_rate": 9.747512766110656e-06,
      "loss": 0.8875,
      "step": 1752
    },
    {
      "epoch": 0.12854352192375065,
      "lm_loss": 1.4596,
      "step": 1752,
      "vm_loss": 0.0598
    },
    {
      "epoch": 0.12854352192375065,
      "lm_loss": 1.2624,
      "step": 1752,
      "vm_loss": 0.1088
    },
    {
      "epoch": 0.12854352192375065,
      "lm_loss": 0.5497,
      "step": 1752,
      "vm_loss": 0.11
    },
    {
      "epoch": 0.12854352192375065,
      "lm_loss": 0.8979,
      "step": 1752,
      "vm_loss": 0.1337
    },
    {
      "epoch": 0.12854352192375065,
      "lm_loss": 0.9774,
      "step": 1752,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.12854352192375065,
      "lm_loss": 2.1035,
      "step": 1752,
      "vm_loss": 0.222
    },
    {
      "epoch": 0.12854352192375065,
      "lm_loss": 0.314,
      "step": 1752,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.12854352192375065,
      "lm_loss": 0.3373,
      "step": 1752,
      "vm_loss": 0.1196
    },
    {
      "epoch": 0.12861689151388978,
      "grad_norm": 1.7915115174802456,
      "learning_rate": 9.747139824222877e-06,
      "loss": 0.9636,
      "step": 1753
    },
    {
      "epoch": 0.1286902611040289,
      "grad_norm": 1.6485923572882557,
      "learning_rate": 9.746766614252357e-06,
      "loss": 0.853,
      "step": 1754
    },
    {
      "epoch": 0.12876363069416805,
      "grad_norm": 1.7315603205919043,
      "learning_rate": 9.74639313622017e-06,
      "loss": 0.9667,
      "step": 1755
    },
    {
      "epoch": 0.12883700028430717,
      "grad_norm": 1.2957795612909975,
      "learning_rate": 9.74601939014741e-06,
      "loss": 0.5772,
      "step": 1756
    },
    {
      "epoch": 0.1289103698744463,
      "grad_norm": 1.570943105819532,
      "learning_rate": 9.74564537605518e-06,
      "loss": 0.8898,
      "step": 1757
    },
    {
      "epoch": 0.1289837394645854,
      "grad_norm": 1.560349705490927,
      "learning_rate": 9.745271093964604e-06,
      "loss": 0.8855,
      "step": 1758
    },
    {
      "epoch": 0.12905710905472453,
      "grad_norm": 1.5487087570317448,
      "learning_rate": 9.744896543896818e-06,
      "loss": 0.8877,
      "step": 1759
    },
    {
      "epoch": 0.12913047864486368,
      "grad_norm": 1.5338526363245037,
      "learning_rate": 9.744521725872974e-06,
      "loss": 0.8386,
      "step": 1760
    },
    {
      "epoch": 0.12913047864486368,
      "lm_loss": 1.2743,
      "step": 1760,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.12913047864486368,
      "lm_loss": 0.8534,
      "step": 1760,
      "vm_loss": 0.1412
    },
    {
      "epoch": 0.12913047864486368,
      "lm_loss": 0.8342,
      "step": 1760,
      "vm_loss": 0.0863
    },
    {
      "epoch": 0.12913047864486368,
      "lm_loss": 0.1949,
      "step": 1760,
      "vm_loss": 0.183
    },
    {
      "epoch": 0.12913047864486368,
      "lm_loss": 0.3598,
      "step": 1760,
      "vm_loss": 0.1185
    },
    {
      "epoch": 0.12913047864486368,
      "lm_loss": 0.5314,
      "step": 1760,
      "vm_loss": 0.1593
    },
    {
      "epoch": 0.12913047864486368,
      "lm_loss": 1.0652,
      "step": 1760,
      "vm_loss": 0.2056
    },
    {
      "epoch": 0.12913047864486368,
      "lm_loss": 1.231,
      "step": 1760,
      "vm_loss": 0.2043
    },
    {
      "epoch": 0.1292038482350028,
      "grad_norm": 1.4982014630312839,
      "learning_rate": 9.744146639914238e-06,
      "loss": 0.8803,
      "step": 1761
    },
    {
      "epoch": 0.12927721782514193,
      "grad_norm": 1.8227076317838373,
      "learning_rate": 9.743771286041792e-06,
      "loss": 0.8914,
      "step": 1762
    },
    {
      "epoch": 0.12935058741528105,
      "grad_norm": 0.7512105463951821,
      "learning_rate": 9.743395664276836e-06,
      "loss": 0.5466,
      "step": 1763
    },
    {
      "epoch": 0.12942395700542017,
      "grad_norm": 1.498038513344772,
      "learning_rate": 9.743019774640577e-06,
      "loss": 0.9036,
      "step": 1764
    },
    {
      "epoch": 0.1294973265955593,
      "grad_norm": 0.6989474952007925,
      "learning_rate": 9.742643617154248e-06,
      "loss": 0.5655,
      "step": 1765
    },
    {
      "epoch": 0.12957069618569844,
      "grad_norm": 1.656660136583307,
      "learning_rate": 9.742267191839088e-06,
      "loss": 0.8786,
      "step": 1766
    },
    {
      "epoch": 0.12964406577583756,
      "grad_norm": 1.812845443845196,
      "learning_rate": 9.741890498716355e-06,
      "loss": 0.9779,
      "step": 1767
    },
    {
      "epoch": 0.12971743536597669,
      "grad_norm": 1.882601071090245,
      "learning_rate": 9.741513537807324e-06,
      "loss": 0.8769,
      "step": 1768
    },
    {
      "epoch": 0.12971743536597669,
      "lm_loss": 1.5735,
      "step": 1768,
      "vm_loss": 0.0713
    },
    {
      "epoch": 0.12971743536597669,
      "lm_loss": 0.5491,
      "step": 1768,
      "vm_loss": 0.1047
    },
    {
      "epoch": 0.12971743536597669,
      "lm_loss": 0.8149,
      "step": 1768,
      "vm_loss": 0.1579
    },
    {
      "epoch": 0.12971743536597669,
      "lm_loss": 0.8413,
      "step": 1768,
      "vm_loss": 0.22
    },
    {
      "epoch": 0.12971743536597669,
      "lm_loss": 0.4042,
      "step": 1768,
      "vm_loss": 0.2546
    },
    {
      "epoch": 0.12971743536597669,
      "lm_loss": 0.2314,
      "step": 1768,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.12971743536597669,
      "lm_loss": 0.5934,
      "step": 1768,
      "vm_loss": 0.0757
    },
    {
      "epoch": 0.12971743536597669,
      "lm_loss": 0.7722,
      "step": 1768,
      "vm_loss": 0.153
    },
    {
      "epoch": 0.1297908049561158,
      "grad_norm": 1.6320816197224406,
      "learning_rate": 9.741136309133279e-06,
      "loss": 0.8995,
      "step": 1769
    },
    {
      "epoch": 0.12986417454625493,
      "grad_norm": 0.6486599152298511,
      "learning_rate": 9.740758812715528e-06,
      "loss": 0.5595,
      "step": 1770
    },
    {
      "epoch": 0.12993754413639408,
      "grad_norm": 1.5714263997091291,
      "learning_rate": 9.740381048575385e-06,
      "loss": 0.7964,
      "step": 1771
    },
    {
      "epoch": 0.1300109137265332,
      "grad_norm": 1.8185539319610864,
      "learning_rate": 9.740003016734184e-06,
      "loss": 0.8661,
      "step": 1772
    },
    {
      "epoch": 0.13008428331667232,
      "grad_norm": 1.9581855861834876,
      "learning_rate": 9.739624717213276e-06,
      "loss": 0.9143,
      "step": 1773
    },
    {
      "epoch": 0.13015765290681144,
      "grad_norm": 0.5988601012494817,
      "learning_rate": 9.739246150034022e-06,
      "loss": 0.5506,
      "step": 1774
    },
    {
      "epoch": 0.13023102249695057,
      "grad_norm": 1.892540494102277,
      "learning_rate": 9.738867315217801e-06,
      "loss": 0.9227,
      "step": 1775
    },
    {
      "epoch": 0.13030439208708972,
      "grad_norm": 1.7097359708939872,
      "learning_rate": 9.738488212786008e-06,
      "loss": 0.8391,
      "step": 1776
    },
    {
      "epoch": 0.13030439208708972,
      "lm_loss": 0.5906,
      "step": 1776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13030439208708972,
      "lm_loss": 0.5772,
      "step": 1776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13030439208708972,
      "lm_loss": 0.6404,
      "step": 1776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13030439208708972,
      "lm_loss": 0.5759,
      "step": 1776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13030439208708972,
      "lm_loss": 0.5232,
      "step": 1776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13030439208708972,
      "lm_loss": 0.5398,
      "step": 1776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13030439208708972,
      "lm_loss": 0.6157,
      "step": 1776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13030439208708972,
      "lm_loss": 0.6703,
      "step": 1776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13037776167722884,
      "grad_norm": 0.6291333282364642,
      "learning_rate": 9.73810884276005e-06,
      "loss": 0.5722,
      "step": 1777
    },
    {
      "epoch": 0.13045113126736796,
      "grad_norm": 1.617130061476935,
      "learning_rate": 9.737729205161352e-06,
      "loss": 0.8802,
      "step": 1778
    },
    {
      "epoch": 0.13052450085750708,
      "grad_norm": 1.6243956432447957,
      "learning_rate": 9.737349300011353e-06,
      "loss": 0.7629,
      "step": 1779
    },
    {
      "epoch": 0.1305978704476462,
      "grad_norm": 1.3955887575942392,
      "learning_rate": 9.736969127331509e-06,
      "loss": 0.7646,
      "step": 1780
    },
    {
      "epoch": 0.13067124003778535,
      "grad_norm": 1.543973288422813,
      "learning_rate": 9.736588687143286e-06,
      "loss": 0.8993,
      "step": 1781
    },
    {
      "epoch": 0.13074460962792447,
      "grad_norm": 1.6684713580119204,
      "learning_rate": 9.736207979468169e-06,
      "loss": 0.8735,
      "step": 1782
    },
    {
      "epoch": 0.1308179792180636,
      "grad_norm": 0.634657954468274,
      "learning_rate": 9.73582700432766e-06,
      "loss": 0.5744,
      "step": 1783
    },
    {
      "epoch": 0.13089134880820272,
      "grad_norm": 1.402407995125086,
      "learning_rate": 9.73544576174327e-06,
      "loss": 0.8382,
      "step": 1784
    },
    {
      "epoch": 0.13089134880820272,
      "lm_loss": 0.1574,
      "step": 1784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13089134880820272,
      "lm_loss": 0.493,
      "step": 1784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13089134880820272,
      "lm_loss": 0.6297,
      "step": 1784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13089134880820272,
      "lm_loss": 0.6283,
      "step": 1784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13089134880820272,
      "lm_loss": 0.5528,
      "step": 1784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13089134880820272,
      "lm_loss": 0.5694,
      "step": 1784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13089134880820272,
      "lm_loss": 0.6149,
      "step": 1784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13089134880820272,
      "lm_loss": 0.5351,
      "step": 1784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13096471839834184,
      "grad_norm": 0.5983540434838518,
      "learning_rate": 9.735064251736533e-06,
      "loss": 0.5538,
      "step": 1785
    },
    {
      "epoch": 0.13103808798848096,
      "grad_norm": 0.6168104154727292,
      "learning_rate": 9.73468247432899e-06,
      "loss": 0.5669,
      "step": 1786
    },
    {
      "epoch": 0.1311114575786201,
      "grad_norm": 1.5632447056606213,
      "learning_rate": 9.734300429542203e-06,
      "loss": 0.9349,
      "step": 1787
    },
    {
      "epoch": 0.13118482716875923,
      "grad_norm": 1.5931457899956456,
      "learning_rate": 9.733918117397748e-06,
      "loss": 0.8832,
      "step": 1788
    },
    {
      "epoch": 0.13125819675889835,
      "grad_norm": 1.7142921078286921,
      "learning_rate": 9.733535537917211e-06,
      "loss": 0.9846,
      "step": 1789
    },
    {
      "epoch": 0.13133156634903748,
      "grad_norm": 1.5661946173763108,
      "learning_rate": 9.7331526911222e-06,
      "loss": 0.9058,
      "step": 1790
    },
    {
      "epoch": 0.1314049359391766,
      "grad_norm": 1.5851001962524531,
      "learning_rate": 9.732769577034336e-06,
      "loss": 0.9264,
      "step": 1791
    },
    {
      "epoch": 0.13147830552931575,
      "grad_norm": 1.5455685437759208,
      "learning_rate": 9.732386195675255e-06,
      "loss": 0.886,
      "step": 1792
    },
    {
      "epoch": 0.13147830552931575,
      "lm_loss": 0.9399,
      "step": 1792,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.13147830552931575,
      "lm_loss": 1.1139,
      "step": 1792,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.13147830552931575,
      "lm_loss": 0.5936,
      "step": 1792,
      "vm_loss": 0.1553
    },
    {
      "epoch": 0.13147830552931575,
      "lm_loss": 1.0182,
      "step": 1792,
      "vm_loss": 0.1917
    },
    {
      "epoch": 0.13147830552931575,
      "lm_loss": 0.8402,
      "step": 1792,
      "vm_loss": 0.2323
    },
    {
      "epoch": 0.13147830552931575,
      "lm_loss": 0.8878,
      "step": 1792,
      "vm_loss": 0.1614
    },
    {
      "epoch": 0.13147830552931575,
      "lm_loss": 0.3276,
      "step": 1792,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.13147830552931575,
      "lm_loss": 0.9362,
      "step": 1792,
      "vm_loss": 0.1409
    },
    {
      "epoch": 0.13155167511945487,
      "grad_norm": 1.5917228913283248,
      "learning_rate": 9.732002547066603e-06,
      "loss": 0.8903,
      "step": 1793
    },
    {
      "epoch": 0.131625044709594,
      "grad_norm": 1.6646330233858866,
      "learning_rate": 9.73161863123005e-06,
      "loss": 0.8514,
      "step": 1794
    },
    {
      "epoch": 0.1316984142997331,
      "grad_norm": 0.7315472436131469,
      "learning_rate": 9.731234448187274e-06,
      "loss": 0.5573,
      "step": 1795
    },
    {
      "epoch": 0.13177178388987223,
      "grad_norm": 0.6596700803414316,
      "learning_rate": 9.730849997959974e-06,
      "loss": 0.5625,
      "step": 1796
    },
    {
      "epoch": 0.13184515348001138,
      "grad_norm": 1.6291240886353724,
      "learning_rate": 9.730465280569857e-06,
      "loss": 0.9057,
      "step": 1797
    },
    {
      "epoch": 0.1319185230701505,
      "grad_norm": 1.4920760219465454,
      "learning_rate": 9.730080296038653e-06,
      "loss": 0.8169,
      "step": 1798
    },
    {
      "epoch": 0.13199189266028963,
      "grad_norm": 0.7028657626346644,
      "learning_rate": 9.729695044388098e-06,
      "loss": 0.587,
      "step": 1799
    },
    {
      "epoch": 0.13206526225042875,
      "grad_norm": 1.8238094031809333,
      "learning_rate": 9.729309525639954e-06,
      "loss": 0.9384,
      "step": 1800
    },
    {
      "epoch": 0.13206526225042875,
      "lm_loss": 0.7119,
      "step": 1800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13206526225042875,
      "lm_loss": 0.6505,
      "step": 1800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13206526225042875,
      "lm_loss": 0.5341,
      "step": 1800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13206526225042875,
      "lm_loss": 0.4108,
      "step": 1800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13206526225042875,
      "lm_loss": 0.6395,
      "step": 1800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13206526225042875,
      "lm_loss": 0.612,
      "step": 1800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13206526225042875,
      "lm_loss": 0.602,
      "step": 1800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13206526225042875,
      "lm_loss": 0.5815,
      "step": 1800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13213863184056787,
      "grad_norm": 0.6787728797859915,
      "learning_rate": 9.728923739815988e-06,
      "loss": 0.5819,
      "step": 1801
    },
    {
      "epoch": 0.13221200143070702,
      "grad_norm": 1.8260282450635001,
      "learning_rate": 9.728537686937987e-06,
      "loss": 0.9139,
      "step": 1802
    },
    {
      "epoch": 0.13228537102084614,
      "grad_norm": 1.6435661912593955,
      "learning_rate": 9.728151367027752e-06,
      "loss": 0.8438,
      "step": 1803
    },
    {
      "epoch": 0.13235874061098526,
      "grad_norm": 1.7965645511010584,
      "learning_rate": 9.7277647801071e-06,
      "loss": 0.9186,
      "step": 1804
    },
    {
      "epoch": 0.1324321102011244,
      "grad_norm": 1.5841080177576174,
      "learning_rate": 9.727377926197864e-06,
      "loss": 0.8701,
      "step": 1805
    },
    {
      "epoch": 0.1325054797912635,
      "grad_norm": 0.6968988356436412,
      "learning_rate": 9.72699080532189e-06,
      "loss": 0.5744,
      "step": 1806
    },
    {
      "epoch": 0.13257884938140263,
      "grad_norm": 1.59967348885228,
      "learning_rate": 9.726603417501038e-06,
      "loss": 0.8217,
      "step": 1807
    },
    {
      "epoch": 0.13265221897154178,
      "grad_norm": 0.6210735029511175,
      "learning_rate": 9.726215762757187e-06,
      "loss": 0.5564,
      "step": 1808
    },
    {
      "epoch": 0.13265221897154178,
      "lm_loss": 0.4998,
      "step": 1808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13265221897154178,
      "lm_loss": 0.6552,
      "step": 1808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13265221897154178,
      "lm_loss": 0.644,
      "step": 1808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13265221897154178,
      "lm_loss": 0.6211,
      "step": 1808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13265221897154178,
      "lm_loss": 0.5957,
      "step": 1808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13265221897154178,
      "lm_loss": 0.5089,
      "step": 1808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13265221897154178,
      "lm_loss": 0.634,
      "step": 1808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13265221897154178,
      "lm_loss": 0.4903,
      "step": 1808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1327255885616809,
      "grad_norm": 0.598315442518538,
      "learning_rate": 9.725827841112226e-06,
      "loss": 0.5655,
      "step": 1809
    },
    {
      "epoch": 0.13279895815182002,
      "grad_norm": 0.6033779124503158,
      "learning_rate": 9.725439652588064e-06,
      "loss": 0.5628,
      "step": 1810
    },
    {
      "epoch": 0.13287232774195915,
      "grad_norm": 1.9862798269693167,
      "learning_rate": 9.725051197206623e-06,
      "loss": 0.9853,
      "step": 1811
    },
    {
      "epoch": 0.13294569733209827,
      "grad_norm": 1.6913710487959204,
      "learning_rate": 9.724662474989839e-06,
      "loss": 0.8477,
      "step": 1812
    },
    {
      "epoch": 0.13301906692223742,
      "grad_norm": 1.8341283587721875,
      "learning_rate": 9.724273485959665e-06,
      "loss": 0.7851,
      "step": 1813
    },
    {
      "epoch": 0.13309243651237654,
      "grad_norm": 1.9537119999347452,
      "learning_rate": 9.723884230138067e-06,
      "loss": 0.8563,
      "step": 1814
    },
    {
      "epoch": 0.13316580610251566,
      "grad_norm": 0.7076224803133236,
      "learning_rate": 9.723494707547027e-06,
      "loss": 0.5706,
      "step": 1815
    },
    {
      "epoch": 0.13323917569265478,
      "grad_norm": 1.6796467894806022,
      "learning_rate": 9.723104918208546e-06,
      "loss": 0.7814,
      "step": 1816
    },
    {
      "epoch": 0.13323917569265478,
      "lm_loss": 0.59,
      "step": 1816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13323917569265478,
      "lm_loss": 0.4868,
      "step": 1816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13323917569265478,
      "lm_loss": 0.5834,
      "step": 1816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13323917569265478,
      "lm_loss": 0.628,
      "step": 1816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13323917569265478,
      "lm_loss": 0.5839,
      "step": 1816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13323917569265478,
      "lm_loss": 0.5603,
      "step": 1816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13323917569265478,
      "lm_loss": 0.67,
      "step": 1816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13323917569265478,
      "lm_loss": 0.5899,
      "step": 1816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1333125452827939,
      "grad_norm": 0.6403468373963995,
      "learning_rate": 9.722714862144633e-06,
      "loss": 0.57,
      "step": 1817
    },
    {
      "epoch": 0.13338591487293305,
      "grad_norm": 0.6211213192191817,
      "learning_rate": 9.722324539377315e-06,
      "loss": 0.5704,
      "step": 1818
    },
    {
      "epoch": 0.13345928446307218,
      "grad_norm": 1.599671788839018,
      "learning_rate": 9.721933949928637e-06,
      "loss": 0.8024,
      "step": 1819
    },
    {
      "epoch": 0.1335326540532113,
      "grad_norm": 1.8678542769680715,
      "learning_rate": 9.721543093820653e-06,
      "loss": 0.9627,
      "step": 1820
    },
    {
      "epoch": 0.13360602364335042,
      "grad_norm": 1.588804946539032,
      "learning_rate": 9.721151971075439e-06,
      "loss": 0.9606,
      "step": 1821
    },
    {
      "epoch": 0.13367939323348954,
      "grad_norm": 0.8743314264827355,
      "learning_rate": 9.720760581715083e-06,
      "loss": 0.5809,
      "step": 1822
    },
    {
      "epoch": 0.1337527628236287,
      "grad_norm": 1.713934519254469,
      "learning_rate": 9.720368925761685e-06,
      "loss": 0.8693,
      "step": 1823
    },
    {
      "epoch": 0.1338261324137678,
      "grad_norm": 0.7135587627700343,
      "learning_rate": 9.719977003237364e-06,
      "loss": 0.5594,
      "step": 1824
    },
    {
      "epoch": 0.1338261324137678,
      "lm_loss": 0.5199,
      "step": 1824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1338261324137678,
      "lm_loss": 0.5736,
      "step": 1824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1338261324137678,
      "lm_loss": 0.4955,
      "step": 1824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1338261324137678,
      "lm_loss": 0.5438,
      "step": 1824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1338261324137678,
      "lm_loss": 0.5762,
      "step": 1824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1338261324137678,
      "lm_loss": 0.6348,
      "step": 1824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1338261324137678,
      "lm_loss": 0.5604,
      "step": 1824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1338261324137678,
      "lm_loss": 0.5014,
      "step": 1824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13389950200390693,
      "grad_norm": 0.6046730333630875,
      "learning_rate": 9.719584814164253e-06,
      "loss": 0.5674,
      "step": 1825
    },
    {
      "epoch": 0.13397287159404606,
      "grad_norm": 1.996427175833602,
      "learning_rate": 9.719192358564498e-06,
      "loss": 0.9603,
      "step": 1826
    },
    {
      "epoch": 0.13404624118418518,
      "grad_norm": 0.7221394238618203,
      "learning_rate": 9.718799636460265e-06,
      "loss": 0.5699,
      "step": 1827
    },
    {
      "epoch": 0.1341196107743243,
      "grad_norm": 1.6277878185471881,
      "learning_rate": 9.718406647873732e-06,
      "loss": 0.9223,
      "step": 1828
    },
    {
      "epoch": 0.13419298036446345,
      "grad_norm": 1.8552356991922245,
      "learning_rate": 9.718013392827087e-06,
      "loss": 0.8789,
      "step": 1829
    },
    {
      "epoch": 0.13426634995460257,
      "grad_norm": 0.7414056583391331,
      "learning_rate": 9.717619871342546e-06,
      "loss": 0.5653,
      "step": 1830
    },
    {
      "epoch": 0.1343397195447417,
      "grad_norm": 0.6733372863076712,
      "learning_rate": 9.717226083442325e-06,
      "loss": 0.5846,
      "step": 1831
    },
    {
      "epoch": 0.13441308913488081,
      "grad_norm": 0.6003870224552963,
      "learning_rate": 9.716832029148666e-06,
      "loss": 0.5676,
      "step": 1832
    },
    {
      "epoch": 0.13441308913488081,
      "lm_loss": 1.4898,
      "step": 1832,
      "vm_loss": 0.1467
    },
    {
      "epoch": 0.13441308913488081,
      "lm_loss": 1.3471,
      "step": 1832,
      "vm_loss": 0.0513
    },
    {
      "epoch": 0.13441308913488081,
      "lm_loss": 0.2077,
      "step": 1832,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.13441308913488081,
      "lm_loss": 0.6531,
      "step": 1832,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.13441308913488081,
      "lm_loss": 0.6781,
      "step": 1832,
      "vm_loss": 0.1047
    },
    {
      "epoch": 0.13441308913488081,
      "lm_loss": 0.5844,
      "step": 1832,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.13441308913488081,
      "lm_loss": 0.2392,
      "step": 1832,
      "vm_loss": 0.0943
    },
    {
      "epoch": 0.13441308913488081,
      "lm_loss": 1.0961,
      "step": 1832,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.13448645872501994,
      "grad_norm": 1.8326631477391793,
      "learning_rate": 9.71643770848382e-06,
      "loss": 0.9152,
      "step": 1833
    },
    {
      "epoch": 0.13455982831515909,
      "grad_norm": 1.8263658306236636,
      "learning_rate": 9.71604312147006e-06,
      "loss": 0.9022,
      "step": 1834
    },
    {
      "epoch": 0.1346331979052982,
      "grad_norm": 0.69196874578836,
      "learning_rate": 9.715648268129662e-06,
      "loss": 0.5658,
      "step": 1835
    },
    {
      "epoch": 0.13470656749543733,
      "grad_norm": 0.6611008981037432,
      "learning_rate": 9.715253148484931e-06,
      "loss": 0.5752,
      "step": 1836
    },
    {
      "epoch": 0.13477993708557645,
      "grad_norm": 1.720358638683923,
      "learning_rate": 9.714857762558177e-06,
      "loss": 0.8098,
      "step": 1837
    },
    {
      "epoch": 0.13485330667571557,
      "grad_norm": 2.142915566851824,
      "learning_rate": 9.714462110371728e-06,
      "loss": 0.8382,
      "step": 1838
    },
    {
      "epoch": 0.13492667626585472,
      "grad_norm": 1.8195052844272426,
      "learning_rate": 9.714066191947928e-06,
      "loss": 0.9816,
      "step": 1839
    },
    {
      "epoch": 0.13500004585599384,
      "grad_norm": 0.6552583898918154,
      "learning_rate": 9.713670007309137e-06,
      "loss": 0.5604,
      "step": 1840
    },
    {
      "epoch": 0.13500004585599384,
      "lm_loss": 0.7804,
      "step": 1840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13500004585599384,
      "lm_loss": 0.6356,
      "step": 1840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13500004585599384,
      "lm_loss": 0.5197,
      "step": 1840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13500004585599384,
      "lm_loss": 0.4734,
      "step": 1840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13500004585599384,
      "lm_loss": 0.6941,
      "step": 1840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13500004585599384,
      "lm_loss": 0.4926,
      "step": 1840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13500004585599384,
      "lm_loss": 0.6767,
      "step": 1840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13500004585599384,
      "lm_loss": 0.6626,
      "step": 1840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13507341544613297,
      "grad_norm": 0.6389651875925871,
      "learning_rate": 9.713273556477727e-06,
      "loss": 0.5758,
      "step": 1841
    },
    {
      "epoch": 0.1351467850362721,
      "grad_norm": 2.1272607782141755,
      "learning_rate": 9.712876839476088e-06,
      "loss": 0.9231,
      "step": 1842
    },
    {
      "epoch": 0.1352201546264112,
      "grad_norm": 2.03884055837719,
      "learning_rate": 9.712479856326621e-06,
      "loss": 0.8763,
      "step": 1843
    },
    {
      "epoch": 0.13529352421655033,
      "grad_norm": 1.9447950650698607,
      "learning_rate": 9.712082607051748e-06,
      "loss": 0.8992,
      "step": 1844
    },
    {
      "epoch": 0.13536689380668948,
      "grad_norm": 0.7101587609725167,
      "learning_rate": 9.7116850916739e-06,
      "loss": 0.5521,
      "step": 1845
    },
    {
      "epoch": 0.1354402633968286,
      "grad_norm": 1.9501239332605256,
      "learning_rate": 9.711287310215526e-06,
      "loss": 0.8173,
      "step": 1846
    },
    {
      "epoch": 0.13551363298696772,
      "grad_norm": 1.9165632854033987,
      "learning_rate": 9.710889262699092e-06,
      "loss": 1.0188,
      "step": 1847
    },
    {
      "epoch": 0.13558700257710685,
      "grad_norm": 0.613548666929407,
      "learning_rate": 9.710490949147073e-06,
      "loss": 0.5291,
      "step": 1848
    },
    {
      "epoch": 0.13558700257710685,
      "lm_loss": 0.7473,
      "step": 1848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13558700257710685,
      "lm_loss": 0.5974,
      "step": 1848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13558700257710685,
      "lm_loss": 0.4429,
      "step": 1848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13558700257710685,
      "lm_loss": 0.5092,
      "step": 1848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13558700257710685,
      "lm_loss": 0.5861,
      "step": 1848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13558700257710685,
      "lm_loss": 0.5596,
      "step": 1848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13558700257710685,
      "lm_loss": 0.46,
      "step": 1848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13558700257710685,
      "lm_loss": 0.6096,
      "step": 1848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13566037216724597,
      "grad_norm": 0.5911236797595352,
      "learning_rate": 9.710092369581966e-06,
      "loss": 0.5476,
      "step": 1849
    },
    {
      "epoch": 0.13573374175738512,
      "grad_norm": 1.7443022455170731,
      "learning_rate": 9.709693524026278e-06,
      "loss": 0.8351,
      "step": 1850
    },
    {
      "epoch": 0.13580711134752424,
      "grad_norm": 1.8043037941050757,
      "learning_rate": 9.709294412502535e-06,
      "loss": 0.9031,
      "step": 1851
    },
    {
      "epoch": 0.13588048093766336,
      "grad_norm": 1.6728790365018966,
      "learning_rate": 9.708895035033272e-06,
      "loss": 0.9584,
      "step": 1852
    },
    {
      "epoch": 0.13595385052780248,
      "grad_norm": 0.6694411069090813,
      "learning_rate": 9.708495391641045e-06,
      "loss": 0.5785,
      "step": 1853
    },
    {
      "epoch": 0.1360272201179416,
      "grad_norm": 1.684022598144045,
      "learning_rate": 9.708095482348423e-06,
      "loss": 0.8794,
      "step": 1854
    },
    {
      "epoch": 0.13610058970808075,
      "grad_norm": 0.6046384496258441,
      "learning_rate": 9.70769530717799e-06,
      "loss": 0.543,
      "step": 1855
    },
    {
      "epoch": 0.13617395929821988,
      "grad_norm": 0.5700805912919785,
      "learning_rate": 9.707294866152346e-06,
      "loss": 0.5408,
      "step": 1856
    },
    {
      "epoch": 0.13617395929821988,
      "lm_loss": 0.2052,
      "step": 1856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13617395929821988,
      "lm_loss": 0.545,
      "step": 1856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13617395929821988,
      "lm_loss": 0.5376,
      "step": 1856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13617395929821988,
      "lm_loss": 0.5185,
      "step": 1856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13617395929821988,
      "lm_loss": 0.6615,
      "step": 1856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13617395929821988,
      "lm_loss": 0.6479,
      "step": 1856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13617395929821988,
      "lm_loss": 0.5934,
      "step": 1856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13617395929821988,
      "lm_loss": 0.5682,
      "step": 1856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.136247328888359,
      "grad_norm": 0.5835767079920321,
      "learning_rate": 9.706894159294102e-06,
      "loss": 0.5512,
      "step": 1857
    },
    {
      "epoch": 0.13632069847849812,
      "grad_norm": 1.8993478227849563,
      "learning_rate": 9.706493186625889e-06,
      "loss": 0.8957,
      "step": 1858
    },
    {
      "epoch": 0.13639406806863724,
      "grad_norm": 1.864352267136974,
      "learning_rate": 9.70609194817035e-06,
      "loss": 0.9264,
      "step": 1859
    },
    {
      "epoch": 0.1364674376587764,
      "grad_norm": 1.60767895304473,
      "learning_rate": 9.705690443950144e-06,
      "loss": 0.916,
      "step": 1860
    },
    {
      "epoch": 0.1365408072489155,
      "grad_norm": 0.679123879298053,
      "learning_rate": 9.705288673987947e-06,
      "loss": 0.5488,
      "step": 1861
    },
    {
      "epoch": 0.13661417683905464,
      "grad_norm": 0.6053941893233915,
      "learning_rate": 9.704886638306444e-06,
      "loss": 0.5423,
      "step": 1862
    },
    {
      "epoch": 0.13668754642919376,
      "grad_norm": 1.9941898252737837,
      "learning_rate": 9.704484336928342e-06,
      "loss": 0.8863,
      "step": 1863
    },
    {
      "epoch": 0.13676091601933288,
      "grad_norm": 1.911546874671488,
      "learning_rate": 9.704081769876359e-06,
      "loss": 0.8941,
      "step": 1864
    },
    {
      "epoch": 0.13676091601933288,
      "lm_loss": 1.5268,
      "step": 1864,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.13676091601933288,
      "lm_loss": 1.6041,
      "step": 1864,
      "vm_loss": 0.1144
    },
    {
      "epoch": 0.13676091601933288,
      "lm_loss": 0.2564,
      "step": 1864,
      "vm_loss": 0.1827
    },
    {
      "epoch": 0.13676091601933288,
      "lm_loss": 0.6773,
      "step": 1864,
      "vm_loss": 0.2645
    },
    {
      "epoch": 0.13676091601933288,
      "lm_loss": 0.9125,
      "step": 1864,
      "vm_loss": 0.2558
    },
    {
      "epoch": 0.13676091601933288,
      "lm_loss": 0.7955,
      "step": 1864,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.13676091601933288,
      "lm_loss": 0.2178,
      "step": 1864,
      "vm_loss": 0.1472
    },
    {
      "epoch": 0.13676091601933288,
      "lm_loss": 0.8712,
      "step": 1864,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.136834285609472,
      "grad_norm": 1.8393129770272438,
      "learning_rate": 9.70367893717323e-06,
      "loss": 1.0081,
      "step": 1865
    },
    {
      "epoch": 0.13690765519961115,
      "grad_norm": 1.626354224482253,
      "learning_rate": 9.703275838841699e-06,
      "loss": 0.8523,
      "step": 1866
    },
    {
      "epoch": 0.13698102478975027,
      "grad_norm": 1.717219011561949,
      "learning_rate": 9.702872474904538e-06,
      "loss": 0.9419,
      "step": 1867
    },
    {
      "epoch": 0.1370543943798894,
      "grad_norm": 1.7061714339533895,
      "learning_rate": 9.702468845384521e-06,
      "loss": 0.8951,
      "step": 1868
    },
    {
      "epoch": 0.13712776397002852,
      "grad_norm": 1.735673862713503,
      "learning_rate": 9.702064950304442e-06,
      "loss": 0.9019,
      "step": 1869
    },
    {
      "epoch": 0.13720113356016764,
      "grad_norm": 1.7315121841062435,
      "learning_rate": 9.70166078968711e-06,
      "loss": 0.8253,
      "step": 1870
    },
    {
      "epoch": 0.1372745031503068,
      "grad_norm": 1.727862371193838,
      "learning_rate": 9.701256363555351e-06,
      "loss": 0.8801,
      "step": 1871
    },
    {
      "epoch": 0.1373478727404459,
      "grad_norm": 1.8781904642334275,
      "learning_rate": 9.700851671932004e-06,
      "loss": 0.9561,
      "step": 1872
    },
    {
      "epoch": 0.1373478727404459,
      "lm_loss": 0.6779,
      "step": 1872,
      "vm_loss": 0.0561
    },
    {
      "epoch": 0.1373478727404459,
      "lm_loss": 0.9854,
      "step": 1872,
      "vm_loss": 0.0943
    },
    {
      "epoch": 0.1373478727404459,
      "lm_loss": 1.1385,
      "step": 1872,
      "vm_loss": 0.1499
    },
    {
      "epoch": 0.1373478727404459,
      "lm_loss": 2.0017,
      "step": 1872,
      "vm_loss": 0.1866
    },
    {
      "epoch": 0.1373478727404459,
      "lm_loss": 0.7094,
      "step": 1872,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.1373478727404459,
      "lm_loss": 0.5953,
      "step": 1872,
      "vm_loss": 0.151
    },
    {
      "epoch": 0.1373478727404459,
      "lm_loss": 0.8457,
      "step": 1872,
      "vm_loss": 0.1234
    },
    {
      "epoch": 0.1373478727404459,
      "lm_loss": 0.0924,
      "step": 1872,
      "vm_loss": 0.1824
    },
    {
      "epoch": 0.13742124233058503,
      "grad_norm": 1.8547912301092393,
      "learning_rate": 9.700446714839919e-06,
      "loss": 0.8076,
      "step": 1873
    },
    {
      "epoch": 0.13749461192072415,
      "grad_norm": 1.7892407929898988,
      "learning_rate": 9.700041492301969e-06,
      "loss": 0.8702,
      "step": 1874
    },
    {
      "epoch": 0.13756798151086327,
      "grad_norm": 1.6272020341682694,
      "learning_rate": 9.699636004341035e-06,
      "loss": 0.8795,
      "step": 1875
    },
    {
      "epoch": 0.13764135110100242,
      "grad_norm": 1.686719868180616,
      "learning_rate": 9.69923025098002e-06,
      "loss": 0.8912,
      "step": 1876
    },
    {
      "epoch": 0.13771472069114155,
      "grad_norm": 0.9636979035191301,
      "learning_rate": 9.698824232241833e-06,
      "loss": 0.5735,
      "step": 1877
    },
    {
      "epoch": 0.13778809028128067,
      "grad_norm": 1.8563742596444666,
      "learning_rate": 9.698417948149406e-06,
      "loss": 0.9348,
      "step": 1878
    },
    {
      "epoch": 0.1378614598714198,
      "grad_norm": 1.751624704230181,
      "learning_rate": 9.698011398725682e-06,
      "loss": 0.8356,
      "step": 1879
    },
    {
      "epoch": 0.1379348294615589,
      "grad_norm": 1.480909702852618,
      "learning_rate": 9.69760458399362e-06,
      "loss": 0.912,
      "step": 1880
    },
    {
      "epoch": 0.1379348294615589,
      "lm_loss": 0.6814,
      "step": 1880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1379348294615589,
      "lm_loss": 0.5467,
      "step": 1880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1379348294615589,
      "lm_loss": 0.6074,
      "step": 1880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1379348294615589,
      "lm_loss": 0.7056,
      "step": 1880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1379348294615589,
      "lm_loss": 0.5639,
      "step": 1880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1379348294615589,
      "lm_loss": 0.665,
      "step": 1880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1379348294615589,
      "lm_loss": 0.6933,
      "step": 1880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1379348294615589,
      "lm_loss": 0.5354,
      "step": 1880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.13800819905169806,
      "grad_norm": 0.7151983634708997,
      "learning_rate": 9.697197503976195e-06,
      "loss": 0.5812,
      "step": 1881
    },
    {
      "epoch": 0.13808156864183718,
      "grad_norm": 0.719910567355689,
      "learning_rate": 9.696790158696393e-06,
      "loss": 0.5769,
      "step": 1882
    },
    {
      "epoch": 0.1381549382319763,
      "grad_norm": 1.7202563098478763,
      "learning_rate": 9.69638254817722e-06,
      "loss": 0.8513,
      "step": 1883
    },
    {
      "epoch": 0.13822830782211543,
      "grad_norm": 1.6070421053380954,
      "learning_rate": 9.695974672441692e-06,
      "loss": 0.8511,
      "step": 1884
    },
    {
      "epoch": 0.13830167741225455,
      "grad_norm": 1.7984860007591439,
      "learning_rate": 9.695566531512849e-06,
      "loss": 0.9814,
      "step": 1885
    },
    {
      "epoch": 0.13837504700239367,
      "grad_norm": 1.7252824929617754,
      "learning_rate": 9.695158125413731e-06,
      "loss": 0.9937,
      "step": 1886
    },
    {
      "epoch": 0.13844841659253282,
      "grad_norm": 2.0162617834757195,
      "learning_rate": 9.694749454167409e-06,
      "loss": 0.9676,
      "step": 1887
    },
    {
      "epoch": 0.13852178618267194,
      "grad_norm": 1.7903099790738584,
      "learning_rate": 9.694340517796959e-06,
      "loss": 0.9179,
      "step": 1888
    },
    {
      "epoch": 0.13852178618267194,
      "lm_loss": 1.4243,
      "step": 1888,
      "vm_loss": 0.0695
    },
    {
      "epoch": 0.13852178618267194,
      "lm_loss": 1.9633,
      "step": 1888,
      "vm_loss": 0.1715
    },
    {
      "epoch": 0.13852178618267194,
      "lm_loss": 0.4953,
      "step": 1888,
      "vm_loss": 0.1747
    },
    {
      "epoch": 0.13852178618267194,
      "lm_loss": 0.5224,
      "step": 1888,
      "vm_loss": 0.209
    },
    {
      "epoch": 0.13852178618267194,
      "lm_loss": 0.472,
      "step": 1888,
      "vm_loss": 0.0858
    },
    {
      "epoch": 0.13852178618267194,
      "lm_loss": 0.658,
      "step": 1888,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.13852178618267194,
      "lm_loss": 1.0733,
      "step": 1888,
      "vm_loss": 0.1515
    },
    {
      "epoch": 0.13852178618267194,
      "lm_loss": 0.1875,
      "step": 1888,
      "vm_loss": 0.179
    },
    {
      "epoch": 0.13859515577281106,
      "grad_norm": 1.6836535050293109,
      "learning_rate": 9.693931316325473e-06,
      "loss": 0.9283,
      "step": 1889
    },
    {
      "epoch": 0.13866852536295018,
      "grad_norm": 1.6898494021920927,
      "learning_rate": 9.693521849776062e-06,
      "loss": 0.8888,
      "step": 1890
    },
    {
      "epoch": 0.1387418949530893,
      "grad_norm": 1.8671176873063116,
      "learning_rate": 9.69311211817185e-06,
      "loss": 0.9697,
      "step": 1891
    },
    {
      "epoch": 0.13881526454322846,
      "grad_norm": 1.7210937909045287,
      "learning_rate": 9.692702121535974e-06,
      "loss": 0.8704,
      "step": 1892
    },
    {
      "epoch": 0.13888863413336758,
      "grad_norm": 1.727819541731709,
      "learning_rate": 9.692291859891586e-06,
      "loss": 0.878,
      "step": 1893
    },
    {
      "epoch": 0.1389620037235067,
      "grad_norm": 1.7491326061016894,
      "learning_rate": 9.691881333261858e-06,
      "loss": 0.8767,
      "step": 1894
    },
    {
      "epoch": 0.13903537331364582,
      "grad_norm": 1.6735437143247893,
      "learning_rate": 9.691470541669972e-06,
      "loss": 0.9268,
      "step": 1895
    },
    {
      "epoch": 0.13910874290378494,
      "grad_norm": 1.760624275157063,
      "learning_rate": 9.691059485139124e-06,
      "loss": 0.838,
      "step": 1896
    },
    {
      "epoch": 0.13910874290378494,
      "lm_loss": 0.9263,
      "step": 1896,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.13910874290378494,
      "lm_loss": 1.1995,
      "step": 1896,
      "vm_loss": 0.2106
    },
    {
      "epoch": 0.13910874290378494,
      "lm_loss": 0.2658,
      "step": 1896,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.13910874290378494,
      "lm_loss": 0.4143,
      "step": 1896,
      "vm_loss": 0.1431
    },
    {
      "epoch": 0.13910874290378494,
      "lm_loss": 0.6168,
      "step": 1896,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.13910874290378494,
      "lm_loss": 1.1045,
      "step": 1896,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.13910874290378494,
      "lm_loss": 1.3183,
      "step": 1896,
      "vm_loss": 0.1347
    },
    {
      "epoch": 0.13910874290378494,
      "lm_loss": 0.1123,
      "step": 1896,
      "vm_loss": 0.0651
    },
    {
      "epoch": 0.1391821124939241,
      "grad_norm": 2.029078118541505,
      "learning_rate": 9.690648163692534e-06,
      "loss": 0.8535,
      "step": 1897
    },
    {
      "epoch": 0.13925548208406321,
      "grad_norm": 1.7348143208432534,
      "learning_rate": 9.690236577353423e-06,
      "loss": 0.9359,
      "step": 1898
    },
    {
      "epoch": 0.13932885167420234,
      "grad_norm": 0.9993534081213826,
      "learning_rate": 9.689824726145038e-06,
      "loss": 0.5838,
      "step": 1899
    },
    {
      "epoch": 0.13940222126434146,
      "grad_norm": 1.7585057115315825,
      "learning_rate": 9.689412610090636e-06,
      "loss": 0.8606,
      "step": 1900
    },
    {
      "epoch": 0.13947559085448058,
      "grad_norm": 1.8496872707532306,
      "learning_rate": 9.68900022921349e-06,
      "loss": 0.9045,
      "step": 1901
    },
    {
      "epoch": 0.13954896044461973,
      "grad_norm": 1.66991649618685,
      "learning_rate": 9.688587583536891e-06,
      "loss": 0.8323,
      "step": 1902
    },
    {
      "epoch": 0.13962233003475885,
      "grad_norm": 1.6919493616200159,
      "learning_rate": 9.688174673084139e-06,
      "loss": 1.0386,
      "step": 1903
    },
    {
      "epoch": 0.13969569962489797,
      "grad_norm": 1.5712410971032522,
      "learning_rate": 9.687761497878554e-06,
      "loss": 0.8871,
      "step": 1904
    },
    {
      "epoch": 0.13969569962489797,
      "lm_loss": 1.8093,
      "step": 1904,
      "vm_loss": 0.0453
    },
    {
      "epoch": 0.13969569962489797,
      "lm_loss": 1.265,
      "step": 1904,
      "vm_loss": 0.2084
    },
    {
      "epoch": 0.13969569962489797,
      "lm_loss": 0.4709,
      "step": 1904,
      "vm_loss": 0.142
    },
    {
      "epoch": 0.13969569962489797,
      "lm_loss": 0.7798,
      "step": 1904,
      "vm_loss": 0.0996
    },
    {
      "epoch": 0.13969569962489797,
      "lm_loss": 1.8631,
      "step": 1904,
      "vm_loss": 0.2036
    },
    {
      "epoch": 0.13969569962489797,
      "lm_loss": 0.3389,
      "step": 1904,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.13969569962489797,
      "lm_loss": 1.1757,
      "step": 1904,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.13969569962489797,
      "lm_loss": 0.8843,
      "step": 1904,
      "vm_loss": 0.2115
    },
    {
      "epoch": 0.1397690692150371,
      "grad_norm": 1.7259660800076146,
      "learning_rate": 9.687348057943468e-06,
      "loss": 0.9206,
      "step": 1905
    },
    {
      "epoch": 0.13984243880517622,
      "grad_norm": 1.772648697754453,
      "learning_rate": 9.686934353302228e-06,
      "loss": 0.8539,
      "step": 1906
    },
    {
      "epoch": 0.13991580839531534,
      "grad_norm": 1.4052442057463506,
      "learning_rate": 9.6865203839782e-06,
      "loss": 0.7055,
      "step": 1907
    },
    {
      "epoch": 0.1399891779854545,
      "grad_norm": 1.712072189417456,
      "learning_rate": 9.686106149994758e-06,
      "loss": 0.9364,
      "step": 1908
    },
    {
      "epoch": 0.1400625475755936,
      "grad_norm": 1.825185470707526,
      "learning_rate": 9.685691651375297e-06,
      "loss": 0.8682,
      "step": 1909
    },
    {
      "epoch": 0.14013591716573273,
      "grad_norm": 1.5530604475174399,
      "learning_rate": 9.685276888143225e-06,
      "loss": 0.8656,
      "step": 1910
    },
    {
      "epoch": 0.14020928675587185,
      "grad_norm": 2.050933252355528,
      "learning_rate": 9.684861860321965e-06,
      "loss": 0.9069,
      "step": 1911
    },
    {
      "epoch": 0.14028265634601098,
      "grad_norm": 0.9651446190539754,
      "learning_rate": 9.684446567934953e-06,
      "loss": 0.5599,
      "step": 1912
    },
    {
      "epoch": 0.14028265634601098,
      "lm_loss": 1.0654,
      "step": 1912,
      "vm_loss": 0.1729
    },
    {
      "epoch": 0.14028265634601098,
      "lm_loss": 1.3012,
      "step": 1912,
      "vm_loss": 0.0436
    },
    {
      "epoch": 0.14028265634601098,
      "lm_loss": 0.3147,
      "step": 1912,
      "vm_loss": 0.1383
    },
    {
      "epoch": 0.14028265634601098,
      "lm_loss": 0.6989,
      "step": 1912,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.14028265634601098,
      "lm_loss": 0.8155,
      "step": 1912,
      "vm_loss": 0.129
    },
    {
      "epoch": 0.14028265634601098,
      "lm_loss": 0.4323,
      "step": 1912,
      "vm_loss": 0.1707
    },
    {
      "epoch": 0.14028265634601098,
      "lm_loss": 0.6407,
      "step": 1912,
      "vm_loss": 0.1515
    },
    {
      "epoch": 0.14028265634601098,
      "lm_loss": 0.3057,
      "step": 1912,
      "vm_loss": 0.0717
    },
    {
      "epoch": 0.14035602593615013,
      "grad_norm": 2.0675256607008503,
      "learning_rate": 9.684031011005642e-06,
      "loss": 0.8533,
      "step": 1913
    },
    {
      "epoch": 0.14042939552628925,
      "grad_norm": 0.7146643659179378,
      "learning_rate": 9.6836151895575e-06,
      "loss": 0.5624,
      "step": 1914
    },
    {
      "epoch": 0.14050276511642837,
      "grad_norm": 1.6326792341517407,
      "learning_rate": 9.68319910361401e-06,
      "loss": 0.8076,
      "step": 1915
    },
    {
      "epoch": 0.1405761347065675,
      "grad_norm": 1.8685957687504409,
      "learning_rate": 9.68278275319867e-06,
      "loss": 0.9531,
      "step": 1916
    },
    {
      "epoch": 0.1406495042967066,
      "grad_norm": 1.635623727221587,
      "learning_rate": 9.68236613833499e-06,
      "loss": 0.8701,
      "step": 1917
    },
    {
      "epoch": 0.14072287388684576,
      "grad_norm": 0.9169613313376077,
      "learning_rate": 9.681949259046498e-06,
      "loss": 0.5766,
      "step": 1918
    },
    {
      "epoch": 0.14079624347698488,
      "grad_norm": 1.6697515943423054,
      "learning_rate": 9.681532115356737e-06,
      "loss": 0.9036,
      "step": 1919
    },
    {
      "epoch": 0.140869613067124,
      "grad_norm": 0.76134765121781,
      "learning_rate": 9.681114707289265e-06,
      "loss": 0.5791,
      "step": 1920
    },
    {
      "epoch": 0.140869613067124,
      "lm_loss": 1.8053,
      "step": 1920,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.140869613067124,
      "lm_loss": 0.6322,
      "step": 1920,
      "vm_loss": 0.1451
    },
    {
      "epoch": 0.140869613067124,
      "lm_loss": 0.2369,
      "step": 1920,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.140869613067124,
      "lm_loss": 0.7433,
      "step": 1920,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.140869613067124,
      "lm_loss": 0.4508,
      "step": 1920,
      "vm_loss": 0.1809
    },
    {
      "epoch": 0.140869613067124,
      "lm_loss": 0.8645,
      "step": 1920,
      "vm_loss": 0.1539
    },
    {
      "epoch": 0.140869613067124,
      "lm_loss": 0.8442,
      "step": 1920,
      "vm_loss": 0.1943
    },
    {
      "epoch": 0.140869613067124,
      "lm_loss": 0.1038,
      "step": 1920,
      "vm_loss": 0.2138
    },
    {
      "epoch": 0.14094298265726313,
      "grad_norm": 1.8191750557141684,
      "learning_rate": 9.680697034867653e-06,
      "loss": 0.8697,
      "step": 1921
    },
    {
      "epoch": 0.14101635224740225,
      "grad_norm": 1.9427232941339885,
      "learning_rate": 9.680279098115486e-06,
      "loss": 0.9535,
      "step": 1922
    },
    {
      "epoch": 0.1410897218375414,
      "grad_norm": 1.9001255122174676,
      "learning_rate": 9.679860897056368e-06,
      "loss": 0.9441,
      "step": 1923
    },
    {
      "epoch": 0.14116309142768052,
      "grad_norm": 1.7713508737927666,
      "learning_rate": 9.679442431713917e-06,
      "loss": 0.8699,
      "step": 1924
    },
    {
      "epoch": 0.14123646101781964,
      "grad_norm": 1.7604318539312558,
      "learning_rate": 9.679023702111762e-06,
      "loss": 0.8867,
      "step": 1925
    },
    {
      "epoch": 0.14130983060795876,
      "grad_norm": 1.7336031073746967,
      "learning_rate": 9.678604708273553e-06,
      "loss": 0.8263,
      "step": 1926
    },
    {
      "epoch": 0.14138320019809789,
      "grad_norm": 1.6056156384365279,
      "learning_rate": 9.678185450222947e-06,
      "loss": 0.9845,
      "step": 1927
    },
    {
      "epoch": 0.141456569788237,
      "grad_norm": 1.4667714924106414,
      "learning_rate": 9.677765927983627e-06,
      "loss": 0.792,
      "step": 1928
    },
    {
      "epoch": 0.141456569788237,
      "lm_loss": 0.3246,
      "step": 1928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.141456569788237,
      "lm_loss": 0.5359,
      "step": 1928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.141456569788237,
      "lm_loss": 0.5865,
      "step": 1928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.141456569788237,
      "lm_loss": 0.5444,
      "step": 1928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.141456569788237,
      "lm_loss": 0.5407,
      "step": 1928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.141456569788237,
      "lm_loss": 0.6828,
      "step": 1928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.141456569788237,
      "lm_loss": 0.519,
      "step": 1928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.141456569788237,
      "lm_loss": 0.4677,
      "step": 1928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14152993937837616,
      "grad_norm": 0.9211084814113375,
      "learning_rate": 9.677346141579277e-06,
      "loss": 0.552,
      "step": 1929
    },
    {
      "epoch": 0.14160330896851528,
      "grad_norm": 2.253084890173707,
      "learning_rate": 9.676926091033609e-06,
      "loss": 0.9914,
      "step": 1930
    },
    {
      "epoch": 0.1416766785586544,
      "grad_norm": 0.7409121140429219,
      "learning_rate": 9.676505776370343e-06,
      "loss": 0.5877,
      "step": 1931
    },
    {
      "epoch": 0.14175004814879352,
      "grad_norm": 0.6444649328314643,
      "learning_rate": 9.676085197613215e-06,
      "loss": 0.5668,
      "step": 1932
    },
    {
      "epoch": 0.14182341773893264,
      "grad_norm": 2.1429447682275105,
      "learning_rate": 9.675664354785973e-06,
      "loss": 0.8677,
      "step": 1933
    },
    {
      "epoch": 0.1418967873290718,
      "grad_norm": 1.836899634934021,
      "learning_rate": 9.67524324791239e-06,
      "loss": 0.8609,
      "step": 1934
    },
    {
      "epoch": 0.14197015691921092,
      "grad_norm": 1.7448600224519848,
      "learning_rate": 9.674821877016238e-06,
      "loss": 0.8798,
      "step": 1935
    },
    {
      "epoch": 0.14204352650935004,
      "grad_norm": 1.0887047263759826,
      "learning_rate": 9.67440024212132e-06,
      "loss": 0.5437,
      "step": 1936
    },
    {
      "epoch": 0.14204352650935004,
      "lm_loss": 0.6775,
      "step": 1936,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.14204352650935004,
      "lm_loss": 1.5217,
      "step": 1936,
      "vm_loss": 0.1762
    },
    {
      "epoch": 0.14204352650935004,
      "lm_loss": 1.0437,
      "step": 1936,
      "vm_loss": 0.1391
    },
    {
      "epoch": 0.14204352650935004,
      "lm_loss": 1.6211,
      "step": 1936,
      "vm_loss": 0.146
    },
    {
      "epoch": 0.14204352650935004,
      "lm_loss": 0.9486,
      "step": 1936,
      "vm_loss": 0.2417
    },
    {
      "epoch": 0.14204352650935004,
      "lm_loss": 0.8988,
      "step": 1936,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.14204352650935004,
      "lm_loss": 0.2615,
      "step": 1936,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.14204352650935004,
      "lm_loss": 1.0913,
      "step": 1936,
      "vm_loss": 0.165
    },
    {
      "epoch": 0.14211689609948916,
      "grad_norm": 1.737182523345536,
      "learning_rate": 9.673978343251444e-06,
      "loss": 0.995,
      "step": 1937
    },
    {
      "epoch": 0.14219026568962828,
      "grad_norm": 1.8043757017459252,
      "learning_rate": 9.673556180430437e-06,
      "loss": 0.9176,
      "step": 1938
    },
    {
      "epoch": 0.14226363527976743,
      "grad_norm": 2.023756127165242,
      "learning_rate": 9.673133753682138e-06,
      "loss": 0.9766,
      "step": 1939
    },
    {
      "epoch": 0.14233700486990655,
      "grad_norm": 1.9582907232283364,
      "learning_rate": 9.672711063030403e-06,
      "loss": 0.9059,
      "step": 1940
    },
    {
      "epoch": 0.14241037446004567,
      "grad_norm": 0.6584392662823566,
      "learning_rate": 9.672288108499104e-06,
      "loss": 0.5506,
      "step": 1941
    },
    {
      "epoch": 0.1424837440501848,
      "grad_norm": 1.811661343439856,
      "learning_rate": 9.671864890112123e-06,
      "loss": 0.8449,
      "step": 1942
    },
    {
      "epoch": 0.14255711364032392,
      "grad_norm": 0.6675348133799498,
      "learning_rate": 9.671441407893362e-06,
      "loss": 0.5494,
      "step": 1943
    },
    {
      "epoch": 0.14263048323046307,
      "grad_norm": 1.6543499521778806,
      "learning_rate": 9.671017661866736e-06,
      "loss": 0.85,
      "step": 1944
    },
    {
      "epoch": 0.14263048323046307,
      "lm_loss": 1.2076,
      "step": 1944,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.14263048323046307,
      "lm_loss": 0.0338,
      "step": 1944,
      "vm_loss": 0.0785
    },
    {
      "epoch": 0.14263048323046307,
      "lm_loss": 1.5264,
      "step": 1944,
      "vm_loss": 0.2235
    },
    {
      "epoch": 0.14263048323046307,
      "lm_loss": 1.0262,
      "step": 1944,
      "vm_loss": 0.139
    },
    {
      "epoch": 0.14263048323046307,
      "lm_loss": 0.6731,
      "step": 1944,
      "vm_loss": 0.1879
    },
    {
      "epoch": 0.14263048323046307,
      "lm_loss": 0.9852,
      "step": 1944,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.14263048323046307,
      "lm_loss": 0.9302,
      "step": 1944,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.14263048323046307,
      "lm_loss": 0.5975,
      "step": 1944,
      "vm_loss": 0.1384
    },
    {
      "epoch": 0.1427038528206022,
      "grad_norm": 1.7724390289528134,
      "learning_rate": 9.670593652056176e-06,
      "loss": 0.9622,
      "step": 1945
    },
    {
      "epoch": 0.1427772224107413,
      "grad_norm": 1.8524893551248556,
      "learning_rate": 9.670169378485625e-06,
      "loss": 0.8704,
      "step": 1946
    },
    {
      "epoch": 0.14285059200088043,
      "grad_norm": 1.677639494555786,
      "learning_rate": 9.669744841179044e-06,
      "loss": 0.8634,
      "step": 1947
    },
    {
      "epoch": 0.14292396159101955,
      "grad_norm": 1.561932316764045,
      "learning_rate": 9.669320040160407e-06,
      "loss": 0.8686,
      "step": 1948
    },
    {
      "epoch": 0.14299733118115868,
      "grad_norm": 1.9273686394489844,
      "learning_rate": 9.668894975453705e-06,
      "loss": 0.9515,
      "step": 1949
    },
    {
      "epoch": 0.14307070077129783,
      "grad_norm": 1.6740312394985286,
      "learning_rate": 9.66846964708294e-06,
      "loss": 0.8907,
      "step": 1950
    },
    {
      "epoch": 0.14314407036143695,
      "grad_norm": 1.6514138465341524,
      "learning_rate": 9.668044055072134e-06,
      "loss": 0.8349,
      "step": 1951
    },
    {
      "epoch": 0.14321743995157607,
      "grad_norm": 1.815617032650084,
      "learning_rate": 9.66761819944532e-06,
      "loss": 1.0067,
      "step": 1952
    },
    {
      "epoch": 0.14321743995157607,
      "lm_loss": 0.462,
      "step": 1952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14321743995157607,
      "lm_loss": 0.6968,
      "step": 1952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14321743995157607,
      "lm_loss": 0.7092,
      "step": 1952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14321743995157607,
      "lm_loss": 0.8372,
      "step": 1952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14321743995157607,
      "lm_loss": 0.6145,
      "step": 1952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14321743995157607,
      "lm_loss": 0.5233,
      "step": 1952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14321743995157607,
      "lm_loss": 0.4319,
      "step": 1952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14321743995157607,
      "lm_loss": 0.6464,
      "step": 1952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1432908095417152,
      "grad_norm": 0.9284326898774319,
      "learning_rate": 9.667192080226547e-06,
      "loss": 0.5638,
      "step": 1953
    },
    {
      "epoch": 0.1433641791318543,
      "grad_norm": 1.5867433481538022,
      "learning_rate": 9.666765697439878e-06,
      "loss": 0.8426,
      "step": 1954
    },
    {
      "epoch": 0.14343754872199346,
      "grad_norm": 1.7434908381955887,
      "learning_rate": 9.666339051109395e-06,
      "loss": 0.9293,
      "step": 1955
    },
    {
      "epoch": 0.14351091831213258,
      "grad_norm": 1.6756023918424754,
      "learning_rate": 9.665912141259188e-06,
      "loss": 0.8437,
      "step": 1956
    },
    {
      "epoch": 0.1435842879022717,
      "grad_norm": 1.4831383595482914,
      "learning_rate": 9.665484967913369e-06,
      "loss": 0.8254,
      "step": 1957
    },
    {
      "epoch": 0.14365765749241083,
      "grad_norm": 1.6828740965377629,
      "learning_rate": 9.665057531096059e-06,
      "loss": 0.9663,
      "step": 1958
    },
    {
      "epoch": 0.14373102708254995,
      "grad_norm": 1.7448496087908372,
      "learning_rate": 9.664629830831396e-06,
      "loss": 0.8061,
      "step": 1959
    },
    {
      "epoch": 0.1438043966726891,
      "grad_norm": 0.6767573215446182,
      "learning_rate": 9.664201867143536e-06,
      "loss": 0.569,
      "step": 1960
    },
    {
      "epoch": 0.1438043966726891,
      "lm_loss": 0.6498,
      "step": 1960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1438043966726891,
      "lm_loss": 0.4433,
      "step": 1960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1438043966726891,
      "lm_loss": 0.6219,
      "step": 1960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1438043966726891,
      "lm_loss": 0.5471,
      "step": 1960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1438043966726891,
      "lm_loss": 0.5754,
      "step": 1960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1438043966726891,
      "lm_loss": 0.6312,
      "step": 1960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1438043966726891,
      "lm_loss": 0.6049,
      "step": 1960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1438043966726891,
      "lm_loss": 0.5942,
      "step": 1960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14387776626282822,
      "grad_norm": 0.6880766036604182,
      "learning_rate": 9.663773640056647e-06,
      "loss": 0.5521,
      "step": 1961
    },
    {
      "epoch": 0.14395113585296734,
      "grad_norm": 1.7887382178867022,
      "learning_rate": 9.66334514959491e-06,
      "loss": 0.952,
      "step": 1962
    },
    {
      "epoch": 0.14402450544310647,
      "grad_norm": 0.6108864348975322,
      "learning_rate": 9.662916395782522e-06,
      "loss": 0.5955,
      "step": 1963
    },
    {
      "epoch": 0.1440978750332456,
      "grad_norm": 1.817462278130477,
      "learning_rate": 9.662487378643701e-06,
      "loss": 0.8785,
      "step": 1964
    },
    {
      "epoch": 0.1441712446233847,
      "grad_norm": 1.528790519775528,
      "learning_rate": 9.66205809820267e-06,
      "loss": 0.9431,
      "step": 1965
    },
    {
      "epoch": 0.14424461421352386,
      "grad_norm": 0.6368778487729809,
      "learning_rate": 9.66162855448367e-06,
      "loss": 0.5567,
      "step": 1966
    },
    {
      "epoch": 0.14431798380366298,
      "grad_norm": 1.7405896125882878,
      "learning_rate": 9.661198747510965e-06,
      "loss": 0.8599,
      "step": 1967
    },
    {
      "epoch": 0.1443913533938021,
      "grad_norm": 1.6423756835330725,
      "learning_rate": 9.660768677308822e-06,
      "loss": 0.8536,
      "step": 1968
    },
    {
      "epoch": 0.1443913533938021,
      "lm_loss": 0.8435,
      "step": 1968,
      "vm_loss": 0.0438
    },
    {
      "epoch": 0.1443913533938021,
      "lm_loss": 0.4275,
      "step": 1968,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.1443913533938021,
      "lm_loss": 1.2882,
      "step": 1968,
      "vm_loss": 0.0446
    },
    {
      "epoch": 0.1443913533938021,
      "lm_loss": 0.7496,
      "step": 1968,
      "vm_loss": 0.0543
    },
    {
      "epoch": 0.1443913533938021,
      "lm_loss": 0.4036,
      "step": 1968,
      "vm_loss": 0.2295
    },
    {
      "epoch": 0.1443913533938021,
      "lm_loss": 0.6999,
      "step": 1968,
      "vm_loss": 0.1405
    },
    {
      "epoch": 0.1443913533938021,
      "lm_loss": 0.1143,
      "step": 1968,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.1443913533938021,
      "lm_loss": 0.0707,
      "step": 1968,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.14446472298394122,
      "grad_norm": 1.6324406144503552,
      "learning_rate": 9.66033834390153e-06,
      "loss": 0.7121,
      "step": 1969
    },
    {
      "epoch": 0.14453809257408035,
      "grad_norm": 0.6186928307580151,
      "learning_rate": 9.659907747313391e-06,
      "loss": 0.5869,
      "step": 1970
    },
    {
      "epoch": 0.1446114621642195,
      "grad_norm": 1.7804781894204267,
      "learning_rate": 9.65947688756872e-06,
      "loss": 0.8997,
      "step": 1971
    },
    {
      "epoch": 0.14468483175435862,
      "grad_norm": 1.7034236752082008,
      "learning_rate": 9.65904576469185e-06,
      "loss": 0.972,
      "step": 1972
    },
    {
      "epoch": 0.14475820134449774,
      "grad_norm": 0.6235043389291263,
      "learning_rate": 9.658614378707129e-06,
      "loss": 0.579,
      "step": 1973
    },
    {
      "epoch": 0.14483157093463686,
      "grad_norm": 1.6963588876643376,
      "learning_rate": 9.658182729638917e-06,
      "loss": 0.8618,
      "step": 1974
    },
    {
      "epoch": 0.14490494052477598,
      "grad_norm": 1.503395883753311,
      "learning_rate": 9.65775081751159e-06,
      "loss": 0.8211,
      "step": 1975
    },
    {
      "epoch": 0.14497831011491513,
      "grad_norm": 1.8762194089318833,
      "learning_rate": 9.65731864234954e-06,
      "loss": 0.9074,
      "step": 1976
    },
    {
      "epoch": 0.14497831011491513,
      "lm_loss": 0.6131,
      "step": 1976,
      "vm_loss": 0.0598
    },
    {
      "epoch": 0.14497831011491513,
      "lm_loss": 1.0715,
      "step": 1976,
      "vm_loss": 0.0852
    },
    {
      "epoch": 0.14497831011491513,
      "lm_loss": 0.7823,
      "step": 1976,
      "vm_loss": 0.1973
    },
    {
      "epoch": 0.14497831011491513,
      "lm_loss": 1.0979,
      "step": 1976,
      "vm_loss": 0.1665
    },
    {
      "epoch": 0.14497831011491513,
      "lm_loss": 0.7806,
      "step": 1976,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.14497831011491513,
      "lm_loss": 0.3488,
      "step": 1976,
      "vm_loss": 0.1424
    },
    {
      "epoch": 0.14497831011491513,
      "lm_loss": 0.7862,
      "step": 1976,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.14497831011491513,
      "lm_loss": 0.9737,
      "step": 1976,
      "vm_loss": 0.0946
    },
    {
      "epoch": 0.14505167970505425,
      "grad_norm": 1.5321360287844588,
      "learning_rate": 9.656886204177172e-06,
      "loss": 0.8421,
      "step": 1977
    },
    {
      "epoch": 0.14512504929519338,
      "grad_norm": 1.7700936544754986,
      "learning_rate": 9.65645350301891e-06,
      "loss": 0.9242,
      "step": 1978
    },
    {
      "epoch": 0.1451984188853325,
      "grad_norm": 0.641882572899425,
      "learning_rate": 9.656020538899183e-06,
      "loss": 0.5647,
      "step": 1979
    },
    {
      "epoch": 0.14527178847547162,
      "grad_norm": 0.6275711383927448,
      "learning_rate": 9.65558731184245e-06,
      "loss": 0.5523,
      "step": 1980
    },
    {
      "epoch": 0.14534515806561077,
      "grad_norm": 0.5999983819743665,
      "learning_rate": 9.655153821873169e-06,
      "loss": 0.5752,
      "step": 1981
    },
    {
      "epoch": 0.1454185276557499,
      "grad_norm": 1.777925785337175,
      "learning_rate": 9.654720069015823e-06,
      "loss": 0.8451,
      "step": 1982
    },
    {
      "epoch": 0.145491897245889,
      "grad_norm": 1.6779895530928985,
      "learning_rate": 9.65428605329491e-06,
      "loss": 0.7982,
      "step": 1983
    },
    {
      "epoch": 0.14556526683602813,
      "grad_norm": 0.6236243916199918,
      "learning_rate": 9.653851774734936e-06,
      "loss": 0.5533,
      "step": 1984
    },
    {
      "epoch": 0.14556526683602813,
      "lm_loss": 1.0853,
      "step": 1984,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.14556526683602813,
      "lm_loss": 0.4786,
      "step": 1984,
      "vm_loss": 0.0889
    },
    {
      "epoch": 0.14556526683602813,
      "lm_loss": 0.3926,
      "step": 1984,
      "vm_loss": 0.1188
    },
    {
      "epoch": 0.14556526683602813,
      "lm_loss": 0.1263,
      "step": 1984,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.14556526683602813,
      "lm_loss": 0.5992,
      "step": 1984,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.14556526683602813,
      "lm_loss": 0.8789,
      "step": 1984,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.14556526683602813,
      "lm_loss": 0.7675,
      "step": 1984,
      "vm_loss": 0.0445
    },
    {
      "epoch": 0.14556526683602813,
      "lm_loss": 0.739,
      "step": 1984,
      "vm_loss": 0.0461
    },
    {
      "epoch": 0.14563863642616726,
      "grad_norm": 1.6625453721904133,
      "learning_rate": 9.653417233360428e-06,
      "loss": 0.8765,
      "step": 1985
    },
    {
      "epoch": 0.14571200601630638,
      "grad_norm": 0.6087431392099135,
      "learning_rate": 9.652982429195922e-06,
      "loss": 0.5719,
      "step": 1986
    },
    {
      "epoch": 0.14578537560644553,
      "grad_norm": 1.6393262094532859,
      "learning_rate": 9.65254736226598e-06,
      "loss": 0.8588,
      "step": 1987
    },
    {
      "epoch": 0.14585874519658465,
      "grad_norm": 1.5873166659366598,
      "learning_rate": 9.652112032595164e-06,
      "loss": 0.9558,
      "step": 1988
    },
    {
      "epoch": 0.14593211478672377,
      "grad_norm": 1.6448170688608148,
      "learning_rate": 9.65167644020806e-06,
      "loss": 0.7821,
      "step": 1989
    },
    {
      "epoch": 0.1460054843768629,
      "grad_norm": 0.5660691500788994,
      "learning_rate": 9.651240585129268e-06,
      "loss": 0.5478,
      "step": 1990
    },
    {
      "epoch": 0.14607885396700201,
      "grad_norm": 1.7786989298535956,
      "learning_rate": 9.650804467383404e-06,
      "loss": 0.9247,
      "step": 1991
    },
    {
      "epoch": 0.14615222355714116,
      "grad_norm": 1.628624028958428,
      "learning_rate": 9.650368086995092e-06,
      "loss": 0.7972,
      "step": 1992
    },
    {
      "epoch": 0.14615222355714116,
      "lm_loss": 2.5715,
      "step": 1992,
      "vm_loss": 0.0488
    },
    {
      "epoch": 0.14615222355714116,
      "lm_loss": 1.034,
      "step": 1992,
      "vm_loss": 0.1223
    },
    {
      "epoch": 0.14615222355714116,
      "lm_loss": 0.7413,
      "step": 1992,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.14615222355714116,
      "lm_loss": 0.3069,
      "step": 1992,
      "vm_loss": 0.0988
    },
    {
      "epoch": 0.14615222355714116,
      "lm_loss": 0.5183,
      "step": 1992,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.14615222355714116,
      "lm_loss": 0.9238,
      "step": 1992,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.14615222355714116,
      "lm_loss": 0.3562,
      "step": 1992,
      "vm_loss": 0.0459
    },
    {
      "epoch": 0.14615222355714116,
      "lm_loss": 0.2942,
      "step": 1992,
      "vm_loss": 0.1656
    },
    {
      "epoch": 0.14622559314728029,
      "grad_norm": 1.6882755022690583,
      "learning_rate": 9.649931443988977e-06,
      "loss": 0.8974,
      "step": 1993
    },
    {
      "epoch": 0.1462989627374194,
      "grad_norm": 1.6260506058559998,
      "learning_rate": 9.649494538389722e-06,
      "loss": 0.8403,
      "step": 1994
    },
    {
      "epoch": 0.14637233232755853,
      "grad_norm": 1.6694505509131823,
      "learning_rate": 9.649057370221992e-06,
      "loss": 0.9,
      "step": 1995
    },
    {
      "epoch": 0.14644570191769765,
      "grad_norm": 1.5309153626599536,
      "learning_rate": 9.64861993951048e-06,
      "loss": 0.8523,
      "step": 1996
    },
    {
      "epoch": 0.1465190715078368,
      "grad_norm": 1.583537098358396,
      "learning_rate": 9.648182246279889e-06,
      "loss": 0.9478,
      "step": 1997
    },
    {
      "epoch": 0.14659244109797592,
      "grad_norm": 0.6523533068248697,
      "learning_rate": 9.647744290554935e-06,
      "loss": 0.5535,
      "step": 1998
    },
    {
      "epoch": 0.14666581068811504,
      "grad_norm": 1.719259124399019,
      "learning_rate": 9.647306072360349e-06,
      "loss": 0.8707,
      "step": 1999
    },
    {
      "epoch": 0.14673918027825417,
      "grad_norm": 1.877229757603639,
      "learning_rate": 9.646867591720883e-06,
      "loss": 0.9006,
      "step": 2000
    },
    {
      "epoch": 0.14673918027825417,
      "lm_loss": 1.4078,
      "step": 2000,
      "vm_loss": 0.0495
    },
    {
      "epoch": 0.14673918027825417,
      "lm_loss": 0.1553,
      "step": 2000,
      "vm_loss": 0.1442
    },
    {
      "epoch": 0.14673918027825417,
      "lm_loss": 0.2985,
      "step": 2000,
      "vm_loss": 0.1444
    },
    {
      "epoch": 0.14673918027825417,
      "lm_loss": 0.8686,
      "step": 2000,
      "vm_loss": 0.0835
    },
    {
      "epoch": 0.14673918027825417,
      "lm_loss": 0.7334,
      "step": 2000,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.14673918027825417,
      "lm_loss": 0.9931,
      "step": 2000,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.14673918027825417,
      "lm_loss": 1.2286,
      "step": 2000,
      "vm_loss": 0.2302
    },
    {
      "epoch": 0.14673918027825417,
      "lm_loss": 0.1181,
      "step": 2000,
      "vm_loss": 0.1451
    },
    {
      "epoch": 0.1468125498683933,
      "grad_norm": 1.8017144614576917,
      "learning_rate": 9.646428848661295e-06,
      "loss": 0.933,
      "step": 2001
    },
    {
      "epoch": 0.14688591945853244,
      "grad_norm": 1.738627073741251,
      "learning_rate": 9.645989843206364e-06,
      "loss": 0.8062,
      "step": 2002
    },
    {
      "epoch": 0.14695928904867156,
      "grad_norm": 1.7192946775995206,
      "learning_rate": 9.64555057538088e-06,
      "loss": 0.8728,
      "step": 2003
    },
    {
      "epoch": 0.14703265863881068,
      "grad_norm": 0.6648929489497584,
      "learning_rate": 9.645111045209652e-06,
      "loss": 0.5929,
      "step": 2004
    },
    {
      "epoch": 0.1471060282289498,
      "grad_norm": 1.7794001587693289,
      "learning_rate": 9.6446712527175e-06,
      "loss": 0.9316,
      "step": 2005
    },
    {
      "epoch": 0.14717939781908893,
      "grad_norm": 1.4675234818959482,
      "learning_rate": 9.644231197929256e-06,
      "loss": 0.907,
      "step": 2006
    },
    {
      "epoch": 0.14725276740922805,
      "grad_norm": 1.464568228634465,
      "learning_rate": 9.643790880869779e-06,
      "loss": 1.0031,
      "step": 2007
    },
    {
      "epoch": 0.1473261369993672,
      "grad_norm": 0.6069240872748561,
      "learning_rate": 9.643350301563929e-06,
      "loss": 0.5463,
      "step": 2008
    },
    {
      "epoch": 0.1473261369993672,
      "lm_loss": 1.4796,
      "step": 2008,
      "vm_loss": 0.0866
    },
    {
      "epoch": 0.1473261369993672,
      "lm_loss": 0.9448,
      "step": 2008,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.1473261369993672,
      "lm_loss": 0.7942,
      "step": 2008,
      "vm_loss": 0.1629
    },
    {
      "epoch": 0.1473261369993672,
      "lm_loss": 0.727,
      "step": 2008,
      "vm_loss": 0.1879
    },
    {
      "epoch": 0.1473261369993672,
      "lm_loss": 1.0105,
      "step": 2008,
      "vm_loss": 0.2044
    },
    {
      "epoch": 0.1473261369993672,
      "lm_loss": 0.2205,
      "step": 2008,
      "vm_loss": 0.1275
    },
    {
      "epoch": 0.1473261369993672,
      "lm_loss": 0.5786,
      "step": 2008,
      "vm_loss": 0.0717
    },
    {
      "epoch": 0.1473261369993672,
      "lm_loss": 0.433,
      "step": 2008,
      "vm_loss": 0.097
    },
    {
      "epoch": 0.14739950658950632,
      "grad_norm": 1.5735129980902633,
      "learning_rate": 9.64290946003659e-06,
      "loss": 0.8478,
      "step": 2009
    },
    {
      "epoch": 0.14747287617964544,
      "grad_norm": 1.7015862624004592,
      "learning_rate": 9.642468356312655e-06,
      "loss": 0.9874,
      "step": 2010
    },
    {
      "epoch": 0.14754624576978456,
      "grad_norm": 1.5543770533641046,
      "learning_rate": 9.642026990417034e-06,
      "loss": 0.8531,
      "step": 2011
    },
    {
      "epoch": 0.14761961535992368,
      "grad_norm": 1.4976741017610335,
      "learning_rate": 9.641585362374654e-06,
      "loss": 0.8394,
      "step": 2012
    },
    {
      "epoch": 0.14769298495006283,
      "grad_norm": 0.6540341243840839,
      "learning_rate": 9.641143472210454e-06,
      "loss": 0.5679,
      "step": 2013
    },
    {
      "epoch": 0.14776635454020196,
      "grad_norm": 1.7501948407749022,
      "learning_rate": 9.64070131994939e-06,
      "loss": 0.8631,
      "step": 2014
    },
    {
      "epoch": 0.14783972413034108,
      "grad_norm": 0.603555672329481,
      "learning_rate": 9.64025890561643e-06,
      "loss": 0.5416,
      "step": 2015
    },
    {
      "epoch": 0.1479130937204802,
      "grad_norm": 0.5639201295508671,
      "learning_rate": 9.639816229236554e-06,
      "loss": 0.5274,
      "step": 2016
    },
    {
      "epoch": 0.1479130937204802,
      "lm_loss": 0.1735,
      "step": 2016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1479130937204802,
      "lm_loss": 0.8231,
      "step": 2016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1479130937204802,
      "lm_loss": 0.4524,
      "step": 2016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1479130937204802,
      "lm_loss": 0.4471,
      "step": 2016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1479130937204802,
      "lm_loss": 0.5937,
      "step": 2016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1479130937204802,
      "lm_loss": 0.552,
      "step": 2016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1479130937204802,
      "lm_loss": 0.4503,
      "step": 2016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1479130937204802,
      "lm_loss": 0.5841,
      "step": 2016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14798646331061932,
      "grad_norm": 0.5843121657181674,
      "learning_rate": 9.639373290834769e-06,
      "loss": 0.564,
      "step": 2017
    },
    {
      "epoch": 0.14805983290075847,
      "grad_norm": 1.5621924414248178,
      "learning_rate": 9.638930090436082e-06,
      "loss": 0.8122,
      "step": 2018
    },
    {
      "epoch": 0.1481332024908976,
      "grad_norm": 0.6604779789437416,
      "learning_rate": 9.638486628065528e-06,
      "loss": 0.5667,
      "step": 2019
    },
    {
      "epoch": 0.1482065720810367,
      "grad_norm": 1.8009064482506876,
      "learning_rate": 9.638042903748145e-06,
      "loss": 0.9689,
      "step": 2020
    },
    {
      "epoch": 0.14827994167117584,
      "grad_norm": 1.6222546335575108,
      "learning_rate": 9.637598917508994e-06,
      "loss": 0.8266,
      "step": 2021
    },
    {
      "epoch": 0.14835331126131496,
      "grad_norm": 1.557883794515253,
      "learning_rate": 9.637154669373147e-06,
      "loss": 0.8398,
      "step": 2022
    },
    {
      "epoch": 0.1484266808514541,
      "grad_norm": 0.6390412860711165,
      "learning_rate": 9.636710159365693e-06,
      "loss": 0.5424,
      "step": 2023
    },
    {
      "epoch": 0.14850005044159323,
      "grad_norm": 0.6283699255591061,
      "learning_rate": 9.636265387511734e-06,
      "loss": 0.5619,
      "step": 2024
    },
    {
      "epoch": 0.14850005044159323,
      "lm_loss": 0.7038,
      "step": 2024,
      "vm_loss": 0.0357
    },
    {
      "epoch": 0.14850005044159323,
      "lm_loss": 1.0265,
      "step": 2024,
      "vm_loss": 0.1865
    },
    {
      "epoch": 0.14850005044159323,
      "lm_loss": 0.2668,
      "step": 2024,
      "vm_loss": 0.1326
    },
    {
      "epoch": 0.14850005044159323,
      "lm_loss": 1.2828,
      "step": 2024,
      "vm_loss": 0.1905
    },
    {
      "epoch": 0.14850005044159323,
      "lm_loss": 0.0342,
      "step": 2024,
      "vm_loss": 0.0661
    },
    {
      "epoch": 0.14850005044159323,
      "lm_loss": 0.3041,
      "step": 2024,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.14850005044159323,
      "lm_loss": 0.4576,
      "step": 2024,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.14850005044159323,
      "lm_loss": 0.2877,
      "step": 2024,
      "vm_loss": 0.2617
    },
    {
      "epoch": 0.14857342003173235,
      "grad_norm": 1.6637409025196348,
      "learning_rate": 9.635820353836385e-06,
      "loss": 0.9251,
      "step": 2025
    },
    {
      "epoch": 0.14864678962187147,
      "grad_norm": 1.6808942318680555,
      "learning_rate": 9.635375058364782e-06,
      "loss": 1.036,
      "step": 2026
    },
    {
      "epoch": 0.1487201592120106,
      "grad_norm": 0.5809102431231792,
      "learning_rate": 9.63492950112207e-06,
      "loss": 0.5495,
      "step": 2027
    },
    {
      "epoch": 0.14879352880214972,
      "grad_norm": 1.6887718381243713,
      "learning_rate": 9.634483682133409e-06,
      "loss": 0.9268,
      "step": 2028
    },
    {
      "epoch": 0.14886689839228887,
      "grad_norm": 0.5940951097870059,
      "learning_rate": 9.63403760142398e-06,
      "loss": 0.5649,
      "step": 2029
    },
    {
      "epoch": 0.148940267982428,
      "grad_norm": 0.5641607865330038,
      "learning_rate": 9.63359125901897e-06,
      "loss": 0.5577,
      "step": 2030
    },
    {
      "epoch": 0.1490136375725671,
      "grad_norm": 1.7766011736851672,
      "learning_rate": 9.633144654943588e-06,
      "loss": 0.8029,
      "step": 2031
    },
    {
      "epoch": 0.14908700716270623,
      "grad_norm": 2.1415565686207287,
      "learning_rate": 9.63269778922305e-06,
      "loss": 0.9848,
      "step": 2032
    },
    {
      "epoch": 0.14908700716270623,
      "lm_loss": 0.6172,
      "step": 2032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14908700716270623,
      "lm_loss": 0.5659,
      "step": 2032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14908700716270623,
      "lm_loss": 0.5838,
      "step": 2032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14908700716270623,
      "lm_loss": 0.4174,
      "step": 2032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14908700716270623,
      "lm_loss": 0.4552,
      "step": 2032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14908700716270623,
      "lm_loss": 0.5358,
      "step": 2032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14908700716270623,
      "lm_loss": 0.4179,
      "step": 2032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14908700716270623,
      "lm_loss": 0.5205,
      "step": 2032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.14916037675284535,
      "grad_norm": 0.5938023372779943,
      "learning_rate": 9.632250661882598e-06,
      "loss": 0.5485,
      "step": 2033
    },
    {
      "epoch": 0.1492337463429845,
      "grad_norm": 1.940255944928084,
      "learning_rate": 9.63180327294748e-06,
      "loss": 0.9119,
      "step": 2034
    },
    {
      "epoch": 0.14930711593312362,
      "grad_norm": 1.748525865731965,
      "learning_rate": 9.63135562244296e-06,
      "loss": 0.8584,
      "step": 2035
    },
    {
      "epoch": 0.14938048552326275,
      "grad_norm": 1.7747744016199822,
      "learning_rate": 9.630907710394319e-06,
      "loss": 0.9213,
      "step": 2036
    },
    {
      "epoch": 0.14945385511340187,
      "grad_norm": 0.6087880890630154,
      "learning_rate": 9.63045953682685e-06,
      "loss": 0.5383,
      "step": 2037
    },
    {
      "epoch": 0.149527224703541,
      "grad_norm": 1.8686963387798303,
      "learning_rate": 9.630011101765866e-06,
      "loss": 0.827,
      "step": 2038
    },
    {
      "epoch": 0.14960059429368014,
      "grad_norm": 1.758132047620512,
      "learning_rate": 9.629562405236687e-06,
      "loss": 0.859,
      "step": 2039
    },
    {
      "epoch": 0.14967396388381926,
      "grad_norm": 1.6338706396091962,
      "learning_rate": 9.629113447264655e-06,
      "loss": 0.9204,
      "step": 2040
    },
    {
      "epoch": 0.14967396388381926,
      "lm_loss": 1.3668,
      "step": 2040,
      "vm_loss": 0.09
    },
    {
      "epoch": 0.14967396388381926,
      "lm_loss": 0.2566,
      "step": 2040,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.14967396388381926,
      "lm_loss": 0.7963,
      "step": 2040,
      "vm_loss": 0.1123
    },
    {
      "epoch": 0.14967396388381926,
      "lm_loss": 0.5401,
      "step": 2040,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.14967396388381926,
      "lm_loss": 1.2255,
      "step": 2040,
      "vm_loss": 0.23
    },
    {
      "epoch": 0.14967396388381926,
      "lm_loss": 0.0768,
      "step": 2040,
      "vm_loss": 0.179
    },
    {
      "epoch": 0.14967396388381926,
      "lm_loss": 1.1871,
      "step": 2040,
      "vm_loss": 0.0914
    },
    {
      "epoch": 0.14967396388381926,
      "lm_loss": 0.3193,
      "step": 2040,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.14974733347395838,
      "grad_norm": 1.741824023428898,
      "learning_rate": 9.628664227875124e-06,
      "loss": 0.8752,
      "step": 2041
    },
    {
      "epoch": 0.1498207030640975,
      "grad_norm": 0.6048556387928882,
      "learning_rate": 9.62821474709346e-06,
      "loss": 0.56,
      "step": 2042
    },
    {
      "epoch": 0.14989407265423663,
      "grad_norm": 1.79864231173568,
      "learning_rate": 9.627765004945047e-06,
      "loss": 0.9221,
      "step": 2043
    },
    {
      "epoch": 0.14996744224437578,
      "grad_norm": 0.6439667611150726,
      "learning_rate": 9.627315001455283e-06,
      "loss": 0.5565,
      "step": 2044
    },
    {
      "epoch": 0.1500408118345149,
      "grad_norm": 0.5927019147998362,
      "learning_rate": 9.626864736649585e-06,
      "loss": 0.5634,
      "step": 2045
    },
    {
      "epoch": 0.15011418142465402,
      "grad_norm": 1.8655204238594534,
      "learning_rate": 9.626414210553375e-06,
      "loss": 0.9446,
      "step": 2046
    },
    {
      "epoch": 0.15018755101479314,
      "grad_norm": 1.6241599312614845,
      "learning_rate": 9.625963423192097e-06,
      "loss": 0.8456,
      "step": 2047
    },
    {
      "epoch": 0.15026092060493226,
      "grad_norm": 0.6079790022876832,
      "learning_rate": 9.62551237459121e-06,
      "loss": 0.5718,
      "step": 2048
    },
    {
      "epoch": 0.15026092060493226,
      "lm_loss": 0.4915,
      "step": 2048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15026092060493226,
      "lm_loss": 0.3278,
      "step": 2048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15026092060493226,
      "lm_loss": 0.3665,
      "step": 2048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15026092060493226,
      "lm_loss": 0.4406,
      "step": 2048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15026092060493226,
      "lm_loss": 0.5725,
      "step": 2048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15026092060493226,
      "lm_loss": 0.4191,
      "step": 2048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15026092060493226,
      "lm_loss": 0.6281,
      "step": 2048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15026092060493226,
      "lm_loss": 0.5382,
      "step": 2048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15033429019507139,
      "grad_norm": 0.6112063715793129,
      "learning_rate": 9.625061064776183e-06,
      "loss": 0.552,
      "step": 2049
    },
    {
      "epoch": 0.15040765978521053,
      "grad_norm": 1.7079238405004904,
      "learning_rate": 9.624609493772504e-06,
      "loss": 0.8157,
      "step": 2050
    },
    {
      "epoch": 0.15048102937534966,
      "grad_norm": 0.6170113298207757,
      "learning_rate": 9.624157661605674e-06,
      "loss": 0.556,
      "step": 2051
    },
    {
      "epoch": 0.15055439896548878,
      "grad_norm": 1.820696151752125,
      "learning_rate": 9.623705568301209e-06,
      "loss": 0.8753,
      "step": 2052
    },
    {
      "epoch": 0.1506277685556279,
      "grad_norm": 1.7993300088658608,
      "learning_rate": 9.623253213884643e-06,
      "loss": 0.88,
      "step": 2053
    },
    {
      "epoch": 0.15070113814576702,
      "grad_norm": 0.5986518529675332,
      "learning_rate": 9.622800598381515e-06,
      "loss": 0.5423,
      "step": 2054
    },
    {
      "epoch": 0.15077450773590617,
      "grad_norm": 1.565831791747312,
      "learning_rate": 9.622347721817391e-06,
      "loss": 0.8734,
      "step": 2055
    },
    {
      "epoch": 0.1508478773260453,
      "grad_norm": 0.6363659798075266,
      "learning_rate": 9.621894584217843e-06,
      "loss": 0.5582,
      "step": 2056
    },
    {
      "epoch": 0.1508478773260453,
      "lm_loss": 0.4847,
      "step": 2056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1508478773260453,
      "lm_loss": 0.5439,
      "step": 2056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1508478773260453,
      "lm_loss": 0.6809,
      "step": 2056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1508478773260453,
      "lm_loss": 0.5674,
      "step": 2056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1508478773260453,
      "lm_loss": 0.5948,
      "step": 2056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1508478773260453,
      "lm_loss": 0.4898,
      "step": 2056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1508478773260453,
      "lm_loss": 0.5098,
      "step": 2056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1508478773260453,
      "lm_loss": 0.5604,
      "step": 2056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15092124691618442,
      "grad_norm": 0.5967506692529596,
      "learning_rate": 9.621441185608464e-06,
      "loss": 0.5616,
      "step": 2057
    },
    {
      "epoch": 0.15099461650632354,
      "grad_norm": 1.7073119906691137,
      "learning_rate": 9.620987526014854e-06,
      "loss": 0.8483,
      "step": 2058
    },
    {
      "epoch": 0.15106798609646266,
      "grad_norm": 1.5710861729885677,
      "learning_rate": 9.620533605462636e-06,
      "loss": 0.8427,
      "step": 2059
    },
    {
      "epoch": 0.1511413556866018,
      "grad_norm": 0.5999182819791978,
      "learning_rate": 9.620079423977441e-06,
      "loss": 0.5606,
      "step": 2060
    },
    {
      "epoch": 0.15121472527674093,
      "grad_norm": 1.6937299759142603,
      "learning_rate": 9.619624981584923e-06,
      "loss": 0.9154,
      "step": 2061
    },
    {
      "epoch": 0.15128809486688005,
      "grad_norm": 1.512702316853248,
      "learning_rate": 9.61917027831074e-06,
      "loss": 0.8539,
      "step": 2062
    },
    {
      "epoch": 0.15136146445701917,
      "grad_norm": 1.5731570261610017,
      "learning_rate": 9.618715314180572e-06,
      "loss": 0.8621,
      "step": 2063
    },
    {
      "epoch": 0.1514348340471583,
      "grad_norm": 1.6247267725168646,
      "learning_rate": 9.618260089220113e-06,
      "loss": 0.8673,
      "step": 2064
    },
    {
      "epoch": 0.1514348340471583,
      "lm_loss": 1.7912,
      "step": 2064,
      "vm_loss": 0.0353
    },
    {
      "epoch": 0.1514348340471583,
      "lm_loss": 0.5454,
      "step": 2064,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.1514348340471583,
      "lm_loss": 0.5203,
      "step": 2064,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.1514348340471583,
      "lm_loss": 0.6904,
      "step": 2064,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.1514348340471583,
      "lm_loss": 0.2323,
      "step": 2064,
      "vm_loss": 0.2056
    },
    {
      "epoch": 0.1514348340471583,
      "lm_loss": 0.7372,
      "step": 2064,
      "vm_loss": 0.1338
    },
    {
      "epoch": 0.1514348340471583,
      "lm_loss": 1.0864,
      "step": 2064,
      "vm_loss": 0.0848
    },
    {
      "epoch": 0.1514348340471583,
      "lm_loss": 0.5517,
      "step": 2064,
      "vm_loss": 0.1249
    },
    {
      "epoch": 0.15150820363729744,
      "grad_norm": 1.5539283449509529,
      "learning_rate": 9.617804603455068e-06,
      "loss": 0.8187,
      "step": 2065
    },
    {
      "epoch": 0.15158157322743657,
      "grad_norm": 0.7052968341009257,
      "learning_rate": 9.617348856911162e-06,
      "loss": 0.5751,
      "step": 2066
    },
    {
      "epoch": 0.1516549428175757,
      "grad_norm": 0.639239569433652,
      "learning_rate": 9.616892849614132e-06,
      "loss": 0.5489,
      "step": 2067
    },
    {
      "epoch": 0.1517283124077148,
      "grad_norm": 1.5692941632136945,
      "learning_rate": 9.61643658158973e-06,
      "loss": 0.9072,
      "step": 2068
    },
    {
      "epoch": 0.15180168199785393,
      "grad_norm": 1.7477034111876997,
      "learning_rate": 9.61598005286372e-06,
      "loss": 0.8061,
      "step": 2069
    },
    {
      "epoch": 0.15187505158799305,
      "grad_norm": 1.7536350626644772,
      "learning_rate": 9.615523263461886e-06,
      "loss": 0.8688,
      "step": 2070
    },
    {
      "epoch": 0.1519484211781322,
      "grad_norm": 0.686386001669242,
      "learning_rate": 9.615066213410025e-06,
      "loss": 0.5422,
      "step": 2071
    },
    {
      "epoch": 0.15202179076827133,
      "grad_norm": 1.6327194282093647,
      "learning_rate": 9.614608902733944e-06,
      "loss": 0.8552,
      "step": 2072
    },
    {
      "epoch": 0.15202179076827133,
      "lm_loss": 1.1088,
      "step": 2072,
      "vm_loss": 0.0455
    },
    {
      "epoch": 0.15202179076827133,
      "lm_loss": 0.4142,
      "step": 2072,
      "vm_loss": 0.0705
    },
    {
      "epoch": 0.15202179076827133,
      "lm_loss": 0.2798,
      "step": 2072,
      "vm_loss": 0.0747
    },
    {
      "epoch": 0.15202179076827133,
      "lm_loss": 0.6212,
      "step": 2072,
      "vm_loss": 0.0521
    },
    {
      "epoch": 0.15202179076827133,
      "lm_loss": 0.4959,
      "step": 2072,
      "vm_loss": 0.0965
    },
    {
      "epoch": 0.15202179076827133,
      "lm_loss": 1.1002,
      "step": 2072,
      "vm_loss": 0.1572
    },
    {
      "epoch": 0.15202179076827133,
      "lm_loss": 0.5663,
      "step": 2072,
      "vm_loss": 0.0856
    },
    {
      "epoch": 0.15202179076827133,
      "lm_loss": 0.7614,
      "step": 2072,
      "vm_loss": 0.0776
    },
    {
      "epoch": 0.15209516035841045,
      "grad_norm": 1.641458285415528,
      "learning_rate": 9.614151331459469e-06,
      "loss": 0.8526,
      "step": 2073
    },
    {
      "epoch": 0.15216852994854957,
      "grad_norm": 1.4366724123243237,
      "learning_rate": 9.613693499612445e-06,
      "loss": 0.8128,
      "step": 2074
    },
    {
      "epoch": 0.1522418995386887,
      "grad_norm": 1.506819102705303,
      "learning_rate": 9.61323540721872e-06,
      "loss": 0.9359,
      "step": 2075
    },
    {
      "epoch": 0.15231526912882784,
      "grad_norm": 1.848586512439693,
      "learning_rate": 9.61277705430417e-06,
      "loss": 0.8952,
      "step": 2076
    },
    {
      "epoch": 0.15238863871896696,
      "grad_norm": 1.4933640909324584,
      "learning_rate": 9.612318440894673e-06,
      "loss": 0.8737,
      "step": 2077
    },
    {
      "epoch": 0.15246200830910608,
      "grad_norm": 0.6809214840631658,
      "learning_rate": 9.611859567016135e-06,
      "loss": 0.5388,
      "step": 2078
    },
    {
      "epoch": 0.1525353778992452,
      "grad_norm": 1.716830018203256,
      "learning_rate": 9.611400432694463e-06,
      "loss": 0.8259,
      "step": 2079
    },
    {
      "epoch": 0.15260874748938433,
      "grad_norm": 1.7934057887322121,
      "learning_rate": 9.61094103795559e-06,
      "loss": 0.8656,
      "step": 2080
    },
    {
      "epoch": 0.15260874748938433,
      "lm_loss": 0.6088,
      "step": 2080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15260874748938433,
      "lm_loss": 0.4603,
      "step": 2080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15260874748938433,
      "lm_loss": 0.4507,
      "step": 2080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15260874748938433,
      "lm_loss": 0.5493,
      "step": 2080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15260874748938433,
      "lm_loss": 0.4738,
      "step": 2080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15260874748938433,
      "lm_loss": 0.8006,
      "step": 2080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15260874748938433,
      "lm_loss": 0.6183,
      "step": 2080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15260874748938433,
      "lm_loss": 0.623,
      "step": 2080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15268211707952348,
      "grad_norm": 0.6045761947852507,
      "learning_rate": 9.610481382825458e-06,
      "loss": 0.5651,
      "step": 2081
    },
    {
      "epoch": 0.1527554866696626,
      "grad_norm": 0.5877432367927914,
      "learning_rate": 9.610021467330026e-06,
      "loss": 0.5522,
      "step": 2082
    },
    {
      "epoch": 0.15282885625980172,
      "grad_norm": 1.6903253832357799,
      "learning_rate": 9.609561291495262e-06,
      "loss": 0.8467,
      "step": 2083
    },
    {
      "epoch": 0.15290222584994084,
      "grad_norm": 1.4643908373741281,
      "learning_rate": 9.60910085534716e-06,
      "loss": 0.8996,
      "step": 2084
    },
    {
      "epoch": 0.15297559544007996,
      "grad_norm": 0.5788788180509763,
      "learning_rate": 9.608640158911717e-06,
      "loss": 0.5289,
      "step": 2085
    },
    {
      "epoch": 0.1530489650302191,
      "grad_norm": 0.5698379013660503,
      "learning_rate": 9.60817920221495e-06,
      "loss": 0.5506,
      "step": 2086
    },
    {
      "epoch": 0.15312233462035824,
      "grad_norm": 1.5213674043951937,
      "learning_rate": 9.607717985282891e-06,
      "loss": 0.7805,
      "step": 2087
    },
    {
      "epoch": 0.15319570421049736,
      "grad_norm": 1.760019130955491,
      "learning_rate": 9.607256508141589e-06,
      "loss": 0.923,
      "step": 2088
    },
    {
      "epoch": 0.15319570421049736,
      "lm_loss": 0.6244,
      "step": 2088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15319570421049736,
      "lm_loss": 0.6032,
      "step": 2088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15319570421049736,
      "lm_loss": 0.5128,
      "step": 2088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15319570421049736,
      "lm_loss": 0.6503,
      "step": 2088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15319570421049736,
      "lm_loss": 0.5462,
      "step": 2088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15319570421049736,
      "lm_loss": 0.516,
      "step": 2088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15319570421049736,
      "lm_loss": 0.4671,
      "step": 2088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15319570421049736,
      "lm_loss": 0.5994,
      "step": 2088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15326907380063648,
      "grad_norm": 0.5807083463735471,
      "learning_rate": 9.606794770817102e-06,
      "loss": 0.5576,
      "step": 2089
    },
    {
      "epoch": 0.1533424433907756,
      "grad_norm": 1.648938697509036,
      "learning_rate": 9.606332773335505e-06,
      "loss": 0.8856,
      "step": 2090
    },
    {
      "epoch": 0.15341581298091472,
      "grad_norm": 1.7022956410925068,
      "learning_rate": 9.605870515722888e-06,
      "loss": 0.8454,
      "step": 2091
    },
    {
      "epoch": 0.15348918257105387,
      "grad_norm": 1.540838154375145,
      "learning_rate": 9.605407998005357e-06,
      "loss": 0.8264,
      "step": 2092
    },
    {
      "epoch": 0.153562552161193,
      "grad_norm": 1.71105546771,
      "learning_rate": 9.604945220209034e-06,
      "loss": 0.801,
      "step": 2093
    },
    {
      "epoch": 0.15363592175133212,
      "grad_norm": 1.9860286719829252,
      "learning_rate": 9.604482182360046e-06,
      "loss": 0.8844,
      "step": 2094
    },
    {
      "epoch": 0.15370929134147124,
      "grad_norm": 1.5639352277844407,
      "learning_rate": 9.604018884484549e-06,
      "loss": 0.8625,
      "step": 2095
    },
    {
      "epoch": 0.15378266093161036,
      "grad_norm": 0.6709791540005761,
      "learning_rate": 9.603555326608704e-06,
      "loss": 0.5645,
      "step": 2096
    },
    {
      "epoch": 0.15378266093161036,
      "lm_loss": 0.2233,
      "step": 2096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15378266093161036,
      "lm_loss": 0.3965,
      "step": 2096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15378266093161036,
      "lm_loss": 0.5768,
      "step": 2096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15378266093161036,
      "lm_loss": 0.4824,
      "step": 2096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15378266093161036,
      "lm_loss": 0.6273,
      "step": 2096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15378266093161036,
      "lm_loss": 0.5508,
      "step": 2096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15378266093161036,
      "lm_loss": 0.5271,
      "step": 2096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15378266093161036,
      "lm_loss": 0.6326,
      "step": 2096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1538560305217495,
      "grad_norm": 0.6129690006252547,
      "learning_rate": 9.60309150875869e-06,
      "loss": 0.5298,
      "step": 2097
    },
    {
      "epoch": 0.15392940011188863,
      "grad_norm": 0.6192968879573043,
      "learning_rate": 9.602627430960696e-06,
      "loss": 0.5674,
      "step": 2098
    },
    {
      "epoch": 0.15400276970202775,
      "grad_norm": 1.623588355977832,
      "learning_rate": 9.602163093240936e-06,
      "loss": 0.8115,
      "step": 2099
    },
    {
      "epoch": 0.15407613929216687,
      "grad_norm": 0.5868133815508111,
      "learning_rate": 9.601698495625628e-06,
      "loss": 0.5672,
      "step": 2100
    },
    {
      "epoch": 0.154149508882306,
      "grad_norm": 0.639585697963578,
      "learning_rate": 9.60123363814101e-06,
      "loss": 0.5563,
      "step": 2101
    },
    {
      "epoch": 0.15422287847244515,
      "grad_norm": 1.8327389905701923,
      "learning_rate": 9.600768520813333e-06,
      "loss": 0.9519,
      "step": 2102
    },
    {
      "epoch": 0.15429624806258427,
      "grad_norm": 1.490600743892167,
      "learning_rate": 9.600303143668865e-06,
      "loss": 0.8652,
      "step": 2103
    },
    {
      "epoch": 0.1543696176527234,
      "grad_norm": 1.5938130906313412,
      "learning_rate": 9.599837506733885e-06,
      "loss": 0.8451,
      "step": 2104
    },
    {
      "epoch": 0.1543696176527234,
      "lm_loss": 0.4856,
      "step": 2104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1543696176527234,
      "lm_loss": 0.5251,
      "step": 2104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1543696176527234,
      "lm_loss": 0.4128,
      "step": 2104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1543696176527234,
      "lm_loss": 0.6271,
      "step": 2104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1543696176527234,
      "lm_loss": 0.6482,
      "step": 2104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1543696176527234,
      "lm_loss": 0.5358,
      "step": 2104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1543696176527234,
      "lm_loss": 0.6867,
      "step": 2104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1543696176527234,
      "lm_loss": 0.5861,
      "step": 2104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1544429872428625,
      "grad_norm": 0.6634077338737469,
      "learning_rate": 9.59937161003469e-06,
      "loss": 0.5555,
      "step": 2105
    },
    {
      "epoch": 0.15451635683300163,
      "grad_norm": 1.649765840661449,
      "learning_rate": 9.598905453597591e-06,
      "loss": 0.9261,
      "step": 2106
    },
    {
      "epoch": 0.15458972642314076,
      "grad_norm": 1.5786707390506538,
      "learning_rate": 9.59843903744891e-06,
      "loss": 0.8551,
      "step": 2107
    },
    {
      "epoch": 0.1546630960132799,
      "grad_norm": 1.584063426610224,
      "learning_rate": 9.597972361614991e-06,
      "loss": 0.8739,
      "step": 2108
    },
    {
      "epoch": 0.15473646560341903,
      "grad_norm": 1.687624165726387,
      "learning_rate": 9.597505426122184e-06,
      "loss": 0.8739,
      "step": 2109
    },
    {
      "epoch": 0.15480983519355815,
      "grad_norm": 1.7851306187855525,
      "learning_rate": 9.597038230996863e-06,
      "loss": 0.9029,
      "step": 2110
    },
    {
      "epoch": 0.15488320478369727,
      "grad_norm": 1.7238089260967078,
      "learning_rate": 9.596570776265406e-06,
      "loss": 0.8355,
      "step": 2111
    },
    {
      "epoch": 0.1549565743738364,
      "grad_norm": 1.6222929598897313,
      "learning_rate": 9.596103061954213e-06,
      "loss": 0.8803,
      "step": 2112
    },
    {
      "epoch": 0.1549565743738364,
      "lm_loss": 1.3515,
      "step": 2112,
      "vm_loss": 0.0743
    },
    {
      "epoch": 0.1549565743738364,
      "lm_loss": 0.9161,
      "step": 2112,
      "vm_loss": 0.1276
    },
    {
      "epoch": 0.1549565743738364,
      "lm_loss": 1.1004,
      "step": 2112,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.1549565743738364,
      "lm_loss": 0.753,
      "step": 2112,
      "vm_loss": 0.1793
    },
    {
      "epoch": 0.1549565743738364,
      "lm_loss": 0.479,
      "step": 2112,
      "vm_loss": 0.085
    },
    {
      "epoch": 0.1549565743738364,
      "lm_loss": 0.6504,
      "step": 2112,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.1549565743738364,
      "lm_loss": 0.9555,
      "step": 2112,
      "vm_loss": 0.2385
    },
    {
      "epoch": 0.1549565743738364,
      "lm_loss": 0.589,
      "step": 2112,
      "vm_loss": 0.1652
    },
    {
      "epoch": 0.15502994396397554,
      "grad_norm": 1.6104757688480136,
      "learning_rate": 9.595635088089701e-06,
      "loss": 0.8592,
      "step": 2113
    },
    {
      "epoch": 0.15510331355411466,
      "grad_norm": 1.5448323652901728,
      "learning_rate": 9.595166854698295e-06,
      "loss": 0.8215,
      "step": 2114
    },
    {
      "epoch": 0.15517668314425379,
      "grad_norm": 1.6629819738968046,
      "learning_rate": 9.594698361806435e-06,
      "loss": 0.8458,
      "step": 2115
    },
    {
      "epoch": 0.1552500527343929,
      "grad_norm": 0.6661842573595275,
      "learning_rate": 9.59422960944058e-06,
      "loss": 0.5742,
      "step": 2116
    },
    {
      "epoch": 0.15532342232453203,
      "grad_norm": 1.5828322912514692,
      "learning_rate": 9.593760597627201e-06,
      "loss": 0.9125,
      "step": 2117
    },
    {
      "epoch": 0.15539679191467118,
      "grad_norm": 0.6334304746434447,
      "learning_rate": 9.593291326392788e-06,
      "loss": 0.5583,
      "step": 2118
    },
    {
      "epoch": 0.1554701615048103,
      "grad_norm": 0.6121664259111947,
      "learning_rate": 9.592821795763835e-06,
      "loss": 0.5725,
      "step": 2119
    },
    {
      "epoch": 0.15554353109494942,
      "grad_norm": 0.5669791220131286,
      "learning_rate": 9.592352005766863e-06,
      "loss": 0.5395,
      "step": 2120
    },
    {
      "epoch": 0.15554353109494942,
      "lm_loss": 0.2858,
      "step": 2120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15554353109494942,
      "lm_loss": 0.3495,
      "step": 2120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15554353109494942,
      "lm_loss": 0.4615,
      "step": 2120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15554353109494942,
      "lm_loss": 0.4561,
      "step": 2120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15554353109494942,
      "lm_loss": 0.4625,
      "step": 2120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15554353109494942,
      "lm_loss": 0.4347,
      "step": 2120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15554353109494942,
      "lm_loss": 0.6772,
      "step": 2120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15554353109494942,
      "lm_loss": 0.4978,
      "step": 2120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15561690068508854,
      "grad_norm": 0.5995890522908884,
      "learning_rate": 9.591881956428401e-06,
      "loss": 0.5368,
      "step": 2121
    },
    {
      "epoch": 0.15569027027522767,
      "grad_norm": 1.7262944350304699,
      "learning_rate": 9.591411647774993e-06,
      "loss": 0.8465,
      "step": 2122
    },
    {
      "epoch": 0.15576363986536682,
      "grad_norm": 1.7478452675304121,
      "learning_rate": 9.590941079833198e-06,
      "loss": 0.7391,
      "step": 2123
    },
    {
      "epoch": 0.15583700945550594,
      "grad_norm": 1.5889642329974945,
      "learning_rate": 9.590470252629591e-06,
      "loss": 0.9524,
      "step": 2124
    },
    {
      "epoch": 0.15591037904564506,
      "grad_norm": 1.5594410853120524,
      "learning_rate": 9.589999166190761e-06,
      "loss": 0.8031,
      "step": 2125
    },
    {
      "epoch": 0.15598374863578418,
      "grad_norm": 0.7343461379317179,
      "learning_rate": 9.589527820543312e-06,
      "loss": 0.5539,
      "step": 2126
    },
    {
      "epoch": 0.1560571182259233,
      "grad_norm": 1.6147578179134696,
      "learning_rate": 9.58905621571386e-06,
      "loss": 0.951,
      "step": 2127
    },
    {
      "epoch": 0.15613048781606242,
      "grad_norm": 1.521548054846317,
      "learning_rate": 9.588584351729038e-06,
      "loss": 0.8067,
      "step": 2128
    },
    {
      "epoch": 0.15613048781606242,
      "lm_loss": 0.7573,
      "step": 2128,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.15613048781606242,
      "lm_loss": 0.3049,
      "step": 2128,
      "vm_loss": 0.1112
    },
    {
      "epoch": 0.15613048781606242,
      "lm_loss": 0.7661,
      "step": 2128,
      "vm_loss": 0.1352
    },
    {
      "epoch": 0.15613048781606242,
      "lm_loss": 0.5845,
      "step": 2128,
      "vm_loss": 0.1526
    },
    {
      "epoch": 0.15613048781606242,
      "lm_loss": 0.5664,
      "step": 2128,
      "vm_loss": 0.1005
    },
    {
      "epoch": 0.15613048781606242,
      "lm_loss": 1.7078,
      "step": 2128,
      "vm_loss": 0.203
    },
    {
      "epoch": 0.15613048781606242,
      "lm_loss": 0.305,
      "step": 2128,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.15613048781606242,
      "lm_loss": 0.4968,
      "step": 2128,
      "vm_loss": 0.1953
    },
    {
      "epoch": 0.15620385740620157,
      "grad_norm": 1.5932438085782192,
      "learning_rate": 9.588112228615495e-06,
      "loss": 0.8835,
      "step": 2129
    },
    {
      "epoch": 0.1562772269963407,
      "grad_norm": 0.6746970351995926,
      "learning_rate": 9.587639846399895e-06,
      "loss": 0.5591,
      "step": 2130
    },
    {
      "epoch": 0.15635059658647982,
      "grad_norm": 1.7567679412565351,
      "learning_rate": 9.58716720510891e-06,
      "loss": 0.8133,
      "step": 2131
    },
    {
      "epoch": 0.15642396617661894,
      "grad_norm": 1.6009154071802392,
      "learning_rate": 9.58669430476923e-06,
      "loss": 0.8704,
      "step": 2132
    },
    {
      "epoch": 0.15649733576675806,
      "grad_norm": 1.4705841756841753,
      "learning_rate": 9.586221145407568e-06,
      "loss": 0.8542,
      "step": 2133
    },
    {
      "epoch": 0.1565707053568972,
      "grad_norm": 0.6389400729234475,
      "learning_rate": 9.585747727050638e-06,
      "loss": 0.5715,
      "step": 2134
    },
    {
      "epoch": 0.15664407494703633,
      "grad_norm": 1.57734594716157,
      "learning_rate": 9.585274049725178e-06,
      "loss": 0.8662,
      "step": 2135
    },
    {
      "epoch": 0.15671744453717545,
      "grad_norm": 0.6105547458839873,
      "learning_rate": 9.584800113457937e-06,
      "loss": 0.569,
      "step": 2136
    },
    {
      "epoch": 0.15671744453717545,
      "lm_loss": 1.5373,
      "step": 2136,
      "vm_loss": 0.0538
    },
    {
      "epoch": 0.15671744453717545,
      "lm_loss": 0.1779,
      "step": 2136,
      "vm_loss": 0.0721
    },
    {
      "epoch": 0.15671744453717545,
      "lm_loss": 0.7507,
      "step": 2136,
      "vm_loss": 0.0654
    },
    {
      "epoch": 0.15671744453717545,
      "lm_loss": 0.7371,
      "step": 2136,
      "vm_loss": 0.1547
    },
    {
      "epoch": 0.15671744453717545,
      "lm_loss": 0.4395,
      "step": 2136,
      "vm_loss": 0.1694
    },
    {
      "epoch": 0.15671744453717545,
      "lm_loss": 1.2608,
      "step": 2136,
      "vm_loss": 0.1323
    },
    {
      "epoch": 0.15671744453717545,
      "lm_loss": 0.3122,
      "step": 2136,
      "vm_loss": 0.1424
    },
    {
      "epoch": 0.15671744453717545,
      "lm_loss": 0.3186,
      "step": 2136,
      "vm_loss": 0.1204
    },
    {
      "epoch": 0.15679081412731458,
      "grad_norm": 1.4870803127479248,
      "learning_rate": 9.584325918275681e-06,
      "loss": 0.8197,
      "step": 2137
    },
    {
      "epoch": 0.1568641837174537,
      "grad_norm": 1.6297458653245789,
      "learning_rate": 9.583851464205184e-06,
      "loss": 0.8594,
      "step": 2138
    },
    {
      "epoch": 0.15693755330759285,
      "grad_norm": 1.6413256428107867,
      "learning_rate": 9.583376751273248e-06,
      "loss": 0.856,
      "step": 2139
    },
    {
      "epoch": 0.15701092289773197,
      "grad_norm": 1.4533457964219882,
      "learning_rate": 9.582901779506673e-06,
      "loss": 0.8632,
      "step": 2140
    },
    {
      "epoch": 0.1570842924878711,
      "grad_norm": 0.6810470438709876,
      "learning_rate": 9.582426548932286e-06,
      "loss": 0.5542,
      "step": 2141
    },
    {
      "epoch": 0.1571576620780102,
      "grad_norm": 0.660590542279113,
      "learning_rate": 9.581951059576925e-06,
      "loss": 0.5523,
      "step": 2142
    },
    {
      "epoch": 0.15723103166814933,
      "grad_norm": 0.5635773271987891,
      "learning_rate": 9.581475311467438e-06,
      "loss": 0.5274,
      "step": 2143
    },
    {
      "epoch": 0.15730440125828848,
      "grad_norm": 1.796543416515566,
      "learning_rate": 9.580999304630694e-06,
      "loss": 1.0203,
      "step": 2144
    },
    {
      "epoch": 0.15730440125828848,
      "lm_loss": 0.6965,
      "step": 2144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15730440125828848,
      "lm_loss": 0.459,
      "step": 2144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15730440125828848,
      "lm_loss": 0.5465,
      "step": 2144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15730440125828848,
      "lm_loss": 0.4419,
      "step": 2144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15730440125828848,
      "lm_loss": 0.4784,
      "step": 2144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15730440125828848,
      "lm_loss": 0.4692,
      "step": 2144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15730440125828848,
      "lm_loss": 0.5292,
      "step": 2144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15730440125828848,
      "lm_loss": 0.4799,
      "step": 2144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1573777708484276,
      "grad_norm": 0.6420932767235052,
      "learning_rate": 9.580523039093576e-06,
      "loss": 0.5595,
      "step": 2145
    },
    {
      "epoch": 0.15745114043856673,
      "grad_norm": 0.6938974897130274,
      "learning_rate": 9.580046514882979e-06,
      "loss": 0.5617,
      "step": 2146
    },
    {
      "epoch": 0.15752451002870585,
      "grad_norm": 1.5139365889287895,
      "learning_rate": 9.579569732025813e-06,
      "loss": 0.752,
      "step": 2147
    },
    {
      "epoch": 0.15759787961884497,
      "grad_norm": 1.4424252341040942,
      "learning_rate": 9.579092690549003e-06,
      "loss": 0.8028,
      "step": 2148
    },
    {
      "epoch": 0.1576712492089841,
      "grad_norm": 1.5895674450405635,
      "learning_rate": 9.57861539047949e-06,
      "loss": 0.8532,
      "step": 2149
    },
    {
      "epoch": 0.15774461879912324,
      "grad_norm": 1.5914516657312872,
      "learning_rate": 9.578137831844224e-06,
      "loss": 0.8519,
      "step": 2150
    },
    {
      "epoch": 0.15781798838926236,
      "grad_norm": 1.54806344390901,
      "learning_rate": 9.577660014670177e-06,
      "loss": 0.8841,
      "step": 2151
    },
    {
      "epoch": 0.1578913579794015,
      "grad_norm": 1.6653770896651783,
      "learning_rate": 9.577181938984335e-06,
      "loss": 0.9511,
      "step": 2152
    },
    {
      "epoch": 0.1578913579794015,
      "lm_loss": 0.1079,
      "step": 2152,
      "vm_loss": 0.0818
    },
    {
      "epoch": 0.1578913579794015,
      "lm_loss": 0.2203,
      "step": 2152,
      "vm_loss": 0.0802
    },
    {
      "epoch": 0.1578913579794015,
      "lm_loss": 1.512,
      "step": 2152,
      "vm_loss": 0.0986
    },
    {
      "epoch": 0.1578913579794015,
      "lm_loss": 0.3948,
      "step": 2152,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.1578913579794015,
      "lm_loss": 0.6936,
      "step": 2152,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.1578913579794015,
      "lm_loss": 0.4957,
      "step": 2152,
      "vm_loss": 0.1437
    },
    {
      "epoch": 0.1578913579794015,
      "lm_loss": 0.6,
      "step": 2152,
      "vm_loss": 0.1509
    },
    {
      "epoch": 0.1578913579794015,
      "lm_loss": 1.0632,
      "step": 2152,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.1579647275695406,
      "grad_norm": 1.3959183883747257,
      "learning_rate": 9.576703604813694e-06,
      "loss": 0.7936,
      "step": 2153
    },
    {
      "epoch": 0.15803809715967973,
      "grad_norm": 1.7232599979890337,
      "learning_rate": 9.576225012185265e-06,
      "loss": 0.8827,
      "step": 2154
    },
    {
      "epoch": 0.15811146674981888,
      "grad_norm": 0.8020936491197582,
      "learning_rate": 9.575746161126078e-06,
      "loss": 0.5505,
      "step": 2155
    },
    {
      "epoch": 0.158184836339958,
      "grad_norm": 0.7133560871248735,
      "learning_rate": 9.575267051663173e-06,
      "loss": 0.5575,
      "step": 2156
    },
    {
      "epoch": 0.15825820593009712,
      "grad_norm": 1.5439097012807572,
      "learning_rate": 9.574787683823606e-06,
      "loss": 0.8128,
      "step": 2157
    },
    {
      "epoch": 0.15833157552023625,
      "grad_norm": 1.648825104958901,
      "learning_rate": 9.574308057634452e-06,
      "loss": 0.8811,
      "step": 2158
    },
    {
      "epoch": 0.15840494511037537,
      "grad_norm": 1.6459749256233134,
      "learning_rate": 9.57382817312279e-06,
      "loss": 0.8766,
      "step": 2159
    },
    {
      "epoch": 0.15847831470051452,
      "grad_norm": 1.6276372745610754,
      "learning_rate": 9.573348030315728e-06,
      "loss": 0.7618,
      "step": 2160
    },
    {
      "epoch": 0.15847831470051452,
      "lm_loss": 1.632,
      "step": 2160,
      "vm_loss": 0.0288
    },
    {
      "epoch": 0.15847831470051452,
      "lm_loss": 1.2325,
      "step": 2160,
      "vm_loss": 0.1399
    },
    {
      "epoch": 0.15847831470051452,
      "lm_loss": 0.5025,
      "step": 2160,
      "vm_loss": 0.0741
    },
    {
      "epoch": 0.15847831470051452,
      "lm_loss": 1.0673,
      "step": 2160,
      "vm_loss": 0.167
    },
    {
      "epoch": 0.15847831470051452,
      "lm_loss": 0.4352,
      "step": 2160,
      "vm_loss": 0.1573
    },
    {
      "epoch": 0.15847831470051452,
      "lm_loss": 0.2804,
      "step": 2160,
      "vm_loss": 0.1605
    },
    {
      "epoch": 0.15847831470051452,
      "lm_loss": 0.4789,
      "step": 2160,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.15847831470051452,
      "lm_loss": 1.5215,
      "step": 2160,
      "vm_loss": 0.2496
    },
    {
      "epoch": 0.15855168429065364,
      "grad_norm": 1.8419126761696196,
      "learning_rate": 9.572867629240374e-06,
      "loss": 0.9205,
      "step": 2161
    },
    {
      "epoch": 0.15862505388079276,
      "grad_norm": 1.6295706073080198,
      "learning_rate": 9.572386969923862e-06,
      "loss": 0.8085,
      "step": 2162
    },
    {
      "epoch": 0.15869842347093188,
      "grad_norm": 1.5968248312100382,
      "learning_rate": 9.571906052393333e-06,
      "loss": 0.8588,
      "step": 2163
    },
    {
      "epoch": 0.158771793061071,
      "grad_norm": 1.52401570795056,
      "learning_rate": 9.57142487667595e-06,
      "loss": 0.8882,
      "step": 2164
    },
    {
      "epoch": 0.15884516265121015,
      "grad_norm": 1.4320318645923866,
      "learning_rate": 9.570943442798882e-06,
      "loss": 0.572,
      "step": 2165
    },
    {
      "epoch": 0.15891853224134928,
      "grad_norm": 1.6568349458818106,
      "learning_rate": 9.570461750789317e-06,
      "loss": 0.858,
      "step": 2166
    },
    {
      "epoch": 0.1589919018314884,
      "grad_norm": 1.8887239338547066,
      "learning_rate": 9.569979800674459e-06,
      "loss": 0.8689,
      "step": 2167
    },
    {
      "epoch": 0.15906527142162752,
      "grad_norm": 1.7669279640604478,
      "learning_rate": 9.569497592481524e-06,
      "loss": 0.8204,
      "step": 2168
    },
    {
      "epoch": 0.15906527142162752,
      "lm_loss": 1.4218,
      "step": 2168,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.15906527142162752,
      "lm_loss": 1.2918,
      "step": 2168,
      "vm_loss": 0.1168
    },
    {
      "epoch": 0.15906527142162752,
      "lm_loss": 0.8576,
      "step": 2168,
      "vm_loss": 0.1928
    },
    {
      "epoch": 0.15906527142162752,
      "lm_loss": 0.235,
      "step": 2168,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.15906527142162752,
      "lm_loss": 1.1217,
      "step": 2168,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.15906527142162752,
      "lm_loss": 0.0318,
      "step": 2168,
      "vm_loss": 0.1278
    },
    {
      "epoch": 0.15906527142162752,
      "lm_loss": 0.7296,
      "step": 2168,
      "vm_loss": 0.1304
    },
    {
      "epoch": 0.15906527142162752,
      "lm_loss": 1.1444,
      "step": 2168,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.15913864101176664,
      "grad_norm": 1.8039073268471408,
      "learning_rate": 9.569015126237744e-06,
      "loss": 0.7989,
      "step": 2169
    },
    {
      "epoch": 0.15921201060190576,
      "grad_norm": 1.694974668080935,
      "learning_rate": 9.568532401970365e-06,
      "loss": 0.872,
      "step": 2170
    },
    {
      "epoch": 0.1592853801920449,
      "grad_norm": 1.5798601035127064,
      "learning_rate": 9.568049419706648e-06,
      "loss": 0.8776,
      "step": 2171
    },
    {
      "epoch": 0.15935874978218403,
      "grad_norm": 1.167316228292312,
      "learning_rate": 9.567566179473866e-06,
      "loss": 0.5245,
      "step": 2172
    },
    {
      "epoch": 0.15943211937232316,
      "grad_norm": 1.555523874846942,
      "learning_rate": 9.567082681299312e-06,
      "loss": 0.7758,
      "step": 2173
    },
    {
      "epoch": 0.15950548896246228,
      "grad_norm": 0.8756918113241,
      "learning_rate": 9.566598925210289e-06,
      "loss": 0.5728,
      "step": 2174
    },
    {
      "epoch": 0.1595788585526014,
      "grad_norm": 1.8633713486648293,
      "learning_rate": 9.566114911234114e-06,
      "loss": 0.7933,
      "step": 2175
    },
    {
      "epoch": 0.15965222814274055,
      "grad_norm": 2.075965719283776,
      "learning_rate": 9.565630639398124e-06,
      "loss": 0.9316,
      "step": 2176
    },
    {
      "epoch": 0.15965222814274055,
      "lm_loss": 0.4703,
      "step": 2176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15965222814274055,
      "lm_loss": 0.6286,
      "step": 2176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15965222814274055,
      "lm_loss": 0.5105,
      "step": 2176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15965222814274055,
      "lm_loss": 0.5125,
      "step": 2176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15965222814274055,
      "lm_loss": 0.4733,
      "step": 2176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15965222814274055,
      "lm_loss": 0.6018,
      "step": 2176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15965222814274055,
      "lm_loss": 0.4809,
      "step": 2176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15965222814274055,
      "lm_loss": 0.7893,
      "step": 2176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.15972559773287967,
      "grad_norm": 0.750906984078962,
      "learning_rate": 9.565146109729664e-06,
      "loss": 0.5485,
      "step": 2177
    },
    {
      "epoch": 0.1597989673230188,
      "grad_norm": 0.8052096003923808,
      "learning_rate": 9.564661322256099e-06,
      "loss": 0.559,
      "step": 2178
    },
    {
      "epoch": 0.15987233691315791,
      "grad_norm": 0.6617744758796618,
      "learning_rate": 9.564176277004804e-06,
      "loss": 0.5464,
      "step": 2179
    },
    {
      "epoch": 0.15994570650329704,
      "grad_norm": 0.6230682937720433,
      "learning_rate": 9.563690974003171e-06,
      "loss": 0.5529,
      "step": 2180
    },
    {
      "epoch": 0.16001907609343619,
      "grad_norm": 2.136525036188987,
      "learning_rate": 9.563205413278607e-06,
      "loss": 0.9068,
      "step": 2181
    },
    {
      "epoch": 0.1600924456835753,
      "grad_norm": 1.646313794754805,
      "learning_rate": 9.562719594858534e-06,
      "loss": 0.8827,
      "step": 2182
    },
    {
      "epoch": 0.16016581527371443,
      "grad_norm": 1.7196441193582046,
      "learning_rate": 9.562233518770385e-06,
      "loss": 0.8462,
      "step": 2183
    },
    {
      "epoch": 0.16023918486385355,
      "grad_norm": 1.7599584725059447,
      "learning_rate": 9.561747185041612e-06,
      "loss": 0.9286,
      "step": 2184
    },
    {
      "epoch": 0.16023918486385355,
      "lm_loss": 1.0966,
      "step": 2184,
      "vm_loss": 0.061
    },
    {
      "epoch": 0.16023918486385355,
      "lm_loss": 1.5051,
      "step": 2184,
      "vm_loss": 0.1703
    },
    {
      "epoch": 0.16023918486385355,
      "lm_loss": 0.988,
      "step": 2184,
      "vm_loss": 0.0779
    },
    {
      "epoch": 0.16023918486385355,
      "lm_loss": 0.5501,
      "step": 2184,
      "vm_loss": 0.1999
    },
    {
      "epoch": 0.16023918486385355,
      "lm_loss": 0.0918,
      "step": 2184,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.16023918486385355,
      "lm_loss": 2.3106,
      "step": 2184,
      "vm_loss": 0.1767
    },
    {
      "epoch": 0.16023918486385355,
      "lm_loss": 0.2948,
      "step": 2184,
      "vm_loss": 0.1138
    },
    {
      "epoch": 0.16023918486385355,
      "lm_loss": 0.7501,
      "step": 2184,
      "vm_loss": 0.1483
    },
    {
      "epoch": 0.16031255445399267,
      "grad_norm": 1.6096066520154544,
      "learning_rate": 9.561260593699679e-06,
      "loss": 0.8428,
      "step": 2185
    },
    {
      "epoch": 0.1603859240441318,
      "grad_norm": 1.1396614963180958,
      "learning_rate": 9.560773744772063e-06,
      "loss": 0.5644,
      "step": 2186
    },
    {
      "epoch": 0.16045929363427094,
      "grad_norm": 1.6638926699789789,
      "learning_rate": 9.56028663828626e-06,
      "loss": 0.7272,
      "step": 2187
    },
    {
      "epoch": 0.16053266322441007,
      "grad_norm": 1.7366581843684101,
      "learning_rate": 9.559799274269777e-06,
      "loss": 0.8199,
      "step": 2188
    },
    {
      "epoch": 0.1606060328145492,
      "grad_norm": 0.6480774135072419,
      "learning_rate": 9.559311652750135e-06,
      "loss": 0.566,
      "step": 2189
    },
    {
      "epoch": 0.1606794024046883,
      "grad_norm": 0.6161722461795723,
      "learning_rate": 9.558823773754875e-06,
      "loss": 0.5467,
      "step": 2190
    },
    {
      "epoch": 0.16075277199482743,
      "grad_norm": 1.4471204223343228,
      "learning_rate": 9.558335637311547e-06,
      "loss": 0.7846,
      "step": 2191
    },
    {
      "epoch": 0.16082614158496658,
      "grad_norm": 1.6965061785829403,
      "learning_rate": 9.557847243447717e-06,
      "loss": 0.8537,
      "step": 2192
    },
    {
      "epoch": 0.16082614158496658,
      "lm_loss": 1.1557,
      "step": 2192,
      "vm_loss": 0.095
    },
    {
      "epoch": 0.16082614158496658,
      "lm_loss": 2.4243,
      "step": 2192,
      "vm_loss": 0.0562
    },
    {
      "epoch": 0.16082614158496658,
      "lm_loss": 1.076,
      "step": 2192,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.16082614158496658,
      "lm_loss": 0.7205,
      "step": 2192,
      "vm_loss": 0.1
    },
    {
      "epoch": 0.16082614158496658,
      "lm_loss": 0.527,
      "step": 2192,
      "vm_loss": 0.1543
    },
    {
      "epoch": 0.16082614158496658,
      "lm_loss": 0.7018,
      "step": 2192,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.16082614158496658,
      "lm_loss": 0.9741,
      "step": 2192,
      "vm_loss": 0.2208
    },
    {
      "epoch": 0.16082614158496658,
      "lm_loss": 1.0384,
      "step": 2192,
      "vm_loss": 0.2503
    },
    {
      "epoch": 0.1608995111751057,
      "grad_norm": 1.6948908476033377,
      "learning_rate": 9.557358592190963e-06,
      "loss": 0.9116,
      "step": 2193
    },
    {
      "epoch": 0.16097288076524482,
      "grad_norm": 0.904694292460374,
      "learning_rate": 9.556869683568889e-06,
      "loss": 0.5341,
      "step": 2194
    },
    {
      "epoch": 0.16104625035538395,
      "grad_norm": 1.7747474221093362,
      "learning_rate": 9.556380517609094e-06,
      "loss": 0.9203,
      "step": 2195
    },
    {
      "epoch": 0.16111961994552307,
      "grad_norm": 1.6920504948994521,
      "learning_rate": 9.555891094339209e-06,
      "loss": 1.0646,
      "step": 2196
    },
    {
      "epoch": 0.16119298953566222,
      "grad_norm": 1.655018084637921,
      "learning_rate": 9.555401413786872e-06,
      "loss": 0.8536,
      "step": 2197
    },
    {
      "epoch": 0.16126635912580134,
      "grad_norm": 1.8149446134480578,
      "learning_rate": 9.554911475979735e-06,
      "loss": 0.8617,
      "step": 2198
    },
    {
      "epoch": 0.16133972871594046,
      "grad_norm": 0.6387514726298024,
      "learning_rate": 9.554421280945467e-06,
      "loss": 0.5639,
      "step": 2199
    },
    {
      "epoch": 0.16141309830607958,
      "grad_norm": 0.6073539306495023,
      "learning_rate": 9.55393082871175e-06,
      "loss": 0.5509,
      "step": 2200
    },
    {
      "epoch": 0.16141309830607958,
      "lm_loss": 1.237,
      "step": 2200,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.16141309830607958,
      "lm_loss": 1.5184,
      "step": 2200,
      "vm_loss": 0.1625
    },
    {
      "epoch": 0.16141309830607958,
      "lm_loss": 0.9405,
      "step": 2200,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.16141309830607958,
      "lm_loss": 1.0346,
      "step": 2200,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.16141309830607958,
      "lm_loss": 0.7567,
      "step": 2200,
      "vm_loss": 0.1341
    },
    {
      "epoch": 0.16141309830607958,
      "lm_loss": 0.8899,
      "step": 2200,
      "vm_loss": 0.1579
    },
    {
      "epoch": 0.16141309830607958,
      "lm_loss": 0.5903,
      "step": 2200,
      "vm_loss": 0.1252
    },
    {
      "epoch": 0.16141309830607958,
      "lm_loss": 0.0781,
      "step": 2200,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.1614864678962187,
      "grad_norm": 1.9583207565031726,
      "learning_rate": 9.553440119306283e-06,
      "loss": 0.966,
      "step": 2201
    },
    {
      "epoch": 0.16155983748635785,
      "grad_norm": 1.4758304948403351,
      "learning_rate": 9.552949152756775e-06,
      "loss": 0.8322,
      "step": 2202
    },
    {
      "epoch": 0.16163320707649698,
      "grad_norm": 1.5842527402791844,
      "learning_rate": 9.552457929090952e-06,
      "loss": 0.9248,
      "step": 2203
    },
    {
      "epoch": 0.1617065766666361,
      "grad_norm": 0.5996874944397917,
      "learning_rate": 9.551966448336558e-06,
      "loss": 0.5523,
      "step": 2204
    },
    {
      "epoch": 0.16177994625677522,
      "grad_norm": 1.6938137956475217,
      "learning_rate": 9.551474710521345e-06,
      "loss": 0.9443,
      "step": 2205
    },
    {
      "epoch": 0.16185331584691434,
      "grad_norm": 1.8718736320827527,
      "learning_rate": 9.550982715673084e-06,
      "loss": 0.8685,
      "step": 2206
    },
    {
      "epoch": 0.16192668543705346,
      "grad_norm": 1.454347052754901,
      "learning_rate": 9.550490463819559e-06,
      "loss": 0.7908,
      "step": 2207
    },
    {
      "epoch": 0.1620000550271926,
      "grad_norm": 1.8949893586922413,
      "learning_rate": 9.549997954988567e-06,
      "loss": 0.9517,
      "step": 2208
    },
    {
      "epoch": 0.1620000550271926,
      "lm_loss": 1.6171,
      "step": 2208,
      "vm_loss": 0.0517
    },
    {
      "epoch": 0.1620000550271926,
      "lm_loss": 0.7084,
      "step": 2208,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.1620000550271926,
      "lm_loss": 0.6152,
      "step": 2208,
      "vm_loss": 0.1482
    },
    {
      "epoch": 0.1620000550271926,
      "lm_loss": 0.0744,
      "step": 2208,
      "vm_loss": 0.1074
    },
    {
      "epoch": 0.1620000550271926,
      "lm_loss": 0.7071,
      "step": 2208,
      "vm_loss": 0.1085
    },
    {
      "epoch": 0.1620000550271926,
      "lm_loss": 0.3578,
      "step": 2208,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.1620000550271926,
      "lm_loss": 1.3267,
      "step": 2208,
      "vm_loss": 0.185
    },
    {
      "epoch": 0.1620000550271926,
      "lm_loss": 1.3889,
      "step": 2208,
      "vm_loss": 0.1556
    },
    {
      "epoch": 0.16207342461733173,
      "grad_norm": 1.6286514863402224,
      "learning_rate": 9.549505189207924e-06,
      "loss": 0.815,
      "step": 2209
    },
    {
      "epoch": 0.16214679420747086,
      "grad_norm": 0.7028384473100328,
      "learning_rate": 9.549012166505455e-06,
      "loss": 0.5597,
      "step": 2210
    },
    {
      "epoch": 0.16222016379760998,
      "grad_norm": 0.6377155482641548,
      "learning_rate": 9.548518886909005e-06,
      "loss": 0.5437,
      "step": 2211
    },
    {
      "epoch": 0.1622935333877491,
      "grad_norm": 1.9240403971761169,
      "learning_rate": 9.548025350446427e-06,
      "loss": 0.9077,
      "step": 2212
    },
    {
      "epoch": 0.16236690297788825,
      "grad_norm": 0.5904147433009326,
      "learning_rate": 9.547531557145597e-06,
      "loss": 0.5437,
      "step": 2213
    },
    {
      "epoch": 0.16244027256802737,
      "grad_norm": 0.5724711963044239,
      "learning_rate": 9.547037507034396e-06,
      "loss": 0.5488,
      "step": 2214
    },
    {
      "epoch": 0.1625136421581665,
      "grad_norm": 1.9355166184304622,
      "learning_rate": 9.546543200140728e-06,
      "loss": 0.9128,
      "step": 2215
    },
    {
      "epoch": 0.16258701174830562,
      "grad_norm": 1.752235043298254,
      "learning_rate": 9.546048636492506e-06,
      "loss": 0.8505,
      "step": 2216
    },
    {
      "epoch": 0.16258701174830562,
      "lm_loss": 1.5531,
      "step": 2216,
      "vm_loss": 0.0711
    },
    {
      "epoch": 0.16258701174830562,
      "lm_loss": 0.1745,
      "step": 2216,
      "vm_loss": 0.1262
    },
    {
      "epoch": 0.16258701174830562,
      "lm_loss": 1.2705,
      "step": 2216,
      "vm_loss": 0.0764
    },
    {
      "epoch": 0.16258701174830562,
      "lm_loss": 0.7922,
      "step": 2216,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.16258701174830562,
      "lm_loss": 0.6217,
      "step": 2216,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.16258701174830562,
      "lm_loss": 1.0703,
      "step": 2216,
      "vm_loss": 0.0991
    },
    {
      "epoch": 0.16258701174830562,
      "lm_loss": 1.1771,
      "step": 2216,
      "vm_loss": 0.1298
    },
    {
      "epoch": 0.16258701174830562,
      "lm_loss": 0.2504,
      "step": 2216,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.16266038133844474,
      "grad_norm": 1.7351215828858242,
      "learning_rate": 9.545553816117658e-06,
      "loss": 0.8417,
      "step": 2217
    },
    {
      "epoch": 0.1627337509285839,
      "grad_norm": 1.6333471697012993,
      "learning_rate": 9.54505873904413e-06,
      "loss": 0.8493,
      "step": 2218
    },
    {
      "epoch": 0.162807120518723,
      "grad_norm": 1.551557253043814,
      "learning_rate": 9.54456340529988e-06,
      "loss": 0.912,
      "step": 2219
    },
    {
      "epoch": 0.16288049010886213,
      "grad_norm": 1.661860760418264,
      "learning_rate": 9.544067814912879e-06,
      "loss": 0.9972,
      "step": 2220
    },
    {
      "epoch": 0.16295385969900125,
      "grad_norm": 0.8216905583389849,
      "learning_rate": 9.543571967911115e-06,
      "loss": 0.5571,
      "step": 2221
    },
    {
      "epoch": 0.16302722928914037,
      "grad_norm": 0.7246962873175001,
      "learning_rate": 9.543075864322592e-06,
      "loss": 0.5407,
      "step": 2222
    },
    {
      "epoch": 0.16310059887927952,
      "grad_norm": 0.6291044403444087,
      "learning_rate": 9.542579504175325e-06,
      "loss": 0.5676,
      "step": 2223
    },
    {
      "epoch": 0.16317396846941865,
      "grad_norm": 0.6011796380528251,
      "learning_rate": 9.542082887497341e-06,
      "loss": 0.5528,
      "step": 2224
    },
    {
      "epoch": 0.16317396846941865,
      "lm_loss": 1.4322,
      "step": 2224,
      "vm_loss": 0.1489
    },
    {
      "epoch": 0.16317396846941865,
      "lm_loss": 1.1724,
      "step": 2224,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.16317396846941865,
      "lm_loss": 0.0272,
      "step": 2224,
      "vm_loss": 0.1412
    },
    {
      "epoch": 0.16317396846941865,
      "lm_loss": 0.2837,
      "step": 2224,
      "vm_loss": 0.1602
    },
    {
      "epoch": 0.16317396846941865,
      "lm_loss": 0.6101,
      "step": 2224,
      "vm_loss": 0.0609
    },
    {
      "epoch": 0.16317396846941865,
      "lm_loss": 0.8929,
      "step": 2224,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.16317396846941865,
      "lm_loss": 0.6158,
      "step": 2224,
      "vm_loss": 0.1547
    },
    {
      "epoch": 0.16317396846941865,
      "lm_loss": 0.8703,
      "step": 2224,
      "vm_loss": 0.2013
    },
    {
      "epoch": 0.16324733805955777,
      "grad_norm": 2.003256293461366,
      "learning_rate": 9.541586014316688e-06,
      "loss": 0.882,
      "step": 2225
    },
    {
      "epoch": 0.1633207076496969,
      "grad_norm": 1.7380362664654425,
      "learning_rate": 9.541088884661427e-06,
      "loss": 0.931,
      "step": 2226
    },
    {
      "epoch": 0.163394077239836,
      "grad_norm": 1.7001301767355814,
      "learning_rate": 9.540591498559632e-06,
      "loss": 0.7642,
      "step": 2227
    },
    {
      "epoch": 0.16346744682997513,
      "grad_norm": 0.8754138167595712,
      "learning_rate": 9.540093856039389e-06,
      "loss": 0.5445,
      "step": 2228
    },
    {
      "epoch": 0.16354081642011428,
      "grad_norm": 0.7920868319996616,
      "learning_rate": 9.539595957128803e-06,
      "loss": 0.5665,
      "step": 2229
    },
    {
      "epoch": 0.1636141860102534,
      "grad_norm": 0.6512797399991396,
      "learning_rate": 9.53909780185599e-06,
      "loss": 0.5563,
      "step": 2230
    },
    {
      "epoch": 0.16368755560039253,
      "grad_norm": 2.3807200225963787,
      "learning_rate": 9.538599390249085e-06,
      "loss": 0.8829,
      "step": 2231
    },
    {
      "epoch": 0.16376092519053165,
      "grad_norm": 2.127589196886961,
      "learning_rate": 9.538100722336232e-06,
      "loss": 0.9616,
      "step": 2232
    },
    {
      "epoch": 0.16376092519053165,
      "lm_loss": 0.3006,
      "step": 2232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16376092519053165,
      "lm_loss": 0.6501,
      "step": 2232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16376092519053165,
      "lm_loss": 0.4512,
      "step": 2232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16376092519053165,
      "lm_loss": 0.4998,
      "step": 2232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16376092519053165,
      "lm_loss": 0.587,
      "step": 2232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16376092519053165,
      "lm_loss": 0.5044,
      "step": 2232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16376092519053165,
      "lm_loss": 0.5185,
      "step": 2232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16376092519053165,
      "lm_loss": 0.5404,
      "step": 2232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16383429478067077,
      "grad_norm": 0.7694038934533066,
      "learning_rate": 9.537601798145593e-06,
      "loss": 0.5416,
      "step": 2233
    },
    {
      "epoch": 0.16390766437080992,
      "grad_norm": 1.7012686695749237,
      "learning_rate": 9.537102617705342e-06,
      "loss": 0.8558,
      "step": 2234
    },
    {
      "epoch": 0.16398103396094904,
      "grad_norm": 1.8027140161356383,
      "learning_rate": 9.536603181043671e-06,
      "loss": 0.927,
      "step": 2235
    },
    {
      "epoch": 0.16405440355108816,
      "grad_norm": 1.6551561870362033,
      "learning_rate": 9.536103488188784e-06,
      "loss": 0.8445,
      "step": 2236
    },
    {
      "epoch": 0.16412777314122728,
      "grad_norm": 0.7766150919917792,
      "learning_rate": 9.535603539168901e-06,
      "loss": 0.5524,
      "step": 2237
    },
    {
      "epoch": 0.1642011427313664,
      "grad_norm": 0.7367043878356313,
      "learning_rate": 9.535103334012251e-06,
      "loss": 0.5519,
      "step": 2238
    },
    {
      "epoch": 0.16427451232150556,
      "grad_norm": 1.753407109006742,
      "learning_rate": 9.534602872747086e-06,
      "loss": 0.8632,
      "step": 2239
    },
    {
      "epoch": 0.16434788191164468,
      "grad_norm": 1.5681636117846816,
      "learning_rate": 9.534102155401668e-06,
      "loss": 0.9234,
      "step": 2240
    },
    {
      "epoch": 0.16434788191164468,
      "lm_loss": 1.1884,
      "step": 2240,
      "vm_loss": 0.1198
    },
    {
      "epoch": 0.16434788191164468,
      "lm_loss": 0.4533,
      "step": 2240,
      "vm_loss": 0.2003
    },
    {
      "epoch": 0.16434788191164468,
      "lm_loss": 1.0032,
      "step": 2240,
      "vm_loss": 0.0567
    },
    {
      "epoch": 0.16434788191164468,
      "lm_loss": 0.8468,
      "step": 2240,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.16434788191164468,
      "lm_loss": 0.7306,
      "step": 2240,
      "vm_loss": 0.1646
    },
    {
      "epoch": 0.16434788191164468,
      "lm_loss": 1.3815,
      "step": 2240,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.16434788191164468,
      "lm_loss": 0.308,
      "step": 2240,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.16434788191164468,
      "lm_loss": 0.1426,
      "step": 2240,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.1644212515017838,
      "grad_norm": 1.659854484746543,
      "learning_rate": 9.533601182004272e-06,
      "loss": 0.8706,
      "step": 2241
    },
    {
      "epoch": 0.16449462109192292,
      "grad_norm": 1.5506567711435304,
      "learning_rate": 9.533099952583189e-06,
      "loss": 0.8669,
      "step": 2242
    },
    {
      "epoch": 0.16456799068206204,
      "grad_norm": 1.7955557860352296,
      "learning_rate": 9.532598467166728e-06,
      "loss": 0.8104,
      "step": 2243
    },
    {
      "epoch": 0.1646413602722012,
      "grad_norm": 1.71023284672044,
      "learning_rate": 9.532096725783204e-06,
      "loss": 0.959,
      "step": 2244
    },
    {
      "epoch": 0.16471472986234031,
      "grad_norm": 1.7662553010158688,
      "learning_rate": 9.531594728460956e-06,
      "loss": 0.8722,
      "step": 2245
    },
    {
      "epoch": 0.16478809945247944,
      "grad_norm": 1.6839899177616446,
      "learning_rate": 9.53109247522833e-06,
      "loss": 0.8645,
      "step": 2246
    },
    {
      "epoch": 0.16486146904261856,
      "grad_norm": 1.457417770447728,
      "learning_rate": 9.530589966113692e-06,
      "loss": 0.8411,
      "step": 2247
    },
    {
      "epoch": 0.16493483863275768,
      "grad_norm": 1.6254490451052515,
      "learning_rate": 9.530087201145419e-06,
      "loss": 0.7121,
      "step": 2248
    },
    {
      "epoch": 0.16493483863275768,
      "lm_loss": 0.9778,
      "step": 2248,
      "vm_loss": 0.0708
    },
    {
      "epoch": 0.16493483863275768,
      "lm_loss": 0.9233,
      "step": 2248,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.16493483863275768,
      "lm_loss": 1.0923,
      "step": 2248,
      "vm_loss": 0.1505
    },
    {
      "epoch": 0.16493483863275768,
      "lm_loss": 0.9419,
      "step": 2248,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.16493483863275768,
      "lm_loss": 0.5085,
      "step": 2248,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.16493483863275768,
      "lm_loss": 0.3091,
      "step": 2248,
      "vm_loss": 0.0755
    },
    {
      "epoch": 0.16493483863275768,
      "lm_loss": 0.9907,
      "step": 2248,
      "vm_loss": 0.201
    },
    {
      "epoch": 0.16493483863275768,
      "lm_loss": 1.1039,
      "step": 2248,
      "vm_loss": 0.2119
    },
    {
      "epoch": 0.1650082082228968,
      "grad_norm": 1.5833049236291914,
      "learning_rate": 9.529584180351902e-06,
      "loss": 0.8262,
      "step": 2249
    },
    {
      "epoch": 0.16508157781303595,
      "grad_norm": 1.527373395879894,
      "learning_rate": 9.529080903761551e-06,
      "loss": 0.7754,
      "step": 2250
    },
    {
      "epoch": 0.16515494740317507,
      "grad_norm": 1.6041128681415353,
      "learning_rate": 9.528577371402784e-06,
      "loss": 0.9075,
      "step": 2251
    },
    {
      "epoch": 0.1652283169933142,
      "grad_norm": 1.8521824367276143,
      "learning_rate": 9.528073583304039e-06,
      "loss": 0.8687,
      "step": 2252
    },
    {
      "epoch": 0.16530168658345332,
      "grad_norm": 1.5602958102642477,
      "learning_rate": 9.527569539493763e-06,
      "loss": 0.7704,
      "step": 2253
    },
    {
      "epoch": 0.16537505617359244,
      "grad_norm": 1.534213488714071,
      "learning_rate": 9.527065240000424e-06,
      "loss": 0.9705,
      "step": 2254
    },
    {
      "epoch": 0.1654484257637316,
      "grad_norm": 1.5470869328809094,
      "learning_rate": 9.526560684852501e-06,
      "loss": 0.8559,
      "step": 2255
    },
    {
      "epoch": 0.1655217953538707,
      "grad_norm": 1.6581833436165054,
      "learning_rate": 9.526055874078486e-06,
      "loss": 0.8737,
      "step": 2256
    },
    {
      "epoch": 0.1655217953538707,
      "lm_loss": 1.5922,
      "step": 2256,
      "vm_loss": 0.0572
    },
    {
      "epoch": 0.1655217953538707,
      "lm_loss": 0.0394,
      "step": 2256,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.1655217953538707,
      "lm_loss": 0.8619,
      "step": 2256,
      "vm_loss": 0.2567
    },
    {
      "epoch": 0.1655217953538707,
      "lm_loss": 0.9962,
      "step": 2256,
      "vm_loss": 0.1682
    },
    {
      "epoch": 0.1655217953538707,
      "lm_loss": 1.1478,
      "step": 2256,
      "vm_loss": 0.1405
    },
    {
      "epoch": 0.1655217953538707,
      "lm_loss": 0.5193,
      "step": 2256,
      "vm_loss": 0.1135
    },
    {
      "epoch": 0.1655217953538707,
      "lm_loss": 0.6826,
      "step": 2256,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.1655217953538707,
      "lm_loss": 0.8475,
      "step": 2256,
      "vm_loss": 0.326
    },
    {
      "epoch": 0.16559516494400983,
      "grad_norm": 1.8335355456957159,
      "learning_rate": 9.525550807706885e-06,
      "loss": 1.0118,
      "step": 2257
    },
    {
      "epoch": 0.16566853453414895,
      "grad_norm": 1.2239755986622893,
      "learning_rate": 9.525045485766225e-06,
      "loss": 0.5696,
      "step": 2258
    },
    {
      "epoch": 0.16574190412428808,
      "grad_norm": 0.8651554573618588,
      "learning_rate": 9.524539908285039e-06,
      "loss": 0.5681,
      "step": 2259
    },
    {
      "epoch": 0.16581527371442722,
      "grad_norm": 1.9936584109877535,
      "learning_rate": 9.52403407529188e-06,
      "loss": 0.9117,
      "step": 2260
    },
    {
      "epoch": 0.16588864330456635,
      "grad_norm": 1.8748101902344336,
      "learning_rate": 9.523527986815313e-06,
      "loss": 0.8714,
      "step": 2261
    },
    {
      "epoch": 0.16596201289470547,
      "grad_norm": 1.4924336184645035,
      "learning_rate": 9.523021642883918e-06,
      "loss": 0.8415,
      "step": 2262
    },
    {
      "epoch": 0.1660353824848446,
      "grad_norm": 2.2274136151973516,
      "learning_rate": 9.52251504352629e-06,
      "loss": 0.8956,
      "step": 2263
    },
    {
      "epoch": 0.1661087520749837,
      "grad_norm": 1.2927608716866896,
      "learning_rate": 9.522008188771036e-06,
      "loss": 0.5553,
      "step": 2264
    },
    {
      "epoch": 0.1661087520749837,
      "lm_loss": 0.5929,
      "step": 2264,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.1661087520749837,
      "lm_loss": 0.8125,
      "step": 2264,
      "vm_loss": 0.0716
    },
    {
      "epoch": 0.1661087520749837,
      "lm_loss": 0.2237,
      "step": 2264,
      "vm_loss": 0.0946
    },
    {
      "epoch": 0.1661087520749837,
      "lm_loss": 1.0735,
      "step": 2264,
      "vm_loss": 0.2021
    },
    {
      "epoch": 0.1661087520749837,
      "lm_loss": 0.4043,
      "step": 2264,
      "vm_loss": 0.1878
    },
    {
      "epoch": 0.1661087520749837,
      "lm_loss": 0.6911,
      "step": 2264,
      "vm_loss": 0.2016
    },
    {
      "epoch": 0.1661087520749837,
      "lm_loss": 1.36,
      "step": 2264,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.1661087520749837,
      "lm_loss": 0.3951,
      "step": 2264,
      "vm_loss": 0.2472
    },
    {
      "epoch": 0.16618212166512286,
      "grad_norm": 2.253580409843883,
      "learning_rate": 9.521501078646783e-06,
      "loss": 0.9134,
      "step": 2265
    },
    {
      "epoch": 0.16625549125526198,
      "grad_norm": 1.66674571206447,
      "learning_rate": 9.520993713182167e-06,
      "loss": 0.8685,
      "step": 2266
    },
    {
      "epoch": 0.1663288608454011,
      "grad_norm": 1.837184182020013,
      "learning_rate": 9.520486092405841e-06,
      "loss": 0.9226,
      "step": 2267
    },
    {
      "epoch": 0.16640223043554023,
      "grad_norm": 1.6002993514415211,
      "learning_rate": 9.519978216346469e-06,
      "loss": 0.9049,
      "step": 2268
    },
    {
      "epoch": 0.16647560002567935,
      "grad_norm": 1.8160110938989917,
      "learning_rate": 9.519470085032733e-06,
      "loss": 0.8664,
      "step": 2269
    },
    {
      "epoch": 0.16654896961581847,
      "grad_norm": 1.893927296637265,
      "learning_rate": 9.51896169849333e-06,
      "loss": 0.7958,
      "step": 2270
    },
    {
      "epoch": 0.16662233920595762,
      "grad_norm": 0.8562000914643925,
      "learning_rate": 9.51845305675697e-06,
      "loss": 0.5666,
      "step": 2271
    },
    {
      "epoch": 0.16669570879609674,
      "grad_norm": 1.7682276069736127,
      "learning_rate": 9.517944159852372e-06,
      "loss": 0.9225,
      "step": 2272
    },
    {
      "epoch": 0.16669570879609674,
      "lm_loss": 1.707,
      "step": 2272,
      "vm_loss": 0.0538
    },
    {
      "epoch": 0.16669570879609674,
      "lm_loss": 0.2771,
      "step": 2272,
      "vm_loss": 0.0771
    },
    {
      "epoch": 0.16669570879609674,
      "lm_loss": 0.7912,
      "step": 2272,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.16669570879609674,
      "lm_loss": 0.0899,
      "step": 2272,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.16669570879609674,
      "lm_loss": 0.5829,
      "step": 2272,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.16669570879609674,
      "lm_loss": 1.2802,
      "step": 2272,
      "vm_loss": 0.0946
    },
    {
      "epoch": 0.16669570879609674,
      "lm_loss": 0.3552,
      "step": 2272,
      "vm_loss": 0.1392
    },
    {
      "epoch": 0.16669570879609674,
      "lm_loss": 0.7996,
      "step": 2272,
      "vm_loss": 0.1767
    },
    {
      "epoch": 0.16676907838623586,
      "grad_norm": 1.5619420223890186,
      "learning_rate": 9.517435007808282e-06,
      "loss": 0.8553,
      "step": 2273
    },
    {
      "epoch": 0.16684244797637499,
      "grad_norm": 1.532130970424405,
      "learning_rate": 9.516925600653449e-06,
      "loss": 0.86,
      "step": 2274
    },
    {
      "epoch": 0.1669158175665141,
      "grad_norm": 1.687521591425339,
      "learning_rate": 9.516415938416642e-06,
      "loss": 0.8722,
      "step": 2275
    },
    {
      "epoch": 0.16698918715665326,
      "grad_norm": 1.7264354879078203,
      "learning_rate": 9.515906021126641e-06,
      "loss": 0.8592,
      "step": 2276
    },
    {
      "epoch": 0.16706255674679238,
      "grad_norm": 1.7104139404536804,
      "learning_rate": 9.515395848812245e-06,
      "loss": 0.953,
      "step": 2277
    },
    {
      "epoch": 0.1671359263369315,
      "grad_norm": 0.7765964984225379,
      "learning_rate": 9.514885421502263e-06,
      "loss": 0.5643,
      "step": 2278
    },
    {
      "epoch": 0.16720929592707062,
      "grad_norm": 0.7030646263943905,
      "learning_rate": 9.51437473922552e-06,
      "loss": 0.5372,
      "step": 2279
    },
    {
      "epoch": 0.16728266551720974,
      "grad_norm": 0.5963221316712989,
      "learning_rate": 9.513863802010857e-06,
      "loss": 0.5459,
      "step": 2280
    },
    {
      "epoch": 0.16728266551720974,
      "lm_loss": 2.2332,
      "step": 2280,
      "vm_loss": 0.0484
    },
    {
      "epoch": 0.16728266551720974,
      "lm_loss": 1.5709,
      "step": 2280,
      "vm_loss": 0.091
    },
    {
      "epoch": 0.16728266551720974,
      "lm_loss": 0.2363,
      "step": 2280,
      "vm_loss": 0.1217
    },
    {
      "epoch": 0.16728266551720974,
      "lm_loss": 1.44,
      "step": 2280,
      "vm_loss": 0.251
    },
    {
      "epoch": 0.16728266551720974,
      "lm_loss": 1.5472,
      "step": 2280,
      "vm_loss": 0.1544
    },
    {
      "epoch": 0.16728266551720974,
      "lm_loss": 0.5159,
      "step": 2280,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.16728266551720974,
      "lm_loss": 0.656,
      "step": 2280,
      "vm_loss": 0.0708
    },
    {
      "epoch": 0.16728266551720974,
      "lm_loss": 0.2391,
      "step": 2280,
      "vm_loss": 0.122
    },
    {
      "epoch": 0.1673560351073489,
      "grad_norm": 1.6724768381743766,
      "learning_rate": 9.513352609887124e-06,
      "loss": 0.9025,
      "step": 2281
    },
    {
      "epoch": 0.16742940469748802,
      "grad_norm": 1.6247204384537681,
      "learning_rate": 9.512841162883194e-06,
      "loss": 0.8717,
      "step": 2282
    },
    {
      "epoch": 0.16750277428762714,
      "grad_norm": 1.6156153484631586,
      "learning_rate": 9.512329461027947e-06,
      "loss": 0.8881,
      "step": 2283
    },
    {
      "epoch": 0.16757614387776626,
      "grad_norm": 1.6244004226184194,
      "learning_rate": 9.51181750435028e-06,
      "loss": 0.9069,
      "step": 2284
    },
    {
      "epoch": 0.16764951346790538,
      "grad_norm": 1.1284258308517798,
      "learning_rate": 9.511305292879108e-06,
      "loss": 0.5746,
      "step": 2285
    },
    {
      "epoch": 0.16772288305804453,
      "grad_norm": 0.8232033613703826,
      "learning_rate": 9.510792826643352e-06,
      "loss": 0.5558,
      "step": 2286
    },
    {
      "epoch": 0.16779625264818365,
      "grad_norm": 1.8329579253932484,
      "learning_rate": 9.510280105671954e-06,
      "loss": 0.8132,
      "step": 2287
    },
    {
      "epoch": 0.16786962223832277,
      "grad_norm": 0.6725032355993829,
      "learning_rate": 9.509767129993868e-06,
      "loss": 0.5393,
      "step": 2288
    },
    {
      "epoch": 0.16786962223832277,
      "lm_loss": 0.4822,
      "step": 2288,
      "vm_loss": 0.0534
    },
    {
      "epoch": 0.16786962223832277,
      "lm_loss": 1.7649,
      "step": 2288,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.16786962223832277,
      "lm_loss": 1.109,
      "step": 2288,
      "vm_loss": 0.1328
    },
    {
      "epoch": 0.16786962223832277,
      "lm_loss": 0.6866,
      "step": 2288,
      "vm_loss": 0.135
    },
    {
      "epoch": 0.16786962223832277,
      "lm_loss": 0.7155,
      "step": 2288,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.16786962223832277,
      "lm_loss": 0.0956,
      "step": 2288,
      "vm_loss": 0.1204
    },
    {
      "epoch": 0.16786962223832277,
      "lm_loss": 1.1505,
      "step": 2288,
      "vm_loss": 0.1728
    },
    {
      "epoch": 0.16786962223832277,
      "lm_loss": 0.2268,
      "step": 2288,
      "vm_loss": 0.0948
    },
    {
      "epoch": 0.1679429918284619,
      "grad_norm": 1.7457609963042362,
      "learning_rate": 9.509253899638066e-06,
      "loss": 0.9172,
      "step": 2289
    },
    {
      "epoch": 0.16801636141860102,
      "grad_norm": 1.8831058847808906,
      "learning_rate": 9.508740414633528e-06,
      "loss": 0.9067,
      "step": 2290
    },
    {
      "epoch": 0.16808973100874014,
      "grad_norm": 0.9856240073671737,
      "learning_rate": 9.508226675009254e-06,
      "loss": 0.5483,
      "step": 2291
    },
    {
      "epoch": 0.1681631005988793,
      "grad_norm": 1.4300644524715054,
      "learning_rate": 9.507712680794256e-06,
      "loss": 0.8808,
      "step": 2292
    },
    {
      "epoch": 0.1682364701890184,
      "grad_norm": 1.5783029671712412,
      "learning_rate": 9.507198432017558e-06,
      "loss": 0.8119,
      "step": 2293
    },
    {
      "epoch": 0.16830983977915753,
      "grad_norm": 1.6496651070231276,
      "learning_rate": 9.506683928708203e-06,
      "loss": 0.8441,
      "step": 2294
    },
    {
      "epoch": 0.16838320936929665,
      "grad_norm": 1.6536733485344774,
      "learning_rate": 9.506169170895246e-06,
      "loss": 0.9134,
      "step": 2295
    },
    {
      "epoch": 0.16845657895943578,
      "grad_norm": 1.5661797939470792,
      "learning_rate": 9.505654158607758e-06,
      "loss": 0.896,
      "step": 2296
    },
    {
      "epoch": 0.16845657895943578,
      "lm_loss": 0.6715,
      "step": 2296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16845657895943578,
      "lm_loss": 0.5232,
      "step": 2296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16845657895943578,
      "lm_loss": 0.5775,
      "step": 2296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16845657895943578,
      "lm_loss": 0.5193,
      "step": 2296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16845657895943578,
      "lm_loss": 0.6687,
      "step": 2296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16845657895943578,
      "lm_loss": 0.4868,
      "step": 2296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16845657895943578,
      "lm_loss": 0.6081,
      "step": 2296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16845657895943578,
      "lm_loss": 0.5227,
      "step": 2296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.16852994854957493,
      "grad_norm": 0.7233637042234161,
      "learning_rate": 9.50513889187482e-06,
      "loss": 0.5802,
      "step": 2297
    },
    {
      "epoch": 0.16860331813971405,
      "grad_norm": 1.5820988658970216,
      "learning_rate": 9.504623370725531e-06,
      "loss": 0.898,
      "step": 2298
    },
    {
      "epoch": 0.16867668772985317,
      "grad_norm": 1.5787409498041913,
      "learning_rate": 9.504107595189007e-06,
      "loss": 0.7969,
      "step": 2299
    },
    {
      "epoch": 0.1687500573199923,
      "grad_norm": 0.6123732827438322,
      "learning_rate": 9.503591565294372e-06,
      "loss": 0.5644,
      "step": 2300
    },
    {
      "epoch": 0.1688234269101314,
      "grad_norm": 0.5772213266245365,
      "learning_rate": 9.503075281070769e-06,
      "loss": 0.5443,
      "step": 2301
    },
    {
      "epoch": 0.16889679650027056,
      "grad_norm": 1.663596122553907,
      "learning_rate": 9.502558742547352e-06,
      "loss": 0.7246,
      "step": 2302
    },
    {
      "epoch": 0.16897016609040968,
      "grad_norm": 0.6238213384110279,
      "learning_rate": 9.502041949753294e-06,
      "loss": 0.5423,
      "step": 2303
    },
    {
      "epoch": 0.1690435356805488,
      "grad_norm": 1.7580126530369318,
      "learning_rate": 9.501524902717775e-06,
      "loss": 0.8775,
      "step": 2304
    },
    {
      "epoch": 0.1690435356805488,
      "lm_loss": 1.2168,
      "step": 2304,
      "vm_loss": 0.0974
    },
    {
      "epoch": 0.1690435356805488,
      "lm_loss": 0.2293,
      "step": 2304,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.1690435356805488,
      "lm_loss": 0.8689,
      "step": 2304,
      "vm_loss": 0.1579
    },
    {
      "epoch": 0.1690435356805488,
      "lm_loss": 0.943,
      "step": 2304,
      "vm_loss": 0.1666
    },
    {
      "epoch": 0.1690435356805488,
      "lm_loss": 0.7044,
      "step": 2304,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.1690435356805488,
      "lm_loss": 0.2876,
      "step": 2304,
      "vm_loss": 0.1524
    },
    {
      "epoch": 0.1690435356805488,
      "lm_loss": 0.4716,
      "step": 2304,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.1690435356805488,
      "lm_loss": 0.7696,
      "step": 2304,
      "vm_loss": 0.1633
    },
    {
      "epoch": 0.16911690527068793,
      "grad_norm": 1.6579066521346695,
      "learning_rate": 9.50100760147e-06,
      "loss": 0.8336,
      "step": 2305
    },
    {
      "epoch": 0.16919027486082705,
      "grad_norm": 1.3750079633777434,
      "learning_rate": 9.500490046039177e-06,
      "loss": 0.7918,
      "step": 2306
    },
    {
      "epoch": 0.16926364445096617,
      "grad_norm": 0.6989991840732959,
      "learning_rate": 9.499972236454538e-06,
      "loss": 0.5548,
      "step": 2307
    },
    {
      "epoch": 0.16933701404110532,
      "grad_norm": 1.677942391549459,
      "learning_rate": 9.499454172745321e-06,
      "loss": 0.8267,
      "step": 2308
    },
    {
      "epoch": 0.16941038363124444,
      "grad_norm": 0.616357526570089,
      "learning_rate": 9.498935854940785e-06,
      "loss": 0.5474,
      "step": 2309
    },
    {
      "epoch": 0.16948375322138357,
      "grad_norm": 1.6161000797182135,
      "learning_rate": 9.4984172830702e-06,
      "loss": 0.8819,
      "step": 2310
    },
    {
      "epoch": 0.1695571228115227,
      "grad_norm": 1.6163919683004575,
      "learning_rate": 9.49789845716285e-06,
      "loss": 0.8539,
      "step": 2311
    },
    {
      "epoch": 0.1696304924016618,
      "grad_norm": 1.5972279248303547,
      "learning_rate": 9.497379377248037e-06,
      "loss": 0.8731,
      "step": 2312
    },
    {
      "epoch": 0.1696304924016618,
      "lm_loss": 0.5146,
      "step": 2312,
      "vm_loss": 0.0589
    },
    {
      "epoch": 0.1696304924016618,
      "lm_loss": 1.3176,
      "step": 2312,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.1696304924016618,
      "lm_loss": 0.1306,
      "step": 2312,
      "vm_loss": 0.0482
    },
    {
      "epoch": 0.1696304924016618,
      "lm_loss": 1.2754,
      "step": 2312,
      "vm_loss": 0.1672
    },
    {
      "epoch": 0.1696304924016618,
      "lm_loss": 1.0053,
      "step": 2312,
      "vm_loss": 0.1857
    },
    {
      "epoch": 0.1696304924016618,
      "lm_loss": 0.3479,
      "step": 2312,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.1696304924016618,
      "lm_loss": 0.7316,
      "step": 2312,
      "vm_loss": 0.1185
    },
    {
      "epoch": 0.1696304924016618,
      "lm_loss": 0.1563,
      "step": 2312,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.16970386199180096,
      "grad_norm": 1.424432211220731,
      "learning_rate": 9.496860043355072e-06,
      "loss": 0.8224,
      "step": 2313
    },
    {
      "epoch": 0.16977723158194008,
      "grad_norm": 1.81688430539798,
      "learning_rate": 9.496340455513282e-06,
      "loss": 0.876,
      "step": 2314
    },
    {
      "epoch": 0.1698506011720792,
      "grad_norm": 0.7078596124876723,
      "learning_rate": 9.495820613752014e-06,
      "loss": 0.5469,
      "step": 2315
    },
    {
      "epoch": 0.16992397076221832,
      "grad_norm": 1.4901651396256568,
      "learning_rate": 9.495300518100622e-06,
      "loss": 0.8591,
      "step": 2316
    },
    {
      "epoch": 0.16999734035235745,
      "grad_norm": 1.6060219437675765,
      "learning_rate": 9.494780168588477e-06,
      "loss": 0.8587,
      "step": 2317
    },
    {
      "epoch": 0.1700707099424966,
      "grad_norm": 1.6129442526814532,
      "learning_rate": 9.494259565244965e-06,
      "loss": 0.9094,
      "step": 2318
    },
    {
      "epoch": 0.17014407953263572,
      "grad_norm": 1.725517321645079,
      "learning_rate": 9.493738708099485e-06,
      "loss": 0.9805,
      "step": 2319
    },
    {
      "epoch": 0.17021744912277484,
      "grad_norm": 1.5778649691424635,
      "learning_rate": 9.49321759718145e-06,
      "loss": 0.8778,
      "step": 2320
    },
    {
      "epoch": 0.17021744912277484,
      "lm_loss": 0.4228,
      "step": 2320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17021744912277484,
      "lm_loss": 0.6695,
      "step": 2320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17021744912277484,
      "lm_loss": 0.4835,
      "step": 2320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17021744912277484,
      "lm_loss": 0.6921,
      "step": 2320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17021744912277484,
      "lm_loss": 0.6465,
      "step": 2320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17021744912277484,
      "lm_loss": 0.5209,
      "step": 2320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17021744912277484,
      "lm_loss": 0.59,
      "step": 2320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17021744912277484,
      "lm_loss": 0.5583,
      "step": 2320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17029081871291396,
      "grad_norm": 0.5888768252736022,
      "learning_rate": 9.492696232520294e-06,
      "loss": 0.5293,
      "step": 2321
    },
    {
      "epoch": 0.17036418830305308,
      "grad_norm": 0.5941064589842092,
      "learning_rate": 9.492174614145452e-06,
      "loss": 0.5438,
      "step": 2322
    },
    {
      "epoch": 0.17043755789319223,
      "grad_norm": 1.7638158657399483,
      "learning_rate": 9.491652742086386e-06,
      "loss": 0.8574,
      "step": 2323
    },
    {
      "epoch": 0.17051092748333135,
      "grad_norm": 1.9996308725135687,
      "learning_rate": 9.491130616372569e-06,
      "loss": 0.9002,
      "step": 2324
    },
    {
      "epoch": 0.17058429707347048,
      "grad_norm": 1.7622281911973776,
      "learning_rate": 9.49060823703348e-06,
      "loss": 0.8766,
      "step": 2325
    },
    {
      "epoch": 0.1706576666636096,
      "grad_norm": 1.3999676267459251,
      "learning_rate": 9.490085604098624e-06,
      "loss": 0.9562,
      "step": 2326
    },
    {
      "epoch": 0.17073103625374872,
      "grad_norm": 1.6611276909563837,
      "learning_rate": 9.489562717597516e-06,
      "loss": 0.8422,
      "step": 2327
    },
    {
      "epoch": 0.17080440584388784,
      "grad_norm": 0.633188120647983,
      "learning_rate": 9.489039577559683e-06,
      "loss": 0.5464,
      "step": 2328
    },
    {
      "epoch": 0.17080440584388784,
      "lm_loss": 1.5171,
      "step": 2328,
      "vm_loss": 0.1313
    },
    {
      "epoch": 0.17080440584388784,
      "lm_loss": 0.6347,
      "step": 2328,
      "vm_loss": 0.0501
    },
    {
      "epoch": 0.17080440584388784,
      "lm_loss": 1.1922,
      "step": 2328,
      "vm_loss": 0.1893
    },
    {
      "epoch": 0.17080440584388784,
      "lm_loss": 0.1286,
      "step": 2328,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.17080440584388784,
      "lm_loss": 0.0302,
      "step": 2328,
      "vm_loss": 0.1229
    },
    {
      "epoch": 0.17080440584388784,
      "lm_loss": 0.5492,
      "step": 2328,
      "vm_loss": 0.0636
    },
    {
      "epoch": 0.17080440584388784,
      "lm_loss": 1.4936,
      "step": 2328,
      "vm_loss": 0.0852
    },
    {
      "epoch": 0.17080440584388784,
      "lm_loss": 0.3356,
      "step": 2328,
      "vm_loss": 0.076
    },
    {
      "epoch": 0.170877775434027,
      "grad_norm": 1.785495801294828,
      "learning_rate": 9.488516184014667e-06,
      "loss": 0.8383,
      "step": 2329
    },
    {
      "epoch": 0.1709511450241661,
      "grad_norm": 1.8485340063267002,
      "learning_rate": 9.487992536992027e-06,
      "loss": 0.8763,
      "step": 2330
    },
    {
      "epoch": 0.17102451461430523,
      "grad_norm": 1.5912727106366304,
      "learning_rate": 9.487468636521334e-06,
      "loss": 0.8563,
      "step": 2331
    },
    {
      "epoch": 0.17109788420444436,
      "grad_norm": 2.0031622733963865,
      "learning_rate": 9.486944482632175e-06,
      "loss": 0.8703,
      "step": 2332
    },
    {
      "epoch": 0.17117125379458348,
      "grad_norm": 1.4983185330690543,
      "learning_rate": 9.486420075354147e-06,
      "loss": 0.831,
      "step": 2333
    },
    {
      "epoch": 0.17124462338472263,
      "grad_norm": 0.7196485033898196,
      "learning_rate": 9.485895414716868e-06,
      "loss": 0.5416,
      "step": 2334
    },
    {
      "epoch": 0.17131799297486175,
      "grad_norm": 1.7708181812497958,
      "learning_rate": 9.485370500749966e-06,
      "loss": 0.8643,
      "step": 2335
    },
    {
      "epoch": 0.17139136256500087,
      "grad_norm": 1.7450802275185897,
      "learning_rate": 9.484845333483084e-06,
      "loss": 0.8408,
      "step": 2336
    },
    {
      "epoch": 0.17139136256500087,
      "lm_loss": 1.8636,
      "step": 2336,
      "vm_loss": 0.0747
    },
    {
      "epoch": 0.17139136256500087,
      "lm_loss": 1.7501,
      "step": 2336,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.17139136256500087,
      "lm_loss": 0.8976,
      "step": 2336,
      "vm_loss": 0.1742
    },
    {
      "epoch": 0.17139136256500087,
      "lm_loss": 1.5249,
      "step": 2336,
      "vm_loss": 0.1489
    },
    {
      "epoch": 0.17139136256500087,
      "lm_loss": 0.185,
      "step": 2336,
      "vm_loss": 0.0895
    },
    {
      "epoch": 0.17139136256500087,
      "lm_loss": 0.4537,
      "step": 2336,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.17139136256500087,
      "lm_loss": 1.1784,
      "step": 2336,
      "vm_loss": 0.1968
    },
    {
      "epoch": 0.17139136256500087,
      "lm_loss": 0.7828,
      "step": 2336,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.17146473215514,
      "grad_norm": 1.9517119361969335,
      "learning_rate": 9.48431991294588e-06,
      "loss": 0.9792,
      "step": 2337
    },
    {
      "epoch": 0.17153810174527911,
      "grad_norm": 1.5622113880029025,
      "learning_rate": 9.483794239168026e-06,
      "loss": 0.8772,
      "step": 2338
    },
    {
      "epoch": 0.17161147133541826,
      "grad_norm": 0.6339407841752912,
      "learning_rate": 9.483268312179206e-06,
      "loss": 0.5358,
      "step": 2339
    },
    {
      "epoch": 0.17168484092555739,
      "grad_norm": 0.6200644175509866,
      "learning_rate": 9.482742132009122e-06,
      "loss": 0.5493,
      "step": 2340
    },
    {
      "epoch": 0.1717582105156965,
      "grad_norm": 0.5911515901856688,
      "learning_rate": 9.482215698687486e-06,
      "loss": 0.5637,
      "step": 2341
    },
    {
      "epoch": 0.17183158010583563,
      "grad_norm": 1.9019521012020721,
      "learning_rate": 9.481689012244033e-06,
      "loss": 0.933,
      "step": 2342
    },
    {
      "epoch": 0.17190494969597475,
      "grad_norm": 1.6949363292892514,
      "learning_rate": 9.481162072708503e-06,
      "loss": 0.8874,
      "step": 2343
    },
    {
      "epoch": 0.1719783192861139,
      "grad_norm": 1.4544729247811108,
      "learning_rate": 9.480634880110651e-06,
      "loss": 0.8312,
      "step": 2344
    },
    {
      "epoch": 0.1719783192861139,
      "lm_loss": 0.1099,
      "step": 2344,
      "vm_loss": 0.0526
    },
    {
      "epoch": 0.1719783192861139,
      "lm_loss": 1.3562,
      "step": 2344,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.1719783192861139,
      "lm_loss": 0.5716,
      "step": 2344,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.1719783192861139,
      "lm_loss": 0.9029,
      "step": 2344,
      "vm_loss": 0.0688
    },
    {
      "epoch": 0.1719783192861139,
      "lm_loss": 0.3132,
      "step": 2344,
      "vm_loss": 0.1015
    },
    {
      "epoch": 0.1719783192861139,
      "lm_loss": 0.9109,
      "step": 2344,
      "vm_loss": 0.1231
    },
    {
      "epoch": 0.1719783192861139,
      "lm_loss": 0.4176,
      "step": 2344,
      "vm_loss": 0.1691
    },
    {
      "epoch": 0.1719783192861139,
      "lm_loss": 0.9621,
      "step": 2344,
      "vm_loss": 0.1317
    },
    {
      "epoch": 0.17205168887625302,
      "grad_norm": 1.4338244290562268,
      "learning_rate": 9.480107434480253e-06,
      "loss": 0.8457,
      "step": 2345
    },
    {
      "epoch": 0.17212505846639214,
      "grad_norm": 1.8102378675757997,
      "learning_rate": 9.479579735847094e-06,
      "loss": 0.9016,
      "step": 2346
    },
    {
      "epoch": 0.17219842805653127,
      "grad_norm": 0.7319498785041589,
      "learning_rate": 9.479051784240972e-06,
      "loss": 0.5242,
      "step": 2347
    },
    {
      "epoch": 0.1722717976466704,
      "grad_norm": 0.715058165072618,
      "learning_rate": 9.478523579691702e-06,
      "loss": 0.5484,
      "step": 2348
    },
    {
      "epoch": 0.1723451672368095,
      "grad_norm": 2.242256143034402,
      "learning_rate": 9.477995122229117e-06,
      "loss": 0.8951,
      "step": 2349
    },
    {
      "epoch": 0.17241853682694866,
      "grad_norm": 1.729734781108882,
      "learning_rate": 9.477466411883058e-06,
      "loss": 0.89,
      "step": 2350
    },
    {
      "epoch": 0.17249190641708778,
      "grad_norm": 1.4755920142096983,
      "learning_rate": 9.476937448683381e-06,
      "loss": 0.7703,
      "step": 2351
    },
    {
      "epoch": 0.1725652760072269,
      "grad_norm": 1.6765391107099938,
      "learning_rate": 9.476408232659962e-06,
      "loss": 0.8614,
      "step": 2352
    },
    {
      "epoch": 0.1725652760072269,
      "lm_loss": 1.5527,
      "step": 2352,
      "vm_loss": 0.1296
    },
    {
      "epoch": 0.1725652760072269,
      "lm_loss": 1.0424,
      "step": 2352,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.1725652760072269,
      "lm_loss": 0.14,
      "step": 2352,
      "vm_loss": 0.1971
    },
    {
      "epoch": 0.1725652760072269,
      "lm_loss": 1.0171,
      "step": 2352,
      "vm_loss": 0.1706
    },
    {
      "epoch": 0.1725652760072269,
      "lm_loss": 0.4638,
      "step": 2352,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.1725652760072269,
      "lm_loss": 0.3003,
      "step": 2352,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.1725652760072269,
      "lm_loss": 0.77,
      "step": 2352,
      "vm_loss": 0.1311
    },
    {
      "epoch": 0.1725652760072269,
      "lm_loss": 0.7996,
      "step": 2352,
      "vm_loss": 0.1256
    },
    {
      "epoch": 0.17263864559736602,
      "grad_norm": 1.7474478968635798,
      "learning_rate": 9.475878763842684e-06,
      "loss": 0.8744,
      "step": 2353
    },
    {
      "epoch": 0.17271201518750515,
      "grad_norm": 1.5926669360715728,
      "learning_rate": 9.475349042261445e-06,
      "loss": 0.8386,
      "step": 2354
    },
    {
      "epoch": 0.1727853847776443,
      "grad_norm": 0.9046613790975603,
      "learning_rate": 9.474819067946164e-06,
      "loss": 0.5518,
      "step": 2355
    },
    {
      "epoch": 0.17285875436778342,
      "grad_norm": 1.575332715037008,
      "learning_rate": 9.474288840926769e-06,
      "loss": 0.8096,
      "step": 2356
    },
    {
      "epoch": 0.17293212395792254,
      "grad_norm": 1.6666004327973984,
      "learning_rate": 9.473758361233203e-06,
      "loss": 0.7719,
      "step": 2357
    },
    {
      "epoch": 0.17300549354806166,
      "grad_norm": 1.680816535215726,
      "learning_rate": 9.473227628895422e-06,
      "loss": 0.8777,
      "step": 2358
    },
    {
      "epoch": 0.17307886313820078,
      "grad_norm": 1.6446054499683158,
      "learning_rate": 9.4726966439434e-06,
      "loss": 0.8625,
      "step": 2359
    },
    {
      "epoch": 0.17315223272833993,
      "grad_norm": 2.421217885850566,
      "learning_rate": 9.472165406407121e-06,
      "loss": 0.8984,
      "step": 2360
    },
    {
      "epoch": 0.17315223272833993,
      "lm_loss": 0.7201,
      "step": 2360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17315223272833993,
      "lm_loss": 0.3595,
      "step": 2360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17315223272833993,
      "lm_loss": 0.5582,
      "step": 2360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17315223272833993,
      "lm_loss": 0.7517,
      "step": 2360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17315223272833993,
      "lm_loss": 0.6365,
      "step": 2360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17315223272833993,
      "lm_loss": 0.5543,
      "step": 2360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17315223272833993,
      "lm_loss": 0.5,
      "step": 2360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17315223272833993,
      "lm_loss": 0.5607,
      "step": 2360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17322560231847905,
      "grad_norm": 0.6192122768799,
      "learning_rate": 9.471633916316588e-06,
      "loss": 0.571,
      "step": 2361
    },
    {
      "epoch": 0.17329897190861818,
      "grad_norm": 1.7055374250189896,
      "learning_rate": 9.471102173701814e-06,
      "loss": 0.8986,
      "step": 2362
    },
    {
      "epoch": 0.1733723414987573,
      "grad_norm": 1.7434344398847434,
      "learning_rate": 9.470570178592826e-06,
      "loss": 0.8682,
      "step": 2363
    },
    {
      "epoch": 0.17344571108889642,
      "grad_norm": 0.6225359114166765,
      "learning_rate": 9.47003793101967e-06,
      "loss": 0.5608,
      "step": 2364
    },
    {
      "epoch": 0.17351908067903557,
      "grad_norm": 1.5645227109871092,
      "learning_rate": 9.469505431012402e-06,
      "loss": 0.9417,
      "step": 2365
    },
    {
      "epoch": 0.1735924502691747,
      "grad_norm": 1.609472423212149,
      "learning_rate": 9.468972678601093e-06,
      "loss": 0.8363,
      "step": 2366
    },
    {
      "epoch": 0.1736658198593138,
      "grad_norm": 1.535484076378304,
      "learning_rate": 9.46843967381583e-06,
      "loss": 0.7726,
      "step": 2367
    },
    {
      "epoch": 0.17373918944945294,
      "grad_norm": 1.6955294601470512,
      "learning_rate": 9.467906416686714e-06,
      "loss": 1.0234,
      "step": 2368
    },
    {
      "epoch": 0.17373918944945294,
      "lm_loss": 0.5701,
      "step": 2368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17373918944945294,
      "lm_loss": 0.4691,
      "step": 2368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17373918944945294,
      "lm_loss": 0.5374,
      "step": 2368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17373918944945294,
      "lm_loss": 0.6687,
      "step": 2368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17373918944945294,
      "lm_loss": 0.4318,
      "step": 2368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17373918944945294,
      "lm_loss": 0.5839,
      "step": 2368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17373918944945294,
      "lm_loss": 0.5976,
      "step": 2368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17373918944945294,
      "lm_loss": 0.5741,
      "step": 2368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17381255903959206,
      "grad_norm": 0.5912680744788774,
      "learning_rate": 9.467372907243858e-06,
      "loss": 0.5608,
      "step": 2369
    },
    {
      "epoch": 0.17388592862973118,
      "grad_norm": 1.8690760184794684,
      "learning_rate": 9.466839145517388e-06,
      "loss": 0.8823,
      "step": 2370
    },
    {
      "epoch": 0.17395929821987033,
      "grad_norm": 1.5955700241955242,
      "learning_rate": 9.466305131537452e-06,
      "loss": 0.8456,
      "step": 2371
    },
    {
      "epoch": 0.17403266781000945,
      "grad_norm": 0.5985256221753034,
      "learning_rate": 9.465770865334203e-06,
      "loss": 0.5574,
      "step": 2372
    },
    {
      "epoch": 0.17410603740014857,
      "grad_norm": 1.7190017242305409,
      "learning_rate": 9.465236346937816e-06,
      "loss": 0.8463,
      "step": 2373
    },
    {
      "epoch": 0.1741794069902877,
      "grad_norm": 1.6921210166662464,
      "learning_rate": 9.464701576378472e-06,
      "loss": 0.8311,
      "step": 2374
    },
    {
      "epoch": 0.17425277658042682,
      "grad_norm": 1.759648584866543,
      "learning_rate": 9.464166553686376e-06,
      "loss": 0.918,
      "step": 2375
    },
    {
      "epoch": 0.17432614617056597,
      "grad_norm": 0.6125491141944223,
      "learning_rate": 9.463631278891737e-06,
      "loss": 0.5612,
      "step": 2376
    },
    {
      "epoch": 0.17432614617056597,
      "lm_loss": 1.0371,
      "step": 2376,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.17432614617056597,
      "lm_loss": 1.4349,
      "step": 2376,
      "vm_loss": 0.1259
    },
    {
      "epoch": 0.17432614617056597,
      "lm_loss": 1.0368,
      "step": 2376,
      "vm_loss": 0.0764
    },
    {
      "epoch": 0.17432614617056597,
      "lm_loss": 1.0951,
      "step": 2376,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.17432614617056597,
      "lm_loss": 0.7107,
      "step": 2376,
      "vm_loss": 0.0613
    },
    {
      "epoch": 0.17432614617056597,
      "lm_loss": 1.3364,
      "step": 2376,
      "vm_loss": 0.2297
    },
    {
      "epoch": 0.17432614617056597,
      "lm_loss": 0.9983,
      "step": 2376,
      "vm_loss": 0.193
    },
    {
      "epoch": 0.17432614617056597,
      "lm_loss": 0.144,
      "step": 2376,
      "vm_loss": 0.0897
    },
    {
      "epoch": 0.1743995157607051,
      "grad_norm": 1.6474825934883304,
      "learning_rate": 9.463095752024787e-06,
      "loss": 0.9005,
      "step": 2377
    },
    {
      "epoch": 0.1744728853508442,
      "grad_norm": 1.6109746654394248,
      "learning_rate": 9.462559973115768e-06,
      "loss": 0.794,
      "step": 2378
    },
    {
      "epoch": 0.17454625494098333,
      "grad_norm": 1.4032198350829586,
      "learning_rate": 9.462023942194934e-06,
      "loss": 0.7095,
      "step": 2379
    },
    {
      "epoch": 0.17461962453112245,
      "grad_norm": 1.472257415374824,
      "learning_rate": 9.461487659292559e-06,
      "loss": 0.7685,
      "step": 2380
    },
    {
      "epoch": 0.1746929941212616,
      "grad_norm": 0.6226838487817655,
      "learning_rate": 9.460951124438927e-06,
      "loss": 0.5404,
      "step": 2381
    },
    {
      "epoch": 0.17476636371140072,
      "grad_norm": 1.5694016684813097,
      "learning_rate": 9.460414337664338e-06,
      "loss": 0.8286,
      "step": 2382
    },
    {
      "epoch": 0.17483973330153985,
      "grad_norm": 1.4677096045187146,
      "learning_rate": 9.459877298999104e-06,
      "loss": 0.7723,
      "step": 2383
    },
    {
      "epoch": 0.17491310289167897,
      "grad_norm": 1.7709394206452655,
      "learning_rate": 9.459340008473557e-06,
      "loss": 0.751,
      "step": 2384
    },
    {
      "epoch": 0.17491310289167897,
      "lm_loss": 0.6161,
      "step": 2384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17491310289167897,
      "lm_loss": 0.6766,
      "step": 2384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17491310289167897,
      "lm_loss": 0.5884,
      "step": 2384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17491310289167897,
      "lm_loss": 0.3378,
      "step": 2384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17491310289167897,
      "lm_loss": 0.5444,
      "step": 2384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17491310289167897,
      "lm_loss": 0.4926,
      "step": 2384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17491310289167897,
      "lm_loss": 0.509,
      "step": 2384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17491310289167897,
      "lm_loss": 0.5786,
      "step": 2384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1749864724818181,
      "grad_norm": 0.6105797990345843,
      "learning_rate": 9.458802466118035e-06,
      "loss": 0.5492,
      "step": 2385
    },
    {
      "epoch": 0.17505984207195724,
      "grad_norm": 1.5629912883026023,
      "learning_rate": 9.458264671962894e-06,
      "loss": 0.9334,
      "step": 2386
    },
    {
      "epoch": 0.17513321166209636,
      "grad_norm": 1.389119840881358,
      "learning_rate": 9.457726626038508e-06,
      "loss": 0.8429,
      "step": 2387
    },
    {
      "epoch": 0.17520658125223548,
      "grad_norm": 1.4888529272357434,
      "learning_rate": 9.45718832837526e-06,
      "loss": 0.8615,
      "step": 2388
    },
    {
      "epoch": 0.1752799508423746,
      "grad_norm": 0.6178221654257212,
      "learning_rate": 9.456649779003548e-06,
      "loss": 0.5214,
      "step": 2389
    },
    {
      "epoch": 0.17535332043251373,
      "grad_norm": 1.536754523545562,
      "learning_rate": 9.456110977953788e-06,
      "loss": 0.8551,
      "step": 2390
    },
    {
      "epoch": 0.17542669002265285,
      "grad_norm": 1.7192215319170936,
      "learning_rate": 9.455571925256403e-06,
      "loss": 0.8164,
      "step": 2391
    },
    {
      "epoch": 0.175500059612792,
      "grad_norm": 1.7880404408291855,
      "learning_rate": 9.45503262094184e-06,
      "loss": 0.8694,
      "step": 2392
    },
    {
      "epoch": 0.175500059612792,
      "lm_loss": 1.6199,
      "step": 2392,
      "vm_loss": 0.0589
    },
    {
      "epoch": 0.175500059612792,
      "lm_loss": 0.6379,
      "step": 2392,
      "vm_loss": 0.155
    },
    {
      "epoch": 0.175500059612792,
      "lm_loss": 0.1143,
      "step": 2392,
      "vm_loss": 0.1409
    },
    {
      "epoch": 0.175500059612792,
      "lm_loss": 0.3737,
      "step": 2392,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.175500059612792,
      "lm_loss": 0.1642,
      "step": 2392,
      "vm_loss": 0.1094
    },
    {
      "epoch": 0.175500059612792,
      "lm_loss": 0.6193,
      "step": 2392,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.175500059612792,
      "lm_loss": 0.8457,
      "step": 2392,
      "vm_loss": 0.2046
    },
    {
      "epoch": 0.175500059612792,
      "lm_loss": 0.3977,
      "step": 2392,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.17557342920293112,
      "grad_norm": 1.6134811740812889,
      "learning_rate": 9.454493065040551e-06,
      "loss": 0.8832,
      "step": 2393
    },
    {
      "epoch": 0.17564679879307024,
      "grad_norm": 1.5380875516860482,
      "learning_rate": 9.453953257583008e-06,
      "loss": 0.8891,
      "step": 2394
    },
    {
      "epoch": 0.17572016838320936,
      "grad_norm": 1.6953696511000413,
      "learning_rate": 9.453413198599694e-06,
      "loss": 0.9295,
      "step": 2395
    },
    {
      "epoch": 0.17579353797334848,
      "grad_norm": 0.6387580692163377,
      "learning_rate": 9.452872888121107e-06,
      "loss": 0.5421,
      "step": 2396
    },
    {
      "epoch": 0.17586690756348763,
      "grad_norm": 1.4710715717966285,
      "learning_rate": 9.452332326177762e-06,
      "loss": 0.8654,
      "step": 2397
    },
    {
      "epoch": 0.17594027715362676,
      "grad_norm": 1.4865614629685318,
      "learning_rate": 9.451791512800184e-06,
      "loss": 0.8898,
      "step": 2398
    },
    {
      "epoch": 0.17601364674376588,
      "grad_norm": 1.6587252509412702,
      "learning_rate": 9.451250448018916e-06,
      "loss": 0.7981,
      "step": 2399
    },
    {
      "epoch": 0.176087016333905,
      "grad_norm": 1.7003850719412612,
      "learning_rate": 9.45070913186451e-06,
      "loss": 0.9426,
      "step": 2400
    },
    {
      "epoch": 0.176087016333905,
      "lm_loss": 1.69,
      "step": 2400,
      "vm_loss": 0.0776
    },
    {
      "epoch": 0.176087016333905,
      "lm_loss": 1.4038,
      "step": 2400,
      "vm_loss": 0.2393
    },
    {
      "epoch": 0.176087016333905,
      "lm_loss": 0.9582,
      "step": 2400,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.176087016333905,
      "lm_loss": 1.4775,
      "step": 2400,
      "vm_loss": 0.1717
    },
    {
      "epoch": 0.176087016333905,
      "lm_loss": 0.7426,
      "step": 2400,
      "vm_loss": 0.1693
    },
    {
      "epoch": 0.176087016333905,
      "lm_loss": 0.833,
      "step": 2400,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.176087016333905,
      "lm_loss": 1.3269,
      "step": 2400,
      "vm_loss": 0.2051
    },
    {
      "epoch": 0.176087016333905,
      "lm_loss": 0.4112,
      "step": 2400,
      "vm_loss": 0.1862
    },
    {
      "epoch": 0.17616038592404412,
      "grad_norm": 1.7127510924680351,
      "learning_rate": 9.450167564367539e-06,
      "loss": 0.9765,
      "step": 2401
    },
    {
      "epoch": 0.17623375551418327,
      "grad_norm": 1.8024548448344722,
      "learning_rate": 9.449625745558583e-06,
      "loss": 0.8613,
      "step": 2402
    },
    {
      "epoch": 0.1763071251043224,
      "grad_norm": 1.4795586750052812,
      "learning_rate": 9.449083675468242e-06,
      "loss": 0.852,
      "step": 2403
    },
    {
      "epoch": 0.17638049469446151,
      "grad_norm": 1.4623222940072242,
      "learning_rate": 9.44854135412713e-06,
      "loss": 0.772,
      "step": 2404
    },
    {
      "epoch": 0.17645386428460064,
      "grad_norm": 1.4698711091810448,
      "learning_rate": 9.447998781565869e-06,
      "loss": 0.9227,
      "step": 2405
    },
    {
      "epoch": 0.17652723387473976,
      "grad_norm": 1.5763401596027917,
      "learning_rate": 9.447455957815103e-06,
      "loss": 0.9633,
      "step": 2406
    },
    {
      "epoch": 0.1766006034648789,
      "grad_norm": 1.5866909297581182,
      "learning_rate": 9.446912882905484e-06,
      "loss": 0.8621,
      "step": 2407
    },
    {
      "epoch": 0.17667397305501803,
      "grad_norm": 0.6910645108961839,
      "learning_rate": 9.446369556867683e-06,
      "loss": 0.5772,
      "step": 2408
    },
    {
      "epoch": 0.17667397305501803,
      "lm_loss": 1.02,
      "step": 2408,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.17667397305501803,
      "lm_loss": 1.6928,
      "step": 2408,
      "vm_loss": 0.0913
    },
    {
      "epoch": 0.17667397305501803,
      "lm_loss": 1.3324,
      "step": 2408,
      "vm_loss": 0.195
    },
    {
      "epoch": 0.17667397305501803,
      "lm_loss": 0.8863,
      "step": 2408,
      "vm_loss": 0.1412
    },
    {
      "epoch": 0.17667397305501803,
      "lm_loss": 0.5981,
      "step": 2408,
      "vm_loss": 0.1518
    },
    {
      "epoch": 0.17667397305501803,
      "lm_loss": 0.3688,
      "step": 2408,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.17667397305501803,
      "lm_loss": 0.2996,
      "step": 2408,
      "vm_loss": 0.1775
    },
    {
      "epoch": 0.17667397305501803,
      "lm_loss": 0.806,
      "step": 2408,
      "vm_loss": 0.1695
    },
    {
      "epoch": 0.17674734264515715,
      "grad_norm": 1.9059884658700774,
      "learning_rate": 9.44582597973238e-06,
      "loss": 0.8899,
      "step": 2409
    },
    {
      "epoch": 0.17682071223529627,
      "grad_norm": 0.600579759165999,
      "learning_rate": 9.445282151530276e-06,
      "loss": 0.5591,
      "step": 2410
    },
    {
      "epoch": 0.1768940818254354,
      "grad_norm": 1.5685080997466943,
      "learning_rate": 9.444738072292079e-06,
      "loss": 0.8352,
      "step": 2411
    },
    {
      "epoch": 0.17696745141557452,
      "grad_norm": 1.6067889592561837,
      "learning_rate": 9.444193742048517e-06,
      "loss": 0.8778,
      "step": 2412
    },
    {
      "epoch": 0.17704082100571367,
      "grad_norm": 1.52226746407857,
      "learning_rate": 9.443649160830328e-06,
      "loss": 0.8722,
      "step": 2413
    },
    {
      "epoch": 0.1771141905958528,
      "grad_norm": 0.6679649877833564,
      "learning_rate": 9.443104328668266e-06,
      "loss": 0.566,
      "step": 2414
    },
    {
      "epoch": 0.1771875601859919,
      "grad_norm": 1.4616162278909888,
      "learning_rate": 9.4425592455931e-06,
      "loss": 0.8177,
      "step": 2415
    },
    {
      "epoch": 0.17726092977613103,
      "grad_norm": 1.5326060094875527,
      "learning_rate": 9.44201391163561e-06,
      "loss": 0.8782,
      "step": 2416
    },
    {
      "epoch": 0.17726092977613103,
      "lm_loss": 0.4926,
      "step": 2416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17726092977613103,
      "lm_loss": 0.3332,
      "step": 2416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17726092977613103,
      "lm_loss": 0.7382,
      "step": 2416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17726092977613103,
      "lm_loss": 0.2235,
      "step": 2416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17726092977613103,
      "lm_loss": 0.6468,
      "step": 2416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17726092977613103,
      "lm_loss": 0.582,
      "step": 2416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17726092977613103,
      "lm_loss": 0.5143,
      "step": 2416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17726092977613103,
      "lm_loss": 0.4125,
      "step": 2416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17733429936627015,
      "grad_norm": 0.6059995631286575,
      "learning_rate": 9.441468326826596e-06,
      "loss": 0.5398,
      "step": 2417
    },
    {
      "epoch": 0.1774076689564093,
      "grad_norm": 1.561386941130974,
      "learning_rate": 9.440922491196866e-06,
      "loss": 1.0448,
      "step": 2418
    },
    {
      "epoch": 0.17748103854654843,
      "grad_norm": 1.4883661391956935,
      "learning_rate": 9.440376404777245e-06,
      "loss": 0.9178,
      "step": 2419
    },
    {
      "epoch": 0.17755440813668755,
      "grad_norm": 0.5828737074774856,
      "learning_rate": 9.439830067598572e-06,
      "loss": 0.5449,
      "step": 2420
    },
    {
      "epoch": 0.17762777772682667,
      "grad_norm": 1.8630959325374825,
      "learning_rate": 9.439283479691698e-06,
      "loss": 0.8276,
      "step": 2421
    },
    {
      "epoch": 0.1777011473169658,
      "grad_norm": 1.5431989889920046,
      "learning_rate": 9.438736641087496e-06,
      "loss": 0.8881,
      "step": 2422
    },
    {
      "epoch": 0.17777451690710494,
      "grad_norm": 0.5946269055619037,
      "learning_rate": 9.438189551816841e-06,
      "loss": 0.5478,
      "step": 2423
    },
    {
      "epoch": 0.17784788649724406,
      "grad_norm": 1.756419856264424,
      "learning_rate": 9.43764221191063e-06,
      "loss": 0.912,
      "step": 2424
    },
    {
      "epoch": 0.17784788649724406,
      "lm_loss": 0.539,
      "step": 2424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17784788649724406,
      "lm_loss": 0.4819,
      "step": 2424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17784788649724406,
      "lm_loss": 0.6503,
      "step": 2424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17784788649724406,
      "lm_loss": 0.4858,
      "step": 2424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17784788649724406,
      "lm_loss": 0.5501,
      "step": 2424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17784788649724406,
      "lm_loss": 0.512,
      "step": 2424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17784788649724406,
      "lm_loss": 0.702,
      "step": 2424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17784788649724406,
      "lm_loss": 0.713,
      "step": 2424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17792125608738318,
      "grad_norm": 0.602776952011787,
      "learning_rate": 9.437094621399777e-06,
      "loss": 0.5524,
      "step": 2425
    },
    {
      "epoch": 0.1779946256775223,
      "grad_norm": 1.4938749193370138,
      "learning_rate": 9.4365467803152e-06,
      "loss": 0.8005,
      "step": 2426
    },
    {
      "epoch": 0.17806799526766143,
      "grad_norm": 1.5633622819495063,
      "learning_rate": 9.435998688687841e-06,
      "loss": 0.8274,
      "step": 2427
    },
    {
      "epoch": 0.17814136485780055,
      "grad_norm": 1.7083023137527549,
      "learning_rate": 9.43545034654865e-06,
      "loss": 0.8803,
      "step": 2428
    },
    {
      "epoch": 0.1782147344479397,
      "grad_norm": 0.6329851959710752,
      "learning_rate": 9.434901753928593e-06,
      "loss": 0.5612,
      "step": 2429
    },
    {
      "epoch": 0.17828810403807882,
      "grad_norm": 1.612429566754053,
      "learning_rate": 9.434352910858651e-06,
      "loss": 0.868,
      "step": 2430
    },
    {
      "epoch": 0.17836147362821794,
      "grad_norm": 1.5742327524702842,
      "learning_rate": 9.43380381736982e-06,
      "loss": 0.8491,
      "step": 2431
    },
    {
      "epoch": 0.17843484321835706,
      "grad_norm": 0.6074270953728987,
      "learning_rate": 9.433254473493106e-06,
      "loss": 0.5416,
      "step": 2432
    },
    {
      "epoch": 0.17843484321835706,
      "lm_loss": 1.7667,
      "step": 2432,
      "vm_loss": 0.0438
    },
    {
      "epoch": 0.17843484321835706,
      "lm_loss": 1.0364,
      "step": 2432,
      "vm_loss": 0.0673
    },
    {
      "epoch": 0.17843484321835706,
      "lm_loss": 0.0416,
      "step": 2432,
      "vm_loss": 0.0413
    },
    {
      "epoch": 0.17843484321835706,
      "lm_loss": 1.6761,
      "step": 2432,
      "vm_loss": 0.0938
    },
    {
      "epoch": 0.17843484321835706,
      "lm_loss": 0.4252,
      "step": 2432,
      "vm_loss": 0.2059
    },
    {
      "epoch": 0.17843484321835706,
      "lm_loss": 0.3186,
      "step": 2432,
      "vm_loss": 0.082
    },
    {
      "epoch": 0.17843484321835706,
      "lm_loss": 0.7139,
      "step": 2432,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.17843484321835706,
      "lm_loss": 2.2603,
      "step": 2432,
      "vm_loss": 0.1319
    },
    {
      "epoch": 0.17850821280849619,
      "grad_norm": 1.512423903675026,
      "learning_rate": 9.432704879259533e-06,
      "loss": 0.7787,
      "step": 2433
    },
    {
      "epoch": 0.17858158239863534,
      "grad_norm": 1.639071861623884,
      "learning_rate": 9.432155034700139e-06,
      "loss": 0.7862,
      "step": 2434
    },
    {
      "epoch": 0.17865495198877446,
      "grad_norm": 1.5297074864108327,
      "learning_rate": 9.431604939845975e-06,
      "loss": 0.8848,
      "step": 2435
    },
    {
      "epoch": 0.17872832157891358,
      "grad_norm": 1.5983385551884972,
      "learning_rate": 9.431054594728103e-06,
      "loss": 0.8011,
      "step": 2436
    },
    {
      "epoch": 0.1788016911690527,
      "grad_norm": 0.6345339798351245,
      "learning_rate": 9.430503999377605e-06,
      "loss": 0.5457,
      "step": 2437
    },
    {
      "epoch": 0.17887506075919182,
      "grad_norm": 0.6025768889494261,
      "learning_rate": 9.429953153825575e-06,
      "loss": 0.5455,
      "step": 2438
    },
    {
      "epoch": 0.17894843034933097,
      "grad_norm": 0.5888745030573316,
      "learning_rate": 9.429402058103122e-06,
      "loss": 0.5391,
      "step": 2439
    },
    {
      "epoch": 0.1790217999394701,
      "grad_norm": 1.8257007429128687,
      "learning_rate": 9.428850712241362e-06,
      "loss": 0.9549,
      "step": 2440
    },
    {
      "epoch": 0.1790217999394701,
      "lm_loss": 2.0501,
      "step": 2440,
      "vm_loss": 0.1502
    },
    {
      "epoch": 0.1790217999394701,
      "lm_loss": 0.8763,
      "step": 2440,
      "vm_loss": 0.1539
    },
    {
      "epoch": 0.1790217999394701,
      "lm_loss": 1.2299,
      "step": 2440,
      "vm_loss": 0.0846
    },
    {
      "epoch": 0.1790217999394701,
      "lm_loss": 0.3464,
      "step": 2440,
      "vm_loss": 0.1198
    },
    {
      "epoch": 0.1790217999394701,
      "lm_loss": 0.3759,
      "step": 2440,
      "vm_loss": 0.2699
    },
    {
      "epoch": 0.1790217999394701,
      "lm_loss": 0.4383,
      "step": 2440,
      "vm_loss": 0.1294
    },
    {
      "epoch": 0.1790217999394701,
      "lm_loss": 0.5188,
      "step": 2440,
      "vm_loss": 0.2046
    },
    {
      "epoch": 0.1790217999394701,
      "lm_loss": 0.4383,
      "step": 2440,
      "vm_loss": 0.1032
    },
    {
      "epoch": 0.17909516952960922,
      "grad_norm": 1.6132626102529541,
      "learning_rate": 9.428299116271435e-06,
      "loss": 0.883,
      "step": 2441
    },
    {
      "epoch": 0.17916853911974834,
      "grad_norm": 1.7807852252397656,
      "learning_rate": 9.427747270224493e-06,
      "loss": 0.8693,
      "step": 2442
    },
    {
      "epoch": 0.17924190870988746,
      "grad_norm": 0.7752395021272488,
      "learning_rate": 9.427195174131695e-06,
      "loss": 0.5523,
      "step": 2443
    },
    {
      "epoch": 0.1793152783000266,
      "grad_norm": 1.502703818438088,
      "learning_rate": 9.426642828024222e-06,
      "loss": 0.8628,
      "step": 2444
    },
    {
      "epoch": 0.17938864789016573,
      "grad_norm": 1.5871797744072702,
      "learning_rate": 9.426090231933268e-06,
      "loss": 0.9731,
      "step": 2445
    },
    {
      "epoch": 0.17946201748030485,
      "grad_norm": 0.6485951613974309,
      "learning_rate": 9.425537385890036e-06,
      "loss": 0.5407,
      "step": 2446
    },
    {
      "epoch": 0.17953538707044397,
      "grad_norm": 1.636881878015504,
      "learning_rate": 9.424984289925747e-06,
      "loss": 0.9415,
      "step": 2447
    },
    {
      "epoch": 0.1796087566605831,
      "grad_norm": 1.5427867724909454,
      "learning_rate": 9.424430944071638e-06,
      "loss": 0.8001,
      "step": 2448
    },
    {
      "epoch": 0.1796087566605831,
      "lm_loss": 0.2124,
      "step": 2448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1796087566605831,
      "lm_loss": 0.5453,
      "step": 2448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1796087566605831,
      "lm_loss": 0.5309,
      "step": 2448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1796087566605831,
      "lm_loss": 0.649,
      "step": 2448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1796087566605831,
      "lm_loss": 0.5788,
      "step": 2448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1796087566605831,
      "lm_loss": 0.6633,
      "step": 2448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1796087566605831,
      "lm_loss": 0.5966,
      "step": 2448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1796087566605831,
      "lm_loss": 0.5799,
      "step": 2448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.17968212625072222,
      "grad_norm": 0.5883097648219602,
      "learning_rate": 9.423877348358956e-06,
      "loss": 0.541,
      "step": 2449
    },
    {
      "epoch": 0.17975549584086137,
      "grad_norm": 0.6174772756705696,
      "learning_rate": 9.423323502818966e-06,
      "loss": 0.5557,
      "step": 2450
    },
    {
      "epoch": 0.1798288654310005,
      "grad_norm": 0.5962723621219043,
      "learning_rate": 9.422769407482944e-06,
      "loss": 0.5473,
      "step": 2451
    },
    {
      "epoch": 0.1799022350211396,
      "grad_norm": 0.5711055313693965,
      "learning_rate": 9.42221506238218e-06,
      "loss": 0.5306,
      "step": 2452
    },
    {
      "epoch": 0.17997560461127873,
      "grad_norm": 1.832076933902017,
      "learning_rate": 9.42166046754798e-06,
      "loss": 0.8425,
      "step": 2453
    },
    {
      "epoch": 0.18004897420141786,
      "grad_norm": 1.508278271328744,
      "learning_rate": 9.421105623011661e-06,
      "loss": 0.7739,
      "step": 2454
    },
    {
      "epoch": 0.180122343791557,
      "grad_norm": 1.5177503285819076,
      "learning_rate": 9.420550528804562e-06,
      "loss": 0.8792,
      "step": 2455
    },
    {
      "epoch": 0.18019571338169613,
      "grad_norm": 1.7267341076281992,
      "learning_rate": 9.419995184958026e-06,
      "loss": 0.8495,
      "step": 2456
    },
    {
      "epoch": 0.18019571338169613,
      "lm_loss": 1.3425,
      "step": 2456,
      "vm_loss": 0.1148
    },
    {
      "epoch": 0.18019571338169613,
      "lm_loss": 1.7716,
      "step": 2456,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.18019571338169613,
      "lm_loss": 0.5929,
      "step": 2456,
      "vm_loss": 0.0616
    },
    {
      "epoch": 0.18019571338169613,
      "lm_loss": 0.6198,
      "step": 2456,
      "vm_loss": 0.0854
    },
    {
      "epoch": 0.18019571338169613,
      "lm_loss": 1.1089,
      "step": 2456,
      "vm_loss": 0.1776
    },
    {
      "epoch": 0.18019571338169613,
      "lm_loss": 0.0428,
      "step": 2456,
      "vm_loss": 0.1131
    },
    {
      "epoch": 0.18019571338169613,
      "lm_loss": 0.2197,
      "step": 2456,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.18019571338169613,
      "lm_loss": 1.1665,
      "step": 2456,
      "vm_loss": 0.2621
    },
    {
      "epoch": 0.18026908297183525,
      "grad_norm": 1.6433782229926608,
      "learning_rate": 9.419439591503417e-06,
      "loss": 0.8772,
      "step": 2457
    },
    {
      "epoch": 0.18034245256197437,
      "grad_norm": 1.3783421230121489,
      "learning_rate": 9.41888374847211e-06,
      "loss": 0.8897,
      "step": 2458
    },
    {
      "epoch": 0.1804158221521135,
      "grad_norm": 1.657313606270818,
      "learning_rate": 9.418327655895492e-06,
      "loss": 0.8171,
      "step": 2459
    },
    {
      "epoch": 0.18048919174225264,
      "grad_norm": 1.6303318513366054,
      "learning_rate": 9.41777131380497e-06,
      "loss": 0.867,
      "step": 2460
    },
    {
      "epoch": 0.18056256133239176,
      "grad_norm": 1.6374920017152317,
      "learning_rate": 9.417214722231965e-06,
      "loss": 0.798,
      "step": 2461
    },
    {
      "epoch": 0.18063593092253089,
      "grad_norm": 1.5015681479809233,
      "learning_rate": 9.416657881207903e-06,
      "loss": 0.9764,
      "step": 2462
    },
    {
      "epoch": 0.18070930051267,
      "grad_norm": 1.6621552405746916,
      "learning_rate": 9.416100790764231e-06,
      "loss": 0.9268,
      "step": 2463
    },
    {
      "epoch": 0.18078267010280913,
      "grad_norm": 1.4936876337545346,
      "learning_rate": 9.415543450932414e-06,
      "loss": 0.925,
      "step": 2464
    },
    {
      "epoch": 0.18078267010280913,
      "lm_loss": 1.6633,
      "step": 2464,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.18078267010280913,
      "lm_loss": 3.1668,
      "step": 2464,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.18078267010280913,
      "lm_loss": 0.5698,
      "step": 2464,
      "vm_loss": 0.1038
    },
    {
      "epoch": 0.18078267010280913,
      "lm_loss": 1.0914,
      "step": 2464,
      "vm_loss": 0.2429
    },
    {
      "epoch": 0.18078267010280913,
      "lm_loss": 1.0427,
      "step": 2464,
      "vm_loss": 0.1442
    },
    {
      "epoch": 0.18078267010280913,
      "lm_loss": 0.8364,
      "step": 2464,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.18078267010280913,
      "lm_loss": 0.5707,
      "step": 2464,
      "vm_loss": 0.097
    },
    {
      "epoch": 0.18078267010280913,
      "lm_loss": 0.7475,
      "step": 2464,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.18085603969294828,
      "grad_norm": 1.6731158337178482,
      "learning_rate": 9.414985861743922e-06,
      "loss": 0.9264,
      "step": 2465
    },
    {
      "epoch": 0.1809294092830874,
      "grad_norm": 1.5971678757042163,
      "learning_rate": 9.414428023230244e-06,
      "loss": 0.8903,
      "step": 2466
    },
    {
      "epoch": 0.18100277887322652,
      "grad_norm": 1.759150988934529,
      "learning_rate": 9.413869935422885e-06,
      "loss": 0.8231,
      "step": 2467
    },
    {
      "epoch": 0.18107614846336564,
      "grad_norm": 0.9508243285031395,
      "learning_rate": 9.41331159835336e-06,
      "loss": 0.5431,
      "step": 2468
    },
    {
      "epoch": 0.18114951805350477,
      "grad_norm": 1.55912433991203,
      "learning_rate": 9.4127530120532e-06,
      "loss": 0.9177,
      "step": 2469
    },
    {
      "epoch": 0.1812228876436439,
      "grad_norm": 1.7327074469581352,
      "learning_rate": 9.412194176553948e-06,
      "loss": 0.9068,
      "step": 2470
    },
    {
      "epoch": 0.18129625723378304,
      "grad_norm": 1.6984137410407125,
      "learning_rate": 9.411635091887163e-06,
      "loss": 0.8624,
      "step": 2471
    },
    {
      "epoch": 0.18136962682392216,
      "grad_norm": 0.6493992943348229,
      "learning_rate": 9.411075758084422e-06,
      "loss": 0.5357,
      "step": 2472
    },
    {
      "epoch": 0.18136962682392216,
      "lm_loss": 1.2433,
      "step": 2472,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.18136962682392216,
      "lm_loss": 1.2549,
      "step": 2472,
      "vm_loss": 0.1048
    },
    {
      "epoch": 0.18136962682392216,
      "lm_loss": 0.6196,
      "step": 2472,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.18136962682392216,
      "lm_loss": 1.0247,
      "step": 2472,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.18136962682392216,
      "lm_loss": 0.0856,
      "step": 2472,
      "vm_loss": 0.1212
    },
    {
      "epoch": 0.18136962682392216,
      "lm_loss": 0.0837,
      "step": 2472,
      "vm_loss": 0.143
    },
    {
      "epoch": 0.18136962682392216,
      "lm_loss": 0.7288,
      "step": 2472,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.18136962682392216,
      "lm_loss": 0.1568,
      "step": 2472,
      "vm_loss": 0.1318
    },
    {
      "epoch": 0.18144299641406128,
      "grad_norm": 1.5973012325333473,
      "learning_rate": 9.41051617517731e-06,
      "loss": 0.8517,
      "step": 2473
    },
    {
      "epoch": 0.1815163660042004,
      "grad_norm": 1.641738346010027,
      "learning_rate": 9.409956343197424e-06,
      "loss": 0.9093,
      "step": 2474
    },
    {
      "epoch": 0.18158973559433952,
      "grad_norm": 1.7776737837774756,
      "learning_rate": 9.409396262176383e-06,
      "loss": 0.7802,
      "step": 2475
    },
    {
      "epoch": 0.18166310518447867,
      "grad_norm": 0.696873474777691,
      "learning_rate": 9.408835932145816e-06,
      "loss": 0.5331,
      "step": 2476
    },
    {
      "epoch": 0.1817364747746178,
      "grad_norm": 1.6833176622820896,
      "learning_rate": 9.408275353137364e-06,
      "loss": 0.8106,
      "step": 2477
    },
    {
      "epoch": 0.18180984436475692,
      "grad_norm": 0.6215506902371953,
      "learning_rate": 9.407714525182687e-06,
      "loss": 0.5496,
      "step": 2478
    },
    {
      "epoch": 0.18188321395489604,
      "grad_norm": 0.5831288497614221,
      "learning_rate": 9.407153448313458e-06,
      "loss": 0.557,
      "step": 2479
    },
    {
      "epoch": 0.18195658354503516,
      "grad_norm": 0.5995164899377361,
      "learning_rate": 9.406592122561355e-06,
      "loss": 0.5433,
      "step": 2480
    },
    {
      "epoch": 0.18195658354503516,
      "lm_loss": 1.118,
      "step": 2480,
      "vm_loss": 0.0478
    },
    {
      "epoch": 0.18195658354503516,
      "lm_loss": 1.5696,
      "step": 2480,
      "vm_loss": 0.2161
    },
    {
      "epoch": 0.18195658354503516,
      "lm_loss": 0.6167,
      "step": 2480,
      "vm_loss": 0.2109
    },
    {
      "epoch": 0.18195658354503516,
      "lm_loss": 0.2842,
      "step": 2480,
      "vm_loss": 0.0725
    },
    {
      "epoch": 0.18195658354503516,
      "lm_loss": 0.0921,
      "step": 2480,
      "vm_loss": 0.182
    },
    {
      "epoch": 0.18195658354503516,
      "lm_loss": 0.4013,
      "step": 2480,
      "vm_loss": 0.0891
    },
    {
      "epoch": 0.18195658354503516,
      "lm_loss": 0.336,
      "step": 2480,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.18195658354503516,
      "lm_loss": 0.2942,
      "step": 2480,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.1820299531351743,
      "grad_norm": 1.6179008559713002,
      "learning_rate": 9.406030547958085e-06,
      "loss": 0.8817,
      "step": 2481
    },
    {
      "epoch": 0.18210332272531343,
      "grad_norm": 0.6315733198242451,
      "learning_rate": 9.405468724535358e-06,
      "loss": 0.5346,
      "step": 2482
    },
    {
      "epoch": 0.18217669231545255,
      "grad_norm": 1.589942306117342,
      "learning_rate": 9.404906652324903e-06,
      "loss": 0.7932,
      "step": 2483
    },
    {
      "epoch": 0.18225006190559168,
      "grad_norm": 0.6198373930867562,
      "learning_rate": 9.40434433135846e-06,
      "loss": 0.5428,
      "step": 2484
    },
    {
      "epoch": 0.1823234314957308,
      "grad_norm": 1.5347680961519852,
      "learning_rate": 9.403781761667784e-06,
      "loss": 0.9027,
      "step": 2485
    },
    {
      "epoch": 0.18239680108586995,
      "grad_norm": 1.598959303165296,
      "learning_rate": 9.403218943284646e-06,
      "loss": 0.8767,
      "step": 2486
    },
    {
      "epoch": 0.18247017067600907,
      "grad_norm": 0.6110945764787418,
      "learning_rate": 9.402655876240833e-06,
      "loss": 0.5359,
      "step": 2487
    },
    {
      "epoch": 0.1825435402661482,
      "grad_norm": 1.505058323161946,
      "learning_rate": 9.402092560568137e-06,
      "loss": 0.8682,
      "step": 2488
    },
    {
      "epoch": 0.1825435402661482,
      "lm_loss": 1.5537,
      "step": 2488,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.1825435402661482,
      "lm_loss": 0.1937,
      "step": 2488,
      "vm_loss": 0.1162
    },
    {
      "epoch": 0.1825435402661482,
      "lm_loss": 0.6014,
      "step": 2488,
      "vm_loss": 0.0966
    },
    {
      "epoch": 0.1825435402661482,
      "lm_loss": 1.1747,
      "step": 2488,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.1825435402661482,
      "lm_loss": 0.579,
      "step": 2488,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.1825435402661482,
      "lm_loss": 0.4496,
      "step": 2488,
      "vm_loss": 0.1196
    },
    {
      "epoch": 0.1825435402661482,
      "lm_loss": 1.275,
      "step": 2488,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.1825435402661482,
      "lm_loss": 0.3833,
      "step": 2488,
      "vm_loss": 0.114
    },
    {
      "epoch": 0.1826169098562873,
      "grad_norm": 1.528958973564177,
      "learning_rate": 9.401528996298375e-06,
      "loss": 0.8609,
      "step": 2489
    },
    {
      "epoch": 0.18269027944642643,
      "grad_norm": 1.505432254435788,
      "learning_rate": 9.400965183463368e-06,
      "loss": 0.8471,
      "step": 2490
    },
    {
      "epoch": 0.18276364903656556,
      "grad_norm": 1.4758689493288721,
      "learning_rate": 9.40040112209496e-06,
      "loss": 0.8151,
      "step": 2491
    },
    {
      "epoch": 0.1828370186267047,
      "grad_norm": 1.558618928421936,
      "learning_rate": 9.399836812225e-06,
      "loss": 0.8688,
      "step": 2492
    },
    {
      "epoch": 0.18291038821684383,
      "grad_norm": 1.4922854580012666,
      "learning_rate": 9.399272253885363e-06,
      "loss": 0.8255,
      "step": 2493
    },
    {
      "epoch": 0.18298375780698295,
      "grad_norm": 1.557642042786737,
      "learning_rate": 9.398707447107926e-06,
      "loss": 0.9106,
      "step": 2494
    },
    {
      "epoch": 0.18305712739712207,
      "grad_norm": 1.6092432189812913,
      "learning_rate": 9.398142391924585e-06,
      "loss": 0.8211,
      "step": 2495
    },
    {
      "epoch": 0.1831304969872612,
      "grad_norm": 1.5766068179969999,
      "learning_rate": 9.397577088367253e-06,
      "loss": 0.7271,
      "step": 2496
    },
    {
      "epoch": 0.1831304969872612,
      "lm_loss": 1.1775,
      "step": 2496,
      "vm_loss": 0.0215
    },
    {
      "epoch": 0.1831304969872612,
      "lm_loss": 1.3866,
      "step": 2496,
      "vm_loss": 0.079
    },
    {
      "epoch": 0.1831304969872612,
      "lm_loss": 0.1651,
      "step": 2496,
      "vm_loss": 0.134
    },
    {
      "epoch": 0.1831304969872612,
      "lm_loss": 0.2459,
      "step": 2496,
      "vm_loss": 0.1419
    },
    {
      "epoch": 0.1831304969872612,
      "lm_loss": 0.5772,
      "step": 2496,
      "vm_loss": 0.0959
    },
    {
      "epoch": 0.1831304969872612,
      "lm_loss": 0.8459,
      "step": 2496,
      "vm_loss": 0.1227
    },
    {
      "epoch": 0.1831304969872612,
      "lm_loss": 0.6211,
      "step": 2496,
      "vm_loss": 0.2158
    },
    {
      "epoch": 0.1831304969872612,
      "lm_loss": 0.4132,
      "step": 2496,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.18320386657740034,
      "grad_norm": 1.6990694692975945,
      "learning_rate": 9.397011536467851e-06,
      "loss": 0.8103,
      "step": 2497
    },
    {
      "epoch": 0.18327723616753946,
      "grad_norm": 1.6146673897361756,
      "learning_rate": 9.396445736258319e-06,
      "loss": 0.9083,
      "step": 2498
    },
    {
      "epoch": 0.1833506057576786,
      "grad_norm": 1.6648578546097565,
      "learning_rate": 9.395879687770611e-06,
      "loss": 0.9003,
      "step": 2499
    },
    {
      "epoch": 0.1834239753478177,
      "grad_norm": 0.9096154991143108,
      "learning_rate": 9.395313391036687e-06,
      "loss": 0.5494,
      "step": 2500
    },
    {
      "epoch": 0.18349734493795683,
      "grad_norm": 1.5501231322290028,
      "learning_rate": 9.394746846088533e-06,
      "loss": 0.8759,
      "step": 2501
    },
    {
      "epoch": 0.18357071452809598,
      "grad_norm": 1.8162246144480783,
      "learning_rate": 9.394180052958142e-06,
      "loss": 0.9514,
      "step": 2502
    },
    {
      "epoch": 0.1836440841182351,
      "grad_norm": 1.5834167573007152,
      "learning_rate": 9.39361301167752e-06,
      "loss": 0.7396,
      "step": 2503
    },
    {
      "epoch": 0.18371745370837422,
      "grad_norm": 1.4898600905216457,
      "learning_rate": 9.39304572227869e-06,
      "loss": 0.9048,
      "step": 2504
    },
    {
      "epoch": 0.18371745370837422,
      "lm_loss": 1.335,
      "step": 2504,
      "vm_loss": 0.0698
    },
    {
      "epoch": 0.18371745370837422,
      "lm_loss": 1.4484,
      "step": 2504,
      "vm_loss": 0.0954
    },
    {
      "epoch": 0.18371745370837422,
      "lm_loss": 0.9827,
      "step": 2504,
      "vm_loss": 0.2775
    },
    {
      "epoch": 0.18371745370837422,
      "lm_loss": 0.3828,
      "step": 2504,
      "vm_loss": 0.0987
    },
    {
      "epoch": 0.18371745370837422,
      "lm_loss": 0.711,
      "step": 2504,
      "vm_loss": 0.167
    },
    {
      "epoch": 0.18371745370837422,
      "lm_loss": 0.8681,
      "step": 2504,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.18371745370837422,
      "lm_loss": 0.402,
      "step": 2504,
      "vm_loss": 0.0434
    },
    {
      "epoch": 0.18371745370837422,
      "lm_loss": 0.061,
      "step": 2504,
      "vm_loss": 0.1699
    },
    {
      "epoch": 0.18379082329851334,
      "grad_norm": 1.7729666107400954,
      "learning_rate": 9.39247818479369e-06,
      "loss": 0.9982,
      "step": 2505
    },
    {
      "epoch": 0.18386419288865247,
      "grad_norm": 1.4642195088913228,
      "learning_rate": 9.39191039925457e-06,
      "loss": 0.8374,
      "step": 2506
    },
    {
      "epoch": 0.18393756247879162,
      "grad_norm": 1.7177098202207322,
      "learning_rate": 9.391342365693393e-06,
      "loss": 0.8278,
      "step": 2507
    },
    {
      "epoch": 0.18401093206893074,
      "grad_norm": 1.6871140890805079,
      "learning_rate": 9.390774084142236e-06,
      "loss": 0.8751,
      "step": 2508
    },
    {
      "epoch": 0.18408430165906986,
      "grad_norm": 1.923863819614483,
      "learning_rate": 9.390205554633193e-06,
      "loss": 0.8905,
      "step": 2509
    },
    {
      "epoch": 0.18415767124920898,
      "grad_norm": 1.7976246125179642,
      "learning_rate": 9.389636777198371e-06,
      "loss": 0.9382,
      "step": 2510
    },
    {
      "epoch": 0.1842310408393481,
      "grad_norm": 1.051692467256347,
      "learning_rate": 9.38906775186989e-06,
      "loss": 0.553,
      "step": 2511
    },
    {
      "epoch": 0.18430441042948723,
      "grad_norm": 1.6769482626187997,
      "learning_rate": 9.388498478679883e-06,
      "loss": 0.9097,
      "step": 2512
    },
    {
      "epoch": 0.18430441042948723,
      "lm_loss": 1.1246,
      "step": 2512,
      "vm_loss": 0.0629
    },
    {
      "epoch": 0.18430441042948723,
      "lm_loss": 0.7784,
      "step": 2512,
      "vm_loss": 0.0891
    },
    {
      "epoch": 0.18430441042948723,
      "lm_loss": 0.0724,
      "step": 2512,
      "vm_loss": 0.2301
    },
    {
      "epoch": 0.18430441042948723,
      "lm_loss": 0.3507,
      "step": 2512,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.18430441042948723,
      "lm_loss": 0.7215,
      "step": 2512,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.18430441042948723,
      "lm_loss": 0.2253,
      "step": 2512,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.18430441042948723,
      "lm_loss": 1.2911,
      "step": 2512,
      "vm_loss": 0.0497
    },
    {
      "epoch": 0.18430441042948723,
      "lm_loss": 0.0988,
      "step": 2512,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.18437778001962637,
      "grad_norm": 1.5223361664166706,
      "learning_rate": 9.387928957660499e-06,
      "loss": 0.8666,
      "step": 2513
    },
    {
      "epoch": 0.1844511496097655,
      "grad_norm": 1.5301124858759558,
      "learning_rate": 9.3873591888439e-06,
      "loss": 0.7404,
      "step": 2514
    },
    {
      "epoch": 0.18452451919990462,
      "grad_norm": 1.783760337388731,
      "learning_rate": 9.386789172262262e-06,
      "loss": 0.8468,
      "step": 2515
    },
    {
      "epoch": 0.18459788879004374,
      "grad_norm": 1.6649905512334913,
      "learning_rate": 9.386218907947777e-06,
      "loss": 0.867,
      "step": 2516
    },
    {
      "epoch": 0.18467125838018286,
      "grad_norm": 1.4628886670424122,
      "learning_rate": 9.385648395932648e-06,
      "loss": 0.9063,
      "step": 2517
    },
    {
      "epoch": 0.184744627970322,
      "grad_norm": 1.6870287622734588,
      "learning_rate": 9.385077636249093e-06,
      "loss": 0.8958,
      "step": 2518
    },
    {
      "epoch": 0.18481799756046113,
      "grad_norm": 1.6462776330964288,
      "learning_rate": 9.384506628929344e-06,
      "loss": 0.8872,
      "step": 2519
    },
    {
      "epoch": 0.18489136715060026,
      "grad_norm": 1.5074075153687223,
      "learning_rate": 9.383935374005649e-06,
      "loss": 0.8032,
      "step": 2520
    },
    {
      "epoch": 0.18489136715060026,
      "lm_loss": 0.9189,
      "step": 2520,
      "vm_loss": 0.1427
    },
    {
      "epoch": 0.18489136715060026,
      "lm_loss": 1.2394,
      "step": 2520,
      "vm_loss": 0.1684
    },
    {
      "epoch": 0.18489136715060026,
      "lm_loss": 0.943,
      "step": 2520,
      "vm_loss": 0.1585
    },
    {
      "epoch": 0.18489136715060026,
      "lm_loss": 0.9701,
      "step": 2520,
      "vm_loss": 0.149
    },
    {
      "epoch": 0.18489136715060026,
      "lm_loss": 1.0603,
      "step": 2520,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.18489136715060026,
      "lm_loss": 0.4078,
      "step": 2520,
      "vm_loss": 0.1116
    },
    {
      "epoch": 0.18489136715060026,
      "lm_loss": 0.6603,
      "step": 2520,
      "vm_loss": 0.1294
    },
    {
      "epoch": 0.18489136715060026,
      "lm_loss": 0.4027,
      "step": 2520,
      "vm_loss": 0.1531
    },
    {
      "epoch": 0.18496473674073938,
      "grad_norm": 1.4750533031117021,
      "learning_rate": 9.383363871510264e-06,
      "loss": 0.8743,
      "step": 2521
    },
    {
      "epoch": 0.1850381063308785,
      "grad_norm": 1.7401181875205551,
      "learning_rate": 9.382792121475467e-06,
      "loss": 0.9397,
      "step": 2522
    },
    {
      "epoch": 0.18511147592101765,
      "grad_norm": 1.024865161107991,
      "learning_rate": 9.382220123933547e-06,
      "loss": 0.5451,
      "step": 2523
    },
    {
      "epoch": 0.18518484551115677,
      "grad_norm": 1.5701043782167285,
      "learning_rate": 9.381647878916803e-06,
      "loss": 0.8427,
      "step": 2524
    },
    {
      "epoch": 0.1852582151012959,
      "grad_norm": 0.6625461263500141,
      "learning_rate": 9.381075386457553e-06,
      "loss": 0.5575,
      "step": 2525
    },
    {
      "epoch": 0.18533158469143501,
      "grad_norm": 1.5839450436317692,
      "learning_rate": 9.380502646588124e-06,
      "loss": 0.801,
      "step": 2526
    },
    {
      "epoch": 0.18540495428157414,
      "grad_norm": 0.6778799793387952,
      "learning_rate": 9.379929659340865e-06,
      "loss": 0.5511,
      "step": 2527
    },
    {
      "epoch": 0.18547832387171329,
      "grad_norm": 0.7801983456028537,
      "learning_rate": 9.379356424748129e-06,
      "loss": 0.5551,
      "step": 2528
    },
    {
      "epoch": 0.18547832387171329,
      "lm_loss": 0.8778,
      "step": 2528,
      "vm_loss": 0.045
    },
    {
      "epoch": 0.18547832387171329,
      "lm_loss": 0.3533,
      "step": 2528,
      "vm_loss": 0.1686
    },
    {
      "epoch": 0.18547832387171329,
      "lm_loss": 0.8461,
      "step": 2528,
      "vm_loss": 0.1825
    },
    {
      "epoch": 0.18547832387171329,
      "lm_loss": 1.5817,
      "step": 2528,
      "vm_loss": 0.144
    },
    {
      "epoch": 0.18547832387171329,
      "lm_loss": 0.3488,
      "step": 2528,
      "vm_loss": 0.1431
    },
    {
      "epoch": 0.18547832387171329,
      "lm_loss": 0.788,
      "step": 2528,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.18547832387171329,
      "lm_loss": 0.2014,
      "step": 2528,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.18547832387171329,
      "lm_loss": 0.8466,
      "step": 2528,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.1855516934618524,
      "grad_norm": 1.7079219850558562,
      "learning_rate": 9.378782942842292e-06,
      "loss": 0.8335,
      "step": 2529
    },
    {
      "epoch": 0.18562506305199153,
      "grad_norm": 1.5582667665992287,
      "learning_rate": 9.378209213655739e-06,
      "loss": 0.8979,
      "step": 2530
    },
    {
      "epoch": 0.18569843264213065,
      "grad_norm": 0.6608779762660859,
      "learning_rate": 9.377635237220868e-06,
      "loss": 0.5613,
      "step": 2531
    },
    {
      "epoch": 0.18577180223226977,
      "grad_norm": 1.5475761849358265,
      "learning_rate": 9.377061013570095e-06,
      "loss": 0.8936,
      "step": 2532
    },
    {
      "epoch": 0.1858451718224089,
      "grad_norm": 1.597089926442097,
      "learning_rate": 9.376486542735847e-06,
      "loss": 0.9037,
      "step": 2533
    },
    {
      "epoch": 0.18591854141254804,
      "grad_norm": 0.632493076949344,
      "learning_rate": 9.375911824750566e-06,
      "loss": 0.5622,
      "step": 2534
    },
    {
      "epoch": 0.18599191100268717,
      "grad_norm": 1.6226269671300093,
      "learning_rate": 9.375336859646706e-06,
      "loss": 0.8425,
      "step": 2535
    },
    {
      "epoch": 0.1860652805928263,
      "grad_norm": 0.6036876115611282,
      "learning_rate": 9.374761647456739e-06,
      "loss": 0.5406,
      "step": 2536
    },
    {
      "epoch": 0.1860652805928263,
      "lm_loss": 1.039,
      "step": 2536,
      "vm_loss": 0.094
    },
    {
      "epoch": 0.1860652805928263,
      "lm_loss": 0.9596,
      "step": 2536,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.1860652805928263,
      "lm_loss": 0.7291,
      "step": 2536,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.1860652805928263,
      "lm_loss": 0.2815,
      "step": 2536,
      "vm_loss": 0.2103
    },
    {
      "epoch": 0.1860652805928263,
      "lm_loss": 0.4254,
      "step": 2536,
      "vm_loss": 0.1575
    },
    {
      "epoch": 0.1860652805928263,
      "lm_loss": 0.975,
      "step": 2536,
      "vm_loss": 0.1108
    },
    {
      "epoch": 0.1860652805928263,
      "lm_loss": 0.2457,
      "step": 2536,
      "vm_loss": 0.0775
    },
    {
      "epoch": 0.1860652805928263,
      "lm_loss": 0.3429,
      "step": 2536,
      "vm_loss": 0.0823
    },
    {
      "epoch": 0.1861386501829654,
      "grad_norm": 1.6240854421658872,
      "learning_rate": 9.374186188213148e-06,
      "loss": 0.7358,
      "step": 2537
    },
    {
      "epoch": 0.18621201977310453,
      "grad_norm": 0.5977393618787312,
      "learning_rate": 9.37361048194843e-06,
      "loss": 0.5494,
      "step": 2538
    },
    {
      "epoch": 0.18628538936324368,
      "grad_norm": 0.576925517056191,
      "learning_rate": 9.373034528695097e-06,
      "loss": 0.5305,
      "step": 2539
    },
    {
      "epoch": 0.1863587589533828,
      "grad_norm": 1.4856774260425853,
      "learning_rate": 9.372458328485674e-06,
      "loss": 0.8009,
      "step": 2540
    },
    {
      "epoch": 0.18643212854352192,
      "grad_norm": 0.5881615087709059,
      "learning_rate": 9.371881881352702e-06,
      "loss": 0.5352,
      "step": 2541
    },
    {
      "epoch": 0.18650549813366105,
      "grad_norm": 1.8135820498135924,
      "learning_rate": 9.371305187328732e-06,
      "loss": 0.923,
      "step": 2542
    },
    {
      "epoch": 0.18657886772380017,
      "grad_norm": 0.608820513983902,
      "learning_rate": 9.370728246446335e-06,
      "loss": 0.5357,
      "step": 2543
    },
    {
      "epoch": 0.18665223731393932,
      "grad_norm": 0.5692875151795167,
      "learning_rate": 9.370151058738089e-06,
      "loss": 0.551,
      "step": 2544
    },
    {
      "epoch": 0.18665223731393932,
      "lm_loss": 1.1336,
      "step": 2544,
      "vm_loss": 0.0661
    },
    {
      "epoch": 0.18665223731393932,
      "lm_loss": 0.7859,
      "step": 2544,
      "vm_loss": 0.1455
    },
    {
      "epoch": 0.18665223731393932,
      "lm_loss": 0.4761,
      "step": 2544,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.18665223731393932,
      "lm_loss": 0.4837,
      "step": 2544,
      "vm_loss": 0.1507
    },
    {
      "epoch": 0.18665223731393932,
      "lm_loss": 0.3841,
      "step": 2544,
      "vm_loss": 0.1143
    },
    {
      "epoch": 0.18665223731393932,
      "lm_loss": 0.7616,
      "step": 2544,
      "vm_loss": 0.0834
    },
    {
      "epoch": 0.18665223731393932,
      "lm_loss": 0.1351,
      "step": 2544,
      "vm_loss": 0.2256
    },
    {
      "epoch": 0.18665223731393932,
      "lm_loss": 0.0887,
      "step": 2544,
      "vm_loss": 0.0476
    },
    {
      "epoch": 0.18672560690407844,
      "grad_norm": 1.791104297555599,
      "learning_rate": 9.369573624236589e-06,
      "loss": 0.7399,
      "step": 2545
    },
    {
      "epoch": 0.18679897649421756,
      "grad_norm": 1.455768215259605,
      "learning_rate": 9.368995942974447e-06,
      "loss": 0.8046,
      "step": 2546
    },
    {
      "epoch": 0.18687234608435668,
      "grad_norm": 1.4876479075706266,
      "learning_rate": 9.368418014984285e-06,
      "loss": 0.8044,
      "step": 2547
    },
    {
      "epoch": 0.1869457156744958,
      "grad_norm": 1.4789887721822577,
      "learning_rate": 9.367839840298738e-06,
      "loss": 0.7858,
      "step": 2548
    },
    {
      "epoch": 0.18701908526463493,
      "grad_norm": 1.493930065281657,
      "learning_rate": 9.367261418950459e-06,
      "loss": 0.9038,
      "step": 2549
    },
    {
      "epoch": 0.18709245485477408,
      "grad_norm": 1.6285720258090892,
      "learning_rate": 9.366682750972112e-06,
      "loss": 0.9029,
      "step": 2550
    },
    {
      "epoch": 0.1871658244449132,
      "grad_norm": 1.707900658947582,
      "learning_rate": 9.366103836396377e-06,
      "loss": 0.9442,
      "step": 2551
    },
    {
      "epoch": 0.18723919403505232,
      "grad_norm": 1.4615257434897069,
      "learning_rate": 9.365524675255944e-06,
      "loss": 0.8971,
      "step": 2552
    },
    {
      "epoch": 0.18723919403505232,
      "lm_loss": 1.3967,
      "step": 2552,
      "vm_loss": 0.0245
    },
    {
      "epoch": 0.18723919403505232,
      "lm_loss": 0.3745,
      "step": 2552,
      "vm_loss": 0.1969
    },
    {
      "epoch": 0.18723919403505232,
      "lm_loss": 0.2367,
      "step": 2552,
      "vm_loss": 0.0988
    },
    {
      "epoch": 0.18723919403505232,
      "lm_loss": 1.0516,
      "step": 2552,
      "vm_loss": 0.1555
    },
    {
      "epoch": 0.18723919403505232,
      "lm_loss": 1.1379,
      "step": 2552,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.18723919403505232,
      "lm_loss": 0.8208,
      "step": 2552,
      "vm_loss": 0.1349
    },
    {
      "epoch": 0.18723919403505232,
      "lm_loss": 0.2181,
      "step": 2552,
      "vm_loss": 0.0635
    },
    {
      "epoch": 0.18723919403505232,
      "lm_loss": 0.2069,
      "step": 2552,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.18731256362519144,
      "grad_norm": 1.5658550364849795,
      "learning_rate": 9.364945267583523e-06,
      "loss": 0.785,
      "step": 2553
    },
    {
      "epoch": 0.18738593321533056,
      "grad_norm": 1.5185556868702739,
      "learning_rate": 9.364365613411832e-06,
      "loss": 0.8731,
      "step": 2554
    },
    {
      "epoch": 0.1874593028054697,
      "grad_norm": 1.6935465381407613,
      "learning_rate": 9.36378571277361e-06,
      "loss": 0.9534,
      "step": 2555
    },
    {
      "epoch": 0.18753267239560883,
      "grad_norm": 1.5657094700478684,
      "learning_rate": 9.363205565701598e-06,
      "loss": 0.79,
      "step": 2556
    },
    {
      "epoch": 0.18760604198574796,
      "grad_norm": 1.4820961573418852,
      "learning_rate": 9.362625172228564e-06,
      "loss": 0.7974,
      "step": 2557
    },
    {
      "epoch": 0.18767941157588708,
      "grad_norm": 1.547486544913681,
      "learning_rate": 9.362044532387282e-06,
      "loss": 0.7759,
      "step": 2558
    },
    {
      "epoch": 0.1877527811660262,
      "grad_norm": 1.6975374336956373,
      "learning_rate": 9.361463646210545e-06,
      "loss": 0.8668,
      "step": 2559
    },
    {
      "epoch": 0.18782615075616535,
      "grad_norm": 1.7043847027745136,
      "learning_rate": 9.360882513731154e-06,
      "loss": 0.7917,
      "step": 2560
    },
    {
      "epoch": 0.18782615075616535,
      "lm_loss": 1.4785,
      "step": 2560,
      "vm_loss": 0.0625
    },
    {
      "epoch": 0.18782615075616535,
      "lm_loss": 0.9778,
      "step": 2560,
      "vm_loss": 0.0371
    },
    {
      "epoch": 0.18782615075616535,
      "lm_loss": 0.2646,
      "step": 2560,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.18782615075616535,
      "lm_loss": 0.1696,
      "step": 2560,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.18782615075616535,
      "lm_loss": 0.0794,
      "step": 2560,
      "vm_loss": 0.0586
    },
    {
      "epoch": 0.18782615075616535,
      "lm_loss": 0.8218,
      "step": 2560,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.18782615075616535,
      "lm_loss": 0.98,
      "step": 2560,
      "vm_loss": 0.0952
    },
    {
      "epoch": 0.18782615075616535,
      "lm_loss": 0.9359,
      "step": 2560,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.18789952034630447,
      "grad_norm": 1.5797720730944047,
      "learning_rate": 9.360301134981926e-06,
      "loss": 0.8949,
      "step": 2561
    },
    {
      "epoch": 0.1879728899364436,
      "grad_norm": 0.8674979280890034,
      "learning_rate": 9.359719509995696e-06,
      "loss": 0.5445,
      "step": 2562
    },
    {
      "epoch": 0.18804625952658272,
      "grad_norm": 1.4440345291235581,
      "learning_rate": 9.35913763880531e-06,
      "loss": 0.8181,
      "step": 2563
    },
    {
      "epoch": 0.18811962911672184,
      "grad_norm": 1.6196250763235391,
      "learning_rate": 9.358555521443624e-06,
      "loss": 0.9309,
      "step": 2564
    },
    {
      "epoch": 0.188192998706861,
      "grad_norm": 0.6340519935370761,
      "learning_rate": 9.357973157943515e-06,
      "loss": 0.5488,
      "step": 2565
    },
    {
      "epoch": 0.1882663682970001,
      "grad_norm": 1.8689338437997367,
      "learning_rate": 9.357390548337868e-06,
      "loss": 0.8532,
      "step": 2566
    },
    {
      "epoch": 0.18833973788713923,
      "grad_norm": 0.6089879106266846,
      "learning_rate": 9.356807692659589e-06,
      "loss": 0.5508,
      "step": 2567
    },
    {
      "epoch": 0.18841310747727835,
      "grad_norm": 1.6878281535315023,
      "learning_rate": 9.356224590941588e-06,
      "loss": 0.775,
      "step": 2568
    },
    {
      "epoch": 0.18841310747727835,
      "lm_loss": 1.4833,
      "step": 2568,
      "vm_loss": 0.0481
    },
    {
      "epoch": 0.18841310747727835,
      "lm_loss": 0.4989,
      "step": 2568,
      "vm_loss": 0.1892
    },
    {
      "epoch": 0.18841310747727835,
      "lm_loss": 0.8676,
      "step": 2568,
      "vm_loss": 0.1059
    },
    {
      "epoch": 0.18841310747727835,
      "lm_loss": 0.4413,
      "step": 2568,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.18841310747727835,
      "lm_loss": 0.322,
      "step": 2568,
      "vm_loss": 0.1447
    },
    {
      "epoch": 0.18841310747727835,
      "lm_loss": 0.574,
      "step": 2568,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.18841310747727835,
      "lm_loss": 0.5743,
      "step": 2568,
      "vm_loss": 0.0927
    },
    {
      "epoch": 0.18841310747727835,
      "lm_loss": 0.6811,
      "step": 2568,
      "vm_loss": 0.1209
    },
    {
      "epoch": 0.18848647706741747,
      "grad_norm": 1.8064989028129026,
      "learning_rate": 9.355641243216798e-06,
      "loss": 0.8051,
      "step": 2569
    },
    {
      "epoch": 0.1885598466575566,
      "grad_norm": 1.4724461554244515,
      "learning_rate": 9.355057649518158e-06,
      "loss": 0.81,
      "step": 2570
    },
    {
      "epoch": 0.18863321624769575,
      "grad_norm": 1.6737761004806297,
      "learning_rate": 9.354473809878627e-06,
      "loss": 0.8446,
      "step": 2571
    },
    {
      "epoch": 0.18870658583783487,
      "grad_norm": 1.6166884299364002,
      "learning_rate": 9.35388972433118e-06,
      "loss": 0.8556,
      "step": 2572
    },
    {
      "epoch": 0.188779955427974,
      "grad_norm": 0.8762591906157089,
      "learning_rate": 9.353305392908795e-06,
      "loss": 0.5361,
      "step": 2573
    },
    {
      "epoch": 0.1888533250181131,
      "grad_norm": 1.5139033436695724,
      "learning_rate": 9.352720815644475e-06,
      "loss": 0.8552,
      "step": 2574
    },
    {
      "epoch": 0.18892669460825223,
      "grad_norm": 1.3936320683983625,
      "learning_rate": 9.352135992571231e-06,
      "loss": 0.7823,
      "step": 2575
    },
    {
      "epoch": 0.18900006419839138,
      "grad_norm": 1.5615686456822229,
      "learning_rate": 9.351550923722089e-06,
      "loss": 0.891,
      "step": 2576
    },
    {
      "epoch": 0.18900006419839138,
      "lm_loss": 1.2343,
      "step": 2576,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.18900006419839138,
      "lm_loss": 0.9307,
      "step": 2576,
      "vm_loss": 0.1215
    },
    {
      "epoch": 0.18900006419839138,
      "lm_loss": 0.3115,
      "step": 2576,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.18900006419839138,
      "lm_loss": 0.8935,
      "step": 2576,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.18900006419839138,
      "lm_loss": 0.5024,
      "step": 2576,
      "vm_loss": 0.1598
    },
    {
      "epoch": 0.18900006419839138,
      "lm_loss": 0.1821,
      "step": 2576,
      "vm_loss": 0.1629
    },
    {
      "epoch": 0.18900006419839138,
      "lm_loss": 0.7935,
      "step": 2576,
      "vm_loss": 0.1136
    },
    {
      "epoch": 0.18900006419839138,
      "lm_loss": 0.9031,
      "step": 2576,
      "vm_loss": 0.2475
    },
    {
      "epoch": 0.1890734337885305,
      "grad_norm": 1.465295043211412,
      "learning_rate": 9.35096560913009e-06,
      "loss": 0.9194,
      "step": 2577
    },
    {
      "epoch": 0.18914680337866963,
      "grad_norm": 1.5030002288423276,
      "learning_rate": 9.35038004882829e-06,
      "loss": 0.7705,
      "step": 2578
    },
    {
      "epoch": 0.18922017296880875,
      "grad_norm": 1.7499452463311567,
      "learning_rate": 9.349794242849752e-06,
      "loss": 0.9162,
      "step": 2579
    },
    {
      "epoch": 0.18929354255894787,
      "grad_norm": 1.596648709388716,
      "learning_rate": 9.349208191227563e-06,
      "loss": 0.82,
      "step": 2580
    },
    {
      "epoch": 0.18936691214908702,
      "grad_norm": 0.7525538868128427,
      "learning_rate": 9.348621893994815e-06,
      "loss": 0.5738,
      "step": 2581
    },
    {
      "epoch": 0.18944028173922614,
      "grad_norm": 1.7439193785264981,
      "learning_rate": 9.348035351184622e-06,
      "loss": 0.8459,
      "step": 2582
    },
    {
      "epoch": 0.18951365132936526,
      "grad_norm": 0.6227357659060612,
      "learning_rate": 9.347448562830104e-06,
      "loss": 0.5317,
      "step": 2583
    },
    {
      "epoch": 0.18958702091950438,
      "grad_norm": 1.5803886698483092,
      "learning_rate": 9.3468615289644e-06,
      "loss": 0.8064,
      "step": 2584
    },
    {
      "epoch": 0.18958702091950438,
      "lm_loss": 0.6518,
      "step": 2584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.18958702091950438,
      "lm_loss": 0.3573,
      "step": 2584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.18958702091950438,
      "lm_loss": 0.534,
      "step": 2584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.18958702091950438,
      "lm_loss": 0.4455,
      "step": 2584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.18958702091950438,
      "lm_loss": 0.6305,
      "step": 2584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.18958702091950438,
      "lm_loss": 0.6204,
      "step": 2584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.18958702091950438,
      "lm_loss": 0.6138,
      "step": 2584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.18958702091950438,
      "lm_loss": 0.4037,
      "step": 2584,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1896603905096435,
      "grad_norm": 0.5749890361814478,
      "learning_rate": 9.34627424962066e-06,
      "loss": 0.5763,
      "step": 2585
    },
    {
      "epoch": 0.18973376009978266,
      "grad_norm": 1.6022914440510285,
      "learning_rate": 9.345686724832049e-06,
      "loss": 0.8637,
      "step": 2586
    },
    {
      "epoch": 0.18980712968992178,
      "grad_norm": 1.634251641443594,
      "learning_rate": 9.345098954631747e-06,
      "loss": 0.7772,
      "step": 2587
    },
    {
      "epoch": 0.1898804992800609,
      "grad_norm": 0.6415298529910916,
      "learning_rate": 9.344510939052947e-06,
      "loss": 0.545,
      "step": 2588
    },
    {
      "epoch": 0.18995386887020002,
      "grad_norm": 1.683528763886828,
      "learning_rate": 9.343922678128854e-06,
      "loss": 0.8696,
      "step": 2589
    },
    {
      "epoch": 0.19002723846033914,
      "grad_norm": 1.7397869851859942,
      "learning_rate": 9.343334171892691e-06,
      "loss": 0.8404,
      "step": 2590
    },
    {
      "epoch": 0.19010060805047826,
      "grad_norm": 0.610017049616809,
      "learning_rate": 9.342745420377692e-06,
      "loss": 0.555,
      "step": 2591
    },
    {
      "epoch": 0.19017397764061741,
      "grad_norm": 1.4663384839890772,
      "learning_rate": 9.342156423617104e-06,
      "loss": 0.8041,
      "step": 2592
    },
    {
      "epoch": 0.19017397764061741,
      "lm_loss": 1.2941,
      "step": 2592,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.19017397764061741,
      "lm_loss": 0.1826,
      "step": 2592,
      "vm_loss": 0.0984
    },
    {
      "epoch": 0.19017397764061741,
      "lm_loss": 0.415,
      "step": 2592,
      "vm_loss": 0.0561
    },
    {
      "epoch": 0.19017397764061741,
      "lm_loss": 0.6073,
      "step": 2592,
      "vm_loss": 0.1652
    },
    {
      "epoch": 0.19017397764061741,
      "lm_loss": 0.1643,
      "step": 2592,
      "vm_loss": 0.1401
    },
    {
      "epoch": 0.19017397764061741,
      "lm_loss": 0.4356,
      "step": 2592,
      "vm_loss": 0.178
    },
    {
      "epoch": 0.19017397764061741,
      "lm_loss": 0.3585,
      "step": 2592,
      "vm_loss": 0.1679
    },
    {
      "epoch": 0.19017397764061741,
      "lm_loss": 0.4327,
      "step": 2592,
      "vm_loss": 0.2361
    },
    {
      "epoch": 0.19024734723075654,
      "grad_norm": 1.6504157523444747,
      "learning_rate": 9.34156718164419e-06,
      "loss": 0.7624,
      "step": 2593
    },
    {
      "epoch": 0.19032071682089566,
      "grad_norm": 0.5888816740810636,
      "learning_rate": 9.340977694492223e-06,
      "loss": 0.5351,
      "step": 2594
    },
    {
      "epoch": 0.19039408641103478,
      "grad_norm": 0.5990543551322995,
      "learning_rate": 9.340387962194497e-06,
      "loss": 0.5433,
      "step": 2595
    },
    {
      "epoch": 0.1904674560011739,
      "grad_norm": 1.702209166497084,
      "learning_rate": 9.339797984784316e-06,
      "loss": 0.8979,
      "step": 2596
    },
    {
      "epoch": 0.19054082559131305,
      "grad_norm": 0.5476679771736932,
      "learning_rate": 9.339207762294993e-06,
      "loss": 0.5322,
      "step": 2597
    },
    {
      "epoch": 0.19061419518145217,
      "grad_norm": 1.7097013286324607,
      "learning_rate": 9.338617294759861e-06,
      "loss": 0.7979,
      "step": 2598
    },
    {
      "epoch": 0.1906875647715913,
      "grad_norm": 1.443635983327831,
      "learning_rate": 9.338026582212268e-06,
      "loss": 0.8396,
      "step": 2599
    },
    {
      "epoch": 0.19076093436173042,
      "grad_norm": 1.5044050067881114,
      "learning_rate": 9.337435624685571e-06,
      "loss": 0.9031,
      "step": 2600
    },
    {
      "epoch": 0.19076093436173042,
      "lm_loss": 1.5079,
      "step": 2600,
      "vm_loss": 0.0717
    },
    {
      "epoch": 0.19076093436173042,
      "lm_loss": 0.5188,
      "step": 2600,
      "vm_loss": 0.0412
    },
    {
      "epoch": 0.19076093436173042,
      "lm_loss": 1.0163,
      "step": 2600,
      "vm_loss": 0.1225
    },
    {
      "epoch": 0.19076093436173042,
      "lm_loss": 0.2296,
      "step": 2600,
      "vm_loss": 0.0676
    },
    {
      "epoch": 0.19076093436173042,
      "lm_loss": 0.9126,
      "step": 2600,
      "vm_loss": 0.2125
    },
    {
      "epoch": 0.19076093436173042,
      "lm_loss": 0.5885,
      "step": 2600,
      "vm_loss": 0.0704
    },
    {
      "epoch": 0.19076093436173042,
      "lm_loss": 1.1654,
      "step": 2600,
      "vm_loss": 0.1249
    },
    {
      "epoch": 0.19076093436173042,
      "lm_loss": 0.8028,
      "step": 2600,
      "vm_loss": 0.1404
    },
    {
      "epoch": 0.19083430395186954,
      "grad_norm": 1.639786633275661,
      "learning_rate": 9.336844422213141e-06,
      "loss": 0.9033,
      "step": 2601
    },
    {
      "epoch": 0.1909076735420087,
      "grad_norm": 1.6212184508546355,
      "learning_rate": 9.336252974828368e-06,
      "loss": 0.8604,
      "step": 2602
    },
    {
      "epoch": 0.1909810431321478,
      "grad_norm": 0.6598192096560204,
      "learning_rate": 9.335661282564651e-06,
      "loss": 0.5476,
      "step": 2603
    },
    {
      "epoch": 0.19105441272228693,
      "grad_norm": 1.507742301535878,
      "learning_rate": 9.335069345455403e-06,
      "loss": 0.9708,
      "step": 2604
    },
    {
      "epoch": 0.19112778231242605,
      "grad_norm": 1.5790676923871632,
      "learning_rate": 9.334477163534055e-06,
      "loss": 0.8239,
      "step": 2605
    },
    {
      "epoch": 0.19120115190256518,
      "grad_norm": 0.5887745739389819,
      "learning_rate": 9.333884736834047e-06,
      "loss": 0.549,
      "step": 2606
    },
    {
      "epoch": 0.19127452149270432,
      "grad_norm": 1.6938613876394553,
      "learning_rate": 9.333292065388835e-06,
      "loss": 0.8257,
      "step": 2607
    },
    {
      "epoch": 0.19134789108284345,
      "grad_norm": 1.4073109699582267,
      "learning_rate": 9.332699149231888e-06,
      "loss": 0.7926,
      "step": 2608
    },
    {
      "epoch": 0.19134789108284345,
      "lm_loss": 0.6324,
      "step": 2608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19134789108284345,
      "lm_loss": 0.4196,
      "step": 2608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19134789108284345,
      "lm_loss": 0.6089,
      "step": 2608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19134789108284345,
      "lm_loss": 0.4138,
      "step": 2608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19134789108284345,
      "lm_loss": 0.5017,
      "step": 2608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19134789108284345,
      "lm_loss": 0.5519,
      "step": 2608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19134789108284345,
      "lm_loss": 0.5835,
      "step": 2608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19134789108284345,
      "lm_loss": 0.5614,
      "step": 2608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19142126067298257,
      "grad_norm": 0.6456639504931018,
      "learning_rate": 9.332105988396692e-06,
      "loss": 0.5429,
      "step": 2609
    },
    {
      "epoch": 0.1914946302631217,
      "grad_norm": 1.7126828873494802,
      "learning_rate": 9.331512582916743e-06,
      "loss": 0.9001,
      "step": 2610
    },
    {
      "epoch": 0.1915679998532608,
      "grad_norm": 0.6121254731508978,
      "learning_rate": 9.330918932825548e-06,
      "loss": 0.5551,
      "step": 2611
    },
    {
      "epoch": 0.19164136944339993,
      "grad_norm": 1.6401945724288043,
      "learning_rate": 9.330325038156638e-06,
      "loss": 0.9372,
      "step": 2612
    },
    {
      "epoch": 0.19171473903353908,
      "grad_norm": 1.609892200022743,
      "learning_rate": 9.32973089894355e-06,
      "loss": 0.865,
      "step": 2613
    },
    {
      "epoch": 0.1917881086236782,
      "grad_norm": 1.4184353225011261,
      "learning_rate": 9.329136515219834e-06,
      "loss": 0.8685,
      "step": 2614
    },
    {
      "epoch": 0.19186147821381733,
      "grad_norm": 1.4850666381752782,
      "learning_rate": 9.32854188701906e-06,
      "loss": 0.8754,
      "step": 2615
    },
    {
      "epoch": 0.19193484780395645,
      "grad_norm": 0.6515815138767397,
      "learning_rate": 9.327947014374804e-06,
      "loss": 0.5402,
      "step": 2616
    },
    {
      "epoch": 0.19193484780395645,
      "lm_loss": 1.3222,
      "step": 2616,
      "vm_loss": 0.0525
    },
    {
      "epoch": 0.19193484780395645,
      "lm_loss": 0.9073,
      "step": 2616,
      "vm_loss": 0.1916
    },
    {
      "epoch": 0.19193484780395645,
      "lm_loss": 0.244,
      "step": 2616,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.19193484780395645,
      "lm_loss": 1.6128,
      "step": 2616,
      "vm_loss": 0.1996
    },
    {
      "epoch": 0.19193484780395645,
      "lm_loss": 1.0465,
      "step": 2616,
      "vm_loss": 0.1052
    },
    {
      "epoch": 0.19193484780395645,
      "lm_loss": 0.2341,
      "step": 2616,
      "vm_loss": 0.1347
    },
    {
      "epoch": 0.19193484780395645,
      "lm_loss": 1.2066,
      "step": 2616,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.19193484780395645,
      "lm_loss": 0.7707,
      "step": 2616,
      "vm_loss": 0.0796
    },
    {
      "epoch": 0.19200821739409557,
      "grad_norm": 1.4469034281778266,
      "learning_rate": 9.327351897320662e-06,
      "loss": 0.8856,
      "step": 2617
    },
    {
      "epoch": 0.19208158698423472,
      "grad_norm": 1.781984880687301,
      "learning_rate": 9.326756535890244e-06,
      "loss": 0.7923,
      "step": 2618
    },
    {
      "epoch": 0.19215495657437384,
      "grad_norm": 1.6620903911166482,
      "learning_rate": 9.326160930117168e-06,
      "loss": 0.8403,
      "step": 2619
    },
    {
      "epoch": 0.19222832616451296,
      "grad_norm": 1.517756772006008,
      "learning_rate": 9.325565080035071e-06,
      "loss": 0.8794,
      "step": 2620
    },
    {
      "epoch": 0.19230169575465209,
      "grad_norm": 1.677833664015872,
      "learning_rate": 9.324968985677603e-06,
      "loss": 0.8465,
      "step": 2621
    },
    {
      "epoch": 0.1923750653447912,
      "grad_norm": 1.4839132744543564,
      "learning_rate": 9.324372647078423e-06,
      "loss": 0.8026,
      "step": 2622
    },
    {
      "epoch": 0.19244843493493036,
      "grad_norm": 0.6146396151370362,
      "learning_rate": 9.323776064271213e-06,
      "loss": 0.5492,
      "step": 2623
    },
    {
      "epoch": 0.19252180452506948,
      "grad_norm": 1.840034797122103,
      "learning_rate": 9.323179237289661e-06,
      "loss": 0.9056,
      "step": 2624
    },
    {
      "epoch": 0.19252180452506948,
      "lm_loss": 0.5966,
      "step": 2624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19252180452506948,
      "lm_loss": 0.7247,
      "step": 2624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19252180452506948,
      "lm_loss": 0.4236,
      "step": 2624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19252180452506948,
      "lm_loss": 0.5312,
      "step": 2624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19252180452506948,
      "lm_loss": 0.4847,
      "step": 2624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19252180452506948,
      "lm_loss": 0.5409,
      "step": 2624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19252180452506948,
      "lm_loss": 0.5323,
      "step": 2624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19252180452506948,
      "lm_loss": 0.5741,
      "step": 2624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1925951741152086,
      "grad_norm": 0.6101697543884255,
      "learning_rate": 9.32258216616747e-06,
      "loss": 0.5506,
      "step": 2625
    },
    {
      "epoch": 0.19266854370534772,
      "grad_norm": 1.7343383872196794,
      "learning_rate": 9.32198485093836e-06,
      "loss": 0.8486,
      "step": 2626
    },
    {
      "epoch": 0.19274191329548684,
      "grad_norm": 1.4793439690471426,
      "learning_rate": 9.32138729163606e-06,
      "loss": 0.7965,
      "step": 2627
    },
    {
      "epoch": 0.192815282885626,
      "grad_norm": 0.5696160582445275,
      "learning_rate": 9.320789488294322e-06,
      "loss": 0.533,
      "step": 2628
    },
    {
      "epoch": 0.19288865247576512,
      "grad_norm": 1.7761483456594471,
      "learning_rate": 9.3201914409469e-06,
      "loss": 0.8881,
      "step": 2629
    },
    {
      "epoch": 0.19296202206590424,
      "grad_norm": 1.654295045399841,
      "learning_rate": 9.319593149627568e-06,
      "loss": 0.9176,
      "step": 2630
    },
    {
      "epoch": 0.19303539165604336,
      "grad_norm": 1.470583457143108,
      "learning_rate": 9.318994614370112e-06,
      "loss": 0.8187,
      "step": 2631
    },
    {
      "epoch": 0.19310876124618248,
      "grad_norm": 0.6696732337710056,
      "learning_rate": 9.318395835208335e-06,
      "loss": 0.555,
      "step": 2632
    },
    {
      "epoch": 0.19310876124618248,
      "lm_loss": 2.0797,
      "step": 2632,
      "vm_loss": 0.0632
    },
    {
      "epoch": 0.19310876124618248,
      "lm_loss": 0.4534,
      "step": 2632,
      "vm_loss": 0.1569
    },
    {
      "epoch": 0.19310876124618248,
      "lm_loss": 0.1381,
      "step": 2632,
      "vm_loss": 0.0596
    },
    {
      "epoch": 0.19310876124618248,
      "lm_loss": 0.2633,
      "step": 2632,
      "vm_loss": 0.253
    },
    {
      "epoch": 0.19310876124618248,
      "lm_loss": 0.5914,
      "step": 2632,
      "vm_loss": 0.1081
    },
    {
      "epoch": 0.19310876124618248,
      "lm_loss": 1.0414,
      "step": 2632,
      "vm_loss": 0.1188
    },
    {
      "epoch": 0.19310876124618248,
      "lm_loss": 0.3676,
      "step": 2632,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.19310876124618248,
      "lm_loss": 0.8368,
      "step": 2632,
      "vm_loss": 0.1517
    },
    {
      "epoch": 0.1931821308363216,
      "grad_norm": 1.494752913231835,
      "learning_rate": 9.317796812176054e-06,
      "loss": 0.8173,
      "step": 2633
    },
    {
      "epoch": 0.19325550042646075,
      "grad_norm": 1.342818506321852,
      "learning_rate": 9.31719754530709e-06,
      "loss": 0.7679,
      "step": 2634
    },
    {
      "epoch": 0.19332887001659987,
      "grad_norm": 1.5756035259716672,
      "learning_rate": 9.316598034635292e-06,
      "loss": 0.8251,
      "step": 2635
    },
    {
      "epoch": 0.193402239606739,
      "grad_norm": 1.7277244157372185,
      "learning_rate": 9.315998280194511e-06,
      "loss": 0.8383,
      "step": 2636
    },
    {
      "epoch": 0.19347560919687812,
      "grad_norm": 1.4508091379364536,
      "learning_rate": 9.315398282018622e-06,
      "loss": 0.845,
      "step": 2637
    },
    {
      "epoch": 0.19354897878701724,
      "grad_norm": 1.9465536190199313,
      "learning_rate": 9.314798040141502e-06,
      "loss": 0.8175,
      "step": 2638
    },
    {
      "epoch": 0.1936223483771564,
      "grad_norm": 0.6062859729830791,
      "learning_rate": 9.314197554597053e-06,
      "loss": 0.5462,
      "step": 2639
    },
    {
      "epoch": 0.1936957179672955,
      "grad_norm": 1.6631867106302767,
      "learning_rate": 9.313596825419183e-06,
      "loss": 0.8351,
      "step": 2640
    },
    {
      "epoch": 0.1936957179672955,
      "lm_loss": 1.5149,
      "step": 2640,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.1936957179672955,
      "lm_loss": 0.4081,
      "step": 2640,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.1936957179672955,
      "lm_loss": 0.125,
      "step": 2640,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.1936957179672955,
      "lm_loss": 0.9529,
      "step": 2640,
      "vm_loss": 0.1845
    },
    {
      "epoch": 0.1936957179672955,
      "lm_loss": 0.3917,
      "step": 2640,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.1936957179672955,
      "lm_loss": 1.1242,
      "step": 2640,
      "vm_loss": 0.1858
    },
    {
      "epoch": 0.1936957179672955,
      "lm_loss": 0.5325,
      "step": 2640,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.1936957179672955,
      "lm_loss": 0.1354,
      "step": 2640,
      "vm_loss": 0.0379
    },
    {
      "epoch": 0.19376908755743463,
      "grad_norm": 1.4355943266659685,
      "learning_rate": 9.312995852641817e-06,
      "loss": 0.8284,
      "step": 2641
    },
    {
      "epoch": 0.19384245714757375,
      "grad_norm": 1.5232791437079027,
      "learning_rate": 9.312394636298896e-06,
      "loss": 0.8055,
      "step": 2642
    },
    {
      "epoch": 0.19391582673771288,
      "grad_norm": 1.5607232492190857,
      "learning_rate": 9.311793176424371e-06,
      "loss": 0.9508,
      "step": 2643
    },
    {
      "epoch": 0.19398919632785203,
      "grad_norm": 1.5562308895685977,
      "learning_rate": 9.311191473052206e-06,
      "loss": 0.9284,
      "step": 2644
    },
    {
      "epoch": 0.19406256591799115,
      "grad_norm": 1.475644164416973,
      "learning_rate": 9.310589526216383e-06,
      "loss": 0.8901,
      "step": 2645
    },
    {
      "epoch": 0.19413593550813027,
      "grad_norm": 1.6454139195528596,
      "learning_rate": 9.309987335950894e-06,
      "loss": 0.9715,
      "step": 2646
    },
    {
      "epoch": 0.1942093050982694,
      "grad_norm": 1.7640626070572674,
      "learning_rate": 9.309384902289747e-06,
      "loss": 0.8871,
      "step": 2647
    },
    {
      "epoch": 0.1942826746884085,
      "grad_norm": 1.9439206947546634,
      "learning_rate": 9.308782225266961e-06,
      "loss": 0.8526,
      "step": 2648
    },
    {
      "epoch": 0.1942826746884085,
      "lm_loss": 0.4937,
      "step": 2648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1942826746884085,
      "lm_loss": 0.5132,
      "step": 2648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1942826746884085,
      "lm_loss": 0.5265,
      "step": 2648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1942826746884085,
      "lm_loss": 0.5058,
      "step": 2648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1942826746884085,
      "lm_loss": 0.5077,
      "step": 2648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1942826746884085,
      "lm_loss": 0.5694,
      "step": 2648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1942826746884085,
      "lm_loss": 0.4589,
      "step": 2648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.1942826746884085,
      "lm_loss": 0.5254,
      "step": 2648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.19435604427854766,
      "grad_norm": 0.6664383700108177,
      "learning_rate": 9.308179304916573e-06,
      "loss": 0.5298,
      "step": 2649
    },
    {
      "epoch": 0.19442941386868678,
      "grad_norm": 1.4293169495543394,
      "learning_rate": 9.30757614127263e-06,
      "loss": 0.8848,
      "step": 2650
    },
    {
      "epoch": 0.1945027834588259,
      "grad_norm": 0.642497516691333,
      "learning_rate": 9.306972734369196e-06,
      "loss": 0.5588,
      "step": 2651
    },
    {
      "epoch": 0.19457615304896503,
      "grad_norm": 1.4393440801600996,
      "learning_rate": 9.306369084240345e-06,
      "loss": 0.777,
      "step": 2652
    },
    {
      "epoch": 0.19464952263910415,
      "grad_norm": 1.6141664860020433,
      "learning_rate": 9.305765190920168e-06,
      "loss": 0.8825,
      "step": 2653
    },
    {
      "epoch": 0.19472289222924327,
      "grad_norm": 1.6291160870215013,
      "learning_rate": 9.305161054442767e-06,
      "loss": 0.9005,
      "step": 2654
    },
    {
      "epoch": 0.19479626181938242,
      "grad_norm": 1.683063053603776,
      "learning_rate": 9.30455667484226e-06,
      "loss": 0.8372,
      "step": 2655
    },
    {
      "epoch": 0.19486963140952154,
      "grad_norm": 1.8216144238072165,
      "learning_rate": 9.303952052152777e-06,
      "loss": 0.8554,
      "step": 2656
    },
    {
      "epoch": 0.19486963140952154,
      "lm_loss": 1.4462,
      "step": 2656,
      "vm_loss": 0.0305
    },
    {
      "epoch": 0.19486963140952154,
      "lm_loss": 2.7059,
      "step": 2656,
      "vm_loss": 0.0965
    },
    {
      "epoch": 0.19486963140952154,
      "lm_loss": 0.1451,
      "step": 2656,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.19486963140952154,
      "lm_loss": 0.3992,
      "step": 2656,
      "vm_loss": 0.096
    },
    {
      "epoch": 0.19486963140952154,
      "lm_loss": 0.9092,
      "step": 2656,
      "vm_loss": 0.1591
    },
    {
      "epoch": 0.19486963140952154,
      "lm_loss": 1.1708,
      "step": 2656,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.19486963140952154,
      "lm_loss": 1.1189,
      "step": 2656,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.19486963140952154,
      "lm_loss": 0.6696,
      "step": 2656,
      "vm_loss": 0.1326
    },
    {
      "epoch": 0.19494300099966066,
      "grad_norm": 1.6565933473290795,
      "learning_rate": 9.303347186408463e-06,
      "loss": 0.9229,
      "step": 2657
    },
    {
      "epoch": 0.1950163705897998,
      "grad_norm": 0.7031656678241954,
      "learning_rate": 9.302742077643478e-06,
      "loss": 0.532,
      "step": 2658
    },
    {
      "epoch": 0.1950897401799389,
      "grad_norm": 1.541860204814638,
      "learning_rate": 9.30213672589199e-06,
      "loss": 0.7409,
      "step": 2659
    },
    {
      "epoch": 0.19516310977007806,
      "grad_norm": 0.6189869716600435,
      "learning_rate": 9.301531131188189e-06,
      "loss": 0.5488,
      "step": 2660
    },
    {
      "epoch": 0.19523647936021718,
      "grad_norm": 1.4889523647544387,
      "learning_rate": 9.30092529356627e-06,
      "loss": 0.9067,
      "step": 2661
    },
    {
      "epoch": 0.1953098489503563,
      "grad_norm": 1.7538268194852864,
      "learning_rate": 9.300319213060452e-06,
      "loss": 0.8006,
      "step": 2662
    },
    {
      "epoch": 0.19538321854049542,
      "grad_norm": 1.7536378212090384,
      "learning_rate": 9.299712889704956e-06,
      "loss": 0.8333,
      "step": 2663
    },
    {
      "epoch": 0.19545658813063455,
      "grad_norm": 1.4892990881826333,
      "learning_rate": 9.299106323534026e-06,
      "loss": 0.8031,
      "step": 2664
    },
    {
      "epoch": 0.19545658813063455,
      "lm_loss": 1.3631,
      "step": 2664,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.19545658813063455,
      "lm_loss": 0.6065,
      "step": 2664,
      "vm_loss": 0.0705
    },
    {
      "epoch": 0.19545658813063455,
      "lm_loss": 0.8101,
      "step": 2664,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.19545658813063455,
      "lm_loss": 0.5043,
      "step": 2664,
      "vm_loss": 0.0818
    },
    {
      "epoch": 0.19545658813063455,
      "lm_loss": 0.7367,
      "step": 2664,
      "vm_loss": 0.0723
    },
    {
      "epoch": 0.19545658813063455,
      "lm_loss": 0.1561,
      "step": 2664,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.19545658813063455,
      "lm_loss": 0.3005,
      "step": 2664,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.19545658813063455,
      "lm_loss": 0.5676,
      "step": 2664,
      "vm_loss": 0.1005
    },
    {
      "epoch": 0.1955299577207737,
      "grad_norm": 1.5503274383603274,
      "learning_rate": 9.298499514581914e-06,
      "loss": 0.8542,
      "step": 2665
    },
    {
      "epoch": 0.19560332731091282,
      "grad_norm": 1.4877413281166825,
      "learning_rate": 9.297892462882892e-06,
      "loss": 0.8543,
      "step": 2666
    },
    {
      "epoch": 0.19567669690105194,
      "grad_norm": 0.7108585170524859,
      "learning_rate": 9.297285168471237e-06,
      "loss": 0.5438,
      "step": 2667
    },
    {
      "epoch": 0.19575006649119106,
      "grad_norm": 1.4873411836583357,
      "learning_rate": 9.296677631381246e-06,
      "loss": 0.7775,
      "step": 2668
    },
    {
      "epoch": 0.19582343608133018,
      "grad_norm": 1.6372972642440182,
      "learning_rate": 9.29606985164723e-06,
      "loss": 0.8368,
      "step": 2669
    },
    {
      "epoch": 0.1958968056714693,
      "grad_norm": 0.6040250537784829,
      "learning_rate": 9.295461829303509e-06,
      "loss": 0.5625,
      "step": 2670
    },
    {
      "epoch": 0.19597017526160845,
      "grad_norm": 0.5895424458788621,
      "learning_rate": 9.294853564384424e-06,
      "loss": 0.5259,
      "step": 2671
    },
    {
      "epoch": 0.19604354485174758,
      "grad_norm": 1.4794270429756098,
      "learning_rate": 9.294245056924319e-06,
      "loss": 0.7822,
      "step": 2672
    },
    {
      "epoch": 0.19604354485174758,
      "lm_loss": 1.306,
      "step": 2672,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.19604354485174758,
      "lm_loss": 0.4826,
      "step": 2672,
      "vm_loss": 0.0852
    },
    {
      "epoch": 0.19604354485174758,
      "lm_loss": 0.3463,
      "step": 2672,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.19604354485174758,
      "lm_loss": 0.5209,
      "step": 2672,
      "vm_loss": 0.1908
    },
    {
      "epoch": 0.19604354485174758,
      "lm_loss": 1.0165,
      "step": 2672,
      "vm_loss": 0.0881
    },
    {
      "epoch": 0.19604354485174758,
      "lm_loss": 0.9594,
      "step": 2672,
      "vm_loss": 0.0981
    },
    {
      "epoch": 0.19604354485174758,
      "lm_loss": 0.5854,
      "step": 2672,
      "vm_loss": 0.1074
    },
    {
      "epoch": 0.19604354485174758,
      "lm_loss": 0.3073,
      "step": 2672,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.1961169144418867,
      "grad_norm": 1.906453583472227,
      "learning_rate": 9.293636306957563e-06,
      "loss": 0.86,
      "step": 2673
    },
    {
      "epoch": 0.19619028403202582,
      "grad_norm": 1.7349427776491138,
      "learning_rate": 9.29302731451853e-06,
      "loss": 0.8964,
      "step": 2674
    },
    {
      "epoch": 0.19626365362216494,
      "grad_norm": 2.119110516165551,
      "learning_rate": 9.292418079641614e-06,
      "loss": 0.9977,
      "step": 2675
    },
    {
      "epoch": 0.1963370232123041,
      "grad_norm": 1.568431990812854,
      "learning_rate": 9.291808602361219e-06,
      "loss": 0.7386,
      "step": 2676
    },
    {
      "epoch": 0.1964103928024432,
      "grad_norm": 0.7977608918481968,
      "learning_rate": 9.291198882711765e-06,
      "loss": 0.5659,
      "step": 2677
    },
    {
      "epoch": 0.19648376239258233,
      "grad_norm": 1.8829579669051804,
      "learning_rate": 9.290588920727681e-06,
      "loss": 0.8315,
      "step": 2678
    },
    {
      "epoch": 0.19655713198272146,
      "grad_norm": 1.6295919721403398,
      "learning_rate": 9.289978716443417e-06,
      "loss": 0.7443,
      "step": 2679
    },
    {
      "epoch": 0.19663050157286058,
      "grad_norm": 1.684821444389276,
      "learning_rate": 9.28936826989343e-06,
      "loss": 0.8615,
      "step": 2680
    },
    {
      "epoch": 0.19663050157286058,
      "lm_loss": 1.2558,
      "step": 2680,
      "vm_loss": 0.0706
    },
    {
      "epoch": 0.19663050157286058,
      "lm_loss": 1.0623,
      "step": 2680,
      "vm_loss": 0.2282
    },
    {
      "epoch": 0.19663050157286058,
      "lm_loss": 1.1938,
      "step": 2680,
      "vm_loss": 0.2358
    },
    {
      "epoch": 0.19663050157286058,
      "lm_loss": 0.4243,
      "step": 2680,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.19663050157286058,
      "lm_loss": 0.3825,
      "step": 2680,
      "vm_loss": 0.2055
    },
    {
      "epoch": 0.19663050157286058,
      "lm_loss": 1.1937,
      "step": 2680,
      "vm_loss": 0.115
    },
    {
      "epoch": 0.19663050157286058,
      "lm_loss": 0.1099,
      "step": 2680,
      "vm_loss": 0.085
    },
    {
      "epoch": 0.19663050157286058,
      "lm_loss": 0.8873,
      "step": 2680,
      "vm_loss": 0.1841
    },
    {
      "epoch": 0.19670387116299973,
      "grad_norm": 1.6741642109093893,
      "learning_rate": 9.288757581112194e-06,
      "loss": 0.8198,
      "step": 2681
    },
    {
      "epoch": 0.19677724075313885,
      "grad_norm": 1.786187250971065,
      "learning_rate": 9.288146650134198e-06,
      "loss": 0.8523,
      "step": 2682
    },
    {
      "epoch": 0.19685061034327797,
      "grad_norm": 0.6101254067112768,
      "learning_rate": 9.287535476993939e-06,
      "loss": 0.5449,
      "step": 2683
    },
    {
      "epoch": 0.1969239799334171,
      "grad_norm": 1.4986056918726902,
      "learning_rate": 9.286924061725935e-06,
      "loss": 0.8136,
      "step": 2684
    },
    {
      "epoch": 0.19699734952355621,
      "grad_norm": 1.5831982327760266,
      "learning_rate": 9.286312404364712e-06,
      "loss": 0.8385,
      "step": 2685
    },
    {
      "epoch": 0.19707071911369536,
      "grad_norm": 0.6063410826916344,
      "learning_rate": 9.285700504944815e-06,
      "loss": 0.5291,
      "step": 2686
    },
    {
      "epoch": 0.19714408870383449,
      "grad_norm": 1.6893084477344906,
      "learning_rate": 9.285088363500795e-06,
      "loss": 0.8364,
      "step": 2687
    },
    {
      "epoch": 0.1972174582939736,
      "grad_norm": 0.5706420824953755,
      "learning_rate": 9.284475980067224e-06,
      "loss": 0.5666,
      "step": 2688
    },
    {
      "epoch": 0.1972174582939736,
      "lm_loss": 1.4177,
      "step": 2688,
      "vm_loss": 0.0523
    },
    {
      "epoch": 0.1972174582939736,
      "lm_loss": 0.951,
      "step": 2688,
      "vm_loss": 0.2195
    },
    {
      "epoch": 0.1972174582939736,
      "lm_loss": 1.0375,
      "step": 2688,
      "vm_loss": 0.1319
    },
    {
      "epoch": 0.1972174582939736,
      "lm_loss": 0.7065,
      "step": 2688,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.1972174582939736,
      "lm_loss": 1.138,
      "step": 2688,
      "vm_loss": 0.1042
    },
    {
      "epoch": 0.1972174582939736,
      "lm_loss": 0.3688,
      "step": 2688,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.1972174582939736,
      "lm_loss": 0.6425,
      "step": 2688,
      "vm_loss": 0.0707
    },
    {
      "epoch": 0.1972174582939736,
      "lm_loss": 0.6145,
      "step": 2688,
      "vm_loss": 0.189
    },
    {
      "epoch": 0.19729082788411273,
      "grad_norm": 1.636416918524559,
      "learning_rate": 9.283863354678683e-06,
      "loss": 0.9524,
      "step": 2689
    },
    {
      "epoch": 0.19736419747425185,
      "grad_norm": 1.4920035764514312,
      "learning_rate": 9.28325048736977e-06,
      "loss": 0.9435,
      "step": 2690
    },
    {
      "epoch": 0.19743756706439097,
      "grad_norm": 1.6595553308917381,
      "learning_rate": 9.282637378175098e-06,
      "loss": 0.7905,
      "step": 2691
    },
    {
      "epoch": 0.19751093665453012,
      "grad_norm": 1.5306209363984085,
      "learning_rate": 9.282024027129282e-06,
      "loss": 0.8699,
      "step": 2692
    },
    {
      "epoch": 0.19758430624466924,
      "grad_norm": 1.940865128094411,
      "learning_rate": 9.281410434266968e-06,
      "loss": 0.8211,
      "step": 2693
    },
    {
      "epoch": 0.19765767583480837,
      "grad_norm": 0.7189049995127983,
      "learning_rate": 9.280796599622804e-06,
      "loss": 0.5332,
      "step": 2694
    },
    {
      "epoch": 0.1977310454249475,
      "grad_norm": 1.9654002235439756,
      "learning_rate": 9.280182523231454e-06,
      "loss": 0.8405,
      "step": 2695
    },
    {
      "epoch": 0.1978044150150866,
      "grad_norm": 0.6265740823533803,
      "learning_rate": 9.279568205127598e-06,
      "loss": 0.5539,
      "step": 2696
    },
    {
      "epoch": 0.1978044150150866,
      "lm_loss": 1.4361,
      "step": 2696,
      "vm_loss": 0.0898
    },
    {
      "epoch": 0.1978044150150866,
      "lm_loss": 0.4975,
      "step": 2696,
      "vm_loss": 0.169
    },
    {
      "epoch": 0.1978044150150866,
      "lm_loss": 0.0154,
      "step": 2696,
      "vm_loss": 0.1517
    },
    {
      "epoch": 0.1978044150150866,
      "lm_loss": 1.0759,
      "step": 2696,
      "vm_loss": 0.1239
    },
    {
      "epoch": 0.1978044150150866,
      "lm_loss": 0.8373,
      "step": 2696,
      "vm_loss": 0.0994
    },
    {
      "epoch": 0.1978044150150866,
      "lm_loss": 0.4344,
      "step": 2696,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.1978044150150866,
      "lm_loss": 1.7693,
      "step": 2696,
      "vm_loss": 0.1635
    },
    {
      "epoch": 0.1978044150150866,
      "lm_loss": 1.2445,
      "step": 2696,
      "vm_loss": 0.1754
    },
    {
      "epoch": 0.19787778460522576,
      "grad_norm": 1.7667608585350854,
      "learning_rate": 9.278953645345926e-06,
      "loss": 0.8827,
      "step": 2697
    },
    {
      "epoch": 0.19795115419536488,
      "grad_norm": 0.5797139984370139,
      "learning_rate": 9.278338843921146e-06,
      "loss": 0.5502,
      "step": 2698
    },
    {
      "epoch": 0.198024523785504,
      "grad_norm": 1.7732699064577593,
      "learning_rate": 9.277723800887977e-06,
      "loss": 0.8956,
      "step": 2699
    },
    {
      "epoch": 0.19809789337564312,
      "grad_norm": 1.4056129930847021,
      "learning_rate": 9.27710851628115e-06,
      "loss": 0.8114,
      "step": 2700
    },
    {
      "epoch": 0.19817126296578225,
      "grad_norm": 1.443424549332283,
      "learning_rate": 9.276492990135413e-06,
      "loss": 0.8694,
      "step": 2701
    },
    {
      "epoch": 0.1982446325559214,
      "grad_norm": 0.7043431538666916,
      "learning_rate": 9.275877222485527e-06,
      "loss": 0.5409,
      "step": 2702
    },
    {
      "epoch": 0.19831800214606052,
      "grad_norm": 1.6623670013207121,
      "learning_rate": 9.275261213366265e-06,
      "loss": 0.8731,
      "step": 2703
    },
    {
      "epoch": 0.19839137173619964,
      "grad_norm": 0.6610538912014526,
      "learning_rate": 9.274644962812415e-06,
      "loss": 0.5798,
      "step": 2704
    },
    {
      "epoch": 0.19839137173619964,
      "lm_loss": 0.7393,
      "step": 2704,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.19839137173619964,
      "lm_loss": 0.1111,
      "step": 2704,
      "vm_loss": 0.173
    },
    {
      "epoch": 0.19839137173619964,
      "lm_loss": 1.1039,
      "step": 2704,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.19839137173619964,
      "lm_loss": 1.1282,
      "step": 2704,
      "vm_loss": 0.1691
    },
    {
      "epoch": 0.19839137173619964,
      "lm_loss": 0.8189,
      "step": 2704,
      "vm_loss": 0.1846
    },
    {
      "epoch": 0.19839137173619964,
      "lm_loss": 0.9388,
      "step": 2704,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.19839137173619964,
      "lm_loss": 0.0369,
      "step": 2704,
      "vm_loss": 0.1377
    },
    {
      "epoch": 0.19839137173619964,
      "lm_loss": 0.7377,
      "step": 2704,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.19846474132633876,
      "grad_norm": 1.5549549877723188,
      "learning_rate": 9.274028470858778e-06,
      "loss": 0.8749,
      "step": 2705
    },
    {
      "epoch": 0.19853811091647788,
      "grad_norm": 1.6155155792500384,
      "learning_rate": 9.27341173754017e-06,
      "loss": 0.9333,
      "step": 2706
    },
    {
      "epoch": 0.19861148050661703,
      "grad_norm": 1.5616616719143908,
      "learning_rate": 9.272794762891416e-06,
      "loss": 0.9488,
      "step": 2707
    },
    {
      "epoch": 0.19868485009675615,
      "grad_norm": 1.6788670284835692,
      "learning_rate": 9.272177546947361e-06,
      "loss": 0.7834,
      "step": 2708
    },
    {
      "epoch": 0.19875821968689528,
      "grad_norm": 1.6490050189891052,
      "learning_rate": 9.27156008974286e-06,
      "loss": 0.9351,
      "step": 2709
    },
    {
      "epoch": 0.1988315892770344,
      "grad_norm": 1.4723204953924394,
      "learning_rate": 9.270942391312782e-06,
      "loss": 0.8426,
      "step": 2710
    },
    {
      "epoch": 0.19890495886717352,
      "grad_norm": 1.6523041794383362,
      "learning_rate": 9.270324451692012e-06,
      "loss": 0.8625,
      "step": 2711
    },
    {
      "epoch": 0.19897832845731264,
      "grad_norm": 1.76787640363328,
      "learning_rate": 9.269706270915443e-06,
      "loss": 0.9203,
      "step": 2712
    },
    {
      "epoch": 0.19897832845731264,
      "lm_loss": 0.5794,
      "step": 2712,
      "vm_loss": 0.033
    },
    {
      "epoch": 0.19897832845731264,
      "lm_loss": 1.4487,
      "step": 2712,
      "vm_loss": 0.0881
    },
    {
      "epoch": 0.19897832845731264,
      "lm_loss": 0.9703,
      "step": 2712,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.19897832845731264,
      "lm_loss": 0.1529,
      "step": 2712,
      "vm_loss": 0.0782
    },
    {
      "epoch": 0.19897832845731264,
      "lm_loss": 1.1162,
      "step": 2712,
      "vm_loss": 0.1476
    },
    {
      "epoch": 0.19897832845731264,
      "lm_loss": 0.4233,
      "step": 2712,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.19897832845731264,
      "lm_loss": 0.1176,
      "step": 2712,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.19897832845731264,
      "lm_loss": 0.5536,
      "step": 2712,
      "vm_loss": 0.1245
    },
    {
      "epoch": 0.1990516980474518,
      "grad_norm": 1.5518020904298888,
      "learning_rate": 9.26908784901799e-06,
      "loss": 0.8116,
      "step": 2713
    },
    {
      "epoch": 0.1991250676375909,
      "grad_norm": 1.6688549429819541,
      "learning_rate": 9.268469186034571e-06,
      "loss": 0.8742,
      "step": 2714
    },
    {
      "epoch": 0.19919843722773004,
      "grad_norm": 0.8761305231864559,
      "learning_rate": 9.267850282000128e-06,
      "loss": 0.5689,
      "step": 2715
    },
    {
      "epoch": 0.19927180681786916,
      "grad_norm": 1.6084447705301905,
      "learning_rate": 9.26723113694961e-06,
      "loss": 0.8961,
      "step": 2716
    },
    {
      "epoch": 0.19934517640800828,
      "grad_norm": 1.703575754073247,
      "learning_rate": 9.266611750917981e-06,
      "loss": 0.9358,
      "step": 2717
    },
    {
      "epoch": 0.19941854599814743,
      "grad_norm": 1.4657337347727482,
      "learning_rate": 9.265992123940221e-06,
      "loss": 0.8524,
      "step": 2718
    },
    {
      "epoch": 0.19949191558828655,
      "grad_norm": 1.7688169955773705,
      "learning_rate": 9.265372256051322e-06,
      "loss": 0.9319,
      "step": 2719
    },
    {
      "epoch": 0.19956528517842567,
      "grad_norm": 1.439896238309071,
      "learning_rate": 9.264752147286289e-06,
      "loss": 0.7998,
      "step": 2720
    },
    {
      "epoch": 0.19956528517842567,
      "lm_loss": 1.2353,
      "step": 2720,
      "vm_loss": 0.1107
    },
    {
      "epoch": 0.19956528517842567,
      "lm_loss": 0.3077,
      "step": 2720,
      "vm_loss": 0.0534
    },
    {
      "epoch": 0.19956528517842567,
      "lm_loss": 1.125,
      "step": 2720,
      "vm_loss": 0.1469
    },
    {
      "epoch": 0.19956528517842567,
      "lm_loss": 0.6001,
      "step": 2720,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.19956528517842567,
      "lm_loss": 0.2452,
      "step": 2720,
      "vm_loss": 0.1644
    },
    {
      "epoch": 0.19956528517842567,
      "lm_loss": 0.5152,
      "step": 2720,
      "vm_loss": 0.2138
    },
    {
      "epoch": 0.19956528517842567,
      "lm_loss": 0.5973,
      "step": 2720,
      "vm_loss": 0.0631
    },
    {
      "epoch": 0.19956528517842567,
      "lm_loss": 1.0288,
      "step": 2720,
      "vm_loss": 0.3255
    },
    {
      "epoch": 0.1996386547685648,
      "grad_norm": 1.6546764575973258,
      "learning_rate": 9.264131797680139e-06,
      "loss": 0.8268,
      "step": 2721
    },
    {
      "epoch": 0.19971202435870392,
      "grad_norm": 1.4982186385657297,
      "learning_rate": 9.263511207267907e-06,
      "loss": 0.7839,
      "step": 2722
    },
    {
      "epoch": 0.19978539394884307,
      "grad_norm": 1.5952222064119346,
      "learning_rate": 9.26289037608464e-06,
      "loss": 0.8842,
      "step": 2723
    },
    {
      "epoch": 0.1998587635389822,
      "grad_norm": 1.5485459147811325,
      "learning_rate": 9.262269304165394e-06,
      "loss": 0.9182,
      "step": 2724
    },
    {
      "epoch": 0.1999321331291213,
      "grad_norm": 1.655820497406885,
      "learning_rate": 9.261647991545248e-06,
      "loss": 0.8503,
      "step": 2725
    },
    {
      "epoch": 0.20000550271926043,
      "grad_norm": 0.7389471037324504,
      "learning_rate": 9.261026438259285e-06,
      "loss": 0.5467,
      "step": 2726
    },
    {
      "epoch": 0.20007887230939955,
      "grad_norm": 1.748656950416708,
      "learning_rate": 9.260404644342606e-06,
      "loss": 0.8581,
      "step": 2727
    },
    {
      "epoch": 0.2001522418995387,
      "grad_norm": 0.6362929505299132,
      "learning_rate": 9.259782609830327e-06,
      "loss": 0.5608,
      "step": 2728
    },
    {
      "epoch": 0.2001522418995387,
      "lm_loss": 1.2466,
      "step": 2728,
      "vm_loss": 0.0762
    },
    {
      "epoch": 0.2001522418995387,
      "lm_loss": 0.0573,
      "step": 2728,
      "vm_loss": 0.0454
    },
    {
      "epoch": 0.2001522418995387,
      "lm_loss": 1.0162,
      "step": 2728,
      "vm_loss": 0.2669
    },
    {
      "epoch": 0.2001522418995387,
      "lm_loss": 0.234,
      "step": 2728,
      "vm_loss": 0.1343
    },
    {
      "epoch": 0.2001522418995387,
      "lm_loss": 0.1554,
      "step": 2728,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.2001522418995387,
      "lm_loss": 0.686,
      "step": 2728,
      "vm_loss": 0.089
    },
    {
      "epoch": 0.2001522418995387,
      "lm_loss": 0.5193,
      "step": 2728,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.2001522418995387,
      "lm_loss": 0.6751,
      "step": 2728,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.20022561148967782,
      "grad_norm": 1.6023574742174298,
      "learning_rate": 9.259160334757575e-06,
      "loss": 0.7881,
      "step": 2729
    },
    {
      "epoch": 0.20029898107981695,
      "grad_norm": 1.6587293086514987,
      "learning_rate": 9.258537819159489e-06,
      "loss": 0.859,
      "step": 2730
    },
    {
      "epoch": 0.20037235066995607,
      "grad_norm": 1.7912686850743713,
      "learning_rate": 9.25791506307123e-06,
      "loss": 0.9315,
      "step": 2731
    },
    {
      "epoch": 0.2004457202600952,
      "grad_norm": 1.5085313052412384,
      "learning_rate": 9.25729206652796e-06,
      "loss": 0.953,
      "step": 2732
    },
    {
      "epoch": 0.2005190898502343,
      "grad_norm": 1.578757911941164,
      "learning_rate": 9.256668829564866e-06,
      "loss": 0.8417,
      "step": 2733
    },
    {
      "epoch": 0.20059245944037346,
      "grad_norm": 1.5049041392231746,
      "learning_rate": 9.25604535221714e-06,
      "loss": 0.81,
      "step": 2734
    },
    {
      "epoch": 0.20066582903051258,
      "grad_norm": 1.7439019081852516,
      "learning_rate": 9.255421634519993e-06,
      "loss": 0.8218,
      "step": 2735
    },
    {
      "epoch": 0.2007391986206517,
      "grad_norm": 1.5696025303499663,
      "learning_rate": 9.254797676508649e-06,
      "loss": 0.8421,
      "step": 2736
    },
    {
      "epoch": 0.2007391986206517,
      "lm_loss": 0.7504,
      "step": 2736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2007391986206517,
      "lm_loss": 0.6916,
      "step": 2736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2007391986206517,
      "lm_loss": 0.5761,
      "step": 2736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2007391986206517,
      "lm_loss": 0.4802,
      "step": 2736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2007391986206517,
      "lm_loss": 0.5831,
      "step": 2736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2007391986206517,
      "lm_loss": 0.5261,
      "step": 2736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2007391986206517,
      "lm_loss": 0.7148,
      "step": 2736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2007391986206517,
      "lm_loss": 0.5003,
      "step": 2736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20081256821079083,
      "grad_norm": 1.0073131083999647,
      "learning_rate": 9.254173478218342e-06,
      "loss": 0.5467,
      "step": 2737
    },
    {
      "epoch": 0.20088593780092995,
      "grad_norm": 1.7897943262822016,
      "learning_rate": 9.253549039684325e-06,
      "loss": 0.7671,
      "step": 2738
    },
    {
      "epoch": 0.2009593073910691,
      "grad_norm": 1.6007706736211225,
      "learning_rate": 9.25292436094186e-06,
      "loss": 0.7998,
      "step": 2739
    },
    {
      "epoch": 0.20103267698120822,
      "grad_norm": 1.7021825606188166,
      "learning_rate": 9.252299442026222e-06,
      "loss": 0.8311,
      "step": 2740
    },
    {
      "epoch": 0.20110604657134734,
      "grad_norm": 1.640050966964961,
      "learning_rate": 9.251674282972707e-06,
      "loss": 0.8205,
      "step": 2741
    },
    {
      "epoch": 0.20117941616148646,
      "grad_norm": 0.7743900068440391,
      "learning_rate": 9.251048883816613e-06,
      "loss": 0.5652,
      "step": 2742
    },
    {
      "epoch": 0.20125278575162558,
      "grad_norm": 1.5969169197296684,
      "learning_rate": 9.250423244593262e-06,
      "loss": 0.9793,
      "step": 2743
    },
    {
      "epoch": 0.20132615534176473,
      "grad_norm": 1.5886185863294908,
      "learning_rate": 9.249797365337986e-06,
      "loss": 0.9856,
      "step": 2744
    },
    {
      "epoch": 0.20132615534176473,
      "lm_loss": 0.4307,
      "step": 2744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20132615534176473,
      "lm_loss": 0.4442,
      "step": 2744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20132615534176473,
      "lm_loss": 0.5554,
      "step": 2744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20132615534176473,
      "lm_loss": 0.3866,
      "step": 2744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20132615534176473,
      "lm_loss": 0.5587,
      "step": 2744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20132615534176473,
      "lm_loss": 0.6746,
      "step": 2744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20132615534176473,
      "lm_loss": 0.5621,
      "step": 2744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20132615534176473,
      "lm_loss": 0.4825,
      "step": 2744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20139952493190386,
      "grad_norm": 0.652614022473919,
      "learning_rate": 9.249171246086128e-06,
      "loss": 0.558,
      "step": 2745
    },
    {
      "epoch": 0.20147289452204298,
      "grad_norm": 1.636701478530409,
      "learning_rate": 9.248544886873045e-06,
      "loss": 0.9066,
      "step": 2746
    },
    {
      "epoch": 0.2015462641121821,
      "grad_norm": 1.3960089294254816,
      "learning_rate": 9.247918287734112e-06,
      "loss": 0.8042,
      "step": 2747
    },
    {
      "epoch": 0.20161963370232122,
      "grad_norm": 0.6429652913401717,
      "learning_rate": 9.247291448704713e-06,
      "loss": 0.5687,
      "step": 2748
    },
    {
      "epoch": 0.20169300329246037,
      "grad_norm": 1.5332018152365563,
      "learning_rate": 9.246664369820249e-06,
      "loss": 0.8017,
      "step": 2749
    },
    {
      "epoch": 0.2017663728825995,
      "grad_norm": 1.6926924830361578,
      "learning_rate": 9.24603705111613e-06,
      "loss": 0.8993,
      "step": 2750
    },
    {
      "epoch": 0.20183974247273861,
      "grad_norm": 1.6126239464462993,
      "learning_rate": 9.245409492627783e-06,
      "loss": 0.9434,
      "step": 2751
    },
    {
      "epoch": 0.20191311206287774,
      "grad_norm": 1.558682101975675,
      "learning_rate": 9.24478169439065e-06,
      "loss": 0.6838,
      "step": 2752
    },
    {
      "epoch": 0.20191311206287774,
      "lm_loss": 1.4822,
      "step": 2752,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.20191311206287774,
      "lm_loss": 0.69,
      "step": 2752,
      "vm_loss": 0.1393
    },
    {
      "epoch": 0.20191311206287774,
      "lm_loss": 0.6821,
      "step": 2752,
      "vm_loss": 0.1498
    },
    {
      "epoch": 0.20191311206287774,
      "lm_loss": 1.1795,
      "step": 2752,
      "vm_loss": 0.1769
    },
    {
      "epoch": 0.20191311206287774,
      "lm_loss": 0.9033,
      "step": 2752,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.20191311206287774,
      "lm_loss": 0.5041,
      "step": 2752,
      "vm_loss": 0.0852
    },
    {
      "epoch": 0.20191311206287774,
      "lm_loss": 0.0991,
      "step": 2752,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.20191311206287774,
      "lm_loss": 0.3987,
      "step": 2752,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.20198648165301686,
      "grad_norm": 1.6039272590672693,
      "learning_rate": 9.244153656440182e-06,
      "loss": 0.7929,
      "step": 2753
    },
    {
      "epoch": 0.20205985124315598,
      "grad_norm": 1.3922346232933382,
      "learning_rate": 9.243525378811846e-06,
      "loss": 0.7768,
      "step": 2754
    },
    {
      "epoch": 0.20213322083329513,
      "grad_norm": 0.6944830524284149,
      "learning_rate": 9.242896861541124e-06,
      "loss": 0.5518,
      "step": 2755
    },
    {
      "epoch": 0.20220659042343425,
      "grad_norm": 0.6198256196151475,
      "learning_rate": 9.242268104663508e-06,
      "loss": 0.5258,
      "step": 2756
    },
    {
      "epoch": 0.20227996001357337,
      "grad_norm": 0.5883559803214344,
      "learning_rate": 9.241639108214508e-06,
      "loss": 0.5426,
      "step": 2757
    },
    {
      "epoch": 0.2023533296037125,
      "grad_norm": 1.7567835034515724,
      "learning_rate": 9.241009872229643e-06,
      "loss": 0.7708,
      "step": 2758
    },
    {
      "epoch": 0.20242669919385162,
      "grad_norm": 1.6335471011281537,
      "learning_rate": 9.240380396744446e-06,
      "loss": 0.8156,
      "step": 2759
    },
    {
      "epoch": 0.20250006878399077,
      "grad_norm": 1.6526996679805632,
      "learning_rate": 9.239750681794469e-06,
      "loss": 0.939,
      "step": 2760
    },
    {
      "epoch": 0.20250006878399077,
      "lm_loss": 1.5263,
      "step": 2760,
      "vm_loss": 0.1016
    },
    {
      "epoch": 0.20250006878399077,
      "lm_loss": 0.7533,
      "step": 2760,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.20250006878399077,
      "lm_loss": 1.1162,
      "step": 2760,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.20250006878399077,
      "lm_loss": 1.2111,
      "step": 2760,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.20250006878399077,
      "lm_loss": 0.8791,
      "step": 2760,
      "vm_loss": 0.0978
    },
    {
      "epoch": 0.20250006878399077,
      "lm_loss": 0.4388,
      "step": 2760,
      "vm_loss": 0.0524
    },
    {
      "epoch": 0.20250006878399077,
      "lm_loss": 0.3954,
      "step": 2760,
      "vm_loss": 0.062
    },
    {
      "epoch": 0.20250006878399077,
      "lm_loss": 0.3745,
      "step": 2760,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.2025734383741299,
      "grad_norm": 1.7343650930041414,
      "learning_rate": 9.23912072741527e-06,
      "loss": 0.8479,
      "step": 2761
    },
    {
      "epoch": 0.202646807964269,
      "grad_norm": 1.9522603451308502,
      "learning_rate": 9.238490533642425e-06,
      "loss": 0.8025,
      "step": 2762
    },
    {
      "epoch": 0.20272017755440813,
      "grad_norm": 1.5488734882132222,
      "learning_rate": 9.237860100511524e-06,
      "loss": 0.9129,
      "step": 2763
    },
    {
      "epoch": 0.20279354714454725,
      "grad_norm": 1.503052963816069,
      "learning_rate": 9.237229428058167e-06,
      "loss": 0.81,
      "step": 2764
    },
    {
      "epoch": 0.2028669167346864,
      "grad_norm": 1.5146355843398474,
      "learning_rate": 9.236598516317973e-06,
      "loss": 0.9031,
      "step": 2765
    },
    {
      "epoch": 0.20294028632482553,
      "grad_norm": 1.7458633240872539,
      "learning_rate": 9.235967365326568e-06,
      "loss": 0.8293,
      "step": 2766
    },
    {
      "epoch": 0.20301365591496465,
      "grad_norm": 0.9559356256437689,
      "learning_rate": 9.235335975119594e-06,
      "loss": 0.5518,
      "step": 2767
    },
    {
      "epoch": 0.20308702550510377,
      "grad_norm": 1.6841276829893062,
      "learning_rate": 9.234704345732712e-06,
      "loss": 0.8422,
      "step": 2768
    },
    {
      "epoch": 0.20308702550510377,
      "lm_loss": 0.5005,
      "step": 2768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20308702550510377,
      "lm_loss": 0.3959,
      "step": 2768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20308702550510377,
      "lm_loss": 0.4753,
      "step": 2768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20308702550510377,
      "lm_loss": 0.3939,
      "step": 2768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20308702550510377,
      "lm_loss": 0.4759,
      "step": 2768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20308702550510377,
      "lm_loss": 0.5049,
      "step": 2768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20308702550510377,
      "lm_loss": 0.5166,
      "step": 2768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20308702550510377,
      "lm_loss": 0.5013,
      "step": 2768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2031603950952429,
      "grad_norm": 0.7377196098220251,
      "learning_rate": 9.234072477201588e-06,
      "loss": 0.5419,
      "step": 2769
    },
    {
      "epoch": 0.20323376468538204,
      "grad_norm": 0.6286336075961666,
      "learning_rate": 9.233440369561903e-06,
      "loss": 0.5486,
      "step": 2770
    },
    {
      "epoch": 0.20330713427552116,
      "grad_norm": 2.1984777299285736,
      "learning_rate": 9.232808022849358e-06,
      "loss": 0.803,
      "step": 2771
    },
    {
      "epoch": 0.20338050386566028,
      "grad_norm": 1.6245025197881202,
      "learning_rate": 9.232175437099661e-06,
      "loss": 0.8547,
      "step": 2772
    },
    {
      "epoch": 0.2034538734557994,
      "grad_norm": 1.4097046006284608,
      "learning_rate": 9.231542612348538e-06,
      "loss": 0.8658,
      "step": 2773
    },
    {
      "epoch": 0.20352724304593853,
      "grad_norm": 1.057138686179434,
      "learning_rate": 9.230909548631723e-06,
      "loss": 0.5446,
      "step": 2774
    },
    {
      "epoch": 0.20360061263607765,
      "grad_norm": 0.9095660972453691,
      "learning_rate": 9.230276245984971e-06,
      "loss": 0.5457,
      "step": 2775
    },
    {
      "epoch": 0.2036739822262168,
      "grad_norm": 0.7017435537154734,
      "learning_rate": 9.22964270444404e-06,
      "loss": 0.5482,
      "step": 2776
    },
    {
      "epoch": 0.2036739822262168,
      "lm_loss": 1.2641,
      "step": 2776,
      "vm_loss": 0.0944
    },
    {
      "epoch": 0.2036739822262168,
      "lm_loss": 0.7753,
      "step": 2776,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.2036739822262168,
      "lm_loss": 0.753,
      "step": 2776,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.2036739822262168,
      "lm_loss": 0.9448,
      "step": 2776,
      "vm_loss": 0.1343
    },
    {
      "epoch": 0.2036739822262168,
      "lm_loss": 0.2474,
      "step": 2776,
      "vm_loss": 0.183
    },
    {
      "epoch": 0.2036739822262168,
      "lm_loss": 0.4063,
      "step": 2776,
      "vm_loss": 0.1212
    },
    {
      "epoch": 0.2036739822262168,
      "lm_loss": 0.8612,
      "step": 2776,
      "vm_loss": 0.0944
    },
    {
      "epoch": 0.2036739822262168,
      "lm_loss": 0.5175,
      "step": 2776,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.20374735181635592,
      "grad_norm": 1.9432381528952032,
      "learning_rate": 9.229008924044711e-06,
      "loss": 0.829,
      "step": 2777
    },
    {
      "epoch": 0.20382072140649504,
      "grad_norm": 1.7572321525063537,
      "learning_rate": 9.228374904822777e-06,
      "loss": 0.9642,
      "step": 2778
    },
    {
      "epoch": 0.20389409099663416,
      "grad_norm": 1.593352527227726,
      "learning_rate": 9.227740646814038e-06,
      "loss": 0.7968,
      "step": 2779
    },
    {
      "epoch": 0.20396746058677329,
      "grad_norm": 1.688610656478481,
      "learning_rate": 9.227106150054317e-06,
      "loss": 0.8829,
      "step": 2780
    },
    {
      "epoch": 0.20404083017691244,
      "grad_norm": 1.5816709314965627,
      "learning_rate": 9.226471414579444e-06,
      "loss": 0.885,
      "step": 2781
    },
    {
      "epoch": 0.20411419976705156,
      "grad_norm": 1.7039008882648985,
      "learning_rate": 9.225836440425261e-06,
      "loss": 0.8701,
      "step": 2782
    },
    {
      "epoch": 0.20418756935719068,
      "grad_norm": 1.6752928985415003,
      "learning_rate": 9.225201227627632e-06,
      "loss": 0.9244,
      "step": 2783
    },
    {
      "epoch": 0.2042609389473298,
      "grad_norm": 1.415502595154655,
      "learning_rate": 9.224565776222425e-06,
      "loss": 0.8095,
      "step": 2784
    },
    {
      "epoch": 0.2042609389473298,
      "lm_loss": 0.6928,
      "step": 2784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2042609389473298,
      "lm_loss": 0.6335,
      "step": 2784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2042609389473298,
      "lm_loss": 0.314,
      "step": 2784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2042609389473298,
      "lm_loss": 0.4119,
      "step": 2784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2042609389473298,
      "lm_loss": 0.5427,
      "step": 2784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2042609389473298,
      "lm_loss": 0.8099,
      "step": 2784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2042609389473298,
      "lm_loss": 0.6049,
      "step": 2784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2042609389473298,
      "lm_loss": 0.6315,
      "step": 2784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20433430853746892,
      "grad_norm": 1.6087217270151675,
      "learning_rate": 9.223930086245525e-06,
      "loss": 0.5912,
      "step": 2785
    },
    {
      "epoch": 0.20440767812760807,
      "grad_norm": 1.4888778414448611,
      "learning_rate": 9.223294157732834e-06,
      "loss": 0.7639,
      "step": 2786
    },
    {
      "epoch": 0.2044810477177472,
      "grad_norm": 1.5442573598100289,
      "learning_rate": 9.222657990720264e-06,
      "loss": 0.7705,
      "step": 2787
    },
    {
      "epoch": 0.20455441730788632,
      "grad_norm": 1.7038450225156163,
      "learning_rate": 9.222021585243739e-06,
      "loss": 0.8656,
      "step": 2788
    },
    {
      "epoch": 0.20462778689802544,
      "grad_norm": 0.973455236884678,
      "learning_rate": 9.2213849413392e-06,
      "loss": 0.5793,
      "step": 2789
    },
    {
      "epoch": 0.20470115648816456,
      "grad_norm": 1.6007216598740175,
      "learning_rate": 9.220748059042599e-06,
      "loss": 0.8745,
      "step": 2790
    },
    {
      "epoch": 0.20477452607830368,
      "grad_norm": 1.611865451591205,
      "learning_rate": 9.220110938389902e-06,
      "loss": 0.8874,
      "step": 2791
    },
    {
      "epoch": 0.20484789566844283,
      "grad_norm": 1.702670012060339,
      "learning_rate": 9.21947357941709e-06,
      "loss": 0.8483,
      "step": 2792
    },
    {
      "epoch": 0.20484789566844283,
      "lm_loss": 1.5732,
      "step": 2792,
      "vm_loss": 0.0423
    },
    {
      "epoch": 0.20484789566844283,
      "lm_loss": 0.8272,
      "step": 2792,
      "vm_loss": 0.0531
    },
    {
      "epoch": 0.20484789566844283,
      "lm_loss": 0.0703,
      "step": 2792,
      "vm_loss": 0.0976
    },
    {
      "epoch": 0.20484789566844283,
      "lm_loss": 0.8883,
      "step": 2792,
      "vm_loss": 0.1151
    },
    {
      "epoch": 0.20484789566844283,
      "lm_loss": 1.527,
      "step": 2792,
      "vm_loss": 0.0733
    },
    {
      "epoch": 0.20484789566844283,
      "lm_loss": 0.7314,
      "step": 2792,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.20484789566844283,
      "lm_loss": 0.1263,
      "step": 2792,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.20484789566844283,
      "lm_loss": 0.5948,
      "step": 2792,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.20492126525858195,
      "grad_norm": 1.6988088136683641,
      "learning_rate": 9.218835982160155e-06,
      "loss": 0.8271,
      "step": 2793
    },
    {
      "epoch": 0.20499463484872107,
      "grad_norm": 0.7418210173806065,
      "learning_rate": 9.218198146655104e-06,
      "loss": 0.5549,
      "step": 2794
    },
    {
      "epoch": 0.2050680044388602,
      "grad_norm": 1.843223171928186,
      "learning_rate": 9.217560072937958e-06,
      "loss": 0.9607,
      "step": 2795
    },
    {
      "epoch": 0.20514137402899932,
      "grad_norm": 0.7328795349808058,
      "learning_rate": 9.216921761044752e-06,
      "loss": 0.5627,
      "step": 2796
    },
    {
      "epoch": 0.20521474361913847,
      "grad_norm": 1.9819361342455024,
      "learning_rate": 9.21628321101153e-06,
      "loss": 0.9855,
      "step": 2797
    },
    {
      "epoch": 0.2052881132092776,
      "grad_norm": 0.661611576515597,
      "learning_rate": 9.215644422874352e-06,
      "loss": 0.5416,
      "step": 2798
    },
    {
      "epoch": 0.2053614827994167,
      "grad_norm": 0.6415057934967922,
      "learning_rate": 9.215005396669294e-06,
      "loss": 0.5397,
      "step": 2799
    },
    {
      "epoch": 0.20543485238955583,
      "grad_norm": 1.733101272417568,
      "learning_rate": 9.214366132432444e-06,
      "loss": 0.8156,
      "step": 2800
    },
    {
      "epoch": 0.20543485238955583,
      "lm_loss": 1.2422,
      "step": 2800,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.20543485238955583,
      "lm_loss": 0.2831,
      "step": 2800,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.20543485238955583,
      "lm_loss": 0.14,
      "step": 2800,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.20543485238955583,
      "lm_loss": 0.5962,
      "step": 2800,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.20543485238955583,
      "lm_loss": 0.9814,
      "step": 2800,
      "vm_loss": 0.0833
    },
    {
      "epoch": 0.20543485238955583,
      "lm_loss": 0.3357,
      "step": 2800,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.20543485238955583,
      "lm_loss": 0.7445,
      "step": 2800,
      "vm_loss": 0.1116
    },
    {
      "epoch": 0.20543485238955583,
      "lm_loss": 1.1637,
      "step": 2800,
      "vm_loss": 0.2245
    },
    {
      "epoch": 0.20550822197969495,
      "grad_norm": 1.9024686136956461,
      "learning_rate": 9.213726630199901e-06,
      "loss": 0.9002,
      "step": 2801
    },
    {
      "epoch": 0.2055815915698341,
      "grad_norm": 1.5810351139020666,
      "learning_rate": 9.21308689000778e-06,
      "loss": 0.8315,
      "step": 2802
    },
    {
      "epoch": 0.20565496115997323,
      "grad_norm": 1.6921917244600082,
      "learning_rate": 9.212446911892209e-06,
      "loss": 0.8458,
      "step": 2803
    },
    {
      "epoch": 0.20572833075011235,
      "grad_norm": 0.6842808471482817,
      "learning_rate": 9.21180669588933e-06,
      "loss": 0.5521,
      "step": 2804
    },
    {
      "epoch": 0.20580170034025147,
      "grad_norm": 1.6177433524092706,
      "learning_rate": 9.211166242035297e-06,
      "loss": 0.7482,
      "step": 2805
    },
    {
      "epoch": 0.2058750699303906,
      "grad_norm": 1.814537396566788,
      "learning_rate": 9.210525550366276e-06,
      "loss": 0.9419,
      "step": 2806
    },
    {
      "epoch": 0.20594843952052974,
      "grad_norm": 1.6139477530830648,
      "learning_rate": 9.20988462091845e-06,
      "loss": 0.9316,
      "step": 2807
    },
    {
      "epoch": 0.20602180911066886,
      "grad_norm": 1.6053498530120223,
      "learning_rate": 9.209243453728014e-06,
      "loss": 0.9193,
      "step": 2808
    },
    {
      "epoch": 0.20602180911066886,
      "lm_loss": 0.7398,
      "step": 2808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20602180911066886,
      "lm_loss": 0.4454,
      "step": 2808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20602180911066886,
      "lm_loss": 0.4756,
      "step": 2808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20602180911066886,
      "lm_loss": 0.6265,
      "step": 2808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20602180911066886,
      "lm_loss": 0.55,
      "step": 2808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20602180911066886,
      "lm_loss": 0.4533,
      "step": 2808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20602180911066886,
      "lm_loss": 0.5743,
      "step": 2808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20602180911066886,
      "lm_loss": 0.5387,
      "step": 2808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20609517870080798,
      "grad_norm": 0.6960412433680145,
      "learning_rate": 9.208602048831176e-06,
      "loss": 0.5433,
      "step": 2809
    },
    {
      "epoch": 0.2061685482909471,
      "grad_norm": 1.6269417431022763,
      "learning_rate": 9.207960406264159e-06,
      "loss": 0.8672,
      "step": 2810
    },
    {
      "epoch": 0.20624191788108623,
      "grad_norm": 1.577455572251361,
      "learning_rate": 9.207318526063195e-06,
      "loss": 0.8428,
      "step": 2811
    },
    {
      "epoch": 0.20631528747122535,
      "grad_norm": 1.5323268167845387,
      "learning_rate": 9.206676408264536e-06,
      "loss": 0.8735,
      "step": 2812
    },
    {
      "epoch": 0.2063886570613645,
      "grad_norm": 1.6668057836178174,
      "learning_rate": 9.206034052904442e-06,
      "loss": 0.9574,
      "step": 2813
    },
    {
      "epoch": 0.20646202665150362,
      "grad_norm": 1.805251406629048,
      "learning_rate": 9.205391460019188e-06,
      "loss": 0.8417,
      "step": 2814
    },
    {
      "epoch": 0.20653539624164274,
      "grad_norm": 0.6697628789573077,
      "learning_rate": 9.204748629645067e-06,
      "loss": 0.5483,
      "step": 2815
    },
    {
      "epoch": 0.20660876583178187,
      "grad_norm": 1.6422701531729866,
      "learning_rate": 9.204105561818374e-06,
      "loss": 0.7555,
      "step": 2816
    },
    {
      "epoch": 0.20660876583178187,
      "lm_loss": 1.2857,
      "step": 2816,
      "vm_loss": 0.0743
    },
    {
      "epoch": 0.20660876583178187,
      "lm_loss": 0.4762,
      "step": 2816,
      "vm_loss": 0.2131
    },
    {
      "epoch": 0.20660876583178187,
      "lm_loss": 1.0922,
      "step": 2816,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.20660876583178187,
      "lm_loss": 1.082,
      "step": 2816,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.20660876583178187,
      "lm_loss": 0.9,
      "step": 2816,
      "vm_loss": 0.1584
    },
    {
      "epoch": 0.20660876583178187,
      "lm_loss": 0.2355,
      "step": 2816,
      "vm_loss": 0.0606
    },
    {
      "epoch": 0.20660876583178187,
      "lm_loss": 0.5065,
      "step": 2816,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.20660876583178187,
      "lm_loss": 1.0475,
      "step": 2816,
      "vm_loss": 0.0995
    },
    {
      "epoch": 0.206682135421921,
      "grad_norm": 1.6810204410644656,
      "learning_rate": 9.203462256575431e-06,
      "loss": 0.8311,
      "step": 2817
    },
    {
      "epoch": 0.20675550501206014,
      "grad_norm": 1.7096854912155397,
      "learning_rate": 9.202818713952565e-06,
      "loss": 0.983,
      "step": 2818
    },
    {
      "epoch": 0.20682887460219926,
      "grad_norm": 1.493734626497853,
      "learning_rate": 9.202174933986118e-06,
      "loss": 0.7837,
      "step": 2819
    },
    {
      "epoch": 0.20690224419233838,
      "grad_norm": 1.758178373728349,
      "learning_rate": 9.201530916712444e-06,
      "loss": 0.8246,
      "step": 2820
    },
    {
      "epoch": 0.2069756137824775,
      "grad_norm": 1.6721595090544823,
      "learning_rate": 9.200886662167915e-06,
      "loss": 0.8784,
      "step": 2821
    },
    {
      "epoch": 0.20704898337261662,
      "grad_norm": 1.4880842511629686,
      "learning_rate": 9.200242170388916e-06,
      "loss": 0.858,
      "step": 2822
    },
    {
      "epoch": 0.20712235296275577,
      "grad_norm": 0.5794618936969264,
      "learning_rate": 9.199597441411836e-06,
      "loss": 0.5313,
      "step": 2823
    },
    {
      "epoch": 0.2071957225528949,
      "grad_norm": 0.603972049773285,
      "learning_rate": 9.19895247527309e-06,
      "loss": 0.5582,
      "step": 2824
    },
    {
      "epoch": 0.2071957225528949,
      "lm_loss": 0.2395,
      "step": 2824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2071957225528949,
      "lm_loss": 0.4593,
      "step": 2824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2071957225528949,
      "lm_loss": 0.5907,
      "step": 2824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2071957225528949,
      "lm_loss": 0.4553,
      "step": 2824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2071957225528949,
      "lm_loss": 0.6628,
      "step": 2824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2071957225528949,
      "lm_loss": 0.5614,
      "step": 2824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2071957225528949,
      "lm_loss": 0.567,
      "step": 2824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2071957225528949,
      "lm_loss": 0.5408,
      "step": 2824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.20726909214303402,
      "grad_norm": 0.5889450556672311,
      "learning_rate": 9.198307272009099e-06,
      "loss": 0.5396,
      "step": 2825
    },
    {
      "epoch": 0.20734246173317314,
      "grad_norm": 0.5539683372508872,
      "learning_rate": 9.197661831656299e-06,
      "loss": 0.5434,
      "step": 2826
    },
    {
      "epoch": 0.20741583132331226,
      "grad_norm": 1.8294752262712588,
      "learning_rate": 9.197016154251141e-06,
      "loss": 0.8599,
      "step": 2827
    },
    {
      "epoch": 0.2074892009134514,
      "grad_norm": 1.9280468871302798,
      "learning_rate": 9.196370239830086e-06,
      "loss": 0.9375,
      "step": 2828
    },
    {
      "epoch": 0.20756257050359053,
      "grad_norm": 0.6152614984450757,
      "learning_rate": 9.195724088429611e-06,
      "loss": 0.5352,
      "step": 2829
    },
    {
      "epoch": 0.20763594009372965,
      "grad_norm": 0.6313847531760435,
      "learning_rate": 9.195077700086209e-06,
      "loss": 0.5706,
      "step": 2830
    },
    {
      "epoch": 0.20770930968386878,
      "grad_norm": 1.733747767910607,
      "learning_rate": 9.194431074836377e-06,
      "loss": 0.9264,
      "step": 2831
    },
    {
      "epoch": 0.2077826792740079,
      "grad_norm": 0.5674200616576499,
      "learning_rate": 9.193784212716637e-06,
      "loss": 0.516,
      "step": 2832
    },
    {
      "epoch": 0.2077826792740079,
      "lm_loss": 1.0849,
      "step": 2832,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.2077826792740079,
      "lm_loss": 0.2935,
      "step": 2832,
      "vm_loss": 0.0714
    },
    {
      "epoch": 0.2077826792740079,
      "lm_loss": 0.9267,
      "step": 2832,
      "vm_loss": 0.2051
    },
    {
      "epoch": 0.2077826792740079,
      "lm_loss": 0.1068,
      "step": 2832,
      "vm_loss": 0.1593
    },
    {
      "epoch": 0.2077826792740079,
      "lm_loss": 0.5246,
      "step": 2832,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.2077826792740079,
      "lm_loss": 0.2552,
      "step": 2832,
      "vm_loss": 0.1513
    },
    {
      "epoch": 0.2077826792740079,
      "lm_loss": 0.4992,
      "step": 2832,
      "vm_loss": 0.1449
    },
    {
      "epoch": 0.2077826792740079,
      "lm_loss": 0.2547,
      "step": 2832,
      "vm_loss": 0.1228
    },
    {
      "epoch": 0.20785604886414702,
      "grad_norm": 1.6459179361127423,
      "learning_rate": 9.193137113763516e-06,
      "loss": 0.7931,
      "step": 2833
    },
    {
      "epoch": 0.20792941845428617,
      "grad_norm": 1.3939361453562609,
      "learning_rate": 9.192489778013557e-06,
      "loss": 0.728,
      "step": 2834
    },
    {
      "epoch": 0.2080027880444253,
      "grad_norm": 0.6017842455758108,
      "learning_rate": 9.191842205503321e-06,
      "loss": 0.5617,
      "step": 2835
    },
    {
      "epoch": 0.2080761576345644,
      "grad_norm": 0.6000879476736304,
      "learning_rate": 9.191194396269371e-06,
      "loss": 0.5556,
      "step": 2836
    },
    {
      "epoch": 0.20814952722470353,
      "grad_norm": 0.5808768114375521,
      "learning_rate": 9.190546350348296e-06,
      "loss": 0.5444,
      "step": 2837
    },
    {
      "epoch": 0.20822289681484266,
      "grad_norm": 0.558387238798408,
      "learning_rate": 9.189898067776691e-06,
      "loss": 0.5454,
      "step": 2838
    },
    {
      "epoch": 0.2082962664049818,
      "grad_norm": 1.7143421270959578,
      "learning_rate": 9.189249548591165e-06,
      "loss": 0.8514,
      "step": 2839
    },
    {
      "epoch": 0.20836963599512093,
      "grad_norm": 0.5932159922154372,
      "learning_rate": 9.188600792828343e-06,
      "loss": 0.5369,
      "step": 2840
    },
    {
      "epoch": 0.20836963599512093,
      "lm_loss": 1.4384,
      "step": 2840,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.20836963599512093,
      "lm_loss": 2.4236,
      "step": 2840,
      "vm_loss": 0.0774
    },
    {
      "epoch": 0.20836963599512093,
      "lm_loss": 0.4365,
      "step": 2840,
      "vm_loss": 0.1664
    },
    {
      "epoch": 0.20836963599512093,
      "lm_loss": 0.0541,
      "step": 2840,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.20836963599512093,
      "lm_loss": 0.7074,
      "step": 2840,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.20836963599512093,
      "lm_loss": 0.9401,
      "step": 2840,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.20836963599512093,
      "lm_loss": 0.2474,
      "step": 2840,
      "vm_loss": 0.1449
    },
    {
      "epoch": 0.20836963599512093,
      "lm_loss": 0.1801,
      "step": 2840,
      "vm_loss": 0.1743
    },
    {
      "epoch": 0.20844300558526005,
      "grad_norm": 1.6297747425973168,
      "learning_rate": 9.187951800524862e-06,
      "loss": 0.9099,
      "step": 2841
    },
    {
      "epoch": 0.20851637517539917,
      "grad_norm": 1.4814002918571783,
      "learning_rate": 9.187302571717371e-06,
      "loss": 0.9312,
      "step": 2842
    },
    {
      "epoch": 0.2085897447655383,
      "grad_norm": 0.6087883355844168,
      "learning_rate": 9.186653106442534e-06,
      "loss": 0.5389,
      "step": 2843
    },
    {
      "epoch": 0.20866311435567744,
      "grad_norm": 1.679648035626757,
      "learning_rate": 9.186003404737028e-06,
      "loss": 0.8105,
      "step": 2844
    },
    {
      "epoch": 0.20873648394581656,
      "grad_norm": 1.5502037266157684,
      "learning_rate": 9.185353466637544e-06,
      "loss": 0.8279,
      "step": 2845
    },
    {
      "epoch": 0.2088098535359557,
      "grad_norm": 1.5819263903600886,
      "learning_rate": 9.184703292180783e-06,
      "loss": 0.7814,
      "step": 2846
    },
    {
      "epoch": 0.2088832231260948,
      "grad_norm": 1.7244205239254855,
      "learning_rate": 9.184052881403465e-06,
      "loss": 0.8306,
      "step": 2847
    },
    {
      "epoch": 0.20895659271623393,
      "grad_norm": 1.627406725723878,
      "learning_rate": 9.18340223434232e-06,
      "loss": 0.9163,
      "step": 2848
    },
    {
      "epoch": 0.20895659271623393,
      "lm_loss": 2.6128,
      "step": 2848,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.20895659271623393,
      "lm_loss": 1.0168,
      "step": 2848,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.20895659271623393,
      "lm_loss": 1.2222,
      "step": 2848,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.20895659271623393,
      "lm_loss": 0.6008,
      "step": 2848,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.20895659271623393,
      "lm_loss": 0.469,
      "step": 2848,
      "vm_loss": 0.2977
    },
    {
      "epoch": 0.20895659271623393,
      "lm_loss": 0.4297,
      "step": 2848,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.20895659271623393,
      "lm_loss": 0.6126,
      "step": 2848,
      "vm_loss": 0.1513
    },
    {
      "epoch": 0.20895659271623393,
      "lm_loss": 0.0842,
      "step": 2848,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.20902996230637308,
      "grad_norm": 1.7610813989830298,
      "learning_rate": 9.18275135103409e-06,
      "loss": 0.9476,
      "step": 2849
    },
    {
      "epoch": 0.2091033318965122,
      "grad_norm": 0.7268906211039292,
      "learning_rate": 9.182100231515532e-06,
      "loss": 0.5317,
      "step": 2850
    },
    {
      "epoch": 0.20917670148665132,
      "grad_norm": 1.658875677600122,
      "learning_rate": 9.181448875823417e-06,
      "loss": 0.8536,
      "step": 2851
    },
    {
      "epoch": 0.20925007107679044,
      "grad_norm": 1.5637411889052832,
      "learning_rate": 9.180797283994532e-06,
      "loss": 0.8609,
      "step": 2852
    },
    {
      "epoch": 0.20932344066692957,
      "grad_norm": 1.4471226773756702,
      "learning_rate": 9.180145456065668e-06,
      "loss": 0.7501,
      "step": 2853
    },
    {
      "epoch": 0.2093968102570687,
      "grad_norm": 1.7847504223196307,
      "learning_rate": 9.179493392073637e-06,
      "loss": 0.7977,
      "step": 2854
    },
    {
      "epoch": 0.20947017984720784,
      "grad_norm": 1.555495972167344,
      "learning_rate": 9.178841092055266e-06,
      "loss": 0.9134,
      "step": 2855
    },
    {
      "epoch": 0.20954354943734696,
      "grad_norm": 1.5902696489055517,
      "learning_rate": 9.178188556047389e-06,
      "loss": 0.6971,
      "step": 2856
    },
    {
      "epoch": 0.20954354943734696,
      "lm_loss": 1.373,
      "step": 2856,
      "vm_loss": 0.0354
    },
    {
      "epoch": 0.20954354943734696,
      "lm_loss": 2.3303,
      "step": 2856,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.20954354943734696,
      "lm_loss": 0.9615,
      "step": 2856,
      "vm_loss": 0.2294
    },
    {
      "epoch": 0.20954354943734696,
      "lm_loss": 0.9137,
      "step": 2856,
      "vm_loss": 0.1126
    },
    {
      "epoch": 0.20954354943734696,
      "lm_loss": 0.885,
      "step": 2856,
      "vm_loss": 0.1487
    },
    {
      "epoch": 0.20954354943734696,
      "lm_loss": 0.1113,
      "step": 2856,
      "vm_loss": 0.1422
    },
    {
      "epoch": 0.20954354943734696,
      "lm_loss": 0.9154,
      "step": 2856,
      "vm_loss": 0.1517
    },
    {
      "epoch": 0.20954354943734696,
      "lm_loss": 0.7414,
      "step": 2856,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.20961691902748608,
      "grad_norm": 1.5476892724039004,
      "learning_rate": 9.177535784086856e-06,
      "loss": 0.9135,
      "step": 2857
    },
    {
      "epoch": 0.2096902886176252,
      "grad_norm": 1.5912933865796361,
      "learning_rate": 9.176882776210535e-06,
      "loss": 0.8382,
      "step": 2858
    },
    {
      "epoch": 0.20976365820776433,
      "grad_norm": 1.5524687731667068,
      "learning_rate": 9.176229532455298e-06,
      "loss": 0.8559,
      "step": 2859
    },
    {
      "epoch": 0.20983702779790347,
      "grad_norm": 0.65050654814605,
      "learning_rate": 9.175576052858037e-06,
      "loss": 0.5534,
      "step": 2860
    },
    {
      "epoch": 0.2099103973880426,
      "grad_norm": 1.4752825651467978,
      "learning_rate": 9.174922337455656e-06,
      "loss": 0.8514,
      "step": 2861
    },
    {
      "epoch": 0.20998376697818172,
      "grad_norm": 0.5875984506641003,
      "learning_rate": 9.17426838628507e-06,
      "loss": 0.5268,
      "step": 2862
    },
    {
      "epoch": 0.21005713656832084,
      "grad_norm": 1.630409393483599,
      "learning_rate": 9.173614199383212e-06,
      "loss": 0.8416,
      "step": 2863
    },
    {
      "epoch": 0.21013050615845996,
      "grad_norm": 1.6412585270189588,
      "learning_rate": 9.172959776787026e-06,
      "loss": 0.8894,
      "step": 2864
    },
    {
      "epoch": 0.21013050615845996,
      "lm_loss": 1.2659,
      "step": 2864,
      "vm_loss": 0.082
    },
    {
      "epoch": 0.21013050615845996,
      "lm_loss": 0.8723,
      "step": 2864,
      "vm_loss": 0.1068
    },
    {
      "epoch": 0.21013050615845996,
      "lm_loss": 0.126,
      "step": 2864,
      "vm_loss": 0.1778
    },
    {
      "epoch": 0.21013050615845996,
      "lm_loss": 0.3757,
      "step": 2864,
      "vm_loss": 0.1394
    },
    {
      "epoch": 0.21013050615845996,
      "lm_loss": 0.0405,
      "step": 2864,
      "vm_loss": 0.1011
    },
    {
      "epoch": 0.21013050615845996,
      "lm_loss": 0.5139,
      "step": 2864,
      "vm_loss": 0.1136
    },
    {
      "epoch": 0.21013050615845996,
      "lm_loss": 1.2255,
      "step": 2864,
      "vm_loss": 0.198
    },
    {
      "epoch": 0.21013050615845996,
      "lm_loss": 0.9001,
      "step": 2864,
      "vm_loss": 0.1461
    },
    {
      "epoch": 0.2102038757485991,
      "grad_norm": 1.6996186736378045,
      "learning_rate": 9.172305118533464e-06,
      "loss": 0.8807,
      "step": 2865
    },
    {
      "epoch": 0.21027724533873823,
      "grad_norm": 1.6951713278376914,
      "learning_rate": 9.171650224659501e-06,
      "loss": 0.8518,
      "step": 2866
    },
    {
      "epoch": 0.21035061492887736,
      "grad_norm": 1.747477713184239,
      "learning_rate": 9.17099509520212e-06,
      "loss": 0.9437,
      "step": 2867
    },
    {
      "epoch": 0.21042398451901648,
      "grad_norm": 1.5671056712998113,
      "learning_rate": 9.170339730198316e-06,
      "loss": 0.9099,
      "step": 2868
    },
    {
      "epoch": 0.2104973541091556,
      "grad_norm": 1.5925500105794153,
      "learning_rate": 9.169684129685099e-06,
      "loss": 0.8493,
      "step": 2869
    },
    {
      "epoch": 0.21057072369929475,
      "grad_norm": 1.5803133884940697,
      "learning_rate": 9.169028293699494e-06,
      "loss": 0.7749,
      "step": 2870
    },
    {
      "epoch": 0.21064409328943387,
      "grad_norm": 1.3874373436350482,
      "learning_rate": 9.168372222278537e-06,
      "loss": 0.8011,
      "step": 2871
    },
    {
      "epoch": 0.210717462879573,
      "grad_norm": 0.6898823661110424,
      "learning_rate": 9.167715915459277e-06,
      "loss": 0.5415,
      "step": 2872
    },
    {
      "epoch": 0.210717462879573,
      "lm_loss": 0.576,
      "step": 2872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.210717462879573,
      "lm_loss": 0.2012,
      "step": 2872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.210717462879573,
      "lm_loss": 0.5936,
      "step": 2872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.210717462879573,
      "lm_loss": 0.4518,
      "step": 2872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.210717462879573,
      "lm_loss": 0.4818,
      "step": 2872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.210717462879573,
      "lm_loss": 0.4782,
      "step": 2872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.210717462879573,
      "lm_loss": 0.563,
      "step": 2872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.210717462879573,
      "lm_loss": 0.6549,
      "step": 2872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2107908324697121,
      "grad_norm": 0.6522663894286319,
      "learning_rate": 9.167059373278777e-06,
      "loss": 0.5407,
      "step": 2873
    },
    {
      "epoch": 0.21086420205985124,
      "grad_norm": 1.5607871672875606,
      "learning_rate": 9.166402595774116e-06,
      "loss": 0.8836,
      "step": 2874
    },
    {
      "epoch": 0.21093757164999036,
      "grad_norm": 0.5541639531871503,
      "learning_rate": 9.165745582982383e-06,
      "loss": 0.5128,
      "step": 2875
    },
    {
      "epoch": 0.2110109412401295,
      "grad_norm": 1.4524264198754437,
      "learning_rate": 9.16508833494068e-06,
      "loss": 0.7333,
      "step": 2876
    },
    {
      "epoch": 0.21108431083026863,
      "grad_norm": 1.687830722383401,
      "learning_rate": 9.164430851686123e-06,
      "loss": 0.879,
      "step": 2877
    },
    {
      "epoch": 0.21115768042040775,
      "grad_norm": 1.7491904094321846,
      "learning_rate": 9.163773133255845e-06,
      "loss": 0.8632,
      "step": 2878
    },
    {
      "epoch": 0.21123105001054687,
      "grad_norm": 1.6876561498306746,
      "learning_rate": 9.163115179686986e-06,
      "loss": 0.9466,
      "step": 2879
    },
    {
      "epoch": 0.211304419600686,
      "grad_norm": 2.0139286869249133,
      "learning_rate": 9.162456991016702e-06,
      "loss": 0.9432,
      "step": 2880
    },
    {
      "epoch": 0.211304419600686,
      "lm_loss": 1.9744,
      "step": 2880,
      "vm_loss": 0.1504
    },
    {
      "epoch": 0.211304419600686,
      "lm_loss": 0.2437,
      "step": 2880,
      "vm_loss": 0.1083
    },
    {
      "epoch": 0.211304419600686,
      "lm_loss": 0.2446,
      "step": 2880,
      "vm_loss": 0.124
    },
    {
      "epoch": 0.211304419600686,
      "lm_loss": 0.1089,
      "step": 2880,
      "vm_loss": 0.1496
    },
    {
      "epoch": 0.211304419600686,
      "lm_loss": 0.2409,
      "step": 2880,
      "vm_loss": 0.1793
    },
    {
      "epoch": 0.211304419600686,
      "lm_loss": 0.5738,
      "step": 2880,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.211304419600686,
      "lm_loss": 0.9229,
      "step": 2880,
      "vm_loss": 0.0857
    },
    {
      "epoch": 0.211304419600686,
      "lm_loss": 0.5158,
      "step": 2880,
      "vm_loss": 0.2269
    },
    {
      "epoch": 0.21137778919082514,
      "grad_norm": 1.4614524683266308,
      "learning_rate": 9.161798567282163e-06,
      "loss": 0.8588,
      "step": 2881
    },
    {
      "epoch": 0.21145115878096427,
      "grad_norm": 0.9263025628301031,
      "learning_rate": 9.161139908520556e-06,
      "loss": 0.5445,
      "step": 2882
    },
    {
      "epoch": 0.2115245283711034,
      "grad_norm": 0.7510743373227373,
      "learning_rate": 9.160481014769071e-06,
      "loss": 0.5524,
      "step": 2883
    },
    {
      "epoch": 0.2115978979612425,
      "grad_norm": 1.8431575691841484,
      "learning_rate": 9.159821886064919e-06,
      "loss": 0.8956,
      "step": 2884
    },
    {
      "epoch": 0.21167126755138163,
      "grad_norm": 1.6216030882076866,
      "learning_rate": 9.159162522445324e-06,
      "loss": 0.8843,
      "step": 2885
    },
    {
      "epoch": 0.21174463714152078,
      "grad_norm": 1.4860686406156889,
      "learning_rate": 9.158502923947523e-06,
      "loss": 0.8859,
      "step": 2886
    },
    {
      "epoch": 0.2118180067316599,
      "grad_norm": 1.5191968697553748,
      "learning_rate": 9.157843090608763e-06,
      "loss": 0.856,
      "step": 2887
    },
    {
      "epoch": 0.21189137632179902,
      "grad_norm": 1.6344159382537478,
      "learning_rate": 9.157183022466305e-06,
      "loss": 0.8039,
      "step": 2888
    },
    {
      "epoch": 0.21189137632179902,
      "lm_loss": 0.5026,
      "step": 2888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21189137632179902,
      "lm_loss": 0.5276,
      "step": 2888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21189137632179902,
      "lm_loss": 0.6171,
      "step": 2888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21189137632179902,
      "lm_loss": 0.6076,
      "step": 2888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21189137632179902,
      "lm_loss": 0.5239,
      "step": 2888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21189137632179902,
      "lm_loss": 0.5074,
      "step": 2888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21189137632179902,
      "lm_loss": 0.5471,
      "step": 2888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21189137632179902,
      "lm_loss": 0.6328,
      "step": 2888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21196474591193815,
      "grad_norm": 1.2926170543918811,
      "learning_rate": 9.156522719557428e-06,
      "loss": 0.5682,
      "step": 2889
    },
    {
      "epoch": 0.21203811550207727,
      "grad_norm": 1.6554276827591552,
      "learning_rate": 9.15586218191942e-06,
      "loss": 0.8639,
      "step": 2890
    },
    {
      "epoch": 0.21211148509221642,
      "grad_norm": 1.4941701652145538,
      "learning_rate": 9.155201409589581e-06,
      "loss": 0.8202,
      "step": 2891
    },
    {
      "epoch": 0.21218485468235554,
      "grad_norm": 1.5137224723438683,
      "learning_rate": 9.15454040260523e-06,
      "loss": 0.7994,
      "step": 2892
    },
    {
      "epoch": 0.21225822427249466,
      "grad_norm": 1.6014306684848323,
      "learning_rate": 9.153879161003692e-06,
      "loss": 0.8724,
      "step": 2893
    },
    {
      "epoch": 0.21233159386263378,
      "grad_norm": 1.458827005057177,
      "learning_rate": 9.153217684822313e-06,
      "loss": 0.9408,
      "step": 2894
    },
    {
      "epoch": 0.2124049634527729,
      "grad_norm": 1.4205334496146096,
      "learning_rate": 9.152555974098445e-06,
      "loss": 0.9443,
      "step": 2895
    },
    {
      "epoch": 0.21247833304291203,
      "grad_norm": 1.5575412116580618,
      "learning_rate": 9.151894028869458e-06,
      "loss": 0.8266,
      "step": 2896
    },
    {
      "epoch": 0.21247833304291203,
      "lm_loss": 0.5587,
      "step": 2896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21247833304291203,
      "lm_loss": 0.3859,
      "step": 2896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21247833304291203,
      "lm_loss": 0.4849,
      "step": 2896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21247833304291203,
      "lm_loss": 0.5499,
      "step": 2896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21247833304291203,
      "lm_loss": 0.5184,
      "step": 2896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21247833304291203,
      "lm_loss": 0.7132,
      "step": 2896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21247833304291203,
      "lm_loss": 0.6048,
      "step": 2896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21247833304291203,
      "lm_loss": 0.5623,
      "step": 2896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21255170263305118,
      "grad_norm": 0.8592270550646692,
      "learning_rate": 9.151231849172733e-06,
      "loss": 0.5592,
      "step": 2897
    },
    {
      "epoch": 0.2126250722231903,
      "grad_norm": 0.701339387115346,
      "learning_rate": 9.150569435045665e-06,
      "loss": 0.5574,
      "step": 2898
    },
    {
      "epoch": 0.21269844181332942,
      "grad_norm": 1.642759898450425,
      "learning_rate": 9.149906786525662e-06,
      "loss": 0.845,
      "step": 2899
    },
    {
      "epoch": 0.21277181140346854,
      "grad_norm": 1.7500593083724099,
      "learning_rate": 9.149243903650147e-06,
      "loss": 0.8713,
      "step": 2900
    },
    {
      "epoch": 0.21284518099360766,
      "grad_norm": 1.6030282080244513,
      "learning_rate": 9.148580786456552e-06,
      "loss": 0.8119,
      "step": 2901
    },
    {
      "epoch": 0.2129185505837468,
      "grad_norm": 1.7247569908560283,
      "learning_rate": 9.147917434982326e-06,
      "loss": 0.9141,
      "step": 2902
    },
    {
      "epoch": 0.21299192017388593,
      "grad_norm": 0.9221073276203361,
      "learning_rate": 9.147253849264933e-06,
      "loss": 0.5302,
      "step": 2903
    },
    {
      "epoch": 0.21306528976402506,
      "grad_norm": 1.6812474499392762,
      "learning_rate": 9.146590029341842e-06,
      "loss": 0.9402,
      "step": 2904
    },
    {
      "epoch": 0.21306528976402506,
      "lm_loss": 0.5395,
      "step": 2904,
      "vm_loss": 0.0584
    },
    {
      "epoch": 0.21306528976402506,
      "lm_loss": 1.251,
      "step": 2904,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.21306528976402506,
      "lm_loss": 0.1435,
      "step": 2904,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.21306528976402506,
      "lm_loss": 0.3462,
      "step": 2904,
      "vm_loss": 0.2138
    },
    {
      "epoch": 0.21306528976402506,
      "lm_loss": 0.0811,
      "step": 2904,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.21306528976402506,
      "lm_loss": 1.1524,
      "step": 2904,
      "vm_loss": 0.1085
    },
    {
      "epoch": 0.21306528976402506,
      "lm_loss": 0.535,
      "step": 2904,
      "vm_loss": 0.1018
    },
    {
      "epoch": 0.21306528976402506,
      "lm_loss": 0.5864,
      "step": 2904,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.21313865935416418,
      "grad_norm": 1.6562110008280027,
      "learning_rate": 9.145925975250542e-06,
      "loss": 0.9289,
      "step": 2905
    },
    {
      "epoch": 0.2132120289443033,
      "grad_norm": 0.6650705597828338,
      "learning_rate": 9.145261687028537e-06,
      "loss": 0.5531,
      "step": 2906
    },
    {
      "epoch": 0.21328539853444245,
      "grad_norm": 1.854757064734743,
      "learning_rate": 9.14459716471334e-06,
      "loss": 0.8427,
      "step": 2907
    },
    {
      "epoch": 0.21335876812458157,
      "grad_norm": 1.4942307448087997,
      "learning_rate": 9.143932408342475e-06,
      "loss": 0.7961,
      "step": 2908
    },
    {
      "epoch": 0.2134321377147207,
      "grad_norm": 1.6419858018747853,
      "learning_rate": 9.143267417953486e-06,
      "loss": 0.8319,
      "step": 2909
    },
    {
      "epoch": 0.21350550730485982,
      "grad_norm": 0.6520353758937235,
      "learning_rate": 9.142602193583925e-06,
      "loss": 0.5276,
      "step": 2910
    },
    {
      "epoch": 0.21357887689499894,
      "grad_norm": 0.628287192315831,
      "learning_rate": 9.141936735271358e-06,
      "loss": 0.5474,
      "step": 2911
    },
    {
      "epoch": 0.21365224648513806,
      "grad_norm": 1.5771438385186434,
      "learning_rate": 9.141271043053367e-06,
      "loss": 0.8417,
      "step": 2912
    },
    {
      "epoch": 0.21365224648513806,
      "lm_loss": 0.1137,
      "step": 2912,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.21365224648513806,
      "lm_loss": 1.1493,
      "step": 2912,
      "vm_loss": 0.1422
    },
    {
      "epoch": 0.21365224648513806,
      "lm_loss": 0.485,
      "step": 2912,
      "vm_loss": 0.162
    },
    {
      "epoch": 0.21365224648513806,
      "lm_loss": 1.2071,
      "step": 2912,
      "vm_loss": 0.1081
    },
    {
      "epoch": 0.21365224648513806,
      "lm_loss": 1.0334,
      "step": 2912,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.21365224648513806,
      "lm_loss": 0.7943,
      "step": 2912,
      "vm_loss": 0.102
    },
    {
      "epoch": 0.21365224648513806,
      "lm_loss": 0.3759,
      "step": 2912,
      "vm_loss": 0.1632
    },
    {
      "epoch": 0.21365224648513806,
      "lm_loss": 1.1541,
      "step": 2912,
      "vm_loss": 0.3305
    },
    {
      "epoch": 0.2137256160752772,
      "grad_norm": 1.7025959200614853,
      "learning_rate": 9.140605116967544e-06,
      "loss": 0.9023,
      "step": 2913
    },
    {
      "epoch": 0.21379898566541633,
      "grad_norm": 1.674206936497695,
      "learning_rate": 9.139938957051498e-06,
      "loss": 0.8454,
      "step": 2914
    },
    {
      "epoch": 0.21387235525555545,
      "grad_norm": 1.6050830406839907,
      "learning_rate": 9.139272563342845e-06,
      "loss": 0.8482,
      "step": 2915
    },
    {
      "epoch": 0.21394572484569457,
      "grad_norm": 0.6283162604283119,
      "learning_rate": 9.13860593587922e-06,
      "loss": 0.5507,
      "step": 2916
    },
    {
      "epoch": 0.2140190944358337,
      "grad_norm": 1.511296132088093,
      "learning_rate": 9.137939074698268e-06,
      "loss": 0.8507,
      "step": 2917
    },
    {
      "epoch": 0.21409246402597285,
      "grad_norm": 1.431729384725092,
      "learning_rate": 9.137271979837651e-06,
      "loss": 0.9046,
      "step": 2918
    },
    {
      "epoch": 0.21416583361611197,
      "grad_norm": 0.6588463742844215,
      "learning_rate": 9.136604651335039e-06,
      "loss": 0.5462,
      "step": 2919
    },
    {
      "epoch": 0.2142392032062511,
      "grad_norm": 0.611634950611222,
      "learning_rate": 9.135937089228116e-06,
      "loss": 0.5464,
      "step": 2920
    },
    {
      "epoch": 0.2142392032062511,
      "lm_loss": 0.6019,
      "step": 2920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2142392032062511,
      "lm_loss": 0.7477,
      "step": 2920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2142392032062511,
      "lm_loss": 0.5586,
      "step": 2920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2142392032062511,
      "lm_loss": 0.5484,
      "step": 2920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2142392032062511,
      "lm_loss": 0.7012,
      "step": 2920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2142392032062511,
      "lm_loss": 0.5613,
      "step": 2920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2142392032062511,
      "lm_loss": 0.5488,
      "step": 2920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2142392032062511,
      "lm_loss": 0.4113,
      "step": 2920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2143125727963902,
      "grad_norm": 0.5843545892095685,
      "learning_rate": 9.135269293554585e-06,
      "loss": 0.5328,
      "step": 2921
    },
    {
      "epoch": 0.21438594238652933,
      "grad_norm": 1.934392540127996,
      "learning_rate": 9.134601264352156e-06,
      "loss": 0.9064,
      "step": 2922
    },
    {
      "epoch": 0.21445931197666848,
      "grad_norm": 0.6157500364149585,
      "learning_rate": 9.133933001658554e-06,
      "loss": 0.5359,
      "step": 2923
    },
    {
      "epoch": 0.2145326815668076,
      "grad_norm": 1.4603680515647406,
      "learning_rate": 9.133264505511517e-06,
      "loss": 0.8041,
      "step": 2924
    },
    {
      "epoch": 0.21460605115694673,
      "grad_norm": 1.6372565312225111,
      "learning_rate": 9.1325957759488e-06,
      "loss": 0.8953,
      "step": 2925
    },
    {
      "epoch": 0.21467942074708585,
      "grad_norm": 1.680323705415168,
      "learning_rate": 9.131926813008163e-06,
      "loss": 0.8918,
      "step": 2926
    },
    {
      "epoch": 0.21475279033722497,
      "grad_norm": 0.7463984129239646,
      "learning_rate": 9.131257616727387e-06,
      "loss": 0.5744,
      "step": 2927
    },
    {
      "epoch": 0.21482615992736412,
      "grad_norm": 1.5429857701713858,
      "learning_rate": 9.130588187144263e-06,
      "loss": 0.7875,
      "step": 2928
    },
    {
      "epoch": 0.21482615992736412,
      "lm_loss": 0.9955,
      "step": 2928,
      "vm_loss": 0.1272
    },
    {
      "epoch": 0.21482615992736412,
      "lm_loss": 0.3445,
      "step": 2928,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.21482615992736412,
      "lm_loss": 0.2143,
      "step": 2928,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.21482615992736412,
      "lm_loss": 0.6108,
      "step": 2928,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.21482615992736412,
      "lm_loss": 0.8303,
      "step": 2928,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.21482615992736412,
      "lm_loss": 1.0171,
      "step": 2928,
      "vm_loss": 0.135
    },
    {
      "epoch": 0.21482615992736412,
      "lm_loss": 0.6682,
      "step": 2928,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.21482615992736412,
      "lm_loss": 1.2097,
      "step": 2928,
      "vm_loss": 0.2134
    },
    {
      "epoch": 0.21489952951750324,
      "grad_norm": 1.642667053712958,
      "learning_rate": 9.129918524296596e-06,
      "loss": 0.9138,
      "step": 2929
    },
    {
      "epoch": 0.21497289910764236,
      "grad_norm": 1.77786197812563,
      "learning_rate": 9.1292486282222e-06,
      "loss": 0.938,
      "step": 2930
    },
    {
      "epoch": 0.21504626869778148,
      "grad_norm": 1.5483590463272243,
      "learning_rate": 9.128578498958909e-06,
      "loss": 0.788,
      "step": 2931
    },
    {
      "epoch": 0.2151196382879206,
      "grad_norm": 1.5147338713329577,
      "learning_rate": 9.127908136544565e-06,
      "loss": 0.8605,
      "step": 2932
    },
    {
      "epoch": 0.21519300787805973,
      "grad_norm": 0.629603435577674,
      "learning_rate": 9.127237541017027e-06,
      "loss": 0.5405,
      "step": 2933
    },
    {
      "epoch": 0.21526637746819888,
      "grad_norm": 1.557616246521555,
      "learning_rate": 9.126566712414164e-06,
      "loss": 0.9776,
      "step": 2934
    },
    {
      "epoch": 0.215339747058338,
      "grad_norm": 0.5908552200880913,
      "learning_rate": 9.12589565077386e-06,
      "loss": 0.5569,
      "step": 2935
    },
    {
      "epoch": 0.21541311664847712,
      "grad_norm": 1.5814821154372267,
      "learning_rate": 9.125224356134011e-06,
      "loss": 0.8758,
      "step": 2936
    },
    {
      "epoch": 0.21541311664847712,
      "lm_loss": 1.4398,
      "step": 2936,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.21541311664847712,
      "lm_loss": 0.1939,
      "step": 2936,
      "vm_loss": 0.1893
    },
    {
      "epoch": 0.21541311664847712,
      "lm_loss": 0.5001,
      "step": 2936,
      "vm_loss": 0.0685
    },
    {
      "epoch": 0.21541311664847712,
      "lm_loss": 0.0672,
      "step": 2936,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.21541311664847712,
      "lm_loss": 0.954,
      "step": 2936,
      "vm_loss": 0.2621
    },
    {
      "epoch": 0.21541311664847712,
      "lm_loss": 0.7963,
      "step": 2936,
      "vm_loss": 0.1536
    },
    {
      "epoch": 0.21541311664847712,
      "lm_loss": 0.9736,
      "step": 2936,
      "vm_loss": 0.1345
    },
    {
      "epoch": 0.21541311664847712,
      "lm_loss": 0.7556,
      "step": 2936,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.21548648623861624,
      "grad_norm": 1.5624975884997543,
      "learning_rate": 9.124552828532524e-06,
      "loss": 0.9158,
      "step": 2937
    },
    {
      "epoch": 0.21555985582875536,
      "grad_norm": 0.5849153084720199,
      "learning_rate": 9.123881068007327e-06,
      "loss": 0.5502,
      "step": 2938
    },
    {
      "epoch": 0.21563322541889451,
      "grad_norm": 1.6151043855806828,
      "learning_rate": 9.123209074596353e-06,
      "loss": 0.8518,
      "step": 2939
    },
    {
      "epoch": 0.21570659500903364,
      "grad_norm": 1.5564150674792634,
      "learning_rate": 9.122536848337552e-06,
      "loss": 0.915,
      "step": 2940
    },
    {
      "epoch": 0.21577996459917276,
      "grad_norm": 0.624874817003811,
      "learning_rate": 9.121864389268887e-06,
      "loss": 0.5623,
      "step": 2941
    },
    {
      "epoch": 0.21585333418931188,
      "grad_norm": 1.5028766487669023,
      "learning_rate": 9.12119169742833e-06,
      "loss": 0.7981,
      "step": 2942
    },
    {
      "epoch": 0.215926703779451,
      "grad_norm": 1.4541586492691252,
      "learning_rate": 9.120518772853872e-06,
      "loss": 0.8246,
      "step": 2943
    },
    {
      "epoch": 0.21600007336959015,
      "grad_norm": 1.6962889416706672,
      "learning_rate": 9.119845615583515e-06,
      "loss": 0.7939,
      "step": 2944
    },
    {
      "epoch": 0.21600007336959015,
      "lm_loss": 0.643,
      "step": 2944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21600007336959015,
      "lm_loss": 0.5561,
      "step": 2944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21600007336959015,
      "lm_loss": 0.5205,
      "step": 2944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21600007336959015,
      "lm_loss": 0.4577,
      "step": 2944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21600007336959015,
      "lm_loss": 0.5787,
      "step": 2944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21600007336959015,
      "lm_loss": 0.4608,
      "step": 2944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21600007336959015,
      "lm_loss": 0.4947,
      "step": 2944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21600007336959015,
      "lm_loss": 0.5738,
      "step": 2944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21607344295972927,
      "grad_norm": 0.6097830515153574,
      "learning_rate": 9.119172225655275e-06,
      "loss": 0.5478,
      "step": 2945
    },
    {
      "epoch": 0.2161468125498684,
      "grad_norm": 1.5524088984035034,
      "learning_rate": 9.118498603107176e-06,
      "loss": 0.9424,
      "step": 2946
    },
    {
      "epoch": 0.21622018214000752,
      "grad_norm": 1.5692864110395373,
      "learning_rate": 9.117824747977264e-06,
      "loss": 0.8703,
      "step": 2947
    },
    {
      "epoch": 0.21629355173014664,
      "grad_norm": 0.5718003680670728,
      "learning_rate": 9.117150660303588e-06,
      "loss": 0.5596,
      "step": 2948
    },
    {
      "epoch": 0.2163669213202858,
      "grad_norm": 1.7103929237377764,
      "learning_rate": 9.11647634012422e-06,
      "loss": 0.829,
      "step": 2949
    },
    {
      "epoch": 0.2164402909104249,
      "grad_norm": 1.5888854331331672,
      "learning_rate": 9.11580178747724e-06,
      "loss": 0.9231,
      "step": 2950
    },
    {
      "epoch": 0.21651366050056403,
      "grad_norm": 0.5618837030687959,
      "learning_rate": 9.11512700240074e-06,
      "loss": 0.5424,
      "step": 2951
    },
    {
      "epoch": 0.21658703009070315,
      "grad_norm": 1.5323813948654061,
      "learning_rate": 9.114451984932826e-06,
      "loss": 0.9163,
      "step": 2952
    },
    {
      "epoch": 0.21658703009070315,
      "lm_loss": 0.6468,
      "step": 2952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21658703009070315,
      "lm_loss": 0.4418,
      "step": 2952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21658703009070315,
      "lm_loss": 0.5576,
      "step": 2952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21658703009070315,
      "lm_loss": 0.5192,
      "step": 2952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21658703009070315,
      "lm_loss": 0.5105,
      "step": 2952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21658703009070315,
      "lm_loss": 0.7041,
      "step": 2952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21658703009070315,
      "lm_loss": 0.5998,
      "step": 2952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21658703009070315,
      "lm_loss": 0.5261,
      "step": 2952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21666039968084227,
      "grad_norm": 0.5822117725174207,
      "learning_rate": 9.11377673511162e-06,
      "loss": 0.5463,
      "step": 2953
    },
    {
      "epoch": 0.2167337692709814,
      "grad_norm": 1.4523080729625337,
      "learning_rate": 9.113101252975253e-06,
      "loss": 0.9192,
      "step": 2954
    },
    {
      "epoch": 0.21680713886112055,
      "grad_norm": 0.5431756934912384,
      "learning_rate": 9.112425538561874e-06,
      "loss": 0.534,
      "step": 2955
    },
    {
      "epoch": 0.21688050845125967,
      "grad_norm": 1.5269458575702481,
      "learning_rate": 9.11174959190964e-06,
      "loss": 0.9095,
      "step": 2956
    },
    {
      "epoch": 0.2169538780413988,
      "grad_norm": 0.5770278084117288,
      "learning_rate": 9.111073413056722e-06,
      "loss": 0.5376,
      "step": 2957
    },
    {
      "epoch": 0.2170272476315379,
      "grad_norm": 0.5692831518777934,
      "learning_rate": 9.110397002041309e-06,
      "loss": 0.5284,
      "step": 2958
    },
    {
      "epoch": 0.21710061722167703,
      "grad_norm": 1.7678207387818163,
      "learning_rate": 9.109720358901599e-06,
      "loss": 0.8416,
      "step": 2959
    },
    {
      "epoch": 0.21717398681181618,
      "grad_norm": 1.7350265732466288,
      "learning_rate": 9.1090434836758e-06,
      "loss": 0.9097,
      "step": 2960
    },
    {
      "epoch": 0.21717398681181618,
      "lm_loss": 0.5239,
      "step": 2960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21717398681181618,
      "lm_loss": 0.5129,
      "step": 2960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21717398681181618,
      "lm_loss": 0.422,
      "step": 2960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21717398681181618,
      "lm_loss": 0.6025,
      "step": 2960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21717398681181618,
      "lm_loss": 0.5827,
      "step": 2960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21717398681181618,
      "lm_loss": 0.3788,
      "step": 2960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21717398681181618,
      "lm_loss": 0.4591,
      "step": 2960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21717398681181618,
      "lm_loss": 0.5059,
      "step": 2960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2172473564019553,
      "grad_norm": 0.5908113829043823,
      "learning_rate": 9.108366376402142e-06,
      "loss": 0.5312,
      "step": 2961
    },
    {
      "epoch": 0.21732072599209443,
      "grad_norm": 1.442366928197949,
      "learning_rate": 9.107689037118862e-06,
      "loss": 0.7261,
      "step": 2962
    },
    {
      "epoch": 0.21739409558223355,
      "grad_norm": 1.7306570331844136,
      "learning_rate": 9.107011465864205e-06,
      "loss": 0.8716,
      "step": 2963
    },
    {
      "epoch": 0.21746746517237267,
      "grad_norm": 0.5837362577827924,
      "learning_rate": 9.106333662676445e-06,
      "loss": 0.5321,
      "step": 2964
    },
    {
      "epoch": 0.21754083476251182,
      "grad_norm": 1.7851850272740362,
      "learning_rate": 9.10565562759385e-06,
      "loss": 0.8999,
      "step": 2965
    },
    {
      "epoch": 0.21761420435265094,
      "grad_norm": 1.4944224385978258,
      "learning_rate": 9.104977360654717e-06,
      "loss": 0.859,
      "step": 2966
    },
    {
      "epoch": 0.21768757394279006,
      "grad_norm": 0.5851864781575459,
      "learning_rate": 9.104298861897344e-06,
      "loss": 0.5409,
      "step": 2967
    },
    {
      "epoch": 0.21776094353292919,
      "grad_norm": 1.3590261566504318,
      "learning_rate": 9.103620131360052e-06,
      "loss": 0.7966,
      "step": 2968
    },
    {
      "epoch": 0.21776094353292919,
      "lm_loss": 1.0893,
      "step": 2968,
      "vm_loss": 0.0586
    },
    {
      "epoch": 0.21776094353292919,
      "lm_loss": 1.1108,
      "step": 2968,
      "vm_loss": 0.1223
    },
    {
      "epoch": 0.21776094353292919,
      "lm_loss": 0.9106,
      "step": 2968,
      "vm_loss": 0.1263
    },
    {
      "epoch": 0.21776094353292919,
      "lm_loss": 0.1418,
      "step": 2968,
      "vm_loss": 0.183
    },
    {
      "epoch": 0.21776094353292919,
      "lm_loss": 1.1026,
      "step": 2968,
      "vm_loss": 0.178
    },
    {
      "epoch": 0.21776094353292919,
      "lm_loss": 1.9421,
      "step": 2968,
      "vm_loss": 0.1579
    },
    {
      "epoch": 0.21776094353292919,
      "lm_loss": 0.3714,
      "step": 2968,
      "vm_loss": 0.0613
    },
    {
      "epoch": 0.21776094353292919,
      "lm_loss": 0.7986,
      "step": 2968,
      "vm_loss": 0.0197
    },
    {
      "epoch": 0.2178343131230683,
      "grad_norm": 1.3782279299827331,
      "learning_rate": 9.102941169081167e-06,
      "loss": 0.795,
      "step": 2969
    },
    {
      "epoch": 0.21790768271320746,
      "grad_norm": 0.5933096410062452,
      "learning_rate": 9.102261975099034e-06,
      "loss": 0.5436,
      "step": 2970
    },
    {
      "epoch": 0.21798105230334658,
      "grad_norm": 0.592620706357368,
      "learning_rate": 9.101582549452008e-06,
      "loss": 0.553,
      "step": 2971
    },
    {
      "epoch": 0.2180544218934857,
      "grad_norm": 1.7269716311408712,
      "learning_rate": 9.10090289217846e-06,
      "loss": 0.8657,
      "step": 2972
    },
    {
      "epoch": 0.21812779148362482,
      "grad_norm": 1.7391252072486978,
      "learning_rate": 9.100223003316769e-06,
      "loss": 0.9128,
      "step": 2973
    },
    {
      "epoch": 0.21820116107376394,
      "grad_norm": 1.639919474959599,
      "learning_rate": 9.09954288290533e-06,
      "loss": 0.7564,
      "step": 2974
    },
    {
      "epoch": 0.21827453066390307,
      "grad_norm": 0.641899797739283,
      "learning_rate": 9.098862530982554e-06,
      "loss": 0.5431,
      "step": 2975
    },
    {
      "epoch": 0.21834790025404222,
      "grad_norm": 1.4756778160288888,
      "learning_rate": 9.098181947586858e-06,
      "loss": 0.9268,
      "step": 2976
    },
    {
      "epoch": 0.21834790025404222,
      "lm_loss": 0.5729,
      "step": 2976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21834790025404222,
      "lm_loss": 0.505,
      "step": 2976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21834790025404222,
      "lm_loss": 0.539,
      "step": 2976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21834790025404222,
      "lm_loss": 0.47,
      "step": 2976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21834790025404222,
      "lm_loss": 0.5462,
      "step": 2976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21834790025404222,
      "lm_loss": 0.6081,
      "step": 2976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21834790025404222,
      "lm_loss": 0.5302,
      "step": 2976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21834790025404222,
      "lm_loss": 0.611,
      "step": 2976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.21842126984418134,
      "grad_norm": 0.5936443100271855,
      "learning_rate": 9.097501132756678e-06,
      "loss": 0.5363,
      "step": 2977
    },
    {
      "epoch": 0.21849463943432046,
      "grad_norm": 1.730554006579594,
      "learning_rate": 9.096820086530462e-06,
      "loss": 0.8952,
      "step": 2978
    },
    {
      "epoch": 0.21856800902445958,
      "grad_norm": 0.5522612580751293,
      "learning_rate": 9.096138808946673e-06,
      "loss": 0.5466,
      "step": 2979
    },
    {
      "epoch": 0.2186413786145987,
      "grad_norm": 1.8416776816755107,
      "learning_rate": 9.095457300043778e-06,
      "loss": 0.8643,
      "step": 2980
    },
    {
      "epoch": 0.21871474820473785,
      "grad_norm": 1.568144477575648,
      "learning_rate": 9.09477555986027e-06,
      "loss": 0.7664,
      "step": 2981
    },
    {
      "epoch": 0.21878811779487697,
      "grad_norm": 1.4150165241364672,
      "learning_rate": 9.094093588434645e-06,
      "loss": 0.7137,
      "step": 2982
    },
    {
      "epoch": 0.2188614873850161,
      "grad_norm": 0.6286873087917082,
      "learning_rate": 9.093411385805415e-06,
      "loss": 0.5392,
      "step": 2983
    },
    {
      "epoch": 0.21893485697515522,
      "grad_norm": 1.474502339523466,
      "learning_rate": 9.092728952011107e-06,
      "loss": 0.8301,
      "step": 2984
    },
    {
      "epoch": 0.21893485697515522,
      "lm_loss": 1.4328,
      "step": 2984,
      "vm_loss": 0.038
    },
    {
      "epoch": 0.21893485697515522,
      "lm_loss": 0.7493,
      "step": 2984,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.21893485697515522,
      "lm_loss": 0.7586,
      "step": 2984,
      "vm_loss": 0.1427
    },
    {
      "epoch": 0.21893485697515522,
      "lm_loss": 0.9383,
      "step": 2984,
      "vm_loss": 0.1327
    },
    {
      "epoch": 0.21893485697515522,
      "lm_loss": 0.963,
      "step": 2984,
      "vm_loss": 0.1227
    },
    {
      "epoch": 0.21893485697515522,
      "lm_loss": 0.5195,
      "step": 2984,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.21893485697515522,
      "lm_loss": 0.7489,
      "step": 2984,
      "vm_loss": 0.1778
    },
    {
      "epoch": 0.21893485697515522,
      "lm_loss": 1.2357,
      "step": 2984,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.21900822656529434,
      "grad_norm": 1.6386819508441735,
      "learning_rate": 9.09204628709026e-06,
      "loss": 0.933,
      "step": 2985
    },
    {
      "epoch": 0.2190815961554335,
      "grad_norm": 0.5820028864143112,
      "learning_rate": 9.091363391081425e-06,
      "loss": 0.5349,
      "step": 2986
    },
    {
      "epoch": 0.2191549657455726,
      "grad_norm": 0.6017101990708333,
      "learning_rate": 9.090680264023168e-06,
      "loss": 0.5623,
      "step": 2987
    },
    {
      "epoch": 0.21922833533571173,
      "grad_norm": 1.9162367347282632,
      "learning_rate": 9.089996905954065e-06,
      "loss": 0.8493,
      "step": 2988
    },
    {
      "epoch": 0.21930170492585085,
      "grad_norm": 1.3803178203709598,
      "learning_rate": 9.089313316912708e-06,
      "loss": 0.6934,
      "step": 2989
    },
    {
      "epoch": 0.21937507451598998,
      "grad_norm": 1.614043005774432,
      "learning_rate": 9.088629496937702e-06,
      "loss": 0.8503,
      "step": 2990
    },
    {
      "epoch": 0.21944844410612913,
      "grad_norm": 0.6081172340746936,
      "learning_rate": 9.087945446067663e-06,
      "loss": 0.5292,
      "step": 2991
    },
    {
      "epoch": 0.21952181369626825,
      "grad_norm": 2.3469095315880772,
      "learning_rate": 9.08726116434122e-06,
      "loss": 0.8369,
      "step": 2992
    },
    {
      "epoch": 0.21952181369626825,
      "lm_loss": 1.7698,
      "step": 2992,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.21952181369626825,
      "lm_loss": 1.5007,
      "step": 2992,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.21952181369626825,
      "lm_loss": 0.4571,
      "step": 2992,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.21952181369626825,
      "lm_loss": 0.3006,
      "step": 2992,
      "vm_loss": 0.1272
    },
    {
      "epoch": 0.21952181369626825,
      "lm_loss": 0.8071,
      "step": 2992,
      "vm_loss": 0.2133
    },
    {
      "epoch": 0.21952181369626825,
      "lm_loss": 1.0596,
      "step": 2992,
      "vm_loss": 0.0834
    },
    {
      "epoch": 0.21952181369626825,
      "lm_loss": 0.9075,
      "step": 2992,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.21952181369626825,
      "lm_loss": 0.7784,
      "step": 2992,
      "vm_loss": 0.1995
    },
    {
      "epoch": 0.21959518328640737,
      "grad_norm": 1.5427491881218418,
      "learning_rate": 9.08657665179702e-06,
      "loss": 0.8287,
      "step": 2993
    },
    {
      "epoch": 0.2196685528765465,
      "grad_norm": 1.8506680061281016,
      "learning_rate": 9.085891908473713e-06,
      "loss": 0.8221,
      "step": 2994
    },
    {
      "epoch": 0.2197419224666856,
      "grad_norm": 0.6357134137246973,
      "learning_rate": 9.085206934409973e-06,
      "loss": 0.5387,
      "step": 2995
    },
    {
      "epoch": 0.21981529205682473,
      "grad_norm": 1.7656906385232907,
      "learning_rate": 9.08452172964448e-06,
      "loss": 0.9317,
      "step": 2996
    },
    {
      "epoch": 0.21988866164696388,
      "grad_norm": 1.7243543994819628,
      "learning_rate": 9.083836294215928e-06,
      "loss": 0.8476,
      "step": 2997
    },
    {
      "epoch": 0.219962031237103,
      "grad_norm": 1.5156344455590565,
      "learning_rate": 9.08315062816303e-06,
      "loss": 0.7869,
      "step": 2998
    },
    {
      "epoch": 0.22003540082724213,
      "grad_norm": 1.673659885678767,
      "learning_rate": 9.082464731524502e-06,
      "loss": 0.9774,
      "step": 2999
    },
    {
      "epoch": 0.22010877041738125,
      "grad_norm": 1.5796017836947138,
      "learning_rate": 9.081778604339082e-06,
      "loss": 0.7099,
      "step": 3000
    },
    {
      "epoch": 0.22010877041738125,
      "lm_loss": 1.5171,
      "step": 3000,
      "vm_loss": 0.0497
    },
    {
      "epoch": 0.22010877041738125,
      "lm_loss": 1.2437,
      "step": 3000,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.22010877041738125,
      "lm_loss": 0.4831,
      "step": 3000,
      "vm_loss": 0.1079
    },
    {
      "epoch": 0.22010877041738125,
      "lm_loss": 0.5868,
      "step": 3000,
      "vm_loss": 0.1428
    },
    {
      "epoch": 0.22010877041738125,
      "lm_loss": 0.5527,
      "step": 3000,
      "vm_loss": 0.2251
    },
    {
      "epoch": 0.22010877041738125,
      "lm_loss": 0.7996,
      "step": 3000,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.22010877041738125,
      "lm_loss": 1.3575,
      "step": 3000,
      "vm_loss": 0.1528
    },
    {
      "epoch": 0.22010877041738125,
      "lm_loss": 0.8558,
      "step": 3000,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.22018214000752037,
      "grad_norm": 1.4803797174409465,
      "learning_rate": 9.081092246645515e-06,
      "loss": 0.894,
      "step": 3001
    },
    {
      "epoch": 0.22025550959765952,
      "grad_norm": 1.4187113110742993,
      "learning_rate": 9.080405658482563e-06,
      "loss": 0.9043,
      "step": 3002
    },
    {
      "epoch": 0.22032887918779864,
      "grad_norm": 1.5750841968643166,
      "learning_rate": 9.079718839888998e-06,
      "loss": 0.9286,
      "step": 3003
    },
    {
      "epoch": 0.22040224877793776,
      "grad_norm": 0.7164016532767326,
      "learning_rate": 9.079031790903608e-06,
      "loss": 0.5348,
      "step": 3004
    },
    {
      "epoch": 0.2204756183680769,
      "grad_norm": 1.6463361092292503,
      "learning_rate": 9.078344511565189e-06,
      "loss": 0.8868,
      "step": 3005
    },
    {
      "epoch": 0.220548987958216,
      "grad_norm": 1.4569626029010303,
      "learning_rate": 9.077657001912558e-06,
      "loss": 0.8702,
      "step": 3006
    },
    {
      "epoch": 0.22062235754835516,
      "grad_norm": 1.4706538544306456,
      "learning_rate": 9.076969261984536e-06,
      "loss": 0.8181,
      "step": 3007
    },
    {
      "epoch": 0.22069572713849428,
      "grad_norm": 0.598554221109875,
      "learning_rate": 9.076281291819963e-06,
      "loss": 0.5406,
      "step": 3008
    },
    {
      "epoch": 0.22069572713849428,
      "lm_loss": 0.5276,
      "step": 3008,
      "vm_loss": 0.034
    },
    {
      "epoch": 0.22069572713849428,
      "lm_loss": 0.4386,
      "step": 3008,
      "vm_loss": 0.1269
    },
    {
      "epoch": 0.22069572713849428,
      "lm_loss": 0.0205,
      "step": 3008,
      "vm_loss": 0.1523
    },
    {
      "epoch": 0.22069572713849428,
      "lm_loss": 0.6575,
      "step": 3008,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.22069572713849428,
      "lm_loss": 1.1028,
      "step": 3008,
      "vm_loss": 0.092
    },
    {
      "epoch": 0.22069572713849428,
      "lm_loss": 0.2921,
      "step": 3008,
      "vm_loss": 0.216
    },
    {
      "epoch": 0.22069572713849428,
      "lm_loss": 0.6862,
      "step": 3008,
      "vm_loss": 0.1012
    },
    {
      "epoch": 0.22069572713849428,
      "lm_loss": 0.2225,
      "step": 3008,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.2207690967286334,
      "grad_norm": 1.5484556077908902,
      "learning_rate": 9.075593091457692e-06,
      "loss": 0.6942,
      "step": 3009
    },
    {
      "epoch": 0.22084246631877252,
      "grad_norm": 1.5569837270263382,
      "learning_rate": 9.074904660936588e-06,
      "loss": 0.8295,
      "step": 3010
    },
    {
      "epoch": 0.22091583590891165,
      "grad_norm": 0.6229747509471609,
      "learning_rate": 9.074216000295524e-06,
      "loss": 0.5499,
      "step": 3011
    },
    {
      "epoch": 0.2209892054990508,
      "grad_norm": 0.56789458794449,
      "learning_rate": 9.073527109573393e-06,
      "loss": 0.5287,
      "step": 3012
    },
    {
      "epoch": 0.22106257508918992,
      "grad_norm": 0.569867653890654,
      "learning_rate": 9.072837988809099e-06,
      "loss": 0.5294,
      "step": 3013
    },
    {
      "epoch": 0.22113594467932904,
      "grad_norm": 1.6914916415927914,
      "learning_rate": 9.072148638041556e-06,
      "loss": 0.7488,
      "step": 3014
    },
    {
      "epoch": 0.22120931426946816,
      "grad_norm": 0.5969681492626455,
      "learning_rate": 9.071459057309696e-06,
      "loss": 0.5401,
      "step": 3015
    },
    {
      "epoch": 0.22128268385960728,
      "grad_norm": 1.7204637063166477,
      "learning_rate": 9.07076924665246e-06,
      "loss": 0.8321,
      "step": 3016
    },
    {
      "epoch": 0.22128268385960728,
      "lm_loss": 1.8091,
      "step": 3016,
      "vm_loss": 0.0856
    },
    {
      "epoch": 0.22128268385960728,
      "lm_loss": 0.3441,
      "step": 3016,
      "vm_loss": 0.1153
    },
    {
      "epoch": 0.22128268385960728,
      "lm_loss": 0.0605,
      "step": 3016,
      "vm_loss": 0.1381
    },
    {
      "epoch": 0.22128268385960728,
      "lm_loss": 0.0852,
      "step": 3016,
      "vm_loss": 0.1654
    },
    {
      "epoch": 0.22128268385960728,
      "lm_loss": 0.149,
      "step": 3016,
      "vm_loss": 0.2327
    },
    {
      "epoch": 0.22128268385960728,
      "lm_loss": 0.5332,
      "step": 3016,
      "vm_loss": 0.1446
    },
    {
      "epoch": 0.22128268385960728,
      "lm_loss": 0.4721,
      "step": 3016,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.22128268385960728,
      "lm_loss": 0.3731,
      "step": 3016,
      "vm_loss": 0.1526
    },
    {
      "epoch": 0.2213560534497464,
      "grad_norm": 1.7471463175212905,
      "learning_rate": 9.070079206108804e-06,
      "loss": 0.8079,
      "step": 3017
    },
    {
      "epoch": 0.22142942303988555,
      "grad_norm": 1.5320943082483696,
      "learning_rate": 9.069388935717695e-06,
      "loss": 0.8765,
      "step": 3018
    },
    {
      "epoch": 0.22150279263002468,
      "grad_norm": 1.5254082454604765,
      "learning_rate": 9.068698435518117e-06,
      "loss": 0.8343,
      "step": 3019
    },
    {
      "epoch": 0.2215761622201638,
      "grad_norm": 1.5897188495132337,
      "learning_rate": 9.068007705549061e-06,
      "loss": 0.908,
      "step": 3020
    },
    {
      "epoch": 0.22164953181030292,
      "grad_norm": 0.648215631726934,
      "learning_rate": 9.067316745849535e-06,
      "loss": 0.5471,
      "step": 3021
    },
    {
      "epoch": 0.22172290140044204,
      "grad_norm": 1.624251109942758,
      "learning_rate": 9.06662555645856e-06,
      "loss": 0.9145,
      "step": 3022
    },
    {
      "epoch": 0.2217962709905812,
      "grad_norm": 0.6317745286254408,
      "learning_rate": 9.06593413741517e-06,
      "loss": 0.5275,
      "step": 3023
    },
    {
      "epoch": 0.2218696405807203,
      "grad_norm": 1.534367191873409,
      "learning_rate": 9.06524248875841e-06,
      "loss": 0.8424,
      "step": 3024
    },
    {
      "epoch": 0.2218696405807203,
      "lm_loss": 1.0869,
      "step": 3024,
      "vm_loss": 0.0543
    },
    {
      "epoch": 0.2218696405807203,
      "lm_loss": 1.5745,
      "step": 3024,
      "vm_loss": 0.1069
    },
    {
      "epoch": 0.2218696405807203,
      "lm_loss": 0.4135,
      "step": 3024,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.2218696405807203,
      "lm_loss": 0.417,
      "step": 3024,
      "vm_loss": 0.1538
    },
    {
      "epoch": 0.2218696405807203,
      "lm_loss": 0.5274,
      "step": 3024,
      "vm_loss": 0.1598
    },
    {
      "epoch": 0.2218696405807203,
      "lm_loss": 0.4629,
      "step": 3024,
      "vm_loss": 0.1216
    },
    {
      "epoch": 0.2218696405807203,
      "lm_loss": 0.888,
      "step": 3024,
      "vm_loss": 0.2008
    },
    {
      "epoch": 0.2218696405807203,
      "lm_loss": 0.1493,
      "step": 3024,
      "vm_loss": 0.1796
    },
    {
      "epoch": 0.22194301017085943,
      "grad_norm": 1.5853166287022529,
      "learning_rate": 9.064550610527338e-06,
      "loss": 0.8259,
      "step": 3025
    },
    {
      "epoch": 0.22201637976099856,
      "grad_norm": 0.5979887703901576,
      "learning_rate": 9.063858502761029e-06,
      "loss": 0.551,
      "step": 3026
    },
    {
      "epoch": 0.22208974935113768,
      "grad_norm": 0.6066709004674689,
      "learning_rate": 9.063166165498564e-06,
      "loss": 0.5365,
      "step": 3027
    },
    {
      "epoch": 0.22216311894127683,
      "grad_norm": 1.7523467016889696,
      "learning_rate": 9.062473598779045e-06,
      "loss": 0.972,
      "step": 3028
    },
    {
      "epoch": 0.22223648853141595,
      "grad_norm": 1.605331948707955,
      "learning_rate": 9.061780802641582e-06,
      "loss": 0.9006,
      "step": 3029
    },
    {
      "epoch": 0.22230985812155507,
      "grad_norm": 1.5661369085386185,
      "learning_rate": 9.061087777125299e-06,
      "loss": 0.8788,
      "step": 3030
    },
    {
      "epoch": 0.2223832277116942,
      "grad_norm": 1.5081964968809958,
      "learning_rate": 9.060394522269328e-06,
      "loss": 0.8022,
      "step": 3031
    },
    {
      "epoch": 0.22245659730183331,
      "grad_norm": 1.7860598250895787,
      "learning_rate": 9.059701038112827e-06,
      "loss": 0.8216,
      "step": 3032
    },
    {
      "epoch": 0.22245659730183331,
      "lm_loss": 1.6097,
      "step": 3032,
      "vm_loss": 0.0789
    },
    {
      "epoch": 0.22245659730183331,
      "lm_loss": 1.3735,
      "step": 3032,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.22245659730183331,
      "lm_loss": 0.3233,
      "step": 3032,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.22245659730183331,
      "lm_loss": 0.7259,
      "step": 3032,
      "vm_loss": 0.1332
    },
    {
      "epoch": 0.22245659730183331,
      "lm_loss": 0.7726,
      "step": 3032,
      "vm_loss": 0.1734
    },
    {
      "epoch": 0.22245659730183331,
      "lm_loss": 0.1594,
      "step": 3032,
      "vm_loss": 0.1017
    },
    {
      "epoch": 0.22245659730183331,
      "lm_loss": 0.7826,
      "step": 3032,
      "vm_loss": 0.15
    },
    {
      "epoch": 0.22245659730183331,
      "lm_loss": 0.0944,
      "step": 3032,
      "vm_loss": 0.0806
    },
    {
      "epoch": 0.22252996689197244,
      "grad_norm": 1.3538379487641787,
      "learning_rate": 9.059007324694952e-06,
      "loss": 0.7556,
      "step": 3033
    },
    {
      "epoch": 0.22260333648211159,
      "grad_norm": 1.643708050702113,
      "learning_rate": 9.058313382054883e-06,
      "loss": 0.8409,
      "step": 3034
    },
    {
      "epoch": 0.2226767060722507,
      "grad_norm": 1.6362691287707876,
      "learning_rate": 9.057619210231807e-06,
      "loss": 0.884,
      "step": 3035
    },
    {
      "epoch": 0.22275007566238983,
      "grad_norm": 0.6882965690354444,
      "learning_rate": 9.056924809264926e-06,
      "loss": 0.5473,
      "step": 3036
    },
    {
      "epoch": 0.22282344525252895,
      "grad_norm": 1.4756897645066553,
      "learning_rate": 9.056230179193454e-06,
      "loss": 0.8003,
      "step": 3037
    },
    {
      "epoch": 0.22289681484266807,
      "grad_norm": 1.6579062943261855,
      "learning_rate": 9.05553532005662e-06,
      "loss": 0.9256,
      "step": 3038
    },
    {
      "epoch": 0.22297018443280722,
      "grad_norm": 1.3978216027914563,
      "learning_rate": 9.05484023189366e-06,
      "loss": 0.7754,
      "step": 3039
    },
    {
      "epoch": 0.22304355402294634,
      "grad_norm": 1.542202995720083,
      "learning_rate": 9.054144914743832e-06,
      "loss": 0.8826,
      "step": 3040
    },
    {
      "epoch": 0.22304355402294634,
      "lm_loss": 0.8273,
      "step": 3040,
      "vm_loss": 0.0565
    },
    {
      "epoch": 0.22304355402294634,
      "lm_loss": 1.3315,
      "step": 3040,
      "vm_loss": 0.0321
    },
    {
      "epoch": 0.22304355402294634,
      "lm_loss": 1.0432,
      "step": 3040,
      "vm_loss": 0.1357
    },
    {
      "epoch": 0.22304355402294634,
      "lm_loss": 0.1138,
      "step": 3040,
      "vm_loss": 0.1324
    },
    {
      "epoch": 0.22304355402294634,
      "lm_loss": 0.8414,
      "step": 3040,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.22304355402294634,
      "lm_loss": 0.2689,
      "step": 3040,
      "vm_loss": 0.1405
    },
    {
      "epoch": 0.22304355402294634,
      "lm_loss": 0.3242,
      "step": 3040,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.22304355402294634,
      "lm_loss": 1.0197,
      "step": 3040,
      "vm_loss": 0.148
    },
    {
      "epoch": 0.22311692361308547,
      "grad_norm": 1.5641707036205372,
      "learning_rate": 9.0534493686464e-06,
      "loss": 0.7917,
      "step": 3041
    },
    {
      "epoch": 0.2231902932032246,
      "grad_norm": 1.5860574735672024,
      "learning_rate": 9.052753593640645e-06,
      "loss": 0.9347,
      "step": 3042
    },
    {
      "epoch": 0.2232636627933637,
      "grad_norm": 1.824398934227548,
      "learning_rate": 9.052057589765856e-06,
      "loss": 0.8307,
      "step": 3043
    },
    {
      "epoch": 0.22333703238350286,
      "grad_norm": 3.3963433605455844,
      "learning_rate": 9.051361357061343e-06,
      "loss": 0.8773,
      "step": 3044
    },
    {
      "epoch": 0.22341040197364198,
      "grad_norm": 1.681561863044313,
      "learning_rate": 9.05066489556642e-06,
      "loss": 0.8804,
      "step": 3045
    },
    {
      "epoch": 0.2234837715637811,
      "grad_norm": 1.653234889048435,
      "learning_rate": 9.04996820532042e-06,
      "loss": 0.8924,
      "step": 3046
    },
    {
      "epoch": 0.22355714115392022,
      "grad_norm": 1.5316680445311575,
      "learning_rate": 9.049271286362684e-06,
      "loss": 0.9456,
      "step": 3047
    },
    {
      "epoch": 0.22363051074405935,
      "grad_norm": 1.6534038098376964,
      "learning_rate": 9.048574138732571e-06,
      "loss": 0.8634,
      "step": 3048
    },
    {
      "epoch": 0.22363051074405935,
      "lm_loss": 0.6959,
      "step": 3048,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.22363051074405935,
      "lm_loss": 0.6615,
      "step": 3048,
      "vm_loss": 0.1617
    },
    {
      "epoch": 0.22363051074405935,
      "lm_loss": 0.609,
      "step": 3048,
      "vm_loss": 0.088
    },
    {
      "epoch": 0.22363051074405935,
      "lm_loss": 1.1288,
      "step": 3048,
      "vm_loss": 0.2316
    },
    {
      "epoch": 0.22363051074405935,
      "lm_loss": 0.1471,
      "step": 3048,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.22363051074405935,
      "lm_loss": 0.4856,
      "step": 3048,
      "vm_loss": 0.1153
    },
    {
      "epoch": 0.22363051074405935,
      "lm_loss": 0.5921,
      "step": 3048,
      "vm_loss": 0.105
    },
    {
      "epoch": 0.22363051074405935,
      "lm_loss": 0.8709,
      "step": 3048,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.2237038803341985,
      "grad_norm": 1.5813944506102648,
      "learning_rate": 9.047876762469451e-06,
      "loss": 0.7935,
      "step": 3049
    },
    {
      "epoch": 0.22377724992433762,
      "grad_norm": 1.5137084713646898,
      "learning_rate": 9.047179157612707e-06,
      "loss": 0.7619,
      "step": 3050
    },
    {
      "epoch": 0.22385061951447674,
      "grad_norm": 1.6698384133615167,
      "learning_rate": 9.04648132420173e-06,
      "loss": 0.852,
      "step": 3051
    },
    {
      "epoch": 0.22392398910461586,
      "grad_norm": 1.7812728610696904,
      "learning_rate": 9.045783262275933e-06,
      "loss": 0.9128,
      "step": 3052
    },
    {
      "epoch": 0.22399735869475498,
      "grad_norm": 1.7214716642020753,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.8766,
      "step": 3053
    },
    {
      "epoch": 0.2240707282848941,
      "grad_norm": 1.5332546888482572,
      "learning_rate": 9.044386453037575e-06,
      "loss": 0.8579,
      "step": 3054
    },
    {
      "epoch": 0.22414409787503325,
      "grad_norm": 1.5965897350154268,
      "learning_rate": 9.043687705803894e-06,
      "loss": 0.8349,
      "step": 3055
    },
    {
      "epoch": 0.22421746746517238,
      "grad_norm": 1.5940354152611564,
      "learning_rate": 9.042988730213155e-06,
      "loss": 0.7929,
      "step": 3056
    },
    {
      "epoch": 0.22421746746517238,
      "lm_loss": 1.5576,
      "step": 3056,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.22421746746517238,
      "lm_loss": 0.602,
      "step": 3056,
      "vm_loss": 0.0896
    },
    {
      "epoch": 0.22421746746517238,
      "lm_loss": 0.9257,
      "step": 3056,
      "vm_loss": 0.2084
    },
    {
      "epoch": 0.22421746746517238,
      "lm_loss": 0.1731,
      "step": 3056,
      "vm_loss": 0.1722
    },
    {
      "epoch": 0.22421746746517238,
      "lm_loss": 1.0633,
      "step": 3056,
      "vm_loss": 0.165
    },
    {
      "epoch": 0.22421746746517238,
      "lm_loss": 0.2081,
      "step": 3056,
      "vm_loss": 0.0838
    },
    {
      "epoch": 0.22421746746517238,
      "lm_loss": 0.7596,
      "step": 3056,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.22421746746517238,
      "lm_loss": 0.5323,
      "step": 3056,
      "vm_loss": 0.072
    },
    {
      "epoch": 0.2242908370553115,
      "grad_norm": 1.7456623368039317,
      "learning_rate": 9.04228952630483e-06,
      "loss": 0.8817,
      "step": 3057
    },
    {
      "epoch": 0.22436420664545062,
      "grad_norm": 1.4636611466964284,
      "learning_rate": 9.041590094118402e-06,
      "loss": 0.8184,
      "step": 3058
    },
    {
      "epoch": 0.22443757623558974,
      "grad_norm": 1.6077169971652978,
      "learning_rate": 9.040890433693377e-06,
      "loss": 0.9075,
      "step": 3059
    },
    {
      "epoch": 0.2245109458257289,
      "grad_norm": 1.521673964829318,
      "learning_rate": 9.040190545069259e-06,
      "loss": 0.879,
      "step": 3060
    },
    {
      "epoch": 0.224584315415868,
      "grad_norm": 0.65244712834373,
      "learning_rate": 9.039490428285577e-06,
      "loss": 0.5618,
      "step": 3061
    },
    {
      "epoch": 0.22465768500600714,
      "grad_norm": 1.880898465283442,
      "learning_rate": 9.038790083381866e-06,
      "loss": 0.8235,
      "step": 3062
    },
    {
      "epoch": 0.22473105459614626,
      "grad_norm": 1.5418212451452051,
      "learning_rate": 9.038089510397677e-06,
      "loss": 0.852,
      "step": 3063
    },
    {
      "epoch": 0.22480442418628538,
      "grad_norm": 1.4784606115172951,
      "learning_rate": 9.037388709372573e-06,
      "loss": 0.7991,
      "step": 3064
    },
    {
      "epoch": 0.22480442418628538,
      "lm_loss": 0.4821,
      "step": 3064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.22480442418628538,
      "lm_loss": 0.4665,
      "step": 3064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.22480442418628538,
      "lm_loss": 0.5084,
      "step": 3064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.22480442418628538,
      "lm_loss": 0.5755,
      "step": 3064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.22480442418628538,
      "lm_loss": 0.6902,
      "step": 3064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.22480442418628538,
      "lm_loss": 0.5508,
      "step": 3064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.22480442418628538,
      "lm_loss": 0.672,
      "step": 3064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.22480442418628538,
      "lm_loss": 0.6046,
      "step": 3064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.22487779377642453,
      "grad_norm": 0.598953851118759,
      "learning_rate": 9.036687680346132e-06,
      "loss": 0.5567,
      "step": 3065
    },
    {
      "epoch": 0.22495116336656365,
      "grad_norm": 1.5911518411871168,
      "learning_rate": 9.035986423357939e-06,
      "loss": 0.8427,
      "step": 3066
    },
    {
      "epoch": 0.22502453295670277,
      "grad_norm": 1.6381768174459361,
      "learning_rate": 9.035284938447598e-06,
      "loss": 0.8184,
      "step": 3067
    },
    {
      "epoch": 0.2250979025468419,
      "grad_norm": 0.6110121008561195,
      "learning_rate": 9.034583225654724e-06,
      "loss": 0.5288,
      "step": 3068
    },
    {
      "epoch": 0.22517127213698102,
      "grad_norm": 1.5423595889673403,
      "learning_rate": 9.033881285018945e-06,
      "loss": 0.8283,
      "step": 3069
    },
    {
      "epoch": 0.22524464172712017,
      "grad_norm": 0.5765318608273766,
      "learning_rate": 9.033179116579898e-06,
      "loss": 0.5571,
      "step": 3070
    },
    {
      "epoch": 0.2253180113172593,
      "grad_norm": 0.5777412313987891,
      "learning_rate": 9.032476720377239e-06,
      "loss": 0.5403,
      "step": 3071
    },
    {
      "epoch": 0.2253913809073984,
      "grad_norm": 1.7307402376996703,
      "learning_rate": 9.031774096450631e-06,
      "loss": 0.841,
      "step": 3072
    },
    {
      "epoch": 0.2253913809073984,
      "lm_loss": 0.9043,
      "step": 3072,
      "vm_loss": 0.0524
    },
    {
      "epoch": 0.2253913809073984,
      "lm_loss": 1.4266,
      "step": 3072,
      "vm_loss": 0.1298
    },
    {
      "epoch": 0.2253913809073984,
      "lm_loss": 0.9147,
      "step": 3072,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.2253913809073984,
      "lm_loss": 0.9558,
      "step": 3072,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.2253913809073984,
      "lm_loss": 0.4242,
      "step": 3072,
      "vm_loss": 0.0685
    },
    {
      "epoch": 0.2253913809073984,
      "lm_loss": 0.4876,
      "step": 3072,
      "vm_loss": 0.2017
    },
    {
      "epoch": 0.2253913809073984,
      "lm_loss": 1.1555,
      "step": 3072,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.2253913809073984,
      "lm_loss": 0.477,
      "step": 3072,
      "vm_loss": 0.2325
    },
    {
      "epoch": 0.22546475049753753,
      "grad_norm": 1.5849992089605736,
      "learning_rate": 9.03107124483976e-06,
      "loss": 0.8348,
      "step": 3073
    },
    {
      "epoch": 0.22553812008767665,
      "grad_norm": 0.5959007314018275,
      "learning_rate": 9.03036816558431e-06,
      "loss": 0.527,
      "step": 3074
    },
    {
      "epoch": 0.22561148967781577,
      "grad_norm": 1.43234067156813,
      "learning_rate": 9.029664858723988e-06,
      "loss": 0.7976,
      "step": 3075
    },
    {
      "epoch": 0.22568485926795492,
      "grad_norm": 1.7643898248288383,
      "learning_rate": 9.028961324298513e-06,
      "loss": 0.8002,
      "step": 3076
    },
    {
      "epoch": 0.22575822885809405,
      "grad_norm": 1.6846090281472599,
      "learning_rate": 9.028257562347615e-06,
      "loss": 0.8901,
      "step": 3077
    },
    {
      "epoch": 0.22583159844823317,
      "grad_norm": 1.7911854931008477,
      "learning_rate": 9.027553572911037e-06,
      "loss": 0.911,
      "step": 3078
    },
    {
      "epoch": 0.2259049680383723,
      "grad_norm": 1.5131056470451285,
      "learning_rate": 9.026849356028534e-06,
      "loss": 0.801,
      "step": 3079
    },
    {
      "epoch": 0.2259783376285114,
      "grad_norm": 1.6963680980348317,
      "learning_rate": 9.026144911739876e-06,
      "loss": 0.8848,
      "step": 3080
    },
    {
      "epoch": 0.2259783376285114,
      "lm_loss": 1.116,
      "step": 3080,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.2259783376285114,
      "lm_loss": 1.1643,
      "step": 3080,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.2259783376285114,
      "lm_loss": 1.1203,
      "step": 3080,
      "vm_loss": 0.2011
    },
    {
      "epoch": 0.2259783376285114,
      "lm_loss": 0.9096,
      "step": 3080,
      "vm_loss": 0.2297
    },
    {
      "epoch": 0.2259783376285114,
      "lm_loss": 0.1186,
      "step": 3080,
      "vm_loss": 0.0656
    },
    {
      "epoch": 0.2259783376285114,
      "lm_loss": 0.74,
      "step": 3080,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.2259783376285114,
      "lm_loss": 1.1631,
      "step": 3080,
      "vm_loss": 0.246
    },
    {
      "epoch": 0.2259783376285114,
      "lm_loss": 0.3404,
      "step": 3080,
      "vm_loss": 0.0834
    },
    {
      "epoch": 0.22605170721865056,
      "grad_norm": 1.9007198861257455,
      "learning_rate": 9.025440240084844e-06,
      "loss": 0.9542,
      "step": 3081
    },
    {
      "epoch": 0.22612507680878968,
      "grad_norm": 1.4954551147572097,
      "learning_rate": 9.024735341103232e-06,
      "loss": 0.8418,
      "step": 3082
    },
    {
      "epoch": 0.2261984463989288,
      "grad_norm": 0.5859702864882198,
      "learning_rate": 9.02403021483485e-06,
      "loss": 0.529,
      "step": 3083
    },
    {
      "epoch": 0.22627181598906793,
      "grad_norm": 1.871439459343772,
      "learning_rate": 9.023324861319514e-06,
      "loss": 0.9043,
      "step": 3084
    },
    {
      "epoch": 0.22634518557920705,
      "grad_norm": 1.8089096163982037,
      "learning_rate": 9.022619280597061e-06,
      "loss": 0.8875,
      "step": 3085
    },
    {
      "epoch": 0.2264185551693462,
      "grad_norm": 1.561685956134809,
      "learning_rate": 9.021913472707336e-06,
      "loss": 0.811,
      "step": 3086
    },
    {
      "epoch": 0.22649192475948532,
      "grad_norm": 1.6134982936543592,
      "learning_rate": 9.021207437690196e-06,
      "loss": 0.7132,
      "step": 3087
    },
    {
      "epoch": 0.22656529434962444,
      "grad_norm": 1.5036443039629739,
      "learning_rate": 9.020501175585514e-06,
      "loss": 0.8286,
      "step": 3088
    },
    {
      "epoch": 0.22656529434962444,
      "lm_loss": 1.19,
      "step": 3088,
      "vm_loss": 0.046
    },
    {
      "epoch": 0.22656529434962444,
      "lm_loss": 0.3713,
      "step": 3088,
      "vm_loss": 0.1278
    },
    {
      "epoch": 0.22656529434962444,
      "lm_loss": 0.4024,
      "step": 3088,
      "vm_loss": 0.1326
    },
    {
      "epoch": 0.22656529434962444,
      "lm_loss": 0.4862,
      "step": 3088,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.22656529434962444,
      "lm_loss": 0.1729,
      "step": 3088,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.22656529434962444,
      "lm_loss": 1.1435,
      "step": 3088,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.22656529434962444,
      "lm_loss": 0.6385,
      "step": 3088,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.22656529434962444,
      "lm_loss": 0.105,
      "step": 3088,
      "vm_loss": 0.0446
    },
    {
      "epoch": 0.22663866393976356,
      "grad_norm": 1.6901467694028909,
      "learning_rate": 9.019794686433174e-06,
      "loss": 0.7564,
      "step": 3089
    },
    {
      "epoch": 0.22671203352990268,
      "grad_norm": 1.632623734895105,
      "learning_rate": 9.019087970273072e-06,
      "loss": 0.9,
      "step": 3090
    },
    {
      "epoch": 0.22678540312004183,
      "grad_norm": 1.5864785839048388,
      "learning_rate": 9.01838102714512e-06,
      "loss": 0.808,
      "step": 3091
    },
    {
      "epoch": 0.22685877271018096,
      "grad_norm": 1.8540594400428771,
      "learning_rate": 9.017673857089242e-06,
      "loss": 0.9584,
      "step": 3092
    },
    {
      "epoch": 0.22693214230032008,
      "grad_norm": 1.5416993137461195,
      "learning_rate": 9.016966460145369e-06,
      "loss": 0.8561,
      "step": 3093
    },
    {
      "epoch": 0.2270055118904592,
      "grad_norm": 0.6315448557499703,
      "learning_rate": 9.016258836353453e-06,
      "loss": 0.5444,
      "step": 3094
    },
    {
      "epoch": 0.22707888148059832,
      "grad_norm": 1.781034537071777,
      "learning_rate": 9.015550985753456e-06,
      "loss": 0.8132,
      "step": 3095
    },
    {
      "epoch": 0.22715225107073744,
      "grad_norm": 1.6871406639474982,
      "learning_rate": 9.014842908385348e-06,
      "loss": 0.8396,
      "step": 3096
    },
    {
      "epoch": 0.22715225107073744,
      "lm_loss": 0.8239,
      "step": 3096,
      "vm_loss": 0.0496
    },
    {
      "epoch": 0.22715225107073744,
      "lm_loss": 1.2298,
      "step": 3096,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.22715225107073744,
      "lm_loss": 1.0433,
      "step": 3096,
      "vm_loss": 0.1561
    },
    {
      "epoch": 0.22715225107073744,
      "lm_loss": 1.1155,
      "step": 3096,
      "vm_loss": 0.141
    },
    {
      "epoch": 0.22715225107073744,
      "lm_loss": 0.4731,
      "step": 3096,
      "vm_loss": 0.1244
    },
    {
      "epoch": 0.22715225107073744,
      "lm_loss": 0.8805,
      "step": 3096,
      "vm_loss": 0.0718
    },
    {
      "epoch": 0.22715225107073744,
      "lm_loss": 0.2354,
      "step": 3096,
      "vm_loss": 0.142
    },
    {
      "epoch": 0.22715225107073744,
      "lm_loss": 0.7491,
      "step": 3096,
      "vm_loss": 0.1272
    },
    {
      "epoch": 0.2272256206608766,
      "grad_norm": 1.7325637391561102,
      "learning_rate": 9.014134604289119e-06,
      "loss": 0.8989,
      "step": 3097
    },
    {
      "epoch": 0.22729899025101571,
      "grad_norm": 1.458915425597555,
      "learning_rate": 9.013426073504769e-06,
      "loss": 0.8026,
      "step": 3098
    },
    {
      "epoch": 0.22737235984115484,
      "grad_norm": 1.533780390136979,
      "learning_rate": 9.01271731607231e-06,
      "loss": 0.8584,
      "step": 3099
    },
    {
      "epoch": 0.22744572943129396,
      "grad_norm": 1.805419431595691,
      "learning_rate": 9.012008332031765e-06,
      "loss": 0.8157,
      "step": 3100
    },
    {
      "epoch": 0.22751909902143308,
      "grad_norm": 0.6366189623476334,
      "learning_rate": 9.011299121423176e-06,
      "loss": 0.5394,
      "step": 3101
    },
    {
      "epoch": 0.22759246861157223,
      "grad_norm": 1.7092972113215068,
      "learning_rate": 9.010589684286591e-06,
      "loss": 0.8899,
      "step": 3102
    },
    {
      "epoch": 0.22766583820171135,
      "grad_norm": 1.6226088059169483,
      "learning_rate": 9.009880020662075e-06,
      "loss": 0.8344,
      "step": 3103
    },
    {
      "epoch": 0.22773920779185047,
      "grad_norm": 1.721923066784378,
      "learning_rate": 9.009170130589702e-06,
      "loss": 0.7844,
      "step": 3104
    },
    {
      "epoch": 0.22773920779185047,
      "lm_loss": 0.8372,
      "step": 3104,
      "vm_loss": 0.0271
    },
    {
      "epoch": 0.22773920779185047,
      "lm_loss": 0.6886,
      "step": 3104,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.22773920779185047,
      "lm_loss": 0.248,
      "step": 3104,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.22773920779185047,
      "lm_loss": 0.1789,
      "step": 3104,
      "vm_loss": 0.0744
    },
    {
      "epoch": 0.22773920779185047,
      "lm_loss": 0.694,
      "step": 3104,
      "vm_loss": 0.1927
    },
    {
      "epoch": 0.22773920779185047,
      "lm_loss": 0.2798,
      "step": 3104,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.22773920779185047,
      "lm_loss": 0.6429,
      "step": 3104,
      "vm_loss": 0.1217
    },
    {
      "epoch": 0.22773920779185047,
      "lm_loss": 0.353,
      "step": 3104,
      "vm_loss": 0.0601
    },
    {
      "epoch": 0.2278125773819896,
      "grad_norm": 1.7515575397906935,
      "learning_rate": 9.008460014109564e-06,
      "loss": 0.8598,
      "step": 3105
    },
    {
      "epoch": 0.22788594697212872,
      "grad_norm": 1.835200535466461,
      "learning_rate": 9.007749671261762e-06,
      "loss": 0.8404,
      "step": 3106
    },
    {
      "epoch": 0.22795931656226787,
      "grad_norm": 1.5423067635983991,
      "learning_rate": 9.007039102086412e-06,
      "loss": 0.9641,
      "step": 3107
    },
    {
      "epoch": 0.228032686152407,
      "grad_norm": 1.8443252471189395,
      "learning_rate": 9.00632830662364e-06,
      "loss": 0.8208,
      "step": 3108
    },
    {
      "epoch": 0.2281060557425461,
      "grad_norm": 0.7348171753340107,
      "learning_rate": 9.005617284913586e-06,
      "loss": 0.5528,
      "step": 3109
    },
    {
      "epoch": 0.22817942533268523,
      "grad_norm": 0.6490556942781528,
      "learning_rate": 9.004906036996406e-06,
      "loss": 0.5396,
      "step": 3110
    },
    {
      "epoch": 0.22825279492282435,
      "grad_norm": 1.540131485785653,
      "learning_rate": 9.004194562912263e-06,
      "loss": 0.8946,
      "step": 3111
    },
    {
      "epoch": 0.2283261645129635,
      "grad_norm": 0.5993809809687731,
      "learning_rate": 9.003482862701338e-06,
      "loss": 0.544,
      "step": 3112
    },
    {
      "epoch": 0.2283261645129635,
      "lm_loss": 1.2345,
      "step": 3112,
      "vm_loss": 0.076
    },
    {
      "epoch": 0.2283261645129635,
      "lm_loss": 0.2214,
      "step": 3112,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.2283261645129635,
      "lm_loss": 0.5253,
      "step": 3112,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.2283261645129635,
      "lm_loss": 0.5127,
      "step": 3112,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.2283261645129635,
      "lm_loss": 0.9243,
      "step": 3112,
      "vm_loss": 0.121
    },
    {
      "epoch": 0.2283261645129635,
      "lm_loss": 0.9675,
      "step": 3112,
      "vm_loss": 0.1459
    },
    {
      "epoch": 0.2283261645129635,
      "lm_loss": 0.5716,
      "step": 3112,
      "vm_loss": 0.2059
    },
    {
      "epoch": 0.2283261645129635,
      "lm_loss": 0.9731,
      "step": 3112,
      "vm_loss": 0.1517
    },
    {
      "epoch": 0.22839953410310262,
      "grad_norm": 1.7126101570133083,
      "learning_rate": 9.002770936403821e-06,
      "loss": 0.8332,
      "step": 3113
    },
    {
      "epoch": 0.22847290369324175,
      "grad_norm": 1.8792013544426733,
      "learning_rate": 9.002058784059915e-06,
      "loss": 0.8692,
      "step": 3114
    },
    {
      "epoch": 0.22854627328338087,
      "grad_norm": 1.639600315079172,
      "learning_rate": 9.00134640570984e-06,
      "loss": 0.8908,
      "step": 3115
    },
    {
      "epoch": 0.22861964287352,
      "grad_norm": 1.4596599044903928,
      "learning_rate": 9.000633801393825e-06,
      "loss": 0.8452,
      "step": 3116
    },
    {
      "epoch": 0.2286930124636591,
      "grad_norm": 1.4517814171244314,
      "learning_rate": 8.99992097115211e-06,
      "loss": 0.8992,
      "step": 3117
    },
    {
      "epoch": 0.22876638205379826,
      "grad_norm": 0.8261760853639166,
      "learning_rate": 8.999207915024952e-06,
      "loss": 0.5491,
      "step": 3118
    },
    {
      "epoch": 0.22883975164393738,
      "grad_norm": 1.6877791292355122,
      "learning_rate": 8.998494633052622e-06,
      "loss": 0.8986,
      "step": 3119
    },
    {
      "epoch": 0.2289131212340765,
      "grad_norm": 1.5550252063741272,
      "learning_rate": 8.997781125275395e-06,
      "loss": 0.9547,
      "step": 3120
    },
    {
      "epoch": 0.2289131212340765,
      "lm_loss": 1.4451,
      "step": 3120,
      "vm_loss": 0.066
    },
    {
      "epoch": 0.2289131212340765,
      "lm_loss": 0.0059,
      "step": 3120,
      "vm_loss": 0.1555
    },
    {
      "epoch": 0.2289131212340765,
      "lm_loss": 0.6469,
      "step": 3120,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.2289131212340765,
      "lm_loss": 0.6619,
      "step": 3120,
      "vm_loss": 0.2644
    },
    {
      "epoch": 0.2289131212340765,
      "lm_loss": 0.0678,
      "step": 3120,
      "vm_loss": 0.0586
    },
    {
      "epoch": 0.2289131212340765,
      "lm_loss": 1.3244,
      "step": 3120,
      "vm_loss": 0.1425
    },
    {
      "epoch": 0.2289131212340765,
      "lm_loss": 0.1502,
      "step": 3120,
      "vm_loss": 0.1238
    },
    {
      "epoch": 0.2289131212340765,
      "lm_loss": 0.6139,
      "step": 3120,
      "vm_loss": 0.0784
    },
    {
      "epoch": 0.22898649082421563,
      "grad_norm": 1.4336477076681493,
      "learning_rate": 8.997067391733568e-06,
      "loss": 0.7251,
      "step": 3121
    },
    {
      "epoch": 0.22905986041435475,
      "grad_norm": 1.5715117896951836,
      "learning_rate": 8.996353432467447e-06,
      "loss": 0.8021,
      "step": 3122
    },
    {
      "epoch": 0.2291332300044939,
      "grad_norm": 1.5641938455312365,
      "learning_rate": 8.995639247517351e-06,
      "loss": 0.8802,
      "step": 3123
    },
    {
      "epoch": 0.22920659959463302,
      "grad_norm": 1.516184586359918,
      "learning_rate": 8.994924836923612e-06,
      "loss": 0.9475,
      "step": 3124
    },
    {
      "epoch": 0.22927996918477214,
      "grad_norm": 0.670471150297613,
      "learning_rate": 8.994210200726575e-06,
      "loss": 0.5534,
      "step": 3125
    },
    {
      "epoch": 0.22935333877491126,
      "grad_norm": 1.5108659268652858,
      "learning_rate": 8.993495338966594e-06,
      "loss": 0.7892,
      "step": 3126
    },
    {
      "epoch": 0.22942670836505039,
      "grad_norm": 1.5985656793638203,
      "learning_rate": 8.992780251684043e-06,
      "loss": 0.8941,
      "step": 3127
    },
    {
      "epoch": 0.22950007795518954,
      "grad_norm": 1.5318726343554465,
      "learning_rate": 8.992064938919305e-06,
      "loss": 0.8243,
      "step": 3128
    },
    {
      "epoch": 0.22950007795518954,
      "lm_loss": 1.2768,
      "step": 3128,
      "vm_loss": 0.0411
    },
    {
      "epoch": 0.22950007795518954,
      "lm_loss": 1.5922,
      "step": 3128,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.22950007795518954,
      "lm_loss": 0.4456,
      "step": 3128,
      "vm_loss": 0.2353
    },
    {
      "epoch": 0.22950007795518954,
      "lm_loss": 0.0563,
      "step": 3128,
      "vm_loss": 0.1444
    },
    {
      "epoch": 0.22950007795518954,
      "lm_loss": 0.8267,
      "step": 3128,
      "vm_loss": 0.1762
    },
    {
      "epoch": 0.22950007795518954,
      "lm_loss": 0.389,
      "step": 3128,
      "vm_loss": 0.1079
    },
    {
      "epoch": 0.22950007795518954,
      "lm_loss": 0.0947,
      "step": 3128,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.22950007795518954,
      "lm_loss": 0.049,
      "step": 3128,
      "vm_loss": 0.1039
    },
    {
      "epoch": 0.22957344754532866,
      "grad_norm": 1.460106893401476,
      "learning_rate": 8.991349400712772e-06,
      "loss": 0.9062,
      "step": 3129
    },
    {
      "epoch": 0.22964681713546778,
      "grad_norm": 0.5921672871649089,
      "learning_rate": 8.990633637104852e-06,
      "loss": 0.5331,
      "step": 3130
    },
    {
      "epoch": 0.2297201867256069,
      "grad_norm": 0.5938413525178365,
      "learning_rate": 8.98991764813597e-06,
      "loss": 0.5491,
      "step": 3131
    },
    {
      "epoch": 0.22979355631574602,
      "grad_norm": 1.6191524236473618,
      "learning_rate": 8.989201433846558e-06,
      "loss": 0.9026,
      "step": 3132
    },
    {
      "epoch": 0.22986692590588514,
      "grad_norm": 0.5513364375013466,
      "learning_rate": 8.988484994277062e-06,
      "loss": 0.5317,
      "step": 3133
    },
    {
      "epoch": 0.2299402954960243,
      "grad_norm": 1.443718114504789,
      "learning_rate": 8.98776832946794e-06,
      "loss": 0.8912,
      "step": 3134
    },
    {
      "epoch": 0.23001366508616342,
      "grad_norm": 0.611467908744224,
      "learning_rate": 8.987051439459665e-06,
      "loss": 0.5439,
      "step": 3135
    },
    {
      "epoch": 0.23008703467630254,
      "grad_norm": 1.3790221296307696,
      "learning_rate": 8.986334324292722e-06,
      "loss": 0.874,
      "step": 3136
    },
    {
      "epoch": 0.23008703467630254,
      "lm_loss": 0.485,
      "step": 3136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23008703467630254,
      "lm_loss": 0.5077,
      "step": 3136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23008703467630254,
      "lm_loss": 0.4105,
      "step": 3136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23008703467630254,
      "lm_loss": 0.5219,
      "step": 3136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23008703467630254,
      "lm_loss": 0.6596,
      "step": 3136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23008703467630254,
      "lm_loss": 0.5534,
      "step": 3136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23008703467630254,
      "lm_loss": 0.6248,
      "step": 3136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23008703467630254,
      "lm_loss": 0.426,
      "step": 3136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23016040426644166,
      "grad_norm": 0.5950290387858699,
      "learning_rate": 8.985616984007609e-06,
      "loss": 0.53,
      "step": 3137
    },
    {
      "epoch": 0.23023377385658078,
      "grad_norm": 0.5591554086445013,
      "learning_rate": 8.984899418644831e-06,
      "loss": 0.5385,
      "step": 3138
    },
    {
      "epoch": 0.23030714344671993,
      "grad_norm": 1.4881490905727588,
      "learning_rate": 8.984181628244917e-06,
      "loss": 0.8921,
      "step": 3139
    },
    {
      "epoch": 0.23038051303685905,
      "grad_norm": 0.5698741719895299,
      "learning_rate": 8.983463612848399e-06,
      "loss": 0.5222,
      "step": 3140
    },
    {
      "epoch": 0.23045388262699817,
      "grad_norm": 1.6208490876444006,
      "learning_rate": 8.982745372495826e-06,
      "loss": 0.8547,
      "step": 3141
    },
    {
      "epoch": 0.2305272522171373,
      "grad_norm": 0.5693359619929691,
      "learning_rate": 8.98202690722776e-06,
      "loss": 0.5314,
      "step": 3142
    },
    {
      "epoch": 0.23060062180727642,
      "grad_norm": 1.5532792184705544,
      "learning_rate": 8.981308217084771e-06,
      "loss": 0.8444,
      "step": 3143
    },
    {
      "epoch": 0.23067399139741557,
      "grad_norm": 1.534000225130178,
      "learning_rate": 8.980589302107448e-06,
      "loss": 0.7247,
      "step": 3144
    },
    {
      "epoch": 0.23067399139741557,
      "lm_loss": 1.1636,
      "step": 3144,
      "vm_loss": 0.1252
    },
    {
      "epoch": 0.23067399139741557,
      "lm_loss": 0.3531,
      "step": 3144,
      "vm_loss": 0.0778
    },
    {
      "epoch": 0.23067399139741557,
      "lm_loss": 0.436,
      "step": 3144,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.23067399139741557,
      "lm_loss": 0.2676,
      "step": 3144,
      "vm_loss": 0.0777
    },
    {
      "epoch": 0.23067399139741557,
      "lm_loss": 1.5432,
      "step": 3144,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.23067399139741557,
      "lm_loss": 0.7202,
      "step": 3144,
      "vm_loss": 0.1918
    },
    {
      "epoch": 0.23067399139741557,
      "lm_loss": 0.2498,
      "step": 3144,
      "vm_loss": 0.1126
    },
    {
      "epoch": 0.23067399139741557,
      "lm_loss": 0.1807,
      "step": 3144,
      "vm_loss": 0.0533
    },
    {
      "epoch": 0.2307473609875547,
      "grad_norm": 1.6186207468986569,
      "learning_rate": 8.97987016233639e-06,
      "loss": 0.8249,
      "step": 3145
    },
    {
      "epoch": 0.2308207305776938,
      "grad_norm": 1.6572036137722617,
      "learning_rate": 8.979150797812209e-06,
      "loss": 0.7983,
      "step": 3146
    },
    {
      "epoch": 0.23089410016783293,
      "grad_norm": 1.6089469286375448,
      "learning_rate": 8.978431208575526e-06,
      "loss": 0.964,
      "step": 3147
    },
    {
      "epoch": 0.23096746975797205,
      "grad_norm": 0.6394949333813591,
      "learning_rate": 8.977711394666981e-06,
      "loss": 0.5147,
      "step": 3148
    },
    {
      "epoch": 0.2310408393481112,
      "grad_norm": 0.6170774324926752,
      "learning_rate": 8.976991356127225e-06,
      "loss": 0.5331,
      "step": 3149
    },
    {
      "epoch": 0.23111420893825033,
      "grad_norm": 1.6282978650210373,
      "learning_rate": 8.976271092996916e-06,
      "loss": 0.8193,
      "step": 3150
    },
    {
      "epoch": 0.23118757852838945,
      "grad_norm": 1.7209093919447769,
      "learning_rate": 8.975550605316731e-06,
      "loss": 0.8237,
      "step": 3151
    },
    {
      "epoch": 0.23126094811852857,
      "grad_norm": 1.6046824329685077,
      "learning_rate": 8.974829893127359e-06,
      "loss": 0.8558,
      "step": 3152
    },
    {
      "epoch": 0.23126094811852857,
      "lm_loss": 0.757,
      "step": 3152,
      "vm_loss": 0.0791
    },
    {
      "epoch": 0.23126094811852857,
      "lm_loss": 0.2509,
      "step": 3152,
      "vm_loss": 0.1066
    },
    {
      "epoch": 0.23126094811852857,
      "lm_loss": 1.245,
      "step": 3152,
      "vm_loss": 0.2694
    },
    {
      "epoch": 0.23126094811852857,
      "lm_loss": 0.8335,
      "step": 3152,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.23126094811852857,
      "lm_loss": 0.914,
      "step": 3152,
      "vm_loss": 0.2511
    },
    {
      "epoch": 0.23126094811852857,
      "lm_loss": 0.2483,
      "step": 3152,
      "vm_loss": 0.0815
    },
    {
      "epoch": 0.23126094811852857,
      "lm_loss": 0.9041,
      "step": 3152,
      "vm_loss": 0.1575
    },
    {
      "epoch": 0.23126094811852857,
      "lm_loss": 0.7017,
      "step": 3152,
      "vm_loss": 0.1362
    },
    {
      "epoch": 0.2313343177086677,
      "grad_norm": 1.745079517999002,
      "learning_rate": 8.974108956469499e-06,
      "loss": 0.8517,
      "step": 3153
    },
    {
      "epoch": 0.2314076872988068,
      "grad_norm": 1.658688482721026,
      "learning_rate": 8.973387795383864e-06,
      "loss": 0.879,
      "step": 3154
    },
    {
      "epoch": 0.23148105688894596,
      "grad_norm": 1.6735903615884502,
      "learning_rate": 8.972666409911181e-06,
      "loss": 0.8827,
      "step": 3155
    },
    {
      "epoch": 0.23155442647908508,
      "grad_norm": 1.599598324161289,
      "learning_rate": 8.971944800092189e-06,
      "loss": 0.9338,
      "step": 3156
    },
    {
      "epoch": 0.2316277960692242,
      "grad_norm": 1.458764224517419,
      "learning_rate": 8.971222965967637e-06,
      "loss": 0.8182,
      "step": 3157
    },
    {
      "epoch": 0.23170116565936333,
      "grad_norm": 1.5011078699798726,
      "learning_rate": 8.970500907578288e-06,
      "loss": 0.8713,
      "step": 3158
    },
    {
      "epoch": 0.23177453524950245,
      "grad_norm": 1.559091780094496,
      "learning_rate": 8.969778624964922e-06,
      "loss": 0.7981,
      "step": 3159
    },
    {
      "epoch": 0.2318479048396416,
      "grad_norm": 0.8415290048983233,
      "learning_rate": 8.969056118168325e-06,
      "loss": 0.5369,
      "step": 3160
    },
    {
      "epoch": 0.2318479048396416,
      "lm_loss": 1.7415,
      "step": 3160,
      "vm_loss": 0.0495
    },
    {
      "epoch": 0.2318479048396416,
      "lm_loss": 0.6134,
      "step": 3160,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.2318479048396416,
      "lm_loss": 0.2322,
      "step": 3160,
      "vm_loss": 0.167
    },
    {
      "epoch": 0.2318479048396416,
      "lm_loss": 1.1301,
      "step": 3160,
      "vm_loss": 0.1251
    },
    {
      "epoch": 0.2318479048396416,
      "lm_loss": 0.7704,
      "step": 3160,
      "vm_loss": 0.1368
    },
    {
      "epoch": 0.2318479048396416,
      "lm_loss": 0.0388,
      "step": 3160,
      "vm_loss": 0.1041
    },
    {
      "epoch": 0.2318479048396416,
      "lm_loss": 0.0919,
      "step": 3160,
      "vm_loss": 0.2025
    },
    {
      "epoch": 0.2318479048396416,
      "lm_loss": 0.4115,
      "step": 3160,
      "vm_loss": 0.1209
    },
    {
      "epoch": 0.23192127442978072,
      "grad_norm": 1.5124012612471,
      "learning_rate": 8.968333387229298e-06,
      "loss": 0.7607,
      "step": 3161
    },
    {
      "epoch": 0.23199464401991984,
      "grad_norm": 0.5660060470787573,
      "learning_rate": 8.96761043218866e-06,
      "loss": 0.5334,
      "step": 3162
    },
    {
      "epoch": 0.23206801361005897,
      "grad_norm": 1.7218331365466952,
      "learning_rate": 8.966887253087236e-06,
      "loss": 0.7523,
      "step": 3163
    },
    {
      "epoch": 0.2321413832001981,
      "grad_norm": 1.7952452385433382,
      "learning_rate": 8.966163849965861e-06,
      "loss": 0.8968,
      "step": 3164
    },
    {
      "epoch": 0.23221475279033724,
      "grad_norm": 0.716017271376165,
      "learning_rate": 8.965440222865393e-06,
      "loss": 0.5407,
      "step": 3165
    },
    {
      "epoch": 0.23228812238047636,
      "grad_norm": 1.6259943774491317,
      "learning_rate": 8.964716371826697e-06,
      "loss": 0.884,
      "step": 3166
    },
    {
      "epoch": 0.23236149197061548,
      "grad_norm": 1.6940055901368927,
      "learning_rate": 8.96399229689065e-06,
      "loss": 0.8733,
      "step": 3167
    },
    {
      "epoch": 0.2324348615607546,
      "grad_norm": 1.7317222763620166,
      "learning_rate": 8.963267998098137e-06,
      "loss": 0.9102,
      "step": 3168
    },
    {
      "epoch": 0.2324348615607546,
      "lm_loss": 0.1008,
      "step": 3168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2324348615607546,
      "lm_loss": 0.5092,
      "step": 3168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2324348615607546,
      "lm_loss": 0.7836,
      "step": 3168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2324348615607546,
      "lm_loss": 0.3955,
      "step": 3168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2324348615607546,
      "lm_loss": 0.6147,
      "step": 3168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2324348615607546,
      "lm_loss": 0.4897,
      "step": 3168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2324348615607546,
      "lm_loss": 0.593,
      "step": 3168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2324348615607546,
      "lm_loss": 0.6011,
      "step": 3168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23250823115089372,
      "grad_norm": 0.6500062960487982,
      "learning_rate": 8.962543475490068e-06,
      "loss": 0.5338,
      "step": 3169
    },
    {
      "epoch": 0.23258160074103287,
      "grad_norm": 1.8040096742991798,
      "learning_rate": 8.961818729107357e-06,
      "loss": 0.918,
      "step": 3170
    },
    {
      "epoch": 0.232654970331172,
      "grad_norm": 1.5351832338750238,
      "learning_rate": 8.961093758990929e-06,
      "loss": 0.8404,
      "step": 3171
    },
    {
      "epoch": 0.23272833992131112,
      "grad_norm": 1.6576591012229394,
      "learning_rate": 8.960368565181727e-06,
      "loss": 0.8851,
      "step": 3172
    },
    {
      "epoch": 0.23280170951145024,
      "grad_norm": 1.6779469490938304,
      "learning_rate": 8.959643147720705e-06,
      "loss": 0.8074,
      "step": 3173
    },
    {
      "epoch": 0.23287507910158936,
      "grad_norm": 1.5825494166517207,
      "learning_rate": 8.95891750664883e-06,
      "loss": 0.8497,
      "step": 3174
    },
    {
      "epoch": 0.23294844869172848,
      "grad_norm": 1.4739993272896892,
      "learning_rate": 8.958191642007076e-06,
      "loss": 0.8722,
      "step": 3175
    },
    {
      "epoch": 0.23302181828186763,
      "grad_norm": 1.419347028277105,
      "learning_rate": 8.957465553836441e-06,
      "loss": 0.885,
      "step": 3176
    },
    {
      "epoch": 0.23302181828186763,
      "lm_loss": 0.3146,
      "step": 3176,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.23302181828186763,
      "lm_loss": 1.1946,
      "step": 3176,
      "vm_loss": 0.0402
    },
    {
      "epoch": 0.23302181828186763,
      "lm_loss": 0.0619,
      "step": 3176,
      "vm_loss": 0.1128
    },
    {
      "epoch": 0.23302181828186763,
      "lm_loss": 0.1984,
      "step": 3176,
      "vm_loss": 0.1649
    },
    {
      "epoch": 0.23302181828186763,
      "lm_loss": 0.8685,
      "step": 3176,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.23302181828186763,
      "lm_loss": 0.9473,
      "step": 3176,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.23302181828186763,
      "lm_loss": 0.9871,
      "step": 3176,
      "vm_loss": 0.163
    },
    {
      "epoch": 0.23302181828186763,
      "lm_loss": 0.3233,
      "step": 3176,
      "vm_loss": 0.1276
    },
    {
      "epoch": 0.23309518787200675,
      "grad_norm": 1.4992017707697414,
      "learning_rate": 8.956739242177924e-06,
      "loss": 0.8056,
      "step": 3177
    },
    {
      "epoch": 0.23316855746214588,
      "grad_norm": 0.7511967885121605,
      "learning_rate": 8.956012707072543e-06,
      "loss": 0.5432,
      "step": 3178
    },
    {
      "epoch": 0.233241927052285,
      "grad_norm": 1.6439912484671195,
      "learning_rate": 8.955285948561328e-06,
      "loss": 0.8204,
      "step": 3179
    },
    {
      "epoch": 0.23331529664242412,
      "grad_norm": 1.545619454218723,
      "learning_rate": 8.954558966685322e-06,
      "loss": 0.8718,
      "step": 3180
    },
    {
      "epoch": 0.23338866623256327,
      "grad_norm": 0.5943877963412453,
      "learning_rate": 8.953831761485576e-06,
      "loss": 0.5484,
      "step": 3181
    },
    {
      "epoch": 0.2334620358227024,
      "grad_norm": 1.8021404208600746,
      "learning_rate": 8.953104333003158e-06,
      "loss": 0.8886,
      "step": 3182
    },
    {
      "epoch": 0.2335354054128415,
      "grad_norm": 1.5488873627064135,
      "learning_rate": 8.952376681279152e-06,
      "loss": 0.8192,
      "step": 3183
    },
    {
      "epoch": 0.23360877500298063,
      "grad_norm": 1.5320300998406777,
      "learning_rate": 8.951648806354644e-06,
      "loss": 0.8748,
      "step": 3184
    },
    {
      "epoch": 0.23360877500298063,
      "lm_loss": 1.2371,
      "step": 3184,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.23360877500298063,
      "lm_loss": 1.4734,
      "step": 3184,
      "vm_loss": 0.0737
    },
    {
      "epoch": 0.23360877500298063,
      "lm_loss": 0.601,
      "step": 3184,
      "vm_loss": 0.1398
    },
    {
      "epoch": 0.23360877500298063,
      "lm_loss": 0.4967,
      "step": 3184,
      "vm_loss": 0.1393
    },
    {
      "epoch": 0.23360877500298063,
      "lm_loss": 0.0853,
      "step": 3184,
      "vm_loss": 0.1302
    },
    {
      "epoch": 0.23360877500298063,
      "lm_loss": 0.7672,
      "step": 3184,
      "vm_loss": 0.0735
    },
    {
      "epoch": 0.23360877500298063,
      "lm_loss": 0.3682,
      "step": 3184,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.23360877500298063,
      "lm_loss": 0.3015,
      "step": 3184,
      "vm_loss": 0.0498
    },
    {
      "epoch": 0.23368214459311976,
      "grad_norm": 1.4186313195885976,
      "learning_rate": 8.950920708270743e-06,
      "loss": 0.8468,
      "step": 3185
    },
    {
      "epoch": 0.2337555141832589,
      "grad_norm": 1.5992014708744784,
      "learning_rate": 8.950192387068564e-06,
      "loss": 0.7461,
      "step": 3186
    },
    {
      "epoch": 0.23382888377339803,
      "grad_norm": 1.5205670404196971,
      "learning_rate": 8.94946384278924e-06,
      "loss": 0.8533,
      "step": 3187
    },
    {
      "epoch": 0.23390225336353715,
      "grad_norm": 1.4389603176237373,
      "learning_rate": 8.94873507547391e-06,
      "loss": 0.7819,
      "step": 3188
    },
    {
      "epoch": 0.23397562295367627,
      "grad_norm": 1.411335094795941,
      "learning_rate": 8.948006085163735e-06,
      "loss": 0.8056,
      "step": 3189
    },
    {
      "epoch": 0.2340489925438154,
      "grad_norm": 1.54390270714021,
      "learning_rate": 8.947276871899875e-06,
      "loss": 0.8883,
      "step": 3190
    },
    {
      "epoch": 0.23412236213395454,
      "grad_norm": 0.7511820268454308,
      "learning_rate": 8.946547435723518e-06,
      "loss": 0.5423,
      "step": 3191
    },
    {
      "epoch": 0.23419573172409366,
      "grad_norm": 1.4159822295080318,
      "learning_rate": 8.945817776675853e-06,
      "loss": 0.802,
      "step": 3192
    },
    {
      "epoch": 0.23419573172409366,
      "lm_loss": 1.9449,
      "step": 3192,
      "vm_loss": 0.0567
    },
    {
      "epoch": 0.23419573172409366,
      "lm_loss": 0.4333,
      "step": 3192,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.23419573172409366,
      "lm_loss": 1.1359,
      "step": 3192,
      "vm_loss": 0.1552
    },
    {
      "epoch": 0.23419573172409366,
      "lm_loss": 0.2952,
      "step": 3192,
      "vm_loss": 0.1428
    },
    {
      "epoch": 0.23419573172409366,
      "lm_loss": 0.3539,
      "step": 3192,
      "vm_loss": 0.0591
    },
    {
      "epoch": 0.23419573172409366,
      "lm_loss": 0.9965,
      "step": 3192,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.23419573172409366,
      "lm_loss": 1.0455,
      "step": 3192,
      "vm_loss": 0.1628
    },
    {
      "epoch": 0.23419573172409366,
      "lm_loss": 0.0806,
      "step": 3192,
      "vm_loss": 0.1018
    },
    {
      "epoch": 0.23426910131423279,
      "grad_norm": 1.5807008634088153,
      "learning_rate": 8.945087894798085e-06,
      "loss": 0.8513,
      "step": 3193
    },
    {
      "epoch": 0.2343424709043719,
      "grad_norm": 1.7111719576077622,
      "learning_rate": 8.944357790131435e-06,
      "loss": 0.8994,
      "step": 3194
    },
    {
      "epoch": 0.23441584049451103,
      "grad_norm": 1.4796979272262158,
      "learning_rate": 8.943627462717132e-06,
      "loss": 0.8767,
      "step": 3195
    },
    {
      "epoch": 0.23448921008465015,
      "grad_norm": 1.4809824515962575,
      "learning_rate": 8.942896912596422e-06,
      "loss": 0.792,
      "step": 3196
    },
    {
      "epoch": 0.2345625796747893,
      "grad_norm": 1.5695035377112854,
      "learning_rate": 8.942166139810558e-06,
      "loss": 0.8604,
      "step": 3197
    },
    {
      "epoch": 0.23463594926492842,
      "grad_norm": 0.6989863026947883,
      "learning_rate": 8.941435144400808e-06,
      "loss": 0.5357,
      "step": 3198
    },
    {
      "epoch": 0.23470931885506754,
      "grad_norm": 1.7778268939310273,
      "learning_rate": 8.940703926408456e-06,
      "loss": 1.0153,
      "step": 3199
    },
    {
      "epoch": 0.23478268844520667,
      "grad_norm": 0.6338140951229138,
      "learning_rate": 8.939972485874793e-06,
      "loss": 0.5367,
      "step": 3200
    },
    {
      "epoch": 0.23478268844520667,
      "lm_loss": 0.3809,
      "step": 3200,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23478268844520667,
      "lm_loss": 0.6625,
      "step": 3200,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23478268844520667,
      "lm_loss": 0.6516,
      "step": 3200,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23478268844520667,
      "lm_loss": 0.5469,
      "step": 3200,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23478268844520667,
      "lm_loss": 0.4019,
      "step": 3200,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23478268844520667,
      "lm_loss": 0.5072,
      "step": 3200,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23478268844520667,
      "lm_loss": 0.4779,
      "step": 3200,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23478268844520667,
      "lm_loss": 0.5302,
      "step": 3200,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2348560580353458,
      "grad_norm": 0.573640826478626,
      "learning_rate": 8.939240822841127e-06,
      "loss": 0.5402,
      "step": 3201
    },
    {
      "epoch": 0.23492942762548494,
      "grad_norm": 1.6772562959945814,
      "learning_rate": 8.938508937348778e-06,
      "loss": 0.7946,
      "step": 3202
    },
    {
      "epoch": 0.23500279721562406,
      "grad_norm": 1.5946523498519778,
      "learning_rate": 8.937776829439074e-06,
      "loss": 0.8386,
      "step": 3203
    },
    {
      "epoch": 0.23507616680576318,
      "grad_norm": 1.5510541212691775,
      "learning_rate": 8.937044499153361e-06,
      "loss": 0.8373,
      "step": 3204
    },
    {
      "epoch": 0.2351495363959023,
      "grad_norm": 1.5586650128645783,
      "learning_rate": 8.936311946532996e-06,
      "loss": 0.8617,
      "step": 3205
    },
    {
      "epoch": 0.23522290598604143,
      "grad_norm": 1.6548514441554811,
      "learning_rate": 8.935579171619347e-06,
      "loss": 0.7785,
      "step": 3206
    },
    {
      "epoch": 0.23529627557618057,
      "grad_norm": 1.5721377284430236,
      "learning_rate": 8.934846174453795e-06,
      "loss": 0.8277,
      "step": 3207
    },
    {
      "epoch": 0.2353696451663197,
      "grad_norm": 0.8393834541763929,
      "learning_rate": 8.934112955077737e-06,
      "loss": 0.5507,
      "step": 3208
    },
    {
      "epoch": 0.2353696451663197,
      "lm_loss": 0.5831,
      "step": 3208,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.2353696451663197,
      "lm_loss": 1.3953,
      "step": 3208,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.2353696451663197,
      "lm_loss": 0.6782,
      "step": 3208,
      "vm_loss": 0.105
    },
    {
      "epoch": 0.2353696451663197,
      "lm_loss": 0.7618,
      "step": 3208,
      "vm_loss": 0.1754
    },
    {
      "epoch": 0.2353696451663197,
      "lm_loss": 0.9896,
      "step": 3208,
      "vm_loss": 0.175
    },
    {
      "epoch": 0.2353696451663197,
      "lm_loss": 0.2091,
      "step": 3208,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.2353696451663197,
      "lm_loss": 0.9857,
      "step": 3208,
      "vm_loss": 0.2251
    },
    {
      "epoch": 0.2353696451663197,
      "lm_loss": 1.0597,
      "step": 3208,
      "vm_loss": 0.182
    },
    {
      "epoch": 0.23544301475645882,
      "grad_norm": 1.6693470021312469,
      "learning_rate": 8.933379513532575e-06,
      "loss": 0.945,
      "step": 3209
    },
    {
      "epoch": 0.23551638434659794,
      "grad_norm": 2.0243006532641203,
      "learning_rate": 8.932645849859733e-06,
      "loss": 0.9061,
      "step": 3210
    },
    {
      "epoch": 0.23558975393673706,
      "grad_norm": 1.6848076163221226,
      "learning_rate": 8.931911964100639e-06,
      "loss": 0.8137,
      "step": 3211
    },
    {
      "epoch": 0.2356631235268762,
      "grad_norm": 1.5493224233587397,
      "learning_rate": 8.93117785629674e-06,
      "loss": 0.8413,
      "step": 3212
    },
    {
      "epoch": 0.23573649311701533,
      "grad_norm": 1.6428169121187681,
      "learning_rate": 8.930443526489494e-06,
      "loss": 0.824,
      "step": 3213
    },
    {
      "epoch": 0.23580986270715446,
      "grad_norm": 0.6327068682803367,
      "learning_rate": 8.929708974720366e-06,
      "loss": 0.5259,
      "step": 3214
    },
    {
      "epoch": 0.23588323229729358,
      "grad_norm": 0.6147529547375007,
      "learning_rate": 8.928974201030842e-06,
      "loss": 0.5134,
      "step": 3215
    },
    {
      "epoch": 0.2359566018874327,
      "grad_norm": 1.724377007670362,
      "learning_rate": 8.928239205462414e-06,
      "loss": 0.8706,
      "step": 3216
    },
    {
      "epoch": 0.2359566018874327,
      "lm_loss": 0.523,
      "step": 3216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2359566018874327,
      "lm_loss": 0.4959,
      "step": 3216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2359566018874327,
      "lm_loss": 0.5006,
      "step": 3216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2359566018874327,
      "lm_loss": 0.5631,
      "step": 3216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2359566018874327,
      "lm_loss": 0.5191,
      "step": 3216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2359566018874327,
      "lm_loss": 0.5118,
      "step": 3216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2359566018874327,
      "lm_loss": 0.5742,
      "step": 3216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2359566018874327,
      "lm_loss": 0.662,
      "step": 3216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23602997147757182,
      "grad_norm": 0.5848595209527717,
      "learning_rate": 8.92750398805659e-06,
      "loss": 0.5544,
      "step": 3217
    },
    {
      "epoch": 0.23610334106771097,
      "grad_norm": 1.7803190905935613,
      "learning_rate": 8.92676854885489e-06,
      "loss": 0.8697,
      "step": 3218
    },
    {
      "epoch": 0.2361767106578501,
      "grad_norm": 1.552497800649197,
      "learning_rate": 8.926032887898846e-06,
      "loss": 0.8276,
      "step": 3219
    },
    {
      "epoch": 0.2362500802479892,
      "grad_norm": 0.6711580400059134,
      "learning_rate": 8.92529700523e-06,
      "loss": 0.5277,
      "step": 3220
    },
    {
      "epoch": 0.23632344983812834,
      "grad_norm": 1.6832034888056788,
      "learning_rate": 8.924560900889914e-06,
      "loss": 0.7423,
      "step": 3221
    },
    {
      "epoch": 0.23639681942826746,
      "grad_norm": 1.4217793756780135,
      "learning_rate": 8.923824574920152e-06,
      "loss": 0.9209,
      "step": 3222
    },
    {
      "epoch": 0.2364701890184066,
      "grad_norm": 1.4378282853703936,
      "learning_rate": 8.923088027362301e-06,
      "loss": 0.9052,
      "step": 3223
    },
    {
      "epoch": 0.23654355860854573,
      "grad_norm": 1.7156401690854064,
      "learning_rate": 8.922351258257952e-06,
      "loss": 0.8386,
      "step": 3224
    },
    {
      "epoch": 0.23654355860854573,
      "lm_loss": 1.5061,
      "step": 3224,
      "vm_loss": 0.0782
    },
    {
      "epoch": 0.23654355860854573,
      "lm_loss": 0.6396,
      "step": 3224,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.23654355860854573,
      "lm_loss": 0.9601,
      "step": 3224,
      "vm_loss": 0.1708
    },
    {
      "epoch": 0.23654355860854573,
      "lm_loss": 0.7696,
      "step": 3224,
      "vm_loss": 0.1698
    },
    {
      "epoch": 0.23654355860854573,
      "lm_loss": 0.3927,
      "step": 3224,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.23654355860854573,
      "lm_loss": 0.7665,
      "step": 3224,
      "vm_loss": 0.1394
    },
    {
      "epoch": 0.23654355860854573,
      "lm_loss": 0.2955,
      "step": 3224,
      "vm_loss": 0.05
    },
    {
      "epoch": 0.23654355860854573,
      "lm_loss": 1.3637,
      "step": 3224,
      "vm_loss": 0.0778
    },
    {
      "epoch": 0.23661692819868485,
      "grad_norm": 1.6144587252480205,
      "learning_rate": 8.921614267648718e-06,
      "loss": 0.813,
      "step": 3225
    },
    {
      "epoch": 0.23669029778882397,
      "grad_norm": 1.720762595823437,
      "learning_rate": 8.920877055576212e-06,
      "loss": 0.8445,
      "step": 3226
    },
    {
      "epoch": 0.2367636673789631,
      "grad_norm": 1.533932104000554,
      "learning_rate": 8.920139622082067e-06,
      "loss": 0.7845,
      "step": 3227
    },
    {
      "epoch": 0.23683703696910224,
      "grad_norm": 0.6424033295835069,
      "learning_rate": 8.919401967207933e-06,
      "loss": 0.5087,
      "step": 3228
    },
    {
      "epoch": 0.23691040655924137,
      "grad_norm": 1.6081032483270945,
      "learning_rate": 8.91866409099546e-06,
      "loss": 0.8821,
      "step": 3229
    },
    {
      "epoch": 0.2369837761493805,
      "grad_norm": 1.6990970684656235,
      "learning_rate": 8.917925993486324e-06,
      "loss": 0.844,
      "step": 3230
    },
    {
      "epoch": 0.2370571457395196,
      "grad_norm": 1.5881973694396398,
      "learning_rate": 8.917187674722203e-06,
      "loss": 0.8079,
      "step": 3231
    },
    {
      "epoch": 0.23713051532965873,
      "grad_norm": 1.7360149946883507,
      "learning_rate": 8.916449134744794e-06,
      "loss": 0.8696,
      "step": 3232
    },
    {
      "epoch": 0.23713051532965873,
      "lm_loss": 1.162,
      "step": 3232,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.23713051532965873,
      "lm_loss": 1.3994,
      "step": 3232,
      "vm_loss": 0.0673
    },
    {
      "epoch": 0.23713051532965873,
      "lm_loss": 0.297,
      "step": 3232,
      "vm_loss": 0.097
    },
    {
      "epoch": 0.23713051532965873,
      "lm_loss": 0.0845,
      "step": 3232,
      "vm_loss": 0.0959
    },
    {
      "epoch": 0.23713051532965873,
      "lm_loss": 0.8475,
      "step": 3232,
      "vm_loss": 0.0955
    },
    {
      "epoch": 0.23713051532965873,
      "lm_loss": 0.8695,
      "step": 3232,
      "vm_loss": 0.1919
    },
    {
      "epoch": 0.23713051532965873,
      "lm_loss": 1.0474,
      "step": 3232,
      "vm_loss": 0.23
    },
    {
      "epoch": 0.23713051532965873,
      "lm_loss": 0.7462,
      "step": 3232,
      "vm_loss": 0.1281
    },
    {
      "epoch": 0.23720388491979788,
      "grad_norm": 1.3518437042148599,
      "learning_rate": 8.915710373595803e-06,
      "loss": 0.7581,
      "step": 3233
    },
    {
      "epoch": 0.237277254509937,
      "grad_norm": 0.6261287084944015,
      "learning_rate": 8.91497139131695e-06,
      "loss": 0.5421,
      "step": 3234
    },
    {
      "epoch": 0.23735062410007612,
      "grad_norm": 1.4782692332343783,
      "learning_rate": 8.91423218794997e-06,
      "loss": 0.8337,
      "step": 3235
    },
    {
      "epoch": 0.23742399369021525,
      "grad_norm": 1.3539912118742807,
      "learning_rate": 8.913492763536602e-06,
      "loss": 0.8556,
      "step": 3236
    },
    {
      "epoch": 0.23749736328035437,
      "grad_norm": 1.535942011078186,
      "learning_rate": 8.912753118118607e-06,
      "loss": 0.8065,
      "step": 3237
    },
    {
      "epoch": 0.2375707328704935,
      "grad_norm": 1.6022492212597061,
      "learning_rate": 8.912013251737754e-06,
      "loss": 0.804,
      "step": 3238
    },
    {
      "epoch": 0.23764410246063264,
      "grad_norm": 0.6204400744367273,
      "learning_rate": 8.911273164435824e-06,
      "loss": 0.5588,
      "step": 3239
    },
    {
      "epoch": 0.23771747205077176,
      "grad_norm": 0.599077102994667,
      "learning_rate": 8.910532856254612e-06,
      "loss": 0.5371,
      "step": 3240
    },
    {
      "epoch": 0.23771747205077176,
      "lm_loss": 0.6444,
      "step": 3240,
      "vm_loss": 0.0251
    },
    {
      "epoch": 0.23771747205077176,
      "lm_loss": 0.84,
      "step": 3240,
      "vm_loss": 0.0835
    },
    {
      "epoch": 0.23771747205077176,
      "lm_loss": 1.0467,
      "step": 3240,
      "vm_loss": 0.1018
    },
    {
      "epoch": 0.23771747205077176,
      "lm_loss": 0.4351,
      "step": 3240,
      "vm_loss": 0.1553
    },
    {
      "epoch": 0.23771747205077176,
      "lm_loss": 1.098,
      "step": 3240,
      "vm_loss": 0.1723
    },
    {
      "epoch": 0.23771747205077176,
      "lm_loss": 0.311,
      "step": 3240,
      "vm_loss": 0.0779
    },
    {
      "epoch": 0.23771747205077176,
      "lm_loss": 0.3388,
      "step": 3240,
      "vm_loss": 0.165
    },
    {
      "epoch": 0.23771747205077176,
      "lm_loss": 0.2911,
      "step": 3240,
      "vm_loss": 0.0566
    },
    {
      "epoch": 0.23779084164091088,
      "grad_norm": 1.6766506216683628,
      "learning_rate": 8.909792327235926e-06,
      "loss": 0.7498,
      "step": 3241
    },
    {
      "epoch": 0.23786421123105,
      "grad_norm": 0.5738693263593246,
      "learning_rate": 8.909051577421586e-06,
      "loss": 0.536,
      "step": 3242
    },
    {
      "epoch": 0.23793758082118913,
      "grad_norm": 1.5379820575382182,
      "learning_rate": 8.908310606853421e-06,
      "loss": 0.8282,
      "step": 3243
    },
    {
      "epoch": 0.23801095041132828,
      "grad_norm": 1.5372692605162042,
      "learning_rate": 8.907569415573278e-06,
      "loss": 0.8258,
      "step": 3244
    },
    {
      "epoch": 0.2380843200014674,
      "grad_norm": 1.5027826423526685,
      "learning_rate": 8.906828003623013e-06,
      "loss": 0.8487,
      "step": 3245
    },
    {
      "epoch": 0.23815768959160652,
      "grad_norm": 0.6059681050030227,
      "learning_rate": 8.906086371044496e-06,
      "loss": 0.5416,
      "step": 3246
    },
    {
      "epoch": 0.23823105918174564,
      "grad_norm": 0.6198107062680961,
      "learning_rate": 8.905344517879608e-06,
      "loss": 0.5319,
      "step": 3247
    },
    {
      "epoch": 0.23830442877188476,
      "grad_norm": 0.6178040853452028,
      "learning_rate": 8.904602444170242e-06,
      "loss": 0.5536,
      "step": 3248
    },
    {
      "epoch": 0.23830442877188476,
      "lm_loss": 0.4461,
      "step": 3248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23830442877188476,
      "lm_loss": 0.546,
      "step": 3248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23830442877188476,
      "lm_loss": 0.5305,
      "step": 3248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23830442877188476,
      "lm_loss": 0.3325,
      "step": 3248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23830442877188476,
      "lm_loss": 0.6702,
      "step": 3248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23830442877188476,
      "lm_loss": 0.5363,
      "step": 3248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23830442877188476,
      "lm_loss": 0.5069,
      "step": 3248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.23830442877188476,
      "lm_loss": 0.653,
      "step": 3248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2383777983620239,
      "grad_norm": 0.591760931209986,
      "learning_rate": 8.903860149958308e-06,
      "loss": 0.5514,
      "step": 3249
    },
    {
      "epoch": 0.23845116795216303,
      "grad_norm": 0.5574477534325464,
      "learning_rate": 8.903117635285723e-06,
      "loss": 0.5231,
      "step": 3250
    },
    {
      "epoch": 0.23852453754230216,
      "grad_norm": 2.0567695300210787,
      "learning_rate": 8.90237490019442e-06,
      "loss": 0.8629,
      "step": 3251
    },
    {
      "epoch": 0.23859790713244128,
      "grad_norm": 1.405458778925607,
      "learning_rate": 8.901631944726344e-06,
      "loss": 0.7478,
      "step": 3252
    },
    {
      "epoch": 0.2386712767225804,
      "grad_norm": 1.6861478376287145,
      "learning_rate": 8.900888768923445e-06,
      "loss": 0.8755,
      "step": 3253
    },
    {
      "epoch": 0.23874464631271952,
      "grad_norm": 1.5147707791219174,
      "learning_rate": 8.9001453728277e-06,
      "loss": 0.921,
      "step": 3254
    },
    {
      "epoch": 0.23881801590285867,
      "grad_norm": 1.8043450932050848,
      "learning_rate": 8.899401756481087e-06,
      "loss": 0.8036,
      "step": 3255
    },
    {
      "epoch": 0.2388913854929978,
      "grad_norm": 1.607485038502908,
      "learning_rate": 8.898657919925602e-06,
      "loss": 0.7964,
      "step": 3256
    },
    {
      "epoch": 0.2388913854929978,
      "lm_loss": 1.6342,
      "step": 3256,
      "vm_loss": 0.0466
    },
    {
      "epoch": 0.2388913854929978,
      "lm_loss": 1.2816,
      "step": 3256,
      "vm_loss": 0.1755
    },
    {
      "epoch": 0.2388913854929978,
      "lm_loss": 0.8188,
      "step": 3256,
      "vm_loss": 0.1931
    },
    {
      "epoch": 0.2388913854929978,
      "lm_loss": 0.1632,
      "step": 3256,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.2388913854929978,
      "lm_loss": 0.5097,
      "step": 3256,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.2388913854929978,
      "lm_loss": 0.8033,
      "step": 3256,
      "vm_loss": 0.1509
    },
    {
      "epoch": 0.2388913854929978,
      "lm_loss": 0.0513,
      "step": 3256,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.2388913854929978,
      "lm_loss": 1.3688,
      "step": 3256,
      "vm_loss": 0.1632
    },
    {
      "epoch": 0.23896475508313691,
      "grad_norm": 1.5731242534442014,
      "learning_rate": 8.897913863203247e-06,
      "loss": 0.9186,
      "step": 3257
    },
    {
      "epoch": 0.23903812467327604,
      "grad_norm": 1.5690183374472897,
      "learning_rate": 8.897169586356042e-06,
      "loss": 0.9103,
      "step": 3258
    },
    {
      "epoch": 0.23911149426341516,
      "grad_norm": 1.46271487475917,
      "learning_rate": 8.896425089426022e-06,
      "loss": 0.856,
      "step": 3259
    },
    {
      "epoch": 0.2391848638535543,
      "grad_norm": 1.4283506431360167,
      "learning_rate": 8.895680372455226e-06,
      "loss": 0.8065,
      "step": 3260
    },
    {
      "epoch": 0.23925823344369343,
      "grad_norm": 1.4884946143393174,
      "learning_rate": 8.894935435485712e-06,
      "loss": 0.7938,
      "step": 3261
    },
    {
      "epoch": 0.23933160303383255,
      "grad_norm": 0.8882545551023953,
      "learning_rate": 8.89419027855955e-06,
      "loss": 0.5558,
      "step": 3262
    },
    {
      "epoch": 0.23940497262397167,
      "grad_norm": 1.5741414586775064,
      "learning_rate": 8.893444901718818e-06,
      "loss": 0.8042,
      "step": 3263
    },
    {
      "epoch": 0.2394783422141108,
      "grad_norm": 0.6754020273771342,
      "learning_rate": 8.892699305005612e-06,
      "loss": 0.5336,
      "step": 3264
    },
    {
      "epoch": 0.2394783422141108,
      "lm_loss": 1.0257,
      "step": 3264,
      "vm_loss": 0.0671
    },
    {
      "epoch": 0.2394783422141108,
      "lm_loss": 0.3444,
      "step": 3264,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.2394783422141108,
      "lm_loss": 2.0753,
      "step": 3264,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.2394783422141108,
      "lm_loss": 0.3026,
      "step": 3264,
      "vm_loss": 0.1826
    },
    {
      "epoch": 0.2394783422141108,
      "lm_loss": 0.624,
      "step": 3264,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.2394783422141108,
      "lm_loss": 0.0964,
      "step": 3264,
      "vm_loss": 0.0818
    },
    {
      "epoch": 0.2394783422141108,
      "lm_loss": 1.0292,
      "step": 3264,
      "vm_loss": 0.045
    },
    {
      "epoch": 0.2394783422141108,
      "lm_loss": 0.2534,
      "step": 3264,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.23955171180424994,
      "grad_norm": 1.6339823406014005,
      "learning_rate": 8.891953488462032e-06,
      "loss": 0.7724,
      "step": 3265
    },
    {
      "epoch": 0.23962508139438907,
      "grad_norm": 1.61707719390653,
      "learning_rate": 8.891207452130203e-06,
      "loss": 0.8373,
      "step": 3266
    },
    {
      "epoch": 0.2396984509845282,
      "grad_norm": 1.655163682886503,
      "learning_rate": 8.890461196052253e-06,
      "loss": 0.8975,
      "step": 3267
    },
    {
      "epoch": 0.2397718205746673,
      "grad_norm": 0.6999689965763555,
      "learning_rate": 8.889714720270325e-06,
      "loss": 0.5385,
      "step": 3268
    },
    {
      "epoch": 0.23984519016480643,
      "grad_norm": 0.6501114695068606,
      "learning_rate": 8.888968024826575e-06,
      "loss": 0.533,
      "step": 3269
    },
    {
      "epoch": 0.23991855975494558,
      "grad_norm": 1.741557832002539,
      "learning_rate": 8.888221109763166e-06,
      "loss": 0.8106,
      "step": 3270
    },
    {
      "epoch": 0.2399919293450847,
      "grad_norm": 0.5911848004378502,
      "learning_rate": 8.887473975122285e-06,
      "loss": 0.5432,
      "step": 3271
    },
    {
      "epoch": 0.24006529893522383,
      "grad_norm": 0.6080541005880873,
      "learning_rate": 8.88672662094612e-06,
      "loss": 0.5471,
      "step": 3272
    },
    {
      "epoch": 0.24006529893522383,
      "lm_loss": 0.6115,
      "step": 3272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24006529893522383,
      "lm_loss": 0.4506,
      "step": 3272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24006529893522383,
      "lm_loss": 0.4867,
      "step": 3272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24006529893522383,
      "lm_loss": 0.5542,
      "step": 3272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24006529893522383,
      "lm_loss": 0.531,
      "step": 3272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24006529893522383,
      "lm_loss": 0.5728,
      "step": 3272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24006529893522383,
      "lm_loss": 0.5808,
      "step": 3272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24006529893522383,
      "lm_loss": 0.5271,
      "step": 3272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24013866852536295,
      "grad_norm": 0.6010648300851219,
      "learning_rate": 8.885979047276878e-06,
      "loss": 0.5406,
      "step": 3273
    },
    {
      "epoch": 0.24021203811550207,
      "grad_norm": 1.7945005427166787,
      "learning_rate": 8.885231254156777e-06,
      "loss": 0.8183,
      "step": 3274
    },
    {
      "epoch": 0.2402854077056412,
      "grad_norm": 0.6299867063889789,
      "learning_rate": 8.884483241628043e-06,
      "loss": 0.5327,
      "step": 3275
    },
    {
      "epoch": 0.24035877729578034,
      "grad_norm": 1.9102647717153376,
      "learning_rate": 8.883735009732922e-06,
      "loss": 0.7386,
      "step": 3276
    },
    {
      "epoch": 0.24043214688591946,
      "grad_norm": 1.5564175679027143,
      "learning_rate": 8.882986558513667e-06,
      "loss": 1.0305,
      "step": 3277
    },
    {
      "epoch": 0.24050551647605858,
      "grad_norm": 1.602878871937813,
      "learning_rate": 8.882237888012546e-06,
      "loss": 0.9091,
      "step": 3278
    },
    {
      "epoch": 0.2405788860661977,
      "grad_norm": 1.661652975150312,
      "learning_rate": 8.881488998271834e-06,
      "loss": 0.8499,
      "step": 3279
    },
    {
      "epoch": 0.24065225565633683,
      "grad_norm": 1.6056368910990992,
      "learning_rate": 8.88073988933383e-06,
      "loss": 0.8555,
      "step": 3280
    },
    {
      "epoch": 0.24065225565633683,
      "lm_loss": 2.0653,
      "step": 3280,
      "vm_loss": 0.0545
    },
    {
      "epoch": 0.24065225565633683,
      "lm_loss": 1.2879,
      "step": 3280,
      "vm_loss": 0.1974
    },
    {
      "epoch": 0.24065225565633683,
      "lm_loss": 1.2485,
      "step": 3280,
      "vm_loss": 0.1803
    },
    {
      "epoch": 0.24065225565633683,
      "lm_loss": 0.2879,
      "step": 3280,
      "vm_loss": 0.1507
    },
    {
      "epoch": 0.24065225565633683,
      "lm_loss": 1.127,
      "step": 3280,
      "vm_loss": 0.2095
    },
    {
      "epoch": 0.24065225565633683,
      "lm_loss": 0.9475,
      "step": 3280,
      "vm_loss": 0.2319
    },
    {
      "epoch": 0.24065225565633683,
      "lm_loss": 0.422,
      "step": 3280,
      "vm_loss": 0.0886
    },
    {
      "epoch": 0.24065225565633683,
      "lm_loss": 0.6214,
      "step": 3280,
      "vm_loss": 0.1439
    },
    {
      "epoch": 0.24072562524647598,
      "grad_norm": 1.4034970058357452,
      "learning_rate": 8.879990561240831e-06,
      "loss": 0.8572,
      "step": 3281
    },
    {
      "epoch": 0.2407989948366151,
      "grad_norm": 1.476543762754883,
      "learning_rate": 8.879241014035158e-06,
      "loss": 0.8272,
      "step": 3282
    },
    {
      "epoch": 0.24087236442675422,
      "grad_norm": 1.4881982250039334,
      "learning_rate": 8.878491247759138e-06,
      "loss": 0.8866,
      "step": 3283
    },
    {
      "epoch": 0.24094573401689334,
      "grad_norm": 1.607053551134281,
      "learning_rate": 8.87774126245511e-06,
      "loss": 0.8692,
      "step": 3284
    },
    {
      "epoch": 0.24101910360703246,
      "grad_norm": 1.4993015800173666,
      "learning_rate": 8.876991058165433e-06,
      "loss": 0.844,
      "step": 3285
    },
    {
      "epoch": 0.24109247319717161,
      "grad_norm": 1.4658334754151805,
      "learning_rate": 8.876240634932468e-06,
      "loss": 0.8328,
      "step": 3286
    },
    {
      "epoch": 0.24116584278731074,
      "grad_norm": 1.453751708867199,
      "learning_rate": 8.875489992798596e-06,
      "loss": 0.8176,
      "step": 3287
    },
    {
      "epoch": 0.24123921237744986,
      "grad_norm": 1.7655907474524304,
      "learning_rate": 8.874739131806206e-06,
      "loss": 0.8629,
      "step": 3288
    },
    {
      "epoch": 0.24123921237744986,
      "lm_loss": 1.8733,
      "step": 3288,
      "vm_loss": 0.0622
    },
    {
      "epoch": 0.24123921237744986,
      "lm_loss": 1.2331,
      "step": 3288,
      "vm_loss": 0.1527
    },
    {
      "epoch": 0.24123921237744986,
      "lm_loss": 0.9526,
      "step": 3288,
      "vm_loss": 0.0881
    },
    {
      "epoch": 0.24123921237744986,
      "lm_loss": 0.2904,
      "step": 3288,
      "vm_loss": 0.1048
    },
    {
      "epoch": 0.24123921237744986,
      "lm_loss": 0.857,
      "step": 3288,
      "vm_loss": 0.2219
    },
    {
      "epoch": 0.24123921237744986,
      "lm_loss": 0.8304,
      "step": 3288,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.24123921237744986,
      "lm_loss": 1.3166,
      "step": 3288,
      "vm_loss": 0.1664
    },
    {
      "epoch": 0.24123921237744986,
      "lm_loss": 1.281,
      "step": 3288,
      "vm_loss": 0.0572
    },
    {
      "epoch": 0.24131258196758898,
      "grad_norm": 1.4600100295605194,
      "learning_rate": 8.873988051997702e-06,
      "loss": 0.801,
      "step": 3289
    },
    {
      "epoch": 0.2413859515577281,
      "grad_norm": 0.7379425982382524,
      "learning_rate": 8.873236753415497e-06,
      "loss": 0.5442,
      "step": 3290
    },
    {
      "epoch": 0.24145932114786725,
      "grad_norm": 1.5545015827832431,
      "learning_rate": 8.872485236102024e-06,
      "loss": 0.8656,
      "step": 3291
    },
    {
      "epoch": 0.24153269073800637,
      "grad_norm": 0.6118159398579943,
      "learning_rate": 8.871733500099718e-06,
      "loss": 0.5457,
      "step": 3292
    },
    {
      "epoch": 0.2416060603281455,
      "grad_norm": 1.5766511522758049,
      "learning_rate": 8.870981545451032e-06,
      "loss": 0.8499,
      "step": 3293
    },
    {
      "epoch": 0.24167942991828462,
      "grad_norm": 1.5439286614926038,
      "learning_rate": 8.870229372198431e-06,
      "loss": 0.793,
      "step": 3294
    },
    {
      "epoch": 0.24175279950842374,
      "grad_norm": 1.6247172844736142,
      "learning_rate": 8.869476980384397e-06,
      "loss": 0.8588,
      "step": 3295
    },
    {
      "epoch": 0.24182616909856286,
      "grad_norm": 1.473190663813444,
      "learning_rate": 8.868724370051411e-06,
      "loss": 0.7884,
      "step": 3296
    },
    {
      "epoch": 0.24182616909856286,
      "lm_loss": 1.3453,
      "step": 3296,
      "vm_loss": 0.0774
    },
    {
      "epoch": 0.24182616909856286,
      "lm_loss": 1.2894,
      "step": 3296,
      "vm_loss": 0.2062
    },
    {
      "epoch": 0.24182616909856286,
      "lm_loss": 0.4416,
      "step": 3296,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.24182616909856286,
      "lm_loss": 1.148,
      "step": 3296,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.24182616909856286,
      "lm_loss": 0.5993,
      "step": 3296,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.24182616909856286,
      "lm_loss": 0.5427,
      "step": 3296,
      "vm_loss": 0.1128
    },
    {
      "epoch": 0.24182616909856286,
      "lm_loss": 0.7234,
      "step": 3296,
      "vm_loss": 0.1027
    },
    {
      "epoch": 0.24182616909856286,
      "lm_loss": 1.0637,
      "step": 3296,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.241899538688702,
      "grad_norm": 1.7208786785196108,
      "learning_rate": 8.867971541241983e-06,
      "loss": 0.8724,
      "step": 3297
    },
    {
      "epoch": 0.24197290827884113,
      "grad_norm": 1.459516839706139,
      "learning_rate": 8.86721849399862e-06,
      "loss": 0.8184,
      "step": 3298
    },
    {
      "epoch": 0.24204627786898025,
      "grad_norm": 1.683976039813056,
      "learning_rate": 8.866465228363853e-06,
      "loss": 0.9109,
      "step": 3299
    },
    {
      "epoch": 0.24211964745911937,
      "grad_norm": 1.5308367252976254,
      "learning_rate": 8.865711744380219e-06,
      "loss": 0.8541,
      "step": 3300
    },
    {
      "epoch": 0.2421930170492585,
      "grad_norm": 1.6140642217053118,
      "learning_rate": 8.86495804209027e-06,
      "loss": 0.8344,
      "step": 3301
    },
    {
      "epoch": 0.24226638663939765,
      "grad_norm": 1.5340241896806734,
      "learning_rate": 8.864204121536568e-06,
      "loss": 0.7802,
      "step": 3302
    },
    {
      "epoch": 0.24233975622953677,
      "grad_norm": 1.3749783588617595,
      "learning_rate": 8.863449982761691e-06,
      "loss": 0.7866,
      "step": 3303
    },
    {
      "epoch": 0.2424131258196759,
      "grad_norm": 1.589168128055083,
      "learning_rate": 8.862695625808225e-06,
      "loss": 0.9018,
      "step": 3304
    },
    {
      "epoch": 0.2424131258196759,
      "lm_loss": 0.4176,
      "step": 3304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2424131258196759,
      "lm_loss": 0.4645,
      "step": 3304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2424131258196759,
      "lm_loss": 0.553,
      "step": 3304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2424131258196759,
      "lm_loss": 0.5776,
      "step": 3304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2424131258196759,
      "lm_loss": 0.5734,
      "step": 3304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2424131258196759,
      "lm_loss": 0.4847,
      "step": 3304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2424131258196759,
      "lm_loss": 0.6764,
      "step": 3304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2424131258196759,
      "lm_loss": 0.4743,
      "step": 3304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.242486495409815,
      "grad_norm": 0.7432719703014041,
      "learning_rate": 8.861941050718772e-06,
      "loss": 0.5482,
      "step": 3305
    },
    {
      "epoch": 0.24255986499995413,
      "grad_norm": 1.6285301372577872,
      "learning_rate": 8.861186257535944e-06,
      "loss": 0.7814,
      "step": 3306
    },
    {
      "epoch": 0.24263323459009328,
      "grad_norm": 0.6077958075102929,
      "learning_rate": 8.860431246302365e-06,
      "loss": 0.5428,
      "step": 3307
    },
    {
      "epoch": 0.2427066041802324,
      "grad_norm": 1.417749612762695,
      "learning_rate": 8.859676017060674e-06,
      "loss": 0.8016,
      "step": 3308
    },
    {
      "epoch": 0.24277997377037153,
      "grad_norm": 0.5862737153277353,
      "learning_rate": 8.85892056985352e-06,
      "loss": 0.5465,
      "step": 3309
    },
    {
      "epoch": 0.24285334336051065,
      "grad_norm": 1.6850591782066848,
      "learning_rate": 8.858164904723564e-06,
      "loss": 0.8652,
      "step": 3310
    },
    {
      "epoch": 0.24292671295064977,
      "grad_norm": 1.5160284432454467,
      "learning_rate": 8.857409021713482e-06,
      "loss": 0.7685,
      "step": 3311
    },
    {
      "epoch": 0.24300008254078892,
      "grad_norm": 1.5023595432666221,
      "learning_rate": 8.85665292086596e-06,
      "loss": 0.8222,
      "step": 3312
    },
    {
      "epoch": 0.24300008254078892,
      "lm_loss": 1.3214,
      "step": 3312,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.24300008254078892,
      "lm_loss": 0.6773,
      "step": 3312,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.24300008254078892,
      "lm_loss": 0.0614,
      "step": 3312,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.24300008254078892,
      "lm_loss": 0.833,
      "step": 3312,
      "vm_loss": 0.0777
    },
    {
      "epoch": 0.24300008254078892,
      "lm_loss": 0.3978,
      "step": 3312,
      "vm_loss": 0.064
    },
    {
      "epoch": 0.24300008254078892,
      "lm_loss": 1.8474,
      "step": 3312,
      "vm_loss": 0.1775
    },
    {
      "epoch": 0.24300008254078892,
      "lm_loss": 0.0443,
      "step": 3312,
      "vm_loss": 0.0632
    },
    {
      "epoch": 0.24300008254078892,
      "lm_loss": 0.179,
      "step": 3312,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.24307345213092804,
      "grad_norm": 1.4592373801177545,
      "learning_rate": 8.855896602223697e-06,
      "loss": 0.7538,
      "step": 3313
    },
    {
      "epoch": 0.24314682172106716,
      "grad_norm": 1.4886894756054945,
      "learning_rate": 8.855140065829402e-06,
      "loss": 0.7989,
      "step": 3314
    },
    {
      "epoch": 0.24322019131120629,
      "grad_norm": 1.7296198451906775,
      "learning_rate": 8.854383311725802e-06,
      "loss": 0.8544,
      "step": 3315
    },
    {
      "epoch": 0.2432935609013454,
      "grad_norm": 1.7232723306990911,
      "learning_rate": 8.85362633995563e-06,
      "loss": 0.9033,
      "step": 3316
    },
    {
      "epoch": 0.24336693049148453,
      "grad_norm": 1.7226227117406259,
      "learning_rate": 8.852869150561638e-06,
      "loss": 0.7566,
      "step": 3317
    },
    {
      "epoch": 0.24344030008162368,
      "grad_norm": 1.5847098891884608,
      "learning_rate": 8.85211174358658e-06,
      "loss": 0.8981,
      "step": 3318
    },
    {
      "epoch": 0.2435136696717628,
      "grad_norm": 1.5147943440471687,
      "learning_rate": 8.851354119073234e-06,
      "loss": 0.8759,
      "step": 3319
    },
    {
      "epoch": 0.24358703926190192,
      "grad_norm": 0.8347764325056065,
      "learning_rate": 8.850596277064384e-06,
      "loss": 0.5357,
      "step": 3320
    },
    {
      "epoch": 0.24358703926190192,
      "lm_loss": 0.5737,
      "step": 3320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24358703926190192,
      "lm_loss": 0.5158,
      "step": 3320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24358703926190192,
      "lm_loss": 0.4211,
      "step": 3320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24358703926190192,
      "lm_loss": 0.6459,
      "step": 3320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24358703926190192,
      "lm_loss": 0.4036,
      "step": 3320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24358703926190192,
      "lm_loss": 0.5016,
      "step": 3320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24358703926190192,
      "lm_loss": 0.4779,
      "step": 3320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24358703926190192,
      "lm_loss": 0.6455,
      "step": 3320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24366040885204104,
      "grad_norm": 0.7295844314402367,
      "learning_rate": 8.849838217602825e-06,
      "loss": 0.537,
      "step": 3321
    },
    {
      "epoch": 0.24373377844218017,
      "grad_norm": 1.7690317085900358,
      "learning_rate": 8.849079940731368e-06,
      "loss": 0.9092,
      "step": 3322
    },
    {
      "epoch": 0.24380714803231932,
      "grad_norm": 1.6572054733595625,
      "learning_rate": 8.848321446492836e-06,
      "loss": 0.8323,
      "step": 3323
    },
    {
      "epoch": 0.24388051762245844,
      "grad_norm": 1.4546599073300583,
      "learning_rate": 8.84756273493006e-06,
      "loss": 0.864,
      "step": 3324
    },
    {
      "epoch": 0.24395388721259756,
      "grad_norm": 1.476076168061029,
      "learning_rate": 8.84680380608589e-06,
      "loss": 0.8382,
      "step": 3325
    },
    {
      "epoch": 0.24402725680273668,
      "grad_norm": 1.5999945517805203,
      "learning_rate": 8.846044660003182e-06,
      "loss": 0.8487,
      "step": 3326
    },
    {
      "epoch": 0.2441006263928758,
      "grad_norm": 1.6270500202631615,
      "learning_rate": 8.845285296724807e-06,
      "loss": 0.8128,
      "step": 3327
    },
    {
      "epoch": 0.24417399598301495,
      "grad_norm": 1.7257588914246713,
      "learning_rate": 8.844525716293649e-06,
      "loss": 0.8655,
      "step": 3328
    },
    {
      "epoch": 0.24417399598301495,
      "lm_loss": 1.3648,
      "step": 3328,
      "vm_loss": 0.1156
    },
    {
      "epoch": 0.24417399598301495,
      "lm_loss": 1.6727,
      "step": 3328,
      "vm_loss": 0.1797
    },
    {
      "epoch": 0.24417399598301495,
      "lm_loss": 0.524,
      "step": 3328,
      "vm_loss": 0.1015
    },
    {
      "epoch": 0.24417399598301495,
      "lm_loss": 0.3221,
      "step": 3328,
      "vm_loss": 0.1391
    },
    {
      "epoch": 0.24417399598301495,
      "lm_loss": 0.3995,
      "step": 3328,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.24417399598301495,
      "lm_loss": 0.6074,
      "step": 3328,
      "vm_loss": 0.1858
    },
    {
      "epoch": 0.24417399598301495,
      "lm_loss": 1.0885,
      "step": 3328,
      "vm_loss": 0.1701
    },
    {
      "epoch": 0.24417399598301495,
      "lm_loss": 1.2354,
      "step": 3328,
      "vm_loss": 0.1583
    },
    {
      "epoch": 0.24424736557315407,
      "grad_norm": 1.6406929123873195,
      "learning_rate": 8.8437659187526e-06,
      "loss": 0.8795,
      "step": 3329
    },
    {
      "epoch": 0.2443207351632932,
      "grad_norm": 1.583494439474213,
      "learning_rate": 8.843005904144574e-06,
      "loss": 0.8467,
      "step": 3330
    },
    {
      "epoch": 0.24439410475343232,
      "grad_norm": 1.1343383243303953,
      "learning_rate": 8.842245672512487e-06,
      "loss": 0.5448,
      "step": 3331
    },
    {
      "epoch": 0.24446747434357144,
      "grad_norm": 0.9966021488866013,
      "learning_rate": 8.84148522389927e-06,
      "loss": 0.5444,
      "step": 3332
    },
    {
      "epoch": 0.2445408439337106,
      "grad_norm": 1.733761437565492,
      "learning_rate": 8.840724558347872e-06,
      "loss": 0.8105,
      "step": 3333
    },
    {
      "epoch": 0.2446142135238497,
      "grad_norm": 0.6736994456818718,
      "learning_rate": 8.839963675901247e-06,
      "loss": 0.5503,
      "step": 3334
    },
    {
      "epoch": 0.24468758311398883,
      "grad_norm": 1.6726667424024915,
      "learning_rate": 8.839202576602362e-06,
      "loss": 0.7905,
      "step": 3335
    },
    {
      "epoch": 0.24476095270412795,
      "grad_norm": 1.0214751761360747,
      "learning_rate": 8.838441260494199e-06,
      "loss": 0.5482,
      "step": 3336
    },
    {
      "epoch": 0.24476095270412795,
      "lm_loss": 1.5831,
      "step": 3336,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.24476095270412795,
      "lm_loss": 0.2917,
      "step": 3336,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.24476095270412795,
      "lm_loss": 0.3741,
      "step": 3336,
      "vm_loss": 0.0757
    },
    {
      "epoch": 0.24476095270412795,
      "lm_loss": 0.0275,
      "step": 3336,
      "vm_loss": 0.1177
    },
    {
      "epoch": 0.24476095270412795,
      "lm_loss": 0.6904,
      "step": 3336,
      "vm_loss": 0.1926
    },
    {
      "epoch": 0.24476095270412795,
      "lm_loss": 0.0939,
      "step": 3336,
      "vm_loss": 0.0924
    },
    {
      "epoch": 0.24476095270412795,
      "lm_loss": 0.5233,
      "step": 3336,
      "vm_loss": 0.1619
    },
    {
      "epoch": 0.24476095270412795,
      "lm_loss": 0.0928,
      "step": 3336,
      "vm_loss": 0.1376
    },
    {
      "epoch": 0.24483432229426708,
      "grad_norm": 1.7735810789191462,
      "learning_rate": 8.837679727619753e-06,
      "loss": 0.903,
      "step": 3337
    },
    {
      "epoch": 0.2449076918844062,
      "grad_norm": 1.5582600061709502,
      "learning_rate": 8.836917978022029e-06,
      "loss": 0.8616,
      "step": 3338
    },
    {
      "epoch": 0.24498106147454535,
      "grad_norm": 1.5539170126348707,
      "learning_rate": 8.836156011744046e-06,
      "loss": 0.8626,
      "step": 3339
    },
    {
      "epoch": 0.24505443106468447,
      "grad_norm": 0.8057385916651754,
      "learning_rate": 8.835393828828831e-06,
      "loss": 0.5375,
      "step": 3340
    },
    {
      "epoch": 0.2451278006548236,
      "grad_norm": 1.8743916566356815,
      "learning_rate": 8.834631429319428e-06,
      "loss": 0.8668,
      "step": 3341
    },
    {
      "epoch": 0.2452011702449627,
      "grad_norm": 1.6187962846393562,
      "learning_rate": 8.833868813258895e-06,
      "loss": 0.8635,
      "step": 3342
    },
    {
      "epoch": 0.24527453983510183,
      "grad_norm": 1.6853839792186114,
      "learning_rate": 8.833105980690293e-06,
      "loss": 0.8122,
      "step": 3343
    },
    {
      "epoch": 0.24534790942524098,
      "grad_norm": 1.6870616471984656,
      "learning_rate": 8.832342931656704e-06,
      "loss": 0.8373,
      "step": 3344
    },
    {
      "epoch": 0.24534790942524098,
      "lm_loss": 1.2128,
      "step": 3344,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.24534790942524098,
      "lm_loss": 0.6021,
      "step": 3344,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.24534790942524098,
      "lm_loss": 0.1333,
      "step": 3344,
      "vm_loss": 0.051
    },
    {
      "epoch": 0.24534790942524098,
      "lm_loss": 0.4864,
      "step": 3344,
      "vm_loss": 0.2001
    },
    {
      "epoch": 0.24534790942524098,
      "lm_loss": 0.7546,
      "step": 3344,
      "vm_loss": 0.156
    },
    {
      "epoch": 0.24534790942524098,
      "lm_loss": 0.1017,
      "step": 3344,
      "vm_loss": 0.0748
    },
    {
      "epoch": 0.24534790942524098,
      "lm_loss": 1.5452,
      "step": 3344,
      "vm_loss": 0.1408
    },
    {
      "epoch": 0.24534790942524098,
      "lm_loss": 0.9448,
      "step": 3344,
      "vm_loss": 0.1189
    },
    {
      "epoch": 0.2454212790153801,
      "grad_norm": 1.5059138679536168,
      "learning_rate": 8.83157966620122e-06,
      "loss": 0.7953,
      "step": 3345
    },
    {
      "epoch": 0.24549464860551923,
      "grad_norm": 1.6240978065766976,
      "learning_rate": 8.830816184366944e-06,
      "loss": 0.8443,
      "step": 3346
    },
    {
      "epoch": 0.24556801819565835,
      "grad_norm": 0.7220229761027529,
      "learning_rate": 8.830052486196989e-06,
      "loss": 0.5227,
      "step": 3347
    },
    {
      "epoch": 0.24564138778579747,
      "grad_norm": 1.8428133899520198,
      "learning_rate": 8.829288571734487e-06,
      "loss": 0.799,
      "step": 3348
    },
    {
      "epoch": 0.24571475737593662,
      "grad_norm": 1.734936609251139,
      "learning_rate": 8.828524441022575e-06,
      "loss": 0.8473,
      "step": 3349
    },
    {
      "epoch": 0.24578812696607574,
      "grad_norm": 0.6553843170861249,
      "learning_rate": 8.827760094104408e-06,
      "loss": 0.5444,
      "step": 3350
    },
    {
      "epoch": 0.24586149655621486,
      "grad_norm": 1.8248650140608722,
      "learning_rate": 8.826995531023149e-06,
      "loss": 0.82,
      "step": 3351
    },
    {
      "epoch": 0.245934866146354,
      "grad_norm": 0.6615592126885822,
      "learning_rate": 8.826230751821974e-06,
      "loss": 0.5567,
      "step": 3352
    },
    {
      "epoch": 0.245934866146354,
      "lm_loss": 1.5288,
      "step": 3352,
      "vm_loss": 0.0441
    },
    {
      "epoch": 0.245934866146354,
      "lm_loss": 1.3137,
      "step": 3352,
      "vm_loss": 0.1682
    },
    {
      "epoch": 0.245934866146354,
      "lm_loss": 0.6258,
      "step": 3352,
      "vm_loss": 0.0745
    },
    {
      "epoch": 0.245934866146354,
      "lm_loss": 0.6742,
      "step": 3352,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.245934866146354,
      "lm_loss": 0.9369,
      "step": 3352,
      "vm_loss": 0.181
    },
    {
      "epoch": 0.245934866146354,
      "lm_loss": 1.1527,
      "step": 3352,
      "vm_loss": 0.2631
    },
    {
      "epoch": 0.245934866146354,
      "lm_loss": 0.1682,
      "step": 3352,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.245934866146354,
      "lm_loss": 0.4637,
      "step": 3352,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.2460082357364931,
      "grad_norm": 1.4098735389579728,
      "learning_rate": 8.825465756544073e-06,
      "loss": 0.8861,
      "step": 3353
    },
    {
      "epoch": 0.24608160532663226,
      "grad_norm": 1.6692430608044642,
      "learning_rate": 8.824700545232647e-06,
      "loss": 0.8589,
      "step": 3354
    },
    {
      "epoch": 0.24615497491677138,
      "grad_norm": 1.4507053801625764,
      "learning_rate": 8.823935117930911e-06,
      "loss": 0.8007,
      "step": 3355
    },
    {
      "epoch": 0.2462283445069105,
      "grad_norm": 1.748847944142736,
      "learning_rate": 8.823169474682087e-06,
      "loss": 0.8119,
      "step": 3356
    },
    {
      "epoch": 0.24630171409704962,
      "grad_norm": 0.7629431491677848,
      "learning_rate": 8.822403615529417e-06,
      "loss": 0.5484,
      "step": 3357
    },
    {
      "epoch": 0.24637508368718875,
      "grad_norm": 1.8384292915394818,
      "learning_rate": 8.821637540516149e-06,
      "loss": 0.8369,
      "step": 3358
    },
    {
      "epoch": 0.24644845327732787,
      "grad_norm": 0.6640827158642569,
      "learning_rate": 8.820871249685543e-06,
      "loss": 0.5343,
      "step": 3359
    },
    {
      "epoch": 0.24652182286746702,
      "grad_norm": 1.6201897566629406,
      "learning_rate": 8.820104743080874e-06,
      "loss": 0.852,
      "step": 3360
    },
    {
      "epoch": 0.24652182286746702,
      "lm_loss": 0.5709,
      "step": 3360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24652182286746702,
      "lm_loss": 0.555,
      "step": 3360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24652182286746702,
      "lm_loss": 0.6203,
      "step": 3360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24652182286746702,
      "lm_loss": 0.4547,
      "step": 3360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24652182286746702,
      "lm_loss": 0.6065,
      "step": 3360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24652182286746702,
      "lm_loss": 0.4719,
      "step": 3360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24652182286746702,
      "lm_loss": 0.6588,
      "step": 3360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24652182286746702,
      "lm_loss": 0.4812,
      "step": 3360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24659519245760614,
      "grad_norm": 0.6125910288653702,
      "learning_rate": 8.819338020745434e-06,
      "loss": 0.5309,
      "step": 3361
    },
    {
      "epoch": 0.24666856204774526,
      "grad_norm": 0.6077468943640527,
      "learning_rate": 8.818571082722516e-06,
      "loss": 0.5308,
      "step": 3362
    },
    {
      "epoch": 0.24674193163788438,
      "grad_norm": 1.6843057179790126,
      "learning_rate": 8.81780392905543e-06,
      "loss": 0.832,
      "step": 3363
    },
    {
      "epoch": 0.2468153012280235,
      "grad_norm": 1.5834692277296605,
      "learning_rate": 8.817036559787504e-06,
      "loss": 0.8697,
      "step": 3364
    },
    {
      "epoch": 0.24688867081816265,
      "grad_norm": 1.401074854489199,
      "learning_rate": 8.81626897496207e-06,
      "loss": 0.8096,
      "step": 3365
    },
    {
      "epoch": 0.24696204040830178,
      "grad_norm": 1.5746597177217172,
      "learning_rate": 8.815501174622476e-06,
      "loss": 0.8522,
      "step": 3366
    },
    {
      "epoch": 0.2470354099984409,
      "grad_norm": 1.5739968585318673,
      "learning_rate": 8.814733158812083e-06,
      "loss": 0.787,
      "step": 3367
    },
    {
      "epoch": 0.24710877958858002,
      "grad_norm": 1.5402058004746084,
      "learning_rate": 8.81396492757426e-06,
      "loss": 0.8812,
      "step": 3368
    },
    {
      "epoch": 0.24710877958858002,
      "lm_loss": 1.1714,
      "step": 3368,
      "vm_loss": 0.0444
    },
    {
      "epoch": 0.24710877958858002,
      "lm_loss": 1.0107,
      "step": 3368,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.24710877958858002,
      "lm_loss": 0.9655,
      "step": 3368,
      "vm_loss": 0.1711
    },
    {
      "epoch": 0.24710877958858002,
      "lm_loss": 0.4518,
      "step": 3368,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.24710877958858002,
      "lm_loss": 0.8905,
      "step": 3368,
      "vm_loss": 0.1988
    },
    {
      "epoch": 0.24710877958858002,
      "lm_loss": 0.544,
      "step": 3368,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.24710877958858002,
      "lm_loss": 1.0232,
      "step": 3368,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.24710877958858002,
      "lm_loss": 0.1575,
      "step": 3368,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.24718214917871914,
      "grad_norm": 1.4400588515791342,
      "learning_rate": 8.813196480952393e-06,
      "loss": 0.8653,
      "step": 3369
    },
    {
      "epoch": 0.2472555187688583,
      "grad_norm": 1.521277092982556,
      "learning_rate": 8.812427818989878e-06,
      "loss": 0.905,
      "step": 3370
    },
    {
      "epoch": 0.2473288883589974,
      "grad_norm": 1.5348978785019376,
      "learning_rate": 8.811658941730122e-06,
      "loss": 0.7899,
      "step": 3371
    },
    {
      "epoch": 0.24740225794913653,
      "grad_norm": 0.7605620006075658,
      "learning_rate": 8.810889849216546e-06,
      "loss": 0.5275,
      "step": 3372
    },
    {
      "epoch": 0.24747562753927566,
      "grad_norm": 1.4842129440934078,
      "learning_rate": 8.810120541492583e-06,
      "loss": 0.8105,
      "step": 3373
    },
    {
      "epoch": 0.24754899712941478,
      "grad_norm": 1.4561138220823067,
      "learning_rate": 8.809351018601677e-06,
      "loss": 0.8137,
      "step": 3374
    },
    {
      "epoch": 0.2476223667195539,
      "grad_norm": 1.6443317530157575,
      "learning_rate": 8.808581280587286e-06,
      "loss": 0.8201,
      "step": 3375
    },
    {
      "epoch": 0.24769573630969305,
      "grad_norm": 1.4258378687162503,
      "learning_rate": 8.807811327492877e-06,
      "loss": 0.7841,
      "step": 3376
    },
    {
      "epoch": 0.24769573630969305,
      "lm_loss": 0.2379,
      "step": 3376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24769573630969305,
      "lm_loss": 0.5281,
      "step": 3376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24769573630969305,
      "lm_loss": 0.5936,
      "step": 3376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24769573630969305,
      "lm_loss": 0.5103,
      "step": 3376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24769573630969305,
      "lm_loss": 0.3424,
      "step": 3376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24769573630969305,
      "lm_loss": 0.6402,
      "step": 3376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24769573630969305,
      "lm_loss": 0.5677,
      "step": 3376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24769573630969305,
      "lm_loss": 0.6113,
      "step": 3376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.24776910589983217,
      "grad_norm": 0.615844858634096,
      "learning_rate": 8.807041159361935e-06,
      "loss": 0.526,
      "step": 3377
    },
    {
      "epoch": 0.2478424754899713,
      "grad_norm": 1.7101884364013158,
      "learning_rate": 8.806270776237949e-06,
      "loss": 0.794,
      "step": 3378
    },
    {
      "epoch": 0.24791584508011041,
      "grad_norm": 0.6011696992070978,
      "learning_rate": 8.805500178164426e-06,
      "loss": 0.5134,
      "step": 3379
    },
    {
      "epoch": 0.24798921467024954,
      "grad_norm": 1.4428235724969245,
      "learning_rate": 8.804729365184884e-06,
      "loss": 0.7442,
      "step": 3380
    },
    {
      "epoch": 0.24806258426038869,
      "grad_norm": 1.837613832460253,
      "learning_rate": 8.803958337342853e-06,
      "loss": 0.8567,
      "step": 3381
    },
    {
      "epoch": 0.2481359538505278,
      "grad_norm": 0.5716006214057396,
      "learning_rate": 8.803187094681873e-06,
      "loss": 0.5376,
      "step": 3382
    },
    {
      "epoch": 0.24820932344066693,
      "grad_norm": 1.626234690675117,
      "learning_rate": 8.8024156372455e-06,
      "loss": 0.8148,
      "step": 3383
    },
    {
      "epoch": 0.24828269303080605,
      "grad_norm": 0.5729334485008969,
      "learning_rate": 8.8016439650773e-06,
      "loss": 0.5521,
      "step": 3384
    },
    {
      "epoch": 0.24828269303080605,
      "lm_loss": 1.4966,
      "step": 3384,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.24828269303080605,
      "lm_loss": 1.3233,
      "step": 3384,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.24828269303080605,
      "lm_loss": 0.7046,
      "step": 3384,
      "vm_loss": 0.126
    },
    {
      "epoch": 0.24828269303080605,
      "lm_loss": 0.2536,
      "step": 3384,
      "vm_loss": 0.1165
    },
    {
      "epoch": 0.24828269303080605,
      "lm_loss": 0.8757,
      "step": 3384,
      "vm_loss": 0.1028
    },
    {
      "epoch": 0.24828269303080605,
      "lm_loss": 0.5326,
      "step": 3384,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.24828269303080605,
      "lm_loss": 1.1235,
      "step": 3384,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.24828269303080605,
      "lm_loss": 1.0894,
      "step": 3384,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.24835606262094517,
      "grad_norm": 1.656502980945408,
      "learning_rate": 8.80087207822085e-06,
      "loss": 0.9237,
      "step": 3385
    },
    {
      "epoch": 0.24842943221108432,
      "grad_norm": 1.589774353208072,
      "learning_rate": 8.800099976719742e-06,
      "loss": 0.8593,
      "step": 3386
    },
    {
      "epoch": 0.24850280180122344,
      "grad_norm": 1.5769276807651866,
      "learning_rate": 8.799327660617577e-06,
      "loss": 0.7936,
      "step": 3387
    },
    {
      "epoch": 0.24857617139136257,
      "grad_norm": 0.608037377251684,
      "learning_rate": 8.79855512995797e-06,
      "loss": 0.5305,
      "step": 3388
    },
    {
      "epoch": 0.2486495409815017,
      "grad_norm": 1.4496105753372277,
      "learning_rate": 8.797782384784549e-06,
      "loss": 0.8491,
      "step": 3389
    },
    {
      "epoch": 0.2487229105716408,
      "grad_norm": 1.5619965685108699,
      "learning_rate": 8.79700942514095e-06,
      "loss": 0.813,
      "step": 3390
    },
    {
      "epoch": 0.24879628016177996,
      "grad_norm": 1.6240228312657847,
      "learning_rate": 8.796236251070827e-06,
      "loss": 0.833,
      "step": 3391
    },
    {
      "epoch": 0.24886964975191908,
      "grad_norm": 0.5732921244874474,
      "learning_rate": 8.79546286261784e-06,
      "loss": 0.5515,
      "step": 3392
    },
    {
      "epoch": 0.24886964975191908,
      "lm_loss": 1.1806,
      "step": 3392,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.24886964975191908,
      "lm_loss": 0.2415,
      "step": 3392,
      "vm_loss": 0.0815
    },
    {
      "epoch": 0.24886964975191908,
      "lm_loss": 0.8333,
      "step": 3392,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.24886964975191908,
      "lm_loss": 0.3911,
      "step": 3392,
      "vm_loss": 0.1349
    },
    {
      "epoch": 0.24886964975191908,
      "lm_loss": 0.9548,
      "step": 3392,
      "vm_loss": 0.2026
    },
    {
      "epoch": 0.24886964975191908,
      "lm_loss": 0.5811,
      "step": 3392,
      "vm_loss": 0.2292
    },
    {
      "epoch": 0.24886964975191908,
      "lm_loss": 1.0698,
      "step": 3392,
      "vm_loss": 0.1848
    },
    {
      "epoch": 0.24886964975191908,
      "lm_loss": 1.1711,
      "step": 3392,
      "vm_loss": 0.1021
    },
    {
      "epoch": 0.2489430193420582,
      "grad_norm": 1.5839549988802957,
      "learning_rate": 8.794689259825666e-06,
      "loss": 0.8842,
      "step": 3393
    },
    {
      "epoch": 0.24901638893219732,
      "grad_norm": 1.5428894209319306,
      "learning_rate": 8.793915442737994e-06,
      "loss": 0.7876,
      "step": 3394
    },
    {
      "epoch": 0.24908975852233645,
      "grad_norm": 1.470883856869266,
      "learning_rate": 8.79314141139852e-06,
      "loss": 0.7123,
      "step": 3395
    },
    {
      "epoch": 0.24916312811247557,
      "grad_norm": 0.6030996286123677,
      "learning_rate": 8.792367165850957e-06,
      "loss": 0.5275,
      "step": 3396
    },
    {
      "epoch": 0.24923649770261472,
      "grad_norm": 1.6155027951624188,
      "learning_rate": 8.791592706139029e-06,
      "loss": 0.9173,
      "step": 3397
    },
    {
      "epoch": 0.24930986729275384,
      "grad_norm": 0.5952294685989583,
      "learning_rate": 8.790818032306472e-06,
      "loss": 0.5179,
      "step": 3398
    },
    {
      "epoch": 0.24938323688289296,
      "grad_norm": 1.5675377936745225,
      "learning_rate": 8.790043144397032e-06,
      "loss": 0.8381,
      "step": 3399
    },
    {
      "epoch": 0.24945660647303208,
      "grad_norm": 0.5801186955430808,
      "learning_rate": 8.789268042454469e-06,
      "loss": 0.5488,
      "step": 3400
    },
    {
      "epoch": 0.24945660647303208,
      "lm_loss": 1.5745,
      "step": 3400,
      "vm_loss": 0.0522
    },
    {
      "epoch": 0.24945660647303208,
      "lm_loss": 1.5426,
      "step": 3400,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.24945660647303208,
      "lm_loss": 0.8334,
      "step": 3400,
      "vm_loss": 0.1088
    },
    {
      "epoch": 0.24945660647303208,
      "lm_loss": 0.7927,
      "step": 3400,
      "vm_loss": 0.0516
    },
    {
      "epoch": 0.24945660647303208,
      "lm_loss": 1.1966,
      "step": 3400,
      "vm_loss": 0.1526
    },
    {
      "epoch": 0.24945660647303208,
      "lm_loss": 1.0416,
      "step": 3400,
      "vm_loss": 0.191
    },
    {
      "epoch": 0.24945660647303208,
      "lm_loss": 0.5335,
      "step": 3400,
      "vm_loss": 0.1934
    },
    {
      "epoch": 0.24945660647303208,
      "lm_loss": 0.1413,
      "step": 3400,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.2495299760631712,
      "grad_norm": 1.4359578269707727,
      "learning_rate": 8.788492726522557e-06,
      "loss": 0.8028,
      "step": 3401
    },
    {
      "epoch": 0.24960334565331035,
      "grad_norm": 1.6646494948468555,
      "learning_rate": 8.787717196645077e-06,
      "loss": 0.7954,
      "step": 3402
    },
    {
      "epoch": 0.24967671524344948,
      "grad_norm": 0.595099762863485,
      "learning_rate": 8.786941452865827e-06,
      "loss": 0.5413,
      "step": 3403
    },
    {
      "epoch": 0.2497500848335886,
      "grad_norm": 1.4150709142282036,
      "learning_rate": 8.786165495228617e-06,
      "loss": 0.7174,
      "step": 3404
    },
    {
      "epoch": 0.24982345442372772,
      "grad_norm": 1.4166787354392745,
      "learning_rate": 8.785389323777265e-06,
      "loss": 0.7783,
      "step": 3405
    },
    {
      "epoch": 0.24989682401386684,
      "grad_norm": 0.5699479219148835,
      "learning_rate": 8.784612938555601e-06,
      "loss": 0.5361,
      "step": 3406
    },
    {
      "epoch": 0.249970193604006,
      "grad_norm": 1.3395056514694326,
      "learning_rate": 8.783836339607475e-06,
      "loss": 0.8116,
      "step": 3407
    },
    {
      "epoch": 0.2500435631941451,
      "grad_norm": 0.5651996554159007,
      "learning_rate": 8.783059526976739e-06,
      "loss": 0.537,
      "step": 3408
    },
    {
      "epoch": 0.2500435631941451,
      "lm_loss": 0.8031,
      "step": 3408,
      "vm_loss": 0.0619
    },
    {
      "epoch": 0.2500435631941451,
      "lm_loss": 0.2589,
      "step": 3408,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.2500435631941451,
      "lm_loss": 0.965,
      "step": 3408,
      "vm_loss": 0.1177
    },
    {
      "epoch": 0.2500435631941451,
      "lm_loss": 0.7485,
      "step": 3408,
      "vm_loss": 0.1562
    },
    {
      "epoch": 0.2500435631941451,
      "lm_loss": 0.5423,
      "step": 3408,
      "vm_loss": 0.199
    },
    {
      "epoch": 0.2500435631941451,
      "lm_loss": 2.4379,
      "step": 3408,
      "vm_loss": 0.164
    },
    {
      "epoch": 0.2500435631941451,
      "lm_loss": 0.1958,
      "step": 3408,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.2500435631941451,
      "lm_loss": 0.0817,
      "step": 3408,
      "vm_loss": 0.0619
    },
    {
      "epoch": 0.2501169327842842,
      "grad_norm": 1.5374981209756011,
      "learning_rate": 8.782282500707262e-06,
      "loss": 0.8637,
      "step": 3409
    },
    {
      "epoch": 0.25019030237442336,
      "grad_norm": 1.5705924100377238,
      "learning_rate": 8.781505260842926e-06,
      "loss": 0.777,
      "step": 3410
    },
    {
      "epoch": 0.2502636719645625,
      "grad_norm": 1.518074384182136,
      "learning_rate": 8.780727807427622e-06,
      "loss": 0.818,
      "step": 3411
    },
    {
      "epoch": 0.2503370415547016,
      "grad_norm": 1.5410316986322448,
      "learning_rate": 8.779950140505258e-06,
      "loss": 0.8858,
      "step": 3412
    },
    {
      "epoch": 0.25041041114484075,
      "grad_norm": 1.3739296346941332,
      "learning_rate": 8.779172260119748e-06,
      "loss": 0.8869,
      "step": 3413
    },
    {
      "epoch": 0.25048378073497984,
      "grad_norm": 1.5623296691202992,
      "learning_rate": 8.778394166315022e-06,
      "loss": 0.8139,
      "step": 3414
    },
    {
      "epoch": 0.250557150325119,
      "grad_norm": 1.5555976719786087,
      "learning_rate": 8.77761585913502e-06,
      "loss": 0.7152,
      "step": 3415
    },
    {
      "epoch": 0.25063051991525814,
      "grad_norm": 1.5197582619343653,
      "learning_rate": 8.776837338623696e-06,
      "loss": 0.8305,
      "step": 3416
    },
    {
      "epoch": 0.25063051991525814,
      "lm_loss": 0.1718,
      "step": 3416,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.25063051991525814,
      "lm_loss": 0.2415,
      "step": 3416,
      "vm_loss": 0.2005
    },
    {
      "epoch": 0.25063051991525814,
      "lm_loss": 1.138,
      "step": 3416,
      "vm_loss": 0.1604
    },
    {
      "epoch": 0.25063051991525814,
      "lm_loss": 0.0909,
      "step": 3416,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.25063051991525814,
      "lm_loss": 0.1538,
      "step": 3416,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.25063051991525814,
      "lm_loss": 0.6295,
      "step": 3416,
      "vm_loss": 0.1376
    },
    {
      "epoch": 0.25063051991525814,
      "lm_loss": 0.1949,
      "step": 3416,
      "vm_loss": 0.0628
    },
    {
      "epoch": 0.25063051991525814,
      "lm_loss": 1.0529,
      "step": 3416,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.25070388950539724,
      "grad_norm": 1.6984770333516042,
      "learning_rate": 8.776058604825013e-06,
      "loss": 0.8298,
      "step": 3417
    },
    {
      "epoch": 0.2507772590955364,
      "grad_norm": 1.671127259754703,
      "learning_rate": 8.775279657782949e-06,
      "loss": 0.8864,
      "step": 3418
    },
    {
      "epoch": 0.2508506286856755,
      "grad_norm": 1.516803136454778,
      "learning_rate": 8.774500497541495e-06,
      "loss": 0.778,
      "step": 3419
    },
    {
      "epoch": 0.25092399827581463,
      "grad_norm": 1.600913613596827,
      "learning_rate": 8.77372112414465e-06,
      "loss": 0.8263,
      "step": 3420
    },
    {
      "epoch": 0.2509973678659538,
      "grad_norm": 0.7565029226047189,
      "learning_rate": 8.772941537636427e-06,
      "loss": 0.5549,
      "step": 3421
    },
    {
      "epoch": 0.2510707374560929,
      "grad_norm": 0.6655783636671734,
      "learning_rate": 8.772161738060853e-06,
      "loss": 0.5497,
      "step": 3422
    },
    {
      "epoch": 0.251144107046232,
      "grad_norm": 1.5529475010340832,
      "learning_rate": 8.771381725461963e-06,
      "loss": 0.9215,
      "step": 3423
    },
    {
      "epoch": 0.2512174766363711,
      "grad_norm": 1.391794676219224,
      "learning_rate": 8.770601499883807e-06,
      "loss": 0.8831,
      "step": 3424
    },
    {
      "epoch": 0.2512174766363711,
      "lm_loss": 0.521,
      "step": 3424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2512174766363711,
      "lm_loss": 0.6491,
      "step": 3424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2512174766363711,
      "lm_loss": 0.3786,
      "step": 3424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2512174766363711,
      "lm_loss": 0.4512,
      "step": 3424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2512174766363711,
      "lm_loss": 0.5716,
      "step": 3424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2512174766363711,
      "lm_loss": 0.636,
      "step": 3424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2512174766363711,
      "lm_loss": 0.6326,
      "step": 3424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2512174766363711,
      "lm_loss": 0.6214,
      "step": 3424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25129084622651027,
      "grad_norm": 0.6481233617324401,
      "learning_rate": 8.769821061370447e-06,
      "loss": 0.5411,
      "step": 3425
    },
    {
      "epoch": 0.2513642158166494,
      "grad_norm": 0.6733575179682362,
      "learning_rate": 8.769040409965957e-06,
      "loss": 0.5391,
      "step": 3426
    },
    {
      "epoch": 0.2514375854067885,
      "grad_norm": 0.6295003616816709,
      "learning_rate": 8.768259545714419e-06,
      "loss": 0.5362,
      "step": 3427
    },
    {
      "epoch": 0.25151095499692766,
      "grad_norm": 1.9302171880646604,
      "learning_rate": 8.767478468659935e-06,
      "loss": 0.916,
      "step": 3428
    },
    {
      "epoch": 0.25158432458706675,
      "grad_norm": 1.6392231387454823,
      "learning_rate": 8.766697178846611e-06,
      "loss": 0.9071,
      "step": 3429
    },
    {
      "epoch": 0.2516576941772059,
      "grad_norm": 0.6921872600554961,
      "learning_rate": 8.765915676318569e-06,
      "loss": 0.5373,
      "step": 3430
    },
    {
      "epoch": 0.25173106376734505,
      "grad_norm": 1.664506169383548,
      "learning_rate": 8.76513396111994e-06,
      "loss": 0.8065,
      "step": 3431
    },
    {
      "epoch": 0.25180443335748415,
      "grad_norm": 0.6725369909766964,
      "learning_rate": 8.764352033294877e-06,
      "loss": 0.5436,
      "step": 3432
    },
    {
      "epoch": 0.25180443335748415,
      "lm_loss": 1.0103,
      "step": 3432,
      "vm_loss": 0.0465
    },
    {
      "epoch": 0.25180443335748415,
      "lm_loss": 0.3816,
      "step": 3432,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.25180443335748415,
      "lm_loss": 0.5081,
      "step": 3432,
      "vm_loss": 0.1544
    },
    {
      "epoch": 0.25180443335748415,
      "lm_loss": 0.6907,
      "step": 3432,
      "vm_loss": 0.1894
    },
    {
      "epoch": 0.25180443335748415,
      "lm_loss": 1.074,
      "step": 3432,
      "vm_loss": 0.1702
    },
    {
      "epoch": 0.25180443335748415,
      "lm_loss": 0.373,
      "step": 3432,
      "vm_loss": 0.2284
    },
    {
      "epoch": 0.25180443335748415,
      "lm_loss": 0.1929,
      "step": 3432,
      "vm_loss": 0.1065
    },
    {
      "epoch": 0.25180443335748415,
      "lm_loss": 0.5151,
      "step": 3432,
      "vm_loss": 0.1005
    },
    {
      "epoch": 0.2518778029476233,
      "grad_norm": 1.6676263235656719,
      "learning_rate": 8.763569892887529e-06,
      "loss": 0.8148,
      "step": 3433
    },
    {
      "epoch": 0.2519511725377624,
      "grad_norm": 1.5404486985190113,
      "learning_rate": 8.76278753994207e-06,
      "loss": 0.8216,
      "step": 3434
    },
    {
      "epoch": 0.25202454212790154,
      "grad_norm": 0.5865237890288142,
      "learning_rate": 8.76200497450268e-06,
      "loss": 0.5169,
      "step": 3435
    },
    {
      "epoch": 0.2520979117180407,
      "grad_norm": 0.6214295769166265,
      "learning_rate": 8.761222196613552e-06,
      "loss": 0.5342,
      "step": 3436
    },
    {
      "epoch": 0.2521712813081798,
      "grad_norm": 1.6847011796012794,
      "learning_rate": 8.760439206318894e-06,
      "loss": 0.8135,
      "step": 3437
    },
    {
      "epoch": 0.25224465089831893,
      "grad_norm": 0.5928311813961699,
      "learning_rate": 8.75965600366292e-06,
      "loss": 0.5274,
      "step": 3438
    },
    {
      "epoch": 0.25231802048845803,
      "grad_norm": 1.4863597521386789,
      "learning_rate": 8.75887258868986e-06,
      "loss": 0.841,
      "step": 3439
    },
    {
      "epoch": 0.2523913900785972,
      "grad_norm": 0.5663628488783722,
      "learning_rate": 8.758088961443958e-06,
      "loss": 0.5111,
      "step": 3440
    },
    {
      "epoch": 0.2523913900785972,
      "lm_loss": 0.5254,
      "step": 3440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2523913900785972,
      "lm_loss": 0.613,
      "step": 3440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2523913900785972,
      "lm_loss": 0.5912,
      "step": 3440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2523913900785972,
      "lm_loss": 0.5944,
      "step": 3440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2523913900785972,
      "lm_loss": 0.3964,
      "step": 3440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2523913900785972,
      "lm_loss": 0.4483,
      "step": 3440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2523913900785972,
      "lm_loss": 0.625,
      "step": 3440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2523913900785972,
      "lm_loss": 0.4978,
      "step": 3440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2524647596687363,
      "grad_norm": 0.5789809277389592,
      "learning_rate": 8.757305121969463e-06,
      "loss": 0.5471,
      "step": 3441
    },
    {
      "epoch": 0.2525381292588754,
      "grad_norm": 1.6599307793702707,
      "learning_rate": 8.756521070310645e-06,
      "loss": 0.9187,
      "step": 3442
    },
    {
      "epoch": 0.25261149884901457,
      "grad_norm": 0.5735208829365585,
      "learning_rate": 8.75573680651178e-06,
      "loss": 0.5357,
      "step": 3443
    },
    {
      "epoch": 0.25268486843915366,
      "grad_norm": 1.5728487422665456,
      "learning_rate": 8.754952330617156e-06,
      "loss": 0.7852,
      "step": 3444
    },
    {
      "epoch": 0.2527582380292928,
      "grad_norm": 0.5639756266411412,
      "learning_rate": 8.754167642671075e-06,
      "loss": 0.536,
      "step": 3445
    },
    {
      "epoch": 0.2528316076194319,
      "grad_norm": 1.6306383928826043,
      "learning_rate": 8.753382742717849e-06,
      "loss": 0.9437,
      "step": 3446
    },
    {
      "epoch": 0.25290497720957106,
      "grad_norm": 0.5759534099788011,
      "learning_rate": 8.752597630801804e-06,
      "loss": 0.5256,
      "step": 3447
    },
    {
      "epoch": 0.2529783467997102,
      "grad_norm": 0.5994288754553578,
      "learning_rate": 8.75181230696728e-06,
      "loss": 0.5339,
      "step": 3448
    },
    {
      "epoch": 0.2529783467997102,
      "lm_loss": 1.5989,
      "step": 3448,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.2529783467997102,
      "lm_loss": 1.1915,
      "step": 3448,
      "vm_loss": 0.0903
    },
    {
      "epoch": 0.2529783467997102,
      "lm_loss": 0.4232,
      "step": 3448,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.2529783467997102,
      "lm_loss": 0.2816,
      "step": 3448,
      "vm_loss": 0.0884
    },
    {
      "epoch": 0.2529783467997102,
      "lm_loss": 0.2556,
      "step": 3448,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.2529783467997102,
      "lm_loss": 0.8114,
      "step": 3448,
      "vm_loss": 0.0809
    },
    {
      "epoch": 0.2529783467997102,
      "lm_loss": 0.12,
      "step": 3448,
      "vm_loss": 0.102
    },
    {
      "epoch": 0.2529783467997102,
      "lm_loss": 0.3376,
      "step": 3448,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.2530517163898493,
      "grad_norm": 1.6578439850890552,
      "learning_rate": 8.751026771258622e-06,
      "loss": 0.8055,
      "step": 3449
    },
    {
      "epoch": 0.25312508597998845,
      "grad_norm": 0.5786878043433178,
      "learning_rate": 8.750241023720194e-06,
      "loss": 0.5536,
      "step": 3450
    },
    {
      "epoch": 0.25319845557012755,
      "grad_norm": 1.794143471240364,
      "learning_rate": 8.749455064396368e-06,
      "loss": 0.8781,
      "step": 3451
    },
    {
      "epoch": 0.2532718251602667,
      "grad_norm": 1.4224678163502964,
      "learning_rate": 8.748668893331528e-06,
      "loss": 0.8127,
      "step": 3452
    },
    {
      "epoch": 0.25334519475040584,
      "grad_norm": 1.5076150742751866,
      "learning_rate": 8.747882510570074e-06,
      "loss": 0.844,
      "step": 3453
    },
    {
      "epoch": 0.25341856434054494,
      "grad_norm": 1.4575790330213894,
      "learning_rate": 8.747095916156412e-06,
      "loss": 0.7283,
      "step": 3454
    },
    {
      "epoch": 0.2534919339306841,
      "grad_norm": 1.4436613650772308,
      "learning_rate": 8.746309110134964e-06,
      "loss": 0.7509,
      "step": 3455
    },
    {
      "epoch": 0.2535653035208232,
      "grad_norm": 1.6139608539440795,
      "learning_rate": 8.745522092550162e-06,
      "loss": 0.8394,
      "step": 3456
    },
    {
      "epoch": 0.2535653035208232,
      "lm_loss": 1.0807,
      "step": 3456,
      "vm_loss": 0.0809
    },
    {
      "epoch": 0.2535653035208232,
      "lm_loss": 0.9459,
      "step": 3456,
      "vm_loss": 0.1219
    },
    {
      "epoch": 0.2535653035208232,
      "lm_loss": 1.0382,
      "step": 3456,
      "vm_loss": 0.1227
    },
    {
      "epoch": 0.2535653035208232,
      "lm_loss": 1.2211,
      "step": 3456,
      "vm_loss": 0.1704
    },
    {
      "epoch": 0.2535653035208232,
      "lm_loss": 1.2442,
      "step": 3456,
      "vm_loss": 0.1137
    },
    {
      "epoch": 0.2535653035208232,
      "lm_loss": 0.0253,
      "step": 3456,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.2535653035208232,
      "lm_loss": 0.4579,
      "step": 3456,
      "vm_loss": 0.1725
    },
    {
      "epoch": 0.2535653035208232,
      "lm_loss": 1.1073,
      "step": 3456,
      "vm_loss": 0.1117
    },
    {
      "epoch": 0.25363867311096233,
      "grad_norm": 1.6369457553788056,
      "learning_rate": 8.744734863446453e-06,
      "loss": 0.8016,
      "step": 3457
    },
    {
      "epoch": 0.2537120427011015,
      "grad_norm": 1.4122203355731913,
      "learning_rate": 8.743947422868293e-06,
      "loss": 0.7993,
      "step": 3458
    },
    {
      "epoch": 0.2537854122912406,
      "grad_norm": 1.5475869578155632,
      "learning_rate": 8.743159770860151e-06,
      "loss": 0.771,
      "step": 3459
    },
    {
      "epoch": 0.2538587818813797,
      "grad_norm": 0.6954574565670885,
      "learning_rate": 8.742371907466505e-06,
      "loss": 0.5415,
      "step": 3460
    },
    {
      "epoch": 0.2539321514715188,
      "grad_norm": 0.6367355660966968,
      "learning_rate": 8.741583832731851e-06,
      "loss": 0.5397,
      "step": 3461
    },
    {
      "epoch": 0.25400552106165797,
      "grad_norm": 1.9138696334045973,
      "learning_rate": 8.740795546700691e-06,
      "loss": 0.8963,
      "step": 3462
    },
    {
      "epoch": 0.2540788906517971,
      "grad_norm": 1.623031718098392,
      "learning_rate": 8.740007049417544e-06,
      "loss": 0.8338,
      "step": 3463
    },
    {
      "epoch": 0.2541522602419362,
      "grad_norm": 0.6596815522003066,
      "learning_rate": 8.739218340926936e-06,
      "loss": 0.5605,
      "step": 3464
    },
    {
      "epoch": 0.2541522602419362,
      "lm_loss": 1.1616,
      "step": 3464,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.2541522602419362,
      "lm_loss": 0.7109,
      "step": 3464,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.2541522602419362,
      "lm_loss": 1.1579,
      "step": 3464,
      "vm_loss": 0.1084
    },
    {
      "epoch": 0.2541522602419362,
      "lm_loss": 0.4834,
      "step": 3464,
      "vm_loss": 0.1733
    },
    {
      "epoch": 0.2541522602419362,
      "lm_loss": 0.2422,
      "step": 3464,
      "vm_loss": 0.065
    },
    {
      "epoch": 0.2541522602419362,
      "lm_loss": 0.9065,
      "step": 3464,
      "vm_loss": 0.2239
    },
    {
      "epoch": 0.2541522602419362,
      "lm_loss": 1.0179,
      "step": 3464,
      "vm_loss": 0.1295
    },
    {
      "epoch": 0.2541522602419362,
      "lm_loss": 1.1723,
      "step": 3464,
      "vm_loss": 0.2052
    },
    {
      "epoch": 0.25422562983207536,
      "grad_norm": 1.564258647574273,
      "learning_rate": 8.738429421273407e-06,
      "loss": 0.8288,
      "step": 3465
    },
    {
      "epoch": 0.25429899942221446,
      "grad_norm": 1.4209559627731214,
      "learning_rate": 8.737640290501512e-06,
      "loss": 0.7739,
      "step": 3466
    },
    {
      "epoch": 0.2543723690123536,
      "grad_norm": 1.67140436855773,
      "learning_rate": 8.736850948655814e-06,
      "loss": 0.9629,
      "step": 3467
    },
    {
      "epoch": 0.25444573860249275,
      "grad_norm": 1.671745702331478,
      "learning_rate": 8.736061395780889e-06,
      "loss": 0.8895,
      "step": 3468
    },
    {
      "epoch": 0.25451910819263185,
      "grad_norm": 1.7049662516689583,
      "learning_rate": 8.735271631921322e-06,
      "loss": 0.9103,
      "step": 3469
    },
    {
      "epoch": 0.254592477782771,
      "grad_norm": 1.6771685063704018,
      "learning_rate": 8.734481657121718e-06,
      "loss": 0.8772,
      "step": 3470
    },
    {
      "epoch": 0.2546658473729101,
      "grad_norm": 1.6207187338261684,
      "learning_rate": 8.733691471426688e-06,
      "loss": 0.8324,
      "step": 3471
    },
    {
      "epoch": 0.25473921696304924,
      "grad_norm": 1.4342873892187153,
      "learning_rate": 8.732901074880852e-06,
      "loss": 0.7646,
      "step": 3472
    },
    {
      "epoch": 0.25473921696304924,
      "lm_loss": 1.1047,
      "step": 3472,
      "vm_loss": 0.049
    },
    {
      "epoch": 0.25473921696304924,
      "lm_loss": 1.3229,
      "step": 3472,
      "vm_loss": 0.173
    },
    {
      "epoch": 0.25473921696304924,
      "lm_loss": 0.583,
      "step": 3472,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.25473921696304924,
      "lm_loss": 1.1135,
      "step": 3472,
      "vm_loss": 0.144
    },
    {
      "epoch": 0.25473921696304924,
      "lm_loss": 0.7746,
      "step": 3472,
      "vm_loss": 0.2052
    },
    {
      "epoch": 0.25473921696304924,
      "lm_loss": 1.6862,
      "step": 3472,
      "vm_loss": 0.1554
    },
    {
      "epoch": 0.25473921696304924,
      "lm_loss": 1.1956,
      "step": 3472,
      "vm_loss": 0.1461
    },
    {
      "epoch": 0.25473921696304924,
      "lm_loss": 0.1021,
      "step": 3472,
      "vm_loss": 0.1182
    },
    {
      "epoch": 0.2548125865531884,
      "grad_norm": 1.3266758922776984,
      "learning_rate": 8.732110467528849e-06,
      "loss": 0.7725,
      "step": 3473
    },
    {
      "epoch": 0.2548859561433275,
      "grad_norm": 0.7968155989302191,
      "learning_rate": 8.731319649415323e-06,
      "loss": 0.5297,
      "step": 3474
    },
    {
      "epoch": 0.25495932573346664,
      "grad_norm": 0.7505444881925782,
      "learning_rate": 8.730528620584939e-06,
      "loss": 0.5488,
      "step": 3475
    },
    {
      "epoch": 0.25503269532360573,
      "grad_norm": 0.5833167927820108,
      "learning_rate": 8.729737381082364e-06,
      "loss": 0.5228,
      "step": 3476
    },
    {
      "epoch": 0.2551060649137449,
      "grad_norm": 1.601012060952373,
      "learning_rate": 8.728945930952283e-06,
      "loss": 0.8582,
      "step": 3477
    },
    {
      "epoch": 0.25517943450388403,
      "grad_norm": 1.5440797605506353,
      "learning_rate": 8.72815427023939e-06,
      "loss": 0.8851,
      "step": 3478
    },
    {
      "epoch": 0.2552528040940231,
      "grad_norm": 1.6154668506315455,
      "learning_rate": 8.727362398988393e-06,
      "loss": 0.9353,
      "step": 3479
    },
    {
      "epoch": 0.25532617368416227,
      "grad_norm": 1.5620634835962761,
      "learning_rate": 8.726570317244012e-06,
      "loss": 0.9787,
      "step": 3480
    },
    {
      "epoch": 0.25532617368416227,
      "lm_loss": 0.5432,
      "step": 3480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25532617368416227,
      "lm_loss": 0.1774,
      "step": 3480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25532617368416227,
      "lm_loss": 0.4539,
      "step": 3480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25532617368416227,
      "lm_loss": 0.5086,
      "step": 3480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25532617368416227,
      "lm_loss": 0.5427,
      "step": 3480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25532617368416227,
      "lm_loss": 0.7028,
      "step": 3480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25532617368416227,
      "lm_loss": 0.6809,
      "step": 3480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25532617368416227,
      "lm_loss": 0.4087,
      "step": 3480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25539954327430137,
      "grad_norm": 1.083394445232287,
      "learning_rate": 8.725778025050974e-06,
      "loss": 0.5346,
      "step": 3481
    },
    {
      "epoch": 0.2554729128644405,
      "grad_norm": 0.9943330178988848,
      "learning_rate": 8.724985522454026e-06,
      "loss": 0.5444,
      "step": 3482
    },
    {
      "epoch": 0.25554628245457967,
      "grad_norm": 0.7683632863482679,
      "learning_rate": 8.72419280949792e-06,
      "loss": 0.5354,
      "step": 3483
    },
    {
      "epoch": 0.25561965204471876,
      "grad_norm": 1.6575779805984012,
      "learning_rate": 8.723399886227424e-06,
      "loss": 0.7967,
      "step": 3484
    },
    {
      "epoch": 0.2556930216348579,
      "grad_norm": 1.5442327077029334,
      "learning_rate": 8.722606752687316e-06,
      "loss": 0.8194,
      "step": 3485
    },
    {
      "epoch": 0.255766391224997,
      "grad_norm": 0.9675180276416738,
      "learning_rate": 8.721813408922385e-06,
      "loss": 0.5545,
      "step": 3486
    },
    {
      "epoch": 0.25583976081513615,
      "grad_norm": 1.0317119842200604,
      "learning_rate": 8.721019854977432e-06,
      "loss": 0.5628,
      "step": 3487
    },
    {
      "epoch": 0.25591313040527525,
      "grad_norm": 1.7573534722874617,
      "learning_rate": 8.720226090897275e-06,
      "loss": 0.8858,
      "step": 3488
    },
    {
      "epoch": 0.25591313040527525,
      "lm_loss": 1.0676,
      "step": 3488,
      "vm_loss": 0.0532
    },
    {
      "epoch": 0.25591313040527525,
      "lm_loss": 0.7746,
      "step": 3488,
      "vm_loss": 0.0848
    },
    {
      "epoch": 0.25591313040527525,
      "lm_loss": 0.9733,
      "step": 3488,
      "vm_loss": 0.1814
    },
    {
      "epoch": 0.25591313040527525,
      "lm_loss": 0.5759,
      "step": 3488,
      "vm_loss": 0.166
    },
    {
      "epoch": 0.25591313040527525,
      "lm_loss": 0.8128,
      "step": 3488,
      "vm_loss": 0.0807
    },
    {
      "epoch": 0.25591313040527525,
      "lm_loss": 0.2209,
      "step": 3488,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.25591313040527525,
      "lm_loss": 0.1435,
      "step": 3488,
      "vm_loss": 0.1362
    },
    {
      "epoch": 0.25591313040527525,
      "lm_loss": 0.9519,
      "step": 3488,
      "vm_loss": 0.1669
    },
    {
      "epoch": 0.2559864999954144,
      "grad_norm": 1.504702925545385,
      "learning_rate": 8.719432116726738e-06,
      "loss": 0.7796,
      "step": 3489
    },
    {
      "epoch": 0.25605986958555355,
      "grad_norm": 1.5138916988806816,
      "learning_rate": 8.718637932510658e-06,
      "loss": 0.7652,
      "step": 3490
    },
    {
      "epoch": 0.25613323917569264,
      "grad_norm": 1.5659425378154745,
      "learning_rate": 8.717843538293884e-06,
      "loss": 0.8096,
      "step": 3491
    },
    {
      "epoch": 0.2562066087658318,
      "grad_norm": 0.9655894671411303,
      "learning_rate": 8.71704893412128e-06,
      "loss": 0.5506,
      "step": 3492
    },
    {
      "epoch": 0.2562799783559709,
      "grad_norm": 0.8029801792580616,
      "learning_rate": 8.716254120037715e-06,
      "loss": 0.535,
      "step": 3493
    },
    {
      "epoch": 0.25635334794611003,
      "grad_norm": 0.6313526019069986,
      "learning_rate": 8.715459096088078e-06,
      "loss": 0.523,
      "step": 3494
    },
    {
      "epoch": 0.2564267175362492,
      "grad_norm": 2.0396226024838446,
      "learning_rate": 8.714663862317266e-06,
      "loss": 0.8537,
      "step": 3495
    },
    {
      "epoch": 0.2565000871263883,
      "grad_norm": 1.6053839187862748,
      "learning_rate": 8.713868418770184e-06,
      "loss": 0.8552,
      "step": 3496
    },
    {
      "epoch": 0.2565000871263883,
      "lm_loss": 0.475,
      "step": 3496,
      "vm_loss": 0.0485
    },
    {
      "epoch": 0.2565000871263883,
      "lm_loss": 0.9815,
      "step": 3496,
      "vm_loss": 0.1349
    },
    {
      "epoch": 0.2565000871263883,
      "lm_loss": 1.0295,
      "step": 3496,
      "vm_loss": 0.1371
    },
    {
      "epoch": 0.2565000871263883,
      "lm_loss": 0.3324,
      "step": 3496,
      "vm_loss": 0.0517
    },
    {
      "epoch": 0.2565000871263883,
      "lm_loss": 0.5956,
      "step": 3496,
      "vm_loss": 0.1408
    },
    {
      "epoch": 0.2565000871263883,
      "lm_loss": 0.3599,
      "step": 3496,
      "vm_loss": 0.1627
    },
    {
      "epoch": 0.2565000871263883,
      "lm_loss": 0.8773,
      "step": 3496,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.2565000871263883,
      "lm_loss": 0.6577,
      "step": 3496,
      "vm_loss": 0.1627
    },
    {
      "epoch": 0.2565734567165274,
      "grad_norm": 1.5339218650973099,
      "learning_rate": 8.713072765491757e-06,
      "loss": 0.7766,
      "step": 3497
    },
    {
      "epoch": 0.2566468263066665,
      "grad_norm": 1.0416642260853888,
      "learning_rate": 8.712276902526914e-06,
      "loss": 0.5301,
      "step": 3498
    },
    {
      "epoch": 0.25672019589680567,
      "grad_norm": 0.9296626881135481,
      "learning_rate": 8.711480829920603e-06,
      "loss": 0.5535,
      "step": 3499
    },
    {
      "epoch": 0.2567935654869448,
      "grad_norm": 2.0181488031578563,
      "learning_rate": 8.710684547717778e-06,
      "loss": 0.8348,
      "step": 3500
    },
    {
      "epoch": 0.2568669350770839,
      "grad_norm": 1.6477701613308402,
      "learning_rate": 8.709888055963406e-06,
      "loss": 0.8677,
      "step": 3501
    },
    {
      "epoch": 0.25694030466722306,
      "grad_norm": 1.6848074472815178,
      "learning_rate": 8.709091354702468e-06,
      "loss": 0.8522,
      "step": 3502
    },
    {
      "epoch": 0.25701367425736216,
      "grad_norm": 1.6481645829569096,
      "learning_rate": 8.708294443979957e-06,
      "loss": 0.7924,
      "step": 3503
    },
    {
      "epoch": 0.2570870438475013,
      "grad_norm": 1.6133439446174935,
      "learning_rate": 8.707497323840875e-06,
      "loss": 0.8794,
      "step": 3504
    },
    {
      "epoch": 0.2570870438475013,
      "lm_loss": 0.3986,
      "step": 3504,
      "vm_loss": 0.0554
    },
    {
      "epoch": 0.2570870438475013,
      "lm_loss": 0.7372,
      "step": 3504,
      "vm_loss": 0.1463
    },
    {
      "epoch": 0.2570870438475013,
      "lm_loss": 0.4395,
      "step": 3504,
      "vm_loss": 0.1138
    },
    {
      "epoch": 0.2570870438475013,
      "lm_loss": 0.7872,
      "step": 3504,
      "vm_loss": 0.1697
    },
    {
      "epoch": 0.2570870438475013,
      "lm_loss": 1.1809,
      "step": 3504,
      "vm_loss": 0.0978
    },
    {
      "epoch": 0.2570870438475013,
      "lm_loss": 1.1728,
      "step": 3504,
      "vm_loss": 0.3022
    },
    {
      "epoch": 0.2570870438475013,
      "lm_loss": 1.576,
      "step": 3504,
      "vm_loss": 0.1066
    },
    {
      "epoch": 0.2570870438475013,
      "lm_loss": 0.8181,
      "step": 3504,
      "vm_loss": 0.1245
    },
    {
      "epoch": 0.25716041343764046,
      "grad_norm": 1.6832989470214992,
      "learning_rate": 8.706699994330237e-06,
      "loss": 0.8747,
      "step": 3505
    },
    {
      "epoch": 0.25723378302777955,
      "grad_norm": 1.919231783057501,
      "learning_rate": 8.705902455493071e-06,
      "loss": 0.9219,
      "step": 3506
    },
    {
      "epoch": 0.2573071526179187,
      "grad_norm": 1.457545245474864,
      "learning_rate": 8.705104707374416e-06,
      "loss": 0.7455,
      "step": 3507
    },
    {
      "epoch": 0.2573805222080578,
      "grad_norm": 1.4863570211287587,
      "learning_rate": 8.704306750019323e-06,
      "loss": 0.5656,
      "step": 3508
    },
    {
      "epoch": 0.25745389179819694,
      "grad_norm": 1.0133568647439688,
      "learning_rate": 8.703508583472855e-06,
      "loss": 0.5523,
      "step": 3509
    },
    {
      "epoch": 0.2575272613883361,
      "grad_norm": 1.6300508005116323,
      "learning_rate": 8.702710207780083e-06,
      "loss": 0.8149,
      "step": 3510
    },
    {
      "epoch": 0.2576006309784752,
      "grad_norm": 0.6907767577596583,
      "learning_rate": 8.7019116229861e-06,
      "loss": 0.5115,
      "step": 3511
    },
    {
      "epoch": 0.25767400056861434,
      "grad_norm": 1.7639122142323342,
      "learning_rate": 8.701112829135996e-06,
      "loss": 0.8176,
      "step": 3512
    },
    {
      "epoch": 0.25767400056861434,
      "lm_loss": 1.364,
      "step": 3512,
      "vm_loss": 0.0591
    },
    {
      "epoch": 0.25767400056861434,
      "lm_loss": 1.5566,
      "step": 3512,
      "vm_loss": 0.0944
    },
    {
      "epoch": 0.25767400056861434,
      "lm_loss": 0.2453,
      "step": 3512,
      "vm_loss": 0.1707
    },
    {
      "epoch": 0.25767400056861434,
      "lm_loss": 0.6477,
      "step": 3512,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.25767400056861434,
      "lm_loss": 0.6504,
      "step": 3512,
      "vm_loss": 0.0917
    },
    {
      "epoch": 0.25767400056861434,
      "lm_loss": 0.3069,
      "step": 3512,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.25767400056861434,
      "lm_loss": 0.8454,
      "step": 3512,
      "vm_loss": 0.2161
    },
    {
      "epoch": 0.25767400056861434,
      "lm_loss": 0.1407,
      "step": 3512,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.25774737015875343,
      "grad_norm": 1.691563673913784,
      "learning_rate": 8.700313826274886e-06,
      "loss": 0.7971,
      "step": 3513
    },
    {
      "epoch": 0.2578207397488926,
      "grad_norm": 1.8390360950784799,
      "learning_rate": 8.699514614447892e-06,
      "loss": 0.901,
      "step": 3514
    },
    {
      "epoch": 0.25789410933903173,
      "grad_norm": 1.6221612391777351,
      "learning_rate": 8.698715193700145e-06,
      "loss": 0.8815,
      "step": 3515
    },
    {
      "epoch": 0.2579674789291708,
      "grad_norm": 1.5890726834778313,
      "learning_rate": 8.697915564076791e-06,
      "loss": 0.8169,
      "step": 3516
    },
    {
      "epoch": 0.25804084851931,
      "grad_norm": 1.6220179990411376,
      "learning_rate": 8.697115725622988e-06,
      "loss": 0.8331,
      "step": 3517
    },
    {
      "epoch": 0.25811421810944907,
      "grad_norm": 1.7746935608465508,
      "learning_rate": 8.696315678383905e-06,
      "loss": 0.9499,
      "step": 3518
    },
    {
      "epoch": 0.2581875876995882,
      "grad_norm": 1.743494789684782,
      "learning_rate": 8.69551542240472e-06,
      "loss": 0.5623,
      "step": 3519
    },
    {
      "epoch": 0.25826095728972737,
      "grad_norm": 1.772826606537224,
      "learning_rate": 8.694714957730628e-06,
      "loss": 0.9085,
      "step": 3520
    },
    {
      "epoch": 0.25826095728972737,
      "lm_loss": 1.1585,
      "step": 3520,
      "vm_loss": 0.0442
    },
    {
      "epoch": 0.25826095728972737,
      "lm_loss": 0.4527,
      "step": 3520,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.25826095728972737,
      "lm_loss": 0.0337,
      "step": 3520,
      "vm_loss": 0.2084
    },
    {
      "epoch": 0.25826095728972737,
      "lm_loss": 0.3124,
      "step": 3520,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.25826095728972737,
      "lm_loss": 0.7349,
      "step": 3520,
      "vm_loss": 0.172
    },
    {
      "epoch": 0.25826095728972737,
      "lm_loss": 0.4687,
      "step": 3520,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.25826095728972737,
      "lm_loss": 0.899,
      "step": 3520,
      "vm_loss": 0.2213
    },
    {
      "epoch": 0.25826095728972737,
      "lm_loss": 0.5637,
      "step": 3520,
      "vm_loss": 0.1901
    },
    {
      "epoch": 0.25833432687986646,
      "grad_norm": 1.7076214017094464,
      "learning_rate": 8.693914284406833e-06,
      "loss": 0.8469,
      "step": 3521
    },
    {
      "epoch": 0.2584076964700056,
      "grad_norm": 1.5686765200379456,
      "learning_rate": 8.69311340247855e-06,
      "loss": 0.8294,
      "step": 3522
    },
    {
      "epoch": 0.2584810660601447,
      "grad_norm": 1.4677686598201845,
      "learning_rate": 8.692312311991007e-06,
      "loss": 0.7634,
      "step": 3523
    },
    {
      "epoch": 0.25855443565028385,
      "grad_norm": 0.6936385514758819,
      "learning_rate": 8.691511012989445e-06,
      "loss": 0.525,
      "step": 3524
    },
    {
      "epoch": 0.258627805240423,
      "grad_norm": 1.7495934104070645,
      "learning_rate": 8.690709505519113e-06,
      "loss": 0.8153,
      "step": 3525
    },
    {
      "epoch": 0.2587011748305621,
      "grad_norm": 1.5695527245182346,
      "learning_rate": 8.689907789625276e-06,
      "loss": 0.9107,
      "step": 3526
    },
    {
      "epoch": 0.25877454442070125,
      "grad_norm": 1.6051638654643623,
      "learning_rate": 8.68910586535321e-06,
      "loss": 0.801,
      "step": 3527
    },
    {
      "epoch": 0.25884791401084034,
      "grad_norm": 0.7931471243777346,
      "learning_rate": 8.688303732748198e-06,
      "loss": 0.5483,
      "step": 3528
    },
    {
      "epoch": 0.25884791401084034,
      "lm_loss": 2.9011,
      "step": 3528,
      "vm_loss": 0.1315
    },
    {
      "epoch": 0.25884791401084034,
      "lm_loss": 0.7706,
      "step": 3528,
      "vm_loss": 0.1524
    },
    {
      "epoch": 0.25884791401084034,
      "lm_loss": 0.954,
      "step": 3528,
      "vm_loss": 0.1977
    },
    {
      "epoch": 0.25884791401084034,
      "lm_loss": 0.5718,
      "step": 3528,
      "vm_loss": 0.2096
    },
    {
      "epoch": 0.25884791401084034,
      "lm_loss": 0.2116,
      "step": 3528,
      "vm_loss": 0.0561
    },
    {
      "epoch": 0.25884791401084034,
      "lm_loss": 0.9098,
      "step": 3528,
      "vm_loss": 0.211
    },
    {
      "epoch": 0.25884791401084034,
      "lm_loss": 1.0027,
      "step": 3528,
      "vm_loss": 0.1751
    },
    {
      "epoch": 0.25884791401084034,
      "lm_loss": 0.6275,
      "step": 3528,
      "vm_loss": 0.1424
    },
    {
      "epoch": 0.2589212836009795,
      "grad_norm": 2.008110086229945,
      "learning_rate": 8.68750139185554e-06,
      "loss": 0.8699,
      "step": 3529
    },
    {
      "epoch": 0.2589946531911186,
      "grad_norm": 0.7293353959973076,
      "learning_rate": 8.686698842720547e-06,
      "loss": 0.5172,
      "step": 3530
    },
    {
      "epoch": 0.25906802278125773,
      "grad_norm": 1.604794318839549,
      "learning_rate": 8.68589608538854e-06,
      "loss": 0.7678,
      "step": 3531
    },
    {
      "epoch": 0.2591413923713969,
      "grad_norm": 1.4100304248518931,
      "learning_rate": 8.685093119904852e-06,
      "loss": 0.8476,
      "step": 3532
    },
    {
      "epoch": 0.259214761961536,
      "grad_norm": 1.5845483737799593,
      "learning_rate": 8.684289946314832e-06,
      "loss": 0.8545,
      "step": 3533
    },
    {
      "epoch": 0.2592881315516751,
      "grad_norm": 1.6765641304610877,
      "learning_rate": 8.683486564663834e-06,
      "loss": 0.81,
      "step": 3534
    },
    {
      "epoch": 0.2593615011418142,
      "grad_norm": 1.4786587368275141,
      "learning_rate": 8.682682974997228e-06,
      "loss": 0.8041,
      "step": 3535
    },
    {
      "epoch": 0.25943487073195337,
      "grad_norm": 1.6801647767727963,
      "learning_rate": 8.681879177360393e-06,
      "loss": 0.9398,
      "step": 3536
    },
    {
      "epoch": 0.25943487073195337,
      "lm_loss": 0.1804,
      "step": 3536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25943487073195337,
      "lm_loss": 0.3793,
      "step": 3536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25943487073195337,
      "lm_loss": 0.588,
      "step": 3536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25943487073195337,
      "lm_loss": 0.4892,
      "step": 3536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25943487073195337,
      "lm_loss": 0.6091,
      "step": 3536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25943487073195337,
      "lm_loss": 0.5574,
      "step": 3536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25943487073195337,
      "lm_loss": 0.6632,
      "step": 3536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.25943487073195337,
      "lm_loss": 0.5971,
      "step": 3536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2595082403220925,
      "grad_norm": 0.8627579431320477,
      "learning_rate": 8.681075171798723e-06,
      "loss": 0.5195,
      "step": 3537
    },
    {
      "epoch": 0.2595816099122316,
      "grad_norm": 1.5967322159995732,
      "learning_rate": 8.680270958357622e-06,
      "loss": 0.8165,
      "step": 3538
    },
    {
      "epoch": 0.25965497950237076,
      "grad_norm": 0.7502170188571357,
      "learning_rate": 8.679466537082507e-06,
      "loss": 0.5281,
      "step": 3539
    },
    {
      "epoch": 0.25972834909250986,
      "grad_norm": 1.5504429941335245,
      "learning_rate": 8.678661908018803e-06,
      "loss": 0.7441,
      "step": 3540
    },
    {
      "epoch": 0.259801718682649,
      "grad_norm": 1.7314182737648622,
      "learning_rate": 8.677857071211953e-06,
      "loss": 0.8811,
      "step": 3541
    },
    {
      "epoch": 0.25987508827278816,
      "grad_norm": 1.7734071336358654,
      "learning_rate": 8.677052026707405e-06,
      "loss": 0.8668,
      "step": 3542
    },
    {
      "epoch": 0.25994845786292725,
      "grad_norm": 1.4564982302627554,
      "learning_rate": 8.676246774550623e-06,
      "loss": 0.915,
      "step": 3543
    },
    {
      "epoch": 0.2600218274530664,
      "grad_norm": 1.4980042155295863,
      "learning_rate": 8.675441314787081e-06,
      "loss": 0.8618,
      "step": 3544
    },
    {
      "epoch": 0.2600218274530664,
      "lm_loss": 0.2442,
      "step": 3544,
      "vm_loss": 0.0391
    },
    {
      "epoch": 0.2600218274530664,
      "lm_loss": 0.3731,
      "step": 3544,
      "vm_loss": 0.0884
    },
    {
      "epoch": 0.2600218274530664,
      "lm_loss": 0.9314,
      "step": 3544,
      "vm_loss": 0.1365
    },
    {
      "epoch": 0.2600218274530664,
      "lm_loss": 0.7823,
      "step": 3544,
      "vm_loss": 0.1826
    },
    {
      "epoch": 0.2600218274530664,
      "lm_loss": 0.1471,
      "step": 3544,
      "vm_loss": 0.0813
    },
    {
      "epoch": 0.2600218274530664,
      "lm_loss": 0.6185,
      "step": 3544,
      "vm_loss": 0.1751
    },
    {
      "epoch": 0.2600218274530664,
      "lm_loss": 0.8984,
      "step": 3544,
      "vm_loss": 0.1989
    },
    {
      "epoch": 0.2600218274530664,
      "lm_loss": 1.1866,
      "step": 3544,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.2600951970432055,
      "grad_norm": 1.5050780769613805,
      "learning_rate": 8.674635647462268e-06,
      "loss": 0.8207,
      "step": 3545
    },
    {
      "epoch": 0.26016856663334464,
      "grad_norm": 1.4119264797825828,
      "learning_rate": 8.673829772621678e-06,
      "loss": 0.8186,
      "step": 3546
    },
    {
      "epoch": 0.2602419362234838,
      "grad_norm": 1.6319166051888403,
      "learning_rate": 8.673023690310825e-06,
      "loss": 0.8069,
      "step": 3547
    },
    {
      "epoch": 0.2603153058136229,
      "grad_norm": 1.4412581249893435,
      "learning_rate": 8.672217400575224e-06,
      "loss": 0.8753,
      "step": 3548
    },
    {
      "epoch": 0.26038867540376204,
      "grad_norm": 1.1157594283935242,
      "learning_rate": 8.671410903460416e-06,
      "loss": 0.5289,
      "step": 3549
    },
    {
      "epoch": 0.26046204499390113,
      "grad_norm": 1.449328066790102,
      "learning_rate": 8.67060419901194e-06,
      "loss": 0.7602,
      "step": 3550
    },
    {
      "epoch": 0.2605354145840403,
      "grad_norm": 1.477254690745163,
      "learning_rate": 8.669797287275357e-06,
      "loss": 0.815,
      "step": 3551
    },
    {
      "epoch": 0.26060878417417943,
      "grad_norm": 0.6522591520287196,
      "learning_rate": 8.668990168296231e-06,
      "loss": 0.5214,
      "step": 3552
    },
    {
      "epoch": 0.26060878417417943,
      "lm_loss": 0.9701,
      "step": 3552,
      "vm_loss": 0.1047
    },
    {
      "epoch": 0.26060878417417943,
      "lm_loss": 0.5874,
      "step": 3552,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.26060878417417943,
      "lm_loss": 0.3601,
      "step": 3552,
      "vm_loss": 0.1618
    },
    {
      "epoch": 0.26060878417417943,
      "lm_loss": 0.6231,
      "step": 3552,
      "vm_loss": 0.1238
    },
    {
      "epoch": 0.26060878417417943,
      "lm_loss": 0.3913,
      "step": 3552,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.26060878417417943,
      "lm_loss": 0.1129,
      "step": 3552,
      "vm_loss": 0.0563
    },
    {
      "epoch": 0.26060878417417943,
      "lm_loss": 1.453,
      "step": 3552,
      "vm_loss": 0.2033
    },
    {
      "epoch": 0.26060878417417943,
      "lm_loss": 0.314,
      "step": 3552,
      "vm_loss": 0.0785
    },
    {
      "epoch": 0.2606821537643185,
      "grad_norm": 1.5265490683934688,
      "learning_rate": 8.668182842120145e-06,
      "loss": 0.8252,
      "step": 3553
    },
    {
      "epoch": 0.2607555233544577,
      "grad_norm": 1.4368909704246524,
      "learning_rate": 8.667375308792689e-06,
      "loss": 0.7554,
      "step": 3554
    },
    {
      "epoch": 0.26082889294459677,
      "grad_norm": 1.7220078632833158,
      "learning_rate": 8.666567568359467e-06,
      "loss": 0.8522,
      "step": 3555
    },
    {
      "epoch": 0.2609022625347359,
      "grad_norm": 1.6853118404428589,
      "learning_rate": 8.665759620866096e-06,
      "loss": 0.8576,
      "step": 3556
    },
    {
      "epoch": 0.26097563212487507,
      "grad_norm": 0.7154516794366707,
      "learning_rate": 8.664951466358201e-06,
      "loss": 0.5415,
      "step": 3557
    },
    {
      "epoch": 0.26104900171501416,
      "grad_norm": 1.5012322604011004,
      "learning_rate": 8.66414310488142e-06,
      "loss": 0.7954,
      "step": 3558
    },
    {
      "epoch": 0.2611223713051533,
      "grad_norm": 1.4918397557857939,
      "learning_rate": 8.663334536481402e-06,
      "loss": 0.8818,
      "step": 3559
    },
    {
      "epoch": 0.2611957408952924,
      "grad_norm": 1.7398532018264103,
      "learning_rate": 8.662525761203812e-06,
      "loss": 0.8671,
      "step": 3560
    },
    {
      "epoch": 0.2611957408952924,
      "lm_loss": 0.5644,
      "step": 3560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2611957408952924,
      "lm_loss": 0.5506,
      "step": 3560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2611957408952924,
      "lm_loss": 0.4423,
      "step": 3560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2611957408952924,
      "lm_loss": 0.5981,
      "step": 3560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2611957408952924,
      "lm_loss": 0.5795,
      "step": 3560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2611957408952924,
      "lm_loss": 0.6254,
      "step": 3560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2611957408952924,
      "lm_loss": 0.8003,
      "step": 3560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2611957408952924,
      "lm_loss": 0.7198,
      "step": 3560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26126911048543155,
      "grad_norm": 0.6182774605446882,
      "learning_rate": 8.661716779094324e-06,
      "loss": 0.5578,
      "step": 3561
    },
    {
      "epoch": 0.2613424800755707,
      "grad_norm": 0.5826864712380405,
      "learning_rate": 8.66090759019862e-06,
      "loss": 0.5327,
      "step": 3562
    },
    {
      "epoch": 0.2614158496657098,
      "grad_norm": 1.6757626838511164,
      "learning_rate": 8.660098194562397e-06,
      "loss": 0.8441,
      "step": 3563
    },
    {
      "epoch": 0.26148921925584895,
      "grad_norm": 1.7828623153243892,
      "learning_rate": 8.659288592231367e-06,
      "loss": 0.8566,
      "step": 3564
    },
    {
      "epoch": 0.26156258884598804,
      "grad_norm": 0.5914617594590542,
      "learning_rate": 8.658478783251246e-06,
      "loss": 0.5196,
      "step": 3565
    },
    {
      "epoch": 0.2616359584361272,
      "grad_norm": 1.6553688395274937,
      "learning_rate": 8.65766876766777e-06,
      "loss": 0.7692,
      "step": 3566
    },
    {
      "epoch": 0.2617093280262663,
      "grad_norm": 0.6165870845574725,
      "learning_rate": 8.656858545526679e-06,
      "loss": 0.5587,
      "step": 3567
    },
    {
      "epoch": 0.26178269761640544,
      "grad_norm": 1.360511917121265,
      "learning_rate": 8.656048116873731e-06,
      "loss": 0.8095,
      "step": 3568
    },
    {
      "epoch": 0.26178269761640544,
      "lm_loss": 1.2528,
      "step": 3568,
      "vm_loss": 0.0476
    },
    {
      "epoch": 0.26178269761640544,
      "lm_loss": 1.239,
      "step": 3568,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.26178269761640544,
      "lm_loss": 0.916,
      "step": 3568,
      "vm_loss": 0.1261
    },
    {
      "epoch": 0.26178269761640544,
      "lm_loss": 0.6358,
      "step": 3568,
      "vm_loss": 0.2264
    },
    {
      "epoch": 0.26178269761640544,
      "lm_loss": 1.1294,
      "step": 3568,
      "vm_loss": 0.1626
    },
    {
      "epoch": 0.26178269761640544,
      "lm_loss": 0.1898,
      "step": 3568,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.26178269761640544,
      "lm_loss": 0.7925,
      "step": 3568,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.26178269761640544,
      "lm_loss": 1.2466,
      "step": 3568,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.2618560672065446,
      "grad_norm": 1.6254566415175473,
      "learning_rate": 8.65523748175469e-06,
      "loss": 0.919,
      "step": 3569
    },
    {
      "epoch": 0.2619294367966837,
      "grad_norm": 1.5056085907192223,
      "learning_rate": 8.65442664021534e-06,
      "loss": 0.7541,
      "step": 3570
    },
    {
      "epoch": 0.26200280638682283,
      "grad_norm": 0.58808588703115,
      "learning_rate": 8.653615592301464e-06,
      "loss": 0.5217,
      "step": 3571
    },
    {
      "epoch": 0.2620761759769619,
      "grad_norm": 1.5006800646040308,
      "learning_rate": 8.65280433805887e-06,
      "loss": 0.8047,
      "step": 3572
    },
    {
      "epoch": 0.26214954556710107,
      "grad_norm": 0.5806918841809191,
      "learning_rate": 8.651992877533366e-06,
      "loss": 0.5557,
      "step": 3573
    },
    {
      "epoch": 0.2622229151572402,
      "grad_norm": 0.5832332443278831,
      "learning_rate": 8.651181210770782e-06,
      "loss": 0.5467,
      "step": 3574
    },
    {
      "epoch": 0.2622962847473793,
      "grad_norm": 0.5657265310450063,
      "learning_rate": 8.650369337816954e-06,
      "loss": 0.5267,
      "step": 3575
    },
    {
      "epoch": 0.26236965433751847,
      "grad_norm": 0.5466207547400299,
      "learning_rate": 8.649557258717728e-06,
      "loss": 0.5283,
      "step": 3576
    },
    {
      "epoch": 0.26236965433751847,
      "lm_loss": 1.2687,
      "step": 3576,
      "vm_loss": 0.0499
    },
    {
      "epoch": 0.26236965433751847,
      "lm_loss": 0.9425,
      "step": 3576,
      "vm_loss": 0.0527
    },
    {
      "epoch": 0.26236965433751847,
      "lm_loss": 0.5003,
      "step": 3576,
      "vm_loss": 0.1404
    },
    {
      "epoch": 0.26236965433751847,
      "lm_loss": 0.9612,
      "step": 3576,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.26236965433751847,
      "lm_loss": 0.5656,
      "step": 3576,
      "vm_loss": 0.0672
    },
    {
      "epoch": 0.26236965433751847,
      "lm_loss": 0.891,
      "step": 3576,
      "vm_loss": 0.1596
    },
    {
      "epoch": 0.26236965433751847,
      "lm_loss": 0.9492,
      "step": 3576,
      "vm_loss": 0.1131
    },
    {
      "epoch": 0.26236965433751847,
      "lm_loss": 0.1739,
      "step": 3576,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.26244302392765756,
      "grad_norm": 1.5756008715616656,
      "learning_rate": 8.648744973518967e-06,
      "loss": 0.8129,
      "step": 3577
    },
    {
      "epoch": 0.2625163935177967,
      "grad_norm": 0.5926515961683865,
      "learning_rate": 8.647932482266541e-06,
      "loss": 0.5356,
      "step": 3578
    },
    {
      "epoch": 0.26258976310793586,
      "grad_norm": 0.5800034353625485,
      "learning_rate": 8.647119785006333e-06,
      "loss": 0.549,
      "step": 3579
    },
    {
      "epoch": 0.26266313269807495,
      "grad_norm": 1.479351957783253,
      "learning_rate": 8.64630688178424e-06,
      "loss": 0.8321,
      "step": 3580
    },
    {
      "epoch": 0.2627365022882141,
      "grad_norm": 1.5664856854008906,
      "learning_rate": 8.645493772646165e-06,
      "loss": 0.8406,
      "step": 3581
    },
    {
      "epoch": 0.2628098718783532,
      "grad_norm": 1.7341620147143486,
      "learning_rate": 8.644680457638031e-06,
      "loss": 0.8812,
      "step": 3582
    },
    {
      "epoch": 0.26288324146849235,
      "grad_norm": 1.4112665812940781,
      "learning_rate": 8.643866936805765e-06,
      "loss": 0.8281,
      "step": 3583
    },
    {
      "epoch": 0.2629566110586315,
      "grad_norm": 1.5274330001557719,
      "learning_rate": 8.643053210195309e-06,
      "loss": 0.7799,
      "step": 3584
    },
    {
      "epoch": 0.2629566110586315,
      "lm_loss": 1.1593,
      "step": 3584,
      "vm_loss": 0.0735
    },
    {
      "epoch": 0.2629566110586315,
      "lm_loss": 0.183,
      "step": 3584,
      "vm_loss": 0.0929
    },
    {
      "epoch": 0.2629566110586315,
      "lm_loss": 0.1382,
      "step": 3584,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.2629566110586315,
      "lm_loss": 0.9887,
      "step": 3584,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.2629566110586315,
      "lm_loss": 0.9203,
      "step": 3584,
      "vm_loss": 0.19
    },
    {
      "epoch": 0.2629566110586315,
      "lm_loss": 1.0065,
      "step": 3584,
      "vm_loss": 0.1926
    },
    {
      "epoch": 0.2629566110586315,
      "lm_loss": 0.7274,
      "step": 3584,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.2629566110586315,
      "lm_loss": 1.1594,
      "step": 3584,
      "vm_loss": 0.0826
    },
    {
      "epoch": 0.2630299806487706,
      "grad_norm": 1.4697624638805602,
      "learning_rate": 8.642239277852618e-06,
      "loss": 0.856,
      "step": 3585
    },
    {
      "epoch": 0.26310335023890974,
      "grad_norm": 1.6403013724285578,
      "learning_rate": 8.641425139823652e-06,
      "loss": 0.8423,
      "step": 3586
    },
    {
      "epoch": 0.26317671982904883,
      "grad_norm": 1.8338831877622617,
      "learning_rate": 8.640610796154393e-06,
      "loss": 0.8933,
      "step": 3587
    },
    {
      "epoch": 0.263250089419188,
      "grad_norm": 0.6223518037482592,
      "learning_rate": 8.639796246890826e-06,
      "loss": 0.5142,
      "step": 3588
    },
    {
      "epoch": 0.26332345900932713,
      "grad_norm": 1.3533432012284354,
      "learning_rate": 8.63898149207895e-06,
      "loss": 0.883,
      "step": 3589
    },
    {
      "epoch": 0.2633968285994662,
      "grad_norm": 0.601567172942334,
      "learning_rate": 8.638166531764779e-06,
      "loss": 0.5241,
      "step": 3590
    },
    {
      "epoch": 0.2634701981896054,
      "grad_norm": 1.5853770936525302,
      "learning_rate": 8.637351365994333e-06,
      "loss": 0.8309,
      "step": 3591
    },
    {
      "epoch": 0.26354356777974447,
      "grad_norm": 1.3252618988910079,
      "learning_rate": 8.636535994813649e-06,
      "loss": 0.854,
      "step": 3592
    },
    {
      "epoch": 0.26354356777974447,
      "lm_loss": 0.5104,
      "step": 3592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26354356777974447,
      "lm_loss": 0.6634,
      "step": 3592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26354356777974447,
      "lm_loss": 0.4892,
      "step": 3592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26354356777974447,
      "lm_loss": 0.5458,
      "step": 3592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26354356777974447,
      "lm_loss": 0.6274,
      "step": 3592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26354356777974447,
      "lm_loss": 0.3983,
      "step": 3592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26354356777974447,
      "lm_loss": 0.461,
      "step": 3592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26354356777974447,
      "lm_loss": 0.5109,
      "step": 3592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2636169373698836,
      "grad_norm": 0.5953626311518065,
      "learning_rate": 8.63572041826877e-06,
      "loss": 0.5364,
      "step": 3593
    },
    {
      "epoch": 0.26369030696002277,
      "grad_norm": 0.5983494761198801,
      "learning_rate": 8.634904636405757e-06,
      "loss": 0.5575,
      "step": 3594
    },
    {
      "epoch": 0.26376367655016186,
      "grad_norm": 1.785357875392721,
      "learning_rate": 8.634088649270676e-06,
      "loss": 0.8675,
      "step": 3595
    },
    {
      "epoch": 0.263837046140301,
      "grad_norm": 1.4871791900345595,
      "learning_rate": 8.63327245690961e-06,
      "loss": 1.0067,
      "step": 3596
    },
    {
      "epoch": 0.2639104157304401,
      "grad_norm": 1.529416921503084,
      "learning_rate": 8.63245605936865e-06,
      "loss": 0.7353,
      "step": 3597
    },
    {
      "epoch": 0.26398378532057926,
      "grad_norm": 1.4979767639523431,
      "learning_rate": 8.631639456693901e-06,
      "loss": 0.8874,
      "step": 3598
    },
    {
      "epoch": 0.2640571549107184,
      "grad_norm": 1.565350284274876,
      "learning_rate": 8.63082264893148e-06,
      "loss": 0.7546,
      "step": 3599
    },
    {
      "epoch": 0.2641305245008575,
      "grad_norm": 0.6328938007534256,
      "learning_rate": 8.630005636127509e-06,
      "loss": 0.5414,
      "step": 3600
    },
    {
      "epoch": 0.2641305245008575,
      "lm_loss": 0.5519,
      "step": 3600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2641305245008575,
      "lm_loss": 0.4369,
      "step": 3600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2641305245008575,
      "lm_loss": 0.5687,
      "step": 3600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2641305245008575,
      "lm_loss": 0.7209,
      "step": 3600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2641305245008575,
      "lm_loss": 0.575,
      "step": 3600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2641305245008575,
      "lm_loss": 0.5381,
      "step": 3600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2641305245008575,
      "lm_loss": 0.3466,
      "step": 3600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2641305245008575,
      "lm_loss": 0.5191,
      "step": 3600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26420389409099665,
      "grad_norm": 0.5860062656808599,
      "learning_rate": 8.629188418328132e-06,
      "loss": 0.5161,
      "step": 3601
    },
    {
      "epoch": 0.26427726368113574,
      "grad_norm": 1.417369381256782,
      "learning_rate": 8.628370995579499e-06,
      "loss": 0.8954,
      "step": 3602
    },
    {
      "epoch": 0.2643506332712749,
      "grad_norm": 1.5425128685011975,
      "learning_rate": 8.627553367927768e-06,
      "loss": 0.8115,
      "step": 3603
    },
    {
      "epoch": 0.26442400286141404,
      "grad_norm": 0.5820382621477694,
      "learning_rate": 8.626735535419115e-06,
      "loss": 0.5432,
      "step": 3604
    },
    {
      "epoch": 0.26449737245155314,
      "grad_norm": 1.529627955326636,
      "learning_rate": 8.625917498099725e-06,
      "loss": 0.8348,
      "step": 3605
    },
    {
      "epoch": 0.2645707420416923,
      "grad_norm": 0.593741805271318,
      "learning_rate": 8.625099256015797e-06,
      "loss": 0.5336,
      "step": 3606
    },
    {
      "epoch": 0.2646441116318314,
      "grad_norm": 1.567098424216777,
      "learning_rate": 8.624280809213533e-06,
      "loss": 0.9153,
      "step": 3607
    },
    {
      "epoch": 0.26471748122197053,
      "grad_norm": 1.3976406443312908,
      "learning_rate": 8.623462157739157e-06,
      "loss": 0.9064,
      "step": 3608
    },
    {
      "epoch": 0.26471748122197053,
      "lm_loss": 1.3293,
      "step": 3608,
      "vm_loss": 0.0412
    },
    {
      "epoch": 0.26471748122197053,
      "lm_loss": 0.5434,
      "step": 3608,
      "vm_loss": 0.1839
    },
    {
      "epoch": 0.26471748122197053,
      "lm_loss": 0.0308,
      "step": 3608,
      "vm_loss": 0.131
    },
    {
      "epoch": 0.26471748122197053,
      "lm_loss": 0.5643,
      "step": 3608,
      "vm_loss": 0.1519
    },
    {
      "epoch": 0.26471748122197053,
      "lm_loss": 0.8753,
      "step": 3608,
      "vm_loss": 0.1015
    },
    {
      "epoch": 0.26471748122197053,
      "lm_loss": 0.3994,
      "step": 3608,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.26471748122197053,
      "lm_loss": 0.2107,
      "step": 3608,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.26471748122197053,
      "lm_loss": 0.6907,
      "step": 3608,
      "vm_loss": 0.0718
    },
    {
      "epoch": 0.2647908508121096,
      "grad_norm": 1.5503845290363734,
      "learning_rate": 8.622643301638902e-06,
      "loss": 0.7531,
      "step": 3609
    },
    {
      "epoch": 0.2648642204022488,
      "grad_norm": 1.834660542963872,
      "learning_rate": 8.621824240959006e-06,
      "loss": 1.0082,
      "step": 3610
    },
    {
      "epoch": 0.2649375899923879,
      "grad_norm": 0.6195649421206748,
      "learning_rate": 8.621004975745728e-06,
      "loss": 0.5471,
      "step": 3611
    },
    {
      "epoch": 0.265010959582527,
      "grad_norm": 1.542720883463997,
      "learning_rate": 8.62018550604533e-06,
      "loss": 0.7824,
      "step": 3612
    },
    {
      "epoch": 0.26508432917266617,
      "grad_norm": 1.5472990494907386,
      "learning_rate": 8.619365831904092e-06,
      "loss": 0.782,
      "step": 3613
    },
    {
      "epoch": 0.26515769876280526,
      "grad_norm": 1.629367812524642,
      "learning_rate": 8.618545953368302e-06,
      "loss": 0.9158,
      "step": 3614
    },
    {
      "epoch": 0.2652310683529444,
      "grad_norm": 1.5821042983575888,
      "learning_rate": 8.617725870484261e-06,
      "loss": 0.7958,
      "step": 3615
    },
    {
      "epoch": 0.26530443794308356,
      "grad_norm": 0.580045897151177,
      "learning_rate": 8.616905583298281e-06,
      "loss": 0.5306,
      "step": 3616
    },
    {
      "epoch": 0.26530443794308356,
      "lm_loss": 0.2939,
      "step": 3616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26530443794308356,
      "lm_loss": 0.4484,
      "step": 3616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26530443794308356,
      "lm_loss": 0.4692,
      "step": 3616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26530443794308356,
      "lm_loss": 0.4181,
      "step": 3616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26530443794308356,
      "lm_loss": 0.6136,
      "step": 3616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26530443794308356,
      "lm_loss": 0.7209,
      "step": 3616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26530443794308356,
      "lm_loss": 0.5881,
      "step": 3616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26530443794308356,
      "lm_loss": 0.4637,
      "step": 3616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.26537780753322265,
      "grad_norm": 0.5806749045737942,
      "learning_rate": 8.616085091856686e-06,
      "loss": 0.546,
      "step": 3617
    },
    {
      "epoch": 0.2654511771233618,
      "grad_norm": 1.507753267986289,
      "learning_rate": 8.615264396205809e-06,
      "loss": 0.8673,
      "step": 3618
    },
    {
      "epoch": 0.2655245467135009,
      "grad_norm": 1.495077045544152,
      "learning_rate": 8.614443496392e-06,
      "loss": 0.6983,
      "step": 3619
    },
    {
      "epoch": 0.26559791630364005,
      "grad_norm": 1.5437802082746523,
      "learning_rate": 8.613622392461615e-06,
      "loss": 0.779,
      "step": 3620
    },
    {
      "epoch": 0.2656712858937792,
      "grad_norm": 1.5529746099632427,
      "learning_rate": 8.612801084461026e-06,
      "loss": 0.8042,
      "step": 3621
    },
    {
      "epoch": 0.2657446554839183,
      "grad_norm": 1.5425515884884575,
      "learning_rate": 8.611979572436611e-06,
      "loss": 0.794,
      "step": 3622
    },
    {
      "epoch": 0.26581802507405744,
      "grad_norm": 1.4212544235615296,
      "learning_rate": 8.611157856434766e-06,
      "loss": 0.8803,
      "step": 3623
    },
    {
      "epoch": 0.26589139466419653,
      "grad_norm": 0.6114158691851536,
      "learning_rate": 8.610335936501892e-06,
      "loss": 0.524,
      "step": 3624
    },
    {
      "epoch": 0.26589139466419653,
      "lm_loss": 0.7833,
      "step": 3624,
      "vm_loss": 0.0395
    },
    {
      "epoch": 0.26589139466419653,
      "lm_loss": 0.2081,
      "step": 3624,
      "vm_loss": 0.1296
    },
    {
      "epoch": 0.26589139466419653,
      "lm_loss": 0.6414,
      "step": 3624,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.26589139466419653,
      "lm_loss": 0.828,
      "step": 3624,
      "vm_loss": 0.0857
    },
    {
      "epoch": 0.26589139466419653,
      "lm_loss": 0.0116,
      "step": 3624,
      "vm_loss": 0.1495
    },
    {
      "epoch": 0.26589139466419653,
      "lm_loss": 0.2267,
      "step": 3624,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.26589139466419653,
      "lm_loss": 0.0141,
      "step": 3624,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.26589139466419653,
      "lm_loss": 0.413,
      "step": 3624,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.2659647642543357,
      "grad_norm": 1.5273479812966066,
      "learning_rate": 8.60951381268441e-06,
      "loss": 0.6724,
      "step": 3625
    },
    {
      "epoch": 0.26603813384447483,
      "grad_norm": 1.4456100574134878,
      "learning_rate": 8.608691485028741e-06,
      "loss": 0.7125,
      "step": 3626
    },
    {
      "epoch": 0.2661115034346139,
      "grad_norm": 1.6272390564177832,
      "learning_rate": 8.607868953581329e-06,
      "loss": 0.8939,
      "step": 3627
    },
    {
      "epoch": 0.2661848730247531,
      "grad_norm": 1.5628018400051658,
      "learning_rate": 8.607046218388621e-06,
      "loss": 0.8829,
      "step": 3628
    },
    {
      "epoch": 0.26625824261489217,
      "grad_norm": 1.576166303964549,
      "learning_rate": 8.606223279497081e-06,
      "loss": 0.9084,
      "step": 3629
    },
    {
      "epoch": 0.2663316122050313,
      "grad_norm": 1.5163026261043597,
      "learning_rate": 8.605400136953183e-06,
      "loss": 0.9624,
      "step": 3630
    },
    {
      "epoch": 0.26640498179517047,
      "grad_norm": 0.5810813072612884,
      "learning_rate": 8.604576790803409e-06,
      "loss": 0.5181,
      "step": 3631
    },
    {
      "epoch": 0.26647835138530956,
      "grad_norm": 1.5522766553283527,
      "learning_rate": 8.603753241094258e-06,
      "loss": 0.9917,
      "step": 3632
    },
    {
      "epoch": 0.26647835138530956,
      "lm_loss": 1.2301,
      "step": 3632,
      "vm_loss": 0.0707
    },
    {
      "epoch": 0.26647835138530956,
      "lm_loss": 1.4319,
      "step": 3632,
      "vm_loss": 0.1893
    },
    {
      "epoch": 0.26647835138530956,
      "lm_loss": 0.1596,
      "step": 3632,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.26647835138530956,
      "lm_loss": 0.1931,
      "step": 3632,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.26647835138530956,
      "lm_loss": 0.1556,
      "step": 3632,
      "vm_loss": 0.0472
    },
    {
      "epoch": 0.26647835138530956,
      "lm_loss": 1.0314,
      "step": 3632,
      "vm_loss": 0.1515
    },
    {
      "epoch": 0.26647835138530956,
      "lm_loss": 0.7761,
      "step": 3632,
      "vm_loss": 0.148
    },
    {
      "epoch": 0.26647835138530956,
      "lm_loss": 0.0871,
      "step": 3632,
      "vm_loss": 0.1038
    },
    {
      "epoch": 0.2665517209754487,
      "grad_norm": 1.5775091135111927,
      "learning_rate": 8.602929487872237e-06,
      "loss": 0.8807,
      "step": 3633
    },
    {
      "epoch": 0.2666250905655878,
      "grad_norm": 0.5711780978450204,
      "learning_rate": 8.602105531183865e-06,
      "loss": 0.5151,
      "step": 3634
    },
    {
      "epoch": 0.26669846015572696,
      "grad_norm": 1.6321536643973034,
      "learning_rate": 8.601281371075674e-06,
      "loss": 0.7562,
      "step": 3635
    },
    {
      "epoch": 0.2667718297458661,
      "grad_norm": 1.4421330939371497,
      "learning_rate": 8.600457007594203e-06,
      "loss": 0.9555,
      "step": 3636
    },
    {
      "epoch": 0.2668451993360052,
      "grad_norm": 1.4693772717154707,
      "learning_rate": 8.59963244078601e-06,
      "loss": 0.8797,
      "step": 3637
    },
    {
      "epoch": 0.26691856892614435,
      "grad_norm": 0.583303692308677,
      "learning_rate": 8.59880767069766e-06,
      "loss": 0.5363,
      "step": 3638
    },
    {
      "epoch": 0.26699193851628344,
      "grad_norm": 0.6049011149352832,
      "learning_rate": 8.597982697375726e-06,
      "loss": 0.5421,
      "step": 3639
    },
    {
      "epoch": 0.2670653081064226,
      "grad_norm": 0.5665061510878131,
      "learning_rate": 8.5971575208668e-06,
      "loss": 0.5198,
      "step": 3640
    },
    {
      "epoch": 0.2670653081064226,
      "lm_loss": 1.6454,
      "step": 3640,
      "vm_loss": 0.0682
    },
    {
      "epoch": 0.2670653081064226,
      "lm_loss": 0.4895,
      "step": 3640,
      "vm_loss": 0.1112
    },
    {
      "epoch": 0.2670653081064226,
      "lm_loss": 1.9683,
      "step": 3640,
      "vm_loss": 0.1094
    },
    {
      "epoch": 0.2670653081064226,
      "lm_loss": 0.9629,
      "step": 3640,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.2670653081064226,
      "lm_loss": 0.0763,
      "step": 3640,
      "vm_loss": 0.1458
    },
    {
      "epoch": 0.2670653081064226,
      "lm_loss": 0.4207,
      "step": 3640,
      "vm_loss": 0.1423
    },
    {
      "epoch": 0.2670653081064226,
      "lm_loss": 0.5786,
      "step": 3640,
      "vm_loss": 0.039
    },
    {
      "epoch": 0.2670653081064226,
      "lm_loss": 0.5264,
      "step": 3640,
      "vm_loss": 0.1644
    },
    {
      "epoch": 0.26713867769656174,
      "grad_norm": 1.8089820160684826,
      "learning_rate": 8.596332141217482e-06,
      "loss": 0.7659,
      "step": 3641
    },
    {
      "epoch": 0.26721204728670084,
      "grad_norm": 1.6706884006151101,
      "learning_rate": 8.59550655847438e-06,
      "loss": 0.7388,
      "step": 3642
    },
    {
      "epoch": 0.26728541687684,
      "grad_norm": 1.5657606213489912,
      "learning_rate": 8.594680772684118e-06,
      "loss": 0.9111,
      "step": 3643
    },
    {
      "epoch": 0.2673587864669791,
      "grad_norm": 0.6055806397479547,
      "learning_rate": 8.593854783893333e-06,
      "loss": 0.5359,
      "step": 3644
    },
    {
      "epoch": 0.26743215605711823,
      "grad_norm": 1.719129369001171,
      "learning_rate": 8.593028592148667e-06,
      "loss": 0.914,
      "step": 3645
    },
    {
      "epoch": 0.2675055256472574,
      "grad_norm": 1.7766109031702058,
      "learning_rate": 8.592202197496778e-06,
      "loss": 0.867,
      "step": 3646
    },
    {
      "epoch": 0.2675788952373965,
      "grad_norm": 0.5905681113380168,
      "learning_rate": 8.591375599984337e-06,
      "loss": 0.5346,
      "step": 3647
    },
    {
      "epoch": 0.2676522648275356,
      "grad_norm": 1.5786523268479824,
      "learning_rate": 8.59054879965802e-06,
      "loss": 0.7339,
      "step": 3648
    },
    {
      "epoch": 0.2676522648275356,
      "lm_loss": 0.4788,
      "step": 3648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2676522648275356,
      "lm_loss": 0.4928,
      "step": 3648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2676522648275356,
      "lm_loss": 0.4814,
      "step": 3648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2676522648275356,
      "lm_loss": 0.4153,
      "step": 3648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2676522648275356,
      "lm_loss": 0.4241,
      "step": 3648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2676522648275356,
      "lm_loss": 0.514,
      "step": 3648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2676522648275356,
      "lm_loss": 0.4694,
      "step": 3648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2676522648275356,
      "lm_loss": 0.3619,
      "step": 3648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2677256344176747,
      "grad_norm": 0.616541617025365,
      "learning_rate": 8.589721796564521e-06,
      "loss": 0.5338,
      "step": 3649
    },
    {
      "epoch": 0.26779900400781387,
      "grad_norm": 1.7354698650012317,
      "learning_rate": 8.588894590750542e-06,
      "loss": 0.7327,
      "step": 3650
    },
    {
      "epoch": 0.26787237359795296,
      "grad_norm": 1.777643359905425,
      "learning_rate": 8.588067182262798e-06,
      "loss": 0.8707,
      "step": 3651
    },
    {
      "epoch": 0.2679457431880921,
      "grad_norm": 1.481468450765493,
      "learning_rate": 8.587239571148014e-06,
      "loss": 0.8118,
      "step": 3652
    },
    {
      "epoch": 0.26801911277823126,
      "grad_norm": 1.5618043932478765,
      "learning_rate": 8.586411757452928e-06,
      "loss": 0.8174,
      "step": 3653
    },
    {
      "epoch": 0.26809248236837036,
      "grad_norm": 1.5235445449713154,
      "learning_rate": 8.58558374122429e-06,
      "loss": 0.8784,
      "step": 3654
    },
    {
      "epoch": 0.2681658519585095,
      "grad_norm": 1.4839353456493665,
      "learning_rate": 8.584755522508856e-06,
      "loss": 0.9612,
      "step": 3655
    },
    {
      "epoch": 0.2682392215486486,
      "grad_norm": 1.5816461879729231,
      "learning_rate": 8.583927101353403e-06,
      "loss": 0.9236,
      "step": 3656
    },
    {
      "epoch": 0.2682392215486486,
      "lm_loss": 1.534,
      "step": 3656,
      "vm_loss": 0.08
    },
    {
      "epoch": 0.2682392215486486,
      "lm_loss": 0.2648,
      "step": 3656,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.2682392215486486,
      "lm_loss": 0.6965,
      "step": 3656,
      "vm_loss": 0.1794
    },
    {
      "epoch": 0.2682392215486486,
      "lm_loss": 0.8622,
      "step": 3656,
      "vm_loss": 0.1314
    },
    {
      "epoch": 0.2682392215486486,
      "lm_loss": 0.8746,
      "step": 3656,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.2682392215486486,
      "lm_loss": 0.7029,
      "step": 3656,
      "vm_loss": 0.126
    },
    {
      "epoch": 0.2682392215486486,
      "lm_loss": 0.4627,
      "step": 3656,
      "vm_loss": 0.1009
    },
    {
      "epoch": 0.2682392215486486,
      "lm_loss": 0.2464,
      "step": 3656,
      "vm_loss": 0.1755
    },
    {
      "epoch": 0.26831259113878775,
      "grad_norm": 1.6988601332218243,
      "learning_rate": 8.583098477804708e-06,
      "loss": 0.88,
      "step": 3657
    },
    {
      "epoch": 0.2683859607289269,
      "grad_norm": 1.4401480964251558,
      "learning_rate": 8.582269651909571e-06,
      "loss": 0.9741,
      "step": 3658
    },
    {
      "epoch": 0.268459330319066,
      "grad_norm": 1.7815726046497387,
      "learning_rate": 8.581440623714794e-06,
      "loss": 0.866,
      "step": 3659
    },
    {
      "epoch": 0.26853269990920514,
      "grad_norm": 1.3938415236826918,
      "learning_rate": 8.580611393267196e-06,
      "loss": 0.7823,
      "step": 3660
    },
    {
      "epoch": 0.26860606949934424,
      "grad_norm": 1.524980755283045,
      "learning_rate": 8.579781960613606e-06,
      "loss": 0.9697,
      "step": 3661
    },
    {
      "epoch": 0.2686794390894834,
      "grad_norm": 1.513043108476052,
      "learning_rate": 8.578952325800863e-06,
      "loss": 0.8788,
      "step": 3662
    },
    {
      "epoch": 0.26875280867962253,
      "grad_norm": 0.6841924567588402,
      "learning_rate": 8.57812248887582e-06,
      "loss": 0.5163,
      "step": 3663
    },
    {
      "epoch": 0.26882617826976163,
      "grad_norm": 1.460395586915183,
      "learning_rate": 8.577292449885338e-06,
      "loss": 0.8194,
      "step": 3664
    },
    {
      "epoch": 0.26882617826976163,
      "lm_loss": 1.8716,
      "step": 3664,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.26882617826976163,
      "lm_loss": 0.7764,
      "step": 3664,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.26882617826976163,
      "lm_loss": 0.7517,
      "step": 3664,
      "vm_loss": 0.1017
    },
    {
      "epoch": 0.26882617826976163,
      "lm_loss": 0.0561,
      "step": 3664,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.26882617826976163,
      "lm_loss": 0.07,
      "step": 3664,
      "vm_loss": 0.1141
    },
    {
      "epoch": 0.26882617826976163,
      "lm_loss": 0.9417,
      "step": 3664,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.26882617826976163,
      "lm_loss": 0.2814,
      "step": 3664,
      "vm_loss": 0.1237
    },
    {
      "epoch": 0.26882617826976163,
      "lm_loss": 1.0019,
      "step": 3664,
      "vm_loss": 0.2042
    },
    {
      "epoch": 0.2688995478599008,
      "grad_norm": 1.4848986580381835,
      "learning_rate": 8.576462208876293e-06,
      "loss": 0.7427,
      "step": 3665
    },
    {
      "epoch": 0.2689729174500399,
      "grad_norm": 0.5808932688197369,
      "learning_rate": 8.57563176589557e-06,
      "loss": 0.5271,
      "step": 3666
    },
    {
      "epoch": 0.269046287040179,
      "grad_norm": 1.6470611005509508,
      "learning_rate": 8.574801120990063e-06,
      "loss": 0.805,
      "step": 3667
    },
    {
      "epoch": 0.26911965663031817,
      "grad_norm": 0.5778739433928407,
      "learning_rate": 8.573970274206689e-06,
      "loss": 0.5293,
      "step": 3668
    },
    {
      "epoch": 0.26919302622045727,
      "grad_norm": 1.6701781360977064,
      "learning_rate": 8.57313922559236e-06,
      "loss": 0.8928,
      "step": 3669
    },
    {
      "epoch": 0.2692663958105964,
      "grad_norm": 1.5542291541982862,
      "learning_rate": 8.57230797519401e-06,
      "loss": 0.9192,
      "step": 3670
    },
    {
      "epoch": 0.2693397654007355,
      "grad_norm": 1.4927948673865663,
      "learning_rate": 8.571476523058585e-06,
      "loss": 0.8398,
      "step": 3671
    },
    {
      "epoch": 0.26941313499087466,
      "grad_norm": 1.755398751091629,
      "learning_rate": 8.570644869233031e-06,
      "loss": 0.8555,
      "step": 3672
    },
    {
      "epoch": 0.26941313499087466,
      "lm_loss": 1.7082,
      "step": 3672,
      "vm_loss": 0.0459
    },
    {
      "epoch": 0.26941313499087466,
      "lm_loss": 0.587,
      "step": 3672,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.26941313499087466,
      "lm_loss": 0.1836,
      "step": 3672,
      "vm_loss": 0.0894
    },
    {
      "epoch": 0.26941313499087466,
      "lm_loss": 0.3223,
      "step": 3672,
      "vm_loss": 0.1548
    },
    {
      "epoch": 0.26941313499087466,
      "lm_loss": 0.3534,
      "step": 3672,
      "vm_loss": 0.2243
    },
    {
      "epoch": 0.26941313499087466,
      "lm_loss": 0.6068,
      "step": 3672,
      "vm_loss": 0.217
    },
    {
      "epoch": 0.26941313499087466,
      "lm_loss": 0.5342,
      "step": 3672,
      "vm_loss": 0.1785
    },
    {
      "epoch": 0.26941313499087466,
      "lm_loss": 0.1866,
      "step": 3672,
      "vm_loss": 0.0935
    },
    {
      "epoch": 0.2694865045810138,
      "grad_norm": 1.7293439356391855,
      "learning_rate": 8.569813013764324e-06,
      "loss": 0.815,
      "step": 3673
    },
    {
      "epoch": 0.2695598741711529,
      "grad_norm": 1.4724758277547663,
      "learning_rate": 8.568980956699432e-06,
      "loss": 0.8782,
      "step": 3674
    },
    {
      "epoch": 0.26963324376129205,
      "grad_norm": 1.5262795732057182,
      "learning_rate": 8.568148698085347e-06,
      "loss": 0.784,
      "step": 3675
    },
    {
      "epoch": 0.26970661335143115,
      "grad_norm": 1.51456396447651,
      "learning_rate": 8.56731623796907e-06,
      "loss": 0.8367,
      "step": 3676
    },
    {
      "epoch": 0.2697799829415703,
      "grad_norm": 1.5468497790625944,
      "learning_rate": 8.566483576397611e-06,
      "loss": 0.8488,
      "step": 3677
    },
    {
      "epoch": 0.26985335253170945,
      "grad_norm": 0.663907074493647,
      "learning_rate": 8.565650713417992e-06,
      "loss": 0.5367,
      "step": 3678
    },
    {
      "epoch": 0.26992672212184854,
      "grad_norm": 1.5251255848319312,
      "learning_rate": 8.564817649077246e-06,
      "loss": 0.858,
      "step": 3679
    },
    {
      "epoch": 0.2700000917119877,
      "grad_norm": 0.6147000489405677,
      "learning_rate": 8.56398438342242e-06,
      "loss": 0.5321,
      "step": 3680
    },
    {
      "epoch": 0.2700000917119877,
      "lm_loss": 0.5237,
      "step": 3680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2700000917119877,
      "lm_loss": 0.5293,
      "step": 3680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2700000917119877,
      "lm_loss": 0.477,
      "step": 3680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2700000917119877,
      "lm_loss": 0.6299,
      "step": 3680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2700000917119877,
      "lm_loss": 0.4453,
      "step": 3680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2700000917119877,
      "lm_loss": 0.4816,
      "step": 3680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2700000917119877,
      "lm_loss": 0.6677,
      "step": 3680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2700000917119877,
      "lm_loss": 0.5103,
      "step": 3680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2700734613021268,
      "grad_norm": 0.5781888465890866,
      "learning_rate": 8.56315091650057e-06,
      "loss": 0.5354,
      "step": 3681
    },
    {
      "epoch": 0.27014683089226593,
      "grad_norm": 1.3641467371346476,
      "learning_rate": 8.562317248358763e-06,
      "loss": 0.7483,
      "step": 3682
    },
    {
      "epoch": 0.2702202004824051,
      "grad_norm": 1.5666373804118494,
      "learning_rate": 8.56148337904408e-06,
      "loss": 0.8123,
      "step": 3683
    },
    {
      "epoch": 0.2702935700725442,
      "grad_norm": 1.3814895183134808,
      "learning_rate": 8.56064930860361e-06,
      "loss": 0.7942,
      "step": 3684
    },
    {
      "epoch": 0.2703669396626833,
      "grad_norm": 1.422468572957216,
      "learning_rate": 8.559815037084456e-06,
      "loss": 0.8293,
      "step": 3685
    },
    {
      "epoch": 0.2704403092528224,
      "grad_norm": 1.4165187824033705,
      "learning_rate": 8.558980564533731e-06,
      "loss": 0.8541,
      "step": 3686
    },
    {
      "epoch": 0.27051367884296157,
      "grad_norm": 1.6264669166801238,
      "learning_rate": 8.55814589099856e-06,
      "loss": 0.8377,
      "step": 3687
    },
    {
      "epoch": 0.27058704843310066,
      "grad_norm": 0.8994413994358417,
      "learning_rate": 8.557311016526079e-06,
      "loss": 0.5431,
      "step": 3688
    },
    {
      "epoch": 0.27058704843310066,
      "lm_loss": 0.5917,
      "step": 3688,
      "vm_loss": 0.0683
    },
    {
      "epoch": 0.27058704843310066,
      "lm_loss": 1.2175,
      "step": 3688,
      "vm_loss": 0.0809
    },
    {
      "epoch": 0.27058704843310066,
      "lm_loss": 0.7219,
      "step": 3688,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.27058704843310066,
      "lm_loss": 0.2471,
      "step": 3688,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.27058704843310066,
      "lm_loss": 0.3223,
      "step": 3688,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.27058704843310066,
      "lm_loss": 0.5577,
      "step": 3688,
      "vm_loss": 0.0495
    },
    {
      "epoch": 0.27058704843310066,
      "lm_loss": 0.7586,
      "step": 3688,
      "vm_loss": 0.1027
    },
    {
      "epoch": 0.27058704843310066,
      "lm_loss": 0.4063,
      "step": 3688,
      "vm_loss": 0.1157
    },
    {
      "epoch": 0.2706604180232398,
      "grad_norm": 1.6461481697614624,
      "learning_rate": 8.556475941163436e-06,
      "loss": 0.8272,
      "step": 3689
    },
    {
      "epoch": 0.27073378761337896,
      "grad_norm": 1.6814567554946886,
      "learning_rate": 8.555640664957787e-06,
      "loss": 0.8778,
      "step": 3690
    },
    {
      "epoch": 0.27080715720351806,
      "grad_norm": 1.3894299543473971,
      "learning_rate": 8.554805187956306e-06,
      "loss": 0.7546,
      "step": 3691
    },
    {
      "epoch": 0.2708805267936572,
      "grad_norm": 0.6161561631093264,
      "learning_rate": 8.553969510206172e-06,
      "loss": 0.5153,
      "step": 3692
    },
    {
      "epoch": 0.2709538963837963,
      "grad_norm": 0.5964087741903572,
      "learning_rate": 8.55313363175458e-06,
      "loss": 0.5229,
      "step": 3693
    },
    {
      "epoch": 0.27102726597393545,
      "grad_norm": 1.6863268421381072,
      "learning_rate": 8.552297552648731e-06,
      "loss": 0.7819,
      "step": 3694
    },
    {
      "epoch": 0.2711006355640746,
      "grad_norm": 0.6506902968766284,
      "learning_rate": 8.551461272935843e-06,
      "loss": 0.5439,
      "step": 3695
    },
    {
      "epoch": 0.2711740051542137,
      "grad_norm": 1.556713853140091,
      "learning_rate": 8.550624792663142e-06,
      "loss": 0.784,
      "step": 3696
    },
    {
      "epoch": 0.2711740051542137,
      "lm_loss": 1.5992,
      "step": 3696,
      "vm_loss": 0.0544
    },
    {
      "epoch": 0.2711740051542137,
      "lm_loss": 0.6208,
      "step": 3696,
      "vm_loss": 0.1215
    },
    {
      "epoch": 0.2711740051542137,
      "lm_loss": 0.9327,
      "step": 3696,
      "vm_loss": 0.0686
    },
    {
      "epoch": 0.2711740051542137,
      "lm_loss": 0.7764,
      "step": 3696,
      "vm_loss": 0.151
    },
    {
      "epoch": 0.2711740051542137,
      "lm_loss": 1.072,
      "step": 3696,
      "vm_loss": 0.1686
    },
    {
      "epoch": 0.2711740051542137,
      "lm_loss": 0.3153,
      "step": 3696,
      "vm_loss": 0.0967
    },
    {
      "epoch": 0.2711740051542137,
      "lm_loss": 0.4051,
      "step": 3696,
      "vm_loss": 0.1424
    },
    {
      "epoch": 0.2711740051542137,
      "lm_loss": 0.1609,
      "step": 3696,
      "vm_loss": 0.1517
    },
    {
      "epoch": 0.27124737474435284,
      "grad_norm": 1.6280147927129012,
      "learning_rate": 8.549788111877866e-06,
      "loss": 0.7673,
      "step": 3697
    },
    {
      "epoch": 0.27132074433449194,
      "grad_norm": 1.5332018743399016,
      "learning_rate": 8.548951230627263e-06,
      "loss": 0.8689,
      "step": 3698
    },
    {
      "epoch": 0.2713941139246311,
      "grad_norm": 1.4958362281722686,
      "learning_rate": 8.548114148958596e-06,
      "loss": 0.7873,
      "step": 3699
    },
    {
      "epoch": 0.27146748351477024,
      "grad_norm": 1.4508436344060802,
      "learning_rate": 8.547276866919136e-06,
      "loss": 0.7421,
      "step": 3700
    },
    {
      "epoch": 0.27154085310490933,
      "grad_norm": 1.6822287909567868,
      "learning_rate": 8.546439384556168e-06,
      "loss": 0.8652,
      "step": 3701
    },
    {
      "epoch": 0.2716142226950485,
      "grad_norm": 1.537797267897428,
      "learning_rate": 8.545601701916984e-06,
      "loss": 0.8695,
      "step": 3702
    },
    {
      "epoch": 0.2716875922851876,
      "grad_norm": 1.6166890753978211,
      "learning_rate": 8.544763819048892e-06,
      "loss": 0.9231,
      "step": 3703
    },
    {
      "epoch": 0.2717609618753267,
      "grad_norm": 1.6276303984443887,
      "learning_rate": 8.543925735999208e-06,
      "loss": 0.884,
      "step": 3704
    },
    {
      "epoch": 0.2717609618753267,
      "lm_loss": 1.3751,
      "step": 3704,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.2717609618753267,
      "lm_loss": 0.0923,
      "step": 3704,
      "vm_loss": 0.1662
    },
    {
      "epoch": 0.2717609618753267,
      "lm_loss": 1.3955,
      "step": 3704,
      "vm_loss": 0.1113
    },
    {
      "epoch": 0.2717609618753267,
      "lm_loss": 0.3001,
      "step": 3704,
      "vm_loss": 0.171
    },
    {
      "epoch": 0.2717609618753267,
      "lm_loss": 0.9006,
      "step": 3704,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.2717609618753267,
      "lm_loss": 1.1805,
      "step": 3704,
      "vm_loss": 0.2214
    },
    {
      "epoch": 0.2717609618753267,
      "lm_loss": 0.7346,
      "step": 3704,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.2717609618753267,
      "lm_loss": 0.1095,
      "step": 3704,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.2718343314654659,
      "grad_norm": 1.5536048092685262,
      "learning_rate": 8.543087452815263e-06,
      "loss": 0.8609,
      "step": 3705
    },
    {
      "epoch": 0.27190770105560497,
      "grad_norm": 1.5263971474944777,
      "learning_rate": 8.542248969544392e-06,
      "loss": 0.8307,
      "step": 3706
    },
    {
      "epoch": 0.2719810706457441,
      "grad_norm": 1.5590714488588462,
      "learning_rate": 8.54141028623395e-06,
      "loss": 0.8426,
      "step": 3707
    },
    {
      "epoch": 0.2720544402358832,
      "grad_norm": 1.5104930000213843,
      "learning_rate": 8.540571402931301e-06,
      "loss": 0.8869,
      "step": 3708
    },
    {
      "epoch": 0.27212780982602236,
      "grad_norm": 1.468153165715172,
      "learning_rate": 8.539732319683817e-06,
      "loss": 0.7877,
      "step": 3709
    },
    {
      "epoch": 0.2722011794161615,
      "grad_norm": 0.7480286991694404,
      "learning_rate": 8.538893036538882e-06,
      "loss": 0.5195,
      "step": 3710
    },
    {
      "epoch": 0.2722745490063006,
      "grad_norm": 1.4156492386277124,
      "learning_rate": 8.538053553543893e-06,
      "loss": 0.7652,
      "step": 3711
    },
    {
      "epoch": 0.27234791859643975,
      "grad_norm": 0.6662119927628222,
      "learning_rate": 8.537213870746259e-06,
      "loss": 0.5382,
      "step": 3712
    },
    {
      "epoch": 0.27234791859643975,
      "lm_loss": 0.957,
      "step": 3712,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.27234791859643975,
      "lm_loss": 0.1291,
      "step": 3712,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.27234791859643975,
      "lm_loss": 0.5926,
      "step": 3712,
      "vm_loss": 0.1414
    },
    {
      "epoch": 0.27234791859643975,
      "lm_loss": 0.8569,
      "step": 3712,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.27234791859643975,
      "lm_loss": 0.4886,
      "step": 3712,
      "vm_loss": 0.0782
    },
    {
      "epoch": 0.27234791859643975,
      "lm_loss": 0.4623,
      "step": 3712,
      "vm_loss": 0.0848
    },
    {
      "epoch": 0.27234791859643975,
      "lm_loss": 1.1336,
      "step": 3712,
      "vm_loss": 0.1458
    },
    {
      "epoch": 0.27234791859643975,
      "lm_loss": 0.0865,
      "step": 3712,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.27242128818657885,
      "grad_norm": 1.477282412639545,
      "learning_rate": 8.536373988193396e-06,
      "loss": 0.7733,
      "step": 3713
    },
    {
      "epoch": 0.272494657776718,
      "grad_norm": 1.5482163022430606,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.7196,
      "step": 3714
    },
    {
      "epoch": 0.27256802736685715,
      "grad_norm": 1.548563425007376,
      "learning_rate": 8.534693624011726e-06,
      "loss": 0.7679,
      "step": 3715
    },
    {
      "epoch": 0.27264139695699624,
      "grad_norm": 1.7390819345527404,
      "learning_rate": 8.533853142477809e-06,
      "loss": 0.8941,
      "step": 3716
    },
    {
      "epoch": 0.2727147665471354,
      "grad_norm": 1.563055534302121,
      "learning_rate": 8.533012461378454e-06,
      "loss": 0.9008,
      "step": 3717
    },
    {
      "epoch": 0.2727881361372745,
      "grad_norm": 0.7501911035251739,
      "learning_rate": 8.532171580761135e-06,
      "loss": 0.5334,
      "step": 3718
    },
    {
      "epoch": 0.27286150572741363,
      "grad_norm": 1.5594538983555284,
      "learning_rate": 8.531330500673341e-06,
      "loss": 0.9168,
      "step": 3719
    },
    {
      "epoch": 0.2729348753175528,
      "grad_norm": 1.4704315595526312,
      "learning_rate": 8.53048922116257e-06,
      "loss": 0.8653,
      "step": 3720
    },
    {
      "epoch": 0.2729348753175528,
      "lm_loss": 0.5906,
      "step": 3720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2729348753175528,
      "lm_loss": 0.5421,
      "step": 3720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2729348753175528,
      "lm_loss": 0.689,
      "step": 3720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2729348753175528,
      "lm_loss": 0.726,
      "step": 3720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2729348753175528,
      "lm_loss": 0.5164,
      "step": 3720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2729348753175528,
      "lm_loss": 0.4845,
      "step": 3720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2729348753175528,
      "lm_loss": 0.575,
      "step": 3720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2729348753175528,
      "lm_loss": 0.5893,
      "step": 3720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2730082449076919,
      "grad_norm": 0.6465278223888473,
      "learning_rate": 8.529647742276326e-06,
      "loss": 0.5394,
      "step": 3721
    },
    {
      "epoch": 0.273081614497831,
      "grad_norm": 1.5253843139291852,
      "learning_rate": 8.528806064062137e-06,
      "loss": 0.7903,
      "step": 3722
    },
    {
      "epoch": 0.2731549840879701,
      "grad_norm": 1.7021341614761842,
      "learning_rate": 8.527964186567528e-06,
      "loss": 0.8459,
      "step": 3723
    },
    {
      "epoch": 0.27322835367810927,
      "grad_norm": 1.3972295839776354,
      "learning_rate": 8.527122109840045e-06,
      "loss": 0.6817,
      "step": 3724
    },
    {
      "epoch": 0.2733017232682484,
      "grad_norm": 1.5705418395569004,
      "learning_rate": 8.52627983392724e-06,
      "loss": 0.8402,
      "step": 3725
    },
    {
      "epoch": 0.2733750928583875,
      "grad_norm": 0.6354822334888242,
      "learning_rate": 8.525437358876683e-06,
      "loss": 0.5173,
      "step": 3726
    },
    {
      "epoch": 0.27344846244852666,
      "grad_norm": 1.3756722613127845,
      "learning_rate": 8.524594684735946e-06,
      "loss": 0.7987,
      "step": 3727
    },
    {
      "epoch": 0.27352183203866576,
      "grad_norm": 1.615891864772541,
      "learning_rate": 8.523751811552618e-06,
      "loss": 0.8515,
      "step": 3728
    },
    {
      "epoch": 0.27352183203866576,
      "lm_loss": 1.139,
      "step": 3728,
      "vm_loss": 0.0651
    },
    {
      "epoch": 0.27352183203866576,
      "lm_loss": 0.779,
      "step": 3728,
      "vm_loss": 0.2038
    },
    {
      "epoch": 0.27352183203866576,
      "lm_loss": 0.682,
      "step": 3728,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.27352183203866576,
      "lm_loss": 0.7016,
      "step": 3728,
      "vm_loss": 0.2194
    },
    {
      "epoch": 0.27352183203866576,
      "lm_loss": 0.3731,
      "step": 3728,
      "vm_loss": 0.1855
    },
    {
      "epoch": 0.27352183203866576,
      "lm_loss": 0.9858,
      "step": 3728,
      "vm_loss": 0.1068
    },
    {
      "epoch": 0.27352183203866576,
      "lm_loss": 1.0998,
      "step": 3728,
      "vm_loss": 0.2235
    },
    {
      "epoch": 0.27352183203866576,
      "lm_loss": 0.0511,
      "step": 3728,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.2735952016288049,
      "grad_norm": 1.3821255634764331,
      "learning_rate": 8.5229087393743e-06,
      "loss": 0.9196,
      "step": 3729
    },
    {
      "epoch": 0.273668571218944,
      "grad_norm": 1.5241218772122564,
      "learning_rate": 8.522065468248602e-06,
      "loss": 0.8351,
      "step": 3730
    },
    {
      "epoch": 0.27374194080908315,
      "grad_norm": 1.4785209383170785,
      "learning_rate": 8.521221998223144e-06,
      "loss": 0.7514,
      "step": 3731
    },
    {
      "epoch": 0.2738153103992223,
      "grad_norm": 1.4954485745687618,
      "learning_rate": 8.52037832934556e-06,
      "loss": 0.8043,
      "step": 3732
    },
    {
      "epoch": 0.2738886799893614,
      "grad_norm": 1.458776802781226,
      "learning_rate": 8.519534461663492e-06,
      "loss": 0.8463,
      "step": 3733
    },
    {
      "epoch": 0.27396204957950054,
      "grad_norm": 1.647248556999029,
      "learning_rate": 8.518690395224599e-06,
      "loss": 0.942,
      "step": 3734
    },
    {
      "epoch": 0.27403541916963964,
      "grad_norm": 1.5054626500577586,
      "learning_rate": 8.517846130076543e-06,
      "loss": 0.8239,
      "step": 3735
    },
    {
      "epoch": 0.2741087887597788,
      "grad_norm": 0.6722984784882349,
      "learning_rate": 8.517001666267007e-06,
      "loss": 0.5226,
      "step": 3736
    },
    {
      "epoch": 0.2741087887597788,
      "lm_loss": 1.2681,
      "step": 3736,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.2741087887597788,
      "lm_loss": 0.9189,
      "step": 3736,
      "vm_loss": 0.0893
    },
    {
      "epoch": 0.2741087887597788,
      "lm_loss": 0.1289,
      "step": 3736,
      "vm_loss": 0.092
    },
    {
      "epoch": 0.2741087887597788,
      "lm_loss": 1.281,
      "step": 3736,
      "vm_loss": 0.0695
    },
    {
      "epoch": 0.2741087887597788,
      "lm_loss": 1.0348,
      "step": 3736,
      "vm_loss": 0.1596
    },
    {
      "epoch": 0.2741087887597788,
      "lm_loss": 1.2731,
      "step": 3736,
      "vm_loss": 0.0958
    },
    {
      "epoch": 0.2741087887597788,
      "lm_loss": 0.4394,
      "step": 3736,
      "vm_loss": 0.0791
    },
    {
      "epoch": 0.2741087887597788,
      "lm_loss": 0.9385,
      "step": 3736,
      "vm_loss": 0.1462
    },
    {
      "epoch": 0.27418215834991794,
      "grad_norm": 1.561702108480245,
      "learning_rate": 8.516157003843675e-06,
      "loss": 0.8261,
      "step": 3737
    },
    {
      "epoch": 0.27425552794005703,
      "grad_norm": 0.621736959058232,
      "learning_rate": 8.51531214285425e-06,
      "loss": 0.5401,
      "step": 3738
    },
    {
      "epoch": 0.2743288975301962,
      "grad_norm": 1.6152986149628152,
      "learning_rate": 8.514467083346445e-06,
      "loss": 0.8151,
      "step": 3739
    },
    {
      "epoch": 0.2744022671203353,
      "grad_norm": 1.5093871593433583,
      "learning_rate": 8.513621825367977e-06,
      "loss": 0.942,
      "step": 3740
    },
    {
      "epoch": 0.2744756367104744,
      "grad_norm": 0.5877928813626994,
      "learning_rate": 8.512776368966585e-06,
      "loss": 0.53,
      "step": 3741
    },
    {
      "epoch": 0.2745490063006136,
      "grad_norm": 1.6410945054279762,
      "learning_rate": 8.511930714190013e-06,
      "loss": 0.7467,
      "step": 3742
    },
    {
      "epoch": 0.27462237589075267,
      "grad_norm": 1.668962222264633,
      "learning_rate": 8.511084861086015e-06,
      "loss": 0.8457,
      "step": 3743
    },
    {
      "epoch": 0.2746957454808918,
      "grad_norm": 1.6518206435670633,
      "learning_rate": 8.51023880970236e-06,
      "loss": 0.833,
      "step": 3744
    },
    {
      "epoch": 0.2746957454808918,
      "lm_loss": 0.4148,
      "step": 3744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2746957454808918,
      "lm_loss": 0.4762,
      "step": 3744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2746957454808918,
      "lm_loss": 0.5672,
      "step": 3744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2746957454808918,
      "lm_loss": 0.5242,
      "step": 3744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2746957454808918,
      "lm_loss": 0.6165,
      "step": 3744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2746957454808918,
      "lm_loss": 0.4114,
      "step": 3744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2746957454808918,
      "lm_loss": 0.4805,
      "step": 3744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2746957454808918,
      "lm_loss": 0.3842,
      "step": 3744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2747691150710309,
      "grad_norm": 0.6283779790656475,
      "learning_rate": 8.509392560086826e-06,
      "loss": 0.5285,
      "step": 3745
    },
    {
      "epoch": 0.27484248466117006,
      "grad_norm": 1.5181610867266095,
      "learning_rate": 8.508546112287203e-06,
      "loss": 0.8387,
      "step": 3746
    },
    {
      "epoch": 0.2749158542513092,
      "grad_norm": 1.6021660075550952,
      "learning_rate": 8.507699466351294e-06,
      "loss": 0.8602,
      "step": 3747
    },
    {
      "epoch": 0.2749892238414483,
      "grad_norm": 0.5995995052359945,
      "learning_rate": 8.506852622326909e-06,
      "loss": 0.5319,
      "step": 3748
    },
    {
      "epoch": 0.27506259343158745,
      "grad_norm": 1.5695059617628455,
      "learning_rate": 8.506005580261872e-06,
      "loss": 0.7462,
      "step": 3749
    },
    {
      "epoch": 0.27513596302172655,
      "grad_norm": 1.6978389273701135,
      "learning_rate": 8.505158340204017e-06,
      "loss": 0.9298,
      "step": 3750
    },
    {
      "epoch": 0.2752093326118657,
      "grad_norm": 1.608939110774477,
      "learning_rate": 8.504310902201189e-06,
      "loss": 0.8024,
      "step": 3751
    },
    {
      "epoch": 0.27528270220200485,
      "grad_norm": 1.572515752935076,
      "learning_rate": 8.503463266301248e-06,
      "loss": 0.8204,
      "step": 3752
    },
    {
      "epoch": 0.27528270220200485,
      "lm_loss": 0.6012,
      "step": 3752,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27528270220200485,
      "lm_loss": 0.5683,
      "step": 3752,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27528270220200485,
      "lm_loss": 0.521,
      "step": 3752,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27528270220200485,
      "lm_loss": 0.4173,
      "step": 3752,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27528270220200485,
      "lm_loss": 0.5696,
      "step": 3752,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27528270220200485,
      "lm_loss": 0.703,
      "step": 3752,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27528270220200485,
      "lm_loss": 0.5006,
      "step": 3752,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27528270220200485,
      "lm_loss": 0.6264,
      "step": 3752,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27535607179214394,
      "grad_norm": 0.6246576730582504,
      "learning_rate": 8.502615432552059e-06,
      "loss": 0.5363,
      "step": 3753
    },
    {
      "epoch": 0.2754294413822831,
      "grad_norm": 1.6577231520312354,
      "learning_rate": 8.501767401001505e-06,
      "loss": 0.828,
      "step": 3754
    },
    {
      "epoch": 0.2755028109724222,
      "grad_norm": 1.3321751918887026,
      "learning_rate": 8.50091917169747e-06,
      "loss": 0.8759,
      "step": 3755
    },
    {
      "epoch": 0.27557618056256133,
      "grad_norm": 1.5236135996338755,
      "learning_rate": 8.50007074468786e-06,
      "loss": 0.8395,
      "step": 3756
    },
    {
      "epoch": 0.2756495501527005,
      "grad_norm": 1.6362217426953007,
      "learning_rate": 8.49922212002059e-06,
      "loss": 0.8197,
      "step": 3757
    },
    {
      "epoch": 0.2757229197428396,
      "grad_norm": 1.487926368942577,
      "learning_rate": 8.498373297743578e-06,
      "loss": 0.8654,
      "step": 3758
    },
    {
      "epoch": 0.27579628933297873,
      "grad_norm": 1.5695268391907387,
      "learning_rate": 8.497524277904764e-06,
      "loss": 0.9431,
      "step": 3759
    },
    {
      "epoch": 0.2758696589231178,
      "grad_norm": 1.5906015569813095,
      "learning_rate": 8.496675060552093e-06,
      "loss": 0.8015,
      "step": 3760
    },
    {
      "epoch": 0.2758696589231178,
      "lm_loss": 0.3385,
      "step": 3760,
      "vm_loss": 0.0457
    },
    {
      "epoch": 0.2758696589231178,
      "lm_loss": 0.0106,
      "step": 3760,
      "vm_loss": 0.0773
    },
    {
      "epoch": 0.2758696589231178,
      "lm_loss": 0.9145,
      "step": 3760,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.2758696589231178,
      "lm_loss": 1.357,
      "step": 3760,
      "vm_loss": 0.1797
    },
    {
      "epoch": 0.2758696589231178,
      "lm_loss": 0.2232,
      "step": 3760,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.2758696589231178,
      "lm_loss": 0.6399,
      "step": 3760,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.2758696589231178,
      "lm_loss": 1.0964,
      "step": 3760,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.2758696589231178,
      "lm_loss": 0.1288,
      "step": 3760,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.27594302851325697,
      "grad_norm": 1.4199706171345159,
      "learning_rate": 8.495825645733522e-06,
      "loss": 0.7552,
      "step": 3761
    },
    {
      "epoch": 0.2760163981033961,
      "grad_norm": 1.6213595509939556,
      "learning_rate": 8.494976033497018e-06,
      "loss": 0.7496,
      "step": 3762
    },
    {
      "epoch": 0.2760897676935352,
      "grad_norm": 1.5266189935602128,
      "learning_rate": 8.494126223890564e-06,
      "loss": 0.8572,
      "step": 3763
    },
    {
      "epoch": 0.27616313728367436,
      "grad_norm": 1.4954602787103082,
      "learning_rate": 8.493276216962148e-06,
      "loss": 0.7632,
      "step": 3764
    },
    {
      "epoch": 0.27623650687381346,
      "grad_norm": 1.4467595016529555,
      "learning_rate": 8.492426012759773e-06,
      "loss": 0.822,
      "step": 3765
    },
    {
      "epoch": 0.2763098764639526,
      "grad_norm": 0.6083727980050602,
      "learning_rate": 8.491575611331452e-06,
      "loss": 0.5491,
      "step": 3766
    },
    {
      "epoch": 0.27638324605409176,
      "grad_norm": 1.359956592494324,
      "learning_rate": 8.49072501272521e-06,
      "loss": 0.7335,
      "step": 3767
    },
    {
      "epoch": 0.27645661564423085,
      "grad_norm": 1.4970201203629818,
      "learning_rate": 8.489874216989082e-06,
      "loss": 0.8403,
      "step": 3768
    },
    {
      "epoch": 0.27645661564423085,
      "lm_loss": 0.1575,
      "step": 3768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27645661564423085,
      "lm_loss": 0.516,
      "step": 3768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27645661564423085,
      "lm_loss": 0.5366,
      "step": 3768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27645661564423085,
      "lm_loss": 0.5458,
      "step": 3768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27645661564423085,
      "lm_loss": 0.449,
      "step": 3768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27645661564423085,
      "lm_loss": 0.5043,
      "step": 3768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27645661564423085,
      "lm_loss": 0.3942,
      "step": 3768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27645661564423085,
      "lm_loss": 0.5831,
      "step": 3768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27652998523437,
      "grad_norm": 0.5664830659053557,
      "learning_rate": 8.489023224171114e-06,
      "loss": 0.5085,
      "step": 3769
    },
    {
      "epoch": 0.2766033548245091,
      "grad_norm": 0.5848188975965936,
      "learning_rate": 8.488172034319366e-06,
      "loss": 0.5311,
      "step": 3770
    },
    {
      "epoch": 0.27667672441464825,
      "grad_norm": 0.5719047412932943,
      "learning_rate": 8.487320647481903e-06,
      "loss": 0.5227,
      "step": 3771
    },
    {
      "epoch": 0.27675009400478734,
      "grad_norm": 0.5593590696325603,
      "learning_rate": 8.486469063706806e-06,
      "loss": 0.5226,
      "step": 3772
    },
    {
      "epoch": 0.2768234635949265,
      "grad_norm": 1.5917430058178108,
      "learning_rate": 8.485617283042169e-06,
      "loss": 0.7763,
      "step": 3773
    },
    {
      "epoch": 0.27689683318506564,
      "grad_norm": 0.5757651942830301,
      "learning_rate": 8.484765305536092e-06,
      "loss": 0.5421,
      "step": 3774
    },
    {
      "epoch": 0.27697020277520473,
      "grad_norm": 1.449111766988218,
      "learning_rate": 8.483913131236685e-06,
      "loss": 0.8526,
      "step": 3775
    },
    {
      "epoch": 0.2770435723653439,
      "grad_norm": 1.4576830963781333,
      "learning_rate": 8.48306076019208e-06,
      "loss": 0.755,
      "step": 3776
    },
    {
      "epoch": 0.2770435723653439,
      "lm_loss": 0.139,
      "step": 3776,
      "vm_loss": 0.051
    },
    {
      "epoch": 0.2770435723653439,
      "lm_loss": 1.8809,
      "step": 3776,
      "vm_loss": 0.1085
    },
    {
      "epoch": 0.2770435723653439,
      "lm_loss": 0.5836,
      "step": 3776,
      "vm_loss": 0.1762
    },
    {
      "epoch": 0.2770435723653439,
      "lm_loss": 1.0681,
      "step": 3776,
      "vm_loss": 0.2005
    },
    {
      "epoch": 0.2770435723653439,
      "lm_loss": 0.8487,
      "step": 3776,
      "vm_loss": 0.2552
    },
    {
      "epoch": 0.2770435723653439,
      "lm_loss": 0.2599,
      "step": 3776,
      "vm_loss": 0.0673
    },
    {
      "epoch": 0.2770435723653439,
      "lm_loss": 0.0193,
      "step": 3776,
      "vm_loss": 0.1316
    },
    {
      "epoch": 0.2770435723653439,
      "lm_loss": 0.5886,
      "step": 3776,
      "vm_loss": 0.1391
    },
    {
      "epoch": 0.277116941955483,
      "grad_norm": 1.5110861898391943,
      "learning_rate": 8.482208192450405e-06,
      "loss": 0.7908,
      "step": 3777
    },
    {
      "epoch": 0.2771903115456221,
      "grad_norm": 0.6307200550872094,
      "learning_rate": 8.481355428059813e-06,
      "loss": 0.5105,
      "step": 3778
    },
    {
      "epoch": 0.2772636811357613,
      "grad_norm": 1.5922022769893642,
      "learning_rate": 8.480502467068455e-06,
      "loss": 0.8382,
      "step": 3779
    },
    {
      "epoch": 0.27733705072590037,
      "grad_norm": 1.8015578165683357,
      "learning_rate": 8.479649309524504e-06,
      "loss": 0.8633,
      "step": 3780
    },
    {
      "epoch": 0.2774104203160395,
      "grad_norm": 1.4055288485369202,
      "learning_rate": 8.478795955476142e-06,
      "loss": 0.866,
      "step": 3781
    },
    {
      "epoch": 0.2774837899061786,
      "grad_norm": 1.630245991566642,
      "learning_rate": 8.477942404971555e-06,
      "loss": 0.9005,
      "step": 3782
    },
    {
      "epoch": 0.27755715949631776,
      "grad_norm": 1.7437476267046264,
      "learning_rate": 8.477088658058949e-06,
      "loss": 0.8491,
      "step": 3783
    },
    {
      "epoch": 0.2776305290864569,
      "grad_norm": 1.7481852888205032,
      "learning_rate": 8.476234714786535e-06,
      "loss": 0.7582,
      "step": 3784
    },
    {
      "epoch": 0.2776305290864569,
      "lm_loss": 1.5534,
      "step": 3784,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.2776305290864569,
      "lm_loss": 1.7415,
      "step": 3784,
      "vm_loss": 0.1741
    },
    {
      "epoch": 0.2776305290864569,
      "lm_loss": 0.7718,
      "step": 3784,
      "vm_loss": 0.1302
    },
    {
      "epoch": 0.2776305290864569,
      "lm_loss": 0.8264,
      "step": 3784,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.2776305290864569,
      "lm_loss": 0.3383,
      "step": 3784,
      "vm_loss": 0.0447
    },
    {
      "epoch": 0.2776305290864569,
      "lm_loss": 1.199,
      "step": 3784,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.2776305290864569,
      "lm_loss": 0.686,
      "step": 3784,
      "vm_loss": 0.1462
    },
    {
      "epoch": 0.2776305290864569,
      "lm_loss": 0.2522,
      "step": 3784,
      "vm_loss": 0.0561
    },
    {
      "epoch": 0.277703898676596,
      "grad_norm": 1.508094450660093,
      "learning_rate": 8.475380575202536e-06,
      "loss": 0.8644,
      "step": 3785
    },
    {
      "epoch": 0.27777726826673516,
      "grad_norm": 1.5380645608082066,
      "learning_rate": 8.474526239355191e-06,
      "loss": 0.8429,
      "step": 3786
    },
    {
      "epoch": 0.27785063785687425,
      "grad_norm": 1.55263096148642,
      "learning_rate": 8.473671707292745e-06,
      "loss": 0.8381,
      "step": 3787
    },
    {
      "epoch": 0.2779240074470134,
      "grad_norm": 0.6690434022697161,
      "learning_rate": 8.472816979063456e-06,
      "loss": 0.5295,
      "step": 3788
    },
    {
      "epoch": 0.27799737703715255,
      "grad_norm": 0.6169866166204268,
      "learning_rate": 8.47196205471559e-06,
      "loss": 0.5475,
      "step": 3789
    },
    {
      "epoch": 0.27807074662729164,
      "grad_norm": 1.697309496244864,
      "learning_rate": 8.471106934297428e-06,
      "loss": 0.8974,
      "step": 3790
    },
    {
      "epoch": 0.2781441162174308,
      "grad_norm": 1.7435398576059224,
      "learning_rate": 8.470251617857264e-06,
      "loss": 0.8529,
      "step": 3791
    },
    {
      "epoch": 0.2782174858075699,
      "grad_norm": 1.5244858349045916,
      "learning_rate": 8.469396105443396e-06,
      "loss": 0.863,
      "step": 3792
    },
    {
      "epoch": 0.2782174858075699,
      "lm_loss": 1.0366,
      "step": 3792,
      "vm_loss": 0.0523
    },
    {
      "epoch": 0.2782174858075699,
      "lm_loss": 1.3183,
      "step": 3792,
      "vm_loss": 0.0686
    },
    {
      "epoch": 0.2782174858075699,
      "lm_loss": 0.8961,
      "step": 3792,
      "vm_loss": 0.0995
    },
    {
      "epoch": 0.2782174858075699,
      "lm_loss": 1.6606,
      "step": 3792,
      "vm_loss": 0.0862
    },
    {
      "epoch": 0.2782174858075699,
      "lm_loss": 1.091,
      "step": 3792,
      "vm_loss": 0.1341
    },
    {
      "epoch": 0.2782174858075699,
      "lm_loss": 1.022,
      "step": 3792,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.2782174858075699,
      "lm_loss": 0.7547,
      "step": 3792,
      "vm_loss": 0.1681
    },
    {
      "epoch": 0.2782174858075699,
      "lm_loss": 0.4911,
      "step": 3792,
      "vm_loss": 0.1656
    },
    {
      "epoch": 0.27829085539770904,
      "grad_norm": 1.6239320570574463,
      "learning_rate": 8.468540397104138e-06,
      "loss": 0.9162,
      "step": 3793
    },
    {
      "epoch": 0.2783642249878482,
      "grad_norm": 1.4646717151174067,
      "learning_rate": 8.467684492887815e-06,
      "loss": 0.8512,
      "step": 3794
    },
    {
      "epoch": 0.2784375945779873,
      "grad_norm": 0.8235589964611882,
      "learning_rate": 8.46682839284276e-06,
      "loss": 0.5253,
      "step": 3795
    },
    {
      "epoch": 0.27851096416812643,
      "grad_norm": 1.7023138102138367,
      "learning_rate": 8.46597209701732e-06,
      "loss": 0.7494,
      "step": 3796
    },
    {
      "epoch": 0.2785843337582655,
      "grad_norm": 1.4796541544931772,
      "learning_rate": 8.465115605459853e-06,
      "loss": 0.8077,
      "step": 3797
    },
    {
      "epoch": 0.2786577033484047,
      "grad_norm": 1.4547174363265671,
      "learning_rate": 8.464258918218726e-06,
      "loss": 0.7939,
      "step": 3798
    },
    {
      "epoch": 0.2787310729385438,
      "grad_norm": 1.535290273133161,
      "learning_rate": 8.463402035342319e-06,
      "loss": 0.8229,
      "step": 3799
    },
    {
      "epoch": 0.2788044425286829,
      "grad_norm": 1.6162760193735435,
      "learning_rate": 8.462544956879024e-06,
      "loss": 0.8361,
      "step": 3800
    },
    {
      "epoch": 0.2788044425286829,
      "lm_loss": 1.5869,
      "step": 3800,
      "vm_loss": 0.0703
    },
    {
      "epoch": 0.2788044425286829,
      "lm_loss": 0.6554,
      "step": 3800,
      "vm_loss": 0.1423
    },
    {
      "epoch": 0.2788044425286829,
      "lm_loss": 0.4329,
      "step": 3800,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.2788044425286829,
      "lm_loss": 0.1149,
      "step": 3800,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.2788044425286829,
      "lm_loss": 0.9555,
      "step": 3800,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.2788044425286829,
      "lm_loss": 0.408,
      "step": 3800,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.2788044425286829,
      "lm_loss": 0.0701,
      "step": 3800,
      "vm_loss": 0.0572
    },
    {
      "epoch": 0.2788044425286829,
      "lm_loss": 0.499,
      "step": 3800,
      "vm_loss": 0.1341
    },
    {
      "epoch": 0.27887781211882207,
      "grad_norm": 1.6104245034614209,
      "learning_rate": 8.461687682877237e-06,
      "loss": 0.7732,
      "step": 3801
    },
    {
      "epoch": 0.27895118170896116,
      "grad_norm": 1.5897169431277822,
      "learning_rate": 8.460830213385376e-06,
      "loss": 0.7574,
      "step": 3802
    },
    {
      "epoch": 0.2790245512991003,
      "grad_norm": 0.6309337605891181,
      "learning_rate": 8.459972548451862e-06,
      "loss": 0.5453,
      "step": 3803
    },
    {
      "epoch": 0.27909792088923946,
      "grad_norm": 0.6071702770179717,
      "learning_rate": 8.45911468812513e-06,
      "loss": 0.5325,
      "step": 3804
    },
    {
      "epoch": 0.27917129047937855,
      "grad_norm": 0.5684438279924311,
      "learning_rate": 8.458256632453624e-06,
      "loss": 0.5194,
      "step": 3805
    },
    {
      "epoch": 0.2792446600695177,
      "grad_norm": 1.6683940362415355,
      "learning_rate": 8.457398381485802e-06,
      "loss": 0.7992,
      "step": 3806
    },
    {
      "epoch": 0.2793180296596568,
      "grad_norm": 1.7500855990914699,
      "learning_rate": 8.456539935270132e-06,
      "loss": 0.8848,
      "step": 3807
    },
    {
      "epoch": 0.27939139924979595,
      "grad_norm": 1.6032078336508873,
      "learning_rate": 8.455681293855092e-06,
      "loss": 0.7825,
      "step": 3808
    },
    {
      "epoch": 0.27939139924979595,
      "lm_loss": 0.1832,
      "step": 3808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27939139924979595,
      "lm_loss": 0.4568,
      "step": 3808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27939139924979595,
      "lm_loss": 0.7092,
      "step": 3808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27939139924979595,
      "lm_loss": 0.5615,
      "step": 3808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27939139924979595,
      "lm_loss": 0.6229,
      "step": 3808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27939139924979595,
      "lm_loss": 0.3009,
      "step": 3808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27939139924979595,
      "lm_loss": 0.5947,
      "step": 3808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27939139924979595,
      "lm_loss": 0.5791,
      "step": 3808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.27946476883993504,
      "grad_norm": 0.6207190531637576,
      "learning_rate": 8.45482245728917e-06,
      "loss": 0.5121,
      "step": 3809
    },
    {
      "epoch": 0.2795381384300742,
      "grad_norm": 1.6326649125411112,
      "learning_rate": 8.45396342562087e-06,
      "loss": 0.7231,
      "step": 3810
    },
    {
      "epoch": 0.27961150802021334,
      "grad_norm": 0.6127555087062085,
      "learning_rate": 8.453104198898701e-06,
      "loss": 0.5282,
      "step": 3811
    },
    {
      "epoch": 0.27968487761035243,
      "grad_norm": 1.576864679366262,
      "learning_rate": 8.452244777171186e-06,
      "loss": 0.8473,
      "step": 3812
    },
    {
      "epoch": 0.2797582472004916,
      "grad_norm": 0.594722121435404,
      "learning_rate": 8.45138516048686e-06,
      "loss": 0.5262,
      "step": 3813
    },
    {
      "epoch": 0.2798316167906307,
      "grad_norm": 1.7512886574815005,
      "learning_rate": 8.450525348894265e-06,
      "loss": 0.8059,
      "step": 3814
    },
    {
      "epoch": 0.2799049863807698,
      "grad_norm": 1.5937905697226842,
      "learning_rate": 8.449665342441962e-06,
      "loss": 0.9272,
      "step": 3815
    },
    {
      "epoch": 0.279978355970909,
      "grad_norm": 0.6225708465461703,
      "learning_rate": 8.448805141178512e-06,
      "loss": 0.551,
      "step": 3816
    },
    {
      "epoch": 0.279978355970909,
      "lm_loss": 0.3346,
      "step": 3816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.279978355970909,
      "lm_loss": 0.6553,
      "step": 3816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.279978355970909,
      "lm_loss": 0.6282,
      "step": 3816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.279978355970909,
      "lm_loss": 0.4778,
      "step": 3816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.279978355970909,
      "lm_loss": 0.621,
      "step": 3816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.279978355970909,
      "lm_loss": 0.5446,
      "step": 3816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.279978355970909,
      "lm_loss": 0.542,
      "step": 3816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.279978355970909,
      "lm_loss": 0.4014,
      "step": 3816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28005172556104807,
      "grad_norm": 0.6244104528673922,
      "learning_rate": 8.447944745152493e-06,
      "loss": 0.5226,
      "step": 3817
    },
    {
      "epoch": 0.2801250951511872,
      "grad_norm": 1.6731463771784518,
      "learning_rate": 8.447084154412499e-06,
      "loss": 0.7216,
      "step": 3818
    },
    {
      "epoch": 0.2801984647413263,
      "grad_norm": 0.569625887840189,
      "learning_rate": 8.446223369007122e-06,
      "loss": 0.5232,
      "step": 3819
    },
    {
      "epoch": 0.28027183433146546,
      "grad_norm": 0.5971060217451604,
      "learning_rate": 8.445362388984982e-06,
      "loss": 0.5312,
      "step": 3820
    },
    {
      "epoch": 0.2803452039216046,
      "grad_norm": 1.5378124468015342,
      "learning_rate": 8.44450121439469e-06,
      "loss": 0.7901,
      "step": 3821
    },
    {
      "epoch": 0.2804185735117437,
      "grad_norm": 1.4741519626869866,
      "learning_rate": 8.44363984528489e-06,
      "loss": 0.7871,
      "step": 3822
    },
    {
      "epoch": 0.28049194310188286,
      "grad_norm": 0.6161535415167354,
      "learning_rate": 8.442778281704217e-06,
      "loss": 0.5265,
      "step": 3823
    },
    {
      "epoch": 0.28056531269202195,
      "grad_norm": 0.6336977316077719,
      "learning_rate": 8.441916523701328e-06,
      "loss": 0.5367,
      "step": 3824
    },
    {
      "epoch": 0.28056531269202195,
      "lm_loss": 1.0028,
      "step": 3824,
      "vm_loss": 0.0585
    },
    {
      "epoch": 0.28056531269202195,
      "lm_loss": 0.2175,
      "step": 3824,
      "vm_loss": 0.0538
    },
    {
      "epoch": 0.28056531269202195,
      "lm_loss": 0.6501,
      "step": 3824,
      "vm_loss": 0.1764
    },
    {
      "epoch": 0.28056531269202195,
      "lm_loss": 0.2644,
      "step": 3824,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.28056531269202195,
      "lm_loss": 0.2076,
      "step": 3824,
      "vm_loss": 0.122
    },
    {
      "epoch": 0.28056531269202195,
      "lm_loss": 0.0297,
      "step": 3824,
      "vm_loss": 0.1021
    },
    {
      "epoch": 0.28056531269202195,
      "lm_loss": 0.442,
      "step": 3824,
      "vm_loss": 0.1478
    },
    {
      "epoch": 0.28056531269202195,
      "lm_loss": 0.1992,
      "step": 3824,
      "vm_loss": 0.131
    },
    {
      "epoch": 0.2806386822821611,
      "grad_norm": 1.6506382653197371,
      "learning_rate": 8.441054571324889e-06,
      "loss": 0.814,
      "step": 3825
    },
    {
      "epoch": 0.28071205187230025,
      "grad_norm": 0.6176308163244874,
      "learning_rate": 8.440192424623579e-06,
      "loss": 0.5285,
      "step": 3826
    },
    {
      "epoch": 0.28078542146243934,
      "grad_norm": 0.5560670633938267,
      "learning_rate": 8.439330083646082e-06,
      "loss": 0.5294,
      "step": 3827
    },
    {
      "epoch": 0.2808587910525785,
      "grad_norm": 1.7177192031765445,
      "learning_rate": 8.438467548441098e-06,
      "loss": 0.8909,
      "step": 3828
    },
    {
      "epoch": 0.2809321606427176,
      "grad_norm": 1.5786605593008047,
      "learning_rate": 8.437604819057336e-06,
      "loss": 0.9692,
      "step": 3829
    },
    {
      "epoch": 0.28100553023285674,
      "grad_norm": 1.6124322419804624,
      "learning_rate": 8.436741895543516e-06,
      "loss": 0.8007,
      "step": 3830
    },
    {
      "epoch": 0.2810788998229959,
      "grad_norm": 0.6946296834230996,
      "learning_rate": 8.435878777948372e-06,
      "loss": 0.5254,
      "step": 3831
    },
    {
      "epoch": 0.281152269413135,
      "grad_norm": 0.5949642135387208,
      "learning_rate": 8.435015466320644e-06,
      "loss": 0.5087,
      "step": 3832
    },
    {
      "epoch": 0.281152269413135,
      "lm_loss": 0.6336,
      "step": 3832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.281152269413135,
      "lm_loss": 0.4678,
      "step": 3832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.281152269413135,
      "lm_loss": 0.6518,
      "step": 3832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.281152269413135,
      "lm_loss": 0.5447,
      "step": 3832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.281152269413135,
      "lm_loss": 0.5281,
      "step": 3832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.281152269413135,
      "lm_loss": 0.5837,
      "step": 3832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.281152269413135,
      "lm_loss": 0.4512,
      "step": 3832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.281152269413135,
      "lm_loss": 0.6012,
      "step": 3832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28122563900327413,
      "grad_norm": 0.57366594966541,
      "learning_rate": 8.434151960709084e-06,
      "loss": 0.5421,
      "step": 3833
    },
    {
      "epoch": 0.2812990085934132,
      "grad_norm": 0.5694356945434349,
      "learning_rate": 8.43328826116246e-06,
      "loss": 0.5274,
      "step": 3834
    },
    {
      "epoch": 0.2813723781835524,
      "grad_norm": 0.570296974933384,
      "learning_rate": 8.432424367729545e-06,
      "loss": 0.5336,
      "step": 3835
    },
    {
      "epoch": 0.2814457477736915,
      "grad_norm": 1.575043316341027,
      "learning_rate": 8.431560280459127e-06,
      "loss": 0.9365,
      "step": 3836
    },
    {
      "epoch": 0.2815191173638306,
      "grad_norm": 1.7313753662886933,
      "learning_rate": 8.4306959994e-06,
      "loss": 0.7678,
      "step": 3837
    },
    {
      "epoch": 0.28159248695396977,
      "grad_norm": 1.587016235693594,
      "learning_rate": 8.429831524600976e-06,
      "loss": 0.8773,
      "step": 3838
    },
    {
      "epoch": 0.28166585654410886,
      "grad_norm": 1.6015921115525975,
      "learning_rate": 8.42896685611087e-06,
      "loss": 0.836,
      "step": 3839
    },
    {
      "epoch": 0.281739226134248,
      "grad_norm": 1.563298329439023,
      "learning_rate": 8.428101993978514e-06,
      "loss": 0.7528,
      "step": 3840
    },
    {
      "epoch": 0.281739226134248,
      "lm_loss": 1.071,
      "step": 3840,
      "vm_loss": 0.0724
    },
    {
      "epoch": 0.281739226134248,
      "lm_loss": 0.682,
      "step": 3840,
      "vm_loss": 0.1202
    },
    {
      "epoch": 0.281739226134248,
      "lm_loss": 0.1792,
      "step": 3840,
      "vm_loss": 0.0503
    },
    {
      "epoch": 0.281739226134248,
      "lm_loss": 0.8642,
      "step": 3840,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.281739226134248,
      "lm_loss": 0.384,
      "step": 3840,
      "vm_loss": 0.0789
    },
    {
      "epoch": 0.281739226134248,
      "lm_loss": 0.1208,
      "step": 3840,
      "vm_loss": 0.1668
    },
    {
      "epoch": 0.281739226134248,
      "lm_loss": 0.4222,
      "step": 3840,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.281739226134248,
      "lm_loss": 0.4956,
      "step": 3840,
      "vm_loss": 0.2203
    },
    {
      "epoch": 0.28181259572438716,
      "grad_norm": 1.49876316243246,
      "learning_rate": 8.42723693825275e-06,
      "loss": 0.8286,
      "step": 3841
    },
    {
      "epoch": 0.28188596531452625,
      "grad_norm": 1.4010466407732265,
      "learning_rate": 8.426371688982429e-06,
      "loss": 0.8811,
      "step": 3842
    },
    {
      "epoch": 0.2819593349046654,
      "grad_norm": 0.776896127443158,
      "learning_rate": 8.425506246216414e-06,
      "loss": 0.5113,
      "step": 3843
    },
    {
      "epoch": 0.2820327044948045,
      "grad_norm": 1.5000317591968924,
      "learning_rate": 8.424640610003576e-06,
      "loss": 0.9451,
      "step": 3844
    },
    {
      "epoch": 0.28210607408494365,
      "grad_norm": 1.5697605321556511,
      "learning_rate": 8.423774780392802e-06,
      "loss": 0.7663,
      "step": 3845
    },
    {
      "epoch": 0.2821794436750828,
      "grad_norm": 1.6522192633208002,
      "learning_rate": 8.422908757432988e-06,
      "loss": 0.8454,
      "step": 3846
    },
    {
      "epoch": 0.2822528132652219,
      "grad_norm": 1.4711463399656701,
      "learning_rate": 8.422042541173041e-06,
      "loss": 0.8028,
      "step": 3847
    },
    {
      "epoch": 0.28232618285536104,
      "grad_norm": 0.6039114916349064,
      "learning_rate": 8.421176131661874e-06,
      "loss": 0.5318,
      "step": 3848
    },
    {
      "epoch": 0.28232618285536104,
      "lm_loss": 1.0312,
      "step": 3848,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.28232618285536104,
      "lm_loss": 0.9604,
      "step": 3848,
      "vm_loss": 0.1446
    },
    {
      "epoch": 0.28232618285536104,
      "lm_loss": 1.5194,
      "step": 3848,
      "vm_loss": 0.1708
    },
    {
      "epoch": 0.28232618285536104,
      "lm_loss": 0.4255,
      "step": 3848,
      "vm_loss": 0.1527
    },
    {
      "epoch": 0.28232618285536104,
      "lm_loss": 1.2933,
      "step": 3848,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.28232618285536104,
      "lm_loss": 0.904,
      "step": 3848,
      "vm_loss": 0.2151
    },
    {
      "epoch": 0.28232618285536104,
      "lm_loss": 0.4391,
      "step": 3848,
      "vm_loss": 0.1661
    },
    {
      "epoch": 0.28232618285536104,
      "lm_loss": 0.0783,
      "step": 3848,
      "vm_loss": 0.1535
    },
    {
      "epoch": 0.28239955244550013,
      "grad_norm": 1.6650823732456799,
      "learning_rate": 8.420309528948422e-06,
      "loss": 0.9155,
      "step": 3849
    },
    {
      "epoch": 0.2824729220356393,
      "grad_norm": 1.664741452437267,
      "learning_rate": 8.419442733081618e-06,
      "loss": 0.8274,
      "step": 3850
    },
    {
      "epoch": 0.2825462916257784,
      "grad_norm": 0.5926490841919985,
      "learning_rate": 8.418575744110417e-06,
      "loss": 0.5243,
      "step": 3851
    },
    {
      "epoch": 0.28261966121591753,
      "grad_norm": 1.6605388741197578,
      "learning_rate": 8.417708562083775e-06,
      "loss": 0.8471,
      "step": 3852
    },
    {
      "epoch": 0.2826930308060567,
      "grad_norm": 1.522366977003717,
      "learning_rate": 8.416841187050668e-06,
      "loss": 0.7846,
      "step": 3853
    },
    {
      "epoch": 0.28276640039619577,
      "grad_norm": 1.9978117788628693,
      "learning_rate": 8.415973619060079e-06,
      "loss": 0.8761,
      "step": 3854
    },
    {
      "epoch": 0.2828397699863349,
      "grad_norm": 1.723137647068127,
      "learning_rate": 8.415105858160999e-06,
      "loss": 0.8461,
      "step": 3855
    },
    {
      "epoch": 0.282913139576474,
      "grad_norm": 0.6070700566442204,
      "learning_rate": 8.414237904402434e-06,
      "loss": 0.5409,
      "step": 3856
    },
    {
      "epoch": 0.282913139576474,
      "lm_loss": 0.6436,
      "step": 3856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.282913139576474,
      "lm_loss": 0.5516,
      "step": 3856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.282913139576474,
      "lm_loss": 0.5986,
      "step": 3856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.282913139576474,
      "lm_loss": 0.4956,
      "step": 3856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.282913139576474,
      "lm_loss": 0.4141,
      "step": 3856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.282913139576474,
      "lm_loss": 0.454,
      "step": 3856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.282913139576474,
      "lm_loss": 0.7515,
      "step": 3856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.282913139576474,
      "lm_loss": 0.6097,
      "step": 3856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28298650916661316,
      "grad_norm": 0.5973656938466405,
      "learning_rate": 8.413369757833398e-06,
      "loss": 0.5343,
      "step": 3857
    },
    {
      "epoch": 0.2830598787567523,
      "grad_norm": 1.5283225495651163,
      "learning_rate": 8.41250141850292e-06,
      "loss": 0.7685,
      "step": 3858
    },
    {
      "epoch": 0.2831332483468914,
      "grad_norm": 1.467660243840062,
      "learning_rate": 8.411632886460036e-06,
      "loss": 0.81,
      "step": 3859
    },
    {
      "epoch": 0.28320661793703056,
      "grad_norm": 1.4359655578004495,
      "learning_rate": 8.410764161753793e-06,
      "loss": 0.8392,
      "step": 3860
    },
    {
      "epoch": 0.28327998752716965,
      "grad_norm": 0.5724154592422412,
      "learning_rate": 8.409895244433253e-06,
      "loss": 0.5336,
      "step": 3861
    },
    {
      "epoch": 0.2833533571173088,
      "grad_norm": 1.4961979354330355,
      "learning_rate": 8.409026134547482e-06,
      "loss": 0.9213,
      "step": 3862
    },
    {
      "epoch": 0.28342672670744795,
      "grad_norm": 0.6036682738731614,
      "learning_rate": 8.408156832145564e-06,
      "loss": 0.5364,
      "step": 3863
    },
    {
      "epoch": 0.28350009629758705,
      "grad_norm": 0.5651913279919389,
      "learning_rate": 8.407287337276591e-06,
      "loss": 0.5367,
      "step": 3864
    },
    {
      "epoch": 0.28350009629758705,
      "lm_loss": 1.562,
      "step": 3864,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.28350009629758705,
      "lm_loss": 0.3867,
      "step": 3864,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.28350009629758705,
      "lm_loss": 0.9124,
      "step": 3864,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.28350009629758705,
      "lm_loss": 0.9865,
      "step": 3864,
      "vm_loss": 0.1268
    },
    {
      "epoch": 0.28350009629758705,
      "lm_loss": 0.6364,
      "step": 3864,
      "vm_loss": 0.1659
    },
    {
      "epoch": 0.28350009629758705,
      "lm_loss": 0.659,
      "step": 3864,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.28350009629758705,
      "lm_loss": 0.3441,
      "step": 3864,
      "vm_loss": 0.182
    },
    {
      "epoch": 0.28350009629758705,
      "lm_loss": 0.373,
      "step": 3864,
      "vm_loss": 0.1662
    },
    {
      "epoch": 0.2835734658877262,
      "grad_norm": 1.6109373352588383,
      "learning_rate": 8.406417649989663e-06,
      "loss": 0.7746,
      "step": 3865
    },
    {
      "epoch": 0.2836468354778653,
      "grad_norm": 1.8087995307648548,
      "learning_rate": 8.405547770333893e-06,
      "loss": 0.9411,
      "step": 3866
    },
    {
      "epoch": 0.28372020506800444,
      "grad_norm": 1.459196544922513,
      "learning_rate": 8.404677698358407e-06,
      "loss": 0.7528,
      "step": 3867
    },
    {
      "epoch": 0.2837935746581436,
      "grad_norm": 1.5132935313530047,
      "learning_rate": 8.40380743411234e-06,
      "loss": 0.7808,
      "step": 3868
    },
    {
      "epoch": 0.2838669442482827,
      "grad_norm": 1.353326964989949,
      "learning_rate": 8.40293697764484e-06,
      "loss": 0.8136,
      "step": 3869
    },
    {
      "epoch": 0.28394031383842183,
      "grad_norm": 1.5771937082696263,
      "learning_rate": 8.402066329005062e-06,
      "loss": 0.8758,
      "step": 3870
    },
    {
      "epoch": 0.2840136834285609,
      "grad_norm": 1.5169734619916428,
      "learning_rate": 8.401195488242169e-06,
      "loss": 0.7673,
      "step": 3871
    },
    {
      "epoch": 0.2840870530187001,
      "grad_norm": 0.6832944792838475,
      "learning_rate": 8.400324455405347e-06,
      "loss": 0.5289,
      "step": 3872
    },
    {
      "epoch": 0.2840870530187001,
      "lm_loss": 0.792,
      "step": 3872,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.2840870530187001,
      "lm_loss": 1.0349,
      "step": 3872,
      "vm_loss": 0.146
    },
    {
      "epoch": 0.2840870530187001,
      "lm_loss": 0.4656,
      "step": 3872,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.2840870530187001,
      "lm_loss": 0.9627,
      "step": 3872,
      "vm_loss": 0.1972
    },
    {
      "epoch": 0.2840870530187001,
      "lm_loss": 1.1351,
      "step": 3872,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.2840870530187001,
      "lm_loss": 1.1044,
      "step": 3872,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.2840870530187001,
      "lm_loss": 0.968,
      "step": 3872,
      "vm_loss": 0.2134
    },
    {
      "epoch": 0.2840870530187001,
      "lm_loss": 0.5858,
      "step": 3872,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.2841604226088392,
      "grad_norm": 1.551684866614203,
      "learning_rate": 8.399453230543782e-06,
      "loss": 0.7672,
      "step": 3873
    },
    {
      "epoch": 0.2842337921989783,
      "grad_norm": 1.454298888956315,
      "learning_rate": 8.398581813706674e-06,
      "loss": 0.8662,
      "step": 3874
    },
    {
      "epoch": 0.28430716178911747,
      "grad_norm": 1.6619802861921613,
      "learning_rate": 8.397710204943233e-06,
      "loss": 0.8279,
      "step": 3875
    },
    {
      "epoch": 0.28438053137925656,
      "grad_norm": 1.4372870113302925,
      "learning_rate": 8.396838404302684e-06,
      "loss": 0.7554,
      "step": 3876
    },
    {
      "epoch": 0.2844539009693957,
      "grad_norm": 0.614305072024752,
      "learning_rate": 8.395966411834257e-06,
      "loss": 0.5552,
      "step": 3877
    },
    {
      "epoch": 0.28452727055953486,
      "grad_norm": 1.5001610814519848,
      "learning_rate": 8.3950942275872e-06,
      "loss": 0.6723,
      "step": 3878
    },
    {
      "epoch": 0.28460064014967396,
      "grad_norm": 1.5566792885235128,
      "learning_rate": 8.394221851610761e-06,
      "loss": 0.888,
      "step": 3879
    },
    {
      "epoch": 0.2846740097398131,
      "grad_norm": 1.5866120028460304,
      "learning_rate": 8.39334928395421e-06,
      "loss": 0.8344,
      "step": 3880
    },
    {
      "epoch": 0.2846740097398131,
      "lm_loss": 1.346,
      "step": 3880,
      "vm_loss": 0.1259
    },
    {
      "epoch": 0.2846740097398131,
      "lm_loss": 0.4312,
      "step": 3880,
      "vm_loss": 0.0516
    },
    {
      "epoch": 0.2846740097398131,
      "lm_loss": 0.8773,
      "step": 3880,
      "vm_loss": 0.2283
    },
    {
      "epoch": 0.2846740097398131,
      "lm_loss": 0.6356,
      "step": 3880,
      "vm_loss": 0.1732
    },
    {
      "epoch": 0.2846740097398131,
      "lm_loss": 0.6162,
      "step": 3880,
      "vm_loss": 0.0889
    },
    {
      "epoch": 0.2846740097398131,
      "lm_loss": 0.3163,
      "step": 3880,
      "vm_loss": 0.0837
    },
    {
      "epoch": 0.2846740097398131,
      "lm_loss": 1.0284,
      "step": 3880,
      "vm_loss": 0.1667
    },
    {
      "epoch": 0.2846740097398131,
      "lm_loss": 0.9433,
      "step": 3880,
      "vm_loss": 0.0726
    },
    {
      "epoch": 0.2847473793299522,
      "grad_norm": 1.635314889438721,
      "learning_rate": 8.392476524666821e-06,
      "loss": 0.8863,
      "step": 3881
    },
    {
      "epoch": 0.28482074892009135,
      "grad_norm": 1.4842611335273743,
      "learning_rate": 8.39160357379788e-06,
      "loss": 0.7423,
      "step": 3882
    },
    {
      "epoch": 0.2848941185102305,
      "grad_norm": 1.6006826183893332,
      "learning_rate": 8.390730431396688e-06,
      "loss": 0.8271,
      "step": 3883
    },
    {
      "epoch": 0.2849674881003696,
      "grad_norm": 1.4638146536533574,
      "learning_rate": 8.38985709751255e-06,
      "loss": 0.8011,
      "step": 3884
    },
    {
      "epoch": 0.28504085769050874,
      "grad_norm": 1.6139171751641326,
      "learning_rate": 8.388983572194789e-06,
      "loss": 0.9218,
      "step": 3885
    },
    {
      "epoch": 0.28511422728064784,
      "grad_norm": 0.708024690567214,
      "learning_rate": 8.388109855492731e-06,
      "loss": 0.539,
      "step": 3886
    },
    {
      "epoch": 0.285187596870787,
      "grad_norm": 0.6056194723544732,
      "learning_rate": 8.387235947455719e-06,
      "loss": 0.5279,
      "step": 3887
    },
    {
      "epoch": 0.28526096646092614,
      "grad_norm": 1.7738826149077986,
      "learning_rate": 8.386361848133106e-06,
      "loss": 0.904,
      "step": 3888
    },
    {
      "epoch": 0.28526096646092614,
      "lm_loss": 1.7274,
      "step": 3888,
      "vm_loss": 0.0584
    },
    {
      "epoch": 0.28526096646092614,
      "lm_loss": 0.3601,
      "step": 3888,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.28526096646092614,
      "lm_loss": 0.7037,
      "step": 3888,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.28526096646092614,
      "lm_loss": 0.5503,
      "step": 3888,
      "vm_loss": 0.176
    },
    {
      "epoch": 0.28526096646092614,
      "lm_loss": 1.909,
      "step": 3888,
      "vm_loss": 0.1389
    },
    {
      "epoch": 0.28526096646092614,
      "lm_loss": 0.0672,
      "step": 3888,
      "vm_loss": 0.0899
    },
    {
      "epoch": 0.28526096646092614,
      "lm_loss": 0.9684,
      "step": 3888,
      "vm_loss": 0.1055
    },
    {
      "epoch": 0.28526096646092614,
      "lm_loss": 0.8057,
      "step": 3888,
      "vm_loss": 0.2262
    },
    {
      "epoch": 0.28533433605106523,
      "grad_norm": 1.5116003804655467,
      "learning_rate": 8.385487557574253e-06,
      "loss": 0.8027,
      "step": 3889
    },
    {
      "epoch": 0.2854077056412044,
      "grad_norm": 1.5740774160189095,
      "learning_rate": 8.384613075828532e-06,
      "loss": 0.779,
      "step": 3890
    },
    {
      "epoch": 0.2854810752313435,
      "grad_norm": 1.4765935284268874,
      "learning_rate": 8.38373840294533e-06,
      "loss": 0.8313,
      "step": 3891
    },
    {
      "epoch": 0.2855544448214826,
      "grad_norm": 0.7568579444259992,
      "learning_rate": 8.38286353897404e-06,
      "loss": 0.5447,
      "step": 3892
    },
    {
      "epoch": 0.2856278144116217,
      "grad_norm": 1.620950265719642,
      "learning_rate": 8.38198848396407e-06,
      "loss": 0.7927,
      "step": 3893
    },
    {
      "epoch": 0.28570118400176087,
      "grad_norm": 1.3705208633900134,
      "learning_rate": 8.381113237964834e-06,
      "loss": 0.8128,
      "step": 3894
    },
    {
      "epoch": 0.2857745535919,
      "grad_norm": 1.6931432004605682,
      "learning_rate": 8.38023780102576e-06,
      "loss": 0.8419,
      "step": 3895
    },
    {
      "epoch": 0.2858479231820391,
      "grad_norm": 1.4649862158316225,
      "learning_rate": 8.379362173196285e-06,
      "loss": 0.7412,
      "step": 3896
    },
    {
      "epoch": 0.2858479231820391,
      "lm_loss": 1.4674,
      "step": 3896,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.2858479231820391,
      "lm_loss": 1.6447,
      "step": 3896,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.2858479231820391,
      "lm_loss": 0.3396,
      "step": 3896,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.2858479231820391,
      "lm_loss": 0.4741,
      "step": 3896,
      "vm_loss": 0.1602
    },
    {
      "epoch": 0.2858479231820391,
      "lm_loss": 0.3429,
      "step": 3896,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.2858479231820391,
      "lm_loss": 0.3292,
      "step": 3896,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.2858479231820391,
      "lm_loss": 0.0972,
      "step": 3896,
      "vm_loss": 0.1038
    },
    {
      "epoch": 0.2858479231820391,
      "lm_loss": 0.3421,
      "step": 3896,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.28592129277217826,
      "grad_norm": 1.4808579249320386,
      "learning_rate": 8.378486354525863e-06,
      "loss": 0.7851,
      "step": 3897
    },
    {
      "epoch": 0.28599466236231735,
      "grad_norm": 1.5739525337777343,
      "learning_rate": 8.377610345063947e-06,
      "loss": 0.7571,
      "step": 3898
    },
    {
      "epoch": 0.2860680319524565,
      "grad_norm": 0.6138679141204306,
      "learning_rate": 8.37673414486001e-06,
      "loss": 0.5127,
      "step": 3899
    },
    {
      "epoch": 0.28614140154259565,
      "grad_norm": 0.5899310038144395,
      "learning_rate": 8.375857753963535e-06,
      "loss": 0.5213,
      "step": 3900
    },
    {
      "epoch": 0.28621477113273475,
      "grad_norm": 0.6036600968887796,
      "learning_rate": 8.374981172424012e-06,
      "loss": 0.5314,
      "step": 3901
    },
    {
      "epoch": 0.2862881407228739,
      "grad_norm": 0.5757000519833853,
      "learning_rate": 8.374104400290945e-06,
      "loss": 0.531,
      "step": 3902
    },
    {
      "epoch": 0.286361510313013,
      "grad_norm": 1.8605137407150614,
      "learning_rate": 8.373227437613847e-06,
      "loss": 0.9315,
      "step": 3903
    },
    {
      "epoch": 0.28643487990315214,
      "grad_norm": 1.547807562650617,
      "learning_rate": 8.37235028444224e-06,
      "loss": 0.8202,
      "step": 3904
    },
    {
      "epoch": 0.28643487990315214,
      "lm_loss": 1.3119,
      "step": 3904,
      "vm_loss": 0.0422
    },
    {
      "epoch": 0.28643487990315214,
      "lm_loss": 1.3658,
      "step": 3904,
      "vm_loss": 0.1572
    },
    {
      "epoch": 0.28643487990315214,
      "lm_loss": 0.2392,
      "step": 3904,
      "vm_loss": 0.0504
    },
    {
      "epoch": 0.28643487990315214,
      "lm_loss": 0.9854,
      "step": 3904,
      "vm_loss": 0.1004
    },
    {
      "epoch": 0.28643487990315214,
      "lm_loss": 1.0237,
      "step": 3904,
      "vm_loss": 0.1791
    },
    {
      "epoch": 0.28643487990315214,
      "lm_loss": 0.1921,
      "step": 3904,
      "vm_loss": 0.1108
    },
    {
      "epoch": 0.28643487990315214,
      "lm_loss": 1.2406,
      "step": 3904,
      "vm_loss": 0.1466
    },
    {
      "epoch": 0.28643487990315214,
      "lm_loss": 1.8128,
      "step": 3904,
      "vm_loss": 0.1042
    },
    {
      "epoch": 0.2865082494932913,
      "grad_norm": 1.6724112358424466,
      "learning_rate": 8.37147294082566e-06,
      "loss": 0.8614,
      "step": 3905
    },
    {
      "epoch": 0.2865816190834304,
      "grad_norm": 1.7204177128030649,
      "learning_rate": 8.370595406813657e-06,
      "loss": 0.7682,
      "step": 3906
    },
    {
      "epoch": 0.28665498867356953,
      "grad_norm": 1.5094980244810556,
      "learning_rate": 8.369717682455783e-06,
      "loss": 0.8264,
      "step": 3907
    },
    {
      "epoch": 0.2867283582637086,
      "grad_norm": 1.6023395183277103,
      "learning_rate": 8.368839767801606e-06,
      "loss": 0.9432,
      "step": 3908
    },
    {
      "epoch": 0.2868017278538478,
      "grad_norm": 1.4368162546475827,
      "learning_rate": 8.367961662900704e-06,
      "loss": 0.8759,
      "step": 3909
    },
    {
      "epoch": 0.2868750974439869,
      "grad_norm": 1.4434694374329822,
      "learning_rate": 8.367083367802668e-06,
      "loss": 0.7962,
      "step": 3910
    },
    {
      "epoch": 0.286948467034126,
      "grad_norm": 1.5638862024033011,
      "learning_rate": 8.366204882557095e-06,
      "loss": 0.8001,
      "step": 3911
    },
    {
      "epoch": 0.28702183662426517,
      "grad_norm": 1.711657299747424,
      "learning_rate": 8.365326207213596e-06,
      "loss": 0.832,
      "step": 3912
    },
    {
      "epoch": 0.28702183662426517,
      "lm_loss": 0.5283,
      "step": 3912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28702183662426517,
      "lm_loss": 0.379,
      "step": 3912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28702183662426517,
      "lm_loss": 0.4862,
      "step": 3912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28702183662426517,
      "lm_loss": 0.4557,
      "step": 3912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28702183662426517,
      "lm_loss": 0.6477,
      "step": 3912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28702183662426517,
      "lm_loss": 0.4129,
      "step": 3912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28702183662426517,
      "lm_loss": 0.4735,
      "step": 3912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28702183662426517,
      "lm_loss": 0.3505,
      "step": 3912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.28709520621440426,
      "grad_norm": 0.7830214889899333,
      "learning_rate": 8.36444734182179e-06,
      "loss": 0.5298,
      "step": 3913
    },
    {
      "epoch": 0.2871685758045434,
      "grad_norm": 0.7328254139179543,
      "learning_rate": 8.363568286431314e-06,
      "loss": 0.5464,
      "step": 3914
    },
    {
      "epoch": 0.28724194539468256,
      "grad_norm": 1.6080072922996085,
      "learning_rate": 8.362689041091805e-06,
      "loss": 0.891,
      "step": 3915
    },
    {
      "epoch": 0.28731531498482166,
      "grad_norm": 1.7713062632060539,
      "learning_rate": 8.361809605852919e-06,
      "loss": 0.8352,
      "step": 3916
    },
    {
      "epoch": 0.2873886845749608,
      "grad_norm": 0.6361280982086086,
      "learning_rate": 8.360929980764319e-06,
      "loss": 0.5064,
      "step": 3917
    },
    {
      "epoch": 0.2874620541650999,
      "grad_norm": 0.6415511379283567,
      "learning_rate": 8.36005016587568e-06,
      "loss": 0.5207,
      "step": 3918
    },
    {
      "epoch": 0.28753542375523905,
      "grad_norm": 0.6317671768281261,
      "learning_rate": 8.359170161236686e-06,
      "loss": 0.5262,
      "step": 3919
    },
    {
      "epoch": 0.2876087933453782,
      "grad_norm": 1.7231498827675928,
      "learning_rate": 8.358289966897034e-06,
      "loss": 0.7825,
      "step": 3920
    },
    {
      "epoch": 0.2876087933453782,
      "lm_loss": 1.0954,
      "step": 3920,
      "vm_loss": 0.1528
    },
    {
      "epoch": 0.2876087933453782,
      "lm_loss": 0.4642,
      "step": 3920,
      "vm_loss": 0.0966
    },
    {
      "epoch": 0.2876087933453782,
      "lm_loss": 0.871,
      "step": 3920,
      "vm_loss": 0.2057
    },
    {
      "epoch": 0.2876087933453782,
      "lm_loss": 1.5956,
      "step": 3920,
      "vm_loss": 0.1676
    },
    {
      "epoch": 0.2876087933453782,
      "lm_loss": 0.8401,
      "step": 3920,
      "vm_loss": 0.1613
    },
    {
      "epoch": 0.2876087933453782,
      "lm_loss": 0.8695,
      "step": 3920,
      "vm_loss": 0.1613
    },
    {
      "epoch": 0.2876087933453782,
      "lm_loss": 0.3428,
      "step": 3920,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.2876087933453782,
      "lm_loss": 0.1192,
      "step": 3920,
      "vm_loss": 0.1337
    },
    {
      "epoch": 0.2876821629355173,
      "grad_norm": 2.149980643504565,
      "learning_rate": 8.357409582906433e-06,
      "loss": 0.8788,
      "step": 3921
    },
    {
      "epoch": 0.28775553252565644,
      "grad_norm": 1.449628048166824,
      "learning_rate": 8.356529009314597e-06,
      "loss": 0.8432,
      "step": 3922
    },
    {
      "epoch": 0.28782890211579554,
      "grad_norm": 1.9695005400342016,
      "learning_rate": 8.355648246171257e-06,
      "loss": 0.8535,
      "step": 3923
    },
    {
      "epoch": 0.2879022717059347,
      "grad_norm": 1.7375599491834657,
      "learning_rate": 8.354767293526149e-06,
      "loss": 0.7524,
      "step": 3924
    },
    {
      "epoch": 0.28797564129607384,
      "grad_norm": 1.6909562580573048,
      "learning_rate": 8.353886151429025e-06,
      "loss": 0.86,
      "step": 3925
    },
    {
      "epoch": 0.28804901088621293,
      "grad_norm": 1.502092048697493,
      "learning_rate": 8.353004819929643e-06,
      "loss": 0.8343,
      "step": 3926
    },
    {
      "epoch": 0.2881223804763521,
      "grad_norm": 1.630394633755537,
      "learning_rate": 8.352123299077776e-06,
      "loss": 0.8188,
      "step": 3927
    },
    {
      "epoch": 0.2881957500664912,
      "grad_norm": 1.5498958860258343,
      "learning_rate": 8.351241588923205e-06,
      "loss": 0.8841,
      "step": 3928
    },
    {
      "epoch": 0.2881957500664912,
      "lm_loss": 0.5153,
      "step": 3928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2881957500664912,
      "lm_loss": 0.509,
      "step": 3928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2881957500664912,
      "lm_loss": 0.6298,
      "step": 3928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2881957500664912,
      "lm_loss": 0.7069,
      "step": 3928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2881957500664912,
      "lm_loss": 0.4527,
      "step": 3928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2881957500664912,
      "lm_loss": 0.4131,
      "step": 3928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2881957500664912,
      "lm_loss": 0.3031,
      "step": 3928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2881957500664912,
      "lm_loss": 0.4877,
      "step": 3928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2882691196566303,
      "grad_norm": 1.0946348114903055,
      "learning_rate": 8.35035968951572e-06,
      "loss": 0.5464,
      "step": 3929
    },
    {
      "epoch": 0.2883424892467694,
      "grad_norm": 1.5765339110551477,
      "learning_rate": 8.34947760090513e-06,
      "loss": 0.92,
      "step": 3930
    },
    {
      "epoch": 0.28841585883690857,
      "grad_norm": 1.4837297691291584,
      "learning_rate": 8.348595323141245e-06,
      "loss": 0.7799,
      "step": 3931
    },
    {
      "epoch": 0.2884892284270477,
      "grad_norm": 1.2749097599627432,
      "learning_rate": 8.347712856273886e-06,
      "loss": 0.8068,
      "step": 3932
    },
    {
      "epoch": 0.2885625980171868,
      "grad_norm": 1.5429834916660417,
      "learning_rate": 8.346830200352894e-06,
      "loss": 0.8129,
      "step": 3933
    },
    {
      "epoch": 0.28863596760732596,
      "grad_norm": 1.5930979388144373,
      "learning_rate": 8.345947355428112e-06,
      "loss": 0.8638,
      "step": 3934
    },
    {
      "epoch": 0.28870933719746505,
      "grad_norm": 1.470658956658613,
      "learning_rate": 8.345064321549396e-06,
      "loss": 0.7985,
      "step": 3935
    },
    {
      "epoch": 0.2887827067876042,
      "grad_norm": 1.6136760082724912,
      "learning_rate": 8.344181098766613e-06,
      "loss": 0.7781,
      "step": 3936
    },
    {
      "epoch": 0.2887827067876042,
      "lm_loss": 1.5236,
      "step": 3936,
      "vm_loss": 0.0359
    },
    {
      "epoch": 0.2887827067876042,
      "lm_loss": 1.4414,
      "step": 3936,
      "vm_loss": 0.0595
    },
    {
      "epoch": 0.2887827067876042,
      "lm_loss": 1.2135,
      "step": 3936,
      "vm_loss": 0.1523
    },
    {
      "epoch": 0.2887827067876042,
      "lm_loss": 1.1019,
      "step": 3936,
      "vm_loss": 0.2386
    },
    {
      "epoch": 0.2887827067876042,
      "lm_loss": 1.0589,
      "step": 3936,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.2887827067876042,
      "lm_loss": 0.1017,
      "step": 3936,
      "vm_loss": 0.0894
    },
    {
      "epoch": 0.2887827067876042,
      "lm_loss": 0.4553,
      "step": 3936,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.2887827067876042,
      "lm_loss": 0.0341,
      "step": 3936,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.28885607637774335,
      "grad_norm": 1.544459472135866,
      "learning_rate": 8.343297687129643e-06,
      "loss": 0.8056,
      "step": 3937
    },
    {
      "epoch": 0.28892944596788245,
      "grad_norm": 1.551138094909275,
      "learning_rate": 8.342414086688374e-06,
      "loss": 0.7983,
      "step": 3938
    },
    {
      "epoch": 0.2890028155580216,
      "grad_norm": 1.4666676314825458,
      "learning_rate": 8.341530297492703e-06,
      "loss": 0.8159,
      "step": 3939
    },
    {
      "epoch": 0.2890761851481607,
      "grad_norm": 0.889492664449201,
      "learning_rate": 8.34064631959254e-06,
      "loss": 0.5503,
      "step": 3940
    },
    {
      "epoch": 0.28914955473829984,
      "grad_norm": 0.7574745051931345,
      "learning_rate": 8.339762153037809e-06,
      "loss": 0.5441,
      "step": 3941
    },
    {
      "epoch": 0.289222924328439,
      "grad_norm": 1.5433253072106454,
      "learning_rate": 8.338877797878436e-06,
      "loss": 0.8313,
      "step": 3942
    },
    {
      "epoch": 0.2892962939185781,
      "grad_norm": 1.534331932326462,
      "learning_rate": 8.337993254164367e-06,
      "loss": 0.9231,
      "step": 3943
    },
    {
      "epoch": 0.28936966350871723,
      "grad_norm": 1.6782672336344935,
      "learning_rate": 8.33710852194555e-06,
      "loss": 0.8246,
      "step": 3944
    },
    {
      "epoch": 0.28936966350871723,
      "lm_loss": 1.1115,
      "step": 3944,
      "vm_loss": 0.0373
    },
    {
      "epoch": 0.28936966350871723,
      "lm_loss": 0.6008,
      "step": 3944,
      "vm_loss": 0.1862
    },
    {
      "epoch": 0.28936966350871723,
      "lm_loss": 0.1691,
      "step": 3944,
      "vm_loss": 0.177
    },
    {
      "epoch": 0.28936966350871723,
      "lm_loss": 0.9277,
      "step": 3944,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.28936966350871723,
      "lm_loss": 0.394,
      "step": 3944,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.28936966350871723,
      "lm_loss": 0.6912,
      "step": 3944,
      "vm_loss": 0.1016
    },
    {
      "epoch": 0.28936966350871723,
      "lm_loss": 0.7182,
      "step": 3944,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.28936966350871723,
      "lm_loss": 0.6114,
      "step": 3944,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.28944303309885633,
      "grad_norm": 1.5891282155384083,
      "learning_rate": 8.336223601271953e-06,
      "loss": 0.865,
      "step": 3945
    },
    {
      "epoch": 0.2895164026889955,
      "grad_norm": 1.4795626040876735,
      "learning_rate": 8.335338492193546e-06,
      "loss": 0.7802,
      "step": 3946
    },
    {
      "epoch": 0.2895897722791346,
      "grad_norm": 1.4931569558270505,
      "learning_rate": 8.334453194760314e-06,
      "loss": 0.8365,
      "step": 3947
    },
    {
      "epoch": 0.2896631418692737,
      "grad_norm": 1.5865319953222068,
      "learning_rate": 8.333567709022255e-06,
      "loss": 0.8693,
      "step": 3948
    },
    {
      "epoch": 0.28973651145941287,
      "grad_norm": 1.474553457898886,
      "learning_rate": 8.33268203502937e-06,
      "loss": 0.8554,
      "step": 3949
    },
    {
      "epoch": 0.28980988104955197,
      "grad_norm": 1.509278302157166,
      "learning_rate": 8.331796172831677e-06,
      "loss": 0.8245,
      "step": 3950
    },
    {
      "epoch": 0.2898832506396911,
      "grad_norm": 1.6647792356565791,
      "learning_rate": 8.330910122479203e-06,
      "loss": 0.9408,
      "step": 3951
    },
    {
      "epoch": 0.28995662022983026,
      "grad_norm": 1.6139599898330406,
      "learning_rate": 8.330023884021988e-06,
      "loss": 0.8156,
      "step": 3952
    },
    {
      "epoch": 0.28995662022983026,
      "lm_loss": 1.0131,
      "step": 3952,
      "vm_loss": 0.0502
    },
    {
      "epoch": 0.28995662022983026,
      "lm_loss": 0.3174,
      "step": 3952,
      "vm_loss": 0.091
    },
    {
      "epoch": 0.28995662022983026,
      "lm_loss": 0.375,
      "step": 3952,
      "vm_loss": 0.0919
    },
    {
      "epoch": 0.28995662022983026,
      "lm_loss": 0.2394,
      "step": 3952,
      "vm_loss": 0.1543
    },
    {
      "epoch": 0.28995662022983026,
      "lm_loss": 1.0745,
      "step": 3952,
      "vm_loss": 0.2193
    },
    {
      "epoch": 0.28995662022983026,
      "lm_loss": 0.2446,
      "step": 3952,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.28995662022983026,
      "lm_loss": 0.4128,
      "step": 3952,
      "vm_loss": 0.122
    },
    {
      "epoch": 0.28995662022983026,
      "lm_loss": 1.1568,
      "step": 3952,
      "vm_loss": 0.1537
    },
    {
      "epoch": 0.29002998981996936,
      "grad_norm": 1.460997746584391,
      "learning_rate": 8.329137457510075e-06,
      "loss": 0.8146,
      "step": 3953
    },
    {
      "epoch": 0.2901033594101085,
      "grad_norm": 1.415560026309733,
      "learning_rate": 8.328250842993526e-06,
      "loss": 0.8596,
      "step": 3954
    },
    {
      "epoch": 0.2901767290002476,
      "grad_norm": 1.5944723882713265,
      "learning_rate": 8.32736404052241e-06,
      "loss": 0.7624,
      "step": 3955
    },
    {
      "epoch": 0.29025009859038675,
      "grad_norm": 1.4031860864553394,
      "learning_rate": 8.326477050146806e-06,
      "loss": 0.8295,
      "step": 3956
    },
    {
      "epoch": 0.2903234681805259,
      "grad_norm": 1.6191094768809533,
      "learning_rate": 8.325589871916805e-06,
      "loss": 0.9288,
      "step": 3957
    },
    {
      "epoch": 0.290396837770665,
      "grad_norm": 1.553461601946525,
      "learning_rate": 8.324702505882509e-06,
      "loss": 0.7648,
      "step": 3958
    },
    {
      "epoch": 0.29047020736080414,
      "grad_norm": 1.8626969688030584,
      "learning_rate": 8.323814952094028e-06,
      "loss": 0.5531,
      "step": 3959
    },
    {
      "epoch": 0.29054357695094324,
      "grad_norm": 1.7223997297660385,
      "learning_rate": 8.322927210601486e-06,
      "loss": 0.8763,
      "step": 3960
    },
    {
      "epoch": 0.29054357695094324,
      "lm_loss": 1.6012,
      "step": 3960,
      "vm_loss": 0.0497
    },
    {
      "epoch": 0.29054357695094324,
      "lm_loss": 1.7209,
      "step": 3960,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.29054357695094324,
      "lm_loss": 0.1312,
      "step": 3960,
      "vm_loss": 0.2247
    },
    {
      "epoch": 0.29054357695094324,
      "lm_loss": 1.3671,
      "step": 3960,
      "vm_loss": 0.2258
    },
    {
      "epoch": 0.29054357695094324,
      "lm_loss": 1.3165,
      "step": 3960,
      "vm_loss": 0.1343
    },
    {
      "epoch": 0.29054357695094324,
      "lm_loss": 1.0947,
      "step": 3960,
      "vm_loss": 0.1829
    },
    {
      "epoch": 0.29054357695094324,
      "lm_loss": 0.6847,
      "step": 3960,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.29054357695094324,
      "lm_loss": 0.2863,
      "step": 3960,
      "vm_loss": 0.1924
    },
    {
      "epoch": 0.2906169465410824,
      "grad_norm": 1.5727805935862122,
      "learning_rate": 8.322039281455016e-06,
      "loss": 0.8739,
      "step": 3961
    },
    {
      "epoch": 0.29069031613122154,
      "grad_norm": 1.3841992098622584,
      "learning_rate": 8.321151164704759e-06,
      "loss": 0.7912,
      "step": 3962
    },
    {
      "epoch": 0.29076368572136063,
      "grad_norm": 1.390453187960736,
      "learning_rate": 8.320262860400872e-06,
      "loss": 0.7479,
      "step": 3963
    },
    {
      "epoch": 0.2908370553114998,
      "grad_norm": 1.5365229826147666,
      "learning_rate": 8.31937436859352e-06,
      "loss": 0.7657,
      "step": 3964
    },
    {
      "epoch": 0.2909104249016389,
      "grad_norm": 1.8436122679550804,
      "learning_rate": 8.318485689332876e-06,
      "loss": 0.8695,
      "step": 3965
    },
    {
      "epoch": 0.290983794491778,
      "grad_norm": 1.5353453954202636,
      "learning_rate": 8.317596822669128e-06,
      "loss": 0.8113,
      "step": 3966
    },
    {
      "epoch": 0.2910571640819172,
      "grad_norm": 1.4385068630030633,
      "learning_rate": 8.31670776865247e-06,
      "loss": 0.8355,
      "step": 3967
    },
    {
      "epoch": 0.29113053367205627,
      "grad_norm": 1.3336269265865393,
      "learning_rate": 8.315818527333111e-06,
      "loss": 0.5359,
      "step": 3968
    },
    {
      "epoch": 0.29113053367205627,
      "lm_loss": 1.0607,
      "step": 3968,
      "vm_loss": 0.0335
    },
    {
      "epoch": 0.29113053367205627,
      "lm_loss": 1.2114,
      "step": 3968,
      "vm_loss": 0.2004
    },
    {
      "epoch": 0.29113053367205627,
      "lm_loss": 0.8277,
      "step": 3968,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.29113053367205627,
      "lm_loss": 0.5512,
      "step": 3968,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.29113053367205627,
      "lm_loss": 1.0917,
      "step": 3968,
      "vm_loss": 0.099
    },
    {
      "epoch": 0.29113053367205627,
      "lm_loss": 1.2524,
      "step": 3968,
      "vm_loss": 0.2485
    },
    {
      "epoch": 0.29113053367205627,
      "lm_loss": 0.2736,
      "step": 3968,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.29113053367205627,
      "lm_loss": 0.0252,
      "step": 3968,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.2912039032621954,
      "grad_norm": 1.6418798756087793,
      "learning_rate": 8.314929098761268e-06,
      "loss": 0.8489,
      "step": 3969
    },
    {
      "epoch": 0.2912772728523345,
      "grad_norm": 1.752435758077675,
      "learning_rate": 8.31403948298717e-06,
      "loss": 0.9214,
      "step": 3970
    },
    {
      "epoch": 0.29135064244247366,
      "grad_norm": 1.490081691341368,
      "learning_rate": 8.313149680061055e-06,
      "loss": 0.7934,
      "step": 3971
    },
    {
      "epoch": 0.29142401203261276,
      "grad_norm": 1.4063578376685724,
      "learning_rate": 8.312259690033172e-06,
      "loss": 0.9017,
      "step": 3972
    },
    {
      "epoch": 0.2914973816227519,
      "grad_norm": 0.6986542164296666,
      "learning_rate": 8.311369512953785e-06,
      "loss": 0.5344,
      "step": 3973
    },
    {
      "epoch": 0.29157075121289106,
      "grad_norm": 1.7810255268588784,
      "learning_rate": 8.310479148873158e-06,
      "loss": 0.9213,
      "step": 3974
    },
    {
      "epoch": 0.29164412080303015,
      "grad_norm": 0.6283062801264102,
      "learning_rate": 8.309588597841576e-06,
      "loss": 0.5488,
      "step": 3975
    },
    {
      "epoch": 0.2917174903931693,
      "grad_norm": 1.5916620353659208,
      "learning_rate": 8.308697859909331e-06,
      "loss": 0.8638,
      "step": 3976
    },
    {
      "epoch": 0.2917174903931693,
      "lm_loss": 1.4571,
      "step": 3976,
      "vm_loss": 0.0679
    },
    {
      "epoch": 0.2917174903931693,
      "lm_loss": 0.2485,
      "step": 3976,
      "vm_loss": 0.1169
    },
    {
      "epoch": 0.2917174903931693,
      "lm_loss": 0.9251,
      "step": 3976,
      "vm_loss": 0.1388
    },
    {
      "epoch": 0.2917174903931693,
      "lm_loss": 0.8017,
      "step": 3976,
      "vm_loss": 0.0703
    },
    {
      "epoch": 0.2917174903931693,
      "lm_loss": 0.9619,
      "step": 3976,
      "vm_loss": 0.1259
    },
    {
      "epoch": 0.2917174903931693,
      "lm_loss": 1.2569,
      "step": 3976,
      "vm_loss": 0.0834
    },
    {
      "epoch": 0.2917174903931693,
      "lm_loss": 0.8603,
      "step": 3976,
      "vm_loss": 0.1376
    },
    {
      "epoch": 0.2917174903931693,
      "lm_loss": 0.6711,
      "step": 3976,
      "vm_loss": 0.1019
    },
    {
      "epoch": 0.2917908599833084,
      "grad_norm": 1.6246813345045585,
      "learning_rate": 8.307806935126723e-06,
      "loss": 0.8951,
      "step": 3977
    },
    {
      "epoch": 0.29186422957344754,
      "grad_norm": 1.3818590191476738,
      "learning_rate": 8.306915823544066e-06,
      "loss": 0.6856,
      "step": 3978
    },
    {
      "epoch": 0.2919375991635867,
      "grad_norm": 1.42814266855178,
      "learning_rate": 8.306024525211682e-06,
      "loss": 0.9094,
      "step": 3979
    },
    {
      "epoch": 0.2920109687537258,
      "grad_norm": 1.5180899084761088,
      "learning_rate": 8.305133040179907e-06,
      "loss": 0.8212,
      "step": 3980
    },
    {
      "epoch": 0.29208433834386494,
      "grad_norm": 1.751232808000277,
      "learning_rate": 8.304241368499081e-06,
      "loss": 0.8878,
      "step": 3981
    },
    {
      "epoch": 0.29215770793400403,
      "grad_norm": 1.491878647351508,
      "learning_rate": 8.303349510219567e-06,
      "loss": 0.8331,
      "step": 3982
    },
    {
      "epoch": 0.2922310775241432,
      "grad_norm": 1.5530015522507996,
      "learning_rate": 8.302457465391722e-06,
      "loss": 0.8804,
      "step": 3983
    },
    {
      "epoch": 0.29230444711428233,
      "grad_norm": 1.709787671184625,
      "learning_rate": 8.301565234065926e-06,
      "loss": 0.9109,
      "step": 3984
    },
    {
      "epoch": 0.29230444711428233,
      "lm_loss": 1.2943,
      "step": 3984,
      "vm_loss": 0.0676
    },
    {
      "epoch": 0.29230444711428233,
      "lm_loss": 0.1787,
      "step": 3984,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.29230444711428233,
      "lm_loss": 1.0,
      "step": 3984,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.29230444711428233,
      "lm_loss": 1.0503,
      "step": 3984,
      "vm_loss": 0.1838
    },
    {
      "epoch": 0.29230444711428233,
      "lm_loss": 0.0653,
      "step": 3984,
      "vm_loss": 0.1733
    },
    {
      "epoch": 0.29230444711428233,
      "lm_loss": 1.1272,
      "step": 3984,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.29230444711428233,
      "lm_loss": 1.1569,
      "step": 3984,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.29230444711428233,
      "lm_loss": 0.91,
      "step": 3984,
      "vm_loss": 0.1212
    },
    {
      "epoch": 0.2923778167044214,
      "grad_norm": 1.4834338531643876,
      "learning_rate": 8.300672816292568e-06,
      "loss": 0.8593,
      "step": 3985
    },
    {
      "epoch": 0.29245118629456057,
      "grad_norm": 1.4788544301325288,
      "learning_rate": 8.29978021212204e-06,
      "loss": 0.7823,
      "step": 3986
    },
    {
      "epoch": 0.29252455588469967,
      "grad_norm": 1.5960833059753958,
      "learning_rate": 8.298887421604751e-06,
      "loss": 0.7198,
      "step": 3987
    },
    {
      "epoch": 0.2925979254748388,
      "grad_norm": 1.5486891831556657,
      "learning_rate": 8.29799444479112e-06,
      "loss": 0.811,
      "step": 3988
    },
    {
      "epoch": 0.29267129506497797,
      "grad_norm": 1.1195338209269348,
      "learning_rate": 8.297101281731576e-06,
      "loss": 0.5401,
      "step": 3989
    },
    {
      "epoch": 0.29274466465511706,
      "grad_norm": 0.919501786501421,
      "learning_rate": 8.296207932476557e-06,
      "loss": 0.5448,
      "step": 3990
    },
    {
      "epoch": 0.2928180342452562,
      "grad_norm": 1.7263421379397137,
      "learning_rate": 8.295314397076513e-06,
      "loss": 0.9637,
      "step": 3991
    },
    {
      "epoch": 0.2928914038353953,
      "grad_norm": 1.5577079384576313,
      "learning_rate": 8.294420675581904e-06,
      "loss": 0.7729,
      "step": 3992
    },
    {
      "epoch": 0.2928914038353953,
      "lm_loss": 0.8129,
      "step": 3992,
      "vm_loss": 0.0483
    },
    {
      "epoch": 0.2928914038353953,
      "lm_loss": 1.3735,
      "step": 3992,
      "vm_loss": 0.1214
    },
    {
      "epoch": 0.2928914038353953,
      "lm_loss": 0.8918,
      "step": 3992,
      "vm_loss": 0.1232
    },
    {
      "epoch": 0.2928914038353953,
      "lm_loss": 0.606,
      "step": 3992,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.2928914038353953,
      "lm_loss": 0.2255,
      "step": 3992,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.2928914038353953,
      "lm_loss": 0.1636,
      "step": 3992,
      "vm_loss": 0.0582
    },
    {
      "epoch": 0.2928914038353953,
      "lm_loss": 0.6159,
      "step": 3992,
      "vm_loss": 0.1804
    },
    {
      "epoch": 0.2928914038353953,
      "lm_loss": 0.0926,
      "step": 3992,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.29296477342553445,
      "grad_norm": 1.5295870915415417,
      "learning_rate": 8.293526768043204e-06,
      "loss": 0.8423,
      "step": 3993
    },
    {
      "epoch": 0.2930381430156736,
      "grad_norm": 1.9552809910226423,
      "learning_rate": 8.292632674510888e-06,
      "loss": 0.8637,
      "step": 3994
    },
    {
      "epoch": 0.2931115126058127,
      "grad_norm": 1.0507373698866473,
      "learning_rate": 8.291738395035451e-06,
      "loss": 0.5368,
      "step": 3995
    },
    {
      "epoch": 0.29318488219595185,
      "grad_norm": 0.9417033169983425,
      "learning_rate": 8.290843929667394e-06,
      "loss": 0.5443,
      "step": 3996
    },
    {
      "epoch": 0.29325825178609094,
      "grad_norm": 1.7038897812364573,
      "learning_rate": 8.289949278457234e-06,
      "loss": 0.8899,
      "step": 3997
    },
    {
      "epoch": 0.2933316213762301,
      "grad_norm": 1.6279808932145368,
      "learning_rate": 8.289054441455489e-06,
      "loss": 0.7927,
      "step": 3998
    },
    {
      "epoch": 0.29340499096636924,
      "grad_norm": 1.587201573659257,
      "learning_rate": 8.288159418712693e-06,
      "loss": 0.7923,
      "step": 3999
    },
    {
      "epoch": 0.29347836055650833,
      "grad_norm": 1.7301689226244208,
      "learning_rate": 8.287264210279393e-06,
      "loss": 0.9007,
      "step": 4000
    },
    {
      "epoch": 0.29347836055650833,
      "lm_loss": 0.7353,
      "step": 4000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29347836055650833,
      "lm_loss": 0.5725,
      "step": 4000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29347836055650833,
      "lm_loss": 0.4841,
      "step": 4000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29347836055650833,
      "lm_loss": 0.4406,
      "step": 4000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29347836055650833,
      "lm_loss": 0.428,
      "step": 4000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29347836055650833,
      "lm_loss": 0.5282,
      "step": 4000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29347836055650833,
      "lm_loss": 0.5791,
      "step": 4000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29347836055650833,
      "lm_loss": 0.6686,
      "step": 4000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2935517301466475,
      "grad_norm": 0.6580418222710264,
      "learning_rate": 8.286368816206144e-06,
      "loss": 0.5311,
      "step": 4001
    },
    {
      "epoch": 0.2936250997367866,
      "grad_norm": 2.004403842765528,
      "learning_rate": 8.285473236543507e-06,
      "loss": 0.8653,
      "step": 4002
    },
    {
      "epoch": 0.2936984693269257,
      "grad_norm": 1.592936930086206,
      "learning_rate": 8.28457747134206e-06,
      "loss": 0.8012,
      "step": 4003
    },
    {
      "epoch": 0.2937718389170649,
      "grad_norm": 1.5606978423976838,
      "learning_rate": 8.28368152065239e-06,
      "loss": 0.8765,
      "step": 4004
    },
    {
      "epoch": 0.29384520850720397,
      "grad_norm": 0.6808446166915473,
      "learning_rate": 8.282785384525093e-06,
      "loss": 0.5228,
      "step": 4005
    },
    {
      "epoch": 0.2939185780973431,
      "grad_norm": 1.6621435851988902,
      "learning_rate": 8.281889063010775e-06,
      "loss": 0.8293,
      "step": 4006
    },
    {
      "epoch": 0.2939919476874822,
      "grad_norm": 0.5970513949048817,
      "learning_rate": 8.280992556160055e-06,
      "loss": 0.5403,
      "step": 4007
    },
    {
      "epoch": 0.29406531727762136,
      "grad_norm": 0.5964250512756073,
      "learning_rate": 8.280095864023558e-06,
      "loss": 0.5428,
      "step": 4008
    },
    {
      "epoch": 0.29406531727762136,
      "lm_loss": 0.4255,
      "step": 4008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29406531727762136,
      "lm_loss": 0.6519,
      "step": 4008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29406531727762136,
      "lm_loss": 0.3741,
      "step": 4008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29406531727762136,
      "lm_loss": 0.2814,
      "step": 4008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29406531727762136,
      "lm_loss": 0.3348,
      "step": 4008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29406531727762136,
      "lm_loss": 0.5856,
      "step": 4008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29406531727762136,
      "lm_loss": 0.396,
      "step": 4008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29406531727762136,
      "lm_loss": 0.5583,
      "step": 4008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2941386868677605,
      "grad_norm": 0.617492455527388,
      "learning_rate": 8.279198986651925e-06,
      "loss": 0.5289,
      "step": 4009
    },
    {
      "epoch": 0.2942120564578996,
      "grad_norm": 1.644859350008741,
      "learning_rate": 8.278301924095806e-06,
      "loss": 0.8665,
      "step": 4010
    },
    {
      "epoch": 0.29428542604803876,
      "grad_norm": 1.5023026983388337,
      "learning_rate": 8.277404676405859e-06,
      "loss": 0.8343,
      "step": 4011
    },
    {
      "epoch": 0.29435879563817785,
      "grad_norm": 0.5973129548022662,
      "learning_rate": 8.276507243632753e-06,
      "loss": 0.5307,
      "step": 4012
    },
    {
      "epoch": 0.294432165228317,
      "grad_norm": 1.6372681149860242,
      "learning_rate": 8.275609625827168e-06,
      "loss": 0.8194,
      "step": 4013
    },
    {
      "epoch": 0.2945055348184561,
      "grad_norm": 1.6711431521257327,
      "learning_rate": 8.274711823039796e-06,
      "loss": 0.934,
      "step": 4014
    },
    {
      "epoch": 0.29457890440859524,
      "grad_norm": 1.5937200985143487,
      "learning_rate": 8.273813835321337e-06,
      "loss": 0.8831,
      "step": 4015
    },
    {
      "epoch": 0.2946522739987344,
      "grad_norm": 1.6805103813857971,
      "learning_rate": 8.272915662722504e-06,
      "loss": 0.8817,
      "step": 4016
    },
    {
      "epoch": 0.2946522739987344,
      "lm_loss": 0.1421,
      "step": 4016,
      "vm_loss": 0.0984
    },
    {
      "epoch": 0.2946522739987344,
      "lm_loss": 1.4618,
      "step": 4016,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.2946522739987344,
      "lm_loss": 0.1532,
      "step": 4016,
      "vm_loss": 0.0776
    },
    {
      "epoch": 0.2946522739987344,
      "lm_loss": 0.2117,
      "step": 4016,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.2946522739987344,
      "lm_loss": 0.2035,
      "step": 4016,
      "vm_loss": 0.1777
    },
    {
      "epoch": 0.2946522739987344,
      "lm_loss": 0.8801,
      "step": 4016,
      "vm_loss": 0.1662
    },
    {
      "epoch": 0.2946522739987344,
      "lm_loss": 1.9786,
      "step": 4016,
      "vm_loss": 0.2246
    },
    {
      "epoch": 0.2946522739987344,
      "lm_loss": 0.1038,
      "step": 4016,
      "vm_loss": 0.139
    },
    {
      "epoch": 0.2947256435888735,
      "grad_norm": 1.5897319262399638,
      "learning_rate": 8.272017305294018e-06,
      "loss": 0.8597,
      "step": 4017
    },
    {
      "epoch": 0.29479901317901264,
      "grad_norm": 1.4531727807867614,
      "learning_rate": 8.271118763086612e-06,
      "loss": 0.7193,
      "step": 4018
    },
    {
      "epoch": 0.29487238276915173,
      "grad_norm": 1.5484331314106008,
      "learning_rate": 8.270220036151028e-06,
      "loss": 0.7802,
      "step": 4019
    },
    {
      "epoch": 0.2949457523592909,
      "grad_norm": 1.4370628801744558,
      "learning_rate": 8.26932112453802e-06,
      "loss": 0.7934,
      "step": 4020
    },
    {
      "epoch": 0.29501912194943003,
      "grad_norm": 1.4691555431534922,
      "learning_rate": 8.268422028298355e-06,
      "loss": 0.8287,
      "step": 4021
    },
    {
      "epoch": 0.2950924915395691,
      "grad_norm": 1.7730684510605241,
      "learning_rate": 8.2675227474828e-06,
      "loss": 0.9326,
      "step": 4022
    },
    {
      "epoch": 0.2951658611297083,
      "grad_norm": 1.7534227478669127,
      "learning_rate": 8.266623282142146e-06,
      "loss": 0.9664,
      "step": 4023
    },
    {
      "epoch": 0.29523923071984737,
      "grad_norm": 1.4600511965971228,
      "learning_rate": 8.265723632327182e-06,
      "loss": 0.8936,
      "step": 4024
    },
    {
      "epoch": 0.29523923071984737,
      "lm_loss": 1.4965,
      "step": 4024,
      "vm_loss": 0.0619
    },
    {
      "epoch": 0.29523923071984737,
      "lm_loss": 1.0668,
      "step": 4024,
      "vm_loss": 0.1852
    },
    {
      "epoch": 0.29523923071984737,
      "lm_loss": 0.9403,
      "step": 4024,
      "vm_loss": 0.149
    },
    {
      "epoch": 0.29523923071984737,
      "lm_loss": 0.1531,
      "step": 4024,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.29523923071984737,
      "lm_loss": 1.0828,
      "step": 4024,
      "vm_loss": 0.2054
    },
    {
      "epoch": 0.29523923071984737,
      "lm_loss": 0.9289,
      "step": 4024,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.29523923071984737,
      "lm_loss": 0.9592,
      "step": 4024,
      "vm_loss": 0.0872
    },
    {
      "epoch": 0.29523923071984737,
      "lm_loss": 0.1207,
      "step": 4024,
      "vm_loss": 0.0634
    },
    {
      "epoch": 0.2953126003099865,
      "grad_norm": 1.5731047203144117,
      "learning_rate": 8.264823798088721e-06,
      "loss": 0.829,
      "step": 4025
    },
    {
      "epoch": 0.29538596990012567,
      "grad_norm": 0.685348030108762,
      "learning_rate": 8.263923779477574e-06,
      "loss": 0.5212,
      "step": 4026
    },
    {
      "epoch": 0.29545933949026476,
      "grad_norm": 1.4430487284603857,
      "learning_rate": 8.263023576544566e-06,
      "loss": 0.7754,
      "step": 4027
    },
    {
      "epoch": 0.2955327090804039,
      "grad_norm": 1.6702306749006348,
      "learning_rate": 8.262123189340539e-06,
      "loss": 0.7903,
      "step": 4028
    },
    {
      "epoch": 0.295606078670543,
      "grad_norm": 1.4385147563368827,
      "learning_rate": 8.261222617916335e-06,
      "loss": 0.8277,
      "step": 4029
    },
    {
      "epoch": 0.29567944826068215,
      "grad_norm": 0.5902835846120924,
      "learning_rate": 8.260321862322814e-06,
      "loss": 0.5166,
      "step": 4030
    },
    {
      "epoch": 0.2957528178508213,
      "grad_norm": 1.4247188804127502,
      "learning_rate": 8.259420922610843e-06,
      "loss": 0.7814,
      "step": 4031
    },
    {
      "epoch": 0.2958261874409604,
      "grad_norm": 0.5804823234405212,
      "learning_rate": 8.258519798831302e-06,
      "loss": 0.5548,
      "step": 4032
    },
    {
      "epoch": 0.2958261874409604,
      "lm_loss": 1.0198,
      "step": 4032,
      "vm_loss": 0.0548
    },
    {
      "epoch": 0.2958261874409604,
      "lm_loss": 0.9093,
      "step": 4032,
      "vm_loss": 0.1429
    },
    {
      "epoch": 0.2958261874409604,
      "lm_loss": 0.2421,
      "step": 4032,
      "vm_loss": 0.17
    },
    {
      "epoch": 0.2958261874409604,
      "lm_loss": 1.0799,
      "step": 4032,
      "vm_loss": 0.2238
    },
    {
      "epoch": 0.2958261874409604,
      "lm_loss": 0.6703,
      "step": 4032,
      "vm_loss": 0.1249
    },
    {
      "epoch": 0.2958261874409604,
      "lm_loss": 0.9231,
      "step": 4032,
      "vm_loss": 0.1997
    },
    {
      "epoch": 0.2958261874409604,
      "lm_loss": 0.0875,
      "step": 4032,
      "vm_loss": 0.1701
    },
    {
      "epoch": 0.2958261874409604,
      "lm_loss": 0.0787,
      "step": 4032,
      "vm_loss": 0.096
    },
    {
      "epoch": 0.29589955703109955,
      "grad_norm": 1.754727952568952,
      "learning_rate": 8.257618491035077e-06,
      "loss": 0.8181,
      "step": 4033
    },
    {
      "epoch": 0.29597292662123864,
      "grad_norm": 1.4377299751746817,
      "learning_rate": 8.25671699927307e-06,
      "loss": 0.8394,
      "step": 4034
    },
    {
      "epoch": 0.2960462962113778,
      "grad_norm": 1.4718089493586968,
      "learning_rate": 8.255815323596185e-06,
      "loss": 0.7784,
      "step": 4035
    },
    {
      "epoch": 0.29611966580151694,
      "grad_norm": 1.4209089897874756,
      "learning_rate": 8.25491346405535e-06,
      "loss": 0.8587,
      "step": 4036
    },
    {
      "epoch": 0.29619303539165603,
      "grad_norm": 1.4939690441404632,
      "learning_rate": 8.25401142070149e-06,
      "loss": 1.0282,
      "step": 4037
    },
    {
      "epoch": 0.2962664049817952,
      "grad_norm": 0.6447423903285648,
      "learning_rate": 8.253109193585547e-06,
      "loss": 0.5425,
      "step": 4038
    },
    {
      "epoch": 0.2963397745719343,
      "grad_norm": 1.5334044099261332,
      "learning_rate": 8.25220678275847e-06,
      "loss": 0.8647,
      "step": 4039
    },
    {
      "epoch": 0.2964131441620734,
      "grad_norm": 1.3656990051810747,
      "learning_rate": 8.251304188271224e-06,
      "loss": 0.8567,
      "step": 4040
    },
    {
      "epoch": 0.2964131441620734,
      "lm_loss": 1.0358,
      "step": 4040,
      "vm_loss": 0.041
    },
    {
      "epoch": 0.2964131441620734,
      "lm_loss": 0.6001,
      "step": 4040,
      "vm_loss": 0.0866
    },
    {
      "epoch": 0.2964131441620734,
      "lm_loss": 0.8685,
      "step": 4040,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.2964131441620734,
      "lm_loss": 1.3364,
      "step": 4040,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.2964131441620734,
      "lm_loss": 0.9221,
      "step": 4040,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.2964131441620734,
      "lm_loss": 0.5035,
      "step": 4040,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.2964131441620734,
      "lm_loss": 1.2249,
      "step": 4040,
      "vm_loss": 0.1645
    },
    {
      "epoch": 0.2964131441620734,
      "lm_loss": 0.346,
      "step": 4040,
      "vm_loss": 0.2091
    },
    {
      "epoch": 0.2964865137522126,
      "grad_norm": 1.5824317863615822,
      "learning_rate": 8.250401410174777e-06,
      "loss": 0.9309,
      "step": 4041
    },
    {
      "epoch": 0.29655988334235167,
      "grad_norm": 1.4861075527789254,
      "learning_rate": 8.249498448520113e-06,
      "loss": 0.9149,
      "step": 4042
    },
    {
      "epoch": 0.2966332529324908,
      "grad_norm": 0.6019413598537797,
      "learning_rate": 8.248595303358225e-06,
      "loss": 0.5124,
      "step": 4043
    },
    {
      "epoch": 0.2967066225226299,
      "grad_norm": 0.5880648376958185,
      "learning_rate": 8.247691974740116e-06,
      "loss": 0.5263,
      "step": 4044
    },
    {
      "epoch": 0.29677999211276906,
      "grad_norm": 0.56018926992643,
      "learning_rate": 8.246788462716797e-06,
      "loss": 0.5439,
      "step": 4045
    },
    {
      "epoch": 0.2968533617029082,
      "grad_norm": 1.654727784928455,
      "learning_rate": 8.245884767339293e-06,
      "loss": 0.8598,
      "step": 4046
    },
    {
      "epoch": 0.2969267312930473,
      "grad_norm": 0.550092746737632,
      "learning_rate": 8.24498088865864e-06,
      "loss": 0.5303,
      "step": 4047
    },
    {
      "epoch": 0.29700010088318646,
      "grad_norm": 1.4034392874107233,
      "learning_rate": 8.244076826725877e-06,
      "loss": 0.9055,
      "step": 4048
    },
    {
      "epoch": 0.29700010088318646,
      "lm_loss": 0.1076,
      "step": 4048,
      "vm_loss": 0.0524
    },
    {
      "epoch": 0.29700010088318646,
      "lm_loss": 0.4811,
      "step": 4048,
      "vm_loss": 0.0683
    },
    {
      "epoch": 0.29700010088318646,
      "lm_loss": 0.3009,
      "step": 4048,
      "vm_loss": 0.1404
    },
    {
      "epoch": 0.29700010088318646,
      "lm_loss": 0.0598,
      "step": 4048,
      "vm_loss": 0.1905
    },
    {
      "epoch": 0.29700010088318646,
      "lm_loss": 0.8896,
      "step": 4048,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.29700010088318646,
      "lm_loss": 0.0826,
      "step": 4048,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.29700010088318646,
      "lm_loss": 0.0531,
      "step": 4048,
      "vm_loss": 0.2195
    },
    {
      "epoch": 0.29700010088318646,
      "lm_loss": 0.3301,
      "step": 4048,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.29707347047332555,
      "grad_norm": 1.6273889623482343,
      "learning_rate": 8.243172581592066e-06,
      "loss": 0.786,
      "step": 4049
    },
    {
      "epoch": 0.2971468400634647,
      "grad_norm": 0.6239246959094069,
      "learning_rate": 8.242268153308264e-06,
      "loss": 0.5588,
      "step": 4050
    },
    {
      "epoch": 0.2972202096536038,
      "grad_norm": 1.7069048811261331,
      "learning_rate": 8.241363541925551e-06,
      "loss": 0.7957,
      "step": 4051
    },
    {
      "epoch": 0.29729357924374294,
      "grad_norm": 0.590791823716327,
      "learning_rate": 8.24045874749501e-06,
      "loss": 0.5475,
      "step": 4052
    },
    {
      "epoch": 0.2973669488338821,
      "grad_norm": 1.5442992155543336,
      "learning_rate": 8.239553770067744e-06,
      "loss": 0.7061,
      "step": 4053
    },
    {
      "epoch": 0.2974403184240212,
      "grad_norm": 1.466914630829618,
      "learning_rate": 8.23864860969485e-06,
      "loss": 0.8018,
      "step": 4054
    },
    {
      "epoch": 0.29751368801416034,
      "grad_norm": 0.5351200917990279,
      "learning_rate": 8.237743266427448e-06,
      "loss": 0.5154,
      "step": 4055
    },
    {
      "epoch": 0.29758705760429943,
      "grad_norm": 1.6392229257829989,
      "learning_rate": 8.236837740316667e-06,
      "loss": 0.8724,
      "step": 4056
    },
    {
      "epoch": 0.29758705760429943,
      "lm_loss": 0.9564,
      "step": 4056,
      "vm_loss": 0.0876
    },
    {
      "epoch": 0.29758705760429943,
      "lm_loss": 1.3569,
      "step": 4056,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.29758705760429943,
      "lm_loss": 0.2972,
      "step": 4056,
      "vm_loss": 0.057
    },
    {
      "epoch": 0.29758705760429943,
      "lm_loss": 0.1141,
      "step": 4056,
      "vm_loss": 0.0974
    },
    {
      "epoch": 0.29758705760429943,
      "lm_loss": 1.1107,
      "step": 4056,
      "vm_loss": 0.0818
    },
    {
      "epoch": 0.29758705760429943,
      "lm_loss": 0.1582,
      "step": 4056,
      "vm_loss": 0.0609
    },
    {
      "epoch": 0.29758705760429943,
      "lm_loss": 0.3856,
      "step": 4056,
      "vm_loss": 0.1697
    },
    {
      "epoch": 0.29758705760429943,
      "lm_loss": 0.8385,
      "step": 4056,
      "vm_loss": 0.1032
    },
    {
      "epoch": 0.2976604271944386,
      "grad_norm": 1.4788766381325746,
      "learning_rate": 8.235932031413643e-06,
      "loss": 0.8073,
      "step": 4057
    },
    {
      "epoch": 0.29773379678457773,
      "grad_norm": 1.4302330816436448,
      "learning_rate": 8.235026139769523e-06,
      "loss": 0.9106,
      "step": 4058
    },
    {
      "epoch": 0.2978071663747168,
      "grad_norm": 1.5579296298553473,
      "learning_rate": 8.234120065435466e-06,
      "loss": 0.8567,
      "step": 4059
    },
    {
      "epoch": 0.297880535964856,
      "grad_norm": 0.6313882895245287,
      "learning_rate": 8.23321380846264e-06,
      "loss": 0.5369,
      "step": 4060
    },
    {
      "epoch": 0.29795390555499507,
      "grad_norm": 0.6256041964952573,
      "learning_rate": 8.232307368902222e-06,
      "loss": 0.5384,
      "step": 4061
    },
    {
      "epoch": 0.2980272751451342,
      "grad_norm": 1.6489478014075307,
      "learning_rate": 8.231400746805403e-06,
      "loss": 0.8201,
      "step": 4062
    },
    {
      "epoch": 0.29810064473527337,
      "grad_norm": 1.5086014923835607,
      "learning_rate": 8.230493942223382e-06,
      "loss": 0.8518,
      "step": 4063
    },
    {
      "epoch": 0.29817401432541246,
      "grad_norm": 1.5502739658105265,
      "learning_rate": 8.229586955207366e-06,
      "loss": 0.7922,
      "step": 4064
    },
    {
      "epoch": 0.29817401432541246,
      "lm_loss": 0.627,
      "step": 4064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29817401432541246,
      "lm_loss": 0.5195,
      "step": 4064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29817401432541246,
      "lm_loss": 0.4589,
      "step": 4064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29817401432541246,
      "lm_loss": 0.6186,
      "step": 4064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29817401432541246,
      "lm_loss": 0.4842,
      "step": 4064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29817401432541246,
      "lm_loss": 0.5583,
      "step": 4064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29817401432541246,
      "lm_loss": 0.467,
      "step": 4064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.29817401432541246,
      "lm_loss": 0.3947,
      "step": 4064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.2982473839155516,
      "grad_norm": 0.6021273114189051,
      "learning_rate": 8.228679785808578e-06,
      "loss": 0.5389,
      "step": 4065
    },
    {
      "epoch": 0.2983207535056907,
      "grad_norm": 0.5885069176658657,
      "learning_rate": 8.227772434078247e-06,
      "loss": 0.5409,
      "step": 4066
    },
    {
      "epoch": 0.29839412309582986,
      "grad_norm": 2.083273938283552,
      "learning_rate": 8.226864900067612e-06,
      "loss": 0.805,
      "step": 4067
    },
    {
      "epoch": 0.298467492685969,
      "grad_norm": 0.5806066813908421,
      "learning_rate": 8.225957183827928e-06,
      "loss": 0.5496,
      "step": 4068
    },
    {
      "epoch": 0.2985408622761081,
      "grad_norm": 1.7060185087528066,
      "learning_rate": 8.22504928541045e-06,
      "loss": 0.8859,
      "step": 4069
    },
    {
      "epoch": 0.29861423186624725,
      "grad_norm": 1.5437299909901474,
      "learning_rate": 8.224141204866453e-06,
      "loss": 0.8316,
      "step": 4070
    },
    {
      "epoch": 0.29868760145638634,
      "grad_norm": 1.5535196662995987,
      "learning_rate": 8.223232942247216e-06,
      "loss": 0.7491,
      "step": 4071
    },
    {
      "epoch": 0.2987609710465255,
      "grad_norm": 1.5013844141150814,
      "learning_rate": 8.222324497604035e-06,
      "loss": 0.8397,
      "step": 4072
    },
    {
      "epoch": 0.2987609710465255,
      "lm_loss": 0.2299,
      "step": 4072,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.2987609710465255,
      "lm_loss": 1.5757,
      "step": 4072,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.2987609710465255,
      "lm_loss": 0.9473,
      "step": 4072,
      "vm_loss": 0.1585
    },
    {
      "epoch": 0.2987609710465255,
      "lm_loss": 0.3736,
      "step": 4072,
      "vm_loss": 0.1342
    },
    {
      "epoch": 0.2987609710465255,
      "lm_loss": 0.835,
      "step": 4072,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.2987609710465255,
      "lm_loss": 0.6062,
      "step": 4072,
      "vm_loss": 0.1189
    },
    {
      "epoch": 0.2987609710465255,
      "lm_loss": 0.0841,
      "step": 4072,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.2987609710465255,
      "lm_loss": 0.085,
      "step": 4072,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.29883434063666464,
      "grad_norm": 1.5359326505502096,
      "learning_rate": 8.22141587098821e-06,
      "loss": 0.8555,
      "step": 4073
    },
    {
      "epoch": 0.29890771022680374,
      "grad_norm": 1.641030402771488,
      "learning_rate": 8.220507062451051e-06,
      "loss": 0.7991,
      "step": 4074
    },
    {
      "epoch": 0.2989810798169429,
      "grad_norm": 1.5250553061639696,
      "learning_rate": 8.219598072043884e-06,
      "loss": 0.7374,
      "step": 4075
    },
    {
      "epoch": 0.299054449407082,
      "grad_norm": 1.776285387767044,
      "learning_rate": 8.21868889981804e-06,
      "loss": 0.7958,
      "step": 4076
    },
    {
      "epoch": 0.29912781899722113,
      "grad_norm": 1.7661240858705003,
      "learning_rate": 8.217779545824864e-06,
      "loss": 0.836,
      "step": 4077
    },
    {
      "epoch": 0.2992011885873603,
      "grad_norm": 1.4987327828902037,
      "learning_rate": 8.216870010115707e-06,
      "loss": 0.8377,
      "step": 4078
    },
    {
      "epoch": 0.2992745581774994,
      "grad_norm": 1.5346900513333412,
      "learning_rate": 8.215960292741933e-06,
      "loss": 0.8535,
      "step": 4079
    },
    {
      "epoch": 0.2993479277676385,
      "grad_norm": 0.6693905442147726,
      "learning_rate": 8.21505039375492e-06,
      "loss": 0.5233,
      "step": 4080
    },
    {
      "epoch": 0.2993479277676385,
      "lm_loss": 1.7814,
      "step": 4080,
      "vm_loss": 0.0966
    },
    {
      "epoch": 0.2993479277676385,
      "lm_loss": 0.5582,
      "step": 4080,
      "vm_loss": 0.1244
    },
    {
      "epoch": 0.2993479277676385,
      "lm_loss": 0.4441,
      "step": 4080,
      "vm_loss": 0.0924
    },
    {
      "epoch": 0.2993479277676385,
      "lm_loss": 0.8152,
      "step": 4080,
      "vm_loss": 0.2109
    },
    {
      "epoch": 0.2993479277676385,
      "lm_loss": 0.7763,
      "step": 4080,
      "vm_loss": 0.087
    },
    {
      "epoch": 0.2993479277676385,
      "lm_loss": 0.8989,
      "step": 4080,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.2993479277676385,
      "lm_loss": 0.1607,
      "step": 4080,
      "vm_loss": 0.0848
    },
    {
      "epoch": 0.2993479277676385,
      "lm_loss": 1.3306,
      "step": 4080,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.2994212973577776,
      "grad_norm": 1.4964679301448902,
      "learning_rate": 8.214140313206047e-06,
      "loss": 0.8094,
      "step": 4081
    },
    {
      "epoch": 0.29949466694791677,
      "grad_norm": 0.5984560986125723,
      "learning_rate": 8.213230051146712e-06,
      "loss": 0.5576,
      "step": 4082
    },
    {
      "epoch": 0.2995680365380559,
      "grad_norm": 1.7599852719591536,
      "learning_rate": 8.212319607628319e-06,
      "loss": 0.931,
      "step": 4083
    },
    {
      "epoch": 0.299641406128195,
      "grad_norm": 1.3901246800906535,
      "learning_rate": 8.211408982702283e-06,
      "loss": 0.7358,
      "step": 4084
    },
    {
      "epoch": 0.29971477571833416,
      "grad_norm": 1.4897731971021142,
      "learning_rate": 8.210498176420027e-06,
      "loss": 0.7659,
      "step": 4085
    },
    {
      "epoch": 0.29978814530847325,
      "grad_norm": 1.661926292472413,
      "learning_rate": 8.20958718883299e-06,
      "loss": 0.9064,
      "step": 4086
    },
    {
      "epoch": 0.2998615148986124,
      "grad_norm": 1.6609773667498495,
      "learning_rate": 8.208676019992614e-06,
      "loss": 0.8004,
      "step": 4087
    },
    {
      "epoch": 0.29993488448875155,
      "grad_norm": 1.511196793822372,
      "learning_rate": 8.207764669950358e-06,
      "loss": 0.8091,
      "step": 4088
    },
    {
      "epoch": 0.29993488448875155,
      "lm_loss": 1.1054,
      "step": 4088,
      "vm_loss": 0.0565
    },
    {
      "epoch": 0.29993488448875155,
      "lm_loss": 0.9982,
      "step": 4088,
      "vm_loss": 0.1209
    },
    {
      "epoch": 0.29993488448875155,
      "lm_loss": 1.1502,
      "step": 4088,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.29993488448875155,
      "lm_loss": 0.5982,
      "step": 4088,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.29993488448875155,
      "lm_loss": 0.846,
      "step": 4088,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.29993488448875155,
      "lm_loss": 0.4894,
      "step": 4088,
      "vm_loss": 0.1304
    },
    {
      "epoch": 0.29993488448875155,
      "lm_loss": 1.0799,
      "step": 4088,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.29993488448875155,
      "lm_loss": 0.9532,
      "step": 4088,
      "vm_loss": 0.2126
    },
    {
      "epoch": 0.30000825407889065,
      "grad_norm": 1.4348020420463754,
      "learning_rate": 8.206853138757687e-06,
      "loss": 0.8477,
      "step": 4089
    },
    {
      "epoch": 0.3000816236690298,
      "grad_norm": 1.4451483843945232,
      "learning_rate": 8.205941426466079e-06,
      "loss": 0.8354,
      "step": 4090
    },
    {
      "epoch": 0.3001549932591689,
      "grad_norm": 0.6976031901715869,
      "learning_rate": 8.205029533127017e-06,
      "loss": 0.534,
      "step": 4091
    },
    {
      "epoch": 0.30022836284930804,
      "grad_norm": 0.6400825084413393,
      "learning_rate": 8.204117458792002e-06,
      "loss": 0.5148,
      "step": 4092
    },
    {
      "epoch": 0.30030173243944713,
      "grad_norm": 0.5916836812751001,
      "learning_rate": 8.203205203512539e-06,
      "loss": 0.5264,
      "step": 4093
    },
    {
      "epoch": 0.3003751020295863,
      "grad_norm": 1.7467165125583684,
      "learning_rate": 8.202292767340145e-06,
      "loss": 0.8154,
      "step": 4094
    },
    {
      "epoch": 0.30044847161972543,
      "grad_norm": 0.5967312189123621,
      "learning_rate": 8.201380150326348e-06,
      "loss": 0.5224,
      "step": 4095
    },
    {
      "epoch": 0.3005218412098645,
      "grad_norm": 0.6360070405453299,
      "learning_rate": 8.200467352522687e-06,
      "loss": 0.5349,
      "step": 4096
    },
    {
      "epoch": 0.3005218412098645,
      "lm_loss": 1.4649,
      "step": 4096,
      "vm_loss": 0.0856
    },
    {
      "epoch": 0.3005218412098645,
      "lm_loss": 0.8729,
      "step": 4096,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.3005218412098645,
      "lm_loss": 0.831,
      "step": 4096,
      "vm_loss": 0.2269
    },
    {
      "epoch": 0.3005218412098645,
      "lm_loss": 0.5247,
      "step": 4096,
      "vm_loss": 0.0335
    },
    {
      "epoch": 0.3005218412098645,
      "lm_loss": 0.9849,
      "step": 4096,
      "vm_loss": 0.1295
    },
    {
      "epoch": 0.3005218412098645,
      "lm_loss": 0.0648,
      "step": 4096,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.3005218412098645,
      "lm_loss": 0.1267,
      "step": 4096,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.3005218412098645,
      "lm_loss": 0.442,
      "step": 4096,
      "vm_loss": 0.1595
    },
    {
      "epoch": 0.3005952108000037,
      "grad_norm": 1.6629069802751073,
      "learning_rate": 8.199554373980707e-06,
      "loss": 0.851,
      "step": 4097
    },
    {
      "epoch": 0.30066858039014277,
      "grad_norm": 1.8473047658648623,
      "learning_rate": 8.198641214751968e-06,
      "loss": 0.7922,
      "step": 4098
    },
    {
      "epoch": 0.3007419499802819,
      "grad_norm": 1.5280312182346973,
      "learning_rate": 8.19772787488804e-06,
      "loss": 0.9168,
      "step": 4099
    },
    {
      "epoch": 0.30081531957042107,
      "grad_norm": 1.4237198773772999,
      "learning_rate": 8.196814354440499e-06,
      "loss": 0.7938,
      "step": 4100
    },
    {
      "epoch": 0.30088868916056016,
      "grad_norm": 1.5874245860467497,
      "learning_rate": 8.195900653460935e-06,
      "loss": 0.7691,
      "step": 4101
    },
    {
      "epoch": 0.3009620587506993,
      "grad_norm": 0.6267269159769898,
      "learning_rate": 8.194986772000946e-06,
      "loss": 0.5202,
      "step": 4102
    },
    {
      "epoch": 0.3010354283408384,
      "grad_norm": 1.4798951847103032,
      "learning_rate": 8.194072710112142e-06,
      "loss": 0.7775,
      "step": 4103
    },
    {
      "epoch": 0.30110879793097756,
      "grad_norm": 0.6180079791185806,
      "learning_rate": 8.193158467846143e-06,
      "loss": 0.5038,
      "step": 4104
    },
    {
      "epoch": 0.30110879793097756,
      "lm_loss": 0.7531,
      "step": 4104,
      "vm_loss": 0.0576
    },
    {
      "epoch": 0.30110879793097756,
      "lm_loss": 0.2892,
      "step": 4104,
      "vm_loss": 0.0519
    },
    {
      "epoch": 0.30110879793097756,
      "lm_loss": 1.1945,
      "step": 4104,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.30110879793097756,
      "lm_loss": 0.9142,
      "step": 4104,
      "vm_loss": 0.0394
    },
    {
      "epoch": 0.30110879793097756,
      "lm_loss": 0.7331,
      "step": 4104,
      "vm_loss": 0.144
    },
    {
      "epoch": 0.30110879793097756,
      "lm_loss": 0.5385,
      "step": 4104,
      "vm_loss": 0.1809
    },
    {
      "epoch": 0.30110879793097756,
      "lm_loss": 0.893,
      "step": 4104,
      "vm_loss": 0.1563
    },
    {
      "epoch": 0.30110879793097756,
      "lm_loss": 0.1139,
      "step": 4104,
      "vm_loss": 0.1801
    },
    {
      "epoch": 0.3011821675211167,
      "grad_norm": 1.4774198118939554,
      "learning_rate": 8.192244045254576e-06,
      "loss": 0.7831,
      "step": 4105
    },
    {
      "epoch": 0.3012555371112558,
      "grad_norm": 1.763244797289,
      "learning_rate": 8.191329442389083e-06,
      "loss": 0.7459,
      "step": 4106
    },
    {
      "epoch": 0.30132890670139495,
      "grad_norm": 1.363066266812824,
      "learning_rate": 8.190414659301314e-06,
      "loss": 0.8528,
      "step": 4107
    },
    {
      "epoch": 0.30140227629153404,
      "grad_norm": 1.4850199165339348,
      "learning_rate": 8.189499696042927e-06,
      "loss": 0.7959,
      "step": 4108
    },
    {
      "epoch": 0.3014756458816732,
      "grad_norm": 1.6568517796292008,
      "learning_rate": 8.188584552665592e-06,
      "loss": 0.9466,
      "step": 4109
    },
    {
      "epoch": 0.30154901547181234,
      "grad_norm": 1.6865491545382483,
      "learning_rate": 8.187669229220993e-06,
      "loss": 0.8908,
      "step": 4110
    },
    {
      "epoch": 0.30162238506195144,
      "grad_norm": 1.5957132853203901,
      "learning_rate": 8.186753725760818e-06,
      "loss": 0.8972,
      "step": 4111
    },
    {
      "epoch": 0.3016957546520906,
      "grad_norm": 1.5488274683176477,
      "learning_rate": 8.18583804233677e-06,
      "loss": 0.7889,
      "step": 4112
    },
    {
      "epoch": 0.3016957546520906,
      "lm_loss": 1.6178,
      "step": 4112,
      "vm_loss": 0.0598
    },
    {
      "epoch": 0.3016957546520906,
      "lm_loss": 1.7679,
      "step": 4112,
      "vm_loss": 0.107
    },
    {
      "epoch": 0.3016957546520906,
      "lm_loss": 0.1403,
      "step": 4112,
      "vm_loss": 0.1934
    },
    {
      "epoch": 0.3016957546520906,
      "lm_loss": 0.8744,
      "step": 4112,
      "vm_loss": 0.1968
    },
    {
      "epoch": 0.3016957546520906,
      "lm_loss": 1.2716,
      "step": 4112,
      "vm_loss": 0.0771
    },
    {
      "epoch": 0.3016957546520906,
      "lm_loss": 0.678,
      "step": 4112,
      "vm_loss": 0.121
    },
    {
      "epoch": 0.3016957546520906,
      "lm_loss": 0.8568,
      "step": 4112,
      "vm_loss": 0.1223
    },
    {
      "epoch": 0.3016957546520906,
      "lm_loss": 0.5204,
      "step": 4112,
      "vm_loss": 0.1188
    },
    {
      "epoch": 0.3017691242422297,
      "grad_norm": 1.4850174335312982,
      "learning_rate": 8.184922179000556e-06,
      "loss": 0.8575,
      "step": 4113
    },
    {
      "epoch": 0.30184249383236883,
      "grad_norm": 1.7211231764371715,
      "learning_rate": 8.1840061358039e-06,
      "loss": 0.865,
      "step": 4114
    },
    {
      "epoch": 0.301915863422508,
      "grad_norm": 0.8196562982333305,
      "learning_rate": 8.183089912798531e-06,
      "loss": 0.5342,
      "step": 4115
    },
    {
      "epoch": 0.3019892330126471,
      "grad_norm": 1.5297652935935342,
      "learning_rate": 8.182173510036194e-06,
      "loss": 0.8418,
      "step": 4116
    },
    {
      "epoch": 0.3020626026027862,
      "grad_norm": 1.5326748351734696,
      "learning_rate": 8.181256927568637e-06,
      "loss": 0.7194,
      "step": 4117
    },
    {
      "epoch": 0.3021359721929253,
      "grad_norm": 1.7058129577675312,
      "learning_rate": 8.180340165447625e-06,
      "loss": 0.8768,
      "step": 4118
    },
    {
      "epoch": 0.30220934178306447,
      "grad_norm": 0.6229924895959487,
      "learning_rate": 8.179423223724926e-06,
      "loss": 0.5186,
      "step": 4119
    },
    {
      "epoch": 0.3022827113732036,
      "grad_norm": 0.6503874025193207,
      "learning_rate": 8.178506102452324e-06,
      "loss": 0.5349,
      "step": 4120
    },
    {
      "epoch": 0.3022827113732036,
      "lm_loss": 1.4103,
      "step": 4120,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.3022827113732036,
      "lm_loss": 0.8382,
      "step": 4120,
      "vm_loss": 0.156
    },
    {
      "epoch": 0.3022827113732036,
      "lm_loss": 0.2609,
      "step": 4120,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.3022827113732036,
      "lm_loss": 0.0365,
      "step": 4120,
      "vm_loss": 0.0646
    },
    {
      "epoch": 0.3022827113732036,
      "lm_loss": 0.0281,
      "step": 4120,
      "vm_loss": 0.1686
    },
    {
      "epoch": 0.3022827113732036,
      "lm_loss": 1.3253,
      "step": 4120,
      "vm_loss": 0.1206
    },
    {
      "epoch": 0.3022827113732036,
      "lm_loss": 0.7315,
      "step": 4120,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.3022827113732036,
      "lm_loss": 0.6715,
      "step": 4120,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.3023560809633427,
      "grad_norm": 1.6060343675801063,
      "learning_rate": 8.17758880168161e-06,
      "loss": 0.7829,
      "step": 4121
    },
    {
      "epoch": 0.30242945055348186,
      "grad_norm": 0.5947235305951626,
      "learning_rate": 8.17667132146459e-06,
      "loss": 0.5231,
      "step": 4122
    },
    {
      "epoch": 0.30250282014362095,
      "grad_norm": 1.556281708471848,
      "learning_rate": 8.175753661853072e-06,
      "loss": 0.7783,
      "step": 4123
    },
    {
      "epoch": 0.3025761897337601,
      "grad_norm": 0.6150810468954819,
      "learning_rate": 8.17483582289888e-06,
      "loss": 0.5252,
      "step": 4124
    },
    {
      "epoch": 0.30264955932389925,
      "grad_norm": 1.4779630507337658,
      "learning_rate": 8.173917804653847e-06,
      "loss": 0.796,
      "step": 4125
    },
    {
      "epoch": 0.30272292891403835,
      "grad_norm": 0.6236025312158983,
      "learning_rate": 8.172999607169815e-06,
      "loss": 0.5398,
      "step": 4126
    },
    {
      "epoch": 0.3027962985041775,
      "grad_norm": 1.487829724284257,
      "learning_rate": 8.172081230498638e-06,
      "loss": 0.7749,
      "step": 4127
    },
    {
      "epoch": 0.3028696680943166,
      "grad_norm": 1.4940387953605838,
      "learning_rate": 8.17116267469218e-06,
      "loss": 0.7472,
      "step": 4128
    },
    {
      "epoch": 0.3028696680943166,
      "lm_loss": 0.551,
      "step": 4128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3028696680943166,
      "lm_loss": 0.5201,
      "step": 4128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3028696680943166,
      "lm_loss": 0.4307,
      "step": 4128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3028696680943166,
      "lm_loss": 0.6078,
      "step": 4128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3028696680943166,
      "lm_loss": 0.4655,
      "step": 4128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3028696680943166,
      "lm_loss": 0.4941,
      "step": 4128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3028696680943166,
      "lm_loss": 0.5723,
      "step": 4128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3028696680943166,
      "lm_loss": 0.6851,
      "step": 4128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30294303768445574,
      "grad_norm": 0.5890978318263708,
      "learning_rate": 8.17024393980231e-06,
      "loss": 0.5255,
      "step": 4129
    },
    {
      "epoch": 0.3030164072745949,
      "grad_norm": 1.3980487169152078,
      "learning_rate": 8.169325025880914e-06,
      "loss": 0.7427,
      "step": 4130
    },
    {
      "epoch": 0.303089776864734,
      "grad_norm": 1.3954729616055088,
      "learning_rate": 8.168405932979886e-06,
      "loss": 0.7592,
      "step": 4131
    },
    {
      "epoch": 0.30316314645487313,
      "grad_norm": 1.598468418020435,
      "learning_rate": 8.167486661151128e-06,
      "loss": 0.762,
      "step": 4132
    },
    {
      "epoch": 0.3032365160450122,
      "grad_norm": 1.6075100854233353,
      "learning_rate": 8.166567210446554e-06,
      "loss": 0.8943,
      "step": 4133
    },
    {
      "epoch": 0.3033098856351514,
      "grad_norm": 1.380535228323382,
      "learning_rate": 8.165647580918088e-06,
      "loss": 0.7877,
      "step": 4134
    },
    {
      "epoch": 0.30338325522529047,
      "grad_norm": 1.6143030766692694,
      "learning_rate": 8.164727772617663e-06,
      "loss": 0.866,
      "step": 4135
    },
    {
      "epoch": 0.3034566248154296,
      "grad_norm": 1.4047920436717507,
      "learning_rate": 8.163807785597225e-06,
      "loss": 0.8777,
      "step": 4136
    },
    {
      "epoch": 0.3034566248154296,
      "lm_loss": 1.6397,
      "step": 4136,
      "vm_loss": 0.0253
    },
    {
      "epoch": 0.3034566248154296,
      "lm_loss": 1.019,
      "step": 4136,
      "vm_loss": 0.2083
    },
    {
      "epoch": 0.3034566248154296,
      "lm_loss": 1.0205,
      "step": 4136,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.3034566248154296,
      "lm_loss": 1.0176,
      "step": 4136,
      "vm_loss": 0.2527
    },
    {
      "epoch": 0.3034566248154296,
      "lm_loss": 0.4809,
      "step": 4136,
      "vm_loss": 0.1451
    },
    {
      "epoch": 0.3034566248154296,
      "lm_loss": 1.1746,
      "step": 4136,
      "vm_loss": 0.1378
    },
    {
      "epoch": 0.3034566248154296,
      "lm_loss": 1.1137,
      "step": 4136,
      "vm_loss": 0.2525
    },
    {
      "epoch": 0.3034566248154296,
      "lm_loss": 1.551,
      "step": 4136,
      "vm_loss": 0.0883
    },
    {
      "epoch": 0.30352999440556877,
      "grad_norm": 1.765589160031154,
      "learning_rate": 8.162887619908724e-06,
      "loss": 0.8874,
      "step": 4137
    },
    {
      "epoch": 0.30360336399570786,
      "grad_norm": 1.5426065203293733,
      "learning_rate": 8.161967275604128e-06,
      "loss": 0.9217,
      "step": 4138
    },
    {
      "epoch": 0.303676733585847,
      "grad_norm": 1.486646769021288,
      "learning_rate": 8.161046752735408e-06,
      "loss": 0.8219,
      "step": 4139
    },
    {
      "epoch": 0.3037501031759861,
      "grad_norm": 1.3346389125917315,
      "learning_rate": 8.160126051354552e-06,
      "loss": 0.7629,
      "step": 4140
    },
    {
      "epoch": 0.30382347276612526,
      "grad_norm": 0.8593053928777818,
      "learning_rate": 8.159205171513548e-06,
      "loss": 0.5385,
      "step": 4141
    },
    {
      "epoch": 0.3038968423562644,
      "grad_norm": 1.3506138604397966,
      "learning_rate": 8.158284113264408e-06,
      "loss": 0.7851,
      "step": 4142
    },
    {
      "epoch": 0.3039702119464035,
      "grad_norm": 1.52663211840172,
      "learning_rate": 8.157362876659143e-06,
      "loss": 0.803,
      "step": 4143
    },
    {
      "epoch": 0.30404358153654265,
      "grad_norm": 1.4579230832662136,
      "learning_rate": 8.156441461749775e-06,
      "loss": 0.8788,
      "step": 4144
    },
    {
      "epoch": 0.30404358153654265,
      "lm_loss": 1.658,
      "step": 4144,
      "vm_loss": 0.1332
    },
    {
      "epoch": 0.30404358153654265,
      "lm_loss": 0.1058,
      "step": 4144,
      "vm_loss": 0.1878
    },
    {
      "epoch": 0.30404358153654265,
      "lm_loss": 1.2157,
      "step": 4144,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.30404358153654265,
      "lm_loss": 0.9579,
      "step": 4144,
      "vm_loss": 0.2006
    },
    {
      "epoch": 0.30404358153654265,
      "lm_loss": 0.5923,
      "step": 4144,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.30404358153654265,
      "lm_loss": 1.0158,
      "step": 4144,
      "vm_loss": 0.2022
    },
    {
      "epoch": 0.30404358153654265,
      "lm_loss": 0.9941,
      "step": 4144,
      "vm_loss": 0.1785
    },
    {
      "epoch": 0.30404358153654265,
      "lm_loss": 0.8547,
      "step": 4144,
      "vm_loss": 0.0764
    },
    {
      "epoch": 0.30411695112668174,
      "grad_norm": 1.4776418788086585,
      "learning_rate": 8.155519868588342e-06,
      "loss": 0.8742,
      "step": 4145
    },
    {
      "epoch": 0.3041903207168209,
      "grad_norm": 0.6171715902449837,
      "learning_rate": 8.154598097226886e-06,
      "loss": 0.5177,
      "step": 4146
    },
    {
      "epoch": 0.30426369030696004,
      "grad_norm": 0.64669041469443,
      "learning_rate": 8.153676147717466e-06,
      "loss": 0.5431,
      "step": 4147
    },
    {
      "epoch": 0.30433705989709914,
      "grad_norm": 1.576401835521372,
      "learning_rate": 8.152754020112142e-06,
      "loss": 0.8301,
      "step": 4148
    },
    {
      "epoch": 0.3044104294872383,
      "grad_norm": 1.7044096150911967,
      "learning_rate": 8.15183171446299e-06,
      "loss": 0.8666,
      "step": 4149
    },
    {
      "epoch": 0.3044837990773774,
      "grad_norm": 1.5709479081455182,
      "learning_rate": 8.150909230822097e-06,
      "loss": 0.8684,
      "step": 4150
    },
    {
      "epoch": 0.30455716866751653,
      "grad_norm": 1.548756696541121,
      "learning_rate": 8.149986569241557e-06,
      "loss": 0.8898,
      "step": 4151
    },
    {
      "epoch": 0.3046305382576557,
      "grad_norm": 1.4331070175007412,
      "learning_rate": 8.149063729773474e-06,
      "loss": 0.8062,
      "step": 4152
    },
    {
      "epoch": 0.3046305382576557,
      "lm_loss": 0.6839,
      "step": 4152,
      "vm_loss": 0.0966
    },
    {
      "epoch": 0.3046305382576557,
      "lm_loss": 1.1557,
      "step": 4152,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.3046305382576557,
      "lm_loss": 0.2931,
      "step": 4152,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.3046305382576557,
      "lm_loss": 0.077,
      "step": 4152,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.3046305382576557,
      "lm_loss": 0.8585,
      "step": 4152,
      "vm_loss": 0.1335
    },
    {
      "epoch": 0.3046305382576557,
      "lm_loss": 0.0843,
      "step": 4152,
      "vm_loss": 0.1342
    },
    {
      "epoch": 0.3046305382576557,
      "lm_loss": 1.0813,
      "step": 4152,
      "vm_loss": 0.0678
    },
    {
      "epoch": 0.3046305382576557,
      "lm_loss": 0.0951,
      "step": 4152,
      "vm_loss": 0.0768
    },
    {
      "epoch": 0.3047039078477948,
      "grad_norm": 1.475130430120089,
      "learning_rate": 8.148140712469965e-06,
      "loss": 0.7832,
      "step": 4153
    },
    {
      "epoch": 0.3047772774379339,
      "grad_norm": 1.6131529833877984,
      "learning_rate": 8.147217517383152e-06,
      "loss": 0.9258,
      "step": 4154
    },
    {
      "epoch": 0.304850647028073,
      "grad_norm": 0.7090066521242504,
      "learning_rate": 8.146294144565172e-06,
      "loss": 0.5102,
      "step": 4155
    },
    {
      "epoch": 0.30492401661821217,
      "grad_norm": 1.592966614326014,
      "learning_rate": 8.145370594068172e-06,
      "loss": 0.8955,
      "step": 4156
    },
    {
      "epoch": 0.3049973862083513,
      "grad_norm": 0.6180170221600296,
      "learning_rate": 8.144446865944302e-06,
      "loss": 0.5364,
      "step": 4157
    },
    {
      "epoch": 0.3050707557984904,
      "grad_norm": 1.5389120845221276,
      "learning_rate": 8.143522960245734e-06,
      "loss": 0.7878,
      "step": 4158
    },
    {
      "epoch": 0.30514412538862956,
      "grad_norm": 1.5590120340854434,
      "learning_rate": 8.142598877024637e-06,
      "loss": 0.8372,
      "step": 4159
    },
    {
      "epoch": 0.30521749497876866,
      "grad_norm": 1.6676196738569202,
      "learning_rate": 8.1416746163332e-06,
      "loss": 0.8632,
      "step": 4160
    },
    {
      "epoch": 0.30521749497876866,
      "lm_loss": 0.4223,
      "step": 4160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30521749497876866,
      "lm_loss": 0.4684,
      "step": 4160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30521749497876866,
      "lm_loss": 0.4839,
      "step": 4160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30521749497876866,
      "lm_loss": 0.5381,
      "step": 4160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30521749497876866,
      "lm_loss": 0.6037,
      "step": 4160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30521749497876866,
      "lm_loss": 0.542,
      "step": 4160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30521749497876866,
      "lm_loss": 0.5154,
      "step": 4160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30521749497876866,
      "lm_loss": 0.6409,
      "step": 4160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3052908645689078,
      "grad_norm": 0.6364600476326328,
      "learning_rate": 8.140750178223615e-06,
      "loss": 0.5306,
      "step": 4161
    },
    {
      "epoch": 0.30536423415904695,
      "grad_norm": 1.823231037583671,
      "learning_rate": 8.139825562748092e-06,
      "loss": 0.8368,
      "step": 4162
    },
    {
      "epoch": 0.30543760374918605,
      "grad_norm": 1.5239716181216794,
      "learning_rate": 8.138900769958844e-06,
      "loss": 0.7998,
      "step": 4163
    },
    {
      "epoch": 0.3055109733393252,
      "grad_norm": 0.6434287761979014,
      "learning_rate": 8.137975799908094e-06,
      "loss": 0.5119,
      "step": 4164
    },
    {
      "epoch": 0.3055843429294643,
      "grad_norm": 1.4531591275680977,
      "learning_rate": 8.137050652648082e-06,
      "loss": 0.7819,
      "step": 4165
    },
    {
      "epoch": 0.30565771251960344,
      "grad_norm": 1.5328069344371453,
      "learning_rate": 8.136125328231049e-06,
      "loss": 0.8147,
      "step": 4166
    },
    {
      "epoch": 0.3057310821097426,
      "grad_norm": 1.3514313024543463,
      "learning_rate": 8.135199826709253e-06,
      "loss": 0.7626,
      "step": 4167
    },
    {
      "epoch": 0.3058044516998817,
      "grad_norm": 1.7808054865420815,
      "learning_rate": 8.134274148134958e-06,
      "loss": 0.8727,
      "step": 4168
    },
    {
      "epoch": 0.3058044516998817,
      "lm_loss": 0.7296,
      "step": 4168,
      "vm_loss": 0.0777
    },
    {
      "epoch": 0.3058044516998817,
      "lm_loss": 0.2758,
      "step": 4168,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.3058044516998817,
      "lm_loss": 0.2528,
      "step": 4168,
      "vm_loss": 0.1745
    },
    {
      "epoch": 0.3058044516998817,
      "lm_loss": 0.9878,
      "step": 4168,
      "vm_loss": 0.2064
    },
    {
      "epoch": 0.3058044516998817,
      "lm_loss": 0.8373,
      "step": 4168,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.3058044516998817,
      "lm_loss": 0.4184,
      "step": 4168,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.3058044516998817,
      "lm_loss": 0.0952,
      "step": 4168,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.3058044516998817,
      "lm_loss": 0.2184,
      "step": 4168,
      "vm_loss": 0.1803
    },
    {
      "epoch": 0.30587782129002083,
      "grad_norm": 1.4956605308819462,
      "learning_rate": 8.133348292560442e-06,
      "loss": 0.8433,
      "step": 4169
    },
    {
      "epoch": 0.30595119088015993,
      "grad_norm": 0.6667081412102267,
      "learning_rate": 8.132422260037985e-06,
      "loss": 0.5223,
      "step": 4170
    },
    {
      "epoch": 0.3060245604702991,
      "grad_norm": 1.417703580010309,
      "learning_rate": 8.131496050619887e-06,
      "loss": 0.7555,
      "step": 4171
    },
    {
      "epoch": 0.3060979300604382,
      "grad_norm": 1.6843075138808616,
      "learning_rate": 8.130569664358452e-06,
      "loss": 0.831,
      "step": 4172
    },
    {
      "epoch": 0.3061712996505773,
      "grad_norm": 1.6000237573809672,
      "learning_rate": 8.129643101305998e-06,
      "loss": 0.8363,
      "step": 4173
    },
    {
      "epoch": 0.30624466924071647,
      "grad_norm": 0.5989600132791554,
      "learning_rate": 8.128716361514844e-06,
      "loss": 0.5332,
      "step": 4174
    },
    {
      "epoch": 0.30631803883085557,
      "grad_norm": 1.7623080351542855,
      "learning_rate": 8.127789445037331e-06,
      "loss": 0.8889,
      "step": 4175
    },
    {
      "epoch": 0.3063914084209947,
      "grad_norm": 1.6577598110170546,
      "learning_rate": 8.1268623519258e-06,
      "loss": 0.8846,
      "step": 4176
    },
    {
      "epoch": 0.3063914084209947,
      "lm_loss": 1.143,
      "step": 4176,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.3063914084209947,
      "lm_loss": 1.4528,
      "step": 4176,
      "vm_loss": 0.1814
    },
    {
      "epoch": 0.3063914084209947,
      "lm_loss": 0.3196,
      "step": 4176,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.3063914084209947,
      "lm_loss": 0.8139,
      "step": 4176,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.3063914084209947,
      "lm_loss": 1.1472,
      "step": 4176,
      "vm_loss": 0.1876
    },
    {
      "epoch": 0.3063914084209947,
      "lm_loss": 0.564,
      "step": 4176,
      "vm_loss": 0.1988
    },
    {
      "epoch": 0.3063914084209947,
      "lm_loss": 0.9552,
      "step": 4176,
      "vm_loss": 0.0952
    },
    {
      "epoch": 0.3063914084209947,
      "lm_loss": 0.8436,
      "step": 4176,
      "vm_loss": 0.1309
    },
    {
      "epoch": 0.3064647780111338,
      "grad_norm": 1.724640838220113,
      "learning_rate": 8.12593508223261e-06,
      "loss": 0.8556,
      "step": 4177
    },
    {
      "epoch": 0.30653814760127296,
      "grad_norm": 1.5352138543512042,
      "learning_rate": 8.125007636010125e-06,
      "loss": 0.8444,
      "step": 4178
    },
    {
      "epoch": 0.3066115171914121,
      "grad_norm": 1.5682461425136327,
      "learning_rate": 8.12408001331072e-06,
      "loss": 0.8471,
      "step": 4179
    },
    {
      "epoch": 0.3066848867815512,
      "grad_norm": 0.7344198286694527,
      "learning_rate": 8.12315221418678e-06,
      "loss": 0.5368,
      "step": 4180
    },
    {
      "epoch": 0.30675825637169035,
      "grad_norm": 0.6628359643093481,
      "learning_rate": 8.122224238690699e-06,
      "loss": 0.5284,
      "step": 4181
    },
    {
      "epoch": 0.30683162596182945,
      "grad_norm": 1.6715691444304728,
      "learning_rate": 8.121296086874884e-06,
      "loss": 0.7792,
      "step": 4182
    },
    {
      "epoch": 0.3069049955519686,
      "grad_norm": 1.7950855781488382,
      "learning_rate": 8.12036775879175e-06,
      "loss": 0.8318,
      "step": 4183
    },
    {
      "epoch": 0.30697836514210775,
      "grad_norm": 1.7976295450366617,
      "learning_rate": 8.119439254493718e-06,
      "loss": 0.7862,
      "step": 4184
    },
    {
      "epoch": 0.30697836514210775,
      "lm_loss": 1.257,
      "step": 4184,
      "vm_loss": 0.0617
    },
    {
      "epoch": 0.30697836514210775,
      "lm_loss": 1.1878,
      "step": 4184,
      "vm_loss": 0.0806
    },
    {
      "epoch": 0.30697836514210775,
      "lm_loss": 0.1552,
      "step": 4184,
      "vm_loss": 0.0773
    },
    {
      "epoch": 0.30697836514210775,
      "lm_loss": 0.1082,
      "step": 4184,
      "vm_loss": 0.2091
    },
    {
      "epoch": 0.30697836514210775,
      "lm_loss": 1.1474,
      "step": 4184,
      "vm_loss": 0.1297
    },
    {
      "epoch": 0.30697836514210775,
      "lm_loss": 0.2875,
      "step": 4184,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.30697836514210775,
      "lm_loss": 0.8594,
      "step": 4184,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.30697836514210775,
      "lm_loss": 0.3141,
      "step": 4184,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.30705173473224684,
      "grad_norm": 1.688883280474365,
      "learning_rate": 8.11851057403323e-06,
      "loss": 0.8302,
      "step": 4185
    },
    {
      "epoch": 0.307125104322386,
      "grad_norm": 0.707337253760239,
      "learning_rate": 8.117581717462724e-06,
      "loss": 0.5301,
      "step": 4186
    },
    {
      "epoch": 0.3071984739125251,
      "grad_norm": 0.6996867377984464,
      "learning_rate": 8.116652684834659e-06,
      "loss": 0.5412,
      "step": 4187
    },
    {
      "epoch": 0.30727184350266423,
      "grad_norm": 1.5042341977066895,
      "learning_rate": 8.115723476201498e-06,
      "loss": 0.8219,
      "step": 4188
    },
    {
      "epoch": 0.3073452130928034,
      "grad_norm": 0.6058380965206266,
      "learning_rate": 8.114794091615718e-06,
      "loss": 0.5198,
      "step": 4189
    },
    {
      "epoch": 0.3074185826829425,
      "grad_norm": 1.6274085323752694,
      "learning_rate": 8.1138645311298e-06,
      "loss": 0.8621,
      "step": 4190
    },
    {
      "epoch": 0.3074919522730816,
      "grad_norm": 0.6208668699513173,
      "learning_rate": 8.112934794796244e-06,
      "loss": 0.5194,
      "step": 4191
    },
    {
      "epoch": 0.3075653218632207,
      "grad_norm": 0.6737550554969965,
      "learning_rate": 8.112004882667548e-06,
      "loss": 0.5455,
      "step": 4192
    },
    {
      "epoch": 0.3075653218632207,
      "lm_loss": 0.5376,
      "step": 4192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3075653218632207,
      "lm_loss": 0.5624,
      "step": 4192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3075653218632207,
      "lm_loss": 0.4608,
      "step": 4192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3075653218632207,
      "lm_loss": 0.5776,
      "step": 4192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3075653218632207,
      "lm_loss": 0.5652,
      "step": 4192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3075653218632207,
      "lm_loss": 0.4971,
      "step": 4192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3075653218632207,
      "lm_loss": 0.3114,
      "step": 4192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3075653218632207,
      "lm_loss": 0.6585,
      "step": 4192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.30763869145335987,
      "grad_norm": 0.6190873797784017,
      "learning_rate": 8.11107479479623e-06,
      "loss": 0.5384,
      "step": 4193
    },
    {
      "epoch": 0.307712061043499,
      "grad_norm": 1.6139380931546843,
      "learning_rate": 8.110144531234813e-06,
      "loss": 0.7501,
      "step": 4194
    },
    {
      "epoch": 0.3077854306336381,
      "grad_norm": 1.7447758327306824,
      "learning_rate": 8.109214092035834e-06,
      "loss": 0.9073,
      "step": 4195
    },
    {
      "epoch": 0.30785880022377726,
      "grad_norm": 1.6239910013686933,
      "learning_rate": 8.108283477251836e-06,
      "loss": 0.7668,
      "step": 4196
    },
    {
      "epoch": 0.30793216981391636,
      "grad_norm": 1.9262866322998853,
      "learning_rate": 8.10735268693537e-06,
      "loss": 0.8138,
      "step": 4197
    },
    {
      "epoch": 0.3080055394040555,
      "grad_norm": 1.4772129241630971,
      "learning_rate": 8.106421721139004e-06,
      "loss": 0.8184,
      "step": 4198
    },
    {
      "epoch": 0.30807890899419466,
      "grad_norm": 1.5368826892196044,
      "learning_rate": 8.10549057991531e-06,
      "loss": 0.8317,
      "step": 4199
    },
    {
      "epoch": 0.30815227858433375,
      "grad_norm": 1.540776929969105,
      "learning_rate": 8.104559263316873e-06,
      "loss": 0.7818,
      "step": 4200
    },
    {
      "epoch": 0.30815227858433375,
      "lm_loss": 1.3561,
      "step": 4200,
      "vm_loss": 0.0216
    },
    {
      "epoch": 0.30815227858433375,
      "lm_loss": 0.6155,
      "step": 4200,
      "vm_loss": 0.1311
    },
    {
      "epoch": 0.30815227858433375,
      "lm_loss": 0.8275,
      "step": 4200,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.30815227858433375,
      "lm_loss": 0.6205,
      "step": 4200,
      "vm_loss": 0.1723
    },
    {
      "epoch": 0.30815227858433375,
      "lm_loss": 0.2042,
      "step": 4200,
      "vm_loss": 0.1019
    },
    {
      "epoch": 0.30815227858433375,
      "lm_loss": 0.8056,
      "step": 4200,
      "vm_loss": 0.2177
    },
    {
      "epoch": 0.30815227858433375,
      "lm_loss": 0.6211,
      "step": 4200,
      "vm_loss": 0.1405
    },
    {
      "epoch": 0.30815227858433375,
      "lm_loss": 0.3724,
      "step": 4200,
      "vm_loss": 0.0659
    },
    {
      "epoch": 0.3082256481744729,
      "grad_norm": 1.6739520663484373,
      "learning_rate": 8.103627771396285e-06,
      "loss": 0.7853,
      "step": 4201
    },
    {
      "epoch": 0.308299017764612,
      "grad_norm": 1.6179340909320201,
      "learning_rate": 8.10269610420615e-06,
      "loss": 0.804,
      "step": 4202
    },
    {
      "epoch": 0.30837238735475114,
      "grad_norm": 0.9861512824290484,
      "learning_rate": 8.101764261799087e-06,
      "loss": 0.5384,
      "step": 4203
    },
    {
      "epoch": 0.3084457569448903,
      "grad_norm": 1.4345478949361028,
      "learning_rate": 8.100832244227711e-06,
      "loss": 0.7289,
      "step": 4204
    },
    {
      "epoch": 0.3085191265350294,
      "grad_norm": 1.3214477535840192,
      "learning_rate": 8.099900051544659e-06,
      "loss": 0.6939,
      "step": 4205
    },
    {
      "epoch": 0.30859249612516854,
      "grad_norm": 1.5743986399906762,
      "learning_rate": 8.098967683802574e-06,
      "loss": 0.8655,
      "step": 4206
    },
    {
      "epoch": 0.30866586571530763,
      "grad_norm": 1.5143077990086304,
      "learning_rate": 8.098035141054113e-06,
      "loss": 0.8375,
      "step": 4207
    },
    {
      "epoch": 0.3087392353054468,
      "grad_norm": 0.6300720897091354,
      "learning_rate": 8.097102423351932e-06,
      "loss": 0.509,
      "step": 4208
    },
    {
      "epoch": 0.3087392353054468,
      "lm_loss": 1.5409,
      "step": 4208,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.3087392353054468,
      "lm_loss": 0.0196,
      "step": 4208,
      "vm_loss": 0.0943
    },
    {
      "epoch": 0.3087392353054468,
      "lm_loss": 1.3185,
      "step": 4208,
      "vm_loss": 0.2492
    },
    {
      "epoch": 0.3087392353054468,
      "lm_loss": 0.7934,
      "step": 4208,
      "vm_loss": 0.0602
    },
    {
      "epoch": 0.3087392353054468,
      "lm_loss": 1.281,
      "step": 4208,
      "vm_loss": 0.0648
    },
    {
      "epoch": 0.3087392353054468,
      "lm_loss": 0.9879,
      "step": 4208,
      "vm_loss": 0.1021
    },
    {
      "epoch": 0.3087392353054468,
      "lm_loss": 1.1371,
      "step": 4208,
      "vm_loss": 0.3484
    },
    {
      "epoch": 0.3087392353054468,
      "lm_loss": 0.192,
      "step": 4208,
      "vm_loss": 0.1469
    },
    {
      "epoch": 0.30881260489558593,
      "grad_norm": 2.0017227941979443,
      "learning_rate": 8.096169530748708e-06,
      "loss": 0.7386,
      "step": 4209
    },
    {
      "epoch": 0.308885974485725,
      "grad_norm": 1.6085226421037082,
      "learning_rate": 8.095236463297122e-06,
      "loss": 0.8323,
      "step": 4210
    },
    {
      "epoch": 0.3089593440758642,
      "grad_norm": 0.6084026761551486,
      "learning_rate": 8.09430322104987e-06,
      "loss": 0.5165,
      "step": 4211
    },
    {
      "epoch": 0.30903271366600327,
      "grad_norm": 0.5773297177527619,
      "learning_rate": 8.09336980405965e-06,
      "loss": 0.5177,
      "step": 4212
    },
    {
      "epoch": 0.3091060832561424,
      "grad_norm": 1.5387955767289296,
      "learning_rate": 8.092436212379177e-06,
      "loss": 0.6998,
      "step": 4213
    },
    {
      "epoch": 0.3091794528462815,
      "grad_norm": 1.524548550908573,
      "learning_rate": 8.091502446061175e-06,
      "loss": 0.7075,
      "step": 4214
    },
    {
      "epoch": 0.30925282243642066,
      "grad_norm": 1.7660499441302921,
      "learning_rate": 8.090568505158372e-06,
      "loss": 0.8409,
      "step": 4215
    },
    {
      "epoch": 0.3093261920265598,
      "grad_norm": 1.5464682604069515,
      "learning_rate": 8.08963438972351e-06,
      "loss": 0.8964,
      "step": 4216
    },
    {
      "epoch": 0.3093261920265598,
      "lm_loss": 1.3422,
      "step": 4216,
      "vm_loss": 0.0801
    },
    {
      "epoch": 0.3093261920265598,
      "lm_loss": 1.6101,
      "step": 4216,
      "vm_loss": 0.2777
    },
    {
      "epoch": 0.3093261920265598,
      "lm_loss": 0.9971,
      "step": 4216,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.3093261920265598,
      "lm_loss": 0.1614,
      "step": 4216,
      "vm_loss": 0.1781
    },
    {
      "epoch": 0.3093261920265598,
      "lm_loss": 0.0609,
      "step": 4216,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.3093261920265598,
      "lm_loss": 0.121,
      "step": 4216,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.3093261920265598,
      "lm_loss": 0.0928,
      "step": 4216,
      "vm_loss": 0.1229
    },
    {
      "epoch": 0.3093261920265598,
      "lm_loss": 0.0547,
      "step": 4216,
      "vm_loss": 0.0778
    },
    {
      "epoch": 0.3093995616166989,
      "grad_norm": 1.5617641039190622,
      "learning_rate": 8.088700099809346e-06,
      "loss": 0.8732,
      "step": 4217
    },
    {
      "epoch": 0.30947293120683805,
      "grad_norm": 1.456502423861864,
      "learning_rate": 8.087765635468637e-06,
      "loss": 0.8159,
      "step": 4218
    },
    {
      "epoch": 0.30954630079697715,
      "grad_norm": 1.5959357004862247,
      "learning_rate": 8.086830996754156e-06,
      "loss": 0.8522,
      "step": 4219
    },
    {
      "epoch": 0.3096196703871163,
      "grad_norm": 1.4424130167448683,
      "learning_rate": 8.085896183718684e-06,
      "loss": 0.8552,
      "step": 4220
    },
    {
      "epoch": 0.30969303997725545,
      "grad_norm": 0.7829181218315092,
      "learning_rate": 8.084961196415014e-06,
      "loss": 0.5174,
      "step": 4221
    },
    {
      "epoch": 0.30976640956739454,
      "grad_norm": 1.6965356487794794,
      "learning_rate": 8.084026034895943e-06,
      "loss": 0.9056,
      "step": 4222
    },
    {
      "epoch": 0.3098397791575337,
      "grad_norm": 1.537392705976836,
      "learning_rate": 8.083090699214285e-06,
      "loss": 0.8558,
      "step": 4223
    },
    {
      "epoch": 0.3099131487476728,
      "grad_norm": 1.4708944863841062,
      "learning_rate": 8.08215518942286e-06,
      "loss": 0.7777,
      "step": 4224
    },
    {
      "epoch": 0.3099131487476728,
      "lm_loss": 0.1867,
      "step": 4224,
      "vm_loss": 0.0661
    },
    {
      "epoch": 0.3099131487476728,
      "lm_loss": 1.2157,
      "step": 4224,
      "vm_loss": 0.0711
    },
    {
      "epoch": 0.3099131487476728,
      "lm_loss": 0.3288,
      "step": 4224,
      "vm_loss": 0.1626
    },
    {
      "epoch": 0.3099131487476728,
      "lm_loss": 0.0921,
      "step": 4224,
      "vm_loss": 0.1734
    },
    {
      "epoch": 0.3099131487476728,
      "lm_loss": 0.4054,
      "step": 4224,
      "vm_loss": 0.1304
    },
    {
      "epoch": 0.3099131487476728,
      "lm_loss": 0.6877,
      "step": 4224,
      "vm_loss": 0.1537
    },
    {
      "epoch": 0.3099131487476728,
      "lm_loss": 0.2821,
      "step": 4224,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.3099131487476728,
      "lm_loss": 0.5457,
      "step": 4224,
      "vm_loss": 0.0266
    },
    {
      "epoch": 0.30998651833781193,
      "grad_norm": 1.2565949939028145,
      "learning_rate": 8.0812195055745e-06,
      "loss": 0.8023,
      "step": 4225
    },
    {
      "epoch": 0.3100598879279511,
      "grad_norm": 1.749154909099758,
      "learning_rate": 8.080283647722042e-06,
      "loss": 0.9449,
      "step": 4226
    },
    {
      "epoch": 0.3101332575180902,
      "grad_norm": 1.5845708405493146,
      "learning_rate": 8.07934761591834e-06,
      "loss": 0.8313,
      "step": 4227
    },
    {
      "epoch": 0.3102066271082293,
      "grad_norm": 1.4629095265054652,
      "learning_rate": 8.078411410216251e-06,
      "loss": 0.8111,
      "step": 4228
    },
    {
      "epoch": 0.3102799966983684,
      "grad_norm": 1.5150009396795536,
      "learning_rate": 8.077475030668647e-06,
      "loss": 0.9246,
      "step": 4229
    },
    {
      "epoch": 0.31035336628850757,
      "grad_norm": 1.4564841153744212,
      "learning_rate": 8.076538477328406e-06,
      "loss": 0.8793,
      "step": 4230
    },
    {
      "epoch": 0.3104267358786467,
      "grad_norm": 1.4912750901426168,
      "learning_rate": 8.07560175024842e-06,
      "loss": 0.7471,
      "step": 4231
    },
    {
      "epoch": 0.3105001054687858,
      "grad_norm": 1.4694994898633922,
      "learning_rate": 8.074664849481585e-06,
      "loss": 0.8631,
      "step": 4232
    },
    {
      "epoch": 0.3105001054687858,
      "lm_loss": 1.5057,
      "step": 4232,
      "vm_loss": 0.095
    },
    {
      "epoch": 0.3105001054687858,
      "lm_loss": 1.1396,
      "step": 4232,
      "vm_loss": 0.0569
    },
    {
      "epoch": 0.3105001054687858,
      "lm_loss": 1.002,
      "step": 4232,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.3105001054687858,
      "lm_loss": 0.8176,
      "step": 4232,
      "vm_loss": 0.1004
    },
    {
      "epoch": 0.3105001054687858,
      "lm_loss": 0.2364,
      "step": 4232,
      "vm_loss": 0.0697
    },
    {
      "epoch": 0.3105001054687858,
      "lm_loss": 0.4856,
      "step": 4232,
      "vm_loss": 0.1094
    },
    {
      "epoch": 0.3105001054687858,
      "lm_loss": 0.0918,
      "step": 4232,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.3105001054687858,
      "lm_loss": 0.4912,
      "step": 4232,
      "vm_loss": 0.1784
    },
    {
      "epoch": 0.31057347505892496,
      "grad_norm": 1.5730173474271822,
      "learning_rate": 8.073727775080811e-06,
      "loss": 0.7949,
      "step": 4233
    },
    {
      "epoch": 0.31064684464906406,
      "grad_norm": 1.7245944041071326,
      "learning_rate": 8.072790527099019e-06,
      "loss": 0.8935,
      "step": 4234
    },
    {
      "epoch": 0.3107202142392032,
      "grad_norm": 1.7557088813101318,
      "learning_rate": 8.071853105589136e-06,
      "loss": 0.9124,
      "step": 4235
    },
    {
      "epoch": 0.31079358382934236,
      "grad_norm": 0.7926396388145415,
      "learning_rate": 8.0709155106041e-06,
      "loss": 0.5139,
      "step": 4236
    },
    {
      "epoch": 0.31086695341948145,
      "grad_norm": 1.4506352490411831,
      "learning_rate": 8.069977742196861e-06,
      "loss": 0.8698,
      "step": 4237
    },
    {
      "epoch": 0.3109403230096206,
      "grad_norm": 1.6472601215843778,
      "learning_rate": 8.069039800420377e-06,
      "loss": 0.7709,
      "step": 4238
    },
    {
      "epoch": 0.3110136925997597,
      "grad_norm": 1.6111086314209102,
      "learning_rate": 8.068101685327615e-06,
      "loss": 0.7784,
      "step": 4239
    },
    {
      "epoch": 0.31108706218989884,
      "grad_norm": 1.7737129353408836,
      "learning_rate": 8.06716339697155e-06,
      "loss": 0.8151,
      "step": 4240
    },
    {
      "epoch": 0.31108706218989884,
      "lm_loss": 0.8637,
      "step": 4240,
      "vm_loss": 0.0519
    },
    {
      "epoch": 0.31108706218989884,
      "lm_loss": 0.0644,
      "step": 4240,
      "vm_loss": 0.0929
    },
    {
      "epoch": 0.31108706218989884,
      "lm_loss": 0.1649,
      "step": 4240,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.31108706218989884,
      "lm_loss": 0.4499,
      "step": 4240,
      "vm_loss": 0.1065
    },
    {
      "epoch": 0.31108706218989884,
      "lm_loss": 0.3024,
      "step": 4240,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.31108706218989884,
      "lm_loss": 0.6959,
      "step": 4240,
      "vm_loss": 0.0848
    },
    {
      "epoch": 0.31108706218989884,
      "lm_loss": 0.7125,
      "step": 4240,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.31108706218989884,
      "lm_loss": 1.1677,
      "step": 4240,
      "vm_loss": 0.1379
    },
    {
      "epoch": 0.311160431780038,
      "grad_norm": 1.428964232874538,
      "learning_rate": 8.066224935405173e-06,
      "loss": 0.7652,
      "step": 4241
    },
    {
      "epoch": 0.3112338013701771,
      "grad_norm": 1.5007804711007984,
      "learning_rate": 8.065286300681482e-06,
      "loss": 0.8495,
      "step": 4242
    },
    {
      "epoch": 0.31130717096031624,
      "grad_norm": 0.5962492140527775,
      "learning_rate": 8.064347492853483e-06,
      "loss": 0.5237,
      "step": 4243
    },
    {
      "epoch": 0.31138054055045533,
      "grad_norm": 1.7161747088090822,
      "learning_rate": 8.063408511974191e-06,
      "loss": 0.8473,
      "step": 4244
    },
    {
      "epoch": 0.3114539101405945,
      "grad_norm": 1.8682342075059386,
      "learning_rate": 8.062469358096636e-06,
      "loss": 0.8972,
      "step": 4245
    },
    {
      "epoch": 0.31152727973073363,
      "grad_norm": 1.7033631559376636,
      "learning_rate": 8.06153003127385e-06,
      "loss": 0.8356,
      "step": 4246
    },
    {
      "epoch": 0.3116006493208727,
      "grad_norm": 0.5560734896980066,
      "learning_rate": 8.060590531558883e-06,
      "loss": 0.5023,
      "step": 4247
    },
    {
      "epoch": 0.3116740189110119,
      "grad_norm": 1.560492160333022,
      "learning_rate": 8.059650859004789e-06,
      "loss": 0.8483,
      "step": 4248
    },
    {
      "epoch": 0.3116740189110119,
      "lm_loss": 0.6394,
      "step": 4248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3116740189110119,
      "lm_loss": 0.4014,
      "step": 4248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3116740189110119,
      "lm_loss": 0.6062,
      "step": 4248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3116740189110119,
      "lm_loss": 0.4354,
      "step": 4248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3116740189110119,
      "lm_loss": 0.3638,
      "step": 4248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3116740189110119,
      "lm_loss": 0.4422,
      "step": 4248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3116740189110119,
      "lm_loss": 0.5094,
      "step": 4248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3116740189110119,
      "lm_loss": 0.5769,
      "step": 4248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31174738850115097,
      "grad_norm": 0.5928098699897565,
      "learning_rate": 8.058711013664633e-06,
      "loss": 0.5417,
      "step": 4249
    },
    {
      "epoch": 0.3118207580912901,
      "grad_norm": 0.5570535280508032,
      "learning_rate": 8.05777099559149e-06,
      "loss": 0.5138,
      "step": 4250
    },
    {
      "epoch": 0.31189412768142927,
      "grad_norm": 0.574032597862052,
      "learning_rate": 8.056830804838449e-06,
      "loss": 0.5341,
      "step": 4251
    },
    {
      "epoch": 0.31196749727156836,
      "grad_norm": 1.5273688808404167,
      "learning_rate": 8.0558904414586e-06,
      "loss": 0.8562,
      "step": 4252
    },
    {
      "epoch": 0.3120408668617075,
      "grad_norm": 0.5692624457364137,
      "learning_rate": 8.054949905505052e-06,
      "loss": 0.5075,
      "step": 4253
    },
    {
      "epoch": 0.3121142364518466,
      "grad_norm": 0.5752583307531994,
      "learning_rate": 8.054009197030916e-06,
      "loss": 0.5354,
      "step": 4254
    },
    {
      "epoch": 0.31218760604198575,
      "grad_norm": 1.651454668446928,
      "learning_rate": 8.053068316089318e-06,
      "loss": 0.8835,
      "step": 4255
    },
    {
      "epoch": 0.31226097563212485,
      "grad_norm": 0.596829864432093,
      "learning_rate": 8.052127262733394e-06,
      "loss": 0.5279,
      "step": 4256
    },
    {
      "epoch": 0.31226097563212485,
      "lm_loss": 1.5556,
      "step": 4256,
      "vm_loss": 0.0476
    },
    {
      "epoch": 0.31226097563212485,
      "lm_loss": 0.9643,
      "step": 4256,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.31226097563212485,
      "lm_loss": 0.0891,
      "step": 4256,
      "vm_loss": 0.1498
    },
    {
      "epoch": 0.31226097563212485,
      "lm_loss": 0.7714,
      "step": 4256,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.31226097563212485,
      "lm_loss": 0.2906,
      "step": 4256,
      "vm_loss": 0.172
    },
    {
      "epoch": 0.31226097563212485,
      "lm_loss": 1.0975,
      "step": 4256,
      "vm_loss": 0.1177
    },
    {
      "epoch": 0.31226097563212485,
      "lm_loss": 0.9025,
      "step": 4256,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.31226097563212485,
      "lm_loss": 0.827,
      "step": 4256,
      "vm_loss": 0.1086
    },
    {
      "epoch": 0.312334345222264,
      "grad_norm": 1.3843124423681634,
      "learning_rate": 8.051186037016282e-06,
      "loss": 0.789,
      "step": 4257
    },
    {
      "epoch": 0.31240771481240315,
      "grad_norm": 1.6989740527978567,
      "learning_rate": 8.050244638991139e-06,
      "loss": 0.9273,
      "step": 4258
    },
    {
      "epoch": 0.31248108440254224,
      "grad_norm": 0.5770493531183012,
      "learning_rate": 8.049303068711127e-06,
      "loss": 0.5153,
      "step": 4259
    },
    {
      "epoch": 0.3125544539926814,
      "grad_norm": 1.412094320016173,
      "learning_rate": 8.04836132622942e-06,
      "loss": 0.838,
      "step": 4260
    },
    {
      "epoch": 0.3126278235828205,
      "grad_norm": 1.576407258483308,
      "learning_rate": 8.0474194115992e-06,
      "loss": 0.7575,
      "step": 4261
    },
    {
      "epoch": 0.31270119317295964,
      "grad_norm": 1.4625485669971627,
      "learning_rate": 8.046477324873658e-06,
      "loss": 0.7904,
      "step": 4262
    },
    {
      "epoch": 0.3127745627630988,
      "grad_norm": 1.5077551748472955,
      "learning_rate": 8.045535066105997e-06,
      "loss": 0.804,
      "step": 4263
    },
    {
      "epoch": 0.3128479323532379,
      "grad_norm": 1.6804435108720672,
      "learning_rate": 8.044592635349428e-06,
      "loss": 0.7451,
      "step": 4264
    },
    {
      "epoch": 0.3128479323532379,
      "lm_loss": 0.2042,
      "step": 4264,
      "vm_loss": 0.0485
    },
    {
      "epoch": 0.3128479323532379,
      "lm_loss": 0.075,
      "step": 4264,
      "vm_loss": 0.1455
    },
    {
      "epoch": 0.3128479323532379,
      "lm_loss": 0.0638,
      "step": 4264,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.3128479323532379,
      "lm_loss": 0.8617,
      "step": 4264,
      "vm_loss": 0.035
    },
    {
      "epoch": 0.3128479323532379,
      "lm_loss": 0.4617,
      "step": 4264,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.3128479323532379,
      "lm_loss": 0.64,
      "step": 4264,
      "vm_loss": 0.1823
    },
    {
      "epoch": 0.3128479323532379,
      "lm_loss": 0.2238,
      "step": 4264,
      "vm_loss": 0.1069
    },
    {
      "epoch": 0.3128479323532379,
      "lm_loss": 0.2769,
      "step": 4264,
      "vm_loss": 0.1187
    },
    {
      "epoch": 0.31292130194337703,
      "grad_norm": 1.272857717731752,
      "learning_rate": 8.043650032657176e-06,
      "loss": 0.7527,
      "step": 4265
    },
    {
      "epoch": 0.3129946715335161,
      "grad_norm": 0.6143431107293335,
      "learning_rate": 8.042707258082468e-06,
      "loss": 0.5331,
      "step": 4266
    },
    {
      "epoch": 0.31306804112365527,
      "grad_norm": 0.5775606146076787,
      "learning_rate": 8.041764311678544e-06,
      "loss": 0.5278,
      "step": 4267
    },
    {
      "epoch": 0.3131414107137944,
      "grad_norm": 1.5439696145654322,
      "learning_rate": 8.040821193498658e-06,
      "loss": 0.87,
      "step": 4268
    },
    {
      "epoch": 0.3132147803039335,
      "grad_norm": 1.5592958878601977,
      "learning_rate": 8.039877903596069e-06,
      "loss": 0.7563,
      "step": 4269
    },
    {
      "epoch": 0.31328814989407267,
      "grad_norm": 1.3802928191999777,
      "learning_rate": 8.038934442024047e-06,
      "loss": 0.842,
      "step": 4270
    },
    {
      "epoch": 0.31336151948421176,
      "grad_norm": 1.6026042180493663,
      "learning_rate": 8.037990808835869e-06,
      "loss": 0.8945,
      "step": 4271
    },
    {
      "epoch": 0.3134348890743509,
      "grad_norm": 0.5882365888646075,
      "learning_rate": 8.037047004084828e-06,
      "loss": 0.5362,
      "step": 4272
    },
    {
      "epoch": 0.3134348890743509,
      "lm_loss": 1.6395,
      "step": 4272,
      "vm_loss": 0.1021
    },
    {
      "epoch": 0.3134348890743509,
      "lm_loss": 1.1829,
      "step": 4272,
      "vm_loss": 0.1634
    },
    {
      "epoch": 0.3134348890743509,
      "lm_loss": 1.1753,
      "step": 4272,
      "vm_loss": 0.2219
    },
    {
      "epoch": 0.3134348890743509,
      "lm_loss": 0.7874,
      "step": 4272,
      "vm_loss": 0.071
    },
    {
      "epoch": 0.3134348890743509,
      "lm_loss": 0.4102,
      "step": 4272,
      "vm_loss": 0.1657
    },
    {
      "epoch": 0.3134348890743509,
      "lm_loss": 0.5432,
      "step": 4272,
      "vm_loss": 0.0479
    },
    {
      "epoch": 0.3134348890743509,
      "lm_loss": 0.2154,
      "step": 4272,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.3134348890743509,
      "lm_loss": 0.354,
      "step": 4272,
      "vm_loss": 0.103
    },
    {
      "epoch": 0.31350825866449006,
      "grad_norm": 1.7005967881049038,
      "learning_rate": 8.036103027824223e-06,
      "loss": 0.8267,
      "step": 4273
    },
    {
      "epoch": 0.31358162825462915,
      "grad_norm": 1.7465712343109019,
      "learning_rate": 8.035158880107359e-06,
      "loss": 0.9639,
      "step": 4274
    },
    {
      "epoch": 0.3136549978447683,
      "grad_norm": 1.4233392160663636,
      "learning_rate": 8.034214560987558e-06,
      "loss": 0.7981,
      "step": 4275
    },
    {
      "epoch": 0.3137283674349074,
      "grad_norm": 0.5928151527356258,
      "learning_rate": 8.033270070518147e-06,
      "loss": 0.5535,
      "step": 4276
    },
    {
      "epoch": 0.31380173702504655,
      "grad_norm": 1.559674051423726,
      "learning_rate": 8.032325408752462e-06,
      "loss": 0.8098,
      "step": 4277
    },
    {
      "epoch": 0.3138751066151857,
      "grad_norm": 0.5803970024514682,
      "learning_rate": 8.031380575743852e-06,
      "loss": 0.5555,
      "step": 4278
    },
    {
      "epoch": 0.3139484762053248,
      "grad_norm": 1.4685028024216429,
      "learning_rate": 8.030435571545675e-06,
      "loss": 0.7891,
      "step": 4279
    },
    {
      "epoch": 0.31402184579546394,
      "grad_norm": 1.4921383599627884,
      "learning_rate": 8.029490396211295e-06,
      "loss": 0.858,
      "step": 4280
    },
    {
      "epoch": 0.31402184579546394,
      "lm_loss": 1.4487,
      "step": 4280,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.31402184579546394,
      "lm_loss": 0.64,
      "step": 4280,
      "vm_loss": 0.1757
    },
    {
      "epoch": 0.31402184579546394,
      "lm_loss": 0.7302,
      "step": 4280,
      "vm_loss": 0.179
    },
    {
      "epoch": 0.31402184579546394,
      "lm_loss": 0.3068,
      "step": 4280,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.31402184579546394,
      "lm_loss": 0.3074,
      "step": 4280,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.31402184579546394,
      "lm_loss": 0.4168,
      "step": 4280,
      "vm_loss": 0.0927
    },
    {
      "epoch": 0.31402184579546394,
      "lm_loss": 0.8538,
      "step": 4280,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.31402184579546394,
      "lm_loss": 0.6704,
      "step": 4280,
      "vm_loss": 0.1573
    },
    {
      "epoch": 0.31409521538560303,
      "grad_norm": 1.6346642176421362,
      "learning_rate": 8.028545049794091e-06,
      "loss": 0.8392,
      "step": 4281
    },
    {
      "epoch": 0.3141685849757422,
      "grad_norm": 1.5497646028285463,
      "learning_rate": 8.027599532347447e-06,
      "loss": 0.9327,
      "step": 4282
    },
    {
      "epoch": 0.31424195456588133,
      "grad_norm": 1.467091147048844,
      "learning_rate": 8.02665384392476e-06,
      "loss": 0.7436,
      "step": 4283
    },
    {
      "epoch": 0.3143153241560204,
      "grad_norm": 1.680739791829985,
      "learning_rate": 8.025707984579437e-06,
      "loss": 0.8881,
      "step": 4284
    },
    {
      "epoch": 0.3143886937461596,
      "grad_norm": 0.6283091650535926,
      "learning_rate": 8.024761954364888e-06,
      "loss": 0.5259,
      "step": 4285
    },
    {
      "epoch": 0.31446206333629867,
      "grad_norm": 0.592898173596429,
      "learning_rate": 8.023815753334542e-06,
      "loss": 0.5192,
      "step": 4286
    },
    {
      "epoch": 0.3145354329264378,
      "grad_norm": 1.6044449333534363,
      "learning_rate": 8.022869381541831e-06,
      "loss": 0.7485,
      "step": 4287
    },
    {
      "epoch": 0.31460880251657697,
      "grad_norm": 0.5664800715401466,
      "learning_rate": 8.0219228390402e-06,
      "loss": 0.5379,
      "step": 4288
    },
    {
      "epoch": 0.31460880251657697,
      "lm_loss": 0.2747,
      "step": 4288,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.31460880251657697,
      "lm_loss": 1.1308,
      "step": 4288,
      "vm_loss": 0.0311
    },
    {
      "epoch": 0.31460880251657697,
      "lm_loss": 1.3221,
      "step": 4288,
      "vm_loss": 0.193
    },
    {
      "epoch": 0.31460880251657697,
      "lm_loss": 0.9775,
      "step": 4288,
      "vm_loss": 0.1143
    },
    {
      "epoch": 0.31460880251657697,
      "lm_loss": 0.8201,
      "step": 4288,
      "vm_loss": 0.217
    },
    {
      "epoch": 0.31460880251657697,
      "lm_loss": 0.0321,
      "step": 4288,
      "vm_loss": 0.1019
    },
    {
      "epoch": 0.31460880251657697,
      "lm_loss": 0.2732,
      "step": 4288,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.31460880251657697,
      "lm_loss": 0.7319,
      "step": 4288,
      "vm_loss": 0.0588
    },
    {
      "epoch": 0.31468217210671606,
      "grad_norm": 1.3818745526057168,
      "learning_rate": 8.020976125883105e-06,
      "loss": 0.8744,
      "step": 4289
    },
    {
      "epoch": 0.3147555416968552,
      "grad_norm": 0.6103618158906177,
      "learning_rate": 8.020029242124006e-06,
      "loss": 0.5287,
      "step": 4290
    },
    {
      "epoch": 0.3148289112869943,
      "grad_norm": 0.5728214551221866,
      "learning_rate": 8.019082187816376e-06,
      "loss": 0.5131,
      "step": 4291
    },
    {
      "epoch": 0.31490228087713346,
      "grad_norm": 0.573378527335336,
      "learning_rate": 8.0181349630137e-06,
      "loss": 0.5371,
      "step": 4292
    },
    {
      "epoch": 0.31497565046727255,
      "grad_norm": 1.7054389957882106,
      "learning_rate": 8.017187567769467e-06,
      "loss": 0.792,
      "step": 4293
    },
    {
      "epoch": 0.3150490200574117,
      "grad_norm": 1.6619209649333504,
      "learning_rate": 8.016240002137178e-06,
      "loss": 0.7995,
      "step": 4294
    },
    {
      "epoch": 0.31512238964755085,
      "grad_norm": 1.6243177167603804,
      "learning_rate": 8.015292266170348e-06,
      "loss": 0.8445,
      "step": 4295
    },
    {
      "epoch": 0.31519575923768994,
      "grad_norm": 0.6396039414820679,
      "learning_rate": 8.0143443599225e-06,
      "loss": 0.5399,
      "step": 4296
    },
    {
      "epoch": 0.31519575923768994,
      "lm_loss": 0.6258,
      "step": 4296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31519575923768994,
      "lm_loss": 0.5764,
      "step": 4296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31519575923768994,
      "lm_loss": 0.4974,
      "step": 4296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31519575923768994,
      "lm_loss": 0.5681,
      "step": 4296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31519575923768994,
      "lm_loss": 0.5645,
      "step": 4296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31519575923768994,
      "lm_loss": 0.5757,
      "step": 4296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31519575923768994,
      "lm_loss": 0.3924,
      "step": 4296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31519575923768994,
      "lm_loss": 0.6189,
      "step": 4296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3152691288278291,
      "grad_norm": 0.6173676550990789,
      "learning_rate": 8.013396283447158e-06,
      "loss": 0.5372,
      "step": 4297
    },
    {
      "epoch": 0.3153424984179682,
      "grad_norm": 2.072284258906257,
      "learning_rate": 8.012448036797863e-06,
      "loss": 0.8071,
      "step": 4298
    },
    {
      "epoch": 0.31541586800810734,
      "grad_norm": 0.5437815305631976,
      "learning_rate": 8.01149962002817e-06,
      "loss": 0.5253,
      "step": 4299
    },
    {
      "epoch": 0.3154892375982465,
      "grad_norm": 1.305905832637326,
      "learning_rate": 8.010551033191635e-06,
      "loss": 0.718,
      "step": 4300
    },
    {
      "epoch": 0.3155626071883856,
      "grad_norm": 1.4002618006203136,
      "learning_rate": 8.009602276341828e-06,
      "loss": 0.8294,
      "step": 4301
    },
    {
      "epoch": 0.31563597677852473,
      "grad_norm": 1.6428568961383667,
      "learning_rate": 8.008653349532326e-06,
      "loss": 0.8493,
      "step": 4302
    },
    {
      "epoch": 0.3157093463686638,
      "grad_norm": 1.479308833341518,
      "learning_rate": 8.007704252816721e-06,
      "loss": 0.7686,
      "step": 4303
    },
    {
      "epoch": 0.315782715958803,
      "grad_norm": 0.6936965139801453,
      "learning_rate": 8.006754986248606e-06,
      "loss": 0.5067,
      "step": 4304
    },
    {
      "epoch": 0.315782715958803,
      "lm_loss": 2.2415,
      "step": 4304,
      "vm_loss": 0.1286
    },
    {
      "epoch": 0.315782715958803,
      "lm_loss": 1.0491,
      "step": 4304,
      "vm_loss": 0.153
    },
    {
      "epoch": 0.315782715958803,
      "lm_loss": 0.8132,
      "step": 4304,
      "vm_loss": 0.1002
    },
    {
      "epoch": 0.315782715958803,
      "lm_loss": 0.5386,
      "step": 4304,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.315782715958803,
      "lm_loss": 0.7277,
      "step": 4304,
      "vm_loss": 0.17
    },
    {
      "epoch": 0.315782715958803,
      "lm_loss": 0.6293,
      "step": 4304,
      "vm_loss": 0.1429
    },
    {
      "epoch": 0.315782715958803,
      "lm_loss": 0.6651,
      "step": 4304,
      "vm_loss": 0.0641
    },
    {
      "epoch": 0.315782715958803,
      "lm_loss": 0.8,
      "step": 4304,
      "vm_loss": 0.1268
    },
    {
      "epoch": 0.3158560855489421,
      "grad_norm": 1.5663509798498738,
      "learning_rate": 8.005805549881593e-06,
      "loss": 0.8689,
      "step": 4305
    },
    {
      "epoch": 0.3159294551390812,
      "grad_norm": 1.5440054225611768,
      "learning_rate": 8.004855943769296e-06,
      "loss": 0.8508,
      "step": 4306
    },
    {
      "epoch": 0.31600282472922037,
      "grad_norm": 1.6733793830768513,
      "learning_rate": 8.00390616796534e-06,
      "loss": 0.7929,
      "step": 4307
    },
    {
      "epoch": 0.31607619431935946,
      "grad_norm": 1.6767643904618241,
      "learning_rate": 8.002956222523367e-06,
      "loss": 0.8195,
      "step": 4308
    },
    {
      "epoch": 0.3161495639094986,
      "grad_norm": 1.5106333745841314,
      "learning_rate": 8.002006107497018e-06,
      "loss": 0.819,
      "step": 4309
    },
    {
      "epoch": 0.31622293349963776,
      "grad_norm": 1.5508790693610945,
      "learning_rate": 8.001055822939951e-06,
      "loss": 0.7694,
      "step": 4310
    },
    {
      "epoch": 0.31629630308977685,
      "grad_norm": 0.5895881640456598,
      "learning_rate": 8.000105368905829e-06,
      "loss": 0.5196,
      "step": 4311
    },
    {
      "epoch": 0.316369672679916,
      "grad_norm": 1.6822376831466286,
      "learning_rate": 7.999154745448325e-06,
      "loss": 0.8524,
      "step": 4312
    },
    {
      "epoch": 0.316369672679916,
      "lm_loss": 1.1822,
      "step": 4312,
      "vm_loss": 0.0511
    },
    {
      "epoch": 0.316369672679916,
      "lm_loss": 0.529,
      "step": 4312,
      "vm_loss": 0.0936
    },
    {
      "epoch": 0.316369672679916,
      "lm_loss": 0.3164,
      "step": 4312,
      "vm_loss": 0.1342
    },
    {
      "epoch": 0.316369672679916,
      "lm_loss": 0.32,
      "step": 4312,
      "vm_loss": 0.16
    },
    {
      "epoch": 0.316369672679916,
      "lm_loss": 0.6213,
      "step": 4312,
      "vm_loss": 0.145
    },
    {
      "epoch": 0.316369672679916,
      "lm_loss": 0.0425,
      "step": 4312,
      "vm_loss": 0.0753
    },
    {
      "epoch": 0.316369672679916,
      "lm_loss": 0.7696,
      "step": 4312,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.316369672679916,
      "lm_loss": 0.3655,
      "step": 4312,
      "vm_loss": 0.1799
    },
    {
      "epoch": 0.3164430422700551,
      "grad_norm": 1.7998372630170696,
      "learning_rate": 7.998203952621129e-06,
      "loss": 0.7876,
      "step": 4313
    },
    {
      "epoch": 0.31651641186019425,
      "grad_norm": 1.5327170547480267,
      "learning_rate": 7.997252990477929e-06,
      "loss": 0.796,
      "step": 4314
    },
    {
      "epoch": 0.3165897814503334,
      "grad_norm": 1.5649108961453366,
      "learning_rate": 7.996301859072429e-06,
      "loss": 0.7133,
      "step": 4315
    },
    {
      "epoch": 0.3166631510404725,
      "grad_norm": 1.5568076723899835,
      "learning_rate": 7.995350558458344e-06,
      "loss": 0.9371,
      "step": 4316
    },
    {
      "epoch": 0.31673652063061164,
      "grad_norm": 1.5189662368330106,
      "learning_rate": 7.994399088689396e-06,
      "loss": 0.8055,
      "step": 4317
    },
    {
      "epoch": 0.31680989022075073,
      "grad_norm": 1.4285586545331768,
      "learning_rate": 7.993447449819314e-06,
      "loss": 0.6773,
      "step": 4318
    },
    {
      "epoch": 0.3168832598108899,
      "grad_norm": 1.4500368658589926,
      "learning_rate": 7.992495641901842e-06,
      "loss": 0.8699,
      "step": 4319
    },
    {
      "epoch": 0.31695662940102903,
      "grad_norm": 1.586917959580657,
      "learning_rate": 7.991543664990731e-06,
      "loss": 0.8542,
      "step": 4320
    },
    {
      "epoch": 0.31695662940102903,
      "lm_loss": 0.6699,
      "step": 4320,
      "vm_loss": 0.0645
    },
    {
      "epoch": 0.31695662940102903,
      "lm_loss": 0.3452,
      "step": 4320,
      "vm_loss": 0.1085
    },
    {
      "epoch": 0.31695662940102903,
      "lm_loss": 1.5127,
      "step": 4320,
      "vm_loss": 0.1523
    },
    {
      "epoch": 0.31695662940102903,
      "lm_loss": 0.9048,
      "step": 4320,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.31695662940102903,
      "lm_loss": 0.767,
      "step": 4320,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.31695662940102903,
      "lm_loss": 0.889,
      "step": 4320,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.31695662940102903,
      "lm_loss": 1.039,
      "step": 4320,
      "vm_loss": 0.209
    },
    {
      "epoch": 0.31695662940102903,
      "lm_loss": 0.8003,
      "step": 4320,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.3170299989911681,
      "grad_norm": 1.4629548119503788,
      "learning_rate": 7.99059151913974e-06,
      "loss": 0.8561,
      "step": 4321
    },
    {
      "epoch": 0.3171033685813073,
      "grad_norm": 1.6375212796497172,
      "learning_rate": 7.989639204402639e-06,
      "loss": 0.8287,
      "step": 4322
    },
    {
      "epoch": 0.31717673817144637,
      "grad_norm": 1.871499728689464,
      "learning_rate": 7.988686720833209e-06,
      "loss": 0.8509,
      "step": 4323
    },
    {
      "epoch": 0.3172501077615855,
      "grad_norm": 0.7286834458251611,
      "learning_rate": 7.98773406848524e-06,
      "loss": 0.5227,
      "step": 4324
    },
    {
      "epoch": 0.31732347735172467,
      "grad_norm": 1.4575298309100766,
      "learning_rate": 7.98678124741253e-06,
      "loss": 0.7855,
      "step": 4325
    },
    {
      "epoch": 0.31739684694186376,
      "grad_norm": 1.8352214027927545,
      "learning_rate": 7.985828257668883e-06,
      "loss": 0.8385,
      "step": 4326
    },
    {
      "epoch": 0.3174702165320029,
      "grad_norm": 1.281637662621381,
      "learning_rate": 7.984875099308122e-06,
      "loss": 0.7662,
      "step": 4327
    },
    {
      "epoch": 0.317543586122142,
      "grad_norm": 1.6954383417794208,
      "learning_rate": 7.983921772384072e-06,
      "loss": 0.8094,
      "step": 4328
    },
    {
      "epoch": 0.317543586122142,
      "lm_loss": 1.4464,
      "step": 4328,
      "vm_loss": 0.0376
    },
    {
      "epoch": 0.317543586122142,
      "lm_loss": 1.1398,
      "step": 4328,
      "vm_loss": 0.1068
    },
    {
      "epoch": 0.317543586122142,
      "lm_loss": 0.4838,
      "step": 4328,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.317543586122142,
      "lm_loss": 0.0271,
      "step": 4328,
      "vm_loss": 0.1616
    },
    {
      "epoch": 0.317543586122142,
      "lm_loss": 1.3553,
      "step": 4328,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.317543586122142,
      "lm_loss": 0.3168,
      "step": 4328,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.317543586122142,
      "lm_loss": 0.3269,
      "step": 4328,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.317543586122142,
      "lm_loss": 0.8423,
      "step": 4328,
      "vm_loss": 0.2455
    },
    {
      "epoch": 0.31761695571228116,
      "grad_norm": 1.5767538751309382,
      "learning_rate": 7.982968276950568e-06,
      "loss": 0.838,
      "step": 4329
    },
    {
      "epoch": 0.3176903253024203,
      "grad_norm": 0.6312345098481331,
      "learning_rate": 7.982014613061462e-06,
      "loss": 0.5247,
      "step": 4330
    },
    {
      "epoch": 0.3177636948925594,
      "grad_norm": 1.5480512504694846,
      "learning_rate": 7.981060780770603e-06,
      "loss": 0.8817,
      "step": 4331
    },
    {
      "epoch": 0.31783706448269855,
      "grad_norm": 1.5890160219885192,
      "learning_rate": 7.98010678013186e-06,
      "loss": 0.8196,
      "step": 4332
    },
    {
      "epoch": 0.31791043407283764,
      "grad_norm": 1.8262994552069136,
      "learning_rate": 7.979152611199109e-06,
      "loss": 0.9136,
      "step": 4333
    },
    {
      "epoch": 0.3179838036629768,
      "grad_norm": 1.8696859498063956,
      "learning_rate": 7.97819827402623e-06,
      "loss": 0.8349,
      "step": 4334
    },
    {
      "epoch": 0.3180571732531159,
      "grad_norm": 1.4066967752716717,
      "learning_rate": 7.977243768667119e-06,
      "loss": 0.8136,
      "step": 4335
    },
    {
      "epoch": 0.31813054284325504,
      "grad_norm": 0.5972336462621137,
      "learning_rate": 7.97628909517568e-06,
      "loss": 0.5245,
      "step": 4336
    },
    {
      "epoch": 0.31813054284325504,
      "lm_loss": 1.3281,
      "step": 4336,
      "vm_loss": 0.0612
    },
    {
      "epoch": 0.31813054284325504,
      "lm_loss": 0.2842,
      "step": 4336,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.31813054284325504,
      "lm_loss": 0.9103,
      "step": 4336,
      "vm_loss": 0.1281
    },
    {
      "epoch": 0.31813054284325504,
      "lm_loss": 0.1948,
      "step": 4336,
      "vm_loss": 0.1536
    },
    {
      "epoch": 0.31813054284325504,
      "lm_loss": 0.372,
      "step": 4336,
      "vm_loss": 0.123
    },
    {
      "epoch": 0.31813054284325504,
      "lm_loss": 0.6504,
      "step": 4336,
      "vm_loss": 0.0895
    },
    {
      "epoch": 0.31813054284325504,
      "lm_loss": 0.0883,
      "step": 4336,
      "vm_loss": 0.0431
    },
    {
      "epoch": 0.31813054284325504,
      "lm_loss": 0.3138,
      "step": 4336,
      "vm_loss": 0.1045
    },
    {
      "epoch": 0.3182039124333942,
      "grad_norm": 1.4902792319241187,
      "learning_rate": 7.975334253605824e-06,
      "loss": 0.7718,
      "step": 4337
    },
    {
      "epoch": 0.3182772820235333,
      "grad_norm": 1.4579975354912573,
      "learning_rate": 7.974379244011476e-06,
      "loss": 0.913,
      "step": 4338
    },
    {
      "epoch": 0.31835065161367243,
      "grad_norm": 1.5851217553933452,
      "learning_rate": 7.973424066446566e-06,
      "loss": 0.9298,
      "step": 4339
    },
    {
      "epoch": 0.3184240212038115,
      "grad_norm": 1.5314884305987073,
      "learning_rate": 7.972468720965035e-06,
      "loss": 0.8917,
      "step": 4340
    },
    {
      "epoch": 0.3184973907939507,
      "grad_norm": 1.6469495690824032,
      "learning_rate": 7.971513207620835e-06,
      "loss": 0.7576,
      "step": 4341
    },
    {
      "epoch": 0.3185707603840898,
      "grad_norm": 0.618873546424384,
      "learning_rate": 7.970557526467923e-06,
      "loss": 0.4996,
      "step": 4342
    },
    {
      "epoch": 0.3186441299742289,
      "grad_norm": 1.447199867938129,
      "learning_rate": 7.969601677560272e-06,
      "loss": 0.8629,
      "step": 4343
    },
    {
      "epoch": 0.31871749956436807,
      "grad_norm": 3.8092101538331296,
      "learning_rate": 7.968645660951862e-06,
      "loss": 0.7613,
      "step": 4344
    },
    {
      "epoch": 0.31871749956436807,
      "lm_loss": 0.2198,
      "step": 4344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31871749956436807,
      "lm_loss": 0.5429,
      "step": 4344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31871749956436807,
      "lm_loss": 0.4527,
      "step": 4344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31871749956436807,
      "lm_loss": 0.4691,
      "step": 4344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31871749956436807,
      "lm_loss": 0.448,
      "step": 4344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31871749956436807,
      "lm_loss": 0.5672,
      "step": 4344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31871749956436807,
      "lm_loss": 0.4714,
      "step": 4344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31871749956436807,
      "lm_loss": 0.4521,
      "step": 4344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31879086915450716,
      "grad_norm": 0.5637116775129196,
      "learning_rate": 7.967689476696678e-06,
      "loss": 0.4934,
      "step": 4345
    },
    {
      "epoch": 0.3188642387446463,
      "grad_norm": 1.6803565614857203,
      "learning_rate": 7.966733124848721e-06,
      "loss": 0.8746,
      "step": 4346
    },
    {
      "epoch": 0.31893760833478546,
      "grad_norm": 0.5651757934648419,
      "learning_rate": 7.965776605461997e-06,
      "loss": 0.5351,
      "step": 4347
    },
    {
      "epoch": 0.31901097792492455,
      "grad_norm": 1.5727065667796167,
      "learning_rate": 7.964819918590523e-06,
      "loss": 0.8471,
      "step": 4348
    },
    {
      "epoch": 0.3190843475150637,
      "grad_norm": 1.3911317571196264,
      "learning_rate": 7.963863064288326e-06,
      "loss": 0.8695,
      "step": 4349
    },
    {
      "epoch": 0.3191577171052028,
      "grad_norm": 1.4174421872218563,
      "learning_rate": 7.962906042609443e-06,
      "loss": 0.786,
      "step": 4350
    },
    {
      "epoch": 0.31923108669534195,
      "grad_norm": 1.580047323496869,
      "learning_rate": 7.961948853607917e-06,
      "loss": 0.8895,
      "step": 4351
    },
    {
      "epoch": 0.3193044562854811,
      "grad_norm": 1.3350112132675256,
      "learning_rate": 7.960991497337805e-06,
      "loss": 0.7337,
      "step": 4352
    },
    {
      "epoch": 0.3193044562854811,
      "lm_loss": 0.6168,
      "step": 4352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3193044562854811,
      "lm_loss": 0.5267,
      "step": 4352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3193044562854811,
      "lm_loss": 0.4242,
      "step": 4352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3193044562854811,
      "lm_loss": 0.691,
      "step": 4352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3193044562854811,
      "lm_loss": 0.5104,
      "step": 4352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3193044562854811,
      "lm_loss": 0.5208,
      "step": 4352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3193044562854811,
      "lm_loss": 0.7013,
      "step": 4352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3193044562854811,
      "lm_loss": 0.6037,
      "step": 4352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3193778258756202,
      "grad_norm": 0.6474013285017316,
      "learning_rate": 7.96003397385317e-06,
      "loss": 0.5374,
      "step": 4353
    },
    {
      "epoch": 0.31945119546575934,
      "grad_norm": 1.647156951272638,
      "learning_rate": 7.959076283208087e-06,
      "loss": 0.8739,
      "step": 4354
    },
    {
      "epoch": 0.31952456505589844,
      "grad_norm": 0.5878366585135327,
      "learning_rate": 7.958118425456635e-06,
      "loss": 0.5075,
      "step": 4355
    },
    {
      "epoch": 0.3195979346460376,
      "grad_norm": 1.7782745528115125,
      "learning_rate": 7.957160400652914e-06,
      "loss": 0.8449,
      "step": 4356
    },
    {
      "epoch": 0.31967130423617673,
      "grad_norm": 1.6335668946586628,
      "learning_rate": 7.95620220885102e-06,
      "loss": 0.7799,
      "step": 4357
    },
    {
      "epoch": 0.31974467382631583,
      "grad_norm": 1.472028823688966,
      "learning_rate": 7.955243850105065e-06,
      "loss": 0.8482,
      "step": 4358
    },
    {
      "epoch": 0.319818043416455,
      "grad_norm": 1.459954163748496,
      "learning_rate": 7.954285324469172e-06,
      "loss": 0.9038,
      "step": 4359
    },
    {
      "epoch": 0.31989141300659407,
      "grad_norm": 1.5768551524288055,
      "learning_rate": 7.953326631997472e-06,
      "loss": 0.8557,
      "step": 4360
    },
    {
      "epoch": 0.31989141300659407,
      "lm_loss": 0.5452,
      "step": 4360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31989141300659407,
      "lm_loss": 0.5537,
      "step": 4360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31989141300659407,
      "lm_loss": 0.5427,
      "step": 4360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31989141300659407,
      "lm_loss": 0.2058,
      "step": 4360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31989141300659407,
      "lm_loss": 0.4557,
      "step": 4360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31989141300659407,
      "lm_loss": 0.5351,
      "step": 4360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31989141300659407,
      "lm_loss": 0.7433,
      "step": 4360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.31989141300659407,
      "lm_loss": 0.4979,
      "step": 4360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3199647825967332,
      "grad_norm": 0.644689476245912,
      "learning_rate": 7.952367772744102e-06,
      "loss": 0.5127,
      "step": 4361
    },
    {
      "epoch": 0.32003815218687237,
      "grad_norm": 1.4966240308013972,
      "learning_rate": 7.951408746763215e-06,
      "loss": 0.7528,
      "step": 4362
    },
    {
      "epoch": 0.32011152177701147,
      "grad_norm": 1.4517921301463648,
      "learning_rate": 7.950449554108962e-06,
      "loss": 0.8635,
      "step": 4363
    },
    {
      "epoch": 0.3201848913671506,
      "grad_norm": 1.4436041455797863,
      "learning_rate": 7.949490194835521e-06,
      "loss": 0.7533,
      "step": 4364
    },
    {
      "epoch": 0.3202582609572897,
      "grad_norm": 1.410335207424134,
      "learning_rate": 7.948530668997063e-06,
      "loss": 0.7802,
      "step": 4365
    },
    {
      "epoch": 0.32033163054742886,
      "grad_norm": 0.5607344405862986,
      "learning_rate": 7.947570976647775e-06,
      "loss": 0.5184,
      "step": 4366
    },
    {
      "epoch": 0.320405000137568,
      "grad_norm": 0.5815383666921284,
      "learning_rate": 7.946611117841854e-06,
      "loss": 0.5323,
      "step": 4367
    },
    {
      "epoch": 0.3204783697277071,
      "grad_norm": 1.596286228215562,
      "learning_rate": 7.945651092633509e-06,
      "loss": 0.7527,
      "step": 4368
    },
    {
      "epoch": 0.3204783697277071,
      "lm_loss": 1.3092,
      "step": 4368,
      "vm_loss": 0.0953
    },
    {
      "epoch": 0.3204783697277071,
      "lm_loss": 1.8098,
      "step": 4368,
      "vm_loss": 0.1735
    },
    {
      "epoch": 0.3204783697277071,
      "lm_loss": 0.288,
      "step": 4368,
      "vm_loss": 0.0935
    },
    {
      "epoch": 0.3204783697277071,
      "lm_loss": 0.8062,
      "step": 4368,
      "vm_loss": 0.1967
    },
    {
      "epoch": 0.3204783697277071,
      "lm_loss": 0.2587,
      "step": 4368,
      "vm_loss": 0.0994
    },
    {
      "epoch": 0.3204783697277071,
      "lm_loss": 0.2345,
      "step": 4368,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.3204783697277071,
      "lm_loss": 0.7769,
      "step": 4368,
      "vm_loss": 0.2695
    },
    {
      "epoch": 0.3204783697277071,
      "lm_loss": 0.0825,
      "step": 4368,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.32055173931784625,
      "grad_norm": 1.6584089842077452,
      "learning_rate": 7.944690901076949e-06,
      "loss": 0.8283,
      "step": 4369
    },
    {
      "epoch": 0.32062510890798535,
      "grad_norm": 1.5294842698873172,
      "learning_rate": 7.943730543226402e-06,
      "loss": 0.8674,
      "step": 4370
    },
    {
      "epoch": 0.3206984784981245,
      "grad_norm": 1.5132882312575398,
      "learning_rate": 7.942770019136101e-06,
      "loss": 0.8407,
      "step": 4371
    },
    {
      "epoch": 0.3207718480882636,
      "grad_norm": 1.3701631886171652,
      "learning_rate": 7.94180932886029e-06,
      "loss": 0.7032,
      "step": 4372
    },
    {
      "epoch": 0.32084521767840274,
      "grad_norm": 0.6341703384001566,
      "learning_rate": 7.940848472453223e-06,
      "loss": 0.5121,
      "step": 4373
    },
    {
      "epoch": 0.3209185872685419,
      "grad_norm": 1.4068293043333437,
      "learning_rate": 7.939887449969156e-06,
      "loss": 0.7854,
      "step": 4374
    },
    {
      "epoch": 0.320991956858681,
      "grad_norm": 1.3680301566770583,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.9673,
      "step": 4375
    },
    {
      "epoch": 0.32106532644882013,
      "grad_norm": 1.5250096732060405,
      "learning_rate": 7.937964906987133e-06,
      "loss": 0.8292,
      "step": 4376
    },
    {
      "epoch": 0.32106532644882013,
      "lm_loss": 0.8052,
      "step": 4376,
      "vm_loss": 0.0643
    },
    {
      "epoch": 0.32106532644882013,
      "lm_loss": 0.2162,
      "step": 4376,
      "vm_loss": 0.054
    },
    {
      "epoch": 0.32106532644882013,
      "lm_loss": 0.1013,
      "step": 4376,
      "vm_loss": 0.1023
    },
    {
      "epoch": 0.32106532644882013,
      "lm_loss": 0.5741,
      "step": 4376,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.32106532644882013,
      "lm_loss": 0.719,
      "step": 4376,
      "vm_loss": 0.1322
    },
    {
      "epoch": 0.32106532644882013,
      "lm_loss": 0.4739,
      "step": 4376,
      "vm_loss": 0.1263
    },
    {
      "epoch": 0.32106532644882013,
      "lm_loss": 0.3,
      "step": 4376,
      "vm_loss": 0.1611
    },
    {
      "epoch": 0.32106532644882013,
      "lm_loss": 1.1708,
      "step": 4376,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.3211386960389592,
      "grad_norm": 1.5354032972860545,
      "learning_rate": 7.937003386597743e-06,
      "loss": 0.7922,
      "step": 4377
    },
    {
      "epoch": 0.3212120656290984,
      "grad_norm": 1.5671132419819442,
      "learning_rate": 7.936041700348502e-06,
      "loss": 0.8276,
      "step": 4378
    },
    {
      "epoch": 0.3212854352192375,
      "grad_norm": 1.5524709889032702,
      "learning_rate": 7.935079848293712e-06,
      "loss": 0.8923,
      "step": 4379
    },
    {
      "epoch": 0.3213588048093766,
      "grad_norm": 1.578027658977009,
      "learning_rate": 7.934117830487698e-06,
      "loss": 0.8936,
      "step": 4380
    },
    {
      "epoch": 0.32143217439951577,
      "grad_norm": 0.6392566165138656,
      "learning_rate": 7.93315564698478e-06,
      "loss": 0.5175,
      "step": 4381
    },
    {
      "epoch": 0.32150554398965486,
      "grad_norm": 1.775856611963697,
      "learning_rate": 7.932193297839301e-06,
      "loss": 0.8278,
      "step": 4382
    },
    {
      "epoch": 0.321578913579794,
      "grad_norm": 1.4828865065716141,
      "learning_rate": 7.931230783105604e-06,
      "loss": 0.7745,
      "step": 4383
    },
    {
      "epoch": 0.32165228316993316,
      "grad_norm": 1.5176687470263166,
      "learning_rate": 7.930268102838047e-06,
      "loss": 0.8492,
      "step": 4384
    },
    {
      "epoch": 0.32165228316993316,
      "lm_loss": 0.4363,
      "step": 4384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32165228316993316,
      "lm_loss": 0.556,
      "step": 4384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32165228316993316,
      "lm_loss": 0.5041,
      "step": 4384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32165228316993316,
      "lm_loss": 0.5473,
      "step": 4384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32165228316993316,
      "lm_loss": 0.4066,
      "step": 4384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32165228316993316,
      "lm_loss": 0.4086,
      "step": 4384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32165228316993316,
      "lm_loss": 0.5042,
      "step": 4384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32165228316993316,
      "lm_loss": 0.6373,
      "step": 4384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32172565276007226,
      "grad_norm": 0.5713395959080612,
      "learning_rate": 7.929305257090992e-06,
      "loss": 0.519,
      "step": 4385
    },
    {
      "epoch": 0.3217990223502114,
      "grad_norm": 1.5726794225165972,
      "learning_rate": 7.928342245918815e-06,
      "loss": 0.8523,
      "step": 4386
    },
    {
      "epoch": 0.3218723919403505,
      "grad_norm": 1.53784924347956,
      "learning_rate": 7.9273790693759e-06,
      "loss": 0.8024,
      "step": 4387
    },
    {
      "epoch": 0.32194576153048965,
      "grad_norm": 0.5665581208185815,
      "learning_rate": 7.92641572751664e-06,
      "loss": 0.5135,
      "step": 4388
    },
    {
      "epoch": 0.3220191311206288,
      "grad_norm": 1.5407662696215816,
      "learning_rate": 7.925452220395436e-06,
      "loss": 0.8651,
      "step": 4389
    },
    {
      "epoch": 0.3220925007107679,
      "grad_norm": 0.5784183524034158,
      "learning_rate": 7.9244885480667e-06,
      "loss": 0.5296,
      "step": 4390
    },
    {
      "epoch": 0.32216587030090704,
      "grad_norm": 1.5463321096202145,
      "learning_rate": 7.923524710584854e-06,
      "loss": 0.9192,
      "step": 4391
    },
    {
      "epoch": 0.32223923989104614,
      "grad_norm": 1.4171787245030456,
      "learning_rate": 7.922560708004327e-06,
      "loss": 0.8144,
      "step": 4392
    },
    {
      "epoch": 0.32223923989104614,
      "lm_loss": 1.3851,
      "step": 4392,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.32223923989104614,
      "lm_loss": 0.3377,
      "step": 4392,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.32223923989104614,
      "lm_loss": 0.4048,
      "step": 4392,
      "vm_loss": 0.0748
    },
    {
      "epoch": 0.32223923989104614,
      "lm_loss": 1.164,
      "step": 4392,
      "vm_loss": 0.1717
    },
    {
      "epoch": 0.32223923989104614,
      "lm_loss": 1.0071,
      "step": 4392,
      "vm_loss": 0.1873
    },
    {
      "epoch": 0.32223923989104614,
      "lm_loss": 0.4224,
      "step": 4392,
      "vm_loss": 0.1376
    },
    {
      "epoch": 0.32223923989104614,
      "lm_loss": 0.7511,
      "step": 4392,
      "vm_loss": 0.2318
    },
    {
      "epoch": 0.32223923989104614,
      "lm_loss": 0.7295,
      "step": 4392,
      "vm_loss": 0.1103
    },
    {
      "epoch": 0.3223126094811853,
      "grad_norm": 1.3843796884955413,
      "learning_rate": 7.921596540379559e-06,
      "loss": 0.7476,
      "step": 4393
    },
    {
      "epoch": 0.32238597907132444,
      "grad_norm": 1.8439611634451951,
      "learning_rate": 7.920632207764999e-06,
      "loss": 0.825,
      "step": 4394
    },
    {
      "epoch": 0.32245934866146353,
      "grad_norm": 0.5961746033089179,
      "learning_rate": 7.919667710215108e-06,
      "loss": 0.524,
      "step": 4395
    },
    {
      "epoch": 0.3225327182516027,
      "grad_norm": 1.6568496253643679,
      "learning_rate": 7.918703047784348e-06,
      "loss": 0.9325,
      "step": 4396
    },
    {
      "epoch": 0.3226060878417418,
      "grad_norm": 1.4768134500723495,
      "learning_rate": 7.9177382205272e-06,
      "loss": 0.7892,
      "step": 4397
    },
    {
      "epoch": 0.3226794574318809,
      "grad_norm": 0.5731651779027197,
      "learning_rate": 7.916773228498149e-06,
      "loss": 0.5379,
      "step": 4398
    },
    {
      "epoch": 0.3227528270220201,
      "grad_norm": 1.7945274461294807,
      "learning_rate": 7.915808071751692e-06,
      "loss": 0.809,
      "step": 4399
    },
    {
      "epoch": 0.32282619661215917,
      "grad_norm": 1.416260391800967,
      "learning_rate": 7.91484275034233e-06,
      "loss": 0.8324,
      "step": 4400
    },
    {
      "epoch": 0.32282619661215917,
      "lm_loss": 1.4512,
      "step": 4400,
      "vm_loss": 0.0562
    },
    {
      "epoch": 0.32282619661215917,
      "lm_loss": 0.5111,
      "step": 4400,
      "vm_loss": 0.0962
    },
    {
      "epoch": 0.32282619661215917,
      "lm_loss": 0.7611,
      "step": 4400,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.32282619661215917,
      "lm_loss": 0.9015,
      "step": 4400,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.32282619661215917,
      "lm_loss": 0.9127,
      "step": 4400,
      "vm_loss": 0.0846
    },
    {
      "epoch": 0.32282619661215917,
      "lm_loss": 0.756,
      "step": 4400,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.32282619661215917,
      "lm_loss": 0.3119,
      "step": 4400,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.32282619661215917,
      "lm_loss": 1.8123,
      "step": 4400,
      "vm_loss": 0.0876
    },
    {
      "epoch": 0.3228995662022983,
      "grad_norm": 1.4848645659415383,
      "learning_rate": 7.913877264324579e-06,
      "loss": 0.8374,
      "step": 4401
    },
    {
      "epoch": 0.3229729357924374,
      "grad_norm": 1.298365077881094,
      "learning_rate": 7.912911613752965e-06,
      "loss": 0.7935,
      "step": 4402
    },
    {
      "epoch": 0.32304630538257656,
      "grad_norm": 1.3890293103087832,
      "learning_rate": 7.911945798682017e-06,
      "loss": 0.7957,
      "step": 4403
    },
    {
      "epoch": 0.3231196749727157,
      "grad_norm": 0.5747331976690705,
      "learning_rate": 7.91097981916628e-06,
      "loss": 0.5187,
      "step": 4404
    },
    {
      "epoch": 0.3231930445628548,
      "grad_norm": 1.492452502655867,
      "learning_rate": 7.910013675260301e-06,
      "loss": 0.8553,
      "step": 4405
    },
    {
      "epoch": 0.32326641415299395,
      "grad_norm": 1.3888811129656549,
      "learning_rate": 7.909047367018644e-06,
      "loss": 0.7742,
      "step": 4406
    },
    {
      "epoch": 0.32333978374313305,
      "grad_norm": 0.5729507669845211,
      "learning_rate": 7.908080894495881e-06,
      "loss": 0.5219,
      "step": 4407
    },
    {
      "epoch": 0.3234131533332722,
      "grad_norm": 1.4107807176857154,
      "learning_rate": 7.907114257746585e-06,
      "loss": 0.8198,
      "step": 4408
    },
    {
      "epoch": 0.3234131533332722,
      "lm_loss": 1.666,
      "step": 4408,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.3234131533332722,
      "lm_loss": 0.5273,
      "step": 4408,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.3234131533332722,
      "lm_loss": 1.2423,
      "step": 4408,
      "vm_loss": 0.0851
    },
    {
      "epoch": 0.3234131533332722,
      "lm_loss": 0.884,
      "step": 4408,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.3234131533332722,
      "lm_loss": 0.1017,
      "step": 4408,
      "vm_loss": 0.0591
    },
    {
      "epoch": 0.3234131533332722,
      "lm_loss": 0.1123,
      "step": 4408,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.3234131533332722,
      "lm_loss": 0.7821,
      "step": 4408,
      "vm_loss": 0.2107
    },
    {
      "epoch": 0.3234131533332722,
      "lm_loss": 0.6999,
      "step": 4408,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.32348652292341135,
      "grad_norm": 1.5102919957554761,
      "learning_rate": 7.906147456825349e-06,
      "loss": 0.7962,
      "step": 4409
    },
    {
      "epoch": 0.32355989251355044,
      "grad_norm": 0.5891112886672532,
      "learning_rate": 7.905180491786769e-06,
      "loss": 0.504,
      "step": 4410
    },
    {
      "epoch": 0.3236332621036896,
      "grad_norm": 1.478563652705751,
      "learning_rate": 7.904213362685451e-06,
      "loss": 0.7382,
      "step": 4411
    },
    {
      "epoch": 0.3237066316938287,
      "grad_norm": 0.5593101555058031,
      "learning_rate": 7.903246069576013e-06,
      "loss": 0.5282,
      "step": 4412
    },
    {
      "epoch": 0.32378000128396783,
      "grad_norm": 0.5559565851668393,
      "learning_rate": 7.90227861251308e-06,
      "loss": 0.524,
      "step": 4413
    },
    {
      "epoch": 0.3238533708741069,
      "grad_norm": 1.5360653079505873,
      "learning_rate": 7.901310991551285e-06,
      "loss": 0.7333,
      "step": 4414
    },
    {
      "epoch": 0.3239267404642461,
      "grad_norm": 1.4995751870756446,
      "learning_rate": 7.900343206745276e-06,
      "loss": 0.837,
      "step": 4415
    },
    {
      "epoch": 0.3240001100543852,
      "grad_norm": 1.4903635925285117,
      "learning_rate": 7.899375258149702e-06,
      "loss": 0.9007,
      "step": 4416
    },
    {
      "epoch": 0.3240001100543852,
      "lm_loss": 1.3056,
      "step": 4416,
      "vm_loss": 0.0833
    },
    {
      "epoch": 0.3240001100543852,
      "lm_loss": 1.5804,
      "step": 4416,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.3240001100543852,
      "lm_loss": 1.5495,
      "step": 4416,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.3240001100543852,
      "lm_loss": 0.1045,
      "step": 4416,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.3240001100543852,
      "lm_loss": 0.1948,
      "step": 4416,
      "vm_loss": 0.1486
    },
    {
      "epoch": 0.3240001100543852,
      "lm_loss": 0.9089,
      "step": 4416,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.3240001100543852,
      "lm_loss": 0.5132,
      "step": 4416,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.3240001100543852,
      "lm_loss": 0.501,
      "step": 4416,
      "vm_loss": 0.0624
    },
    {
      "epoch": 0.3240734796445243,
      "grad_norm": 1.3311266578910899,
      "learning_rate": 7.898407145819227e-06,
      "loss": 0.8261,
      "step": 4417
    },
    {
      "epoch": 0.32414684923466347,
      "grad_norm": 1.4818897648009832,
      "learning_rate": 7.897438869808522e-06,
      "loss": 0.8347,
      "step": 4418
    },
    {
      "epoch": 0.32422021882480256,
      "grad_norm": 1.630225462998335,
      "learning_rate": 7.89647043017227e-06,
      "loss": 0.7944,
      "step": 4419
    },
    {
      "epoch": 0.3242935884149417,
      "grad_norm": 1.4466769589704258,
      "learning_rate": 7.895501826965158e-06,
      "loss": 0.924,
      "step": 4420
    },
    {
      "epoch": 0.32436695800508086,
      "grad_norm": 1.4222884222300662,
      "learning_rate": 7.894533060241889e-06,
      "loss": 0.8815,
      "step": 4421
    },
    {
      "epoch": 0.32444032759521996,
      "grad_norm": 1.6669055308557494,
      "learning_rate": 7.893564130057169e-06,
      "loss": 0.7943,
      "step": 4422
    },
    {
      "epoch": 0.3245136971853591,
      "grad_norm": 1.59352576586555,
      "learning_rate": 7.892595036465718e-06,
      "loss": 0.8344,
      "step": 4423
    },
    {
      "epoch": 0.3245870667754982,
      "grad_norm": 0.643300215738706,
      "learning_rate": 7.891625779522259e-06,
      "loss": 0.5227,
      "step": 4424
    },
    {
      "epoch": 0.3245870667754982,
      "lm_loss": 0.4667,
      "step": 4424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3245870667754982,
      "lm_loss": 0.3816,
      "step": 4424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3245870667754982,
      "lm_loss": 0.6223,
      "step": 4424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3245870667754982,
      "lm_loss": 0.4628,
      "step": 4424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3245870667754982,
      "lm_loss": 0.5947,
      "step": 4424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3245870667754982,
      "lm_loss": 0.5403,
      "step": 4424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3245870667754982,
      "lm_loss": 0.5614,
      "step": 4424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3245870667754982,
      "lm_loss": 0.557,
      "step": 4424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32466043636563735,
      "grad_norm": 0.6046476688294936,
      "learning_rate": 7.890656359281534e-06,
      "loss": 0.522,
      "step": 4425
    },
    {
      "epoch": 0.3247338059557765,
      "grad_norm": 0.5895953804447307,
      "learning_rate": 7.889686775798285e-06,
      "loss": 0.5287,
      "step": 4426
    },
    {
      "epoch": 0.3248071755459156,
      "grad_norm": 1.5693353047334664,
      "learning_rate": 7.888717029127267e-06,
      "loss": 0.8098,
      "step": 4427
    },
    {
      "epoch": 0.32488054513605474,
      "grad_norm": 1.4331977873771,
      "learning_rate": 7.887747119323245e-06,
      "loss": 0.8894,
      "step": 4428
    },
    {
      "epoch": 0.32495391472619384,
      "grad_norm": 1.7342305743656332,
      "learning_rate": 7.886777046440993e-06,
      "loss": 0.9143,
      "step": 4429
    },
    {
      "epoch": 0.325027284316333,
      "grad_norm": 1.4803011594081272,
      "learning_rate": 7.88580681053529e-06,
      "loss": 0.846,
      "step": 4430
    },
    {
      "epoch": 0.32510065390647214,
      "grad_norm": 1.5345239346534036,
      "learning_rate": 7.884836411660933e-06,
      "loss": 0.7808,
      "step": 4431
    },
    {
      "epoch": 0.32517402349661123,
      "grad_norm": 1.5624301188454157,
      "learning_rate": 7.883865849872717e-06,
      "loss": 0.7645,
      "step": 4432
    },
    {
      "epoch": 0.32517402349661123,
      "lm_loss": 1.1464,
      "step": 4432,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.32517402349661123,
      "lm_loss": 1.0551,
      "step": 4432,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.32517402349661123,
      "lm_loss": 0.146,
      "step": 4432,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.32517402349661123,
      "lm_loss": 0.6001,
      "step": 4432,
      "vm_loss": 0.1563
    },
    {
      "epoch": 0.32517402349661123,
      "lm_loss": 0.2998,
      "step": 4432,
      "vm_loss": 0.1533
    },
    {
      "epoch": 0.32517402349661123,
      "lm_loss": 0.9698,
      "step": 4432,
      "vm_loss": 0.2086
    },
    {
      "epoch": 0.32517402349661123,
      "lm_loss": 0.4201,
      "step": 4432,
      "vm_loss": 0.1131
    },
    {
      "epoch": 0.32517402349661123,
      "lm_loss": 0.8258,
      "step": 4432,
      "vm_loss": 0.2176
    },
    {
      "epoch": 0.3252473930867504,
      "grad_norm": 1.4407995056654455,
      "learning_rate": 7.882895125225454e-06,
      "loss": 0.8127,
      "step": 4433
    },
    {
      "epoch": 0.3253207626768895,
      "grad_norm": 1.5742874517572703,
      "learning_rate": 7.881924237773965e-06,
      "loss": 0.8841,
      "step": 4434
    },
    {
      "epoch": 0.3253941322670286,
      "grad_norm": 1.7004440351507217,
      "learning_rate": 7.880953187573079e-06,
      "loss": 0.7893,
      "step": 4435
    },
    {
      "epoch": 0.3254675018571678,
      "grad_norm": 1.3669022064514338,
      "learning_rate": 7.879981974677631e-06,
      "loss": 0.7252,
      "step": 4436
    },
    {
      "epoch": 0.32554087144730687,
      "grad_norm": 1.989456132937726,
      "learning_rate": 7.879010599142468e-06,
      "loss": 0.7733,
      "step": 4437
    },
    {
      "epoch": 0.325614241037446,
      "grad_norm": 1.1156178649691508,
      "learning_rate": 7.878039061022445e-06,
      "loss": 0.5308,
      "step": 4438
    },
    {
      "epoch": 0.3256876106275851,
      "grad_norm": 0.84819737050569,
      "learning_rate": 7.877067360372432e-06,
      "loss": 0.5413,
      "step": 4439
    },
    {
      "epoch": 0.32576098021772426,
      "grad_norm": 0.5880882724696872,
      "learning_rate": 7.876095497247299e-06,
      "loss": 0.5261,
      "step": 4440
    },
    {
      "epoch": 0.32576098021772426,
      "lm_loss": 0.3668,
      "step": 4440,
      "vm_loss": 0.0259
    },
    {
      "epoch": 0.32576098021772426,
      "lm_loss": 1.39,
      "step": 4440,
      "vm_loss": 0.1918
    },
    {
      "epoch": 0.32576098021772426,
      "lm_loss": 1.1361,
      "step": 4440,
      "vm_loss": 0.1478
    },
    {
      "epoch": 0.32576098021772426,
      "lm_loss": 1.1317,
      "step": 4440,
      "vm_loss": 0.1957
    },
    {
      "epoch": 0.32576098021772426,
      "lm_loss": 0.8246,
      "step": 4440,
      "vm_loss": 0.1343
    },
    {
      "epoch": 0.32576098021772426,
      "lm_loss": 0.5858,
      "step": 4440,
      "vm_loss": 0.0553
    },
    {
      "epoch": 0.32576098021772426,
      "lm_loss": 0.1478,
      "step": 4440,
      "vm_loss": 0.1843
    },
    {
      "epoch": 0.32576098021772426,
      "lm_loss": 0.721,
      "step": 4440,
      "vm_loss": 0.1293
    },
    {
      "epoch": 0.3258343498078634,
      "grad_norm": 1.619535454842886,
      "learning_rate": 7.875123471701931e-06,
      "loss": 0.8366,
      "step": 4441
    },
    {
      "epoch": 0.3259077193980025,
      "grad_norm": 1.6836826047629765,
      "learning_rate": 7.874151283791219e-06,
      "loss": 0.8668,
      "step": 4442
    },
    {
      "epoch": 0.32598108898814165,
      "grad_norm": 1.7405398336525952,
      "learning_rate": 7.873178933570068e-06,
      "loss": 0.8448,
      "step": 4443
    },
    {
      "epoch": 0.32605445857828075,
      "grad_norm": 1.3625406766274644,
      "learning_rate": 7.872206421093387e-06,
      "loss": 0.5647,
      "step": 4444
    },
    {
      "epoch": 0.3261278281684199,
      "grad_norm": 1.7465988705382036,
      "learning_rate": 7.871233746416096e-06,
      "loss": 0.8288,
      "step": 4445
    },
    {
      "epoch": 0.32620119775855905,
      "grad_norm": 1.0422636509945649,
      "learning_rate": 7.870260909593124e-06,
      "loss": 0.5399,
      "step": 4446
    },
    {
      "epoch": 0.32627456734869814,
      "grad_norm": 1.4764344212799843,
      "learning_rate": 7.869287910679413e-06,
      "loss": 0.8589,
      "step": 4447
    },
    {
      "epoch": 0.3263479369388373,
      "grad_norm": 1.6657835833213002,
      "learning_rate": 7.868314749729906e-06,
      "loss": 0.8091,
      "step": 4448
    },
    {
      "epoch": 0.3263479369388373,
      "lm_loss": 1.501,
      "step": 4448,
      "vm_loss": 0.0463
    },
    {
      "epoch": 0.3263479369388373,
      "lm_loss": 0.3844,
      "step": 4448,
      "vm_loss": 0.1624
    },
    {
      "epoch": 0.3263479369388373,
      "lm_loss": 0.2689,
      "step": 4448,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0.3263479369388373,
      "lm_loss": 0.8473,
      "step": 4448,
      "vm_loss": 0.1594
    },
    {
      "epoch": 0.3263479369388373,
      "lm_loss": 0.7483,
      "step": 4448,
      "vm_loss": 0.1855
    },
    {
      "epoch": 0.3263479369388373,
      "lm_loss": 0.9197,
      "step": 4448,
      "vm_loss": 0.1031
    },
    {
      "epoch": 0.3263479369388373,
      "lm_loss": 0.8893,
      "step": 4448,
      "vm_loss": 0.1622
    },
    {
      "epoch": 0.3263479369388373,
      "lm_loss": 0.1727,
      "step": 4448,
      "vm_loss": 0.0736
    },
    {
      "epoch": 0.3264213065289764,
      "grad_norm": 1.5933817072440943,
      "learning_rate": 7.867341426799562e-06,
      "loss": 0.8097,
      "step": 4449
    },
    {
      "epoch": 0.32649467611911553,
      "grad_norm": 0.7660376637864216,
      "learning_rate": 7.866367941943348e-06,
      "loss": 0.5243,
      "step": 4450
    },
    {
      "epoch": 0.3265680457092547,
      "grad_norm": 1.7799433726595162,
      "learning_rate": 7.865394295216237e-06,
      "loss": 0.8261,
      "step": 4451
    },
    {
      "epoch": 0.3266414152993938,
      "grad_norm": 1.6104877913063722,
      "learning_rate": 7.864420486673214e-06,
      "loss": 0.9012,
      "step": 4452
    },
    {
      "epoch": 0.3267147848895329,
      "grad_norm": 1.716150980300306,
      "learning_rate": 7.863446516369271e-06,
      "loss": 0.7984,
      "step": 4453
    },
    {
      "epoch": 0.326788154479672,
      "grad_norm": 1.5287455038887314,
      "learning_rate": 7.862472384359414e-06,
      "loss": 0.8149,
      "step": 4454
    },
    {
      "epoch": 0.32686152406981117,
      "grad_norm": 1.8643556130508088,
      "learning_rate": 7.861498090698652e-06,
      "loss": 0.8458,
      "step": 4455
    },
    {
      "epoch": 0.32693489365995027,
      "grad_norm": 0.9440980411593806,
      "learning_rate": 7.860523635442008e-06,
      "loss": 0.5293,
      "step": 4456
    },
    {
      "epoch": 0.32693489365995027,
      "lm_loss": 0.5379,
      "step": 4456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32693489365995027,
      "lm_loss": 0.4629,
      "step": 4456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32693489365995027,
      "lm_loss": 0.6394,
      "step": 4456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32693489365995027,
      "lm_loss": 0.6268,
      "step": 4456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32693489365995027,
      "lm_loss": 0.5227,
      "step": 4456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32693489365995027,
      "lm_loss": 0.4834,
      "step": 4456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32693489365995027,
      "lm_loss": 0.6519,
      "step": 4456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32693489365995027,
      "lm_loss": 0.5171,
      "step": 4456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3270082632500894,
      "grad_norm": 0.8342130802477862,
      "learning_rate": 7.859549018644507e-06,
      "loss": 0.5291,
      "step": 4457
    },
    {
      "epoch": 0.32708163284022856,
      "grad_norm": 1.6856032059936803,
      "learning_rate": 7.858574240361195e-06,
      "loss": 0.7719,
      "step": 4458
    },
    {
      "epoch": 0.32715500243036766,
      "grad_norm": 1.5405888791837,
      "learning_rate": 7.857599300647114e-06,
      "loss": 0.8873,
      "step": 4459
    },
    {
      "epoch": 0.3272283720205068,
      "grad_norm": 1.5792915773912575,
      "learning_rate": 7.856624199557326e-06,
      "loss": 0.8366,
      "step": 4460
    },
    {
      "epoch": 0.3273017416106459,
      "grad_norm": 1.3788764797463346,
      "learning_rate": 7.855648937146893e-06,
      "loss": 0.8277,
      "step": 4461
    },
    {
      "epoch": 0.32737511120078505,
      "grad_norm": 1.5281068115985812,
      "learning_rate": 7.854673513470894e-06,
      "loss": 0.8172,
      "step": 4462
    },
    {
      "epoch": 0.3274484807909242,
      "grad_norm": 1.4941949506082983,
      "learning_rate": 7.853697928584411e-06,
      "loss": 0.764,
      "step": 4463
    },
    {
      "epoch": 0.3275218503810633,
      "grad_norm": 1.4748844874453002,
      "learning_rate": 7.852722182542541e-06,
      "loss": 0.791,
      "step": 4464
    },
    {
      "epoch": 0.3275218503810633,
      "lm_loss": 0.6511,
      "step": 4464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3275218503810633,
      "lm_loss": 0.4008,
      "step": 4464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3275218503810633,
      "lm_loss": 0.5052,
      "step": 4464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3275218503810633,
      "lm_loss": 0.532,
      "step": 4464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3275218503810633,
      "lm_loss": 0.3813,
      "step": 4464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3275218503810633,
      "lm_loss": 0.6718,
      "step": 4464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3275218503810633,
      "lm_loss": 0.7048,
      "step": 4464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3275218503810633,
      "lm_loss": 0.6922,
      "step": 4464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.32759521997120244,
      "grad_norm": 0.909125065323743,
      "learning_rate": 7.851746275400383e-06,
      "loss": 0.5537,
      "step": 4465
    },
    {
      "epoch": 0.32766858956134154,
      "grad_norm": 1.573643449500872,
      "learning_rate": 7.85077020721305e-06,
      "loss": 0.7941,
      "step": 4466
    },
    {
      "epoch": 0.3277419591514807,
      "grad_norm": 1.642319350808445,
      "learning_rate": 7.849793978035664e-06,
      "loss": 0.8101,
      "step": 4467
    },
    {
      "epoch": 0.32781532874161984,
      "grad_norm": 1.615476417063594,
      "learning_rate": 7.848817587923355e-06,
      "loss": 0.8468,
      "step": 4468
    },
    {
      "epoch": 0.32788869833175893,
      "grad_norm": 1.487176930890913,
      "learning_rate": 7.847841036931263e-06,
      "loss": 0.8006,
      "step": 4469
    },
    {
      "epoch": 0.3279620679218981,
      "grad_norm": 1.4918069513107222,
      "learning_rate": 7.846864325114534e-06,
      "loss": 0.8309,
      "step": 4470
    },
    {
      "epoch": 0.3280354375120372,
      "grad_norm": 1.4370153590122299,
      "learning_rate": 7.845887452528326e-06,
      "loss": 0.7594,
      "step": 4471
    },
    {
      "epoch": 0.3281088071021763,
      "grad_norm": 1.5784443131406256,
      "learning_rate": 7.844910419227806e-06,
      "loss": 0.7559,
      "step": 4472
    },
    {
      "epoch": 0.3281088071021763,
      "lm_loss": 0.1491,
      "step": 4472,
      "vm_loss": 0.0556
    },
    {
      "epoch": 0.3281088071021763,
      "lm_loss": 0.816,
      "step": 4472,
      "vm_loss": 0.04
    },
    {
      "epoch": 0.3281088071021763,
      "lm_loss": 0.814,
      "step": 4472,
      "vm_loss": 0.1162
    },
    {
      "epoch": 0.3281088071021763,
      "lm_loss": 0.613,
      "step": 4472,
      "vm_loss": 0.0958
    },
    {
      "epoch": 0.3281088071021763,
      "lm_loss": 1.088,
      "step": 4472,
      "vm_loss": 0.1927
    },
    {
      "epoch": 0.3281088071021763,
      "lm_loss": 0.7441,
      "step": 4472,
      "vm_loss": 0.1511
    },
    {
      "epoch": 0.3281088071021763,
      "lm_loss": 0.7006,
      "step": 4472,
      "vm_loss": 0.1909
    },
    {
      "epoch": 0.3281088071021763,
      "lm_loss": 0.1114,
      "step": 4472,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.3281821766923155,
      "grad_norm": 1.6111340170152457,
      "learning_rate": 7.84393322526815e-06,
      "loss": 0.8801,
      "step": 4473
    },
    {
      "epoch": 0.32825554628245457,
      "grad_norm": 1.5512828444114757,
      "learning_rate": 7.842955870704541e-06,
      "loss": 0.8888,
      "step": 4474
    },
    {
      "epoch": 0.3283289158725937,
      "grad_norm": 1.5614143712788244,
      "learning_rate": 7.841978355592175e-06,
      "loss": 0.8214,
      "step": 4475
    },
    {
      "epoch": 0.3284022854627328,
      "grad_norm": 1.5096583808185937,
      "learning_rate": 7.841000679986251e-06,
      "loss": 0.8447,
      "step": 4476
    },
    {
      "epoch": 0.32847565505287196,
      "grad_norm": 1.5191000700067325,
      "learning_rate": 7.840022843941985e-06,
      "loss": 0.7412,
      "step": 4477
    },
    {
      "epoch": 0.3285490246430111,
      "grad_norm": 1.7543296866408071,
      "learning_rate": 7.839044847514595e-06,
      "loss": 0.9137,
      "step": 4478
    },
    {
      "epoch": 0.3286223942331502,
      "grad_norm": 1.5383114585493627,
      "learning_rate": 7.838066690759311e-06,
      "loss": 0.8566,
      "step": 4479
    },
    {
      "epoch": 0.32869576382328936,
      "grad_norm": 1.4681909549903642,
      "learning_rate": 7.837088373731374e-06,
      "loss": 0.8807,
      "step": 4480
    },
    {
      "epoch": 0.32869576382328936,
      "lm_loss": 1.2979,
      "step": 4480,
      "vm_loss": 0.064
    },
    {
      "epoch": 0.32869576382328936,
      "lm_loss": 0.071,
      "step": 4480,
      "vm_loss": 0.1619
    },
    {
      "epoch": 0.32869576382328936,
      "lm_loss": 1.3087,
      "step": 4480,
      "vm_loss": 0.1609
    },
    {
      "epoch": 0.32869576382328936,
      "lm_loss": 0.8294,
      "step": 4480,
      "vm_loss": 0.1708
    },
    {
      "epoch": 0.32869576382328936,
      "lm_loss": 0.14,
      "step": 4480,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.32869576382328936,
      "lm_loss": 0.5251,
      "step": 4480,
      "vm_loss": 0.1531
    },
    {
      "epoch": 0.32869576382328936,
      "lm_loss": 0.4167,
      "step": 4480,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.32869576382328936,
      "lm_loss": 0.9711,
      "step": 4480,
      "vm_loss": 0.1503
    },
    {
      "epoch": 0.32876913341342845,
      "grad_norm": 1.573470399753398,
      "learning_rate": 7.83610989648603e-06,
      "loss": 0.8981,
      "step": 4481
    },
    {
      "epoch": 0.3288425030035676,
      "grad_norm": 1.4963241411850292,
      "learning_rate": 7.835131259078537e-06,
      "loss": 0.6932,
      "step": 4482
    },
    {
      "epoch": 0.32891587259370675,
      "grad_norm": 1.4751276574376575,
      "learning_rate": 7.83415246156416e-06,
      "loss": 0.7571,
      "step": 4483
    },
    {
      "epoch": 0.32898924218384584,
      "grad_norm": 1.5465522851446374,
      "learning_rate": 7.833173503998174e-06,
      "loss": 0.8205,
      "step": 4484
    },
    {
      "epoch": 0.329062611773985,
      "grad_norm": 1.5135155845632708,
      "learning_rate": 7.832194386435864e-06,
      "loss": 0.8212,
      "step": 4485
    },
    {
      "epoch": 0.3291359813641241,
      "grad_norm": 1.5069849555248698,
      "learning_rate": 7.831215108932526e-06,
      "loss": 0.9203,
      "step": 4486
    },
    {
      "epoch": 0.32920935095426324,
      "grad_norm": 0.7869303142051166,
      "learning_rate": 7.830235671543458e-06,
      "loss": 0.5438,
      "step": 4487
    },
    {
      "epoch": 0.3292827205444024,
      "grad_norm": 0.6958064321357321,
      "learning_rate": 7.829256074323973e-06,
      "loss": 0.5207,
      "step": 4488
    },
    {
      "epoch": 0.3292827205444024,
      "lm_loss": 0.525,
      "step": 4488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3292827205444024,
      "lm_loss": 0.5176,
      "step": 4488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3292827205444024,
      "lm_loss": 0.4502,
      "step": 4488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3292827205444024,
      "lm_loss": 0.4734,
      "step": 4488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3292827205444024,
      "lm_loss": 0.4635,
      "step": 4488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3292827205444024,
      "lm_loss": 0.4082,
      "step": 4488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3292827205444024,
      "lm_loss": 0.5293,
      "step": 4488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3292827205444024,
      "lm_loss": 0.4833,
      "step": 4488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3293560901345415,
      "grad_norm": 0.5940566575907008,
      "learning_rate": 7.828276317329388e-06,
      "loss": 0.5183,
      "step": 4489
    },
    {
      "epoch": 0.32942945972468063,
      "grad_norm": 1.5849817966138309,
      "learning_rate": 7.827296400615038e-06,
      "loss": 0.8132,
      "step": 4490
    },
    {
      "epoch": 0.3295028293148197,
      "grad_norm": 1.9228233738083842,
      "learning_rate": 7.826316324236259e-06,
      "loss": 0.7611,
      "step": 4491
    },
    {
      "epoch": 0.3295761989049589,
      "grad_norm": 1.6314048004204316,
      "learning_rate": 7.825336088248397e-06,
      "loss": 0.8004,
      "step": 4492
    },
    {
      "epoch": 0.32964956849509797,
      "grad_norm": 2.1658281308283156,
      "learning_rate": 7.824355692706808e-06,
      "loss": 0.8276,
      "step": 4493
    },
    {
      "epoch": 0.3297229380852371,
      "grad_norm": 1.494827800702589,
      "learning_rate": 7.82337513766686e-06,
      "loss": 0.8232,
      "step": 4494
    },
    {
      "epoch": 0.32979630767537627,
      "grad_norm": 1.6122032611989487,
      "learning_rate": 7.822394423183927e-06,
      "loss": 0.8629,
      "step": 4495
    },
    {
      "epoch": 0.32986967726551536,
      "grad_norm": 1.7799813896271055,
      "learning_rate": 7.82141354931339e-06,
      "loss": 0.8539,
      "step": 4496
    },
    {
      "epoch": 0.32986967726551536,
      "lm_loss": 1.5979,
      "step": 4496,
      "vm_loss": 0.1101
    },
    {
      "epoch": 0.32986967726551536,
      "lm_loss": 0.3744,
      "step": 4496,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.32986967726551536,
      "lm_loss": 0.9717,
      "step": 4496,
      "vm_loss": 0.1371
    },
    {
      "epoch": 0.32986967726551536,
      "lm_loss": 0.8783,
      "step": 4496,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.32986967726551536,
      "lm_loss": 0.8376,
      "step": 4496,
      "vm_loss": 0.2277
    },
    {
      "epoch": 0.32986967726551536,
      "lm_loss": 0.1883,
      "step": 4496,
      "vm_loss": 0.1597
    },
    {
      "epoch": 0.32986967726551536,
      "lm_loss": 0.436,
      "step": 4496,
      "vm_loss": 0.1219
    },
    {
      "epoch": 0.32986967726551536,
      "lm_loss": 0.9091,
      "step": 4496,
      "vm_loss": 0.1353
    },
    {
      "epoch": 0.3299430468556545,
      "grad_norm": 1.5742645031982831,
      "learning_rate": 7.820432516110642e-06,
      "loss": 0.8369,
      "step": 4497
    },
    {
      "epoch": 0.3300164164457936,
      "grad_norm": 1.483323985964741,
      "learning_rate": 7.819451323631087e-06,
      "loss": 0.8638,
      "step": 4498
    },
    {
      "epoch": 0.33008978603593275,
      "grad_norm": 1.432170982205942,
      "learning_rate": 7.818469971930134e-06,
      "loss": 0.8773,
      "step": 4499
    },
    {
      "epoch": 0.3301631556260719,
      "grad_norm": 1.2484671671846481,
      "learning_rate": 7.817488461063201e-06,
      "loss": 0.5217,
      "step": 4500
    },
    {
      "epoch": 0.330236525216211,
      "grad_norm": 1.0163655709422652,
      "learning_rate": 7.816506791085717e-06,
      "loss": 0.5311,
      "step": 4501
    },
    {
      "epoch": 0.33030989480635015,
      "grad_norm": 2.4107961283616586,
      "learning_rate": 7.815524962053119e-06,
      "loss": 0.8305,
      "step": 4502
    },
    {
      "epoch": 0.33038326439648924,
      "grad_norm": 1.7177504722706083,
      "learning_rate": 7.814542974020855e-06,
      "loss": 0.7502,
      "step": 4503
    },
    {
      "epoch": 0.3304566339866284,
      "grad_norm": 0.9330453282508266,
      "learning_rate": 7.813560827044377e-06,
      "loss": 0.5266,
      "step": 4504
    },
    {
      "epoch": 0.3304566339866284,
      "lm_loss": 1.1949,
      "step": 4504,
      "vm_loss": 0.1471
    },
    {
      "epoch": 0.3304566339866284,
      "lm_loss": 1.6307,
      "step": 4504,
      "vm_loss": 0.1292
    },
    {
      "epoch": 0.3304566339866284,
      "lm_loss": 0.9422,
      "step": 4504,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.3304566339866284,
      "lm_loss": 1.1729,
      "step": 4504,
      "vm_loss": 0.1456
    },
    {
      "epoch": 0.3304566339866284,
      "lm_loss": 1.5871,
      "step": 4504,
      "vm_loss": 0.1601
    },
    {
      "epoch": 0.3304566339866284,
      "lm_loss": 0.4916,
      "step": 4504,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.3304566339866284,
      "lm_loss": 0.4571,
      "step": 4504,
      "vm_loss": 0.072
    },
    {
      "epoch": 0.3304566339866284,
      "lm_loss": 0.4627,
      "step": 4504,
      "vm_loss": 0.1788
    },
    {
      "epoch": 0.33053000357676754,
      "grad_norm": 1.9655433905113746,
      "learning_rate": 7.812578521179152e-06,
      "loss": 0.8818,
      "step": 4505
    },
    {
      "epoch": 0.33060337316690663,
      "grad_norm": 1.557108372294289,
      "learning_rate": 7.811596056480656e-06,
      "loss": 0.8795,
      "step": 4506
    },
    {
      "epoch": 0.3306767427570458,
      "grad_norm": 1.0633012341597095,
      "learning_rate": 7.810613433004364e-06,
      "loss": 0.5392,
      "step": 4507
    },
    {
      "epoch": 0.3307501123471849,
      "grad_norm": 1.616665562129354,
      "learning_rate": 7.809630650805774e-06,
      "loss": 0.7501,
      "step": 4508
    },
    {
      "epoch": 0.330823481937324,
      "grad_norm": 1.6344603109400564,
      "learning_rate": 7.80864770994038e-06,
      "loss": 0.8845,
      "step": 4509
    },
    {
      "epoch": 0.3308968515274632,
      "grad_norm": 0.7309828498683553,
      "learning_rate": 7.807664610463697e-06,
      "loss": 0.5262,
      "step": 4510
    },
    {
      "epoch": 0.33097022111760227,
      "grad_norm": 1.519952045206693,
      "learning_rate": 7.806681352431238e-06,
      "loss": 0.741,
      "step": 4511
    },
    {
      "epoch": 0.3310435907077414,
      "grad_norm": 1.4738779302121585,
      "learning_rate": 7.805697935898535e-06,
      "loss": 0.7903,
      "step": 4512
    },
    {
      "epoch": 0.3310435907077414,
      "lm_loss": 1.0894,
      "step": 4512,
      "vm_loss": 0.0872
    },
    {
      "epoch": 0.3310435907077414,
      "lm_loss": 0.8204,
      "step": 4512,
      "vm_loss": 0.0723
    },
    {
      "epoch": 0.3310435907077414,
      "lm_loss": 0.7875,
      "step": 4512,
      "vm_loss": 0.1744
    },
    {
      "epoch": 0.3310435907077414,
      "lm_loss": 0.9601,
      "step": 4512,
      "vm_loss": 0.2056
    },
    {
      "epoch": 0.3310435907077414,
      "lm_loss": 0.7108,
      "step": 4512,
      "vm_loss": 0.1819
    },
    {
      "epoch": 0.3310435907077414,
      "lm_loss": 0.1675,
      "step": 4512,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.3310435907077414,
      "lm_loss": 1.2822,
      "step": 4512,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.3310435907077414,
      "lm_loss": 0.7799,
      "step": 4512,
      "vm_loss": 0.1729
    },
    {
      "epoch": 0.3311169602978805,
      "grad_norm": 1.6646013876052237,
      "learning_rate": 7.80471436092112e-06,
      "loss": 0.8067,
      "step": 4513
    },
    {
      "epoch": 0.33119032988801966,
      "grad_norm": 1.7152926317250745,
      "learning_rate": 7.803730627554539e-06,
      "loss": 0.8365,
      "step": 4514
    },
    {
      "epoch": 0.3312636994781588,
      "grad_norm": 0.7739833216489364,
      "learning_rate": 7.802746735854347e-06,
      "loss": 0.5242,
      "step": 4515
    },
    {
      "epoch": 0.3313370690682979,
      "grad_norm": 1.6860448104257715,
      "learning_rate": 7.801762685876106e-06,
      "loss": 0.8725,
      "step": 4516
    },
    {
      "epoch": 0.33141043865843706,
      "grad_norm": 1.6305236433757133,
      "learning_rate": 7.800778477675386e-06,
      "loss": 0.8076,
      "step": 4517
    },
    {
      "epoch": 0.33148380824857615,
      "grad_norm": 0.6278771134085586,
      "learning_rate": 7.799794111307771e-06,
      "loss": 0.5413,
      "step": 4518
    },
    {
      "epoch": 0.3315571778387153,
      "grad_norm": 1.4659588728925412,
      "learning_rate": 7.798809586828848e-06,
      "loss": 0.7841,
      "step": 4519
    },
    {
      "epoch": 0.33163054742885445,
      "grad_norm": 1.5067647351575324,
      "learning_rate": 7.797824904294218e-06,
      "loss": 0.7025,
      "step": 4520
    },
    {
      "epoch": 0.33163054742885445,
      "lm_loss": 1.1643,
      "step": 4520,
      "vm_loss": 0.0599
    },
    {
      "epoch": 0.33163054742885445,
      "lm_loss": 0.6134,
      "step": 4520,
      "vm_loss": 0.1358
    },
    {
      "epoch": 0.33163054742885445,
      "lm_loss": 0.1163,
      "step": 4520,
      "vm_loss": 0.1117
    },
    {
      "epoch": 0.33163054742885445,
      "lm_loss": 1.0211,
      "step": 4520,
      "vm_loss": 0.3232
    },
    {
      "epoch": 0.33163054742885445,
      "lm_loss": 0.8711,
      "step": 4520,
      "vm_loss": 0.2082
    },
    {
      "epoch": 0.33163054742885445,
      "lm_loss": 0.7439,
      "step": 4520,
      "vm_loss": 0.1232
    },
    {
      "epoch": 0.33163054742885445,
      "lm_loss": 0.8611,
      "step": 4520,
      "vm_loss": 0.1783
    },
    {
      "epoch": 0.33163054742885445,
      "lm_loss": 0.3308,
      "step": 4520,
      "vm_loss": 0.157
    },
    {
      "epoch": 0.33170391701899354,
      "grad_norm": 1.5521014119662824,
      "learning_rate": 7.796840063759486e-06,
      "loss": 0.7654,
      "step": 4521
    },
    {
      "epoch": 0.3317772866091327,
      "grad_norm": 0.6578169532251195,
      "learning_rate": 7.795855065280268e-06,
      "loss": 0.5223,
      "step": 4522
    },
    {
      "epoch": 0.3318506561992718,
      "grad_norm": 0.6721054429726538,
      "learning_rate": 7.794869908912193e-06,
      "loss": 0.5342,
      "step": 4523
    },
    {
      "epoch": 0.33192402578941094,
      "grad_norm": 1.6415185480355934,
      "learning_rate": 7.793884594710892e-06,
      "loss": 0.7708,
      "step": 4524
    },
    {
      "epoch": 0.3319973953795501,
      "grad_norm": 1.5250463662813942,
      "learning_rate": 7.792899122732008e-06,
      "loss": 0.8612,
      "step": 4525
    },
    {
      "epoch": 0.3320707649696892,
      "grad_norm": 0.5990130326591995,
      "learning_rate": 7.791913493031195e-06,
      "loss": 0.5229,
      "step": 4526
    },
    {
      "epoch": 0.33214413455982833,
      "grad_norm": 1.3464134042179323,
      "learning_rate": 7.790927705664113e-06,
      "loss": 0.8176,
      "step": 4527
    },
    {
      "epoch": 0.3322175041499674,
      "grad_norm": 1.4601923341968233,
      "learning_rate": 7.789941760686431e-06,
      "loss": 0.7957,
      "step": 4528
    },
    {
      "epoch": 0.3322175041499674,
      "lm_loss": 1.7898,
      "step": 4528,
      "vm_loss": 0.0297
    },
    {
      "epoch": 0.3322175041499674,
      "lm_loss": 0.8446,
      "step": 4528,
      "vm_loss": 0.1295
    },
    {
      "epoch": 0.3322175041499674,
      "lm_loss": 1.1064,
      "step": 4528,
      "vm_loss": 0.1688
    },
    {
      "epoch": 0.3322175041499674,
      "lm_loss": 0.657,
      "step": 4528,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.3322175041499674,
      "lm_loss": 0.1259,
      "step": 4528,
      "vm_loss": 0.1077
    },
    {
      "epoch": 0.3322175041499674,
      "lm_loss": 0.4721,
      "step": 4528,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.3322175041499674,
      "lm_loss": 0.0355,
      "step": 4528,
      "vm_loss": 0.0457
    },
    {
      "epoch": 0.3322175041499674,
      "lm_loss": 0.6059,
      "step": 4528,
      "vm_loss": 0.137
    },
    {
      "epoch": 0.3322908737401066,
      "grad_norm": 1.3841450573292078,
      "learning_rate": 7.788955658153829e-06,
      "loss": 0.8286,
      "step": 4529
    },
    {
      "epoch": 0.3323642433302457,
      "grad_norm": 1.526955388537814,
      "learning_rate": 7.787969398121995e-06,
      "loss": 0.7958,
      "step": 4530
    },
    {
      "epoch": 0.3324376129203848,
      "grad_norm": 1.4960484752697556,
      "learning_rate": 7.786982980646624e-06,
      "loss": 0.726,
      "step": 4531
    },
    {
      "epoch": 0.33251098251052397,
      "grad_norm": 0.6876741321503081,
      "learning_rate": 7.785996405783423e-06,
      "loss": 0.5274,
      "step": 4532
    },
    {
      "epoch": 0.33258435210066306,
      "grad_norm": 1.687985955528997,
      "learning_rate": 7.785009673588105e-06,
      "loss": 0.8076,
      "step": 4533
    },
    {
      "epoch": 0.3326577216908022,
      "grad_norm": 0.6203090330189999,
      "learning_rate": 7.784022784116394e-06,
      "loss": 0.5098,
      "step": 4534
    },
    {
      "epoch": 0.3327310912809413,
      "grad_norm": 1.6537271659748471,
      "learning_rate": 7.783035737424022e-06,
      "loss": 0.8987,
      "step": 4535
    },
    {
      "epoch": 0.33280446087108045,
      "grad_norm": 1.7092083954482769,
      "learning_rate": 7.78204853356673e-06,
      "loss": 0.8002,
      "step": 4536
    },
    {
      "epoch": 0.33280446087108045,
      "lm_loss": 1.3829,
      "step": 4536,
      "vm_loss": 0.0492
    },
    {
      "epoch": 0.33280446087108045,
      "lm_loss": 0.3605,
      "step": 4536,
      "vm_loss": 0.1448
    },
    {
      "epoch": 0.33280446087108045,
      "lm_loss": 0.386,
      "step": 4536,
      "vm_loss": 0.0586
    },
    {
      "epoch": 0.33280446087108045,
      "lm_loss": 0.6268,
      "step": 4536,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.33280446087108045,
      "lm_loss": 0.0685,
      "step": 4536,
      "vm_loss": 0.1923
    },
    {
      "epoch": 0.33280446087108045,
      "lm_loss": 0.5326,
      "step": 4536,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.33280446087108045,
      "lm_loss": 0.3461,
      "step": 4536,
      "vm_loss": 0.0716
    },
    {
      "epoch": 0.33280446087108045,
      "lm_loss": 1.2122,
      "step": 4536,
      "vm_loss": 0.0689
    },
    {
      "epoch": 0.3328778304612196,
      "grad_norm": 1.4594148117400845,
      "learning_rate": 7.781061172600265e-06,
      "loss": 0.7581,
      "step": 4537
    },
    {
      "epoch": 0.3329512000513587,
      "grad_norm": 1.3976416891578671,
      "learning_rate": 7.780073654580392e-06,
      "loss": 0.7721,
      "step": 4538
    },
    {
      "epoch": 0.33302456964149785,
      "grad_norm": 1.433495470340721,
      "learning_rate": 7.779085979562874e-06,
      "loss": 0.7044,
      "step": 4539
    },
    {
      "epoch": 0.33309793923163694,
      "grad_norm": 1.4206044783576257,
      "learning_rate": 7.778098147603488e-06,
      "loss": 0.8539,
      "step": 4540
    },
    {
      "epoch": 0.3331713088217761,
      "grad_norm": 1.6868769360447804,
      "learning_rate": 7.77711015875802e-06,
      "loss": 0.9246,
      "step": 4541
    },
    {
      "epoch": 0.33324467841191524,
      "grad_norm": 1.6539352635947526,
      "learning_rate": 7.776122013082263e-06,
      "loss": 0.8681,
      "step": 4542
    },
    {
      "epoch": 0.33331804800205433,
      "grad_norm": 0.7003619826831768,
      "learning_rate": 7.775133710632023e-06,
      "loss": 0.5317,
      "step": 4543
    },
    {
      "epoch": 0.3333914175921935,
      "grad_norm": 0.6519460104737775,
      "learning_rate": 7.774145251463108e-06,
      "loss": 0.5284,
      "step": 4544
    },
    {
      "epoch": 0.3333914175921935,
      "lm_loss": 1.4382,
      "step": 4544,
      "vm_loss": 0.0595
    },
    {
      "epoch": 0.3333914175921935,
      "lm_loss": 0.4129,
      "step": 4544,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.3333914175921935,
      "lm_loss": 1.1773,
      "step": 4544,
      "vm_loss": 0.1477
    },
    {
      "epoch": 0.3333914175921935,
      "lm_loss": 0.0687,
      "step": 4544,
      "vm_loss": 0.2173
    },
    {
      "epoch": 0.3333914175921935,
      "lm_loss": 1.9185,
      "step": 4544,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.3333914175921935,
      "lm_loss": 0.7338,
      "step": 4544,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.3333914175921935,
      "lm_loss": 0.6873,
      "step": 4544,
      "vm_loss": 0.0768
    },
    {
      "epoch": 0.3333914175921935,
      "lm_loss": 0.081,
      "step": 4544,
      "vm_loss": 0.1289
    },
    {
      "epoch": 0.3334647871823326,
      "grad_norm": 1.7709610443743409,
      "learning_rate": 7.77315663563134e-06,
      "loss": 0.8466,
      "step": 4545
    },
    {
      "epoch": 0.33353815677247173,
      "grad_norm": 0.5650335524477176,
      "learning_rate": 7.772167863192552e-06,
      "loss": 0.524,
      "step": 4546
    },
    {
      "epoch": 0.3336115263626109,
      "grad_norm": 1.5048172236013748,
      "learning_rate": 7.771178934202578e-06,
      "loss": 0.8606,
      "step": 4547
    },
    {
      "epoch": 0.33368489595274997,
      "grad_norm": 1.6806848040475684,
      "learning_rate": 7.770189848717266e-06,
      "loss": 0.8761,
      "step": 4548
    },
    {
      "epoch": 0.3337582655428891,
      "grad_norm": 1.441870493590572,
      "learning_rate": 7.769200606792476e-06,
      "loss": 0.8333,
      "step": 4549
    },
    {
      "epoch": 0.3338316351330282,
      "grad_norm": 1.60153665149486,
      "learning_rate": 7.768211208484066e-06,
      "loss": 0.8043,
      "step": 4550
    },
    {
      "epoch": 0.33390500472316736,
      "grad_norm": 0.8237782090116865,
      "learning_rate": 7.767221653847917e-06,
      "loss": 0.5245,
      "step": 4551
    },
    {
      "epoch": 0.3339783743133065,
      "grad_norm": 1.5158225875744766,
      "learning_rate": 7.766231942939906e-06,
      "loss": 0.8021,
      "step": 4552
    },
    {
      "epoch": 0.3339783743133065,
      "lm_loss": 1.1326,
      "step": 4552,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.3339783743133065,
      "lm_loss": 0.4772,
      "step": 4552,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.3339783743133065,
      "lm_loss": 0.4424,
      "step": 4552,
      "vm_loss": 0.078
    },
    {
      "epoch": 0.3339783743133065,
      "lm_loss": 0.6238,
      "step": 4552,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.3339783743133065,
      "lm_loss": 0.4949,
      "step": 4552,
      "vm_loss": 0.0481
    },
    {
      "epoch": 0.3339783743133065,
      "lm_loss": 0.7061,
      "step": 4552,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.3339783743133065,
      "lm_loss": 0.2697,
      "step": 4552,
      "vm_loss": 0.0532
    },
    {
      "epoch": 0.3339783743133065,
      "lm_loss": 0.7358,
      "step": 4552,
      "vm_loss": 0.123
    },
    {
      "epoch": 0.3340517439034456,
      "grad_norm": 1.5545783240587268,
      "learning_rate": 7.765242075815928e-06,
      "loss": 0.763,
      "step": 4553
    },
    {
      "epoch": 0.33412511349358476,
      "grad_norm": 0.6554761017425109,
      "learning_rate": 7.76425205253188e-06,
      "loss": 0.5275,
      "step": 4554
    },
    {
      "epoch": 0.33419848308372385,
      "grad_norm": 1.3015271086351996,
      "learning_rate": 7.763261873143675e-06,
      "loss": 0.7852,
      "step": 4555
    },
    {
      "epoch": 0.334271852673863,
      "grad_norm": 1.4967044814468176,
      "learning_rate": 7.762271537707229e-06,
      "loss": 0.8571,
      "step": 4556
    },
    {
      "epoch": 0.33434522226400215,
      "grad_norm": 1.4632799899002176,
      "learning_rate": 7.761281046278468e-06,
      "loss": 0.7826,
      "step": 4557
    },
    {
      "epoch": 0.33441859185414125,
      "grad_norm": 1.2964717596128854,
      "learning_rate": 7.760290398913328e-06,
      "loss": 0.7238,
      "step": 4558
    },
    {
      "epoch": 0.3344919614442804,
      "grad_norm": 0.7975982426287597,
      "learning_rate": 7.759299595667752e-06,
      "loss": 0.517,
      "step": 4559
    },
    {
      "epoch": 0.3345653310344195,
      "grad_norm": 0.7317654342635894,
      "learning_rate": 7.758308636597698e-06,
      "loss": 0.5104,
      "step": 4560
    },
    {
      "epoch": 0.3345653310344195,
      "lm_loss": 1.2503,
      "step": 4560,
      "vm_loss": 0.0617
    },
    {
      "epoch": 0.3345653310344195,
      "lm_loss": 0.9354,
      "step": 4560,
      "vm_loss": 0.2333
    },
    {
      "epoch": 0.3345653310344195,
      "lm_loss": 1.0802,
      "step": 4560,
      "vm_loss": 0.1759
    },
    {
      "epoch": 0.3345653310344195,
      "lm_loss": 0.4188,
      "step": 4560,
      "vm_loss": 0.1166
    },
    {
      "epoch": 0.3345653310344195,
      "lm_loss": 0.8915,
      "step": 4560,
      "vm_loss": 0.1384
    },
    {
      "epoch": 0.3345653310344195,
      "lm_loss": 0.5614,
      "step": 4560,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.3345653310344195,
      "lm_loss": 0.7104,
      "step": 4560,
      "vm_loss": 0.2362
    },
    {
      "epoch": 0.3345653310344195,
      "lm_loss": 0.0652,
      "step": 4560,
      "vm_loss": 0.1736
    },
    {
      "epoch": 0.33463870062455864,
      "grad_norm": 1.4753282251457454,
      "learning_rate": 7.757317521759122e-06,
      "loss": 0.8095,
      "step": 4561
    },
    {
      "epoch": 0.3347120702146978,
      "grad_norm": 1.4384732059326129,
      "learning_rate": 7.756326251207998e-06,
      "loss": 0.9115,
      "step": 4562
    },
    {
      "epoch": 0.3347854398048369,
      "grad_norm": 1.5627746836409282,
      "learning_rate": 7.755334825000305e-06,
      "loss": 0.8526,
      "step": 4563
    },
    {
      "epoch": 0.33485880939497603,
      "grad_norm": 1.5788045541786804,
      "learning_rate": 7.75434324319203e-06,
      "loss": 0.8213,
      "step": 4564
    },
    {
      "epoch": 0.3349321789851151,
      "grad_norm": 1.5690626889004073,
      "learning_rate": 7.753351505839172e-06,
      "loss": 0.888,
      "step": 4565
    },
    {
      "epoch": 0.3350055485752543,
      "grad_norm": 1.2030929394673011,
      "learning_rate": 7.752359612997735e-06,
      "loss": 0.6476,
      "step": 4566
    },
    {
      "epoch": 0.3350789181653934,
      "grad_norm": 0.970775105186002,
      "learning_rate": 7.751367564723735e-06,
      "loss": 0.5272,
      "step": 4567
    },
    {
      "epoch": 0.3351522877555325,
      "grad_norm": 1.6541688605281901,
      "learning_rate": 7.750375361073194e-06,
      "loss": 0.8532,
      "step": 4568
    },
    {
      "epoch": 0.3351522877555325,
      "lm_loss": 2.0455,
      "step": 4568,
      "vm_loss": 0.1248
    },
    {
      "epoch": 0.3351522877555325,
      "lm_loss": 0.1997,
      "step": 4568,
      "vm_loss": 0.0549
    },
    {
      "epoch": 0.3351522877555325,
      "lm_loss": 0.8613,
      "step": 4568,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.3351522877555325,
      "lm_loss": 0.2889,
      "step": 4568,
      "vm_loss": 0.0884
    },
    {
      "epoch": 0.3351522877555325,
      "lm_loss": 0.9451,
      "step": 4568,
      "vm_loss": 0.2152
    },
    {
      "epoch": 0.3351522877555325,
      "lm_loss": 0.3225,
      "step": 4568,
      "vm_loss": 0.0539
    },
    {
      "epoch": 0.3351522877555325,
      "lm_loss": 0.8725,
      "step": 4568,
      "vm_loss": 0.1916
    },
    {
      "epoch": 0.3351522877555325,
      "lm_loss": 0.4193,
      "step": 4568,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.33522565734567167,
      "grad_norm": 1.653596815426783,
      "learning_rate": 7.749383002102147e-06,
      "loss": 0.8251,
      "step": 4569
    },
    {
      "epoch": 0.33529902693581076,
      "grad_norm": 0.620920975036808,
      "learning_rate": 7.748390487866632e-06,
      "loss": 0.5282,
      "step": 4570
    },
    {
      "epoch": 0.3353723965259499,
      "grad_norm": 1.6720792832189406,
      "learning_rate": 7.7473978184227e-06,
      "loss": 0.7687,
      "step": 4571
    },
    {
      "epoch": 0.33544576611608906,
      "grad_norm": 1.5800610888879008,
      "learning_rate": 7.746404993826408e-06,
      "loss": 0.7421,
      "step": 4572
    },
    {
      "epoch": 0.33551913570622816,
      "grad_norm": 0.6019299259391498,
      "learning_rate": 7.745412014133827e-06,
      "loss": 0.5273,
      "step": 4573
    },
    {
      "epoch": 0.3355925052963673,
      "grad_norm": 0.634022430325258,
      "learning_rate": 7.74441887940103e-06,
      "loss": 0.5438,
      "step": 4574
    },
    {
      "epoch": 0.3356658748865064,
      "grad_norm": 1.4247518616799806,
      "learning_rate": 7.743425589684102e-06,
      "loss": 0.8297,
      "step": 4575
    },
    {
      "epoch": 0.33573924447664555,
      "grad_norm": 1.4102573710485748,
      "learning_rate": 7.742432145039139e-06,
      "loss": 0.8046,
      "step": 4576
    },
    {
      "epoch": 0.33573924447664555,
      "lm_loss": 1.1766,
      "step": 4576,
      "vm_loss": 0.1045
    },
    {
      "epoch": 0.33573924447664555,
      "lm_loss": 0.3055,
      "step": 4576,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.33573924447664555,
      "lm_loss": 1.1279,
      "step": 4576,
      "vm_loss": 0.2069
    },
    {
      "epoch": 0.33573924447664555,
      "lm_loss": 0.8401,
      "step": 4576,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.33573924447664555,
      "lm_loss": 1.0195,
      "step": 4576,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.33573924447664555,
      "lm_loss": 0.6771,
      "step": 4576,
      "vm_loss": 0.1328
    },
    {
      "epoch": 0.33573924447664555,
      "lm_loss": 0.5215,
      "step": 4576,
      "vm_loss": 0.0699
    },
    {
      "epoch": 0.33573924447664555,
      "lm_loss": 0.5707,
      "step": 4576,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.33581261406678464,
      "grad_norm": 1.4897799288785705,
      "learning_rate": 7.741438545522238e-06,
      "loss": 0.7919,
      "step": 4577
    },
    {
      "epoch": 0.3358859836569238,
      "grad_norm": 1.4369602627760605,
      "learning_rate": 7.740444791189515e-06,
      "loss": 0.8584,
      "step": 4578
    },
    {
      "epoch": 0.33595935324706294,
      "grad_norm": 1.6904277992448786,
      "learning_rate": 7.739450882097088e-06,
      "loss": 0.8963,
      "step": 4579
    },
    {
      "epoch": 0.33603272283720204,
      "grad_norm": 1.4651737086884995,
      "learning_rate": 7.738456818301085e-06,
      "loss": 0.8361,
      "step": 4580
    },
    {
      "epoch": 0.3361060924273412,
      "grad_norm": 1.3468413379452233,
      "learning_rate": 7.737462599857643e-06,
      "loss": 0.8987,
      "step": 4581
    },
    {
      "epoch": 0.3361794620174803,
      "grad_norm": 1.5530024450636726,
      "learning_rate": 7.736468226822911e-06,
      "loss": 0.8476,
      "step": 4582
    },
    {
      "epoch": 0.33625283160761943,
      "grad_norm": 1.8115531953128299,
      "learning_rate": 7.73547369925304e-06,
      "loss": 0.8174,
      "step": 4583
    },
    {
      "epoch": 0.3363262011977586,
      "grad_norm": 0.6846759777216311,
      "learning_rate": 7.734479017204196e-06,
      "loss": 0.5431,
      "step": 4584
    },
    {
      "epoch": 0.3363262011977586,
      "lm_loss": 1.3255,
      "step": 4584,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.3363262011977586,
      "lm_loss": 1.5658,
      "step": 4584,
      "vm_loss": 0.0638
    },
    {
      "epoch": 0.3363262011977586,
      "lm_loss": 0.1158,
      "step": 4584,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.3363262011977586,
      "lm_loss": 0.0649,
      "step": 4584,
      "vm_loss": 0.1576
    },
    {
      "epoch": 0.3363262011977586,
      "lm_loss": 0.3958,
      "step": 4584,
      "vm_loss": 0.2319
    },
    {
      "epoch": 0.3363262011977586,
      "lm_loss": 0.58,
      "step": 4584,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.3363262011977586,
      "lm_loss": 1.1975,
      "step": 4584,
      "vm_loss": 0.2301
    },
    {
      "epoch": 0.3363262011977586,
      "lm_loss": 1.1298,
      "step": 4584,
      "vm_loss": 0.1765
    },
    {
      "epoch": 0.3363995707878977,
      "grad_norm": 1.529510479045709,
      "learning_rate": 7.73348418073255e-06,
      "loss": 0.9881,
      "step": 4585
    },
    {
      "epoch": 0.3364729403780368,
      "grad_norm": 1.5749141215313558,
      "learning_rate": 7.732489189894281e-06,
      "loss": 0.804,
      "step": 4586
    },
    {
      "epoch": 0.3365463099681759,
      "grad_norm": 0.5773702585884183,
      "learning_rate": 7.731494044745582e-06,
      "loss": 0.5063,
      "step": 4587
    },
    {
      "epoch": 0.33661967955831507,
      "grad_norm": 1.4251274484800913,
      "learning_rate": 7.73049874534265e-06,
      "loss": 0.834,
      "step": 4588
    },
    {
      "epoch": 0.3366930491484542,
      "grad_norm": 1.7128302438065506,
      "learning_rate": 7.72950329174169e-06,
      "loss": 0.8367,
      "step": 4589
    },
    {
      "epoch": 0.3367664187385933,
      "grad_norm": 0.587000602054969,
      "learning_rate": 7.728507683998922e-06,
      "loss": 0.5258,
      "step": 4590
    },
    {
      "epoch": 0.33683978832873246,
      "grad_norm": 1.3151213504019423,
      "learning_rate": 7.727511922170567e-06,
      "loss": 0.802,
      "step": 4591
    },
    {
      "epoch": 0.33691315791887155,
      "grad_norm": 1.5710829107973818,
      "learning_rate": 7.72651600631286e-06,
      "loss": 0.8344,
      "step": 4592
    },
    {
      "epoch": 0.33691315791887155,
      "lm_loss": 0.667,
      "step": 4592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33691315791887155,
      "lm_loss": 0.4881,
      "step": 4592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33691315791887155,
      "lm_loss": 0.5719,
      "step": 4592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33691315791887155,
      "lm_loss": 0.6362,
      "step": 4592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33691315791887155,
      "lm_loss": 0.4683,
      "step": 4592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33691315791887155,
      "lm_loss": 0.4814,
      "step": 4592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33691315791887155,
      "lm_loss": 0.6162,
      "step": 4592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33691315791887155,
      "lm_loss": 0.5759,
      "step": 4592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3369865275090107,
      "grad_norm": 0.5766515987540399,
      "learning_rate": 7.725519936482042e-06,
      "loss": 0.5182,
      "step": 4593
    },
    {
      "epoch": 0.33705989709914985,
      "grad_norm": 0.5669096452220254,
      "learning_rate": 7.724523712734362e-06,
      "loss": 0.5146,
      "step": 4594
    },
    {
      "epoch": 0.33713326668928895,
      "grad_norm": 0.5675592632190496,
      "learning_rate": 7.723527335126082e-06,
      "loss": 0.5183,
      "step": 4595
    },
    {
      "epoch": 0.3372066362794281,
      "grad_norm": 0.5569564682841144,
      "learning_rate": 7.722530803713468e-06,
      "loss": 0.5476,
      "step": 4596
    },
    {
      "epoch": 0.3372800058695672,
      "grad_norm": 0.5720721707131768,
      "learning_rate": 7.7215341185528e-06,
      "loss": 0.5235,
      "step": 4597
    },
    {
      "epoch": 0.33735337545970634,
      "grad_norm": 2.065865926297886,
      "learning_rate": 7.720537279700357e-06,
      "loss": 0.8437,
      "step": 4598
    },
    {
      "epoch": 0.3374267450498455,
      "grad_norm": 1.6385074152716765,
      "learning_rate": 7.719540287212439e-06,
      "loss": 0.8483,
      "step": 4599
    },
    {
      "epoch": 0.3375001146399846,
      "grad_norm": 1.8049818053505264,
      "learning_rate": 7.718543141145344e-06,
      "loss": 0.8922,
      "step": 4600
    },
    {
      "epoch": 0.3375001146399846,
      "lm_loss": 0.6192,
      "step": 4600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3375001146399846,
      "lm_loss": 0.5199,
      "step": 4600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3375001146399846,
      "lm_loss": 0.5766,
      "step": 4600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3375001146399846,
      "lm_loss": 0.5179,
      "step": 4600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3375001146399846,
      "lm_loss": 0.5217,
      "step": 4600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3375001146399846,
      "lm_loss": 0.5866,
      "step": 4600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3375001146399846,
      "lm_loss": 0.3739,
      "step": 4600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3375001146399846,
      "lm_loss": 0.5386,
      "step": 4600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33757348423012373,
      "grad_norm": 0.6084627231269933,
      "learning_rate": 7.717545841555387e-06,
      "loss": 0.5317,
      "step": 4601
    },
    {
      "epoch": 0.3376468538202628,
      "grad_norm": 0.6260786996236114,
      "learning_rate": 7.716548388498886e-06,
      "loss": 0.5141,
      "step": 4602
    },
    {
      "epoch": 0.337720223410402,
      "grad_norm": 1.6922926370741216,
      "learning_rate": 7.71555078203217e-06,
      "loss": 0.7886,
      "step": 4603
    },
    {
      "epoch": 0.3377935930005411,
      "grad_norm": 1.4036331264481976,
      "learning_rate": 7.714553022211574e-06,
      "loss": 0.7635,
      "step": 4604
    },
    {
      "epoch": 0.3378669625906802,
      "grad_norm": 1.5187644251528547,
      "learning_rate": 7.713555109093451e-06,
      "loss": 0.8557,
      "step": 4605
    },
    {
      "epoch": 0.33794033218081937,
      "grad_norm": 0.5674034746288972,
      "learning_rate": 7.712557042734147e-06,
      "loss": 0.5422,
      "step": 4606
    },
    {
      "epoch": 0.33801370177095846,
      "grad_norm": 0.5617865947318837,
      "learning_rate": 7.711558823190031e-06,
      "loss": 0.5128,
      "step": 4607
    },
    {
      "epoch": 0.3380870713610976,
      "grad_norm": 1.7300496535608307,
      "learning_rate": 7.710560450517474e-06,
      "loss": 0.8954,
      "step": 4608
    },
    {
      "epoch": 0.3380870713610976,
      "lm_loss": 0.3597,
      "step": 4608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3380870713610976,
      "lm_loss": 0.365,
      "step": 4608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3380870713610976,
      "lm_loss": 0.5799,
      "step": 4608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3380870713610976,
      "lm_loss": 0.7523,
      "step": 4608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3380870713610976,
      "lm_loss": 0.4041,
      "step": 4608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3380870713610976,
      "lm_loss": 0.4492,
      "step": 4608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3380870713610976,
      "lm_loss": 0.4327,
      "step": 4608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3380870713610976,
      "lm_loss": 0.5993,
      "step": 4608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33816044095123676,
      "grad_norm": 0.5710529880972901,
      "learning_rate": 7.709561924772855e-06,
      "loss": 0.5114,
      "step": 4609
    },
    {
      "epoch": 0.33823381054137586,
      "grad_norm": 1.435202686123152,
      "learning_rate": 7.708563246012564e-06,
      "loss": 0.7567,
      "step": 4610
    },
    {
      "epoch": 0.338307180131515,
      "grad_norm": 0.5719060119131548,
      "learning_rate": 7.707564414292998e-06,
      "loss": 0.5235,
      "step": 4611
    },
    {
      "epoch": 0.3383805497216541,
      "grad_norm": 1.6960645336801063,
      "learning_rate": 7.706565429670565e-06,
      "loss": 0.9383,
      "step": 4612
    },
    {
      "epoch": 0.33845391931179325,
      "grad_norm": 1.7428671800516178,
      "learning_rate": 7.70556629220168e-06,
      "loss": 0.7835,
      "step": 4613
    },
    {
      "epoch": 0.33852728890193234,
      "grad_norm": 1.5585435780638366,
      "learning_rate": 7.704567001942767e-06,
      "loss": 0.7479,
      "step": 4614
    },
    {
      "epoch": 0.3386006584920715,
      "grad_norm": 1.566101835314829,
      "learning_rate": 7.703567558950255e-06,
      "loss": 0.7725,
      "step": 4615
    },
    {
      "epoch": 0.33867402808221064,
      "grad_norm": 1.5793316910551927,
      "learning_rate": 7.702567963280589e-06,
      "loss": 0.8051,
      "step": 4616
    },
    {
      "epoch": 0.33867402808221064,
      "lm_loss": 0.6896,
      "step": 4616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33867402808221064,
      "lm_loss": 0.4111,
      "step": 4616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33867402808221064,
      "lm_loss": 0.554,
      "step": 4616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33867402808221064,
      "lm_loss": 0.4662,
      "step": 4616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33867402808221064,
      "lm_loss": 0.5308,
      "step": 4616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33867402808221064,
      "lm_loss": 0.4472,
      "step": 4616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33867402808221064,
      "lm_loss": 0.454,
      "step": 4616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33867402808221064,
      "lm_loss": 0.5958,
      "step": 4616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.33874739767234974,
      "grad_norm": 0.6491032990929005,
      "learning_rate": 7.701568214990218e-06,
      "loss": 0.5319,
      "step": 4617
    },
    {
      "epoch": 0.3388207672624889,
      "grad_norm": 0.5650211691451812,
      "learning_rate": 7.700568314135599e-06,
      "loss": 0.515,
      "step": 4618
    },
    {
      "epoch": 0.338894136852628,
      "grad_norm": 0.5745676761212036,
      "learning_rate": 7.6995682607732e-06,
      "loss": 0.5163,
      "step": 4619
    },
    {
      "epoch": 0.33896750644276713,
      "grad_norm": 1.6009665450039814,
      "learning_rate": 7.698568054959496e-06,
      "loss": 0.8416,
      "step": 4620
    },
    {
      "epoch": 0.3390408760329063,
      "grad_norm": 1.5474368575026167,
      "learning_rate": 7.69756769675097e-06,
      "loss": 0.7842,
      "step": 4621
    },
    {
      "epoch": 0.3391142456230454,
      "grad_norm": 0.6126058608892194,
      "learning_rate": 7.696567186204117e-06,
      "loss": 0.5159,
      "step": 4622
    },
    {
      "epoch": 0.3391876152131845,
      "grad_norm": 1.6363951842390467,
      "learning_rate": 7.695566523375436e-06,
      "loss": 0.7822,
      "step": 4623
    },
    {
      "epoch": 0.3392609848033236,
      "grad_norm": 1.6067433863001737,
      "learning_rate": 7.694565708321438e-06,
      "loss": 0.8094,
      "step": 4624
    },
    {
      "epoch": 0.3392609848033236,
      "lm_loss": 1.2169,
      "step": 4624,
      "vm_loss": 0.0884
    },
    {
      "epoch": 0.3392609848033236,
      "lm_loss": 0.5966,
      "step": 4624,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.3392609848033236,
      "lm_loss": 1.0219,
      "step": 4624,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.3392609848033236,
      "lm_loss": 0.5553,
      "step": 4624,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.3392609848033236,
      "lm_loss": 0.2458,
      "step": 4624,
      "vm_loss": 0.0651
    },
    {
      "epoch": 0.3392609848033236,
      "lm_loss": 0.9346,
      "step": 4624,
      "vm_loss": 0.1354
    },
    {
      "epoch": 0.3392609848033236,
      "lm_loss": 0.1568,
      "step": 4624,
      "vm_loss": 0.0567
    },
    {
      "epoch": 0.3392609848033236,
      "lm_loss": 1.2001,
      "step": 4624,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.33933435439346277,
      "grad_norm": 1.5094639660768125,
      "learning_rate": 7.693564741098643e-06,
      "loss": 0.7547,
      "step": 4625
    },
    {
      "epoch": 0.3394077239836019,
      "grad_norm": 0.6211110380665178,
      "learning_rate": 7.692563621763575e-06,
      "loss": 0.52,
      "step": 4626
    },
    {
      "epoch": 0.339481093573741,
      "grad_norm": 1.6137231309526383,
      "learning_rate": 7.69156235037277e-06,
      "loss": 0.7545,
      "step": 4627
    },
    {
      "epoch": 0.33955446316388016,
      "grad_norm": 0.5822864660719865,
      "learning_rate": 7.690560926982776e-06,
      "loss": 0.5238,
      "step": 4628
    },
    {
      "epoch": 0.33962783275401925,
      "grad_norm": 1.5237002125406887,
      "learning_rate": 7.689559351650142e-06,
      "loss": 0.8771,
      "step": 4629
    },
    {
      "epoch": 0.3397012023441584,
      "grad_norm": 1.4491534030805429,
      "learning_rate": 7.688557624431432e-06,
      "loss": 0.7665,
      "step": 4630
    },
    {
      "epoch": 0.33977457193429755,
      "grad_norm": 0.5978506067594924,
      "learning_rate": 7.687555745383214e-06,
      "loss": 0.5385,
      "step": 4631
    },
    {
      "epoch": 0.33984794152443665,
      "grad_norm": 1.6270881413605307,
      "learning_rate": 7.686553714562068e-06,
      "loss": 0.8321,
      "step": 4632
    },
    {
      "epoch": 0.33984794152443665,
      "lm_loss": 1.5435,
      "step": 4632,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.33984794152443665,
      "lm_loss": 1.3617,
      "step": 4632,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.33984794152443665,
      "lm_loss": 0.1204,
      "step": 4632,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.33984794152443665,
      "lm_loss": 0.4211,
      "step": 4632,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.33984794152443665,
      "lm_loss": 0.4523,
      "step": 4632,
      "vm_loss": 0.1321
    },
    {
      "epoch": 0.33984794152443665,
      "lm_loss": 1.3502,
      "step": 4632,
      "vm_loss": 0.1396
    },
    {
      "epoch": 0.33984794152443665,
      "lm_loss": 0.0947,
      "step": 4632,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.33984794152443665,
      "lm_loss": 0.827,
      "step": 4632,
      "vm_loss": 0.2656
    },
    {
      "epoch": 0.3399213111145758,
      "grad_norm": 1.598868903714505,
      "learning_rate": 7.68555153202458e-06,
      "loss": 0.7922,
      "step": 4633
    },
    {
      "epoch": 0.3399946807047149,
      "grad_norm": 1.4859789755434651,
      "learning_rate": 7.684549197827347e-06,
      "loss": 0.7841,
      "step": 4634
    },
    {
      "epoch": 0.34006805029485404,
      "grad_norm": 0.5922853066540102,
      "learning_rate": 7.68354671202697e-06,
      "loss": 0.5099,
      "step": 4635
    },
    {
      "epoch": 0.3401414198849932,
      "grad_norm": 0.5820427992046322,
      "learning_rate": 7.682544074680068e-06,
      "loss": 0.5267,
      "step": 4636
    },
    {
      "epoch": 0.3402147894751323,
      "grad_norm": 0.5531477710827444,
      "learning_rate": 7.681541285843257e-06,
      "loss": 0.5273,
      "step": 4637
    },
    {
      "epoch": 0.34028815906527143,
      "grad_norm": 1.7442778490925541,
      "learning_rate": 7.68053834557317e-06,
      "loss": 0.8215,
      "step": 4638
    },
    {
      "epoch": 0.34036152865541053,
      "grad_norm": 1.5264431313752729,
      "learning_rate": 7.679535253926445e-06,
      "loss": 0.7877,
      "step": 4639
    },
    {
      "epoch": 0.3404348982455497,
      "grad_norm": 1.4954317502667223,
      "learning_rate": 7.678532010959726e-06,
      "loss": 0.8656,
      "step": 4640
    },
    {
      "epoch": 0.3404348982455497,
      "lm_loss": 0.965,
      "step": 4640,
      "vm_loss": 0.0525
    },
    {
      "epoch": 0.3404348982455497,
      "lm_loss": 0.0831,
      "step": 4640,
      "vm_loss": 0.073
    },
    {
      "epoch": 0.3404348982455497,
      "lm_loss": 0.4149,
      "step": 4640,
      "vm_loss": 0.1856
    },
    {
      "epoch": 0.3404348982455497,
      "lm_loss": 0.3225,
      "step": 4640,
      "vm_loss": 0.0775
    },
    {
      "epoch": 0.3404348982455497,
      "lm_loss": 0.3249,
      "step": 4640,
      "vm_loss": 0.0413
    },
    {
      "epoch": 0.3404348982455497,
      "lm_loss": 0.7685,
      "step": 4640,
      "vm_loss": 0.1094
    },
    {
      "epoch": 0.3404348982455497,
      "lm_loss": 0.8029,
      "step": 4640,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.3404348982455497,
      "lm_loss": 0.9418,
      "step": 4640,
      "vm_loss": 0.2033
    },
    {
      "epoch": 0.3405082678356888,
      "grad_norm": 1.6210029101348247,
      "learning_rate": 7.677528616729673e-06,
      "loss": 0.8978,
      "step": 4641
    },
    {
      "epoch": 0.3405816374258279,
      "grad_norm": 0.6407529976340058,
      "learning_rate": 7.676525071292949e-06,
      "loss": 0.5127,
      "step": 4642
    },
    {
      "epoch": 0.34065500701596707,
      "grad_norm": 0.6391658354380039,
      "learning_rate": 7.675521374706222e-06,
      "loss": 0.5073,
      "step": 4643
    },
    {
      "epoch": 0.34072837660610616,
      "grad_norm": 1.8820380208134933,
      "learning_rate": 7.67451752702618e-06,
      "loss": 0.814,
      "step": 4644
    },
    {
      "epoch": 0.3408017461962453,
      "grad_norm": 1.4369115199333862,
      "learning_rate": 7.67351352830951e-06,
      "loss": 0.8058,
      "step": 4645
    },
    {
      "epoch": 0.34087511578638446,
      "grad_norm": 0.5806350428701329,
      "learning_rate": 7.67250937861291e-06,
      "loss": 0.5158,
      "step": 4646
    },
    {
      "epoch": 0.34094848537652356,
      "grad_norm": 1.466470677494857,
      "learning_rate": 7.671505077993086e-06,
      "loss": 0.8279,
      "step": 4647
    },
    {
      "epoch": 0.3410218549666627,
      "grad_norm": 1.471766145470709,
      "learning_rate": 7.670500626506756e-06,
      "loss": 0.7926,
      "step": 4648
    },
    {
      "epoch": 0.3410218549666627,
      "lm_loss": 1.3768,
      "step": 4648,
      "vm_loss": 0.0482
    },
    {
      "epoch": 0.3410218549666627,
      "lm_loss": 0.1906,
      "step": 4648,
      "vm_loss": 0.0869
    },
    {
      "epoch": 0.3410218549666627,
      "lm_loss": 1.0305,
      "step": 4648,
      "vm_loss": 0.2042
    },
    {
      "epoch": 0.3410218549666627,
      "lm_loss": 0.9052,
      "step": 4648,
      "vm_loss": 0.144
    },
    {
      "epoch": 0.3410218549666627,
      "lm_loss": 1.0324,
      "step": 4648,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.3410218549666627,
      "lm_loss": 0.8831,
      "step": 4648,
      "vm_loss": 0.2108
    },
    {
      "epoch": 0.3410218549666627,
      "lm_loss": 0.7974,
      "step": 4648,
      "vm_loss": 0.0445
    },
    {
      "epoch": 0.3410218549666627,
      "lm_loss": 0.1801,
      "step": 4648,
      "vm_loss": 0.0582
    },
    {
      "epoch": 0.3410952245568018,
      "grad_norm": 1.6112126898523995,
      "learning_rate": 7.66949602421064e-06,
      "loss": 0.8719,
      "step": 4649
    },
    {
      "epoch": 0.34116859414694095,
      "grad_norm": 0.6294195533082464,
      "learning_rate": 7.668491271161476e-06,
      "loss": 0.5499,
      "step": 4650
    },
    {
      "epoch": 0.3412419637370801,
      "grad_norm": 1.4710885884406317,
      "learning_rate": 7.667486367416e-06,
      "loss": 0.7646,
      "step": 4651
    },
    {
      "epoch": 0.3413153333272192,
      "grad_norm": 1.336563090427456,
      "learning_rate": 7.666481313030963e-06,
      "loss": 0.781,
      "step": 4652
    },
    {
      "epoch": 0.34138870291735834,
      "grad_norm": 1.516858077355268,
      "learning_rate": 7.66547610806312e-06,
      "loss": 0.7981,
      "step": 4653
    },
    {
      "epoch": 0.34146207250749744,
      "grad_norm": 1.5373104424156612,
      "learning_rate": 7.664470752569245e-06,
      "loss": 0.9424,
      "step": 4654
    },
    {
      "epoch": 0.3415354420976366,
      "grad_norm": 1.6987493853437299,
      "learning_rate": 7.663465246606105e-06,
      "loss": 0.9058,
      "step": 4655
    },
    {
      "epoch": 0.3416088116877757,
      "grad_norm": 0.5709231015884701,
      "learning_rate": 7.662459590230487e-06,
      "loss": 0.5048,
      "step": 4656
    },
    {
      "epoch": 0.3416088116877757,
      "lm_loss": 0.5799,
      "step": 4656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3416088116877757,
      "lm_loss": 0.4362,
      "step": 4656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3416088116877757,
      "lm_loss": 0.3872,
      "step": 4656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3416088116877757,
      "lm_loss": 0.4116,
      "step": 4656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3416088116877757,
      "lm_loss": 0.5601,
      "step": 4656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3416088116877757,
      "lm_loss": 0.5551,
      "step": 4656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3416088116877757,
      "lm_loss": 0.5289,
      "step": 4656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3416088116877757,
      "lm_loss": 0.4942,
      "step": 4656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34168218127791483,
      "grad_norm": 0.5731178023541091,
      "learning_rate": 7.661453783499182e-06,
      "loss": 0.5286,
      "step": 4657
    },
    {
      "epoch": 0.341755550868054,
      "grad_norm": 1.6437405813343904,
      "learning_rate": 7.660447826468992e-06,
      "loss": 0.8249,
      "step": 4658
    },
    {
      "epoch": 0.3418289204581931,
      "grad_norm": 0.573836930635808,
      "learning_rate": 7.659441719196724e-06,
      "loss": 0.536,
      "step": 4659
    },
    {
      "epoch": 0.3419022900483322,
      "grad_norm": 1.3676782478843672,
      "learning_rate": 7.658435461739198e-06,
      "loss": 0.8787,
      "step": 4660
    },
    {
      "epoch": 0.3419756596384713,
      "grad_norm": 0.5696528186959572,
      "learning_rate": 7.657429054153235e-06,
      "loss": 0.5239,
      "step": 4661
    },
    {
      "epoch": 0.34204902922861047,
      "grad_norm": 1.5843847109185476,
      "learning_rate": 7.656422496495674e-06,
      "loss": 0.8105,
      "step": 4662
    },
    {
      "epoch": 0.3421223988187496,
      "grad_norm": 1.66574352808267,
      "learning_rate": 7.655415788823356e-06,
      "loss": 0.8999,
      "step": 4663
    },
    {
      "epoch": 0.3421957684088887,
      "grad_norm": 1.612109391113099,
      "learning_rate": 7.65440893119313e-06,
      "loss": 0.7863,
      "step": 4664
    },
    {
      "epoch": 0.3421957684088887,
      "lm_loss": 1.1413,
      "step": 4664,
      "vm_loss": 0.0502
    },
    {
      "epoch": 0.3421957684088887,
      "lm_loss": 0.1937,
      "step": 4664,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.3421957684088887,
      "lm_loss": 0.8996,
      "step": 4664,
      "vm_loss": 0.1458
    },
    {
      "epoch": 0.3421957684088887,
      "lm_loss": 0.3221,
      "step": 4664,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.3421957684088887,
      "lm_loss": 0.3622,
      "step": 4664,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.3421957684088887,
      "lm_loss": 0.3575,
      "step": 4664,
      "vm_loss": 0.1411
    },
    {
      "epoch": 0.3421957684088887,
      "lm_loss": 0.3994,
      "step": 4664,
      "vm_loss": 0.2081
    },
    {
      "epoch": 0.3421957684088887,
      "lm_loss": 0.6675,
      "step": 4664,
      "vm_loss": 0.1977
    },
    {
      "epoch": 0.34226913799902786,
      "grad_norm": 1.7240999242394766,
      "learning_rate": 7.653401923661863e-06,
      "loss": 0.8195,
      "step": 4665
    },
    {
      "epoch": 0.34234250758916696,
      "grad_norm": 1.7419542247787414,
      "learning_rate": 7.652394766286415e-06,
      "loss": 0.7775,
      "step": 4666
    },
    {
      "epoch": 0.3424158771793061,
      "grad_norm": 1.4720362644648493,
      "learning_rate": 7.651387459123667e-06,
      "loss": 0.8876,
      "step": 4667
    },
    {
      "epoch": 0.34248924676944525,
      "grad_norm": 1.5235563366582328,
      "learning_rate": 7.650380002230503e-06,
      "loss": 0.7728,
      "step": 4668
    },
    {
      "epoch": 0.34256261635958435,
      "grad_norm": 0.6377320063478472,
      "learning_rate": 7.649372395663816e-06,
      "loss": 0.5247,
      "step": 4669
    },
    {
      "epoch": 0.3426359859497235,
      "grad_norm": 1.6870186436771757,
      "learning_rate": 7.64836463948051e-06,
      "loss": 0.9091,
      "step": 4670
    },
    {
      "epoch": 0.3427093555398626,
      "grad_norm": 1.6600792816674594,
      "learning_rate": 7.647356733737493e-06,
      "loss": 0.971,
      "step": 4671
    },
    {
      "epoch": 0.34278272513000174,
      "grad_norm": 1.419757135210559,
      "learning_rate": 7.646348678491685e-06,
      "loss": 0.7359,
      "step": 4672
    },
    {
      "epoch": 0.34278272513000174,
      "lm_loss": 1.536,
      "step": 4672,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.34278272513000174,
      "lm_loss": 0.1386,
      "step": 4672,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.34278272513000174,
      "lm_loss": 1.0033,
      "step": 4672,
      "vm_loss": 0.044
    },
    {
      "epoch": 0.34278272513000174,
      "lm_loss": 0.376,
      "step": 4672,
      "vm_loss": 0.068
    },
    {
      "epoch": 0.34278272513000174,
      "lm_loss": 0.2299,
      "step": 4672,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.34278272513000174,
      "lm_loss": 0.6169,
      "step": 4672,
      "vm_loss": 0.2021
    },
    {
      "epoch": 0.34278272513000174,
      "lm_loss": 0.4006,
      "step": 4672,
      "vm_loss": 0.0958
    },
    {
      "epoch": 0.34278272513000174,
      "lm_loss": 0.5287,
      "step": 4672,
      "vm_loss": 0.0455
    },
    {
      "epoch": 0.3428560947201409,
      "grad_norm": 1.5610750264638624,
      "learning_rate": 7.645340473800014e-06,
      "loss": 0.7379,
      "step": 4673
    },
    {
      "epoch": 0.34292946431028,
      "grad_norm": 1.676076806397433,
      "learning_rate": 7.644332119719418e-06,
      "loss": 0.8302,
      "step": 4674
    },
    {
      "epoch": 0.34300283390041914,
      "grad_norm": 1.448126179792965,
      "learning_rate": 7.643323616306835e-06,
      "loss": 0.859,
      "step": 4675
    },
    {
      "epoch": 0.34307620349055823,
      "grad_norm": 1.4375401453978993,
      "learning_rate": 7.642314963619222e-06,
      "loss": 0.7719,
      "step": 4676
    },
    {
      "epoch": 0.3431495730806974,
      "grad_norm": 1.4874353365509556,
      "learning_rate": 7.64130616171354e-06,
      "loss": 0.8276,
      "step": 4677
    },
    {
      "epoch": 0.34322294267083653,
      "grad_norm": 1.592629368935548,
      "learning_rate": 7.640297210646759e-06,
      "loss": 0.8128,
      "step": 4678
    },
    {
      "epoch": 0.3432963122609756,
      "grad_norm": 0.6591252553998278,
      "learning_rate": 7.639288110475855e-06,
      "loss": 0.514,
      "step": 4679
    },
    {
      "epoch": 0.34336968185111477,
      "grad_norm": 0.6867082872620321,
      "learning_rate": 7.638278861257814e-06,
      "loss": 0.5181,
      "step": 4680
    },
    {
      "epoch": 0.34336968185111477,
      "lm_loss": 0.6069,
      "step": 4680,
      "vm_loss": 0.0631
    },
    {
      "epoch": 0.34336968185111477,
      "lm_loss": 1.3043,
      "step": 4680,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.34336968185111477,
      "lm_loss": 0.2886,
      "step": 4680,
      "vm_loss": 0.0665
    },
    {
      "epoch": 0.34336968185111477,
      "lm_loss": 0.9587,
      "step": 4680,
      "vm_loss": 0.1464
    },
    {
      "epoch": 0.34336968185111477,
      "lm_loss": 0.7995,
      "step": 4680,
      "vm_loss": 0.148
    },
    {
      "epoch": 0.34336968185111477,
      "lm_loss": 1.0414,
      "step": 4680,
      "vm_loss": 0.1278
    },
    {
      "epoch": 0.34336968185111477,
      "lm_loss": 0.075,
      "step": 4680,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.34336968185111477,
      "lm_loss": 0.1198,
      "step": 4680,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.34344305144125387,
      "grad_norm": 1.6377515473425066,
      "learning_rate": 7.637269463049633e-06,
      "loss": 0.8156,
      "step": 4681
    },
    {
      "epoch": 0.343516421031393,
      "grad_norm": 0.5985497790597176,
      "learning_rate": 7.636259915908315e-06,
      "loss": 0.5268,
      "step": 4682
    },
    {
      "epoch": 0.34358979062153217,
      "grad_norm": 1.5980818783008706,
      "learning_rate": 7.63525021989087e-06,
      "loss": 0.7963,
      "step": 4683
    },
    {
      "epoch": 0.34366316021167126,
      "grad_norm": 1.6603251165219617,
      "learning_rate": 7.63424037505432e-06,
      "loss": 0.9166,
      "step": 4684
    },
    {
      "epoch": 0.3437365298018104,
      "grad_norm": 1.5033048782607448,
      "learning_rate": 7.633230381455691e-06,
      "loss": 0.7998,
      "step": 4685
    },
    {
      "epoch": 0.3438098993919495,
      "grad_norm": 1.4330585163541552,
      "learning_rate": 7.632220239152024e-06,
      "loss": 0.7655,
      "step": 4686
    },
    {
      "epoch": 0.34388326898208865,
      "grad_norm": 1.602810653870274,
      "learning_rate": 7.63120994820036e-06,
      "loss": 0.8444,
      "step": 4687
    },
    {
      "epoch": 0.3439566385722278,
      "grad_norm": 1.8097901380768835,
      "learning_rate": 7.630199508657753e-06,
      "loss": 0.7901,
      "step": 4688
    },
    {
      "epoch": 0.3439566385722278,
      "lm_loss": 0.4818,
      "step": 4688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3439566385722278,
      "lm_loss": 0.5808,
      "step": 4688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3439566385722278,
      "lm_loss": 0.4483,
      "step": 4688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3439566385722278,
      "lm_loss": 0.473,
      "step": 4688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3439566385722278,
      "lm_loss": 0.467,
      "step": 4688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3439566385722278,
      "lm_loss": 0.6405,
      "step": 4688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3439566385722278,
      "lm_loss": 0.3762,
      "step": 4688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3439566385722278,
      "lm_loss": 0.5815,
      "step": 4688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3440300081623669,
      "grad_norm": 0.8441866676712012,
      "learning_rate": 7.629188920581267e-06,
      "loss": 0.5239,
      "step": 4689
    },
    {
      "epoch": 0.34410337775250605,
      "grad_norm": 0.7473874875792343,
      "learning_rate": 7.628178184027972e-06,
      "loss": 0.5255,
      "step": 4690
    },
    {
      "epoch": 0.34417674734264514,
      "grad_norm": 1.6202984717116633,
      "learning_rate": 7.627167299054946e-06,
      "loss": 0.7976,
      "step": 4691
    },
    {
      "epoch": 0.3442501169327843,
      "grad_norm": 1.5233686124286998,
      "learning_rate": 7.626156265719277e-06,
      "loss": 0.7601,
      "step": 4692
    },
    {
      "epoch": 0.34432348652292344,
      "grad_norm": 1.7402721176845717,
      "learning_rate": 7.625145084078058e-06,
      "loss": 0.7896,
      "step": 4693
    },
    {
      "epoch": 0.34439685611306253,
      "grad_norm": 1.4753287229918195,
      "learning_rate": 7.624133754188399e-06,
      "loss": 0.7081,
      "step": 4694
    },
    {
      "epoch": 0.3444702257032017,
      "grad_norm": 1.5758580215256983,
      "learning_rate": 7.623122276107405e-06,
      "loss": 0.8372,
      "step": 4695
    },
    {
      "epoch": 0.3445435952933408,
      "grad_norm": 1.5747842826719833,
      "learning_rate": 7.6221106498922e-06,
      "loss": 0.7256,
      "step": 4696
    },
    {
      "epoch": 0.3445435952933408,
      "lm_loss": 0.167,
      "step": 4696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3445435952933408,
      "lm_loss": 0.4657,
      "step": 4696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3445435952933408,
      "lm_loss": 0.6649,
      "step": 4696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3445435952933408,
      "lm_loss": 0.4579,
      "step": 4696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3445435952933408,
      "lm_loss": 0.5629,
      "step": 4696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3445435952933408,
      "lm_loss": 0.4596,
      "step": 4696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3445435952933408,
      "lm_loss": 0.4941,
      "step": 4696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3445435952933408,
      "lm_loss": 0.536,
      "step": 4696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3446169648834799,
      "grad_norm": 0.8495012714427608,
      "learning_rate": 7.6210988755999145e-06,
      "loss": 0.5184,
      "step": 4697
    },
    {
      "epoch": 0.344690334473619,
      "grad_norm": 0.7618642299991842,
      "learning_rate": 7.620086953287685e-06,
      "loss": 0.4879,
      "step": 4698
    },
    {
      "epoch": 0.34476370406375817,
      "grad_norm": 1.5730614082616257,
      "learning_rate": 7.619074883012656e-06,
      "loss": 0.8512,
      "step": 4699
    },
    {
      "epoch": 0.3448370736538973,
      "grad_norm": 0.5876247668793438,
      "learning_rate": 7.618062664831982e-06,
      "loss": 0.5218,
      "step": 4700
    },
    {
      "epoch": 0.3449104432440364,
      "grad_norm": 1.723854564167055,
      "learning_rate": 7.6170502988028265e-06,
      "loss": 0.7656,
      "step": 4701
    },
    {
      "epoch": 0.34498381283417556,
      "grad_norm": 1.4979240046053903,
      "learning_rate": 7.616037784982359e-06,
      "loss": 0.6989,
      "step": 4702
    },
    {
      "epoch": 0.34505718242431466,
      "grad_norm": 1.87388004484428,
      "learning_rate": 7.61502512342776e-06,
      "loss": 0.7877,
      "step": 4703
    },
    {
      "epoch": 0.3451305520144538,
      "grad_norm": 1.7096827601663394,
      "learning_rate": 7.614012314196214e-06,
      "loss": 0.8851,
      "step": 4704
    },
    {
      "epoch": 0.3451305520144538,
      "lm_loss": 0.6171,
      "step": 4704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3451305520144538,
      "lm_loss": 0.4937,
      "step": 4704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3451305520144538,
      "lm_loss": 0.5474,
      "step": 4704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3451305520144538,
      "lm_loss": 0.4599,
      "step": 4704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3451305520144538,
      "lm_loss": 0.6393,
      "step": 4704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3451305520144538,
      "lm_loss": 0.5976,
      "step": 4704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3451305520144538,
      "lm_loss": 0.5304,
      "step": 4704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3451305520144538,
      "lm_loss": 0.587,
      "step": 4704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34520392160459296,
      "grad_norm": 0.921312655592833,
      "learning_rate": 7.612999357344921e-06,
      "loss": 0.5566,
      "step": 4705
    },
    {
      "epoch": 0.34527729119473205,
      "grad_norm": 0.8365264163639181,
      "learning_rate": 7.611986252931084e-06,
      "loss": 0.5451,
      "step": 4706
    },
    {
      "epoch": 0.3453506607848712,
      "grad_norm": 1.5295340988797717,
      "learning_rate": 7.610973001011913e-06,
      "loss": 0.7568,
      "step": 4707
    },
    {
      "epoch": 0.3454240303750103,
      "grad_norm": 1.4878003902281363,
      "learning_rate": 7.609959601644632e-06,
      "loss": 0.7139,
      "step": 4708
    },
    {
      "epoch": 0.34549739996514944,
      "grad_norm": 0.6637983936878099,
      "learning_rate": 7.608946054886468e-06,
      "loss": 0.5293,
      "step": 4709
    },
    {
      "epoch": 0.3455707695552886,
      "grad_norm": 2.0709588871484472,
      "learning_rate": 7.60793236079466e-06,
      "loss": 0.8399,
      "step": 4710
    },
    {
      "epoch": 0.3456441391454277,
      "grad_norm": 1.603873417243513,
      "learning_rate": 7.606918519426454e-06,
      "loss": 0.8378,
      "step": 4711
    },
    {
      "epoch": 0.34571750873556684,
      "grad_norm": 1.488763139264996,
      "learning_rate": 7.605904530839101e-06,
      "loss": 0.8181,
      "step": 4712
    },
    {
      "epoch": 0.34571750873556684,
      "lm_loss": 0.65,
      "step": 4712,
      "vm_loss": 0.0419
    },
    {
      "epoch": 0.34571750873556684,
      "lm_loss": 0.0078,
      "step": 4712,
      "vm_loss": 0.0839
    },
    {
      "epoch": 0.34571750873556684,
      "lm_loss": 1.2633,
      "step": 4712,
      "vm_loss": 0.1824
    },
    {
      "epoch": 0.34571750873556684,
      "lm_loss": 0.7084,
      "step": 4712,
      "vm_loss": 0.0666
    },
    {
      "epoch": 0.34571750873556684,
      "lm_loss": 0.5183,
      "step": 4712,
      "vm_loss": 0.1682
    },
    {
      "epoch": 0.34571750873556684,
      "lm_loss": 0.7649,
      "step": 4712,
      "vm_loss": 0.0722
    },
    {
      "epoch": 0.34571750873556684,
      "lm_loss": 0.439,
      "step": 4712,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.34571750873556684,
      "lm_loss": 0.2937,
      "step": 4712,
      "vm_loss": 0.1685
    },
    {
      "epoch": 0.34579087832570593,
      "grad_norm": 1.580518634578087,
      "learning_rate": 7.604890395089867e-06,
      "loss": 0.7855,
      "step": 4713
    },
    {
      "epoch": 0.3458642479158451,
      "grad_norm": 1.53324185708274,
      "learning_rate": 7.603876112236021e-06,
      "loss": 0.8662,
      "step": 4714
    },
    {
      "epoch": 0.34593761750598423,
      "grad_norm": 1.3921903117997134,
      "learning_rate": 7.602861682334843e-06,
      "loss": 0.7563,
      "step": 4715
    },
    {
      "epoch": 0.3460109870961233,
      "grad_norm": 1.6030561666590801,
      "learning_rate": 7.601847105443621e-06,
      "loss": 0.9032,
      "step": 4716
    },
    {
      "epoch": 0.3460843566862625,
      "grad_norm": 0.8376969093192715,
      "learning_rate": 7.6008323816196476e-06,
      "loss": 0.5243,
      "step": 4717
    },
    {
      "epoch": 0.34615772627640157,
      "grad_norm": 1.5539976395713146,
      "learning_rate": 7.599817510920229e-06,
      "loss": 0.7673,
      "step": 4718
    },
    {
      "epoch": 0.3462310958665407,
      "grad_norm": 0.6768105561146897,
      "learning_rate": 7.598802493402678e-06,
      "loss": 0.5245,
      "step": 4719
    },
    {
      "epoch": 0.34630446545667987,
      "grad_norm": 1.7688777145537116,
      "learning_rate": 7.597787329124314e-06,
      "loss": 0.8156,
      "step": 4720
    },
    {
      "epoch": 0.34630446545667987,
      "lm_loss": 0.3385,
      "step": 4720,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.34630446545667987,
      "lm_loss": 1.4018,
      "step": 4720,
      "vm_loss": 0.0642
    },
    {
      "epoch": 0.34630446545667987,
      "lm_loss": 0.2903,
      "step": 4720,
      "vm_loss": 0.0625
    },
    {
      "epoch": 0.34630446545667987,
      "lm_loss": 0.3733,
      "step": 4720,
      "vm_loss": 0.0634
    },
    {
      "epoch": 0.34630446545667987,
      "lm_loss": 0.2423,
      "step": 4720,
      "vm_loss": 0.087
    },
    {
      "epoch": 0.34630446545667987,
      "lm_loss": 0.1805,
      "step": 4720,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.34630446545667987,
      "lm_loss": 0.101,
      "step": 4720,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.34630446545667987,
      "lm_loss": 0.1344,
      "step": 4720,
      "vm_loss": 0.1269
    },
    {
      "epoch": 0.34637783504681896,
      "grad_norm": 1.4052014817144505,
      "learning_rate": 7.5967720181424665e-06,
      "loss": 0.7866,
      "step": 4721
    },
    {
      "epoch": 0.3464512046369581,
      "grad_norm": 1.4066850519032175,
      "learning_rate": 7.5957565605144736e-06,
      "loss": 0.8119,
      "step": 4722
    },
    {
      "epoch": 0.3465245742270972,
      "grad_norm": 1.6370772497842083,
      "learning_rate": 7.594740956297677e-06,
      "loss": 0.8441,
      "step": 4723
    },
    {
      "epoch": 0.34659794381723635,
      "grad_norm": 1.513176208007793,
      "learning_rate": 7.5937252055494335e-06,
      "loss": 0.7465,
      "step": 4724
    },
    {
      "epoch": 0.3466713134073755,
      "grad_norm": 1.7136686188047863,
      "learning_rate": 7.592709308327107e-06,
      "loss": 0.8497,
      "step": 4725
    },
    {
      "epoch": 0.3467446829975146,
      "grad_norm": 1.7305897488201263,
      "learning_rate": 7.591693264688061e-06,
      "loss": 0.8108,
      "step": 4726
    },
    {
      "epoch": 0.34681805258765375,
      "grad_norm": 1.6847043631647403,
      "learning_rate": 7.590677074689681e-06,
      "loss": 0.7655,
      "step": 4727
    },
    {
      "epoch": 0.34689142217779284,
      "grad_norm": 1.6210852278887529,
      "learning_rate": 7.58966073838935e-06,
      "loss": 0.7825,
      "step": 4728
    },
    {
      "epoch": 0.34689142217779284,
      "lm_loss": 0.7517,
      "step": 4728,
      "vm_loss": 0.049
    },
    {
      "epoch": 0.34689142217779284,
      "lm_loss": 0.3387,
      "step": 4728,
      "vm_loss": 0.1701
    },
    {
      "epoch": 0.34689142217779284,
      "lm_loss": 0.6857,
      "step": 4728,
      "vm_loss": 0.1271
    },
    {
      "epoch": 0.34689142217779284,
      "lm_loss": 0.9191,
      "step": 4728,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.34689142217779284,
      "lm_loss": 0.9185,
      "step": 4728,
      "vm_loss": 0.1671
    },
    {
      "epoch": 0.34689142217779284,
      "lm_loss": 0.7824,
      "step": 4728,
      "vm_loss": 0.0638
    },
    {
      "epoch": 0.34689142217779284,
      "lm_loss": 1.2651,
      "step": 4728,
      "vm_loss": 0.0644
    },
    {
      "epoch": 0.34689142217779284,
      "lm_loss": 0.3202,
      "step": 4728,
      "vm_loss": 0.1362
    },
    {
      "epoch": 0.346964791767932,
      "grad_norm": 1.5784953752203998,
      "learning_rate": 7.588644255844464e-06,
      "loss": 0.7843,
      "step": 4729
    },
    {
      "epoch": 0.34703816135807114,
      "grad_norm": 0.949227260710175,
      "learning_rate": 7.587627627112426e-06,
      "loss": 0.5125,
      "step": 4730
    },
    {
      "epoch": 0.34711153094821023,
      "grad_norm": 1.5562689106517302,
      "learning_rate": 7.5866108522506475e-06,
      "loss": 0.732,
      "step": 4731
    },
    {
      "epoch": 0.3471849005383494,
      "grad_norm": 1.651012516276137,
      "learning_rate": 7.585593931316548e-06,
      "loss": 0.8658,
      "step": 4732
    },
    {
      "epoch": 0.3472582701284885,
      "grad_norm": 0.6934147311484582,
      "learning_rate": 7.584576864367557e-06,
      "loss": 0.5365,
      "step": 4733
    },
    {
      "epoch": 0.3473316397186276,
      "grad_norm": 1.6403009081048219,
      "learning_rate": 7.5835596514611095e-06,
      "loss": 0.842,
      "step": 4734
    },
    {
      "epoch": 0.3474050093087667,
      "grad_norm": 0.5953526571017793,
      "learning_rate": 7.582542292654652e-06,
      "loss": 0.5249,
      "step": 4735
    },
    {
      "epoch": 0.34747837889890587,
      "grad_norm": 1.5820053056176906,
      "learning_rate": 7.581524788005634e-06,
      "loss": 0.7999,
      "step": 4736
    },
    {
      "epoch": 0.34747837889890587,
      "lm_loss": 0.4742,
      "step": 4736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34747837889890587,
      "lm_loss": 0.3795,
      "step": 4736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34747837889890587,
      "lm_loss": 0.638,
      "step": 4736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34747837889890587,
      "lm_loss": 0.3902,
      "step": 4736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34747837889890587,
      "lm_loss": 0.5246,
      "step": 4736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34747837889890587,
      "lm_loss": 0.542,
      "step": 4736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34747837889890587,
      "lm_loss": 0.4993,
      "step": 4736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34747837889890587,
      "lm_loss": 0.53,
      "step": 4736,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.347551748489045,
      "grad_norm": 0.6059779090537201,
      "learning_rate": 7.580507137571519e-06,
      "loss": 0.5122,
      "step": 4737
    },
    {
      "epoch": 0.3476251180791841,
      "grad_norm": 0.6511015327833648,
      "learning_rate": 7.579489341409776e-06,
      "loss": 0.5344,
      "step": 4738
    },
    {
      "epoch": 0.34769848766932326,
      "grad_norm": 1.522775426029662,
      "learning_rate": 7.578471399577879e-06,
      "loss": 0.7317,
      "step": 4739
    },
    {
      "epoch": 0.34777185725946236,
      "grad_norm": 1.8063303888738271,
      "learning_rate": 7.5774533121333195e-06,
      "loss": 0.8264,
      "step": 4740
    },
    {
      "epoch": 0.3478452268496015,
      "grad_norm": 1.5443522381132813,
      "learning_rate": 7.576435079133586e-06,
      "loss": 0.8043,
      "step": 4741
    },
    {
      "epoch": 0.34791859643974066,
      "grad_norm": 1.5760598183440382,
      "learning_rate": 7.575416700636186e-06,
      "loss": 0.8536,
      "step": 4742
    },
    {
      "epoch": 0.34799196602987975,
      "grad_norm": 1.4825349799236918,
      "learning_rate": 7.574398176698625e-06,
      "loss": 0.7809,
      "step": 4743
    },
    {
      "epoch": 0.3480653356200189,
      "grad_norm": 1.3541284331510564,
      "learning_rate": 7.573379507378424e-06,
      "loss": 0.8139,
      "step": 4744
    },
    {
      "epoch": 0.3480653356200189,
      "lm_loss": 1.972,
      "step": 4744,
      "vm_loss": 0.0725
    },
    {
      "epoch": 0.3480653356200189,
      "lm_loss": 1.3195,
      "step": 4744,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.3480653356200189,
      "lm_loss": 0.6392,
      "step": 4744,
      "vm_loss": 0.069
    },
    {
      "epoch": 0.3480653356200189,
      "lm_loss": 0.2826,
      "step": 4744,
      "vm_loss": 0.1116
    },
    {
      "epoch": 0.3480653356200189,
      "lm_loss": 0.8868,
      "step": 4744,
      "vm_loss": 0.1802
    },
    {
      "epoch": 0.3480653356200189,
      "lm_loss": 0.1441,
      "step": 4744,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.3480653356200189,
      "lm_loss": 1.1163,
      "step": 4744,
      "vm_loss": 0.0658
    },
    {
      "epoch": 0.3480653356200189,
      "lm_loss": 0.4024,
      "step": 4744,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.348138705210158,
      "grad_norm": 1.665121775431218,
      "learning_rate": 7.572360692733109e-06,
      "loss": 0.8051,
      "step": 4745
    },
    {
      "epoch": 0.34821207480029714,
      "grad_norm": 1.391806026625563,
      "learning_rate": 7.571341732820217e-06,
      "loss": 0.7759,
      "step": 4746
    },
    {
      "epoch": 0.3482854443904363,
      "grad_norm": 0.6759363396151373,
      "learning_rate": 7.570322627697289e-06,
      "loss": 0.5182,
      "step": 4747
    },
    {
      "epoch": 0.3483588139805754,
      "grad_norm": 1.5799792627923717,
      "learning_rate": 7.569303377421875e-06,
      "loss": 0.7888,
      "step": 4748
    },
    {
      "epoch": 0.34843218357071454,
      "grad_norm": 1.647647438159706,
      "learning_rate": 7.568283982051538e-06,
      "loss": 0.8432,
      "step": 4749
    },
    {
      "epoch": 0.34850555316085363,
      "grad_norm": 1.4410579104566636,
      "learning_rate": 7.567264441643845e-06,
      "loss": 0.8334,
      "step": 4750
    },
    {
      "epoch": 0.3485789227509928,
      "grad_norm": 1.5450008433280933,
      "learning_rate": 7.56624475625637e-06,
      "loss": 0.737,
      "step": 4751
    },
    {
      "epoch": 0.34865229234113193,
      "grad_norm": 1.7315442498218505,
      "learning_rate": 7.565224925946698e-06,
      "loss": 0.8602,
      "step": 4752
    },
    {
      "epoch": 0.34865229234113193,
      "lm_loss": 1.4269,
      "step": 4752,
      "vm_loss": 0.0424
    },
    {
      "epoch": 0.34865229234113193,
      "lm_loss": 0.4814,
      "step": 4752,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.34865229234113193,
      "lm_loss": 0.5897,
      "step": 4752,
      "vm_loss": 0.1304
    },
    {
      "epoch": 0.34865229234113193,
      "lm_loss": 0.8639,
      "step": 4752,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.34865229234113193,
      "lm_loss": 1.0393,
      "step": 4752,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.34865229234113193,
      "lm_loss": 0.1597,
      "step": 4752,
      "vm_loss": 0.066
    },
    {
      "epoch": 0.34865229234113193,
      "lm_loss": 1.0864,
      "step": 4752,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.34865229234113193,
      "lm_loss": 0.0932,
      "step": 4752,
      "vm_loss": 0.065
    },
    {
      "epoch": 0.348725661931271,
      "grad_norm": 1.4868857377357902,
      "learning_rate": 7.564204950772423e-06,
      "loss": 0.8055,
      "step": 4753
    },
    {
      "epoch": 0.3487990315214102,
      "grad_norm": 1.6248198860398495,
      "learning_rate": 7.563184830791144e-06,
      "loss": 0.793,
      "step": 4754
    },
    {
      "epoch": 0.34887240111154927,
      "grad_norm": 1.5316572749021597,
      "learning_rate": 7.56216456606047e-06,
      "loss": 0.7679,
      "step": 4755
    },
    {
      "epoch": 0.3489457707016884,
      "grad_norm": 1.561973786404289,
      "learning_rate": 7.561144156638018e-06,
      "loss": 0.8421,
      "step": 4756
    },
    {
      "epoch": 0.34901914029182757,
      "grad_norm": 0.7104031655750531,
      "learning_rate": 7.560123602581413e-06,
      "loss": 0.5273,
      "step": 4757
    },
    {
      "epoch": 0.34909250988196666,
      "grad_norm": 0.6226568615807095,
      "learning_rate": 7.559102903948287e-06,
      "loss": 0.5228,
      "step": 4758
    },
    {
      "epoch": 0.3491658794721058,
      "grad_norm": 0.5799006093758561,
      "learning_rate": 7.558082060796283e-06,
      "loss": 0.5152,
      "step": 4759
    },
    {
      "epoch": 0.3492392490622449,
      "grad_norm": 1.5528141657426906,
      "learning_rate": 7.5570610731830495e-06,
      "loss": 0.8407,
      "step": 4760
    },
    {
      "epoch": 0.3492392490622449,
      "lm_loss": 0.206,
      "step": 4760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3492392490622449,
      "lm_loss": 0.6074,
      "step": 4760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3492392490622449,
      "lm_loss": 0.4834,
      "step": 4760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3492392490622449,
      "lm_loss": 0.4658,
      "step": 4760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3492392490622449,
      "lm_loss": 0.5925,
      "step": 4760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3492392490622449,
      "lm_loss": 0.5104,
      "step": 4760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3492392490622449,
      "lm_loss": 0.5407,
      "step": 4760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3492392490622449,
      "lm_loss": 0.4059,
      "step": 4760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.34931261865238405,
      "grad_norm": 0.6077879799267867,
      "learning_rate": 7.556039941166246e-06,
      "loss": 0.5361,
      "step": 4761
    },
    {
      "epoch": 0.3493859882425232,
      "grad_norm": 0.5989427766722256,
      "learning_rate": 7.555018664803537e-06,
      "loss": 0.522,
      "step": 4762
    },
    {
      "epoch": 0.3494593578326623,
      "grad_norm": 1.572417730485971,
      "learning_rate": 7.553997244152595e-06,
      "loss": 0.9806,
      "step": 4763
    },
    {
      "epoch": 0.34953272742280145,
      "grad_norm": 1.4560996170476574,
      "learning_rate": 7.552975679271106e-06,
      "loss": 0.719,
      "step": 4764
    },
    {
      "epoch": 0.34960609701294054,
      "grad_norm": 1.4632741223741668,
      "learning_rate": 7.551953970216758e-06,
      "loss": 0.8088,
      "step": 4765
    },
    {
      "epoch": 0.3496794666030797,
      "grad_norm": 1.3519178480521088,
      "learning_rate": 7.550932117047247e-06,
      "loss": 0.7271,
      "step": 4766
    },
    {
      "epoch": 0.34975283619321884,
      "grad_norm": 0.5806064512232519,
      "learning_rate": 7.549910119820284e-06,
      "loss": 0.5151,
      "step": 4767
    },
    {
      "epoch": 0.34982620578335794,
      "grad_norm": 0.6211449087140085,
      "learning_rate": 7.548887978593581e-06,
      "loss": 0.5498,
      "step": 4768
    },
    {
      "epoch": 0.34982620578335794,
      "lm_loss": 1.3464,
      "step": 4768,
      "vm_loss": 0.0896
    },
    {
      "epoch": 0.34982620578335794,
      "lm_loss": 0.2347,
      "step": 4768,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.34982620578335794,
      "lm_loss": 0.1757,
      "step": 4768,
      "vm_loss": 0.1541
    },
    {
      "epoch": 0.34982620578335794,
      "lm_loss": 0.6583,
      "step": 4768,
      "vm_loss": 0.1433
    },
    {
      "epoch": 0.34982620578335794,
      "lm_loss": 0.3404,
      "step": 4768,
      "vm_loss": 0.1742
    },
    {
      "epoch": 0.34982620578335794,
      "lm_loss": 0.6979,
      "step": 4768,
      "vm_loss": 0.0735
    },
    {
      "epoch": 0.34982620578335794,
      "lm_loss": 0.0209,
      "step": 4768,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.34982620578335794,
      "lm_loss": 0.4294,
      "step": 4768,
      "vm_loss": 0.1126
    },
    {
      "epoch": 0.3498995753734971,
      "grad_norm": 1.4981155337226753,
      "learning_rate": 7.5478656934248626e-06,
      "loss": 0.6682,
      "step": 4769
    },
    {
      "epoch": 0.3499729449636362,
      "grad_norm": 1.3615141845067213,
      "learning_rate": 7.546843264371858e-06,
      "loss": 0.6866,
      "step": 4770
    },
    {
      "epoch": 0.35004631455377533,
      "grad_norm": 0.5656479223512694,
      "learning_rate": 7.545820691492307e-06,
      "loss": 0.5358,
      "step": 4771
    },
    {
      "epoch": 0.3501196841439145,
      "grad_norm": 1.6246053592302196,
      "learning_rate": 7.544797974843958e-06,
      "loss": 0.806,
      "step": 4772
    },
    {
      "epoch": 0.35019305373405357,
      "grad_norm": 1.5062024978123307,
      "learning_rate": 7.543775114484564e-06,
      "loss": 0.7158,
      "step": 4773
    },
    {
      "epoch": 0.3502664233241927,
      "grad_norm": 0.5899701779795677,
      "learning_rate": 7.542752110471891e-06,
      "loss": 0.5285,
      "step": 4774
    },
    {
      "epoch": 0.3503397929143318,
      "grad_norm": 0.5717797850506068,
      "learning_rate": 7.5417289628637084e-06,
      "loss": 0.5155,
      "step": 4775
    },
    {
      "epoch": 0.35041316250447097,
      "grad_norm": 0.5884207550038606,
      "learning_rate": 7.5407056717177965e-06,
      "loss": 0.5362,
      "step": 4776
    },
    {
      "epoch": 0.35041316250447097,
      "lm_loss": 0.8811,
      "step": 4776,
      "vm_loss": 0.0858
    },
    {
      "epoch": 0.35041316250447097,
      "lm_loss": 0.4285,
      "step": 4776,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.35041316250447097,
      "lm_loss": 0.2643,
      "step": 4776,
      "vm_loss": 0.1563
    },
    {
      "epoch": 0.35041316250447097,
      "lm_loss": 0.3867,
      "step": 4776,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.35041316250447097,
      "lm_loss": 0.4886,
      "step": 4776,
      "vm_loss": 0.1312
    },
    {
      "epoch": 0.35041316250447097,
      "lm_loss": 0.2347,
      "step": 4776,
      "vm_loss": 0.094
    },
    {
      "epoch": 0.35041316250447097,
      "lm_loss": 0.5703,
      "step": 4776,
      "vm_loss": 0.151
    },
    {
      "epoch": 0.35041316250447097,
      "lm_loss": 1.0655,
      "step": 4776,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.35048653209461006,
      "grad_norm": 1.49391519879668,
      "learning_rate": 7.539682237091943e-06,
      "loss": 0.745,
      "step": 4777
    },
    {
      "epoch": 0.3505599016847492,
      "grad_norm": 1.4692184099217216,
      "learning_rate": 7.538658659043946e-06,
      "loss": 0.7487,
      "step": 4778
    },
    {
      "epoch": 0.35063327127488836,
      "grad_norm": 1.4925754542345755,
      "learning_rate": 7.537634937631606e-06,
      "loss": 0.7751,
      "step": 4779
    },
    {
      "epoch": 0.35070664086502745,
      "grad_norm": 1.5914404995610514,
      "learning_rate": 7.5366110729127385e-06,
      "loss": 0.8589,
      "step": 4780
    },
    {
      "epoch": 0.3507800104551666,
      "grad_norm": 1.6364574444670035,
      "learning_rate": 7.53558706494516e-06,
      "loss": 0.9304,
      "step": 4781
    },
    {
      "epoch": 0.3508533800453057,
      "grad_norm": 1.426420156594953,
      "learning_rate": 7.534562913786701e-06,
      "loss": 0.6931,
      "step": 4782
    },
    {
      "epoch": 0.35092674963544485,
      "grad_norm": 1.2954091914388097,
      "learning_rate": 7.533538619495198e-06,
      "loss": 0.8342,
      "step": 4783
    },
    {
      "epoch": 0.351000119225584,
      "grad_norm": 0.7163835659255763,
      "learning_rate": 7.532514182128495e-06,
      "loss": 0.5154,
      "step": 4784
    },
    {
      "epoch": 0.351000119225584,
      "lm_loss": 0.2035,
      "step": 4784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.351000119225584,
      "lm_loss": 0.4834,
      "step": 4784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.351000119225584,
      "lm_loss": 0.6336,
      "step": 4784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.351000119225584,
      "lm_loss": 0.6797,
      "step": 4784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.351000119225584,
      "lm_loss": 0.6268,
      "step": 4784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.351000119225584,
      "lm_loss": 0.5701,
      "step": 4784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.351000119225584,
      "lm_loss": 0.3902,
      "step": 4784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.351000119225584,
      "lm_loss": 0.5961,
      "step": 4784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3510734888157231,
      "grad_norm": 0.6841158106656637,
      "learning_rate": 7.531489601744444e-06,
      "loss": 0.5211,
      "step": 4785
    },
    {
      "epoch": 0.35114685840586224,
      "grad_norm": 0.5776802387963216,
      "learning_rate": 7.5304648784009075e-06,
      "loss": 0.51,
      "step": 4786
    },
    {
      "epoch": 0.35122022799600133,
      "grad_norm": 1.3685331252946429,
      "learning_rate": 7.529440012155751e-06,
      "loss": 0.702,
      "step": 4787
    },
    {
      "epoch": 0.3512935975861405,
      "grad_norm": 0.6616422329395281,
      "learning_rate": 7.528415003066854e-06,
      "loss": 0.5147,
      "step": 4788
    },
    {
      "epoch": 0.35136696717627963,
      "grad_norm": 0.7154884478019454,
      "learning_rate": 7.527389851192099e-06,
      "loss": 0.54,
      "step": 4789
    },
    {
      "epoch": 0.3514403367664187,
      "grad_norm": 1.4723834672858824,
      "learning_rate": 7.526364556589381e-06,
      "loss": 0.7364,
      "step": 4790
    },
    {
      "epoch": 0.3515137063565579,
      "grad_norm": 1.4780340452463687,
      "learning_rate": 7.5253391193166005e-06,
      "loss": 0.8343,
      "step": 4791
    },
    {
      "epoch": 0.35158707594669697,
      "grad_norm": 1.589857763425939,
      "learning_rate": 7.524313539431666e-06,
      "loss": 0.7801,
      "step": 4792
    },
    {
      "epoch": 0.35158707594669697,
      "lm_loss": 1.515,
      "step": 4792,
      "vm_loss": 0.0329
    },
    {
      "epoch": 0.35158707594669697,
      "lm_loss": 0.3783,
      "step": 4792,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.35158707594669697,
      "lm_loss": 0.457,
      "step": 4792,
      "vm_loss": 0.1302
    },
    {
      "epoch": 0.35158707594669697,
      "lm_loss": 0.5617,
      "step": 4792,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.35158707594669697,
      "lm_loss": 0.7648,
      "step": 4792,
      "vm_loss": 0.2169
    },
    {
      "epoch": 0.35158707594669697,
      "lm_loss": 0.0536,
      "step": 4792,
      "vm_loss": 0.2379
    },
    {
      "epoch": 0.35158707594669697,
      "lm_loss": 0.8342,
      "step": 4792,
      "vm_loss": 0.0978
    },
    {
      "epoch": 0.35158707594669697,
      "lm_loss": 0.6717,
      "step": 4792,
      "vm_loss": 0.2042
    },
    {
      "epoch": 0.3516604455368361,
      "grad_norm": 1.430637770101639,
      "learning_rate": 7.5232878169924925e-06,
      "loss": 0.7279,
      "step": 4793
    },
    {
      "epoch": 0.35173381512697527,
      "grad_norm": 1.4515985535111127,
      "learning_rate": 7.52226195205701e-06,
      "loss": 0.8752,
      "step": 4794
    },
    {
      "epoch": 0.35180718471711436,
      "grad_norm": 0.6391870189201095,
      "learning_rate": 7.521235944683147e-06,
      "loss": 0.5109,
      "step": 4795
    },
    {
      "epoch": 0.3518805543072535,
      "grad_norm": 0.6243793795416475,
      "learning_rate": 7.520209794928848e-06,
      "loss": 0.5171,
      "step": 4796
    },
    {
      "epoch": 0.3519539238973926,
      "grad_norm": 0.5878980333564733,
      "learning_rate": 7.519183502852059e-06,
      "loss": 0.5167,
      "step": 4797
    },
    {
      "epoch": 0.35202729348753176,
      "grad_norm": 1.7248652796146442,
      "learning_rate": 7.51815706851074e-06,
      "loss": 0.8172,
      "step": 4798
    },
    {
      "epoch": 0.3521006630776709,
      "grad_norm": 1.6868258651289536,
      "learning_rate": 7.517130491962854e-06,
      "loss": 0.8293,
      "step": 4799
    },
    {
      "epoch": 0.35217403266781,
      "grad_norm": 1.4550144505154579,
      "learning_rate": 7.516103773266377e-06,
      "loss": 0.8798,
      "step": 4800
    },
    {
      "epoch": 0.35217403266781,
      "lm_loss": 0.6372,
      "step": 4800,
      "vm_loss": 0.0378
    },
    {
      "epoch": 0.35217403266781,
      "lm_loss": 0.8603,
      "step": 4800,
      "vm_loss": 0.124
    },
    {
      "epoch": 0.35217403266781,
      "lm_loss": 0.2321,
      "step": 4800,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.35217403266781,
      "lm_loss": 0.503,
      "step": 4800,
      "vm_loss": 0.0974
    },
    {
      "epoch": 0.35217403266781,
      "lm_loss": 0.8014,
      "step": 4800,
      "vm_loss": 0.0918
    },
    {
      "epoch": 0.35217403266781,
      "lm_loss": 0.1213,
      "step": 4800,
      "vm_loss": 0.2245
    },
    {
      "epoch": 0.35217403266781,
      "lm_loss": 0.25,
      "step": 4800,
      "vm_loss": 0.1115
    },
    {
      "epoch": 0.35217403266781,
      "lm_loss": 0.5756,
      "step": 4800,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.35224740225794915,
      "grad_norm": 1.2481257030097332,
      "learning_rate": 7.515076912479289e-06,
      "loss": 0.6974,
      "step": 4801
    },
    {
      "epoch": 0.35232077184808824,
      "grad_norm": 0.6561591169984844,
      "learning_rate": 7.514049909659578e-06,
      "loss": 0.5416,
      "step": 4802
    },
    {
      "epoch": 0.3523941414382274,
      "grad_norm": 1.4923146277415207,
      "learning_rate": 7.513022764865243e-06,
      "loss": 0.7954,
      "step": 4803
    },
    {
      "epoch": 0.35246751102836654,
      "grad_norm": 0.593957889601943,
      "learning_rate": 7.511995478154291e-06,
      "loss": 0.4951,
      "step": 4804
    },
    {
      "epoch": 0.35254088061850564,
      "grad_norm": 1.6728139286125285,
      "learning_rate": 7.5109680495847325e-06,
      "loss": 0.8599,
      "step": 4805
    },
    {
      "epoch": 0.3526142502086448,
      "grad_norm": 1.3308230708359827,
      "learning_rate": 7.509940479214588e-06,
      "loss": 0.7878,
      "step": 4806
    },
    {
      "epoch": 0.3526876197987839,
      "grad_norm": 0.5726905590418282,
      "learning_rate": 7.50891276710189e-06,
      "loss": 0.5132,
      "step": 4807
    },
    {
      "epoch": 0.35276098938892303,
      "grad_norm": 1.536610148588957,
      "learning_rate": 7.507884913304675e-06,
      "loss": 0.8922,
      "step": 4808
    },
    {
      "epoch": 0.35276098938892303,
      "lm_loss": 1.915,
      "step": 4808,
      "vm_loss": 0.0608
    },
    {
      "epoch": 0.35276098938892303,
      "lm_loss": 1.3129,
      "step": 4808,
      "vm_loss": 0.0649
    },
    {
      "epoch": 0.35276098938892303,
      "lm_loss": 1.4621,
      "step": 4808,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.35276098938892303,
      "lm_loss": 0.0778,
      "step": 4808,
      "vm_loss": 0.0492
    },
    {
      "epoch": 0.35276098938892303,
      "lm_loss": 0.8474,
      "step": 4808,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.35276098938892303,
      "lm_loss": 1.2213,
      "step": 4808,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.35276098938892303,
      "lm_loss": 0.1936,
      "step": 4808,
      "vm_loss": 0.0554
    },
    {
      "epoch": 0.35276098938892303,
      "lm_loss": 0.5664,
      "step": 4808,
      "vm_loss": 0.1732
    },
    {
      "epoch": 0.3528343589790622,
      "grad_norm": 1.5322942777818929,
      "learning_rate": 7.506856917880989e-06,
      "loss": 0.8372,
      "step": 4809
    },
    {
      "epoch": 0.3529077285692013,
      "grad_norm": 1.705635049610554,
      "learning_rate": 7.505828780888884e-06,
      "loss": 0.8439,
      "step": 4810
    },
    {
      "epoch": 0.3529810981593404,
      "grad_norm": 1.4886292695909924,
      "learning_rate": 7.5048005023864216e-06,
      "loss": 0.8816,
      "step": 4811
    },
    {
      "epoch": 0.3530544677494795,
      "grad_norm": 1.3652460508114246,
      "learning_rate": 7.503772082431673e-06,
      "loss": 0.8175,
      "step": 4812
    },
    {
      "epoch": 0.35312783733961867,
      "grad_norm": 1.7274658643349425,
      "learning_rate": 7.502743521082714e-06,
      "loss": 0.785,
      "step": 4813
    },
    {
      "epoch": 0.3532012069297578,
      "grad_norm": 1.5043450649415675,
      "learning_rate": 7.501714818397629e-06,
      "loss": 0.7777,
      "step": 4814
    },
    {
      "epoch": 0.3532745765198969,
      "grad_norm": 1.6099039308778906,
      "learning_rate": 7.500685974434514e-06,
      "loss": 0.7907,
      "step": 4815
    },
    {
      "epoch": 0.35334794611003606,
      "grad_norm": 1.4225484534049206,
      "learning_rate": 7.499656989251469e-06,
      "loss": 0.7096,
      "step": 4816
    },
    {
      "epoch": 0.35334794611003606,
      "lm_loss": 1.2251,
      "step": 4816,
      "vm_loss": 0.0566
    },
    {
      "epoch": 0.35334794611003606,
      "lm_loss": 0.994,
      "step": 4816,
      "vm_loss": 0.043
    },
    {
      "epoch": 0.35334794611003606,
      "lm_loss": 1.1029,
      "step": 4816,
      "vm_loss": 0.1567
    },
    {
      "epoch": 0.35334794611003606,
      "lm_loss": 0.4356,
      "step": 4816,
      "vm_loss": 0.0726
    },
    {
      "epoch": 0.35334794611003606,
      "lm_loss": 0.4199,
      "step": 4816,
      "vm_loss": 0.0835
    },
    {
      "epoch": 0.35334794611003606,
      "lm_loss": 0.3725,
      "step": 4816,
      "vm_loss": 0.171
    },
    {
      "epoch": 0.35334794611003606,
      "lm_loss": 1.032,
      "step": 4816,
      "vm_loss": 0.1481
    },
    {
      "epoch": 0.35334794611003606,
      "lm_loss": 1.1612,
      "step": 4816,
      "vm_loss": 0.1148
    },
    {
      "epoch": 0.35342131570017515,
      "grad_norm": 1.3403251148414952,
      "learning_rate": 7.498627862906604e-06,
      "loss": 0.876,
      "step": 4817
    },
    {
      "epoch": 0.3534946852903143,
      "grad_norm": 1.4871353212571095,
      "learning_rate": 7.497598595458036e-06,
      "loss": 0.8431,
      "step": 4818
    },
    {
      "epoch": 0.3535680548804534,
      "grad_norm": 1.5407385978345907,
      "learning_rate": 7.496569186963889e-06,
      "loss": 0.8257,
      "step": 4819
    },
    {
      "epoch": 0.35364142447059255,
      "grad_norm": 1.6582477383862482,
      "learning_rate": 7.495539637482299e-06,
      "loss": 0.8056,
      "step": 4820
    },
    {
      "epoch": 0.3537147940607317,
      "grad_norm": 1.6200382800699369,
      "learning_rate": 7.494509947071405e-06,
      "loss": 0.736,
      "step": 4821
    },
    {
      "epoch": 0.3537881636508708,
      "grad_norm": 1.3870750253821051,
      "learning_rate": 7.493480115789355e-06,
      "loss": 0.7182,
      "step": 4822
    },
    {
      "epoch": 0.35386153324100994,
      "grad_norm": 1.3931391261021864,
      "learning_rate": 7.4924501436943094e-06,
      "loss": 0.8383,
      "step": 4823
    },
    {
      "epoch": 0.35393490283114903,
      "grad_norm": 0.759396985224538,
      "learning_rate": 7.491420030844432e-06,
      "loss": 0.5617,
      "step": 4824
    },
    {
      "epoch": 0.35393490283114903,
      "lm_loss": 1.5592,
      "step": 4824,
      "vm_loss": 0.0837
    },
    {
      "epoch": 0.35393490283114903,
      "lm_loss": 1.3162,
      "step": 4824,
      "vm_loss": 0.1492
    },
    {
      "epoch": 0.35393490283114903,
      "lm_loss": 1.267,
      "step": 4824,
      "vm_loss": 0.1579
    },
    {
      "epoch": 0.35393490283114903,
      "lm_loss": 0.9854,
      "step": 4824,
      "vm_loss": 0.1023
    },
    {
      "epoch": 0.35393490283114903,
      "lm_loss": 0.729,
      "step": 4824,
      "vm_loss": 0.1846
    },
    {
      "epoch": 0.35393490283114903,
      "lm_loss": 0.0346,
      "step": 4824,
      "vm_loss": 0.1674
    },
    {
      "epoch": 0.35393490283114903,
      "lm_loss": 0.6466,
      "step": 4824,
      "vm_loss": 0.1656
    },
    {
      "epoch": 0.35393490283114903,
      "lm_loss": 0.1032,
      "step": 4824,
      "vm_loss": 0.1511
    },
    {
      "epoch": 0.3540082724212882,
      "grad_norm": 1.4087462911528705,
      "learning_rate": 7.490389777297895e-06,
      "loss": 0.9327,
      "step": 4825
    },
    {
      "epoch": 0.35408164201142733,
      "grad_norm": 0.6308494947511389,
      "learning_rate": 7.489359383112879e-06,
      "loss": 0.5237,
      "step": 4826
    },
    {
      "epoch": 0.3541550116015664,
      "grad_norm": 1.5563321356109798,
      "learning_rate": 7.4883288483475744e-06,
      "loss": 0.819,
      "step": 4827
    },
    {
      "epoch": 0.3542283811917056,
      "grad_norm": 1.4641450744441273,
      "learning_rate": 7.487298173060179e-06,
      "loss": 0.8006,
      "step": 4828
    },
    {
      "epoch": 0.35430175078184467,
      "grad_norm": 1.5451979467287,
      "learning_rate": 7.486267357308896e-06,
      "loss": 0.7737,
      "step": 4829
    },
    {
      "epoch": 0.3543751203719838,
      "grad_norm": 1.5078474071724273,
      "learning_rate": 7.485236401151937e-06,
      "loss": 0.7896,
      "step": 4830
    },
    {
      "epoch": 0.35444848996212297,
      "grad_norm": 1.2846467277609934,
      "learning_rate": 7.484205304647524e-06,
      "loss": 0.8257,
      "step": 4831
    },
    {
      "epoch": 0.35452185955226206,
      "grad_norm": 1.5306213492559306,
      "learning_rate": 7.483174067853885e-06,
      "loss": 0.76,
      "step": 4832
    },
    {
      "epoch": 0.35452185955226206,
      "lm_loss": 0.4295,
      "step": 4832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.35452185955226206,
      "lm_loss": 0.5282,
      "step": 4832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.35452185955226206,
      "lm_loss": 0.5006,
      "step": 4832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.35452185955226206,
      "lm_loss": 0.5195,
      "step": 4832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.35452185955226206,
      "lm_loss": 0.4896,
      "step": 4832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.35452185955226206,
      "lm_loss": 0.5631,
      "step": 4832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.35452185955226206,
      "lm_loss": 0.4992,
      "step": 4832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.35452185955226206,
      "lm_loss": 0.5211,
      "step": 4832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3545952291424012,
      "grad_norm": 0.7523278263414981,
      "learning_rate": 7.4821426908292575e-06,
      "loss": 0.5184,
      "step": 4833
    },
    {
      "epoch": 0.3546685987325403,
      "grad_norm": 1.5084375905543999,
      "learning_rate": 7.481111173631885e-06,
      "loss": 0.9056,
      "step": 4834
    },
    {
      "epoch": 0.35474196832267946,
      "grad_norm": 1.4655206112779433,
      "learning_rate": 7.4800795163200205e-06,
      "loss": 0.7749,
      "step": 4835
    },
    {
      "epoch": 0.3548153379128186,
      "grad_norm": 1.4280110028395436,
      "learning_rate": 7.479047718951924e-06,
      "loss": 0.777,
      "step": 4836
    },
    {
      "epoch": 0.3548887075029577,
      "grad_norm": 0.6434425511845536,
      "learning_rate": 7.478015781585864e-06,
      "loss": 0.5155,
      "step": 4837
    },
    {
      "epoch": 0.35496207709309685,
      "grad_norm": 1.402066748053251,
      "learning_rate": 7.476983704280114e-06,
      "loss": 0.7638,
      "step": 4838
    },
    {
      "epoch": 0.35503544668323594,
      "grad_norm": 1.475355573208153,
      "learning_rate": 7.475951487092962e-06,
      "loss": 0.7649,
      "step": 4839
    },
    {
      "epoch": 0.3551088162733751,
      "grad_norm": 1.575644630776506,
      "learning_rate": 7.474919130082698e-06,
      "loss": 0.9035,
      "step": 4840
    },
    {
      "epoch": 0.3551088162733751,
      "lm_loss": 1.1272,
      "step": 4840,
      "vm_loss": 0.1128
    },
    {
      "epoch": 0.3551088162733751,
      "lm_loss": 1.3847,
      "step": 4840,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.3551088162733751,
      "lm_loss": 0.054,
      "step": 4840,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.3551088162733751,
      "lm_loss": 0.4718,
      "step": 4840,
      "vm_loss": 0.067
    },
    {
      "epoch": 0.3551088162733751,
      "lm_loss": 0.892,
      "step": 4840,
      "vm_loss": 0.1185
    },
    {
      "epoch": 0.3551088162733751,
      "lm_loss": 0.4797,
      "step": 4840,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.3551088162733751,
      "lm_loss": 0.4687,
      "step": 4840,
      "vm_loss": 0.1675
    },
    {
      "epoch": 0.3551088162733751,
      "lm_loss": 0.5614,
      "step": 4840,
      "vm_loss": 0.1237
    },
    {
      "epoch": 0.35518218586351424,
      "grad_norm": 1.2987168524494805,
      "learning_rate": 7.473886633307621e-06,
      "loss": 0.7697,
      "step": 4841
    },
    {
      "epoch": 0.35525555545365334,
      "grad_norm": 1.5943045803780356,
      "learning_rate": 7.47285399682604e-06,
      "loss": 0.8319,
      "step": 4842
    },
    {
      "epoch": 0.3553289250437925,
      "grad_norm": 1.4427462992163649,
      "learning_rate": 7.471821220696268e-06,
      "loss": 0.9239,
      "step": 4843
    },
    {
      "epoch": 0.3554022946339316,
      "grad_norm": 0.6043743935282809,
      "learning_rate": 7.470788304976632e-06,
      "loss": 0.5039,
      "step": 4844
    },
    {
      "epoch": 0.35547566422407073,
      "grad_norm": 0.6209866020750094,
      "learning_rate": 7.469755249725461e-06,
      "loss": 0.5303,
      "step": 4845
    },
    {
      "epoch": 0.3555490338142099,
      "grad_norm": 1.608932151133279,
      "learning_rate": 7.4687220550010954e-06,
      "loss": 0.8524,
      "step": 4846
    },
    {
      "epoch": 0.355622403404349,
      "grad_norm": 1.5046219030793841,
      "learning_rate": 7.467688720861881e-06,
      "loss": 0.7672,
      "step": 4847
    },
    {
      "epoch": 0.3556957729944881,
      "grad_norm": 1.44987230244583,
      "learning_rate": 7.466655247366175e-06,
      "loss": 0.852,
      "step": 4848
    },
    {
      "epoch": 0.3556957729944881,
      "lm_loss": 1.4315,
      "step": 4848,
      "vm_loss": 0.0442
    },
    {
      "epoch": 0.3556957729944881,
      "lm_loss": 0.2748,
      "step": 4848,
      "vm_loss": 0.0572
    },
    {
      "epoch": 0.3556957729944881,
      "lm_loss": 0.2535,
      "step": 4848,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.3556957729944881,
      "lm_loss": 0.64,
      "step": 4848,
      "vm_loss": 0.0802
    },
    {
      "epoch": 0.3556957729944881,
      "lm_loss": 0.9759,
      "step": 4848,
      "vm_loss": 0.1775
    },
    {
      "epoch": 0.3556957729944881,
      "lm_loss": 0.0669,
      "step": 4848,
      "vm_loss": 0.2292
    },
    {
      "epoch": 0.3556957729944881,
      "lm_loss": 1.1481,
      "step": 4848,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.3556957729944881,
      "lm_loss": 0.6348,
      "step": 4848,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.3557691425846272,
      "grad_norm": 1.3689214974226407,
      "learning_rate": 7.465621634572336e-06,
      "loss": 0.7132,
      "step": 4849
    },
    {
      "epoch": 0.35584251217476637,
      "grad_norm": 0.6109217108357726,
      "learning_rate": 7.464587882538739e-06,
      "loss": 0.5243,
      "step": 4850
    },
    {
      "epoch": 0.3559158817649055,
      "grad_norm": 0.6057979291941705,
      "learning_rate": 7.463553991323761e-06,
      "loss": 0.5026,
      "step": 4851
    },
    {
      "epoch": 0.3559892513550446,
      "grad_norm": 0.617108097012837,
      "learning_rate": 7.462519960985787e-06,
      "loss": 0.5117,
      "step": 4852
    },
    {
      "epoch": 0.35606262094518376,
      "grad_norm": 1.588681477557239,
      "learning_rate": 7.461485791583214e-06,
      "loss": 0.8192,
      "step": 4853
    },
    {
      "epoch": 0.35613599053532285,
      "grad_norm": 1.5402425120642456,
      "learning_rate": 7.46045148317444e-06,
      "loss": 0.8973,
      "step": 4854
    },
    {
      "epoch": 0.356209360125462,
      "grad_norm": 0.6039905600377571,
      "learning_rate": 7.45941703581788e-06,
      "loss": 0.5259,
      "step": 4855
    },
    {
      "epoch": 0.3562827297156011,
      "grad_norm": 0.6330497136724603,
      "learning_rate": 7.4583824495719495e-06,
      "loss": 0.5229,
      "step": 4856
    },
    {
      "epoch": 0.3562827297156011,
      "lm_loss": 1.2619,
      "step": 4856,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.3562827297156011,
      "lm_loss": 1.4842,
      "step": 4856,
      "vm_loss": 0.1749
    },
    {
      "epoch": 0.3562827297156011,
      "lm_loss": 0.2617,
      "step": 4856,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.3562827297156011,
      "lm_loss": 0.2864,
      "step": 4856,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0.3562827297156011,
      "lm_loss": 0.1327,
      "step": 4856,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.3562827297156011,
      "lm_loss": 0.3989,
      "step": 4856,
      "vm_loss": 0.1153
    },
    {
      "epoch": 0.3562827297156011,
      "lm_loss": 1.0485,
      "step": 4856,
      "vm_loss": 0.2124
    },
    {
      "epoch": 0.3562827297156011,
      "lm_loss": 0.0421,
      "step": 4856,
      "vm_loss": 0.0974
    },
    {
      "epoch": 0.35635609930574025,
      "grad_norm": 1.4893282498188463,
      "learning_rate": 7.457347724495072e-06,
      "loss": 0.8931,
      "step": 4857
    },
    {
      "epoch": 0.3564294688958794,
      "grad_norm": 0.6086058680161233,
      "learning_rate": 7.4563128606456845e-06,
      "loss": 0.5112,
      "step": 4858
    },
    {
      "epoch": 0.3565028384860185,
      "grad_norm": 0.5483616608362997,
      "learning_rate": 7.455277858082227e-06,
      "loss": 0.517,
      "step": 4859
    },
    {
      "epoch": 0.35657620807615764,
      "grad_norm": 1.4242564379981175,
      "learning_rate": 7.454242716863147e-06,
      "loss": 0.7756,
      "step": 4860
    },
    {
      "epoch": 0.35664957766629674,
      "grad_norm": 0.58477943388791,
      "learning_rate": 7.453207437046904e-06,
      "loss": 0.5089,
      "step": 4861
    },
    {
      "epoch": 0.3567229472564359,
      "grad_norm": 0.6097379370817555,
      "learning_rate": 7.452172018691961e-06,
      "loss": 0.5285,
      "step": 4862
    },
    {
      "epoch": 0.35679631684657503,
      "grad_norm": 0.6046777914296831,
      "learning_rate": 7.451136461856793e-06,
      "loss": 0.512,
      "step": 4863
    },
    {
      "epoch": 0.35686968643671413,
      "grad_norm": 1.5039585831808295,
      "learning_rate": 7.4501007665998785e-06,
      "loss": 0.8294,
      "step": 4864
    },
    {
      "epoch": 0.35686968643671413,
      "lm_loss": 1.0624,
      "step": 4864,
      "vm_loss": 0.0389
    },
    {
      "epoch": 0.35686968643671413,
      "lm_loss": 0.8634,
      "step": 4864,
      "vm_loss": 0.1312
    },
    {
      "epoch": 0.35686968643671413,
      "lm_loss": 0.5859,
      "step": 4864,
      "vm_loss": 0.0474
    },
    {
      "epoch": 0.35686968643671413,
      "lm_loss": 0.9901,
      "step": 4864,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.35686968643671413,
      "lm_loss": 0.2526,
      "step": 4864,
      "vm_loss": 0.1023
    },
    {
      "epoch": 0.35686968643671413,
      "lm_loss": 0.6625,
      "step": 4864,
      "vm_loss": 0.1322
    },
    {
      "epoch": 0.35686968643671413,
      "lm_loss": 0.8542,
      "step": 4864,
      "vm_loss": 0.1081
    },
    {
      "epoch": 0.35686968643671413,
      "lm_loss": 0.8747,
      "step": 4864,
      "vm_loss": 0.1397
    },
    {
      "epoch": 0.3569430560268533,
      "grad_norm": 1.3433011715505554,
      "learning_rate": 7.4490649329797035e-06,
      "loss": 0.6884,
      "step": 4865
    },
    {
      "epoch": 0.35701642561699237,
      "grad_norm": 1.6363266616179057,
      "learning_rate": 7.448028961054769e-06,
      "loss": 0.8215,
      "step": 4866
    },
    {
      "epoch": 0.3570897952071315,
      "grad_norm": 1.5303480359287776,
      "learning_rate": 7.4469928508835775e-06,
      "loss": 0.8378,
      "step": 4867
    },
    {
      "epoch": 0.35716316479727067,
      "grad_norm": 1.4661647404096845,
      "learning_rate": 7.445956602524638e-06,
      "loss": 0.8114,
      "step": 4868
    },
    {
      "epoch": 0.35723653438740977,
      "grad_norm": 1.4532125324866942,
      "learning_rate": 7.444920216036473e-06,
      "loss": 0.8861,
      "step": 4869
    },
    {
      "epoch": 0.3573099039775489,
      "grad_norm": 1.4401435868139723,
      "learning_rate": 7.4438836914776065e-06,
      "loss": 0.7678,
      "step": 4870
    },
    {
      "epoch": 0.357383273567688,
      "grad_norm": 1.4475705396498841,
      "learning_rate": 7.442847028906579e-06,
      "loss": 0.8103,
      "step": 4871
    },
    {
      "epoch": 0.35745664315782716,
      "grad_norm": 1.6837808846058093,
      "learning_rate": 7.44181022838193e-06,
      "loss": 0.8971,
      "step": 4872
    },
    {
      "epoch": 0.35745664315782716,
      "lm_loss": 1.0746,
      "step": 4872,
      "vm_loss": 0.0773
    },
    {
      "epoch": 0.35745664315782716,
      "lm_loss": 0.2681,
      "step": 4872,
      "vm_loss": 0.3149
    },
    {
      "epoch": 0.35745664315782716,
      "lm_loss": 0.9461,
      "step": 4872,
      "vm_loss": 0.2356
    },
    {
      "epoch": 0.35745664315782716,
      "lm_loss": 0.2932,
      "step": 4872,
      "vm_loss": 0.1976
    },
    {
      "epoch": 0.35745664315782716,
      "lm_loss": 0.7591,
      "step": 4872,
      "vm_loss": 0.157
    },
    {
      "epoch": 0.35745664315782716,
      "lm_loss": 0.6921,
      "step": 4872,
      "vm_loss": 0.1206
    },
    {
      "epoch": 0.35745664315782716,
      "lm_loss": 0.606,
      "step": 4872,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.35745664315782716,
      "lm_loss": 0.8445,
      "step": 4872,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.3575300127479663,
      "grad_norm": 1.4491842883207018,
      "learning_rate": 7.440773289962207e-06,
      "loss": 0.7899,
      "step": 4873
    },
    {
      "epoch": 0.3576033823381054,
      "grad_norm": 1.4762738243166762,
      "learning_rate": 7.439736213705975e-06,
      "loss": 0.9167,
      "step": 4874
    },
    {
      "epoch": 0.35767675192824455,
      "grad_norm": 1.3220369876552058,
      "learning_rate": 7.438698999671796e-06,
      "loss": 0.8404,
      "step": 4875
    },
    {
      "epoch": 0.35775012151838365,
      "grad_norm": 0.8444680436395274,
      "learning_rate": 7.437661647918244e-06,
      "loss": 0.5508,
      "step": 4876
    },
    {
      "epoch": 0.3578234911085228,
      "grad_norm": 1.619112830850955,
      "learning_rate": 7.436624158503903e-06,
      "loss": 0.8145,
      "step": 4877
    },
    {
      "epoch": 0.35789686069866194,
      "grad_norm": 0.6556956910535505,
      "learning_rate": 7.435586531487362e-06,
      "loss": 0.5145,
      "step": 4878
    },
    {
      "epoch": 0.35797023028880104,
      "grad_norm": 1.5783045166677618,
      "learning_rate": 7.434548766927219e-06,
      "loss": 0.8099,
      "step": 4879
    },
    {
      "epoch": 0.3580435998789402,
      "grad_norm": 1.482706753103602,
      "learning_rate": 7.433510864882076e-06,
      "loss": 0.8066,
      "step": 4880
    },
    {
      "epoch": 0.3580435998789402,
      "lm_loss": 1.5071,
      "step": 4880,
      "vm_loss": 0.1084
    },
    {
      "epoch": 0.3580435998789402,
      "lm_loss": 0.4709,
      "step": 4880,
      "vm_loss": 0.0773
    },
    {
      "epoch": 0.3580435998789402,
      "lm_loss": 0.7355,
      "step": 4880,
      "vm_loss": 0.1838
    },
    {
      "epoch": 0.3580435998789402,
      "lm_loss": 0.0544,
      "step": 4880,
      "vm_loss": 0.0863
    },
    {
      "epoch": 0.3580435998789402,
      "lm_loss": 0.238,
      "step": 4880,
      "vm_loss": 0.2135
    },
    {
      "epoch": 0.3580435998789402,
      "lm_loss": 0.0631,
      "step": 4880,
      "vm_loss": 0.0737
    },
    {
      "epoch": 0.3580435998789402,
      "lm_loss": 0.9383,
      "step": 4880,
      "vm_loss": 0.1688
    },
    {
      "epoch": 0.3580435998789402,
      "lm_loss": 0.3516,
      "step": 4880,
      "vm_loss": 0.064
    },
    {
      "epoch": 0.3581169694690793,
      "grad_norm": 1.6859032914814385,
      "learning_rate": 7.432472825410548e-06,
      "loss": 0.7622,
      "step": 4881
    },
    {
      "epoch": 0.35819033905921843,
      "grad_norm": 1.4556399345823376,
      "learning_rate": 7.431434648571258e-06,
      "loss": 0.7966,
      "step": 4882
    },
    {
      "epoch": 0.3582637086493576,
      "grad_norm": 1.4830210983557368,
      "learning_rate": 7.430396334422829e-06,
      "loss": 0.8678,
      "step": 4883
    },
    {
      "epoch": 0.3583370782394967,
      "grad_norm": 1.735273430254466,
      "learning_rate": 7.429357883023902e-06,
      "loss": 0.9822,
      "step": 4884
    },
    {
      "epoch": 0.3584104478296358,
      "grad_norm": 1.5283327019524888,
      "learning_rate": 7.428319294433118e-06,
      "loss": 0.7877,
      "step": 4885
    },
    {
      "epoch": 0.3584838174197749,
      "grad_norm": 1.344653269023588,
      "learning_rate": 7.42728056870913e-06,
      "loss": 0.8167,
      "step": 4886
    },
    {
      "epoch": 0.35855718700991407,
      "grad_norm": 1.4715981609780078,
      "learning_rate": 7.426241705910599e-06,
      "loss": 0.8552,
      "step": 4887
    },
    {
      "epoch": 0.3586305566000532,
      "grad_norm": 1.1005622828153228,
      "learning_rate": 7.425202706096188e-06,
      "loss": 0.5281,
      "step": 4888
    },
    {
      "epoch": 0.3586305566000532,
      "lm_loss": 1.5606,
      "step": 4888,
      "vm_loss": 0.0856
    },
    {
      "epoch": 0.3586305566000532,
      "lm_loss": 1.4118,
      "step": 4888,
      "vm_loss": 0.1666
    },
    {
      "epoch": 0.3586305566000532,
      "lm_loss": 0.4323,
      "step": 4888,
      "vm_loss": 0.0533
    },
    {
      "epoch": 0.3586305566000532,
      "lm_loss": 0.6225,
      "step": 4888,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.3586305566000532,
      "lm_loss": 1.0494,
      "step": 4888,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.3586305566000532,
      "lm_loss": 0.2395,
      "step": 4888,
      "vm_loss": 0.1492
    },
    {
      "epoch": 0.3586305566000532,
      "lm_loss": 0.6175,
      "step": 4888,
      "vm_loss": 0.2421
    },
    {
      "epoch": 0.3586305566000532,
      "lm_loss": 0.1169,
      "step": 4888,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.3587039261901923,
      "grad_norm": 1.4626639943458417,
      "learning_rate": 7.4241635693245766e-06,
      "loss": 0.819,
      "step": 4889
    },
    {
      "epoch": 0.35877729578033146,
      "grad_norm": 0.6596756586051482,
      "learning_rate": 7.423124295654444e-06,
      "loss": 0.5117,
      "step": 4890
    },
    {
      "epoch": 0.35885066537047056,
      "grad_norm": 1.604629984503066,
      "learning_rate": 7.422084885144481e-06,
      "loss": 0.7586,
      "step": 4891
    },
    {
      "epoch": 0.3589240349606097,
      "grad_norm": 1.725296007133786,
      "learning_rate": 7.421045337853387e-06,
      "loss": 0.8743,
      "step": 4892
    },
    {
      "epoch": 0.35899740455074886,
      "grad_norm": 1.4127531713627355,
      "learning_rate": 7.420005653839867e-06,
      "loss": 0.7783,
      "step": 4893
    },
    {
      "epoch": 0.35907077414088795,
      "grad_norm": 1.600730158900239,
      "learning_rate": 7.418965833162633e-06,
      "loss": 0.8778,
      "step": 4894
    },
    {
      "epoch": 0.3591441437310271,
      "grad_norm": 1.435029298953874,
      "learning_rate": 7.417925875880411e-06,
      "loss": 0.7535,
      "step": 4895
    },
    {
      "epoch": 0.3592175133211662,
      "grad_norm": 1.0625804205240397,
      "learning_rate": 7.416885782051926e-06,
      "loss": 0.5232,
      "step": 4896
    },
    {
      "epoch": 0.3592175133211662,
      "lm_loss": 0.9312,
      "step": 4896,
      "vm_loss": 0.0949
    },
    {
      "epoch": 0.3592175133211662,
      "lm_loss": 1.362,
      "step": 4896,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.3592175133211662,
      "lm_loss": 0.042,
      "step": 4896,
      "vm_loss": 0.0642
    },
    {
      "epoch": 0.3592175133211662,
      "lm_loss": 0.6396,
      "step": 4896,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.3592175133211662,
      "lm_loss": 0.6357,
      "step": 4896,
      "vm_loss": 0.1474
    },
    {
      "epoch": 0.3592175133211662,
      "lm_loss": 1.2196,
      "step": 4896,
      "vm_loss": 0.0895
    },
    {
      "epoch": 0.3592175133211662,
      "lm_loss": 0.0826,
      "step": 4896,
      "vm_loss": 0.1231
    },
    {
      "epoch": 0.3592175133211662,
      "lm_loss": 0.6644,
      "step": 4896,
      "vm_loss": 0.2509
    },
    {
      "epoch": 0.35929088291130534,
      "grad_norm": 1.4700923982919891,
      "learning_rate": 7.415845551735914e-06,
      "loss": 0.9004,
      "step": 4897
    },
    {
      "epoch": 0.35936425250144444,
      "grad_norm": 1.487345173335547,
      "learning_rate": 7.414805184991123e-06,
      "loss": 0.8429,
      "step": 4898
    },
    {
      "epoch": 0.3594376220915836,
      "grad_norm": 0.6498287336340824,
      "learning_rate": 7.413764681876302e-06,
      "loss": 0.5146,
      "step": 4899
    },
    {
      "epoch": 0.35951099168172274,
      "grad_norm": 2.4502273394393668,
      "learning_rate": 7.4127240424502115e-06,
      "loss": 0.8665,
      "step": 4900
    },
    {
      "epoch": 0.35958436127186183,
      "grad_norm": 1.8730124245423125,
      "learning_rate": 7.41168326677162e-06,
      "loss": 0.7717,
      "step": 4901
    },
    {
      "epoch": 0.359657730862001,
      "grad_norm": 1.8007721634788345,
      "learning_rate": 7.410642354899301e-06,
      "loss": 0.7527,
      "step": 4902
    },
    {
      "epoch": 0.3597311004521401,
      "grad_norm": 0.7100304667563655,
      "learning_rate": 7.409601306892039e-06,
      "loss": 0.5293,
      "step": 4903
    },
    {
      "epoch": 0.3598044700422792,
      "grad_norm": 0.7145487086258859,
      "learning_rate": 7.408560122808625e-06,
      "loss": 0.5185,
      "step": 4904
    },
    {
      "epoch": 0.3598044700422792,
      "lm_loss": 1.3796,
      "step": 4904,
      "vm_loss": 0.0516
    },
    {
      "epoch": 0.3598044700422792,
      "lm_loss": 0.7585,
      "step": 4904,
      "vm_loss": 0.1192
    },
    {
      "epoch": 0.3598044700422792,
      "lm_loss": 0.1464,
      "step": 4904,
      "vm_loss": 0.17
    },
    {
      "epoch": 0.3598044700422792,
      "lm_loss": 1.4931,
      "step": 4904,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.3598044700422792,
      "lm_loss": 0.7404,
      "step": 4904,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.3598044700422792,
      "lm_loss": 0.4782,
      "step": 4904,
      "vm_loss": 0.1782
    },
    {
      "epoch": 0.3598044700422792,
      "lm_loss": 1.0222,
      "step": 4904,
      "vm_loss": 0.1915
    },
    {
      "epoch": 0.3598044700422792,
      "lm_loss": 1.5518,
      "step": 4904,
      "vm_loss": 0.196
    },
    {
      "epoch": 0.3598778396324184,
      "grad_norm": 1.499666926297249,
      "learning_rate": 7.407518802707854e-06,
      "loss": 0.7942,
      "step": 4905
    },
    {
      "epoch": 0.35995120922255747,
      "grad_norm": 1.4166672276601575,
      "learning_rate": 7.406477346648535e-06,
      "loss": 0.7003,
      "step": 4906
    },
    {
      "epoch": 0.3600245788126966,
      "grad_norm": 0.6097249455726775,
      "learning_rate": 7.405435754689482e-06,
      "loss": 0.5255,
      "step": 4907
    },
    {
      "epoch": 0.3600979484028357,
      "grad_norm": 1.5365894588324809,
      "learning_rate": 7.404394026889513e-06,
      "loss": 0.7914,
      "step": 4908
    },
    {
      "epoch": 0.36017131799297486,
      "grad_norm": 1.6286382279755693,
      "learning_rate": 7.40335216330746e-06,
      "loss": 0.7658,
      "step": 4909
    },
    {
      "epoch": 0.360244687583114,
      "grad_norm": 1.4083961219729701,
      "learning_rate": 7.402310164002158e-06,
      "loss": 0.7781,
      "step": 4910
    },
    {
      "epoch": 0.3603180571732531,
      "grad_norm": 0.7145790912732171,
      "learning_rate": 7.401268029032454e-06,
      "loss": 0.5106,
      "step": 4911
    },
    {
      "epoch": 0.36039142676339225,
      "grad_norm": 1.5565182442058494,
      "learning_rate": 7.400225758457196e-06,
      "loss": 0.7477,
      "step": 4912
    },
    {
      "epoch": 0.36039142676339225,
      "lm_loss": 1.6995,
      "step": 4912,
      "vm_loss": 0.0367
    },
    {
      "epoch": 0.36039142676339225,
      "lm_loss": 1.7225,
      "step": 4912,
      "vm_loss": 0.1384
    },
    {
      "epoch": 0.36039142676339225,
      "lm_loss": 0.1568,
      "step": 4912,
      "vm_loss": 0.1358
    },
    {
      "epoch": 0.36039142676339225,
      "lm_loss": 1.1658,
      "step": 4912,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.36039142676339225,
      "lm_loss": 0.7697,
      "step": 4912,
      "vm_loss": 0.1621
    },
    {
      "epoch": 0.36039142676339225,
      "lm_loss": 0.0967,
      "step": 4912,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.36039142676339225,
      "lm_loss": 0.466,
      "step": 4912,
      "vm_loss": 0.1334
    },
    {
      "epoch": 0.36039142676339225,
      "lm_loss": 0.6562,
      "step": 4912,
      "vm_loss": 0.1711
    },
    {
      "epoch": 0.36046479635353135,
      "grad_norm": 1.3374759012982824,
      "learning_rate": 7.399183352335244e-06,
      "loss": 0.8151,
      "step": 4913
    },
    {
      "epoch": 0.3605381659436705,
      "grad_norm": 1.6438450913945144,
      "learning_rate": 7.3981408107254684e-06,
      "loss": 0.8854,
      "step": 4914
    },
    {
      "epoch": 0.36061153553380965,
      "grad_norm": 0.5788238380510186,
      "learning_rate": 7.397098133686742e-06,
      "loss": 0.5234,
      "step": 4915
    },
    {
      "epoch": 0.36068490512394874,
      "grad_norm": 1.4628474006865513,
      "learning_rate": 7.3960553212779465e-06,
      "loss": 0.8168,
      "step": 4916
    },
    {
      "epoch": 0.3607582747140879,
      "grad_norm": 0.5548756328756068,
      "learning_rate": 7.395012373557975e-06,
      "loss": 0.5045,
      "step": 4917
    },
    {
      "epoch": 0.360831644304227,
      "grad_norm": 1.428721331553038,
      "learning_rate": 7.393969290585723e-06,
      "loss": 0.775,
      "step": 4918
    },
    {
      "epoch": 0.36090501389436613,
      "grad_norm": 1.4565532824322398,
      "learning_rate": 7.392926072420097e-06,
      "loss": 0.8149,
      "step": 4919
    },
    {
      "epoch": 0.3609783834845053,
      "grad_norm": 1.8347060901460788,
      "learning_rate": 7.391882719120009e-06,
      "loss": 0.821,
      "step": 4920
    },
    {
      "epoch": 0.3609783834845053,
      "lm_loss": 0.2213,
      "step": 4920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3609783834845053,
      "lm_loss": 0.374,
      "step": 4920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3609783834845053,
      "lm_loss": 0.5885,
      "step": 4920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3609783834845053,
      "lm_loss": 0.5417,
      "step": 4920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3609783834845053,
      "lm_loss": 0.6102,
      "step": 4920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3609783834845053,
      "lm_loss": 0.5936,
      "step": 4920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3609783834845053,
      "lm_loss": 0.482,
      "step": 4920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3609783834845053,
      "lm_loss": 0.6704,
      "step": 4920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3610517530746444,
      "grad_norm": 0.6136873371091776,
      "learning_rate": 7.39083923074438e-06,
      "loss": 0.5227,
      "step": 4921
    },
    {
      "epoch": 0.3611251226647835,
      "grad_norm": 0.5922171460899374,
      "learning_rate": 7.389795607352139e-06,
      "loss": 0.5114,
      "step": 4922
    },
    {
      "epoch": 0.3611984922549226,
      "grad_norm": 0.5846724812441861,
      "learning_rate": 7.388751849002224e-06,
      "loss": 0.52,
      "step": 4923
    },
    {
      "epoch": 0.36127186184506177,
      "grad_norm": 0.5453487531094197,
      "learning_rate": 7.387707955753574e-06,
      "loss": 0.5124,
      "step": 4924
    },
    {
      "epoch": 0.3613452314352009,
      "grad_norm": 1.7476432386057899,
      "learning_rate": 7.386663927665143e-06,
      "loss": 0.7815,
      "step": 4925
    },
    {
      "epoch": 0.36141860102534,
      "grad_norm": 1.643101988637567,
      "learning_rate": 7.38561976479589e-06,
      "loss": 0.9329,
      "step": 4926
    },
    {
      "epoch": 0.36149197061547916,
      "grad_norm": 1.4371658703269865,
      "learning_rate": 7.384575467204781e-06,
      "loss": 0.8727,
      "step": 4927
    },
    {
      "epoch": 0.36156534020561826,
      "grad_norm": 0.7036304719048531,
      "learning_rate": 7.3835310349507905e-06,
      "loss": 0.5267,
      "step": 4928
    },
    {
      "epoch": 0.36156534020561826,
      "lm_loss": 0.6805,
      "step": 4928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36156534020561826,
      "lm_loss": 0.5108,
      "step": 4928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36156534020561826,
      "lm_loss": 0.4579,
      "step": 4928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36156534020561826,
      "lm_loss": 0.4665,
      "step": 4928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36156534020561826,
      "lm_loss": 0.4422,
      "step": 4928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36156534020561826,
      "lm_loss": 0.4931,
      "step": 4928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36156534020561826,
      "lm_loss": 0.5493,
      "step": 4928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36156534020561826,
      "lm_loss": 0.4338,
      "step": 4928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3616387097957574,
      "grad_norm": 0.7111587360254545,
      "learning_rate": 7.382486468092899e-06,
      "loss": 0.524,
      "step": 4929
    },
    {
      "epoch": 0.36171207938589656,
      "grad_norm": 1.6322542849800756,
      "learning_rate": 7.381441766690096e-06,
      "loss": 0.7415,
      "step": 4930
    },
    {
      "epoch": 0.36178544897603565,
      "grad_norm": 1.76178012909602,
      "learning_rate": 7.3803969308013814e-06,
      "loss": 0.8342,
      "step": 4931
    },
    {
      "epoch": 0.3618588185661748,
      "grad_norm": 0.5760738038018526,
      "learning_rate": 7.379351960485754e-06,
      "loss": 0.5147,
      "step": 4932
    },
    {
      "epoch": 0.3619321881563139,
      "grad_norm": 0.598020127461025,
      "learning_rate": 7.378306855802231e-06,
      "loss": 0.5056,
      "step": 4933
    },
    {
      "epoch": 0.36200555774645304,
      "grad_norm": 0.5809232166029592,
      "learning_rate": 7.37726161680983e-06,
      "loss": 0.511,
      "step": 4934
    },
    {
      "epoch": 0.3620789273365922,
      "grad_norm": 1.8523526196056799,
      "learning_rate": 7.376216243567578e-06,
      "loss": 0.8715,
      "step": 4935
    },
    {
      "epoch": 0.3621522969267313,
      "grad_norm": 1.6487538965483879,
      "learning_rate": 7.3751707361345095e-06,
      "loss": 0.7693,
      "step": 4936
    },
    {
      "epoch": 0.3621522969267313,
      "lm_loss": 0.2057,
      "step": 4936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3621522969267313,
      "lm_loss": 0.5166,
      "step": 4936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3621522969267313,
      "lm_loss": 0.4598,
      "step": 4936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3621522969267313,
      "lm_loss": 0.5459,
      "step": 4936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3621522969267313,
      "lm_loss": 0.4374,
      "step": 4936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3621522969267313,
      "lm_loss": 0.4003,
      "step": 4936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3621522969267313,
      "lm_loss": 0.4528,
      "step": 4936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3621522969267313,
      "lm_loss": 0.4133,
      "step": 4936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36222566651687044,
      "grad_norm": 0.5466698523815284,
      "learning_rate": 7.374125094569668e-06,
      "loss": 0.5081,
      "step": 4937
    },
    {
      "epoch": 0.36229903610700953,
      "grad_norm": 0.5787937662604644,
      "learning_rate": 7.3730793189321025e-06,
      "loss": 0.5233,
      "step": 4938
    },
    {
      "epoch": 0.3623724056971487,
      "grad_norm": 0.5753884364899978,
      "learning_rate": 7.372033409280872e-06,
      "loss": 0.5353,
      "step": 4939
    },
    {
      "epoch": 0.3624457752872878,
      "grad_norm": 2.2199992035936913,
      "learning_rate": 7.3709873656750396e-06,
      "loss": 0.7931,
      "step": 4940
    },
    {
      "epoch": 0.3625191448774269,
      "grad_norm": 2.074066749043749,
      "learning_rate": 7.36994118817368e-06,
      "loss": 0.7812,
      "step": 4941
    },
    {
      "epoch": 0.3625925144675661,
      "grad_norm": 1.402616066788165,
      "learning_rate": 7.368894876835873e-06,
      "loss": 0.7691,
      "step": 4942
    },
    {
      "epoch": 0.36266588405770517,
      "grad_norm": 0.5960210780618973,
      "learning_rate": 7.367848431720703e-06,
      "loss": 0.5109,
      "step": 4943
    },
    {
      "epoch": 0.3627392536478443,
      "grad_norm": 2.003924062661456,
      "learning_rate": 7.366801852887271e-06,
      "loss": 0.8558,
      "step": 4944
    },
    {
      "epoch": 0.3627392536478443,
      "lm_loss": 1.1824,
      "step": 4944,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.3627392536478443,
      "lm_loss": 1.4409,
      "step": 4944,
      "vm_loss": 0.1
    },
    {
      "epoch": 0.3627392536478443,
      "lm_loss": 1.1487,
      "step": 4944,
      "vm_loss": 0.1352
    },
    {
      "epoch": 0.3627392536478443,
      "lm_loss": 0.4138,
      "step": 4944,
      "vm_loss": 0.2017
    },
    {
      "epoch": 0.3627392536478443,
      "lm_loss": 0.1032,
      "step": 4944,
      "vm_loss": 0.1601
    },
    {
      "epoch": 0.3627392536478443,
      "lm_loss": 0.6631,
      "step": 4944,
      "vm_loss": 0.1162
    },
    {
      "epoch": 0.3627392536478443,
      "lm_loss": 1.2744,
      "step": 4944,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.3627392536478443,
      "lm_loss": 0.8469,
      "step": 4944,
      "vm_loss": 0.2083
    },
    {
      "epoch": 0.3628126232379834,
      "grad_norm": 2.096750904635911,
      "learning_rate": 7.365755140394676e-06,
      "loss": 0.8844,
      "step": 4945
    },
    {
      "epoch": 0.36288599282812256,
      "grad_norm": 1.452253183525648,
      "learning_rate": 7.36470829430203e-06,
      "loss": 0.8098,
      "step": 4946
    },
    {
      "epoch": 0.3629593624182617,
      "grad_norm": 1.7997289295117704,
      "learning_rate": 7.363661314668451e-06,
      "loss": 0.7825,
      "step": 4947
    },
    {
      "epoch": 0.3630327320084008,
      "grad_norm": 1.5539009678245583,
      "learning_rate": 7.362614201553062e-06,
      "loss": 0.8503,
      "step": 4948
    },
    {
      "epoch": 0.36310610159853995,
      "grad_norm": 1.7240276558725476,
      "learning_rate": 7.361566955014999e-06,
      "loss": 0.7259,
      "step": 4949
    },
    {
      "epoch": 0.36317947118867905,
      "grad_norm": 1.685766725464822,
      "learning_rate": 7.360519575113403e-06,
      "loss": 0.8719,
      "step": 4950
    },
    {
      "epoch": 0.3632528407788182,
      "grad_norm": 1.5287764282090124,
      "learning_rate": 7.35947206190742e-06,
      "loss": 0.7797,
      "step": 4951
    },
    {
      "epoch": 0.36332621036895735,
      "grad_norm": 1.6749068020094375,
      "learning_rate": 7.358424415456207e-06,
      "loss": 0.8525,
      "step": 4952
    },
    {
      "epoch": 0.36332621036895735,
      "lm_loss": 0.0397,
      "step": 4952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36332621036895735,
      "lm_loss": 0.5138,
      "step": 4952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36332621036895735,
      "lm_loss": 0.4666,
      "step": 4952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36332621036895735,
      "lm_loss": 0.4388,
      "step": 4952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36332621036895735,
      "lm_loss": 0.4434,
      "step": 4952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36332621036895735,
      "lm_loss": 0.5736,
      "step": 4952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36332621036895735,
      "lm_loss": 0.4723,
      "step": 4952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36332621036895735,
      "lm_loss": 0.4844,
      "step": 4952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36339957995909644,
      "grad_norm": 0.6652246025352111,
      "learning_rate": 7.357376635818924e-06,
      "loss": 0.4999,
      "step": 4953
    },
    {
      "epoch": 0.3634729495492356,
      "grad_norm": 0.6187966663496205,
      "learning_rate": 7.356328723054748e-06,
      "loss": 0.5328,
      "step": 4954
    },
    {
      "epoch": 0.3635463191393747,
      "grad_norm": 0.5437020723195403,
      "learning_rate": 7.355280677222853e-06,
      "loss": 0.5088,
      "step": 4955
    },
    {
      "epoch": 0.36361968872951383,
      "grad_norm": 1.9778136119920913,
      "learning_rate": 7.354232498382424e-06,
      "loss": 0.7912,
      "step": 4956
    },
    {
      "epoch": 0.363693058319653,
      "grad_norm": 1.7138195207389118,
      "learning_rate": 7.3531841865926564e-06,
      "loss": 0.8429,
      "step": 4957
    },
    {
      "epoch": 0.3637664279097921,
      "grad_norm": 1.502386181856719,
      "learning_rate": 7.35213574191275e-06,
      "loss": 0.7719,
      "step": 4958
    },
    {
      "epoch": 0.36383979749993123,
      "grad_norm": 1.6323198538551478,
      "learning_rate": 7.351087164401914e-06,
      "loss": 0.8968,
      "step": 4959
    },
    {
      "epoch": 0.3639131670900703,
      "grad_norm": 1.6136319861019286,
      "learning_rate": 7.350038454119364e-06,
      "loss": 0.8666,
      "step": 4960
    },
    {
      "epoch": 0.3639131670900703,
      "lm_loss": 1.8061,
      "step": 4960,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.3639131670900703,
      "lm_loss": 0.1685,
      "step": 4960,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.3639131670900703,
      "lm_loss": 1.2101,
      "step": 4960,
      "vm_loss": 0.1994
    },
    {
      "epoch": 0.3639131670900703,
      "lm_loss": 1.0401,
      "step": 4960,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.3639131670900703,
      "lm_loss": 0.6069,
      "step": 4960,
      "vm_loss": 0.1467
    },
    {
      "epoch": 0.3639131670900703,
      "lm_loss": 0.4405,
      "step": 4960,
      "vm_loss": 0.0403
    },
    {
      "epoch": 0.3639131670900703,
      "lm_loss": 0.9328,
      "step": 4960,
      "vm_loss": 0.153
    },
    {
      "epoch": 0.3639131670900703,
      "lm_loss": 0.6608,
      "step": 4960,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.36398653668020947,
      "grad_norm": 1.5028915436716717,
      "learning_rate": 7.34898961112432e-06,
      "loss": 0.7372,
      "step": 4961
    },
    {
      "epoch": 0.3640599062703486,
      "grad_norm": 1.7245381449617587,
      "learning_rate": 7.347940635476019e-06,
      "loss": 0.7485,
      "step": 4962
    },
    {
      "epoch": 0.3641332758604877,
      "grad_norm": 0.7593671534744618,
      "learning_rate": 7.346891527233692e-06,
      "loss": 0.5392,
      "step": 4963
    },
    {
      "epoch": 0.36420664545062686,
      "grad_norm": 1.484487676366324,
      "learning_rate": 7.34584228645659e-06,
      "loss": 0.7257,
      "step": 4964
    },
    {
      "epoch": 0.36428001504076596,
      "grad_norm": 1.5640040078478115,
      "learning_rate": 7.344792913203967e-06,
      "loss": 0.7969,
      "step": 4965
    },
    {
      "epoch": 0.3643533846309051,
      "grad_norm": 0.6146513912495105,
      "learning_rate": 7.343743407535078e-06,
      "loss": 0.5259,
      "step": 4966
    },
    {
      "epoch": 0.36442675422104426,
      "grad_norm": 1.5779487328122763,
      "learning_rate": 7.342693769509197e-06,
      "loss": 0.7775,
      "step": 4967
    },
    {
      "epoch": 0.36450012381118335,
      "grad_norm": 1.5944195389842646,
      "learning_rate": 7.341643999185596e-06,
      "loss": 0.782,
      "step": 4968
    },
    {
      "epoch": 0.36450012381118335,
      "lm_loss": 1.9383,
      "step": 4968,
      "vm_loss": 0.0498
    },
    {
      "epoch": 0.36450012381118335,
      "lm_loss": 0.0455,
      "step": 4968,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.36450012381118335,
      "lm_loss": 0.7617,
      "step": 4968,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.36450012381118335,
      "lm_loss": 0.3658,
      "step": 4968,
      "vm_loss": 0.0943
    },
    {
      "epoch": 0.36450012381118335,
      "lm_loss": 0.3886,
      "step": 4968,
      "vm_loss": 0.0798
    },
    {
      "epoch": 0.36450012381118335,
      "lm_loss": 0.0275,
      "step": 4968,
      "vm_loss": 0.1233
    },
    {
      "epoch": 0.36450012381118335,
      "lm_loss": 0.275,
      "step": 4968,
      "vm_loss": 0.1824
    },
    {
      "epoch": 0.36450012381118335,
      "lm_loss": 0.1954,
      "step": 4968,
      "vm_loss": 0.1176
    },
    {
      "epoch": 0.3645734934013225,
      "grad_norm": 1.4396878553108674,
      "learning_rate": 7.340594096623559e-06,
      "loss": 0.7105,
      "step": 4969
    },
    {
      "epoch": 0.3646468629914616,
      "grad_norm": 1.5269109962763845,
      "learning_rate": 7.339544061882377e-06,
      "loss": 0.8475,
      "step": 4970
    },
    {
      "epoch": 0.36472023258160075,
      "grad_norm": 1.5679092443186728,
      "learning_rate": 7.33849389502135e-06,
      "loss": 0.8187,
      "step": 4971
    },
    {
      "epoch": 0.3647936021717399,
      "grad_norm": 0.6479381708655816,
      "learning_rate": 7.33744359609978e-06,
      "loss": 0.5107,
      "step": 4972
    },
    {
      "epoch": 0.364866971761879,
      "grad_norm": 1.5078878301820315,
      "learning_rate": 7.336393165176983e-06,
      "loss": 0.7454,
      "step": 4973
    },
    {
      "epoch": 0.36494034135201814,
      "grad_norm": 1.5866590414808852,
      "learning_rate": 7.3353426023122765e-06,
      "loss": 0.8509,
      "step": 4974
    },
    {
      "epoch": 0.36501371094215723,
      "grad_norm": 1.522585009571749,
      "learning_rate": 7.33429190756499e-06,
      "loss": 0.8331,
      "step": 4975
    },
    {
      "epoch": 0.3650870805322964,
      "grad_norm": 1.5133130635743823,
      "learning_rate": 7.3332410809944585e-06,
      "loss": 0.8684,
      "step": 4976
    },
    {
      "epoch": 0.3650870805322964,
      "lm_loss": 1.3049,
      "step": 4976,
      "vm_loss": 0.0856
    },
    {
      "epoch": 0.3650870805322964,
      "lm_loss": 0.3273,
      "step": 4976,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.3650870805322964,
      "lm_loss": 0.6332,
      "step": 4976,
      "vm_loss": 0.1865
    },
    {
      "epoch": 0.3650870805322964,
      "lm_loss": 0.5745,
      "step": 4976,
      "vm_loss": 0.1619
    },
    {
      "epoch": 0.3650870805322964,
      "lm_loss": 0.6674,
      "step": 4976,
      "vm_loss": 0.0645
    },
    {
      "epoch": 0.3650870805322964,
      "lm_loss": 0.1925,
      "step": 4976,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.3650870805322964,
      "lm_loss": 1.3597,
      "step": 4976,
      "vm_loss": 0.2156
    },
    {
      "epoch": 0.3650870805322964,
      "lm_loss": 0.1871,
      "step": 4976,
      "vm_loss": 0.1549
    },
    {
      "epoch": 0.3651604501224355,
      "grad_norm": 1.6461978721723798,
      "learning_rate": 7.332190122660024e-06,
      "loss": 0.7974,
      "step": 4977
    },
    {
      "epoch": 0.3652338197125746,
      "grad_norm": 1.4936909064103137,
      "learning_rate": 7.33113903262104e-06,
      "loss": 0.9328,
      "step": 4978
    },
    {
      "epoch": 0.3653071893027138,
      "grad_norm": 0.6034272090989378,
      "learning_rate": 7.330087810936861e-06,
      "loss": 0.5302,
      "step": 4979
    },
    {
      "epoch": 0.36538055889285287,
      "grad_norm": 1.5598704605692697,
      "learning_rate": 7.329036457666853e-06,
      "loss": 0.7624,
      "step": 4980
    },
    {
      "epoch": 0.365453928482992,
      "grad_norm": 1.45403669290184,
      "learning_rate": 7.3279849728703874e-06,
      "loss": 0.7998,
      "step": 4981
    },
    {
      "epoch": 0.3655272980731311,
      "grad_norm": 1.687795491944054,
      "learning_rate": 7.326933356606847e-06,
      "loss": 0.8919,
      "step": 4982
    },
    {
      "epoch": 0.36560066766327026,
      "grad_norm": 1.6688448763732513,
      "learning_rate": 7.325881608935616e-06,
      "loss": 0.9714,
      "step": 4983
    },
    {
      "epoch": 0.3656740372534094,
      "grad_norm": 0.5550841964303413,
      "learning_rate": 7.324829729916092e-06,
      "loss": 0.5193,
      "step": 4984
    },
    {
      "epoch": 0.3656740372534094,
      "lm_loss": 1.3114,
      "step": 4984,
      "vm_loss": 0.1434
    },
    {
      "epoch": 0.3656740372534094,
      "lm_loss": 0.2483,
      "step": 4984,
      "vm_loss": 0.1042
    },
    {
      "epoch": 0.3656740372534094,
      "lm_loss": 0.1454,
      "step": 4984,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.3656740372534094,
      "lm_loss": 0.777,
      "step": 4984,
      "vm_loss": 0.1725
    },
    {
      "epoch": 0.3656740372534094,
      "lm_loss": 0.8252,
      "step": 4984,
      "vm_loss": 0.1533
    },
    {
      "epoch": 0.3656740372534094,
      "lm_loss": 1.2226,
      "step": 4984,
      "vm_loss": 0.0986
    },
    {
      "epoch": 0.3656740372534094,
      "lm_loss": 0.7267,
      "step": 4984,
      "vm_loss": 0.1562
    },
    {
      "epoch": 0.3656740372534094,
      "lm_loss": 0.1216,
      "step": 4984,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.3657474068435485,
      "grad_norm": 1.5825920087057632,
      "learning_rate": 7.323777719607674e-06,
      "loss": 0.7676,
      "step": 4985
    },
    {
      "epoch": 0.36582077643368766,
      "grad_norm": 1.653348544648081,
      "learning_rate": 7.322725578069775e-06,
      "loss": 0.7467,
      "step": 4986
    },
    {
      "epoch": 0.36589414602382675,
      "grad_norm": 1.561887093940873,
      "learning_rate": 7.321673305361811e-06,
      "loss": 0.8556,
      "step": 4987
    },
    {
      "epoch": 0.3659675156139659,
      "grad_norm": 0.541354404930762,
      "learning_rate": 7.320620901543202e-06,
      "loss": 0.4902,
      "step": 4988
    },
    {
      "epoch": 0.36604088520410505,
      "grad_norm": 1.4939988446162529,
      "learning_rate": 7.319568366673389e-06,
      "loss": 0.868,
      "step": 4989
    },
    {
      "epoch": 0.36611425479424414,
      "grad_norm": 0.5690897662173133,
      "learning_rate": 7.318515700811803e-06,
      "loss": 0.5143,
      "step": 4990
    },
    {
      "epoch": 0.3661876243843833,
      "grad_norm": 0.5589699115482986,
      "learning_rate": 7.3174629040178935e-06,
      "loss": 0.4993,
      "step": 4991
    },
    {
      "epoch": 0.3662609939745224,
      "grad_norm": 1.5708061117354655,
      "learning_rate": 7.316409976351117e-06,
      "loss": 0.7924,
      "step": 4992
    },
    {
      "epoch": 0.3662609939745224,
      "lm_loss": 1.9179,
      "step": 4992,
      "vm_loss": 0.0672
    },
    {
      "epoch": 0.3662609939745224,
      "lm_loss": 0.6147,
      "step": 4992,
      "vm_loss": 0.0531
    },
    {
      "epoch": 0.3662609939745224,
      "lm_loss": 1.0639,
      "step": 4992,
      "vm_loss": 0.2581
    },
    {
      "epoch": 0.3662609939745224,
      "lm_loss": 0.035,
      "step": 4992,
      "vm_loss": 0.14
    },
    {
      "epoch": 0.3662609939745224,
      "lm_loss": 1.0407,
      "step": 4992,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.3662609939745224,
      "lm_loss": 0.6225,
      "step": 4992,
      "vm_loss": 0.1428
    },
    {
      "epoch": 0.3662609939745224,
      "lm_loss": 0.5225,
      "step": 4992,
      "vm_loss": 0.1843
    },
    {
      "epoch": 0.3662609939745224,
      "lm_loss": 1.2751,
      "step": 4992,
      "vm_loss": 0.0861
    },
    {
      "epoch": 0.36633436356466154,
      "grad_norm": 1.5006134435682437,
      "learning_rate": 7.315356917870929e-06,
      "loss": 0.6948,
      "step": 4993
    },
    {
      "epoch": 0.3664077331548007,
      "grad_norm": 0.5494571801762148,
      "learning_rate": 7.314303728636805e-06,
      "loss": 0.5073,
      "step": 4994
    },
    {
      "epoch": 0.3664811027449398,
      "grad_norm": 1.6098190803414776,
      "learning_rate": 7.313250408708218e-06,
      "loss": 0.8937,
      "step": 4995
    },
    {
      "epoch": 0.36655447233507893,
      "grad_norm": 1.5869419841642698,
      "learning_rate": 7.312196958144651e-06,
      "loss": 0.7727,
      "step": 4996
    },
    {
      "epoch": 0.366627841925218,
      "grad_norm": 0.5974474259850561,
      "learning_rate": 7.311143377005595e-06,
      "loss": 0.5077,
      "step": 4997
    },
    {
      "epoch": 0.3667012115153572,
      "grad_norm": 0.6081753415887535,
      "learning_rate": 7.31008966535055e-06,
      "loss": 0.5235,
      "step": 4998
    },
    {
      "epoch": 0.3667745811054963,
      "grad_norm": 0.5960546512901235,
      "learning_rate": 7.30903582323902e-06,
      "loss": 0.5197,
      "step": 4999
    },
    {
      "epoch": 0.3668479506956354,
      "grad_norm": 0.539433192457367,
      "learning_rate": 7.30798185073052e-06,
      "loss": 0.5268,
      "step": 5000
    },
    {
      "epoch": 0.3668479506956354,
      "lm_loss": 1.3069,
      "step": 5000,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.3668479506956354,
      "lm_loss": 0.0664,
      "step": 5000,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.3668479506956354,
      "lm_loss": 0.9442,
      "step": 5000,
      "vm_loss": 0.1772
    },
    {
      "epoch": 0.3668479506956354,
      "lm_loss": 0.7347,
      "step": 5000,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.3668479506956354,
      "lm_loss": 0.6526,
      "step": 5000,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.3668479506956354,
      "lm_loss": 0.2605,
      "step": 5000,
      "vm_loss": 0.09
    },
    {
      "epoch": 0.3668479506956354,
      "lm_loss": 0.6984,
      "step": 5000,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.3668479506956354,
      "lm_loss": 0.7751,
      "step": 5000,
      "vm_loss": 0.1563
    },
    {
      "epoch": 0.36692132028577457,
      "grad_norm": 2.001091168836865,
      "learning_rate": 7.306927747884568e-06,
      "loss": 0.8161,
      "step": 5001
    },
    {
      "epoch": 0.36699468987591366,
      "grad_norm": 0.5912764977813599,
      "learning_rate": 7.305873514760694e-06,
      "loss": 0.5126,
      "step": 5002
    },
    {
      "epoch": 0.3670680594660528,
      "grad_norm": 0.5897891213986179,
      "learning_rate": 7.304819151418432e-06,
      "loss": 0.5135,
      "step": 5003
    },
    {
      "epoch": 0.36714142905619196,
      "grad_norm": 1.7181729678863167,
      "learning_rate": 7.303764657917323e-06,
      "loss": 0.7965,
      "step": 5004
    },
    {
      "epoch": 0.36721479864633105,
      "grad_norm": 1.4362182550067006,
      "learning_rate": 7.302710034316922e-06,
      "loss": 0.8109,
      "step": 5005
    },
    {
      "epoch": 0.3672881682364702,
      "grad_norm": 1.516653674335136,
      "learning_rate": 7.3016552806767805e-06,
      "loss": 0.8957,
      "step": 5006
    },
    {
      "epoch": 0.3673615378266093,
      "grad_norm": 1.4796705763482858,
      "learning_rate": 7.3006003970564654e-06,
      "loss": 0.7796,
      "step": 5007
    },
    {
      "epoch": 0.36743490741674845,
      "grad_norm": 1.456475020920409,
      "learning_rate": 7.299545383515549e-06,
      "loss": 0.8546,
      "step": 5008
    },
    {
      "epoch": 0.36743490741674845,
      "lm_loss": 1.9302,
      "step": 5008,
      "vm_loss": 0.0582
    },
    {
      "epoch": 0.36743490741674845,
      "lm_loss": 1.7612,
      "step": 5008,
      "vm_loss": 0.0964
    },
    {
      "epoch": 0.36743490741674845,
      "lm_loss": 0.921,
      "step": 5008,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.36743490741674845,
      "lm_loss": 0.4496,
      "step": 5008,
      "vm_loss": 0.2181
    },
    {
      "epoch": 0.36743490741674845,
      "lm_loss": 1.3679,
      "step": 5008,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.36743490741674845,
      "lm_loss": 0.8391,
      "step": 5008,
      "vm_loss": 0.078
    },
    {
      "epoch": 0.36743490741674845,
      "lm_loss": 0.02,
      "step": 5008,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.36743490741674845,
      "lm_loss": 0.3718,
      "step": 5008,
      "vm_loss": 0.114
    },
    {
      "epoch": 0.3675082770068876,
      "grad_norm": 1.737941439761087,
      "learning_rate": 7.2984902401136115e-06,
      "loss": 0.8557,
      "step": 5009
    },
    {
      "epoch": 0.3675816465970267,
      "grad_norm": 1.44399843515672,
      "learning_rate": 7.297434966910235e-06,
      "loss": 0.7898,
      "step": 5010
    },
    {
      "epoch": 0.36765501618716584,
      "grad_norm": 1.4527776461858393,
      "learning_rate": 7.296379563965019e-06,
      "loss": 0.8123,
      "step": 5011
    },
    {
      "epoch": 0.36772838577730493,
      "grad_norm": 1.4305812471173895,
      "learning_rate": 7.29532403133756e-06,
      "loss": 0.7627,
      "step": 5012
    },
    {
      "epoch": 0.3678017553674441,
      "grad_norm": 1.5380685340387834,
      "learning_rate": 7.29426836908747e-06,
      "loss": 0.8273,
      "step": 5013
    },
    {
      "epoch": 0.36787512495758323,
      "grad_norm": 1.447677486110089,
      "learning_rate": 7.293212577274364e-06,
      "loss": 0.7329,
      "step": 5014
    },
    {
      "epoch": 0.3679484945477223,
      "grad_norm": 1.508583060765809,
      "learning_rate": 7.292156655957864e-06,
      "loss": 0.8149,
      "step": 5015
    },
    {
      "epoch": 0.3680218641378615,
      "grad_norm": 0.6174209803680857,
      "learning_rate": 7.291100605197601e-06,
      "loss": 0.5347,
      "step": 5016
    },
    {
      "epoch": 0.3680218641378615,
      "lm_loss": 0.9695,
      "step": 5016,
      "vm_loss": 0.0755
    },
    {
      "epoch": 0.3680218641378615,
      "lm_loss": 0.9066,
      "step": 5016,
      "vm_loss": 0.0796
    },
    {
      "epoch": 0.3680218641378615,
      "lm_loss": 0.8627,
      "step": 5016,
      "vm_loss": 0.1638
    },
    {
      "epoch": 0.3680218641378615,
      "lm_loss": 1.0426,
      "step": 5016,
      "vm_loss": 0.0815
    },
    {
      "epoch": 0.3680218641378615,
      "lm_loss": 0.9883,
      "step": 5016,
      "vm_loss": 0.1846
    },
    {
      "epoch": 0.3680218641378615,
      "lm_loss": 0.3124,
      "step": 5016,
      "vm_loss": 0.2374
    },
    {
      "epoch": 0.3680218641378615,
      "lm_loss": 0.0626,
      "step": 5016,
      "vm_loss": 0.1362
    },
    {
      "epoch": 0.3680218641378615,
      "lm_loss": 0.6348,
      "step": 5016,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.36809523372800057,
      "grad_norm": 1.6684527414119181,
      "learning_rate": 7.290044425053214e-06,
      "loss": 0.7426,
      "step": 5017
    },
    {
      "epoch": 0.3681686033181397,
      "grad_norm": 1.3513605998910907,
      "learning_rate": 7.288988115584345e-06,
      "loss": 0.768,
      "step": 5018
    },
    {
      "epoch": 0.3682419729082788,
      "grad_norm": 1.6543845491928733,
      "learning_rate": 7.287931676850652e-06,
      "loss": 0.872,
      "step": 5019
    },
    {
      "epoch": 0.36831534249841796,
      "grad_norm": 1.5550325091216817,
      "learning_rate": 7.286875108911787e-06,
      "loss": 0.8815,
      "step": 5020
    },
    {
      "epoch": 0.3683887120885571,
      "grad_norm": 0.6062896583674836,
      "learning_rate": 7.285818411827425e-06,
      "loss": 0.5216,
      "step": 5021
    },
    {
      "epoch": 0.3684620816786962,
      "grad_norm": 1.462787736831689,
      "learning_rate": 7.284761585657234e-06,
      "loss": 0.8923,
      "step": 5022
    },
    {
      "epoch": 0.36853545126883536,
      "grad_norm": 1.683222027280668,
      "learning_rate": 7.2837046304609e-06,
      "loss": 0.8713,
      "step": 5023
    },
    {
      "epoch": 0.36860882085897445,
      "grad_norm": 1.3704610601146623,
      "learning_rate": 7.282647546298108e-06,
      "loss": 0.8143,
      "step": 5024
    },
    {
      "epoch": 0.36860882085897445,
      "lm_loss": 0.1202,
      "step": 5024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36860882085897445,
      "lm_loss": 0.5309,
      "step": 5024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36860882085897445,
      "lm_loss": 0.4612,
      "step": 5024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36860882085897445,
      "lm_loss": 0.5402,
      "step": 5024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36860882085897445,
      "lm_loss": 0.2938,
      "step": 5024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36860882085897445,
      "lm_loss": 0.5875,
      "step": 5024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36860882085897445,
      "lm_loss": 0.5806,
      "step": 5024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.36860882085897445,
      "lm_loss": 0.562,
      "step": 5024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3686821904491136,
      "grad_norm": 0.5790612508108167,
      "learning_rate": 7.2815903332285586e-06,
      "loss": 0.5087,
      "step": 5025
    },
    {
      "epoch": 0.36875556003925275,
      "grad_norm": 1.6156500951490564,
      "learning_rate": 7.28053299131195e-06,
      "loss": 0.8336,
      "step": 5026
    },
    {
      "epoch": 0.36882892962939184,
      "grad_norm": 1.5265161484100795,
      "learning_rate": 7.279475520607998e-06,
      "loss": 0.8747,
      "step": 5027
    },
    {
      "epoch": 0.368902299219531,
      "grad_norm": 0.5896325464104887,
      "learning_rate": 7.278417921176417e-06,
      "loss": 0.5113,
      "step": 5028
    },
    {
      "epoch": 0.3689756688096701,
      "grad_norm": 1.412158394309689,
      "learning_rate": 7.277360193076936e-06,
      "loss": 0.8925,
      "step": 5029
    },
    {
      "epoch": 0.36904903839980924,
      "grad_norm": 1.5126055081151413,
      "learning_rate": 7.276302336369283e-06,
      "loss": 0.7308,
      "step": 5030
    },
    {
      "epoch": 0.3691224079899484,
      "grad_norm": 1.6931901217924412,
      "learning_rate": 7.2752443511132e-06,
      "loss": 0.7712,
      "step": 5031
    },
    {
      "epoch": 0.3691957775800875,
      "grad_norm": 1.5943521191973864,
      "learning_rate": 7.2741862373684345e-06,
      "loss": 0.8204,
      "step": 5032
    },
    {
      "epoch": 0.3691957775800875,
      "lm_loss": 1.3463,
      "step": 5032,
      "vm_loss": 0.0375
    },
    {
      "epoch": 0.3691957775800875,
      "lm_loss": 0.2345,
      "step": 5032,
      "vm_loss": 0.1849
    },
    {
      "epoch": 0.3691957775800875,
      "lm_loss": 0.9468,
      "step": 5032,
      "vm_loss": 0.1715
    },
    {
      "epoch": 0.3691957775800875,
      "lm_loss": 0.8718,
      "step": 5032,
      "vm_loss": 0.1249
    },
    {
      "epoch": 0.3691957775800875,
      "lm_loss": 0.5607,
      "step": 5032,
      "vm_loss": 0.175
    },
    {
      "epoch": 0.3691957775800875,
      "lm_loss": 0.1398,
      "step": 5032,
      "vm_loss": 0.0858
    },
    {
      "epoch": 0.3691957775800875,
      "lm_loss": 0.0789,
      "step": 5032,
      "vm_loss": 0.0758
    },
    {
      "epoch": 0.3691957775800875,
      "lm_loss": 0.1302,
      "step": 5032,
      "vm_loss": 0.0632
    },
    {
      "epoch": 0.36926914717022663,
      "grad_norm": 1.4921787950652943,
      "learning_rate": 7.2731279951947395e-06,
      "loss": 0.799,
      "step": 5033
    },
    {
      "epoch": 0.3693425167603657,
      "grad_norm": 1.3636503275596032,
      "learning_rate": 7.2720696246518775e-06,
      "loss": 0.7937,
      "step": 5034
    },
    {
      "epoch": 0.3694158863505049,
      "grad_norm": 1.4058508535656953,
      "learning_rate": 7.2710111257996185e-06,
      "loss": 0.7931,
      "step": 5035
    },
    {
      "epoch": 0.369489255940644,
      "grad_norm": 0.6414203315847284,
      "learning_rate": 7.269952498697734e-06,
      "loss": 0.5136,
      "step": 5036
    },
    {
      "epoch": 0.3695626255307831,
      "grad_norm": 0.5944271516156622,
      "learning_rate": 7.268893743406013e-06,
      "loss": 0.526,
      "step": 5037
    },
    {
      "epoch": 0.36963599512092227,
      "grad_norm": 0.5696844895144354,
      "learning_rate": 7.267834859984244e-06,
      "loss": 0.5084,
      "step": 5038
    },
    {
      "epoch": 0.36970936471106136,
      "grad_norm": 1.4012002012871916,
      "learning_rate": 7.266775848492223e-06,
      "loss": 0.8114,
      "step": 5039
    },
    {
      "epoch": 0.3697827343012005,
      "grad_norm": 1.6293661756420392,
      "learning_rate": 7.265716708989756e-06,
      "loss": 0.9705,
      "step": 5040
    },
    {
      "epoch": 0.3697827343012005,
      "lm_loss": 2.8702,
      "step": 5040,
      "vm_loss": 0.0543
    },
    {
      "epoch": 0.3697827343012005,
      "lm_loss": 0.4221,
      "step": 5040,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.3697827343012005,
      "lm_loss": 0.8794,
      "step": 5040,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.3697827343012005,
      "lm_loss": 0.573,
      "step": 5040,
      "vm_loss": 0.1719
    },
    {
      "epoch": 0.3697827343012005,
      "lm_loss": 0.4352,
      "step": 5040,
      "vm_loss": 0.0886
    },
    {
      "epoch": 0.3697827343012005,
      "lm_loss": 0.2077,
      "step": 5040,
      "vm_loss": 0.1709
    },
    {
      "epoch": 0.3697827343012005,
      "lm_loss": 0.9025,
      "step": 5040,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.3697827343012005,
      "lm_loss": 1.0106,
      "step": 5040,
      "vm_loss": 0.1844
    },
    {
      "epoch": 0.36985610389133966,
      "grad_norm": 1.584968319547136,
      "learning_rate": 7.264657441536656e-06,
      "loss": 0.8552,
      "step": 5041
    },
    {
      "epoch": 0.36992947348147875,
      "grad_norm": 1.567693340417847,
      "learning_rate": 7.263598046192742e-06,
      "loss": 0.8047,
      "step": 5042
    },
    {
      "epoch": 0.3700028430716179,
      "grad_norm": 1.60823354256983,
      "learning_rate": 7.262538523017841e-06,
      "loss": 0.8173,
      "step": 5043
    },
    {
      "epoch": 0.370076212661757,
      "grad_norm": 0.7220440115641473,
      "learning_rate": 7.2614788720717855e-06,
      "loss": 0.5271,
      "step": 5044
    },
    {
      "epoch": 0.37014958225189615,
      "grad_norm": 1.3615162985577927,
      "learning_rate": 7.260419093414418e-06,
      "loss": 0.7865,
      "step": 5045
    },
    {
      "epoch": 0.3702229518420353,
      "grad_norm": 0.6351656047543732,
      "learning_rate": 7.259359187105587e-06,
      "loss": 0.5214,
      "step": 5046
    },
    {
      "epoch": 0.3702963214321744,
      "grad_norm": 1.7441314522979625,
      "learning_rate": 7.258299153205146e-06,
      "loss": 0.9424,
      "step": 5047
    },
    {
      "epoch": 0.37036969102231354,
      "grad_norm": 1.604245824878018,
      "learning_rate": 7.257238991772961e-06,
      "loss": 0.7102,
      "step": 5048
    },
    {
      "epoch": 0.37036969102231354,
      "lm_loss": 1.7658,
      "step": 5048,
      "vm_loss": 0.0681
    },
    {
      "epoch": 0.37036969102231354,
      "lm_loss": 0.0299,
      "step": 5048,
      "vm_loss": 0.1486
    },
    {
      "epoch": 0.37036969102231354,
      "lm_loss": 0.9609,
      "step": 5048,
      "vm_loss": 0.1821
    },
    {
      "epoch": 0.37036969102231354,
      "lm_loss": 1.2267,
      "step": 5048,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.37036969102231354,
      "lm_loss": 0.9186,
      "step": 5048,
      "vm_loss": 0.0938
    },
    {
      "epoch": 0.37036969102231354,
      "lm_loss": 0.2756,
      "step": 5048,
      "vm_loss": 0.2089
    },
    {
      "epoch": 0.37036969102231354,
      "lm_loss": 0.4467,
      "step": 5048,
      "vm_loss": 0.2392
    },
    {
      "epoch": 0.37036969102231354,
      "lm_loss": 1.012,
      "step": 5048,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.37044306061245263,
      "grad_norm": 1.5484166681876106,
      "learning_rate": 7.256178702868899e-06,
      "loss": 0.7773,
      "step": 5049
    },
    {
      "epoch": 0.3705164302025918,
      "grad_norm": 1.2387336999505278,
      "learning_rate": 7.255118286552839e-06,
      "loss": 0.7851,
      "step": 5050
    },
    {
      "epoch": 0.37058979979273093,
      "grad_norm": 1.4918807780426027,
      "learning_rate": 7.254057742884665e-06,
      "loss": 0.8003,
      "step": 5051
    },
    {
      "epoch": 0.37066316938287003,
      "grad_norm": 1.447349178242144,
      "learning_rate": 7.252997071924267e-06,
      "loss": 0.8776,
      "step": 5052
    },
    {
      "epoch": 0.3707365389730092,
      "grad_norm": 1.545608905027029,
      "learning_rate": 7.251936273731546e-06,
      "loss": 0.8173,
      "step": 5053
    },
    {
      "epoch": 0.37080990856314827,
      "grad_norm": 0.6631550455426177,
      "learning_rate": 7.250875348366406e-06,
      "loss": 0.5304,
      "step": 5054
    },
    {
      "epoch": 0.3708832781532874,
      "grad_norm": 0.6450714228129409,
      "learning_rate": 7.249814295888761e-06,
      "loss": 0.5193,
      "step": 5055
    },
    {
      "epoch": 0.37095664774342657,
      "grad_norm": 1.485054587202709,
      "learning_rate": 7.248753116358531e-06,
      "loss": 0.8759,
      "step": 5056
    },
    {
      "epoch": 0.37095664774342657,
      "lm_loss": 1.479,
      "step": 5056,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.37095664774342657,
      "lm_loss": 1.7276,
      "step": 5056,
      "vm_loss": 0.1549
    },
    {
      "epoch": 0.37095664774342657,
      "lm_loss": 0.4642,
      "step": 5056,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.37095664774342657,
      "lm_loss": 0.5423,
      "step": 5056,
      "vm_loss": 0.1323
    },
    {
      "epoch": 0.37095664774342657,
      "lm_loss": 1.027,
      "step": 5056,
      "vm_loss": 0.1802
    },
    {
      "epoch": 0.37095664774342657,
      "lm_loss": 0.7179,
      "step": 5056,
      "vm_loss": 0.2289
    },
    {
      "epoch": 0.37095664774342657,
      "lm_loss": 0.8949,
      "step": 5056,
      "vm_loss": 0.1957
    },
    {
      "epoch": 0.37095664774342657,
      "lm_loss": 0.6638,
      "step": 5056,
      "vm_loss": 0.1924
    },
    {
      "epoch": 0.37103001733356566,
      "grad_norm": 1.598956625450301,
      "learning_rate": 7.247691809835645e-06,
      "loss": 0.8652,
      "step": 5057
    },
    {
      "epoch": 0.3711033869237048,
      "grad_norm": 1.5425440043198302,
      "learning_rate": 7.246630376380035e-06,
      "loss": 0.7588,
      "step": 5058
    },
    {
      "epoch": 0.3711767565138439,
      "grad_norm": 1.3836736508829945,
      "learning_rate": 7.2455688160516445e-06,
      "loss": 0.7452,
      "step": 5059
    },
    {
      "epoch": 0.37125012610398306,
      "grad_norm": 1.4109412386924995,
      "learning_rate": 7.244507128910421e-06,
      "loss": 0.8216,
      "step": 5060
    },
    {
      "epoch": 0.37132349569412215,
      "grad_norm": 0.7034403500043422,
      "learning_rate": 7.243445315016322e-06,
      "loss": 0.5133,
      "step": 5061
    },
    {
      "epoch": 0.3713968652842613,
      "grad_norm": 0.687888691884651,
      "learning_rate": 7.2423833744293105e-06,
      "loss": 0.5164,
      "step": 5062
    },
    {
      "epoch": 0.37147023487440045,
      "grad_norm": 1.491671607230002,
      "learning_rate": 7.241321307209356e-06,
      "loss": 0.7714,
      "step": 5063
    },
    {
      "epoch": 0.37154360446453955,
      "grad_norm": 1.4019191198874177,
      "learning_rate": 7.240259113416437e-06,
      "loss": 0.7228,
      "step": 5064
    },
    {
      "epoch": 0.37154360446453955,
      "lm_loss": 1.0676,
      "step": 5064,
      "vm_loss": 0.0607
    },
    {
      "epoch": 0.37154360446453955,
      "lm_loss": 0.7659,
      "step": 5064,
      "vm_loss": 0.0816
    },
    {
      "epoch": 0.37154360446453955,
      "lm_loss": 0.5731,
      "step": 5064,
      "vm_loss": 0.0809
    },
    {
      "epoch": 0.37154360446453955,
      "lm_loss": 0.6737,
      "step": 5064,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.37154360446453955,
      "lm_loss": 0.4004,
      "step": 5064,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.37154360446453955,
      "lm_loss": 0.3365,
      "step": 5064,
      "vm_loss": 0.2181
    },
    {
      "epoch": 0.37154360446453955,
      "lm_loss": 0.1507,
      "step": 5064,
      "vm_loss": 0.1551
    },
    {
      "epoch": 0.37154360446453955,
      "lm_loss": 0.1838,
      "step": 5064,
      "vm_loss": 0.0597
    },
    {
      "epoch": 0.3716169740546787,
      "grad_norm": 1.6471642712843104,
      "learning_rate": 7.239196793110539e-06,
      "loss": 0.7473,
      "step": 5065
    },
    {
      "epoch": 0.3716903436448178,
      "grad_norm": 1.3830104808326062,
      "learning_rate": 7.238134346351651e-06,
      "loss": 0.7995,
      "step": 5066
    },
    {
      "epoch": 0.37176371323495694,
      "grad_norm": 1.466774442025071,
      "learning_rate": 7.237071773199775e-06,
      "loss": 0.7957,
      "step": 5067
    },
    {
      "epoch": 0.3718370828250961,
      "grad_norm": 1.5809332301770862,
      "learning_rate": 7.236009073714916e-06,
      "loss": 0.7932,
      "step": 5068
    },
    {
      "epoch": 0.3719104524152352,
      "grad_norm": 1.591341555763101,
      "learning_rate": 7.234946247957087e-06,
      "loss": 0.8394,
      "step": 5069
    },
    {
      "epoch": 0.37198382200537433,
      "grad_norm": 1.5793094383542694,
      "learning_rate": 7.233883295986308e-06,
      "loss": 0.7175,
      "step": 5070
    },
    {
      "epoch": 0.3720571915955134,
      "grad_norm": 1.5664663271053094,
      "learning_rate": 7.232820217862607e-06,
      "loss": 0.8082,
      "step": 5071
    },
    {
      "epoch": 0.3721305611856526,
      "grad_norm": 1.631650949730846,
      "learning_rate": 7.23175701364602e-06,
      "loss": 0.7862,
      "step": 5072
    },
    {
      "epoch": 0.3721305611856526,
      "lm_loss": 0.3423,
      "step": 5072,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.3721305611856526,
      "lm_loss": 0.0443,
      "step": 5072,
      "vm_loss": 0.114
    },
    {
      "epoch": 0.3721305611856526,
      "lm_loss": 0.4025,
      "step": 5072,
      "vm_loss": 0.0994
    },
    {
      "epoch": 0.3721305611856526,
      "lm_loss": 0.1289,
      "step": 5072,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.3721305611856526,
      "lm_loss": 0.6748,
      "step": 5072,
      "vm_loss": 0.2209
    },
    {
      "epoch": 0.3721305611856526,
      "lm_loss": 0.2185,
      "step": 5072,
      "vm_loss": 0.1274
    },
    {
      "epoch": 0.3721305611856526,
      "lm_loss": 0.1057,
      "step": 5072,
      "vm_loss": 0.0966
    },
    {
      "epoch": 0.3721305611856526,
      "lm_loss": 1.4993,
      "step": 5072,
      "vm_loss": 0.0641
    },
    {
      "epoch": 0.3722039307757917,
      "grad_norm": 1.6499837550132008,
      "learning_rate": 7.230693683396588e-06,
      "loss": 0.82,
      "step": 5073
    },
    {
      "epoch": 0.3722773003659308,
      "grad_norm": 1.5496063899088623,
      "learning_rate": 7.229630227174358e-06,
      "loss": 0.8199,
      "step": 5074
    },
    {
      "epoch": 0.37235066995606997,
      "grad_norm": 1.5058435481961032,
      "learning_rate": 7.228566645039388e-06,
      "loss": 0.8041,
      "step": 5075
    },
    {
      "epoch": 0.37242403954620906,
      "grad_norm": 1.48068893468518,
      "learning_rate": 7.22750293705174e-06,
      "loss": 0.8184,
      "step": 5076
    },
    {
      "epoch": 0.3724974091363482,
      "grad_norm": 1.519930496113516,
      "learning_rate": 7.226439103271487e-06,
      "loss": 0.782,
      "step": 5077
    },
    {
      "epoch": 0.37257077872648736,
      "grad_norm": 0.9834048556079575,
      "learning_rate": 7.225375143758701e-06,
      "loss": 0.5108,
      "step": 5078
    },
    {
      "epoch": 0.37264414831662646,
      "grad_norm": 1.5576245967787503,
      "learning_rate": 7.224311058573472e-06,
      "loss": 0.7721,
      "step": 5079
    },
    {
      "epoch": 0.3727175179067656,
      "grad_norm": 1.700405729705537,
      "learning_rate": 7.223246847775889e-06,
      "loss": 0.7799,
      "step": 5080
    },
    {
      "epoch": 0.3727175179067656,
      "lm_loss": 1.2502,
      "step": 5080,
      "vm_loss": 0.0476
    },
    {
      "epoch": 0.3727175179067656,
      "lm_loss": 0.2928,
      "step": 5080,
      "vm_loss": 0.0508
    },
    {
      "epoch": 0.3727175179067656,
      "lm_loss": 0.3576,
      "step": 5080,
      "vm_loss": 0.1141
    },
    {
      "epoch": 0.3727175179067656,
      "lm_loss": 0.7053,
      "step": 5080,
      "vm_loss": 0.1599
    },
    {
      "epoch": 0.3727175179067656,
      "lm_loss": 0.7559,
      "step": 5080,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.3727175179067656,
      "lm_loss": 0.1294,
      "step": 5080,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.3727175179067656,
      "lm_loss": 0.3407,
      "step": 5080,
      "vm_loss": 0.1814
    },
    {
      "epoch": 0.3727175179067656,
      "lm_loss": 1.331,
      "step": 5080,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.3727908874969047,
      "grad_norm": 1.6314036511192282,
      "learning_rate": 7.22218251142605e-06,
      "loss": 0.8257,
      "step": 5081
    },
    {
      "epoch": 0.37286425708704385,
      "grad_norm": 1.4186138301896056,
      "learning_rate": 7.221118049584061e-06,
      "loss": 0.8868,
      "step": 5082
    },
    {
      "epoch": 0.372937626677183,
      "grad_norm": 1.7399305273273942,
      "learning_rate": 7.220053462310036e-06,
      "loss": 0.7082,
      "step": 5083
    },
    {
      "epoch": 0.3730109962673221,
      "grad_norm": 1.3825896339043986,
      "learning_rate": 7.2189887496640946e-06,
      "loss": 0.7871,
      "step": 5084
    },
    {
      "epoch": 0.37308436585746124,
      "grad_norm": 1.5751996134058124,
      "learning_rate": 7.217923911706362e-06,
      "loss": 0.8128,
      "step": 5085
    },
    {
      "epoch": 0.37315773544760034,
      "grad_norm": 0.6848194906084986,
      "learning_rate": 7.216858948496975e-06,
      "loss": 0.5164,
      "step": 5086
    },
    {
      "epoch": 0.3732311050377395,
      "grad_norm": 1.685740060782227,
      "learning_rate": 7.215793860096071e-06,
      "loss": 0.8226,
      "step": 5087
    },
    {
      "epoch": 0.37330447462787864,
      "grad_norm": 1.5562113971532214,
      "learning_rate": 7.214728646563803e-06,
      "loss": 0.8299,
      "step": 5088
    },
    {
      "epoch": 0.37330447462787864,
      "lm_loss": 1.1873,
      "step": 5088,
      "vm_loss": 0.049
    },
    {
      "epoch": 0.37330447462787864,
      "lm_loss": 1.4169,
      "step": 5088,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.37330447462787864,
      "lm_loss": 0.4291,
      "step": 5088,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.37330447462787864,
      "lm_loss": 0.0993,
      "step": 5088,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.37330447462787864,
      "lm_loss": 0.5496,
      "step": 5088,
      "vm_loss": 0.1243
    },
    {
      "epoch": 0.37330447462787864,
      "lm_loss": 0.8608,
      "step": 5088,
      "vm_loss": 0.1844
    },
    {
      "epoch": 0.37330447462787864,
      "lm_loss": 0.5333,
      "step": 5088,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.37330447462787864,
      "lm_loss": 0.1042,
      "step": 5088,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.37337784421801773,
      "grad_norm": 1.5713103564526993,
      "learning_rate": 7.213663307960321e-06,
      "loss": 0.7844,
      "step": 5089
    },
    {
      "epoch": 0.3734512138081569,
      "grad_norm": 1.5728936643692337,
      "learning_rate": 7.2125978443457915e-06,
      "loss": 0.8055,
      "step": 5090
    },
    {
      "epoch": 0.373524583398296,
      "grad_norm": 0.599266880828232,
      "learning_rate": 7.211532255780383e-06,
      "loss": 0.4991,
      "step": 5091
    },
    {
      "epoch": 0.3735979529884351,
      "grad_norm": 1.5485561192167017,
      "learning_rate": 7.21046654232427e-06,
      "loss": 0.8643,
      "step": 5092
    },
    {
      "epoch": 0.37367132257857427,
      "grad_norm": 1.541482792760771,
      "learning_rate": 7.209400704037637e-06,
      "loss": 0.7029,
      "step": 5093
    },
    {
      "epoch": 0.37374469216871337,
      "grad_norm": 0.5754784058225269,
      "learning_rate": 7.208334740980676e-06,
      "loss": 0.5177,
      "step": 5094
    },
    {
      "epoch": 0.3738180617588525,
      "grad_norm": 0.5842928823439225,
      "learning_rate": 7.207268653213582e-06,
      "loss": 0.5278,
      "step": 5095
    },
    {
      "epoch": 0.3738914313489916,
      "grad_norm": 1.4142181005005703,
      "learning_rate": 7.206202440796562e-06,
      "loss": 0.7732,
      "step": 5096
    },
    {
      "epoch": 0.3738914313489916,
      "lm_loss": 0.4845,
      "step": 5096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3738914313489916,
      "lm_loss": 0.5584,
      "step": 5096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3738914313489916,
      "lm_loss": 0.7178,
      "step": 5096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3738914313489916,
      "lm_loss": 0.4181,
      "step": 5096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3738914313489916,
      "lm_loss": 0.4632,
      "step": 5096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3738914313489916,
      "lm_loss": 0.5063,
      "step": 5096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3738914313489916,
      "lm_loss": 0.502,
      "step": 5096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3738914313489916,
      "lm_loss": 0.5069,
      "step": 5096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.37396480093913076,
      "grad_norm": 0.5497112422044276,
      "learning_rate": 7.205136103789826e-06,
      "loss": 0.5108,
      "step": 5097
    },
    {
      "epoch": 0.37403817052926985,
      "grad_norm": 1.5140946632739631,
      "learning_rate": 7.204069642253592e-06,
      "loss": 0.8356,
      "step": 5098
    },
    {
      "epoch": 0.374111540119409,
      "grad_norm": 1.8365941266098365,
      "learning_rate": 7.20300305624809e-06,
      "loss": 0.7767,
      "step": 5099
    },
    {
      "epoch": 0.37418490970954815,
      "grad_norm": 1.6074316690526527,
      "learning_rate": 7.201936345833549e-06,
      "loss": 0.828,
      "step": 5100
    },
    {
      "epoch": 0.37425827929968725,
      "grad_norm": 0.5667979817477757,
      "learning_rate": 7.200869511070208e-06,
      "loss": 0.5137,
      "step": 5101
    },
    {
      "epoch": 0.3743316488898264,
      "grad_norm": 1.646682633658499,
      "learning_rate": 7.199802552018315e-06,
      "loss": 0.8227,
      "step": 5102
    },
    {
      "epoch": 0.3744050184799655,
      "grad_norm": 1.5710613623024319,
      "learning_rate": 7.198735468738126e-06,
      "loss": 0.8707,
      "step": 5103
    },
    {
      "epoch": 0.37447838807010464,
      "grad_norm": 1.4794586398215466,
      "learning_rate": 7.197668261289901e-06,
      "loss": 0.7707,
      "step": 5104
    },
    {
      "epoch": 0.37447838807010464,
      "lm_loss": 1.4483,
      "step": 5104,
      "vm_loss": 0.0724
    },
    {
      "epoch": 0.37447838807010464,
      "lm_loss": 1.3075,
      "step": 5104,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.37447838807010464,
      "lm_loss": 0.408,
      "step": 5104,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.37447838807010464,
      "lm_loss": 0.3226,
      "step": 5104,
      "vm_loss": 0.1471
    },
    {
      "epoch": 0.37447838807010464,
      "lm_loss": 0.3596,
      "step": 5104,
      "vm_loss": 0.1117
    },
    {
      "epoch": 0.37447838807010464,
      "lm_loss": 1.063,
      "step": 5104,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.37447838807010464,
      "lm_loss": 0.6257,
      "step": 5104,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.37447838807010464,
      "lm_loss": 0.6398,
      "step": 5104,
      "vm_loss": 0.149
    },
    {
      "epoch": 0.3745517576602438,
      "grad_norm": 1.370311007179936,
      "learning_rate": 7.196600929733906e-06,
      "loss": 0.8374,
      "step": 5105
    },
    {
      "epoch": 0.3746251272503829,
      "grad_norm": 1.6205566960317392,
      "learning_rate": 7.195533474130414e-06,
      "loss": 0.7403,
      "step": 5106
    },
    {
      "epoch": 0.37469849684052203,
      "grad_norm": 0.607455007514133,
      "learning_rate": 7.194465894539713e-06,
      "loss": 0.5261,
      "step": 5107
    },
    {
      "epoch": 0.3747718664306611,
      "grad_norm": 1.6727228393522553,
      "learning_rate": 7.193398191022085e-06,
      "loss": 0.7644,
      "step": 5108
    },
    {
      "epoch": 0.3748452360208003,
      "grad_norm": 1.529017613989226,
      "learning_rate": 7.192330363637832e-06,
      "loss": 0.7567,
      "step": 5109
    },
    {
      "epoch": 0.3749186056109394,
      "grad_norm": 1.543101729394607,
      "learning_rate": 7.191262412447254e-06,
      "loss": 0.8981,
      "step": 5110
    },
    {
      "epoch": 0.3749919752010785,
      "grad_norm": 0.5826880859353307,
      "learning_rate": 7.1901943375106595e-06,
      "loss": 0.5285,
      "step": 5111
    },
    {
      "epoch": 0.37506534479121767,
      "grad_norm": 1.5647691096871892,
      "learning_rate": 7.189126138888368e-06,
      "loss": 0.7274,
      "step": 5112
    },
    {
      "epoch": 0.37506534479121767,
      "lm_loss": 1.1633,
      "step": 5112,
      "vm_loss": 0.0568
    },
    {
      "epoch": 0.37506534479121767,
      "lm_loss": 1.4482,
      "step": 5112,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.37506534479121767,
      "lm_loss": 0.2935,
      "step": 5112,
      "vm_loss": 0.0427
    },
    {
      "epoch": 0.37506534479121767,
      "lm_loss": 0.8622,
      "step": 5112,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.37506534479121767,
      "lm_loss": 0.1801,
      "step": 5112,
      "vm_loss": 0.1601
    },
    {
      "epoch": 0.37506534479121767,
      "lm_loss": 0.9647,
      "step": 5112,
      "vm_loss": 0.1591
    },
    {
      "epoch": 0.37506534479121767,
      "lm_loss": 0.1361,
      "step": 5112,
      "vm_loss": 0.0925
    },
    {
      "epoch": 0.37506534479121767,
      "lm_loss": 1.1679,
      "step": 5112,
      "vm_loss": 0.2226
    },
    {
      "epoch": 0.37513871438135676,
      "grad_norm": 1.4006484963988637,
      "learning_rate": 7.188057816640703e-06,
      "loss": 0.8275,
      "step": 5113
    },
    {
      "epoch": 0.3752120839714959,
      "grad_norm": 1.521139575006532,
      "learning_rate": 7.186989370827993e-06,
      "loss": 0.9526,
      "step": 5114
    },
    {
      "epoch": 0.37528545356163506,
      "grad_norm": 1.5049091259219296,
      "learning_rate": 7.185920801510577e-06,
      "loss": 0.7386,
      "step": 5115
    },
    {
      "epoch": 0.37535882315177416,
      "grad_norm": 0.5560766089134193,
      "learning_rate": 7.1848521087488e-06,
      "loss": 0.5164,
      "step": 5116
    },
    {
      "epoch": 0.3754321927419133,
      "grad_norm": 0.5671572366703526,
      "learning_rate": 7.183783292603015e-06,
      "loss": 0.4909,
      "step": 5117
    },
    {
      "epoch": 0.3755055623320524,
      "grad_norm": 1.738140918305986,
      "learning_rate": 7.1827143531335785e-06,
      "loss": 0.8328,
      "step": 5118
    },
    {
      "epoch": 0.37557893192219155,
      "grad_norm": 1.781337734212752,
      "learning_rate": 7.181645290400856e-06,
      "loss": 0.8922,
      "step": 5119
    },
    {
      "epoch": 0.3756523015123307,
      "grad_norm": 0.5611770330112303,
      "learning_rate": 7.180576104465224e-06,
      "loss": 0.5184,
      "step": 5120
    },
    {
      "epoch": 0.3756523015123307,
      "lm_loss": 1.3425,
      "step": 5120,
      "vm_loss": 0.0831
    },
    {
      "epoch": 0.3756523015123307,
      "lm_loss": 1.0386,
      "step": 5120,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.3756523015123307,
      "lm_loss": 0.9393,
      "step": 5120,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.3756523015123307,
      "lm_loss": 0.4835,
      "step": 5120,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.3756523015123307,
      "lm_loss": 1.0983,
      "step": 5120,
      "vm_loss": 0.1955
    },
    {
      "epoch": 0.3756523015123307,
      "lm_loss": 0.7421,
      "step": 5120,
      "vm_loss": 0.1838
    },
    {
      "epoch": 0.3756523015123307,
      "lm_loss": 0.5211,
      "step": 5120,
      "vm_loss": 0.2236
    },
    {
      "epoch": 0.3756523015123307,
      "lm_loss": 0.8522,
      "step": 5120,
      "vm_loss": 0.1425
    },
    {
      "epoch": 0.3757256711024698,
      "grad_norm": 1.4991170954426316,
      "learning_rate": 7.179506795387057e-06,
      "loss": 0.8967,
      "step": 5121
    },
    {
      "epoch": 0.37579904069260894,
      "grad_norm": 0.5475551465736682,
      "learning_rate": 7.178437363226745e-06,
      "loss": 0.4997,
      "step": 5122
    },
    {
      "epoch": 0.37587241028274804,
      "grad_norm": 1.4507970107603705,
      "learning_rate": 7.1773678080446795e-06,
      "loss": 0.824,
      "step": 5123
    },
    {
      "epoch": 0.3759457798728872,
      "grad_norm": 1.5747811086174055,
      "learning_rate": 7.176298129901261e-06,
      "loss": 0.8107,
      "step": 5124
    },
    {
      "epoch": 0.37601914946302634,
      "grad_norm": 0.5426549360217681,
      "learning_rate": 7.175228328856899e-06,
      "loss": 0.5102,
      "step": 5125
    },
    {
      "epoch": 0.37609251905316543,
      "grad_norm": 1.743275500676904,
      "learning_rate": 7.174158404972007e-06,
      "loss": 0.8452,
      "step": 5126
    },
    {
      "epoch": 0.3761658886433046,
      "grad_norm": 0.5562298659856304,
      "learning_rate": 7.173088358307003e-06,
      "loss": 0.4915,
      "step": 5127
    },
    {
      "epoch": 0.3762392582334437,
      "grad_norm": 1.541301390443972,
      "learning_rate": 7.17201818892232e-06,
      "loss": 0.7838,
      "step": 5128
    },
    {
      "epoch": 0.3762392582334437,
      "lm_loss": 1.4935,
      "step": 5128,
      "vm_loss": 0.0181
    },
    {
      "epoch": 0.3762392582334437,
      "lm_loss": 0.0999,
      "step": 5128,
      "vm_loss": 0.1301
    },
    {
      "epoch": 0.3762392582334437,
      "lm_loss": 0.8282,
      "step": 5128,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.3762392582334437,
      "lm_loss": 0.7005,
      "step": 5128,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.3762392582334437,
      "lm_loss": 1.0377,
      "step": 5128,
      "vm_loss": 0.2103
    },
    {
      "epoch": 0.3762392582334437,
      "lm_loss": 0.7317,
      "step": 5128,
      "vm_loss": 0.1461
    },
    {
      "epoch": 0.3762392582334437,
      "lm_loss": 0.1875,
      "step": 5128,
      "vm_loss": 0.1614
    },
    {
      "epoch": 0.3762392582334437,
      "lm_loss": 0.3864,
      "step": 5128,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.3763126278235828,
      "grad_norm": 1.4496442971888517,
      "learning_rate": 7.170947896878392e-06,
      "loss": 0.7936,
      "step": 5129
    },
    {
      "epoch": 0.376385997413722,
      "grad_norm": 1.5345507048453606,
      "learning_rate": 7.169877482235657e-06,
      "loss": 0.8096,
      "step": 5130
    },
    {
      "epoch": 0.37645936700386107,
      "grad_norm": 0.5891215761580145,
      "learning_rate": 7.168806945054569e-06,
      "loss": 0.5009,
      "step": 5131
    },
    {
      "epoch": 0.3765327365940002,
      "grad_norm": 1.5349671025859917,
      "learning_rate": 7.167736285395582e-06,
      "loss": 0.8873,
      "step": 5132
    },
    {
      "epoch": 0.3766061061841393,
      "grad_norm": 1.483943593210227,
      "learning_rate": 7.166665503319158e-06,
      "loss": 0.7672,
      "step": 5133
    },
    {
      "epoch": 0.37667947577427846,
      "grad_norm": 1.4269580344343127,
      "learning_rate": 7.1655945988857675e-06,
      "loss": 0.8393,
      "step": 5134
    },
    {
      "epoch": 0.3767528453644176,
      "grad_norm": 1.4738724774165672,
      "learning_rate": 7.1645235721558855e-06,
      "loss": 0.8748,
      "step": 5135
    },
    {
      "epoch": 0.3768262149545567,
      "grad_norm": 0.5495987928075441,
      "learning_rate": 7.16345242319e-06,
      "loss": 0.5168,
      "step": 5136
    },
    {
      "epoch": 0.3768262149545567,
      "lm_loss": 1.1452,
      "step": 5136,
      "vm_loss": 0.0737
    },
    {
      "epoch": 0.3768262149545567,
      "lm_loss": 1.4604,
      "step": 5136,
      "vm_loss": 0.134
    },
    {
      "epoch": 0.3768262149545567,
      "lm_loss": 0.0752,
      "step": 5136,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.3768262149545567,
      "lm_loss": 0.8081,
      "step": 5136,
      "vm_loss": 0.0556
    },
    {
      "epoch": 0.3768262149545567,
      "lm_loss": 0.4773,
      "step": 5136,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.3768262149545567,
      "lm_loss": 0.5004,
      "step": 5136,
      "vm_loss": 0.1182
    },
    {
      "epoch": 0.3768262149545567,
      "lm_loss": 0.6503,
      "step": 5136,
      "vm_loss": 0.1548
    },
    {
      "epoch": 0.3768262149545567,
      "lm_loss": 0.4768,
      "step": 5136,
      "vm_loss": 0.1017
    },
    {
      "epoch": 0.37689958454469585,
      "grad_norm": 1.4420169563864615,
      "learning_rate": 7.1623811520485965e-06,
      "loss": 0.7843,
      "step": 5137
    },
    {
      "epoch": 0.37697295413483495,
      "grad_norm": 1.4448481361641121,
      "learning_rate": 7.161309758792174e-06,
      "loss": 0.7246,
      "step": 5138
    },
    {
      "epoch": 0.3770463237249741,
      "grad_norm": 1.446874725662626,
      "learning_rate": 7.160238243481238e-06,
      "loss": 0.765,
      "step": 5139
    },
    {
      "epoch": 0.3771196933151132,
      "grad_norm": 1.4224636703401627,
      "learning_rate": 7.159166606176299e-06,
      "loss": 0.7283,
      "step": 5140
    },
    {
      "epoch": 0.37719306290525234,
      "grad_norm": 1.8059017585079304,
      "learning_rate": 7.158094846937873e-06,
      "loss": 0.8621,
      "step": 5141
    },
    {
      "epoch": 0.3772664324953915,
      "grad_norm": 1.380400141693886,
      "learning_rate": 7.157022965826488e-06,
      "loss": 0.6604,
      "step": 5142
    },
    {
      "epoch": 0.3773398020855306,
      "grad_norm": 0.5782481873967587,
      "learning_rate": 7.155950962902673e-06,
      "loss": 0.5209,
      "step": 5143
    },
    {
      "epoch": 0.37741317167566973,
      "grad_norm": 1.6841363871759556,
      "learning_rate": 7.15487883822697e-06,
      "loss": 0.7858,
      "step": 5144
    },
    {
      "epoch": 0.37741317167566973,
      "lm_loss": 1.1906,
      "step": 5144,
      "vm_loss": 0.0529
    },
    {
      "epoch": 0.37741317167566973,
      "lm_loss": 0.3524,
      "step": 5144,
      "vm_loss": 0.1857
    },
    {
      "epoch": 0.37741317167566973,
      "lm_loss": 0.288,
      "step": 5144,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.37741317167566973,
      "lm_loss": 1.0335,
      "step": 5144,
      "vm_loss": 0.2334
    },
    {
      "epoch": 0.37741317167566973,
      "lm_loss": 0.8408,
      "step": 5144,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.37741317167566973,
      "lm_loss": 0.165,
      "step": 5144,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.37741317167566973,
      "lm_loss": 0.6315,
      "step": 5144,
      "vm_loss": 0.1634
    },
    {
      "epoch": 0.37741317167566973,
      "lm_loss": 0.6325,
      "step": 5144,
      "vm_loss": 0.1463
    },
    {
      "epoch": 0.37748654126580883,
      "grad_norm": 1.512317109086862,
      "learning_rate": 7.153806591859921e-06,
      "loss": 0.8302,
      "step": 5145
    },
    {
      "epoch": 0.377559910855948,
      "grad_norm": 0.5635450139819186,
      "learning_rate": 7.15273422386208e-06,
      "loss": 0.5124,
      "step": 5146
    },
    {
      "epoch": 0.3776332804460871,
      "grad_norm": 1.4841635304141312,
      "learning_rate": 7.151661734294007e-06,
      "loss": 0.8567,
      "step": 5147
    },
    {
      "epoch": 0.3777066500362262,
      "grad_norm": 0.5724764006701568,
      "learning_rate": 7.150589123216267e-06,
      "loss": 0.4972,
      "step": 5148
    },
    {
      "epoch": 0.37778001962636537,
      "grad_norm": 1.4216773011972785,
      "learning_rate": 7.149516390689433e-06,
      "loss": 0.842,
      "step": 5149
    },
    {
      "epoch": 0.37785338921650446,
      "grad_norm": 1.4793998657603447,
      "learning_rate": 7.148443536774086e-06,
      "loss": 0.851,
      "step": 5150
    },
    {
      "epoch": 0.3779267588066436,
      "grad_norm": 1.5579295521288068,
      "learning_rate": 7.147370561530811e-06,
      "loss": 0.7984,
      "step": 5151
    },
    {
      "epoch": 0.37800012839678276,
      "grad_norm": 1.4138865905366351,
      "learning_rate": 7.1462974650202035e-06,
      "loss": 0.775,
      "step": 5152
    },
    {
      "epoch": 0.37800012839678276,
      "lm_loss": 1.5013,
      "step": 5152,
      "vm_loss": 0.048
    },
    {
      "epoch": 0.37800012839678276,
      "lm_loss": 0.1052,
      "step": 5152,
      "vm_loss": 0.1038
    },
    {
      "epoch": 0.37800012839678276,
      "lm_loss": 0.2696,
      "step": 5152,
      "vm_loss": 0.1332
    },
    {
      "epoch": 0.37800012839678276,
      "lm_loss": 0.889,
      "step": 5152,
      "vm_loss": 0.1865
    },
    {
      "epoch": 0.37800012839678276,
      "lm_loss": 0.2573,
      "step": 5152,
      "vm_loss": 0.0631
    },
    {
      "epoch": 0.37800012839678276,
      "lm_loss": 0.5034,
      "step": 5152,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.37800012839678276,
      "lm_loss": 0.6299,
      "step": 5152,
      "vm_loss": 0.1489
    },
    {
      "epoch": 0.37800012839678276,
      "lm_loss": 0.7272,
      "step": 5152,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.37807349798692186,
      "grad_norm": 1.5201510299703263,
      "learning_rate": 7.145224247302862e-06,
      "loss": 0.8583,
      "step": 5153
    },
    {
      "epoch": 0.378146867577061,
      "grad_norm": 1.5160241426837664,
      "learning_rate": 7.144150908439395e-06,
      "loss": 0.8996,
      "step": 5154
    },
    {
      "epoch": 0.3782202371672001,
      "grad_norm": 1.4804782358027597,
      "learning_rate": 7.143077448490417e-06,
      "loss": 0.9264,
      "step": 5155
    },
    {
      "epoch": 0.37829360675733925,
      "grad_norm": 1.5732257166540478,
      "learning_rate": 7.142003867516548e-06,
      "loss": 0.8192,
      "step": 5156
    },
    {
      "epoch": 0.3783669763474784,
      "grad_norm": 1.42857945036302,
      "learning_rate": 7.140930165578415e-06,
      "loss": 0.7341,
      "step": 5157
    },
    {
      "epoch": 0.3784403459376175,
      "grad_norm": 1.528804861239525,
      "learning_rate": 7.1398563427366555e-06,
      "loss": 0.823,
      "step": 5158
    },
    {
      "epoch": 0.37851371552775664,
      "grad_norm": 1.473896817576545,
      "learning_rate": 7.138782399051908e-06,
      "loss": 0.8253,
      "step": 5159
    },
    {
      "epoch": 0.37858708511789574,
      "grad_norm": 0.7160227043597205,
      "learning_rate": 7.137708334584821e-06,
      "loss": 0.5155,
      "step": 5160
    },
    {
      "epoch": 0.37858708511789574,
      "lm_loss": 0.4694,
      "step": 5160,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.37858708511789574,
      "lm_loss": 1.0006,
      "step": 5160,
      "vm_loss": 0.1419
    },
    {
      "epoch": 0.37858708511789574,
      "lm_loss": 0.3084,
      "step": 5160,
      "vm_loss": 0.0689
    },
    {
      "epoch": 0.37858708511789574,
      "lm_loss": 0.6671,
      "step": 5160,
      "vm_loss": 0.1009
    },
    {
      "epoch": 0.37858708511789574,
      "lm_loss": 0.1405,
      "step": 5160,
      "vm_loss": 0.1266
    },
    {
      "epoch": 0.37858708511789574,
      "lm_loss": 0.2404,
      "step": 5160,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.37858708511789574,
      "lm_loss": 0.1544,
      "step": 5160,
      "vm_loss": 0.2444
    },
    {
      "epoch": 0.37858708511789574,
      "lm_loss": 1.0343,
      "step": 5160,
      "vm_loss": 0.1756
    },
    {
      "epoch": 0.3786604547080349,
      "grad_norm": 1.478978764232544,
      "learning_rate": 7.136634149396051e-06,
      "loss": 0.8188,
      "step": 5161
    },
    {
      "epoch": 0.37873382429817404,
      "grad_norm": 1.399967491156691,
      "learning_rate": 7.135559843546259e-06,
      "loss": 0.7972,
      "step": 5162
    },
    {
      "epoch": 0.37880719388831313,
      "grad_norm": 0.58402435697062,
      "learning_rate": 7.134485417096115e-06,
      "loss": 0.5202,
      "step": 5163
    },
    {
      "epoch": 0.3788805634784523,
      "grad_norm": 1.557198055043154,
      "learning_rate": 7.133410870106293e-06,
      "loss": 0.8454,
      "step": 5164
    },
    {
      "epoch": 0.3789539330685914,
      "grad_norm": 1.5129501988432041,
      "learning_rate": 7.1323362026374754e-06,
      "loss": 0.7698,
      "step": 5165
    },
    {
      "epoch": 0.3790273026587305,
      "grad_norm": 1.4295053545335883,
      "learning_rate": 7.131261414750352e-06,
      "loss": 0.8705,
      "step": 5166
    },
    {
      "epoch": 0.3791006722488697,
      "grad_norm": 1.6032544298683533,
      "learning_rate": 7.13018650650562e-06,
      "loss": 0.8238,
      "step": 5167
    },
    {
      "epoch": 0.37917404183900877,
      "grad_norm": 1.4543220261695209,
      "learning_rate": 7.129111477963978e-06,
      "loss": 0.7529,
      "step": 5168
    },
    {
      "epoch": 0.37917404183900877,
      "lm_loss": 0.6053,
      "step": 5168,
      "vm_loss": 0.057
    },
    {
      "epoch": 0.37917404183900877,
      "lm_loss": 0.285,
      "step": 5168,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.37917404183900877,
      "lm_loss": 0.5726,
      "step": 5168,
      "vm_loss": 0.2264
    },
    {
      "epoch": 0.37917404183900877,
      "lm_loss": 0.6328,
      "step": 5168,
      "vm_loss": 0.1575
    },
    {
      "epoch": 0.37917404183900877,
      "lm_loss": 0.8097,
      "step": 5168,
      "vm_loss": 0.1988
    },
    {
      "epoch": 0.37917404183900877,
      "lm_loss": 0.0146,
      "step": 5168,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.37917404183900877,
      "lm_loss": 0.7488,
      "step": 5168,
      "vm_loss": 0.0531
    },
    {
      "epoch": 0.37917404183900877,
      "lm_loss": 0.8518,
      "step": 5168,
      "vm_loss": 0.1571
    },
    {
      "epoch": 0.3792474114291479,
      "grad_norm": 1.5866274597864227,
      "learning_rate": 7.12803632918614e-06,
      "loss": 0.7613,
      "step": 5169
    },
    {
      "epoch": 0.379320781019287,
      "grad_norm": 1.5330916820981542,
      "learning_rate": 7.12696106023282e-06,
      "loss": 0.7605,
      "step": 5170
    },
    {
      "epoch": 0.37939415060942616,
      "grad_norm": 0.660544483815724,
      "learning_rate": 7.125885671164742e-06,
      "loss": 0.4975,
      "step": 5171
    },
    {
      "epoch": 0.3794675201995653,
      "grad_norm": 1.5873779914927502,
      "learning_rate": 7.124810162042635e-06,
      "loss": 0.7773,
      "step": 5172
    },
    {
      "epoch": 0.3795408897897044,
      "grad_norm": 1.4777488840187127,
      "learning_rate": 7.1237345329272355e-06,
      "loss": 0.7692,
      "step": 5173
    },
    {
      "epoch": 0.37961425937984355,
      "grad_norm": 1.4318922880647973,
      "learning_rate": 7.122658783879289e-06,
      "loss": 0.7839,
      "step": 5174
    },
    {
      "epoch": 0.37968762896998265,
      "grad_norm": 1.4047032077109822,
      "learning_rate": 7.121582914959544e-06,
      "loss": 0.8411,
      "step": 5175
    },
    {
      "epoch": 0.3797609985601218,
      "grad_norm": 0.5986026202571969,
      "learning_rate": 7.120506926228757e-06,
      "loss": 0.5227,
      "step": 5176
    },
    {
      "epoch": 0.3797609985601218,
      "lm_loss": 1.3713,
      "step": 5176,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.3797609985601218,
      "lm_loss": 1.0203,
      "step": 5176,
      "vm_loss": 0.0503
    },
    {
      "epoch": 0.3797609985601218,
      "lm_loss": 0.8435,
      "step": 5176,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.3797609985601218,
      "lm_loss": 0.8728,
      "step": 5176,
      "vm_loss": 0.1858
    },
    {
      "epoch": 0.3797609985601218,
      "lm_loss": 1.5623,
      "step": 5176,
      "vm_loss": 0.1265
    },
    {
      "epoch": 0.3797609985601218,
      "lm_loss": 0.1338,
      "step": 5176,
      "vm_loss": 0.044
    },
    {
      "epoch": 0.3797609985601218,
      "lm_loss": 0.5541,
      "step": 5176,
      "vm_loss": 0.1268
    },
    {
      "epoch": 0.3797609985601218,
      "lm_loss": 0.2929,
      "step": 5176,
      "vm_loss": 0.1526
    },
    {
      "epoch": 0.37983436815026095,
      "grad_norm": 1.5217650287185684,
      "learning_rate": 7.119430817747693e-06,
      "loss": 0.7674,
      "step": 5177
    },
    {
      "epoch": 0.37990773774040004,
      "grad_norm": 1.4901529947327152,
      "learning_rate": 7.118354589577122e-06,
      "loss": 0.8262,
      "step": 5178
    },
    {
      "epoch": 0.3799811073305392,
      "grad_norm": 1.524021783024243,
      "learning_rate": 7.117278241777823e-06,
      "loss": 0.8616,
      "step": 5179
    },
    {
      "epoch": 0.3800544769206783,
      "grad_norm": 1.3994282290486657,
      "learning_rate": 7.116201774410577e-06,
      "loss": 0.8527,
      "step": 5180
    },
    {
      "epoch": 0.38012784651081744,
      "grad_norm": 0.5677290418188656,
      "learning_rate": 7.115125187536175e-06,
      "loss": 0.5143,
      "step": 5181
    },
    {
      "epoch": 0.38020121610095653,
      "grad_norm": 1.4552429266005689,
      "learning_rate": 7.1140484812154166e-06,
      "loss": 0.7873,
      "step": 5182
    },
    {
      "epoch": 0.3802745856910957,
      "grad_norm": 1.3985984035477848,
      "learning_rate": 7.112971655509105e-06,
      "loss": 0.8096,
      "step": 5183
    },
    {
      "epoch": 0.38034795528123483,
      "grad_norm": 1.5205311881486816,
      "learning_rate": 7.111894710478051e-06,
      "loss": 0.794,
      "step": 5184
    },
    {
      "epoch": 0.38034795528123483,
      "lm_loss": 0.9574,
      "step": 5184,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.38034795528123483,
      "lm_loss": 0.6013,
      "step": 5184,
      "vm_loss": 0.1236
    },
    {
      "epoch": 0.38034795528123483,
      "lm_loss": 1.1198,
      "step": 5184,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.38034795528123483,
      "lm_loss": 0.1491,
      "step": 5184,
      "vm_loss": 0.1496
    },
    {
      "epoch": 0.38034795528123483,
      "lm_loss": 0.1402,
      "step": 5184,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.38034795528123483,
      "lm_loss": 0.6587,
      "step": 5184,
      "vm_loss": 0.0914
    },
    {
      "epoch": 0.38034795528123483,
      "lm_loss": 0.9111,
      "step": 5184,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.38034795528123483,
      "lm_loss": 0.5354,
      "step": 5184,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.3804213248713739,
      "grad_norm": 1.628196396852385,
      "learning_rate": 7.110817646183074e-06,
      "loss": 0.8394,
      "step": 5185
    },
    {
      "epoch": 0.38049469446151307,
      "grad_norm": 1.6848151197860655,
      "learning_rate": 7.109740462684995e-06,
      "loss": 0.7317,
      "step": 5186
    },
    {
      "epoch": 0.38056806405165217,
      "grad_norm": 0.5843948093061038,
      "learning_rate": 7.108663160044648e-06,
      "loss": 0.51,
      "step": 5187
    },
    {
      "epoch": 0.3806414336417913,
      "grad_norm": 1.4014300934239812,
      "learning_rate": 7.107585738322872e-06,
      "loss": 0.7214,
      "step": 5188
    },
    {
      "epoch": 0.38071480323193047,
      "grad_norm": 1.3261585031083003,
      "learning_rate": 7.1065081975805086e-06,
      "loss": 0.7753,
      "step": 5189
    },
    {
      "epoch": 0.38078817282206956,
      "grad_norm": 0.5664747657542536,
      "learning_rate": 7.10543053787841e-06,
      "loss": 0.5005,
      "step": 5190
    },
    {
      "epoch": 0.3808615424122087,
      "grad_norm": 1.6254465747021123,
      "learning_rate": 7.104352759277437e-06,
      "loss": 0.8628,
      "step": 5191
    },
    {
      "epoch": 0.3809349120023478,
      "grad_norm": 0.561682246514013,
      "learning_rate": 7.10327486183845e-06,
      "loss": 0.5245,
      "step": 5192
    },
    {
      "epoch": 0.3809349120023478,
      "lm_loss": 1.0286,
      "step": 5192,
      "vm_loss": 0.0408
    },
    {
      "epoch": 0.3809349120023478,
      "lm_loss": 1.1091,
      "step": 5192,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.3809349120023478,
      "lm_loss": 0.3372,
      "step": 5192,
      "vm_loss": 0.1039
    },
    {
      "epoch": 0.3809349120023478,
      "lm_loss": 0.3201,
      "step": 5192,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.3809349120023478,
      "lm_loss": 1.1688,
      "step": 5192,
      "vm_loss": 0.1251
    },
    {
      "epoch": 0.3809349120023478,
      "lm_loss": 0.7479,
      "step": 5192,
      "vm_loss": 0.2322
    },
    {
      "epoch": 0.3809349120023478,
      "lm_loss": 0.3553,
      "step": 5192,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.3809349120023478,
      "lm_loss": 0.7242,
      "step": 5192,
      "vm_loss": 0.2284
    },
    {
      "epoch": 0.38100828159248695,
      "grad_norm": 1.5752944882848225,
      "learning_rate": 7.102196845622324e-06,
      "loss": 0.8939,
      "step": 5193
    },
    {
      "epoch": 0.3810816511826261,
      "grad_norm": 1.5936862942537269,
      "learning_rate": 7.101118710689936e-06,
      "loss": 0.865,
      "step": 5194
    },
    {
      "epoch": 0.3811550207727652,
      "grad_norm": 0.5859842502354032,
      "learning_rate": 7.100040457102172e-06,
      "loss": 0.5102,
      "step": 5195
    },
    {
      "epoch": 0.38122839036290435,
      "grad_norm": 1.3970096778523708,
      "learning_rate": 7.098962084919923e-06,
      "loss": 0.7766,
      "step": 5196
    },
    {
      "epoch": 0.38130175995304344,
      "grad_norm": 1.4659922022620067,
      "learning_rate": 7.097883594204085e-06,
      "loss": 0.6984,
      "step": 5197
    },
    {
      "epoch": 0.3813751295431826,
      "grad_norm": 1.566223542859782,
      "learning_rate": 7.096804985015566e-06,
      "loss": 0.7308,
      "step": 5198
    },
    {
      "epoch": 0.38144849913332174,
      "grad_norm": 1.5545370510578778,
      "learning_rate": 7.095726257415276e-06,
      "loss": 0.7668,
      "step": 5199
    },
    {
      "epoch": 0.38152186872346083,
      "grad_norm": 0.6281784538937237,
      "learning_rate": 7.094647411464136e-06,
      "loss": 0.5127,
      "step": 5200
    },
    {
      "epoch": 0.38152186872346083,
      "lm_loss": 2.7343,
      "step": 5200,
      "vm_loss": 0.0612
    },
    {
      "epoch": 0.38152186872346083,
      "lm_loss": 1.5432,
      "step": 5200,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.38152186872346083,
      "lm_loss": 0.5569,
      "step": 5200,
      "vm_loss": 0.0598
    },
    {
      "epoch": 0.38152186872346083,
      "lm_loss": 0.7946,
      "step": 5200,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.38152186872346083,
      "lm_loss": 1.0574,
      "step": 5200,
      "vm_loss": 0.1773
    },
    {
      "epoch": 0.38152186872346083,
      "lm_loss": 0.0817,
      "step": 5200,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.38152186872346083,
      "lm_loss": 0.785,
      "step": 5200,
      "vm_loss": 0.1766
    },
    {
      "epoch": 0.38152186872346083,
      "lm_loss": 0.5904,
      "step": 5200,
      "vm_loss": 0.0713
    },
    {
      "epoch": 0.3815952383136,
      "grad_norm": 1.3760274057432293,
      "learning_rate": 7.093568447223069e-06,
      "loss": 0.8566,
      "step": 5201
    },
    {
      "epoch": 0.3816686079037391,
      "grad_norm": 0.637000100901594,
      "learning_rate": 7.092489364753006e-06,
      "loss": 0.5191,
      "step": 5202
    },
    {
      "epoch": 0.3817419774938782,
      "grad_norm": 1.5497923390097863,
      "learning_rate": 7.091410164114888e-06,
      "loss": 0.7901,
      "step": 5203
    },
    {
      "epoch": 0.3818153470840174,
      "grad_norm": 1.4437085520316921,
      "learning_rate": 7.090330845369658e-06,
      "loss": 0.7388,
      "step": 5204
    },
    {
      "epoch": 0.38188871667415647,
      "grad_norm": 1.5083774260736729,
      "learning_rate": 7.0892514085782685e-06,
      "loss": 0.8777,
      "step": 5205
    },
    {
      "epoch": 0.3819620862642956,
      "grad_norm": 1.6757132520464546,
      "learning_rate": 7.088171853801679e-06,
      "loss": 0.8564,
      "step": 5206
    },
    {
      "epoch": 0.3820354558544347,
      "grad_norm": 0.6203305326503133,
      "learning_rate": 7.0870921811008535e-06,
      "loss": 0.5146,
      "step": 5207
    },
    {
      "epoch": 0.38210882544457386,
      "grad_norm": 1.475914670793849,
      "learning_rate": 7.086012390536762e-06,
      "loss": 0.7968,
      "step": 5208
    },
    {
      "epoch": 0.38210882544457386,
      "lm_loss": 0.5869,
      "step": 5208,
      "vm_loss": 0.0591
    },
    {
      "epoch": 0.38210882544457386,
      "lm_loss": 0.8474,
      "step": 5208,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.38210882544457386,
      "lm_loss": 0.4671,
      "step": 5208,
      "vm_loss": 0.1985
    },
    {
      "epoch": 0.38210882544457386,
      "lm_loss": 0.5425,
      "step": 5208,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.38210882544457386,
      "lm_loss": 1.1271,
      "step": 5208,
      "vm_loss": 0.1211
    },
    {
      "epoch": 0.38210882544457386,
      "lm_loss": 0.4074,
      "step": 5208,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.38210882544457386,
      "lm_loss": 0.068,
      "step": 5208,
      "vm_loss": 0.1564
    },
    {
      "epoch": 0.38210882544457386,
      "lm_loss": 0.3743,
      "step": 5208,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.382182195034713,
      "grad_norm": 1.4352176620790733,
      "learning_rate": 7.084932482170385e-06,
      "loss": 0.7081,
      "step": 5209
    },
    {
      "epoch": 0.3822555646248521,
      "grad_norm": 0.6348582709115933,
      "learning_rate": 7.08385245606271e-06,
      "loss": 0.5241,
      "step": 5210
    },
    {
      "epoch": 0.38232893421499126,
      "grad_norm": 1.445985825519703,
      "learning_rate": 7.082772312274725e-06,
      "loss": 0.7921,
      "step": 5211
    },
    {
      "epoch": 0.38240230380513035,
      "grad_norm": 0.5652541266134871,
      "learning_rate": 7.0816920508674296e-06,
      "loss": 0.5023,
      "step": 5212
    },
    {
      "epoch": 0.3824756733952695,
      "grad_norm": 0.5856417159160909,
      "learning_rate": 7.0806116719018285e-06,
      "loss": 0.5341,
      "step": 5213
    },
    {
      "epoch": 0.38254904298540865,
      "grad_norm": 1.3642304272055719,
      "learning_rate": 7.079531175438934e-06,
      "loss": 0.8052,
      "step": 5214
    },
    {
      "epoch": 0.38262241257554774,
      "grad_norm": 1.6459645193898476,
      "learning_rate": 7.078450561539765e-06,
      "loss": 0.8657,
      "step": 5215
    },
    {
      "epoch": 0.3826957821656869,
      "grad_norm": 1.566805638896517,
      "learning_rate": 7.0773698302653435e-06,
      "loss": 0.8315,
      "step": 5216
    },
    {
      "epoch": 0.3826957821656869,
      "lm_loss": 0.5317,
      "step": 5216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3826957821656869,
      "lm_loss": 0.3912,
      "step": 5216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3826957821656869,
      "lm_loss": 0.4698,
      "step": 5216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3826957821656869,
      "lm_loss": 0.6153,
      "step": 5216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3826957821656869,
      "lm_loss": 0.5145,
      "step": 5216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3826957821656869,
      "lm_loss": 0.4052,
      "step": 5216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3826957821656869,
      "lm_loss": 0.5302,
      "step": 5216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3826957821656869,
      "lm_loss": 0.5281,
      "step": 5216,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.382769151755826,
      "grad_norm": 0.6537916989596088,
      "learning_rate": 7.076288981676704e-06,
      "loss": 0.534,
      "step": 5217
    },
    {
      "epoch": 0.38284252134596514,
      "grad_norm": 1.6244667390608991,
      "learning_rate": 7.075208015834884e-06,
      "loss": 0.8114,
      "step": 5218
    },
    {
      "epoch": 0.38291589093610423,
      "grad_norm": 1.8828614995062836,
      "learning_rate": 7.0741269328009286e-06,
      "loss": 0.7896,
      "step": 5219
    },
    {
      "epoch": 0.3829892605262434,
      "grad_norm": 0.59842385511375,
      "learning_rate": 7.073045732635889e-06,
      "loss": 0.5035,
      "step": 5220
    },
    {
      "epoch": 0.38306263011638253,
      "grad_norm": 0.6269384396172989,
      "learning_rate": 7.071964415400822e-06,
      "loss": 0.5123,
      "step": 5221
    },
    {
      "epoch": 0.3831359997065216,
      "grad_norm": 0.5408733808101804,
      "learning_rate": 7.070882981156795e-06,
      "loss": 0.5064,
      "step": 5222
    },
    {
      "epoch": 0.3832093692966608,
      "grad_norm": 0.5676769712056521,
      "learning_rate": 7.069801429964878e-06,
      "loss": 0.5019,
      "step": 5223
    },
    {
      "epoch": 0.38328273888679987,
      "grad_norm": 0.5746350185295435,
      "learning_rate": 7.068719761886146e-06,
      "loss": 0.5121,
      "step": 5224
    },
    {
      "epoch": 0.38328273888679987,
      "lm_loss": 0.6108,
      "step": 5224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38328273888679987,
      "lm_loss": 0.4635,
      "step": 5224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38328273888679987,
      "lm_loss": 0.5742,
      "step": 5224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38328273888679987,
      "lm_loss": 0.3981,
      "step": 5224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38328273888679987,
      "lm_loss": 0.3773,
      "step": 5224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38328273888679987,
      "lm_loss": 0.5628,
      "step": 5224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38328273888679987,
      "lm_loss": 0.3522,
      "step": 5224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38328273888679987,
      "lm_loss": 0.6352,
      "step": 5224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.383356108476939,
      "grad_norm": 0.5748572870042628,
      "learning_rate": 7.067637976981687e-06,
      "loss": 0.5238,
      "step": 5225
    },
    {
      "epoch": 0.38342947806707817,
      "grad_norm": 2.1734205631263475,
      "learning_rate": 7.066556075312591e-06,
      "loss": 0.7988,
      "step": 5226
    },
    {
      "epoch": 0.38350284765721726,
      "grad_norm": 1.7881392422053783,
      "learning_rate": 7.065474056939956e-06,
      "loss": 0.8723,
      "step": 5227
    },
    {
      "epoch": 0.3835762172473564,
      "grad_norm": 1.7305553760880639,
      "learning_rate": 7.064391921924886e-06,
      "loss": 0.7226,
      "step": 5228
    },
    {
      "epoch": 0.3836495868374955,
      "grad_norm": 0.5749720141752075,
      "learning_rate": 7.063309670328491e-06,
      "loss": 0.5145,
      "step": 5229
    },
    {
      "epoch": 0.38372295642763465,
      "grad_norm": 2.3008647511805713,
      "learning_rate": 7.06222730221189e-06,
      "loss": 0.8555,
      "step": 5230
    },
    {
      "epoch": 0.3837963260177738,
      "grad_norm": 1.7528467748741665,
      "learning_rate": 7.061144817636205e-06,
      "loss": 0.8136,
      "step": 5231
    },
    {
      "epoch": 0.3838696956079129,
      "grad_norm": 1.6411109386716478,
      "learning_rate": 7.060062216662568e-06,
      "loss": 0.7585,
      "step": 5232
    },
    {
      "epoch": 0.3838696956079129,
      "lm_loss": 1.6389,
      "step": 5232,
      "vm_loss": 0.1009
    },
    {
      "epoch": 0.3838696956079129,
      "lm_loss": 1.0195,
      "step": 5232,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.3838696956079129,
      "lm_loss": 0.2352,
      "step": 5232,
      "vm_loss": 0.144
    },
    {
      "epoch": 0.3838696956079129,
      "lm_loss": 0.0871,
      "step": 5232,
      "vm_loss": 0.1778
    },
    {
      "epoch": 0.3838696956079129,
      "lm_loss": 0.3739,
      "step": 5232,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.3838696956079129,
      "lm_loss": 0.4033,
      "step": 5232,
      "vm_loss": 0.0835
    },
    {
      "epoch": 0.3838696956079129,
      "lm_loss": 0.5237,
      "step": 5232,
      "vm_loss": 0.1435
    },
    {
      "epoch": 0.3838696956079129,
      "lm_loss": 0.1287,
      "step": 5232,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.38394306519805205,
      "grad_norm": 1.4561096011556014,
      "learning_rate": 7.0589794993521175e-06,
      "loss": 0.8181,
      "step": 5233
    },
    {
      "epoch": 0.38401643478819114,
      "grad_norm": 0.5883641506263501,
      "learning_rate": 7.0578966657659955e-06,
      "loss": 0.5015,
      "step": 5234
    },
    {
      "epoch": 0.3840898043783303,
      "grad_norm": 1.5564479018463628,
      "learning_rate": 7.056813715965351e-06,
      "loss": 0.9107,
      "step": 5235
    },
    {
      "epoch": 0.38416317396846944,
      "grad_norm": 1.8884422449921017,
      "learning_rate": 7.055730650011344e-06,
      "loss": 0.8127,
      "step": 5236
    },
    {
      "epoch": 0.38423654355860853,
      "grad_norm": 1.6203729080109774,
      "learning_rate": 7.0546474679651344e-06,
      "loss": 0.8077,
      "step": 5237
    },
    {
      "epoch": 0.3843099131487477,
      "grad_norm": 1.4212769473556597,
      "learning_rate": 7.053564169887897e-06,
      "loss": 0.7242,
      "step": 5238
    },
    {
      "epoch": 0.3843832827388868,
      "grad_norm": 1.4713962437765031,
      "learning_rate": 7.052480755840803e-06,
      "loss": 0.656,
      "step": 5239
    },
    {
      "epoch": 0.3844566523290259,
      "grad_norm": 1.7118218536571748,
      "learning_rate": 7.051397225885037e-06,
      "loss": 0.7686,
      "step": 5240
    },
    {
      "epoch": 0.3844566523290259,
      "lm_loss": 0.5886,
      "step": 5240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3844566523290259,
      "lm_loss": 0.4494,
      "step": 5240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3844566523290259,
      "lm_loss": 0.4033,
      "step": 5240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3844566523290259,
      "lm_loss": 0.441,
      "step": 5240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3844566523290259,
      "lm_loss": 0.6002,
      "step": 5240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3844566523290259,
      "lm_loss": 0.4322,
      "step": 5240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3844566523290259,
      "lm_loss": 0.7216,
      "step": 5240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3844566523290259,
      "lm_loss": 0.4968,
      "step": 5240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3845300219191651,
      "grad_norm": 0.6426441176865008,
      "learning_rate": 7.050313580081793e-06,
      "loss": 0.532,
      "step": 5241
    },
    {
      "epoch": 0.38460339150930417,
      "grad_norm": 1.3945517482252656,
      "learning_rate": 7.0492298184922606e-06,
      "loss": 0.7599,
      "step": 5242
    },
    {
      "epoch": 0.3846767610994433,
      "grad_norm": 0.6081319188383119,
      "learning_rate": 7.0481459411776465e-06,
      "loss": 0.5224,
      "step": 5243
    },
    {
      "epoch": 0.3847501306895824,
      "grad_norm": 1.7604732900154103,
      "learning_rate": 7.047061948199159e-06,
      "loss": 0.813,
      "step": 5244
    },
    {
      "epoch": 0.38482350027972156,
      "grad_norm": 1.5231518756659845,
      "learning_rate": 7.045977839618012e-06,
      "loss": 0.8376,
      "step": 5245
    },
    {
      "epoch": 0.3848968698698607,
      "grad_norm": 0.5739109163202396,
      "learning_rate": 7.044893615495432e-06,
      "loss": 0.5073,
      "step": 5246
    },
    {
      "epoch": 0.3849702394599998,
      "grad_norm": 1.4866320058382025,
      "learning_rate": 7.043809275892646e-06,
      "loss": 0.7795,
      "step": 5247
    },
    {
      "epoch": 0.38504360905013896,
      "grad_norm": 0.6558855669160297,
      "learning_rate": 7.0427248208708856e-06,
      "loss": 0.5102,
      "step": 5248
    },
    {
      "epoch": 0.38504360905013896,
      "lm_loss": 0.2141,
      "step": 5248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38504360905013896,
      "lm_loss": 0.6987,
      "step": 5248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38504360905013896,
      "lm_loss": 0.4822,
      "step": 5248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38504360905013896,
      "lm_loss": 0.652,
      "step": 5248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38504360905013896,
      "lm_loss": 0.6154,
      "step": 5248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38504360905013896,
      "lm_loss": 0.4013,
      "step": 5248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38504360905013896,
      "lm_loss": 0.3382,
      "step": 5248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38504360905013896,
      "lm_loss": 0.5192,
      "step": 5248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38511697864027805,
      "grad_norm": 0.5838491612002088,
      "learning_rate": 7.041640250491398e-06,
      "loss": 0.5023,
      "step": 5249
    },
    {
      "epoch": 0.3851903482304172,
      "grad_norm": 1.588864527464154,
      "learning_rate": 7.04055556481543e-06,
      "loss": 0.8332,
      "step": 5250
    },
    {
      "epoch": 0.38526371782055635,
      "grad_norm": 0.553034730885302,
      "learning_rate": 7.039470763904235e-06,
      "loss": 0.4982,
      "step": 5251
    },
    {
      "epoch": 0.38533708741069544,
      "grad_norm": 1.5777243981519042,
      "learning_rate": 7.038385847819074e-06,
      "loss": 0.8372,
      "step": 5252
    },
    {
      "epoch": 0.3854104570008346,
      "grad_norm": 1.5403396719147753,
      "learning_rate": 7.037300816621217e-06,
      "loss": 0.9193,
      "step": 5253
    },
    {
      "epoch": 0.3854838265909737,
      "grad_norm": 0.672688635717062,
      "learning_rate": 7.03621567037194e-06,
      "loss": 0.501,
      "step": 5254
    },
    {
      "epoch": 0.38555719618111284,
      "grad_norm": 0.6322963964715853,
      "learning_rate": 7.03513040913252e-06,
      "loss": 0.5341,
      "step": 5255
    },
    {
      "epoch": 0.385630565771252,
      "grad_norm": 1.613565678148655,
      "learning_rate": 7.034045032964245e-06,
      "loss": 0.7616,
      "step": 5256
    },
    {
      "epoch": 0.385630565771252,
      "lm_loss": 1.6881,
      "step": 5256,
      "vm_loss": 0.0403
    },
    {
      "epoch": 0.385630565771252,
      "lm_loss": 0.0686,
      "step": 5256,
      "vm_loss": 0.1538
    },
    {
      "epoch": 0.385630565771252,
      "lm_loss": 0.8091,
      "step": 5256,
      "vm_loss": 0.1478
    },
    {
      "epoch": 0.385630565771252,
      "lm_loss": 0.0794,
      "step": 5256,
      "vm_loss": 0.1531
    },
    {
      "epoch": 0.385630565771252,
      "lm_loss": 0.3361,
      "step": 5256,
      "vm_loss": 0.1315
    },
    {
      "epoch": 0.385630565771252,
      "lm_loss": 0.699,
      "step": 5256,
      "vm_loss": 0.1269
    },
    {
      "epoch": 0.385630565771252,
      "lm_loss": 0.5901,
      "step": 5256,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.385630565771252,
      "lm_loss": 0.325,
      "step": 5256,
      "vm_loss": 0.1227
    },
    {
      "epoch": 0.3857039353613911,
      "grad_norm": 1.585002858912973,
      "learning_rate": 7.03295954192841e-06,
      "loss": 0.8146,
      "step": 5257
    },
    {
      "epoch": 0.38577730495153023,
      "grad_norm": 1.5139507117724211,
      "learning_rate": 7.031873936086315e-06,
      "loss": 0.8073,
      "step": 5258
    },
    {
      "epoch": 0.3858506745416693,
      "grad_norm": 0.6510823202127098,
      "learning_rate": 7.030788215499268e-06,
      "loss": 0.5229,
      "step": 5259
    },
    {
      "epoch": 0.3859240441318085,
      "grad_norm": 2.3603053186533476,
      "learning_rate": 7.0297023802285815e-06,
      "loss": 0.913,
      "step": 5260
    },
    {
      "epoch": 0.38599741372194757,
      "grad_norm": 0.594494974913835,
      "learning_rate": 7.028616430335573e-06,
      "loss": 0.5059,
      "step": 5261
    },
    {
      "epoch": 0.3860707833120867,
      "grad_norm": 0.5744942465607843,
      "learning_rate": 7.027530365881573e-06,
      "loss": 0.5122,
      "step": 5262
    },
    {
      "epoch": 0.38614415290222587,
      "grad_norm": 0.5719497582364697,
      "learning_rate": 7.026444186927912e-06,
      "loss": 0.5102,
      "step": 5263
    },
    {
      "epoch": 0.38621752249236496,
      "grad_norm": 1.7971877779302665,
      "learning_rate": 7.025357893535928e-06,
      "loss": 0.7955,
      "step": 5264
    },
    {
      "epoch": 0.38621752249236496,
      "lm_loss": 1.6058,
      "step": 5264,
      "vm_loss": 0.07
    },
    {
      "epoch": 0.38621752249236496,
      "lm_loss": 1.7352,
      "step": 5264,
      "vm_loss": 0.0605
    },
    {
      "epoch": 0.38621752249236496,
      "lm_loss": 0.8359,
      "step": 5264,
      "vm_loss": 0.2387
    },
    {
      "epoch": 0.38621752249236496,
      "lm_loss": 0.9153,
      "step": 5264,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.38621752249236496,
      "lm_loss": 0.5665,
      "step": 5264,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.38621752249236496,
      "lm_loss": 0.6869,
      "step": 5264,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.38621752249236496,
      "lm_loss": 0.2275,
      "step": 5264,
      "vm_loss": 0.1335
    },
    {
      "epoch": 0.38621752249236496,
      "lm_loss": 0.1039,
      "step": 5264,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.3862908920825041,
      "grad_norm": 1.5380757159526346,
      "learning_rate": 7.024271485766969e-06,
      "loss": 0.8782,
      "step": 5265
    },
    {
      "epoch": 0.3863642616726432,
      "grad_norm": 0.642737916746398,
      "learning_rate": 7.023184963682385e-06,
      "loss": 0.5401,
      "step": 5266
    },
    {
      "epoch": 0.38643763126278236,
      "grad_norm": 1.463535955692027,
      "learning_rate": 7.0220983273435375e-06,
      "loss": 0.7995,
      "step": 5267
    },
    {
      "epoch": 0.3865110008529215,
      "grad_norm": 1.5605802550399275,
      "learning_rate": 7.0210115768117894e-06,
      "loss": 0.8752,
      "step": 5268
    },
    {
      "epoch": 0.3865843704430606,
      "grad_norm": 1.843690683020869,
      "learning_rate": 7.019924712148511e-06,
      "loss": 0.7471,
      "step": 5269
    },
    {
      "epoch": 0.38665774003319975,
      "grad_norm": 1.4975267650774282,
      "learning_rate": 7.018837733415083e-06,
      "loss": 0.9019,
      "step": 5270
    },
    {
      "epoch": 0.38673110962333884,
      "grad_norm": 0.6323634345961764,
      "learning_rate": 7.017750640672889e-06,
      "loss": 0.5065,
      "step": 5271
    },
    {
      "epoch": 0.386804479213478,
      "grad_norm": 1.546364966593114,
      "learning_rate": 7.016663433983319e-06,
      "loss": 0.8488,
      "step": 5272
    },
    {
      "epoch": 0.386804479213478,
      "lm_loss": 0.1867,
      "step": 5272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.386804479213478,
      "lm_loss": 0.541,
      "step": 5272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.386804479213478,
      "lm_loss": 0.5826,
      "step": 5272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.386804479213478,
      "lm_loss": 0.4745,
      "step": 5272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.386804479213478,
      "lm_loss": 0.4843,
      "step": 5272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.386804479213478,
      "lm_loss": 0.5401,
      "step": 5272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.386804479213478,
      "lm_loss": 0.4682,
      "step": 5272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.386804479213478,
      "lm_loss": 0.471,
      "step": 5272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.38687784880361714,
      "grad_norm": 0.5908072300129996,
      "learning_rate": 7.01557611340777e-06,
      "loss": 0.4926,
      "step": 5273
    },
    {
      "epoch": 0.38695121839375624,
      "grad_norm": 0.5824041055041274,
      "learning_rate": 7.0144886790076475e-06,
      "loss": 0.531,
      "step": 5274
    },
    {
      "epoch": 0.3870245879838954,
      "grad_norm": 1.2431630563432143,
      "learning_rate": 7.013401130844361e-06,
      "loss": 0.7453,
      "step": 5275
    },
    {
      "epoch": 0.3870979575740345,
      "grad_norm": 1.6330317363390454,
      "learning_rate": 7.012313468979326e-06,
      "loss": 0.9026,
      "step": 5276
    },
    {
      "epoch": 0.38717132716417363,
      "grad_norm": 1.533432851527553,
      "learning_rate": 7.011225693473964e-06,
      "loss": 0.8124,
      "step": 5277
    },
    {
      "epoch": 0.3872446967543128,
      "grad_norm": 1.504775536290155,
      "learning_rate": 7.010137804389709e-06,
      "loss": 0.8121,
      "step": 5278
    },
    {
      "epoch": 0.3873180663444519,
      "grad_norm": 1.525724148187808,
      "learning_rate": 7.009049801787994e-06,
      "loss": 0.8444,
      "step": 5279
    },
    {
      "epoch": 0.387391435934591,
      "grad_norm": 1.5668933165144265,
      "learning_rate": 7.0079616857302605e-06,
      "loss": 0.9195,
      "step": 5280
    },
    {
      "epoch": 0.387391435934591,
      "lm_loss": 1.4134,
      "step": 5280,
      "vm_loss": 0.0732
    },
    {
      "epoch": 0.387391435934591,
      "lm_loss": 1.3994,
      "step": 5280,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.387391435934591,
      "lm_loss": 0.851,
      "step": 5280,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.387391435934591,
      "lm_loss": 0.7665,
      "step": 5280,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.387391435934591,
      "lm_loss": 0.8053,
      "step": 5280,
      "vm_loss": 0.1744
    },
    {
      "epoch": 0.387391435934591,
      "lm_loss": 0.9852,
      "step": 5280,
      "vm_loss": 0.2306
    },
    {
      "epoch": 0.387391435934591,
      "lm_loss": 0.7457,
      "step": 5280,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.387391435934591,
      "lm_loss": 0.7964,
      "step": 5280,
      "vm_loss": 0.1698
    },
    {
      "epoch": 0.3874648055247301,
      "grad_norm": 1.6564129041468383,
      "learning_rate": 7.006873456277957e-06,
      "loss": 0.8919,
      "step": 5281
    },
    {
      "epoch": 0.38753817511486927,
      "grad_norm": 1.5008318128593783,
      "learning_rate": 7.005785113492542e-06,
      "loss": 0.7664,
      "step": 5282
    },
    {
      "epoch": 0.3876115447050084,
      "grad_norm": 1.7274282849358737,
      "learning_rate": 7.004696657435472e-06,
      "loss": 0.8611,
      "step": 5283
    },
    {
      "epoch": 0.3876849142951475,
      "grad_norm": 1.4384362079488866,
      "learning_rate": 7.003608088168219e-06,
      "loss": 0.8768,
      "step": 5284
    },
    {
      "epoch": 0.38775828388528666,
      "grad_norm": 1.430362350335634,
      "learning_rate": 7.002519405752253e-06,
      "loss": 0.5081,
      "step": 5285
    },
    {
      "epoch": 0.38783165347542575,
      "grad_norm": 1.7494313383141153,
      "learning_rate": 7.001430610249059e-06,
      "loss": 0.793,
      "step": 5286
    },
    {
      "epoch": 0.3879050230655649,
      "grad_norm": 1.6625480251107092,
      "learning_rate": 7.000341701720122e-06,
      "loss": 0.8328,
      "step": 5287
    },
    {
      "epoch": 0.38797839265570405,
      "grad_norm": 1.5352023712937526,
      "learning_rate": 6.999252680226935e-06,
      "loss": 0.8649,
      "step": 5288
    },
    {
      "epoch": 0.38797839265570405,
      "lm_loss": 1.3061,
      "step": 5288,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.38797839265570405,
      "lm_loss": 0.9834,
      "step": 5288,
      "vm_loss": 0.2182
    },
    {
      "epoch": 0.38797839265570405,
      "lm_loss": 0.7213,
      "step": 5288,
      "vm_loss": 0.2185
    },
    {
      "epoch": 0.38797839265570405,
      "lm_loss": 0.2364,
      "step": 5288,
      "vm_loss": 0.1245
    },
    {
      "epoch": 0.38797839265570405,
      "lm_loss": 1.1147,
      "step": 5288,
      "vm_loss": 0.1834
    },
    {
      "epoch": 0.38797839265570405,
      "lm_loss": 0.4476,
      "step": 5288,
      "vm_loss": 0.098
    },
    {
      "epoch": 0.38797839265570405,
      "lm_loss": 0.5298,
      "step": 5288,
      "vm_loss": 0.2447
    },
    {
      "epoch": 0.38797839265570405,
      "lm_loss": 0.1022,
      "step": 5288,
      "vm_loss": 0.114
    },
    {
      "epoch": 0.38805176224584315,
      "grad_norm": 1.31896282933461,
      "learning_rate": 6.998163545830998e-06,
      "loss": 0.7634,
      "step": 5289
    },
    {
      "epoch": 0.3881251318359823,
      "grad_norm": 1.6138833754570603,
      "learning_rate": 6.9970742985938176e-06,
      "loss": 0.7461,
      "step": 5290
    },
    {
      "epoch": 0.3881985014261214,
      "grad_norm": 1.4537027206022686,
      "learning_rate": 6.9959849385769055e-06,
      "loss": 0.7676,
      "step": 5291
    },
    {
      "epoch": 0.38827187101626054,
      "grad_norm": 0.6985590744241379,
      "learning_rate": 6.99489546584178e-06,
      "loss": 0.5203,
      "step": 5292
    },
    {
      "epoch": 0.3883452406063997,
      "grad_norm": 1.607198000204387,
      "learning_rate": 6.993805880449968e-06,
      "loss": 0.659,
      "step": 5293
    },
    {
      "epoch": 0.3884186101965388,
      "grad_norm": 0.6422150383621261,
      "learning_rate": 6.9927161824630005e-06,
      "loss": 0.5193,
      "step": 5294
    },
    {
      "epoch": 0.38849197978667793,
      "grad_norm": 1.5882639157062626,
      "learning_rate": 6.991626371942416e-06,
      "loss": 0.8208,
      "step": 5295
    },
    {
      "epoch": 0.388565349376817,
      "grad_norm": 1.577566600996507,
      "learning_rate": 6.990536448949756e-06,
      "loss": 0.7568,
      "step": 5296
    },
    {
      "epoch": 0.388565349376817,
      "lm_loss": 1.3009,
      "step": 5296,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.388565349376817,
      "lm_loss": 0.6879,
      "step": 5296,
      "vm_loss": 0.0609
    },
    {
      "epoch": 0.388565349376817,
      "lm_loss": 0.2415,
      "step": 5296,
      "vm_loss": 0.1992
    },
    {
      "epoch": 0.388565349376817,
      "lm_loss": 0.8717,
      "step": 5296,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.388565349376817,
      "lm_loss": 0.1533,
      "step": 5296,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.388565349376817,
      "lm_loss": 0.5734,
      "step": 5296,
      "vm_loss": 0.0863
    },
    {
      "epoch": 0.388565349376817,
      "lm_loss": 0.4964,
      "step": 5296,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.388565349376817,
      "lm_loss": 0.5702,
      "step": 5296,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.3886387189669562,
      "grad_norm": 1.4078919051345602,
      "learning_rate": 6.989446413546575e-06,
      "loss": 0.7959,
      "step": 5297
    },
    {
      "epoch": 0.3887120885570953,
      "grad_norm": 1.5505391015326224,
      "learning_rate": 6.98835626579443e-06,
      "loss": 0.8029,
      "step": 5298
    },
    {
      "epoch": 0.3887854581472344,
      "grad_norm": 1.570344067546083,
      "learning_rate": 6.987266005754879e-06,
      "loss": 0.8071,
      "step": 5299
    },
    {
      "epoch": 0.38885882773737357,
      "grad_norm": 1.4389890576665842,
      "learning_rate": 6.986175633489497e-06,
      "loss": 0.8001,
      "step": 5300
    },
    {
      "epoch": 0.38893219732751266,
      "grad_norm": 0.7735901608006673,
      "learning_rate": 6.985085149059858e-06,
      "loss": 0.5252,
      "step": 5301
    },
    {
      "epoch": 0.3890055669176518,
      "grad_norm": 0.7170480079959141,
      "learning_rate": 6.983994552527544e-06,
      "loss": 0.5109,
      "step": 5302
    },
    {
      "epoch": 0.3890789365077909,
      "grad_norm": 0.6117154794949778,
      "learning_rate": 6.982903843954146e-06,
      "loss": 0.5045,
      "step": 5303
    },
    {
      "epoch": 0.38915230609793006,
      "grad_norm": 1.6623323336362572,
      "learning_rate": 6.981813023401256e-06,
      "loss": 0.791,
      "step": 5304
    },
    {
      "epoch": 0.38915230609793006,
      "lm_loss": 1.4003,
      "step": 5304,
      "vm_loss": 0.0815
    },
    {
      "epoch": 0.38915230609793006,
      "lm_loss": 1.2649,
      "step": 5304,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.38915230609793006,
      "lm_loss": 0.6033,
      "step": 5304,
      "vm_loss": 0.1768
    },
    {
      "epoch": 0.38915230609793006,
      "lm_loss": 0.7049,
      "step": 5304,
      "vm_loss": 0.1028
    },
    {
      "epoch": 0.38915230609793006,
      "lm_loss": 1.0432,
      "step": 5304,
      "vm_loss": 0.1329
    },
    {
      "epoch": 0.38915230609793006,
      "lm_loss": 0.2582,
      "step": 5304,
      "vm_loss": 0.1123
    },
    {
      "epoch": 0.38915230609793006,
      "lm_loss": 0.2343,
      "step": 5304,
      "vm_loss": 0.0869
    },
    {
      "epoch": 0.38915230609793006,
      "lm_loss": 0.7278,
      "step": 5304,
      "vm_loss": 0.1113
    },
    {
      "epoch": 0.3892256756880692,
      "grad_norm": 1.5440219120018723,
      "learning_rate": 6.980722090930478e-06,
      "loss": 0.8891,
      "step": 5305
    },
    {
      "epoch": 0.3892990452782083,
      "grad_norm": 1.5188939384679674,
      "learning_rate": 6.979631046603416e-06,
      "loss": 0.8147,
      "step": 5306
    },
    {
      "epoch": 0.38937241486834745,
      "grad_norm": 1.7004418022457646,
      "learning_rate": 6.978539890481688e-06,
      "loss": 0.8241,
      "step": 5307
    },
    {
      "epoch": 0.38944578445848654,
      "grad_norm": 0.8542137014535953,
      "learning_rate": 6.9774486226269135e-06,
      "loss": 0.5079,
      "step": 5308
    },
    {
      "epoch": 0.3895191540486257,
      "grad_norm": 0.7827052570778438,
      "learning_rate": 6.976357243100718e-06,
      "loss": 0.5394,
      "step": 5309
    },
    {
      "epoch": 0.38959252363876484,
      "grad_norm": 1.5839764462889987,
      "learning_rate": 6.975265751964732e-06,
      "loss": 0.8272,
      "step": 5310
    },
    {
      "epoch": 0.38966589322890394,
      "grad_norm": 0.5788704695744071,
      "learning_rate": 6.974174149280599e-06,
      "loss": 0.4986,
      "step": 5311
    },
    {
      "epoch": 0.3897392628190431,
      "grad_norm": 1.7297859942934224,
      "learning_rate": 6.973082435109964e-06,
      "loss": 0.7909,
      "step": 5312
    },
    {
      "epoch": 0.3897392628190431,
      "lm_loss": 0.5519,
      "step": 5312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3897392628190431,
      "lm_loss": 0.6285,
      "step": 5312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3897392628190431,
      "lm_loss": 0.515,
      "step": 5312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3897392628190431,
      "lm_loss": 0.4125,
      "step": 5312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3897392628190431,
      "lm_loss": 0.5398,
      "step": 5312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3897392628190431,
      "lm_loss": 0.4021,
      "step": 5312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3897392628190431,
      "lm_loss": 0.5605,
      "step": 5312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3897392628190431,
      "lm_loss": 0.53,
      "step": 5312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3898126324091822,
      "grad_norm": 0.6342270950898922,
      "learning_rate": 6.971990609514476e-06,
      "loss": 0.5136,
      "step": 5313
    },
    {
      "epoch": 0.38988600199932133,
      "grad_norm": 1.5402391175708727,
      "learning_rate": 6.970898672555796e-06,
      "loss": 0.7849,
      "step": 5314
    },
    {
      "epoch": 0.3899593715894605,
      "grad_norm": 1.5553492098953403,
      "learning_rate": 6.969806624295585e-06,
      "loss": 0.6731,
      "step": 5315
    },
    {
      "epoch": 0.3900327411795996,
      "grad_norm": 1.3792346368415298,
      "learning_rate": 6.968714464795517e-06,
      "loss": 0.8284,
      "step": 5316
    },
    {
      "epoch": 0.3901061107697387,
      "grad_norm": 0.66026003162583,
      "learning_rate": 6.96762219411727e-06,
      "loss": 0.5049,
      "step": 5317
    },
    {
      "epoch": 0.3901794803598778,
      "grad_norm": 1.486639032228896,
      "learning_rate": 6.966529812322521e-06,
      "loss": 0.7903,
      "step": 5318
    },
    {
      "epoch": 0.39025284995001697,
      "grad_norm": 1.6537459357043462,
      "learning_rate": 6.965437319472965e-06,
      "loss": 0.8368,
      "step": 5319
    },
    {
      "epoch": 0.3903262195401561,
      "grad_norm": 1.4768673787316486,
      "learning_rate": 6.964344715630297e-06,
      "loss": 0.8171,
      "step": 5320
    },
    {
      "epoch": 0.3903262195401561,
      "lm_loss": 1.6015,
      "step": 5320,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.3903262195401561,
      "lm_loss": 0.2589,
      "step": 5320,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.3903262195401561,
      "lm_loss": 0.5955,
      "step": 5320,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.3903262195401561,
      "lm_loss": 2.0979,
      "step": 5320,
      "vm_loss": 0.1575
    },
    {
      "epoch": 0.3903262195401561,
      "lm_loss": 1.1753,
      "step": 5320,
      "vm_loss": 0.068
    },
    {
      "epoch": 0.3903262195401561,
      "lm_loss": 0.1132,
      "step": 5320,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.3903262195401561,
      "lm_loss": 0.1059,
      "step": 5320,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.3903262195401561,
      "lm_loss": 0.5099,
      "step": 5320,
      "vm_loss": 0.1676
    },
    {
      "epoch": 0.3903995891302952,
      "grad_norm": 1.621371403803702,
      "learning_rate": 6.963252000856218e-06,
      "loss": 0.955,
      "step": 5321
    },
    {
      "epoch": 0.39047295872043436,
      "grad_norm": 1.6125522664960494,
      "learning_rate": 6.962159175212437e-06,
      "loss": 0.9199,
      "step": 5322
    },
    {
      "epoch": 0.39054632831057345,
      "grad_norm": 1.5425467026153052,
      "learning_rate": 6.961066238760667e-06,
      "loss": 0.8388,
      "step": 5323
    },
    {
      "epoch": 0.3906196979007126,
      "grad_norm": 1.4886342254747158,
      "learning_rate": 6.959973191562632e-06,
      "loss": 0.8231,
      "step": 5324
    },
    {
      "epoch": 0.39069306749085175,
      "grad_norm": 0.6013450943024957,
      "learning_rate": 6.958880033680055e-06,
      "loss": 0.5074,
      "step": 5325
    },
    {
      "epoch": 0.39076643708099085,
      "grad_norm": 0.6036035563613391,
      "learning_rate": 6.957786765174671e-06,
      "loss": 0.5105,
      "step": 5326
    },
    {
      "epoch": 0.39083980667113,
      "grad_norm": 1.766135074228775,
      "learning_rate": 6.956693386108222e-06,
      "loss": 0.8179,
      "step": 5327
    },
    {
      "epoch": 0.3909131762612691,
      "grad_norm": 1.418504678200895,
      "learning_rate": 6.95559989654245e-06,
      "loss": 0.7663,
      "step": 5328
    },
    {
      "epoch": 0.3909131762612691,
      "lm_loss": 1.1272,
      "step": 5328,
      "vm_loss": 0.0245
    },
    {
      "epoch": 0.3909131762612691,
      "lm_loss": 0.2831,
      "step": 5328,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.3909131762612691,
      "lm_loss": 0.0169,
      "step": 5328,
      "vm_loss": 0.1095
    },
    {
      "epoch": 0.3909131762612691,
      "lm_loss": 0.0964,
      "step": 5328,
      "vm_loss": 0.0692
    },
    {
      "epoch": 0.3909131762612691,
      "lm_loss": 0.1712,
      "step": 5328,
      "vm_loss": 0.1706
    },
    {
      "epoch": 0.3909131762612691,
      "lm_loss": 0.8123,
      "step": 5328,
      "vm_loss": 0.1771
    },
    {
      "epoch": 0.3909131762612691,
      "lm_loss": 0.0298,
      "step": 5328,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.3909131762612691,
      "lm_loss": 0.5183,
      "step": 5328,
      "vm_loss": 0.1689
    },
    {
      "epoch": 0.39098654585140824,
      "grad_norm": 1.451105021964182,
      "learning_rate": 6.954506296539112e-06,
      "loss": 0.724,
      "step": 5329
    },
    {
      "epoch": 0.3910599154415474,
      "grad_norm": 1.3922082032699314,
      "learning_rate": 6.9534125861599595e-06,
      "loss": 0.7822,
      "step": 5330
    },
    {
      "epoch": 0.3911332850316865,
      "grad_norm": 1.4201301052998314,
      "learning_rate": 6.952318765466762e-06,
      "loss": 0.7828,
      "step": 5331
    },
    {
      "epoch": 0.39120665462182563,
      "grad_norm": 1.556929337716057,
      "learning_rate": 6.951224834521291e-06,
      "loss": 0.8352,
      "step": 5332
    },
    {
      "epoch": 0.3912800242119647,
      "grad_norm": 1.53604888652649,
      "learning_rate": 6.950130793385318e-06,
      "loss": 0.8599,
      "step": 5333
    },
    {
      "epoch": 0.3913533938021039,
      "grad_norm": 1.5340463643494757,
      "learning_rate": 6.949036642120632e-06,
      "loss": 0.8656,
      "step": 5334
    },
    {
      "epoch": 0.391426763392243,
      "grad_norm": 1.4458599399102514,
      "learning_rate": 6.94794238078902e-06,
      "loss": 0.7459,
      "step": 5335
    },
    {
      "epoch": 0.3915001329823821,
      "grad_norm": 1.5549167072292434,
      "learning_rate": 6.946848009452276e-06,
      "loss": 0.8845,
      "step": 5336
    },
    {
      "epoch": 0.3915001329823821,
      "lm_loss": 0.3077,
      "step": 5336,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.3915001329823821,
      "lm_loss": 0.2347,
      "step": 5336,
      "vm_loss": 0.1169
    },
    {
      "epoch": 0.3915001329823821,
      "lm_loss": 0.0857,
      "step": 5336,
      "vm_loss": 0.0371
    },
    {
      "epoch": 0.3915001329823821,
      "lm_loss": 0.1839,
      "step": 5336,
      "vm_loss": 0.1638
    },
    {
      "epoch": 0.3915001329823821,
      "lm_loss": 0.3407,
      "step": 5336,
      "vm_loss": 0.1868
    },
    {
      "epoch": 0.3915001329823821,
      "lm_loss": 0.1707,
      "step": 5336,
      "vm_loss": 0.2375
    },
    {
      "epoch": 0.3915001329823821,
      "lm_loss": 0.2784,
      "step": 5336,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.3915001329823821,
      "lm_loss": 0.0508,
      "step": 5336,
      "vm_loss": 0.175
    },
    {
      "epoch": 0.39157350257252127,
      "grad_norm": 1.3113978348620108,
      "learning_rate": 6.945753528172206e-06,
      "loss": 0.6547,
      "step": 5337
    },
    {
      "epoch": 0.39164687216266036,
      "grad_norm": 1.4737705079508854,
      "learning_rate": 6.944658937010614e-06,
      "loss": 0.819,
      "step": 5338
    },
    {
      "epoch": 0.3917202417527995,
      "grad_norm": 1.5013196971536997,
      "learning_rate": 6.943564236029317e-06,
      "loss": 0.8206,
      "step": 5339
    },
    {
      "epoch": 0.3917936113429386,
      "grad_norm": 0.6894772970987071,
      "learning_rate": 6.942469425290134e-06,
      "loss": 0.5186,
      "step": 5340
    },
    {
      "epoch": 0.39186698093307776,
      "grad_norm": 1.424049470416708,
      "learning_rate": 6.941374504854891e-06,
      "loss": 0.791,
      "step": 5341
    },
    {
      "epoch": 0.3919403505232169,
      "grad_norm": 1.5325001370948232,
      "learning_rate": 6.940279474785423e-06,
      "loss": 0.8469,
      "step": 5342
    },
    {
      "epoch": 0.392013720113356,
      "grad_norm": 1.402298701714157,
      "learning_rate": 6.939184335143569e-06,
      "loss": 0.7303,
      "step": 5343
    },
    {
      "epoch": 0.39208708970349515,
      "grad_norm": 1.3510717980608775,
      "learning_rate": 6.938089085991171e-06,
      "loss": 0.7369,
      "step": 5344
    },
    {
      "epoch": 0.39208708970349515,
      "lm_loss": 1.4929,
      "step": 5344,
      "vm_loss": 0.0483
    },
    {
      "epoch": 0.39208708970349515,
      "lm_loss": 0.6605,
      "step": 5344,
      "vm_loss": 0.0813
    },
    {
      "epoch": 0.39208708970349515,
      "lm_loss": 0.7134,
      "step": 5344,
      "vm_loss": 0.1315
    },
    {
      "epoch": 0.39208708970349515,
      "lm_loss": 0.5488,
      "step": 5344,
      "vm_loss": 0.1284
    },
    {
      "epoch": 0.39208708970349515,
      "lm_loss": 0.8197,
      "step": 5344,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.39208708970349515,
      "lm_loss": 1.0189,
      "step": 5344,
      "vm_loss": 0.1168
    },
    {
      "epoch": 0.39208708970349515,
      "lm_loss": 0.4129,
      "step": 5344,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.39208708970349515,
      "lm_loss": 0.0254,
      "step": 5344,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.39216045929363424,
      "grad_norm": 1.4035834921679216,
      "learning_rate": 6.9369937273900845e-06,
      "loss": 0.7713,
      "step": 5345
    },
    {
      "epoch": 0.3922338288837734,
      "grad_norm": 0.5791490632873806,
      "learning_rate": 6.935898259402165e-06,
      "loss": 0.5194,
      "step": 5346
    },
    {
      "epoch": 0.39230719847391254,
      "grad_norm": 1.4419494684741907,
      "learning_rate": 6.934802682089277e-06,
      "loss": 0.7752,
      "step": 5347
    },
    {
      "epoch": 0.39238056806405164,
      "grad_norm": 0.5757030469944691,
      "learning_rate": 6.9337069955132895e-06,
      "loss": 0.5222,
      "step": 5348
    },
    {
      "epoch": 0.3924539376541908,
      "grad_norm": 0.5653251757902963,
      "learning_rate": 6.9326111997360775e-06,
      "loss": 0.5,
      "step": 5349
    },
    {
      "epoch": 0.3925273072443299,
      "grad_norm": 1.4923851944975577,
      "learning_rate": 6.931515294819528e-06,
      "loss": 0.8036,
      "step": 5350
    },
    {
      "epoch": 0.39260067683446903,
      "grad_norm": 1.4630372572802866,
      "learning_rate": 6.930419280825525e-06,
      "loss": 0.7965,
      "step": 5351
    },
    {
      "epoch": 0.3926740464246082,
      "grad_norm": 1.57535051250449,
      "learning_rate": 6.9293231578159645e-06,
      "loss": 0.7402,
      "step": 5352
    },
    {
      "epoch": 0.3926740464246082,
      "lm_loss": 1.2295,
      "step": 5352,
      "vm_loss": 0.0891
    },
    {
      "epoch": 0.3926740464246082,
      "lm_loss": 0.1412,
      "step": 5352,
      "vm_loss": 0.0514
    },
    {
      "epoch": 0.3926740464246082,
      "lm_loss": 0.9014,
      "step": 5352,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.3926740464246082,
      "lm_loss": 0.3781,
      "step": 5352,
      "vm_loss": 0.0566
    },
    {
      "epoch": 0.3926740464246082,
      "lm_loss": 0.0379,
      "step": 5352,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.3926740464246082,
      "lm_loss": 0.1583,
      "step": 5352,
      "vm_loss": 0.0683
    },
    {
      "epoch": 0.3926740464246082,
      "lm_loss": 0.7928,
      "step": 5352,
      "vm_loss": 0.124
    },
    {
      "epoch": 0.3926740464246082,
      "lm_loss": 0.5698,
      "step": 5352,
      "vm_loss": 0.1066
    },
    {
      "epoch": 0.3927474160147473,
      "grad_norm": 1.6061041536100895,
      "learning_rate": 6.928226925852748e-06,
      "loss": 0.8112,
      "step": 5353
    },
    {
      "epoch": 0.3928207856048864,
      "grad_norm": 1.5928065402136107,
      "learning_rate": 6.927130584997782e-06,
      "loss": 0.8954,
      "step": 5354
    },
    {
      "epoch": 0.3928941551950255,
      "grad_norm": 1.4785358758737805,
      "learning_rate": 6.926034135312979e-06,
      "loss": 0.6889,
      "step": 5355
    },
    {
      "epoch": 0.39296752478516467,
      "grad_norm": 1.739829679595001,
      "learning_rate": 6.924937576860259e-06,
      "loss": 0.9564,
      "step": 5356
    },
    {
      "epoch": 0.3930408943753038,
      "grad_norm": 0.7139036317955306,
      "learning_rate": 6.923840909701547e-06,
      "loss": 0.5156,
      "step": 5357
    },
    {
      "epoch": 0.3931142639654429,
      "grad_norm": 1.5380434101318465,
      "learning_rate": 6.922744133898774e-06,
      "loss": 0.8901,
      "step": 5358
    },
    {
      "epoch": 0.39318763355558206,
      "grad_norm": 1.3604738961323029,
      "learning_rate": 6.9216472495138785e-06,
      "loss": 0.812,
      "step": 5359
    },
    {
      "epoch": 0.39326100314572116,
      "grad_norm": 1.4545387609126756,
      "learning_rate": 6.920550256608803e-06,
      "loss": 0.8839,
      "step": 5360
    },
    {
      "epoch": 0.39326100314572116,
      "lm_loss": 0.2237,
      "step": 5360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39326100314572116,
      "lm_loss": 0.4487,
      "step": 5360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39326100314572116,
      "lm_loss": 0.4361,
      "step": 5360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39326100314572116,
      "lm_loss": 0.4427,
      "step": 5360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39326100314572116,
      "lm_loss": 0.2796,
      "step": 5360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39326100314572116,
      "lm_loss": 0.6002,
      "step": 5360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39326100314572116,
      "lm_loss": 0.6704,
      "step": 5360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39326100314572116,
      "lm_loss": 0.509,
      "step": 5360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3933343727358603,
      "grad_norm": 0.5737364839083339,
      "learning_rate": 6.919453155245499e-06,
      "loss": 0.4947,
      "step": 5361
    },
    {
      "epoch": 0.39340774232599945,
      "grad_norm": 0.5823400836069068,
      "learning_rate": 6.918355945485923e-06,
      "loss": 0.5312,
      "step": 5362
    },
    {
      "epoch": 0.39348111191613855,
      "grad_norm": 1.4227363634570351,
      "learning_rate": 6.917258627392034e-06,
      "loss": 0.7125,
      "step": 5363
    },
    {
      "epoch": 0.3935544815062777,
      "grad_norm": 0.5843511608962537,
      "learning_rate": 6.916161201025804e-06,
      "loss": 0.4986,
      "step": 5364
    },
    {
      "epoch": 0.3936278510964168,
      "grad_norm": 0.5764243287485845,
      "learning_rate": 6.9150636664492055e-06,
      "loss": 0.5151,
      "step": 5365
    },
    {
      "epoch": 0.39370122068655594,
      "grad_norm": 1.4392403287660571,
      "learning_rate": 6.913966023724218e-06,
      "loss": 0.7998,
      "step": 5366
    },
    {
      "epoch": 0.3937745902766951,
      "grad_norm": 1.2964428516087743,
      "learning_rate": 6.91286827291283e-06,
      "loss": 0.7173,
      "step": 5367
    },
    {
      "epoch": 0.3938479598668342,
      "grad_norm": 1.4090505869156484,
      "learning_rate": 6.911770414077033e-06,
      "loss": 0.7884,
      "step": 5368
    },
    {
      "epoch": 0.3938479598668342,
      "lm_loss": 1.6908,
      "step": 5368,
      "vm_loss": 0.1148
    },
    {
      "epoch": 0.3938479598668342,
      "lm_loss": 0.6916,
      "step": 5368,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.3938479598668342,
      "lm_loss": 0.0834,
      "step": 5368,
      "vm_loss": 0.0854
    },
    {
      "epoch": 0.3938479598668342,
      "lm_loss": 0.7611,
      "step": 5368,
      "vm_loss": 0.1254
    },
    {
      "epoch": 0.3938479598668342,
      "lm_loss": 0.9664,
      "step": 5368,
      "vm_loss": 0.1831
    },
    {
      "epoch": 0.3938479598668342,
      "lm_loss": 0.5987,
      "step": 5368,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.3938479598668342,
      "lm_loss": 1.0008,
      "step": 5368,
      "vm_loss": 0.2026
    },
    {
      "epoch": 0.3938479598668342,
      "lm_loss": 0.2425,
      "step": 5368,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.39392132945697333,
      "grad_norm": 1.6249753984232749,
      "learning_rate": 6.910672447278827e-06,
      "loss": 0.794,
      "step": 5369
    },
    {
      "epoch": 0.39399469904711243,
      "grad_norm": 0.6050919339982531,
      "learning_rate": 6.909574372580216e-06,
      "loss": 0.4983,
      "step": 5370
    },
    {
      "epoch": 0.3940680686372516,
      "grad_norm": 0.5658762263217012,
      "learning_rate": 6.908476190043211e-06,
      "loss": 0.48,
      "step": 5371
    },
    {
      "epoch": 0.39414143822739073,
      "grad_norm": 1.5447851123187706,
      "learning_rate": 6.90737789972983e-06,
      "loss": 0.7842,
      "step": 5372
    },
    {
      "epoch": 0.3942148078175298,
      "grad_norm": 1.4123872684988532,
      "learning_rate": 6.906279501702096e-06,
      "loss": 0.8752,
      "step": 5373
    },
    {
      "epoch": 0.39428817740766897,
      "grad_norm": 1.4269653151201667,
      "learning_rate": 6.905180996022035e-06,
      "loss": 0.8112,
      "step": 5374
    },
    {
      "epoch": 0.39436154699780807,
      "grad_norm": 1.49239755261545,
      "learning_rate": 6.904082382751687e-06,
      "loss": 0.7876,
      "step": 5375
    },
    {
      "epoch": 0.3944349165879472,
      "grad_norm": 0.5801306160206291,
      "learning_rate": 6.902983661953091e-06,
      "loss": 0.5081,
      "step": 5376
    },
    {
      "epoch": 0.3944349165879472,
      "lm_loss": 0.6531,
      "step": 5376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3944349165879472,
      "lm_loss": 0.3657,
      "step": 5376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3944349165879472,
      "lm_loss": 0.6923,
      "step": 5376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3944349165879472,
      "lm_loss": 0.6338,
      "step": 5376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3944349165879472,
      "lm_loss": 0.4686,
      "step": 5376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3944349165879472,
      "lm_loss": 0.5605,
      "step": 5376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3944349165879472,
      "lm_loss": 0.4033,
      "step": 5376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3944349165879472,
      "lm_loss": 0.4636,
      "step": 5376,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39450828617808636,
      "grad_norm": 0.5641861024461722,
      "learning_rate": 6.901884833688295e-06,
      "loss": 0.5194,
      "step": 5377
    },
    {
      "epoch": 0.39458165576822546,
      "grad_norm": 1.3887383124425388,
      "learning_rate": 6.900785898019353e-06,
      "loss": 0.8185,
      "step": 5378
    },
    {
      "epoch": 0.3946550253583646,
      "grad_norm": 0.5660386871495746,
      "learning_rate": 6.8996868550083226e-06,
      "loss": 0.5176,
      "step": 5379
    },
    {
      "epoch": 0.3947283949485037,
      "grad_norm": 1.5732696412890554,
      "learning_rate": 6.898587704717273e-06,
      "loss": 0.738,
      "step": 5380
    },
    {
      "epoch": 0.39480176453864285,
      "grad_norm": 1.3460054474272831,
      "learning_rate": 6.897488447208274e-06,
      "loss": 0.7229,
      "step": 5381
    },
    {
      "epoch": 0.39487513412878195,
      "grad_norm": 1.3996950853425802,
      "learning_rate": 6.896389082543401e-06,
      "loss": 0.803,
      "step": 5382
    },
    {
      "epoch": 0.3949485037189211,
      "grad_norm": 1.4036431199075332,
      "learning_rate": 6.895289610784742e-06,
      "loss": 0.7406,
      "step": 5383
    },
    {
      "epoch": 0.39502187330906025,
      "grad_norm": 1.4819725629571925,
      "learning_rate": 6.894190031994385e-06,
      "loss": 0.8774,
      "step": 5384
    },
    {
      "epoch": 0.39502187330906025,
      "lm_loss": 0.6688,
      "step": 5384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39502187330906025,
      "lm_loss": 0.3777,
      "step": 5384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39502187330906025,
      "lm_loss": 0.7526,
      "step": 5384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39502187330906025,
      "lm_loss": 0.4113,
      "step": 5384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39502187330906025,
      "lm_loss": 0.5451,
      "step": 5384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39502187330906025,
      "lm_loss": 0.442,
      "step": 5384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39502187330906025,
      "lm_loss": 0.5105,
      "step": 5384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39502187330906025,
      "lm_loss": 0.437,
      "step": 5384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39509524289919934,
      "grad_norm": 0.6247023689073752,
      "learning_rate": 6.893090346234425e-06,
      "loss": 0.5236,
      "step": 5385
    },
    {
      "epoch": 0.3951686124893385,
      "grad_norm": 0.5714805524136006,
      "learning_rate": 6.891990553566966e-06,
      "loss": 0.5233,
      "step": 5386
    },
    {
      "epoch": 0.3952419820794776,
      "grad_norm": 1.4573927439145071,
      "learning_rate": 6.890890654054114e-06,
      "loss": 0.8029,
      "step": 5387
    },
    {
      "epoch": 0.39531535166961673,
      "grad_norm": 1.5449268293472862,
      "learning_rate": 6.889790647757985e-06,
      "loss": 0.7505,
      "step": 5388
    },
    {
      "epoch": 0.3953887212597559,
      "grad_norm": 1.5736143750977907,
      "learning_rate": 6.8886905347406985e-06,
      "loss": 0.7937,
      "step": 5389
    },
    {
      "epoch": 0.395462090849895,
      "grad_norm": 1.3295662845068714,
      "learning_rate": 6.887590315064378e-06,
      "loss": 0.8394,
      "step": 5390
    },
    {
      "epoch": 0.3955354604400341,
      "grad_norm": 1.5186955786710723,
      "learning_rate": 6.886489988791161e-06,
      "loss": 0.8268,
      "step": 5391
    },
    {
      "epoch": 0.3956088300301732,
      "grad_norm": 1.4728927278840838,
      "learning_rate": 6.885389555983182e-06,
      "loss": 0.7813,
      "step": 5392
    },
    {
      "epoch": 0.3956088300301732,
      "lm_loss": 0.4021,
      "step": 5392,
      "vm_loss": 0.051
    },
    {
      "epoch": 0.3956088300301732,
      "lm_loss": 0.668,
      "step": 5392,
      "vm_loss": 0.0661
    },
    {
      "epoch": 0.3956088300301732,
      "lm_loss": 0.0209,
      "step": 5392,
      "vm_loss": 0.1617
    },
    {
      "epoch": 0.3956088300301732,
      "lm_loss": 0.6003,
      "step": 5392,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.3956088300301732,
      "lm_loss": 0.8924,
      "step": 5392,
      "vm_loss": 0.1243
    },
    {
      "epoch": 0.3956088300301732,
      "lm_loss": 0.0546,
      "step": 5392,
      "vm_loss": 0.0468
    },
    {
      "epoch": 0.3956088300301732,
      "lm_loss": 0.774,
      "step": 5392,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.3956088300301732,
      "lm_loss": 0.8839,
      "step": 5392,
      "vm_loss": 0.1918
    },
    {
      "epoch": 0.39568219962031237,
      "grad_norm": 1.5518147153502273,
      "learning_rate": 6.884289016702584e-06,
      "loss": 0.8141,
      "step": 5393
    },
    {
      "epoch": 0.3957555692104515,
      "grad_norm": 0.6719089991443448,
      "learning_rate": 6.883188371011519e-06,
      "loss": 0.4942,
      "step": 5394
    },
    {
      "epoch": 0.3958289388005906,
      "grad_norm": 0.5996128296011909,
      "learning_rate": 6.8820876189721434e-06,
      "loss": 0.4936,
      "step": 5395
    },
    {
      "epoch": 0.39590230839072976,
      "grad_norm": 1.7109214016009686,
      "learning_rate": 6.88098676064662e-06,
      "loss": 0.8381,
      "step": 5396
    },
    {
      "epoch": 0.39597567798086886,
      "grad_norm": 1.4338060092550438,
      "learning_rate": 6.8798857960971146e-06,
      "loss": 0.733,
      "step": 5397
    },
    {
      "epoch": 0.396049047571008,
      "grad_norm": 1.467552765950721,
      "learning_rate": 6.878784725385803e-06,
      "loss": 0.881,
      "step": 5398
    },
    {
      "epoch": 0.39612241716114716,
      "grad_norm": 1.6989169476372872,
      "learning_rate": 6.877683548574866e-06,
      "loss": 0.7789,
      "step": 5399
    },
    {
      "epoch": 0.39619578675128625,
      "grad_norm": 1.5471015049674548,
      "learning_rate": 6.876582265726489e-06,
      "loss": 0.7915,
      "step": 5400
    },
    {
      "epoch": 0.39619578675128625,
      "lm_loss": 1.1237,
      "step": 5400,
      "vm_loss": 0.0464
    },
    {
      "epoch": 0.39619578675128625,
      "lm_loss": 1.3823,
      "step": 5400,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.39619578675128625,
      "lm_loss": 1.0219,
      "step": 5400,
      "vm_loss": 0.2122
    },
    {
      "epoch": 0.39619578675128625,
      "lm_loss": 0.8896,
      "step": 5400,
      "vm_loss": 0.1198
    },
    {
      "epoch": 0.39619578675128625,
      "lm_loss": 0.5145,
      "step": 5400,
      "vm_loss": 0.0527
    },
    {
      "epoch": 0.39619578675128625,
      "lm_loss": 0.9367,
      "step": 5400,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.39619578675128625,
      "lm_loss": 0.7863,
      "step": 5400,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.39619578675128625,
      "lm_loss": 0.6561,
      "step": 5400,
      "vm_loss": 0.1682
    },
    {
      "epoch": 0.3962691563414254,
      "grad_norm": 1.2735172248022537,
      "learning_rate": 6.875480876902864e-06,
      "loss": 0.762,
      "step": 5401
    },
    {
      "epoch": 0.3963425259315645,
      "grad_norm": 0.8919449965970631,
      "learning_rate": 6.87437938216619e-06,
      "loss": 0.5009,
      "step": 5402
    },
    {
      "epoch": 0.39641589552170364,
      "grad_norm": 1.5927899061397794,
      "learning_rate": 6.873277781578669e-06,
      "loss": 0.8069,
      "step": 5403
    },
    {
      "epoch": 0.3964892651118428,
      "grad_norm": 1.5637170756763585,
      "learning_rate": 6.8721760752025135e-06,
      "loss": 0.7477,
      "step": 5404
    },
    {
      "epoch": 0.3965626347019819,
      "grad_norm": 1.6843745342465848,
      "learning_rate": 6.871074263099939e-06,
      "loss": 0.8304,
      "step": 5405
    },
    {
      "epoch": 0.39663600429212104,
      "grad_norm": 1.5564655462421002,
      "learning_rate": 6.869972345333167e-06,
      "loss": 0.7706,
      "step": 5406
    },
    {
      "epoch": 0.39670937388226013,
      "grad_norm": 1.684321738348245,
      "learning_rate": 6.868870321964425e-06,
      "loss": 0.7972,
      "step": 5407
    },
    {
      "epoch": 0.3967827434723993,
      "grad_norm": 0.6229191102650272,
      "learning_rate": 6.867768193055948e-06,
      "loss": 0.5234,
      "step": 5408
    },
    {
      "epoch": 0.3967827434723993,
      "lm_loss": 0.935,
      "step": 5408,
      "vm_loss": 0.0297
    },
    {
      "epoch": 0.3967827434723993,
      "lm_loss": 0.9645,
      "step": 5408,
      "vm_loss": 0.1144
    },
    {
      "epoch": 0.3967827434723993,
      "lm_loss": 0.8932,
      "step": 5408,
      "vm_loss": 0.1422
    },
    {
      "epoch": 0.3967827434723993,
      "lm_loss": 0.8949,
      "step": 5408,
      "vm_loss": 0.1454
    },
    {
      "epoch": 0.3967827434723993,
      "lm_loss": 0.9963,
      "step": 5408,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.3967827434723993,
      "lm_loss": 0.5521,
      "step": 5408,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.3967827434723993,
      "lm_loss": 0.5463,
      "step": 5408,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.3967827434723993,
      "lm_loss": 0.5048,
      "step": 5408,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.39685611306253843,
      "grad_norm": 1.4076528478938628,
      "learning_rate": 6.866665958669976e-06,
      "loss": 0.7738,
      "step": 5409
    },
    {
      "epoch": 0.3969294826526775,
      "grad_norm": 1.4211220222349523,
      "learning_rate": 6.865563618868755e-06,
      "loss": 0.7728,
      "step": 5410
    },
    {
      "epoch": 0.3970028522428167,
      "grad_norm": 1.3691469368294016,
      "learning_rate": 6.864461173714535e-06,
      "loss": 0.7681,
      "step": 5411
    },
    {
      "epoch": 0.39707622183295577,
      "grad_norm": 0.646641995227023,
      "learning_rate": 6.863358623269577e-06,
      "loss": 0.5304,
      "step": 5412
    },
    {
      "epoch": 0.3971495914230949,
      "grad_norm": 1.6500262830455596,
      "learning_rate": 6.862255967596142e-06,
      "loss": 0.8068,
      "step": 5413
    },
    {
      "epoch": 0.39722296101323407,
      "grad_norm": 0.5790560292970869,
      "learning_rate": 6.8611532067565016e-06,
      "loss": 0.5138,
      "step": 5414
    },
    {
      "epoch": 0.39729633060337316,
      "grad_norm": 1.561956454360258,
      "learning_rate": 6.860050340812929e-06,
      "loss": 0.8214,
      "step": 5415
    },
    {
      "epoch": 0.3973697001935123,
      "grad_norm": 1.4618083280904142,
      "learning_rate": 6.858947369827708e-06,
      "loss": 0.9471,
      "step": 5416
    },
    {
      "epoch": 0.3973697001935123,
      "lm_loss": 0.0989,
      "step": 5416,
      "vm_loss": 0.0411
    },
    {
      "epoch": 0.3973697001935123,
      "lm_loss": 1.3103,
      "step": 5416,
      "vm_loss": 0.189
    },
    {
      "epoch": 0.3973697001935123,
      "lm_loss": 0.2709,
      "step": 5416,
      "vm_loss": 0.1949
    },
    {
      "epoch": 0.3973697001935123,
      "lm_loss": 0.7747,
      "step": 5416,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.3973697001935123,
      "lm_loss": 0.8203,
      "step": 5416,
      "vm_loss": 0.1405
    },
    {
      "epoch": 0.3973697001935123,
      "lm_loss": 0.2242,
      "step": 5416,
      "vm_loss": 0.0732
    },
    {
      "epoch": 0.3973697001935123,
      "lm_loss": 0.44,
      "step": 5416,
      "vm_loss": 0.148
    },
    {
      "epoch": 0.3973697001935123,
      "lm_loss": 0.8727,
      "step": 5416,
      "vm_loss": 0.131
    },
    {
      "epoch": 0.3974430697836514,
      "grad_norm": 1.4392558563219549,
      "learning_rate": 6.857844293863125e-06,
      "loss": 0.8711,
      "step": 5417
    },
    {
      "epoch": 0.39751643937379055,
      "grad_norm": 1.4030163324102143,
      "learning_rate": 6.856741112981476e-06,
      "loss": 0.7666,
      "step": 5418
    },
    {
      "epoch": 0.3975898089639297,
      "grad_norm": 1.4681794613843393,
      "learning_rate": 6.855637827245055e-06,
      "loss": 0.8782,
      "step": 5419
    },
    {
      "epoch": 0.3976631785540688,
      "grad_norm": 1.4575775778829096,
      "learning_rate": 6.8545344367161725e-06,
      "loss": 0.8008,
      "step": 5420
    },
    {
      "epoch": 0.39773654814420795,
      "grad_norm": 1.5167086947979385,
      "learning_rate": 6.8534309414571375e-06,
      "loss": 0.8033,
      "step": 5421
    },
    {
      "epoch": 0.39780991773434704,
      "grad_norm": 1.5408031944944833,
      "learning_rate": 6.852327341530267e-06,
      "loss": 0.811,
      "step": 5422
    },
    {
      "epoch": 0.3978832873244862,
      "grad_norm": 1.4579336237262497,
      "learning_rate": 6.851223636997885e-06,
      "loss": 0.7934,
      "step": 5423
    },
    {
      "epoch": 0.3979566569146253,
      "grad_norm": 1.6045731343127791,
      "learning_rate": 6.850119827922318e-06,
      "loss": 0.8279,
      "step": 5424
    },
    {
      "epoch": 0.3979566569146253,
      "lm_loss": 1.4637,
      "step": 5424,
      "vm_loss": 0.096
    },
    {
      "epoch": 0.3979566569146253,
      "lm_loss": 0.8978,
      "step": 5424,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.3979566569146253,
      "lm_loss": 0.5362,
      "step": 5424,
      "vm_loss": 0.198
    },
    {
      "epoch": 0.3979566569146253,
      "lm_loss": 0.1144,
      "step": 5424,
      "vm_loss": 0.1796
    },
    {
      "epoch": 0.3979566569146253,
      "lm_loss": 0.1373,
      "step": 5424,
      "vm_loss": 0.161
    },
    {
      "epoch": 0.3979566569146253,
      "lm_loss": 0.4303,
      "step": 5424,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.3979566569146253,
      "lm_loss": 0.759,
      "step": 5424,
      "vm_loss": 0.1228
    },
    {
      "epoch": 0.3979566569146253,
      "lm_loss": 0.8842,
      "step": 5424,
      "vm_loss": 0.0526
    },
    {
      "epoch": 0.39803002650476443,
      "grad_norm": 1.4941173740029225,
      "learning_rate": 6.849015914365904e-06,
      "loss": 0.7221,
      "step": 5425
    },
    {
      "epoch": 0.3981033960949036,
      "grad_norm": 0.7023203161154481,
      "learning_rate": 6.847911896390983e-06,
      "loss": 0.4999,
      "step": 5426
    },
    {
      "epoch": 0.3981767656850427,
      "grad_norm": 1.600338700786137,
      "learning_rate": 6.8468077740598995e-06,
      "loss": 0.8502,
      "step": 5427
    },
    {
      "epoch": 0.3982501352751818,
      "grad_norm": 1.4248732216634712,
      "learning_rate": 6.845703547435007e-06,
      "loss": 0.7161,
      "step": 5428
    },
    {
      "epoch": 0.3983235048653209,
      "grad_norm": 1.679019198719237,
      "learning_rate": 6.844599216578667e-06,
      "loss": 0.7388,
      "step": 5429
    },
    {
      "epoch": 0.39839687445546007,
      "grad_norm": 0.6039606607171693,
      "learning_rate": 6.843494781553238e-06,
      "loss": 0.5207,
      "step": 5430
    },
    {
      "epoch": 0.3984702440455992,
      "grad_norm": 0.6018865289861133,
      "learning_rate": 6.8423902424210955e-06,
      "loss": 0.51,
      "step": 5431
    },
    {
      "epoch": 0.3985436136357383,
      "grad_norm": 1.4951348677500271,
      "learning_rate": 6.841285599244614e-06,
      "loss": 0.7838,
      "step": 5432
    },
    {
      "epoch": 0.3985436136357383,
      "lm_loss": 1.7524,
      "step": 5432,
      "vm_loss": 0.0522
    },
    {
      "epoch": 0.3985436136357383,
      "lm_loss": 1.0172,
      "step": 5432,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.3985436136357383,
      "lm_loss": 0.5295,
      "step": 5432,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.3985436136357383,
      "lm_loss": 0.2999,
      "step": 5432,
      "vm_loss": 0.0988
    },
    {
      "epoch": 0.3985436136357383,
      "lm_loss": 0.7261,
      "step": 5432,
      "vm_loss": 0.156
    },
    {
      "epoch": 0.3985436136357383,
      "lm_loss": 0.4459,
      "step": 5432,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.3985436136357383,
      "lm_loss": 0.2717,
      "step": 5432,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.3985436136357383,
      "lm_loss": 0.7134,
      "step": 5432,
      "vm_loss": 0.2047
    },
    {
      "epoch": 0.39861698322587746,
      "grad_norm": 1.586836295931421,
      "learning_rate": 6.840180852086173e-06,
      "loss": 0.8381,
      "step": 5433
    },
    {
      "epoch": 0.39869035281601656,
      "grad_norm": 0.6266652465860985,
      "learning_rate": 6.839076001008163e-06,
      "loss": 0.525,
      "step": 5434
    },
    {
      "epoch": 0.3987637224061557,
      "grad_norm": 1.4940947594927774,
      "learning_rate": 6.837971046072975e-06,
      "loss": 0.7598,
      "step": 5435
    },
    {
      "epoch": 0.39883709199629486,
      "grad_norm": 1.628337240561198,
      "learning_rate": 6.836865987343012e-06,
      "loss": 0.8411,
      "step": 5436
    },
    {
      "epoch": 0.39891046158643395,
      "grad_norm": 1.5605374040434934,
      "learning_rate": 6.835760824880678e-06,
      "loss": 0.803,
      "step": 5437
    },
    {
      "epoch": 0.3989838311765731,
      "grad_norm": 0.5972406465430962,
      "learning_rate": 6.834655558748383e-06,
      "loss": 0.5156,
      "step": 5438
    },
    {
      "epoch": 0.3990572007667122,
      "grad_norm": 1.6403380537678853,
      "learning_rate": 6.833550189008546e-06,
      "loss": 0.839,
      "step": 5439
    },
    {
      "epoch": 0.39913057035685134,
      "grad_norm": 0.5784773224259162,
      "learning_rate": 6.8324447157235894e-06,
      "loss": 0.5193,
      "step": 5440
    },
    {
      "epoch": 0.39913057035685134,
      "lm_loss": 1.2363,
      "step": 5440,
      "vm_loss": 0.0864
    },
    {
      "epoch": 0.39913057035685134,
      "lm_loss": 0.0895,
      "step": 5440,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.39913057035685134,
      "lm_loss": 0.7842,
      "step": 5440,
      "vm_loss": 0.1488
    },
    {
      "epoch": 0.39913057035685134,
      "lm_loss": 0.7849,
      "step": 5440,
      "vm_loss": 0.18
    },
    {
      "epoch": 0.39913057035685134,
      "lm_loss": 1.1,
      "step": 5440,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.39913057035685134,
      "lm_loss": 0.9637,
      "step": 5440,
      "vm_loss": 0.1388
    },
    {
      "epoch": 0.39913057035685134,
      "lm_loss": 0.8013,
      "step": 5440,
      "vm_loss": 0.0688
    },
    {
      "epoch": 0.39913057035685134,
      "lm_loss": 0.0889,
      "step": 5440,
      "vm_loss": 0.0841
    },
    {
      "epoch": 0.3992039399469905,
      "grad_norm": 1.469769337432712,
      "learning_rate": 6.8313391389559415e-06,
      "loss": 0.714,
      "step": 5441
    },
    {
      "epoch": 0.3992773095371296,
      "grad_norm": 1.468408619442701,
      "learning_rate": 6.830233458768037e-06,
      "loss": 0.779,
      "step": 5442
    },
    {
      "epoch": 0.39935067912726874,
      "grad_norm": 0.6021910500816224,
      "learning_rate": 6.829127675222316e-06,
      "loss": 0.5164,
      "step": 5443
    },
    {
      "epoch": 0.39942404871740783,
      "grad_norm": 1.5877951266684531,
      "learning_rate": 6.828021788381227e-06,
      "loss": 0.8667,
      "step": 5444
    },
    {
      "epoch": 0.399497418307547,
      "grad_norm": 1.5882606078909516,
      "learning_rate": 6.8269157983072194e-06,
      "loss": 0.8141,
      "step": 5445
    },
    {
      "epoch": 0.39957078789768613,
      "grad_norm": 1.6874132320401474,
      "learning_rate": 6.8258097050627515e-06,
      "loss": 0.746,
      "step": 5446
    },
    {
      "epoch": 0.3996441574878252,
      "grad_norm": 1.4880167381845584,
      "learning_rate": 6.82470350871029e-06,
      "loss": 0.8563,
      "step": 5447
    },
    {
      "epoch": 0.3997175270779644,
      "grad_norm": 0.5936173779138479,
      "learning_rate": 6.8235972093123024e-06,
      "loss": 0.5175,
      "step": 5448
    },
    {
      "epoch": 0.3997175270779644,
      "lm_loss": 0.5121,
      "step": 5448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3997175270779644,
      "lm_loss": 0.443,
      "step": 5448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3997175270779644,
      "lm_loss": 0.4604,
      "step": 5448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3997175270779644,
      "lm_loss": 0.5072,
      "step": 5448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3997175270779644,
      "lm_loss": 0.593,
      "step": 5448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3997175270779644,
      "lm_loss": 0.4269,
      "step": 5448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3997175270779644,
      "lm_loss": 0.6384,
      "step": 5448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.3997175270779644,
      "lm_loss": 0.4328,
      "step": 5448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.39979089666810347,
      "grad_norm": 0.5874945832532064,
      "learning_rate": 6.822490806931262e-06,
      "loss": 0.511,
      "step": 5449
    },
    {
      "epoch": 0.3998642662582426,
      "grad_norm": 1.3574870105698593,
      "learning_rate": 6.821384301629654e-06,
      "loss": 0.7799,
      "step": 5450
    },
    {
      "epoch": 0.39993763584838177,
      "grad_norm": 0.562806477105758,
      "learning_rate": 6.820277693469964e-06,
      "loss": 0.4984,
      "step": 5451
    },
    {
      "epoch": 0.40001100543852086,
      "grad_norm": 0.5783166727761246,
      "learning_rate": 6.819170982514686e-06,
      "loss": 0.5092,
      "step": 5452
    },
    {
      "epoch": 0.40008437502866,
      "grad_norm": 1.5539930304116873,
      "learning_rate": 6.818064168826316e-06,
      "loss": 0.791,
      "step": 5453
    },
    {
      "epoch": 0.4001577446187991,
      "grad_norm": 1.6577970892617353,
      "learning_rate": 6.81695725246736e-06,
      "loss": 0.7426,
      "step": 5454
    },
    {
      "epoch": 0.40023111420893825,
      "grad_norm": 1.4112205782481742,
      "learning_rate": 6.815850233500328e-06,
      "loss": 0.6977,
      "step": 5455
    },
    {
      "epoch": 0.4003044837990774,
      "grad_norm": 1.585440155224818,
      "learning_rate": 6.814743111987738e-06,
      "loss": 0.8497,
      "step": 5456
    },
    {
      "epoch": 0.4003044837990774,
      "lm_loss": 0.5917,
      "step": 5456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4003044837990774,
      "lm_loss": 0.4443,
      "step": 5456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4003044837990774,
      "lm_loss": 0.5838,
      "step": 5456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4003044837990774,
      "lm_loss": 0.4705,
      "step": 5456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4003044837990774,
      "lm_loss": 0.5438,
      "step": 5456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4003044837990774,
      "lm_loss": 0.5593,
      "step": 5456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4003044837990774,
      "lm_loss": 0.5562,
      "step": 5456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4003044837990774,
      "lm_loss": 0.481,
      "step": 5456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4003778533892165,
      "grad_norm": 0.604170722273441,
      "learning_rate": 6.813635887992108e-06,
      "loss": 0.5092,
      "step": 5457
    },
    {
      "epoch": 0.40045122297935565,
      "grad_norm": 0.6096696152754278,
      "learning_rate": 6.812528561575971e-06,
      "loss": 0.5275,
      "step": 5458
    },
    {
      "epoch": 0.40052459256949474,
      "grad_norm": 0.5844102144885686,
      "learning_rate": 6.811421132801855e-06,
      "loss": 0.5152,
      "step": 5459
    },
    {
      "epoch": 0.4005979621596339,
      "grad_norm": 1.8866544563674321,
      "learning_rate": 6.810313601732303e-06,
      "loss": 0.7388,
      "step": 5460
    },
    {
      "epoch": 0.400671331749773,
      "grad_norm": 1.5487404263547921,
      "learning_rate": 6.809205968429859e-06,
      "loss": 0.894,
      "step": 5461
    },
    {
      "epoch": 0.40074470133991213,
      "grad_norm": 1.5534934277261072,
      "learning_rate": 6.8080982329570714e-06,
      "loss": 0.8207,
      "step": 5462
    },
    {
      "epoch": 0.4008180709300513,
      "grad_norm": 1.6704581306502364,
      "learning_rate": 6.8069903953765004e-06,
      "loss": 0.956,
      "step": 5463
    },
    {
      "epoch": 0.4008914405201904,
      "grad_norm": 1.6158827156697808,
      "learning_rate": 6.8058824557507074e-06,
      "loss": 0.8498,
      "step": 5464
    },
    {
      "epoch": 0.4008914405201904,
      "lm_loss": 0.6332,
      "step": 5464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4008914405201904,
      "lm_loss": 0.4616,
      "step": 5464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4008914405201904,
      "lm_loss": 0.5405,
      "step": 5464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4008914405201904,
      "lm_loss": 0.4548,
      "step": 5464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4008914405201904,
      "lm_loss": 0.2654,
      "step": 5464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4008914405201904,
      "lm_loss": 0.6889,
      "step": 5464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4008914405201904,
      "lm_loss": 0.7239,
      "step": 5464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4008914405201904,
      "lm_loss": 0.6241,
      "step": 5464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.40096481011032953,
      "grad_norm": 0.7418534262751878,
      "learning_rate": 6.8047744141422585e-06,
      "loss": 0.5277,
      "step": 5465
    },
    {
      "epoch": 0.4010381797004686,
      "grad_norm": 1.5678402477372093,
      "learning_rate": 6.80366627061373e-06,
      "loss": 0.7393,
      "step": 5466
    },
    {
      "epoch": 0.40111154929060777,
      "grad_norm": 0.6856149359762245,
      "learning_rate": 6.8025580252277015e-06,
      "loss": 0.4895,
      "step": 5467
    },
    {
      "epoch": 0.4011849188807469,
      "grad_norm": 1.5132870651249748,
      "learning_rate": 6.801449678046755e-06,
      "loss": 0.7963,
      "step": 5468
    },
    {
      "epoch": 0.401258288470886,
      "grad_norm": 1.6653218853122742,
      "learning_rate": 6.800341229133486e-06,
      "loss": 0.8338,
      "step": 5469
    },
    {
      "epoch": 0.40133165806102516,
      "grad_norm": 0.5652508638133975,
      "learning_rate": 6.799232678550487e-06,
      "loss": 0.5096,
      "step": 5470
    },
    {
      "epoch": 0.40140502765116426,
      "grad_norm": 0.5694162733890076,
      "learning_rate": 6.798124026360366e-06,
      "loss": 0.5115,
      "step": 5471
    },
    {
      "epoch": 0.4014783972413034,
      "grad_norm": 0.5877420085133847,
      "learning_rate": 6.797015272625727e-06,
      "loss": 0.5195,
      "step": 5472
    },
    {
      "epoch": 0.4014783972413034,
      "lm_loss": 0.7028,
      "step": 5472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4014783972413034,
      "lm_loss": 0.5129,
      "step": 5472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4014783972413034,
      "lm_loss": 0.5353,
      "step": 5472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4014783972413034,
      "lm_loss": 0.473,
      "step": 5472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4014783972413034,
      "lm_loss": 0.4812,
      "step": 5472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4014783972413034,
      "lm_loss": 0.3515,
      "step": 5472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4014783972413034,
      "lm_loss": 0.4568,
      "step": 5472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4014783972413034,
      "lm_loss": 0.7282,
      "step": 5472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.40155176683144256,
      "grad_norm": 0.5633000317361345,
      "learning_rate": 6.795906417409184e-06,
      "loss": 0.5128,
      "step": 5473
    },
    {
      "epoch": 0.40162513642158165,
      "grad_norm": 1.6291723053490297,
      "learning_rate": 6.794797460773359e-06,
      "loss": 0.7637,
      "step": 5474
    },
    {
      "epoch": 0.4016985060117208,
      "grad_norm": 1.596450290363637,
      "learning_rate": 6.793688402780877e-06,
      "loss": 0.728,
      "step": 5475
    },
    {
      "epoch": 0.4017718756018599,
      "grad_norm": 1.4676474641219388,
      "learning_rate": 6.792579243494368e-06,
      "loss": 0.8845,
      "step": 5476
    },
    {
      "epoch": 0.40184524519199905,
      "grad_norm": 1.531044281902299,
      "learning_rate": 6.7914699829764704e-06,
      "loss": 0.8768,
      "step": 5477
    },
    {
      "epoch": 0.4019186147821382,
      "grad_norm": 1.8883941410164005,
      "learning_rate": 6.790360621289826e-06,
      "loss": 0.8774,
      "step": 5478
    },
    {
      "epoch": 0.4019919843722773,
      "grad_norm": 1.5906220120912797,
      "learning_rate": 6.789251158497084e-06,
      "loss": 0.8256,
      "step": 5479
    },
    {
      "epoch": 0.40206535396241644,
      "grad_norm": 1.8577305272277962,
      "learning_rate": 6.788141594660899e-06,
      "loss": 0.8593,
      "step": 5480
    },
    {
      "epoch": 0.40206535396241644,
      "lm_loss": 0.6546,
      "step": 5480,
      "vm_loss": 0.0774
    },
    {
      "epoch": 0.40206535396241644,
      "lm_loss": 1.2849,
      "step": 5480,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.40206535396241644,
      "lm_loss": 1.006,
      "step": 5480,
      "vm_loss": 0.145
    },
    {
      "epoch": 0.40206535396241644,
      "lm_loss": 0.3169,
      "step": 5480,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.40206535396241644,
      "lm_loss": 2.1676,
      "step": 5480,
      "vm_loss": 0.1637
    },
    {
      "epoch": 0.40206535396241644,
      "lm_loss": 0.5922,
      "step": 5480,
      "vm_loss": 0.202
    },
    {
      "epoch": 0.40206535396241644,
      "lm_loss": 0.9546,
      "step": 5480,
      "vm_loss": 0.2127
    },
    {
      "epoch": 0.40206535396241644,
      "lm_loss": 0.1037,
      "step": 5480,
      "vm_loss": 0.1608
    },
    {
      "epoch": 0.40213872355255553,
      "grad_norm": 1.4073965672893087,
      "learning_rate": 6.787031929843929e-06,
      "loss": 0.8163,
      "step": 5481
    },
    {
      "epoch": 0.4022120931426947,
      "grad_norm": 1.3546890068634405,
      "learning_rate": 6.785922164108841e-06,
      "loss": 0.7573,
      "step": 5482
    },
    {
      "epoch": 0.40228546273283383,
      "grad_norm": 1.542287017065773,
      "learning_rate": 6.7848122975183035e-06,
      "loss": 0.705,
      "step": 5483
    },
    {
      "epoch": 0.4023588323229729,
      "grad_norm": 1.5464311789486742,
      "learning_rate": 6.783702330134998e-06,
      "loss": 0.8732,
      "step": 5484
    },
    {
      "epoch": 0.4024322019131121,
      "grad_norm": 1.7771746385230658,
      "learning_rate": 6.782592262021605e-06,
      "loss": 0.805,
      "step": 5485
    },
    {
      "epoch": 0.40250557150325117,
      "grad_norm": 1.4548142452013637,
      "learning_rate": 6.78148209324081e-06,
      "loss": 0.782,
      "step": 5486
    },
    {
      "epoch": 0.4025789410933903,
      "grad_norm": 1.6135724616153888,
      "learning_rate": 6.7803718238553105e-06,
      "loss": 0.7372,
      "step": 5487
    },
    {
      "epoch": 0.40265231068352947,
      "grad_norm": 1.6503011286198088,
      "learning_rate": 6.7792614539278055e-06,
      "loss": 0.8172,
      "step": 5488
    },
    {
      "epoch": 0.40265231068352947,
      "lm_loss": 0.7525,
      "step": 5488,
      "vm_loss": 0.0481
    },
    {
      "epoch": 0.40265231068352947,
      "lm_loss": 1.519,
      "step": 5488,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.40265231068352947,
      "lm_loss": 0.7894,
      "step": 5488,
      "vm_loss": 0.1541
    },
    {
      "epoch": 0.40265231068352947,
      "lm_loss": 0.6519,
      "step": 5488,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.40265231068352947,
      "lm_loss": 0.9748,
      "step": 5488,
      "vm_loss": 0.1365
    },
    {
      "epoch": 0.40265231068352947,
      "lm_loss": 0.0532,
      "step": 5488,
      "vm_loss": 0.2357
    },
    {
      "epoch": 0.40265231068352947,
      "lm_loss": 0.8116,
      "step": 5488,
      "vm_loss": 0.0703
    },
    {
      "epoch": 0.40265231068352947,
      "lm_loss": 0.2619,
      "step": 5488,
      "vm_loss": 0.1367
    },
    {
      "epoch": 0.40272568027366856,
      "grad_norm": 1.3702238381048266,
      "learning_rate": 6.778150983520999e-06,
      "loss": 0.85,
      "step": 5489
    },
    {
      "epoch": 0.4027990498638077,
      "grad_norm": 1.352343164655112,
      "learning_rate": 6.777040412697604e-06,
      "loss": 0.6969,
      "step": 5490
    },
    {
      "epoch": 0.4028724194539468,
      "grad_norm": 1.5386923783876623,
      "learning_rate": 6.775929741520335e-06,
      "loss": 0.6971,
      "step": 5491
    },
    {
      "epoch": 0.40294578904408596,
      "grad_norm": 1.6300959773929624,
      "learning_rate": 6.774818970051917e-06,
      "loss": 0.8139,
      "step": 5492
    },
    {
      "epoch": 0.4030191586342251,
      "grad_norm": 1.4710805318393,
      "learning_rate": 6.773708098355077e-06,
      "loss": 0.7788,
      "step": 5493
    },
    {
      "epoch": 0.4030925282243642,
      "grad_norm": 1.9187550323978577,
      "learning_rate": 6.772597126492545e-06,
      "loss": 0.887,
      "step": 5494
    },
    {
      "epoch": 0.40316589781450335,
      "grad_norm": 0.78828280604363,
      "learning_rate": 6.771486054527065e-06,
      "loss": 0.5104,
      "step": 5495
    },
    {
      "epoch": 0.40323926740464244,
      "grad_norm": 0.7116392858200205,
      "learning_rate": 6.770374882521382e-06,
      "loss": 0.5423,
      "step": 5496
    },
    {
      "epoch": 0.40323926740464244,
      "lm_loss": 0.9685,
      "step": 5496,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.40323926740464244,
      "lm_loss": 1.5266,
      "step": 5496,
      "vm_loss": 0.1074
    },
    {
      "epoch": 0.40323926740464244,
      "lm_loss": 0.0476,
      "step": 5496,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.40323926740464244,
      "lm_loss": 0.5918,
      "step": 5496,
      "vm_loss": 0.3064
    },
    {
      "epoch": 0.40323926740464244,
      "lm_loss": 1.031,
      "step": 5496,
      "vm_loss": 0.1188
    },
    {
      "epoch": 0.40323926740464244,
      "lm_loss": 1.0745,
      "step": 5496,
      "vm_loss": 0.1797
    },
    {
      "epoch": 0.40323926740464244,
      "lm_loss": 0.4236,
      "step": 5496,
      "vm_loss": 0.254
    },
    {
      "epoch": 0.40323926740464244,
      "lm_loss": 0.3944,
      "step": 5496,
      "vm_loss": 0.1586
    },
    {
      "epoch": 0.4033126369947816,
      "grad_norm": 1.9706046853756558,
      "learning_rate": 6.769263610538243e-06,
      "loss": 0.9234,
      "step": 5497
    },
    {
      "epoch": 0.40338600658492074,
      "grad_norm": 1.5364406099102532,
      "learning_rate": 6.768152238640409e-06,
      "loss": 0.8192,
      "step": 5498
    },
    {
      "epoch": 0.40345937617505984,
      "grad_norm": 1.5295933792417622,
      "learning_rate": 6.767040766890636e-06,
      "loss": 0.7738,
      "step": 5499
    },
    {
      "epoch": 0.403532745765199,
      "grad_norm": 0.7042028940113164,
      "learning_rate": 6.765929195351697e-06,
      "loss": 0.514,
      "step": 5500
    },
    {
      "epoch": 0.4036061153553381,
      "grad_norm": 1.6087480128085796,
      "learning_rate": 6.764817524086363e-06,
      "loss": 0.788,
      "step": 5501
    },
    {
      "epoch": 0.40367948494547723,
      "grad_norm": 1.683694931851507,
      "learning_rate": 6.763705753157413e-06,
      "loss": 0.8363,
      "step": 5502
    },
    {
      "epoch": 0.4037528545356163,
      "grad_norm": 1.5213626789369428,
      "learning_rate": 6.76259388262763e-06,
      "loss": 0.7741,
      "step": 5503
    },
    {
      "epoch": 0.4038262241257555,
      "grad_norm": 1.536734142770269,
      "learning_rate": 6.7614819125598075e-06,
      "loss": 0.8481,
      "step": 5504
    },
    {
      "epoch": 0.4038262241257555,
      "lm_loss": 1.4437,
      "step": 5504,
      "vm_loss": 0.1051
    },
    {
      "epoch": 0.4038262241257555,
      "lm_loss": 0.8135,
      "step": 5504,
      "vm_loss": 0.0976
    },
    {
      "epoch": 0.4038262241257555,
      "lm_loss": 0.1119,
      "step": 5504,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.4038262241257555,
      "lm_loss": 0.717,
      "step": 5504,
      "vm_loss": 0.141
    },
    {
      "epoch": 0.4038262241257555,
      "lm_loss": 0.7945,
      "step": 5504,
      "vm_loss": 0.1723
    },
    {
      "epoch": 0.4038262241257555,
      "lm_loss": 0.4975,
      "step": 5504,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.4038262241257555,
      "lm_loss": 0.4891,
      "step": 5504,
      "vm_loss": 0.1446
    },
    {
      "epoch": 0.4038262241257555,
      "lm_loss": 0.4295,
      "step": 5504,
      "vm_loss": 0.1333
    },
    {
      "epoch": 0.4038995937158946,
      "grad_norm": 1.5957007968462642,
      "learning_rate": 6.760369843016736e-06,
      "loss": 0.7944,
      "step": 5505
    },
    {
      "epoch": 0.4039729633060337,
      "grad_norm": 0.6902747460644287,
      "learning_rate": 6.759257674061224e-06,
      "loss": 0.5162,
      "step": 5506
    },
    {
      "epoch": 0.40404633289617287,
      "grad_norm": 0.6579801337446856,
      "learning_rate": 6.758145405756073e-06,
      "loss": 0.5354,
      "step": 5507
    },
    {
      "epoch": 0.40411970248631196,
      "grad_norm": 1.5638094734582475,
      "learning_rate": 6.7570330381640945e-06,
      "loss": 0.7563,
      "step": 5508
    },
    {
      "epoch": 0.4041930720764511,
      "grad_norm": 0.5969065642795905,
      "learning_rate": 6.755920571348111e-06,
      "loss": 0.4905,
      "step": 5509
    },
    {
      "epoch": 0.40426644166659026,
      "grad_norm": 1.519333508182415,
      "learning_rate": 6.754808005370946e-06,
      "loss": 0.8545,
      "step": 5510
    },
    {
      "epoch": 0.40433981125672935,
      "grad_norm": 1.7830827916771774,
      "learning_rate": 6.7536953402954255e-06,
      "loss": 0.8035,
      "step": 5511
    },
    {
      "epoch": 0.4044131808468685,
      "grad_norm": 1.5918223804998863,
      "learning_rate": 6.752582576184385e-06,
      "loss": 0.7142,
      "step": 5512
    },
    {
      "epoch": 0.4044131808468685,
      "lm_loss": 1.2442,
      "step": 5512,
      "vm_loss": 0.1137
    },
    {
      "epoch": 0.4044131808468685,
      "lm_loss": 1.984,
      "step": 5512,
      "vm_loss": 0.1573
    },
    {
      "epoch": 0.4044131808468685,
      "lm_loss": 0.1719,
      "step": 5512,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.4044131808468685,
      "lm_loss": 0.3259,
      "step": 5512,
      "vm_loss": 0.1326
    },
    {
      "epoch": 0.4044131808468685,
      "lm_loss": 0.9118,
      "step": 5512,
      "vm_loss": 0.1782
    },
    {
      "epoch": 0.4044131808468685,
      "lm_loss": 0.6226,
      "step": 5512,
      "vm_loss": 0.1581
    },
    {
      "epoch": 0.4044131808468685,
      "lm_loss": 0.8904,
      "step": 5512,
      "vm_loss": 0.1861
    },
    {
      "epoch": 0.4044131808468685,
      "lm_loss": 0.0264,
      "step": 5512,
      "vm_loss": 0.161
    },
    {
      "epoch": 0.4044865504370076,
      "grad_norm": 1.532570138463848,
      "learning_rate": 6.7514697131006675e-06,
      "loss": 0.8749,
      "step": 5513
    },
    {
      "epoch": 0.40455992002714675,
      "grad_norm": 0.733754254670224,
      "learning_rate": 6.7503567511071185e-06,
      "loss": 0.5072,
      "step": 5514
    },
    {
      "epoch": 0.4046332896172859,
      "grad_norm": 1.606873505141044,
      "learning_rate": 6.74924369026659e-06,
      "loss": 0.775,
      "step": 5515
    },
    {
      "epoch": 0.404706659207425,
      "grad_norm": 1.5152500018394515,
      "learning_rate": 6.748130530641935e-06,
      "loss": 0.7603,
      "step": 5516
    },
    {
      "epoch": 0.40478002879756414,
      "grad_norm": 1.6342968683804726,
      "learning_rate": 6.747017272296023e-06,
      "loss": 0.792,
      "step": 5517
    },
    {
      "epoch": 0.40485339838770323,
      "grad_norm": 1.4008025009372689,
      "learning_rate": 6.7459039152917164e-06,
      "loss": 0.8605,
      "step": 5518
    },
    {
      "epoch": 0.4049267679778424,
      "grad_norm": 1.521644675979047,
      "learning_rate": 6.744790459691894e-06,
      "loss": 0.7401,
      "step": 5519
    },
    {
      "epoch": 0.40500013756798153,
      "grad_norm": 1.4921060868254483,
      "learning_rate": 6.743676905559433e-06,
      "loss": 0.7856,
      "step": 5520
    },
    {
      "epoch": 0.40500013756798153,
      "lm_loss": 1.4664,
      "step": 5520,
      "vm_loss": 0.0726
    },
    {
      "epoch": 0.40500013756798153,
      "lm_loss": 0.7133,
      "step": 5520,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.40500013756798153,
      "lm_loss": 0.4731,
      "step": 5520,
      "vm_loss": 0.155
    },
    {
      "epoch": 0.40500013756798153,
      "lm_loss": 0.1383,
      "step": 5520,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.40500013756798153,
      "lm_loss": 0.8337,
      "step": 5520,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.40500013756798153,
      "lm_loss": 1.0072,
      "step": 5520,
      "vm_loss": 0.0645
    },
    {
      "epoch": 0.40500013756798153,
      "lm_loss": 0.8557,
      "step": 5520,
      "vm_loss": 0.1966
    },
    {
      "epoch": 0.40500013756798153,
      "lm_loss": 0.6111,
      "step": 5520,
      "vm_loss": 0.1196
    },
    {
      "epoch": 0.4050735071581206,
      "grad_norm": 1.6156963998739502,
      "learning_rate": 6.742563252957218e-06,
      "loss": 0.765,
      "step": 5521
    },
    {
      "epoch": 0.4051468767482598,
      "grad_norm": 1.5735800948805017,
      "learning_rate": 6.741449501948143e-06,
      "loss": 0.8153,
      "step": 5522
    },
    {
      "epoch": 0.40522024633839887,
      "grad_norm": 0.6630426870539697,
      "learning_rate": 6.740335652595101e-06,
      "loss": 0.5125,
      "step": 5523
    },
    {
      "epoch": 0.405293615928538,
      "grad_norm": 0.6594834575552112,
      "learning_rate": 6.739221704960993e-06,
      "loss": 0.5106,
      "step": 5524
    },
    {
      "epoch": 0.40536698551867717,
      "grad_norm": 1.4057413287552623,
      "learning_rate": 6.73810765910873e-06,
      "loss": 0.7189,
      "step": 5525
    },
    {
      "epoch": 0.40544035510881626,
      "grad_norm": 1.6265686391039462,
      "learning_rate": 6.7369935151012225e-06,
      "loss": 0.7247,
      "step": 5526
    },
    {
      "epoch": 0.4055137246989554,
      "grad_norm": 0.5698807184481003,
      "learning_rate": 6.7358792730013915e-06,
      "loss": 0.5132,
      "step": 5527
    },
    {
      "epoch": 0.4055870942890945,
      "grad_norm": 1.5938823219911566,
      "learning_rate": 6.734764932872157e-06,
      "loss": 0.8214,
      "step": 5528
    },
    {
      "epoch": 0.4055870942890945,
      "lm_loss": 2.1673,
      "step": 5528,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.4055870942890945,
      "lm_loss": 0.2905,
      "step": 5528,
      "vm_loss": 0.1049
    },
    {
      "epoch": 0.4055870942890945,
      "lm_loss": 0.0881,
      "step": 5528,
      "vm_loss": 0.0652
    },
    {
      "epoch": 0.4055870942890945,
      "lm_loss": 0.2648,
      "step": 5528,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.4055870942890945,
      "lm_loss": 0.5023,
      "step": 5528,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.4055870942890945,
      "lm_loss": 0.414,
      "step": 5528,
      "vm_loss": 0.2297
    },
    {
      "epoch": 0.4055870942890945,
      "lm_loss": 0.1237,
      "step": 5528,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.4055870942890945,
      "lm_loss": 1.0532,
      "step": 5528,
      "vm_loss": 0.0984
    },
    {
      "epoch": 0.40566046387923366,
      "grad_norm": 1.5019051328414552,
      "learning_rate": 6.73365049477645e-06,
      "loss": 0.8136,
      "step": 5529
    },
    {
      "epoch": 0.4057338334693728,
      "grad_norm": 0.6193614200630373,
      "learning_rate": 6.732535958777209e-06,
      "loss": 0.5321,
      "step": 5530
    },
    {
      "epoch": 0.4058072030595119,
      "grad_norm": 1.70577050438156,
      "learning_rate": 6.73142132493737e-06,
      "loss": 0.7972,
      "step": 5531
    },
    {
      "epoch": 0.40588057264965105,
      "grad_norm": 0.5779174889775602,
      "learning_rate": 6.7303065933198806e-06,
      "loss": 0.5022,
      "step": 5532
    },
    {
      "epoch": 0.40595394223979014,
      "grad_norm": 1.583505483547795,
      "learning_rate": 6.729191763987694e-06,
      "loss": 0.7526,
      "step": 5533
    },
    {
      "epoch": 0.4060273118299293,
      "grad_norm": 1.538323781759825,
      "learning_rate": 6.7280768370037665e-06,
      "loss": 0.8038,
      "step": 5534
    },
    {
      "epoch": 0.40610068142006844,
      "grad_norm": 1.7762906851663887,
      "learning_rate": 6.7269618124310586e-06,
      "loss": 0.7596,
      "step": 5535
    },
    {
      "epoch": 0.40617405101020754,
      "grad_norm": 1.5432320072778603,
      "learning_rate": 6.725846690332542e-06,
      "loss": 0.7896,
      "step": 5536
    },
    {
      "epoch": 0.40617405101020754,
      "lm_loss": 0.2812,
      "step": 5536,
      "vm_loss": 0.0284
    },
    {
      "epoch": 0.40617405101020754,
      "lm_loss": 1.5882,
      "step": 5536,
      "vm_loss": 0.1614
    },
    {
      "epoch": 0.40617405101020754,
      "lm_loss": 0.2502,
      "step": 5536,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.40617405101020754,
      "lm_loss": 0.5436,
      "step": 5536,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.40617405101020754,
      "lm_loss": 0.1514,
      "step": 5536,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.40617405101020754,
      "lm_loss": 0.11,
      "step": 5536,
      "vm_loss": 0.0808
    },
    {
      "epoch": 0.40617405101020754,
      "lm_loss": 1.1082,
      "step": 5536,
      "vm_loss": 0.1873
    },
    {
      "epoch": 0.40617405101020754,
      "lm_loss": 1.0313,
      "step": 5536,
      "vm_loss": 0.0553
    },
    {
      "epoch": 0.4062474206003467,
      "grad_norm": 1.351230580654456,
      "learning_rate": 6.724731470771189e-06,
      "loss": 0.7432,
      "step": 5537
    },
    {
      "epoch": 0.4063207901904858,
      "grad_norm": 1.389993343164693,
      "learning_rate": 6.723616153809978e-06,
      "loss": 0.792,
      "step": 5538
    },
    {
      "epoch": 0.40639415978062493,
      "grad_norm": 1.5739535722169957,
      "learning_rate": 6.722500739511895e-06,
      "loss": 0.8146,
      "step": 5539
    },
    {
      "epoch": 0.4064675293707641,
      "grad_norm": 0.6975789460214691,
      "learning_rate": 6.7213852279399275e-06,
      "loss": 0.5029,
      "step": 5540
    },
    {
      "epoch": 0.4065408989609032,
      "grad_norm": 1.6197970991086705,
      "learning_rate": 6.7202696191570745e-06,
      "loss": 0.8075,
      "step": 5541
    },
    {
      "epoch": 0.4066142685510423,
      "grad_norm": 1.554432042319536,
      "learning_rate": 6.719153913226337e-06,
      "loss": 0.8,
      "step": 5542
    },
    {
      "epoch": 0.4066876381411814,
      "grad_norm": 1.6366880325593813,
      "learning_rate": 6.718038110210719e-06,
      "loss": 0.8135,
      "step": 5543
    },
    {
      "epoch": 0.40676100773132057,
      "grad_norm": 0.5699403757377127,
      "learning_rate": 6.716922210173236e-06,
      "loss": 0.5055,
      "step": 5544
    },
    {
      "epoch": 0.40676100773132057,
      "lm_loss": 1.0294,
      "step": 5544,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.40676100773132057,
      "lm_loss": 0.2414,
      "step": 5544,
      "vm_loss": 0.1519
    },
    {
      "epoch": 0.40676100773132057,
      "lm_loss": 1.0154,
      "step": 5544,
      "vm_loss": 0.0673
    },
    {
      "epoch": 0.40676100773132057,
      "lm_loss": 0.2832,
      "step": 5544,
      "vm_loss": 0.1378
    },
    {
      "epoch": 0.40676100773132057,
      "lm_loss": 0.541,
      "step": 5544,
      "vm_loss": 0.1072
    },
    {
      "epoch": 0.40676100773132057,
      "lm_loss": 1.3231,
      "step": 5544,
      "vm_loss": 0.0947
    },
    {
      "epoch": 0.40676100773132057,
      "lm_loss": 0.9976,
      "step": 5544,
      "vm_loss": 0.1736
    },
    {
      "epoch": 0.40676100773132057,
      "lm_loss": 0.7109,
      "step": 5544,
      "vm_loss": 0.1324
    },
    {
      "epoch": 0.40683437732145966,
      "grad_norm": 1.4662242916511918,
      "learning_rate": 6.715806213176902e-06,
      "loss": 0.8832,
      "step": 5545
    },
    {
      "epoch": 0.4069077469115988,
      "grad_norm": 1.6583392170869868,
      "learning_rate": 6.714690119284743e-06,
      "loss": 0.837,
      "step": 5546
    },
    {
      "epoch": 0.40698111650173796,
      "grad_norm": 1.5113575594025224,
      "learning_rate": 6.713573928559787e-06,
      "loss": 0.8357,
      "step": 5547
    },
    {
      "epoch": 0.40705448609187705,
      "grad_norm": 0.6147579609516555,
      "learning_rate": 6.712457641065066e-06,
      "loss": 0.5177,
      "step": 5548
    },
    {
      "epoch": 0.4071278556820162,
      "grad_norm": 1.4405670602405891,
      "learning_rate": 6.711341256863623e-06,
      "loss": 0.8533,
      "step": 5549
    },
    {
      "epoch": 0.4072012252721553,
      "grad_norm": 0.5890815393862399,
      "learning_rate": 6.7102247760185016e-06,
      "loss": 0.4947,
      "step": 5550
    },
    {
      "epoch": 0.40727459486229445,
      "grad_norm": 1.5445771000401483,
      "learning_rate": 6.709108198592749e-06,
      "loss": 0.7548,
      "step": 5551
    },
    {
      "epoch": 0.4073479644524336,
      "grad_norm": 0.5804934410436241,
      "learning_rate": 6.707991524649426e-06,
      "loss": 0.517,
      "step": 5552
    },
    {
      "epoch": 0.4073479644524336,
      "lm_loss": 0.6084,
      "step": 5552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4073479644524336,
      "lm_loss": 0.36,
      "step": 5552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4073479644524336,
      "lm_loss": 0.5364,
      "step": 5552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4073479644524336,
      "lm_loss": 0.6359,
      "step": 5552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4073479644524336,
      "lm_loss": 0.4455,
      "step": 5552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4073479644524336,
      "lm_loss": 0.3809,
      "step": 5552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4073479644524336,
      "lm_loss": 0.5472,
      "step": 5552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4073479644524336,
      "lm_loss": 0.523,
      "step": 5552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4074213340425727,
      "grad_norm": 0.5604100506176587,
      "learning_rate": 6.7068747542515915e-06,
      "loss": 0.5053,
      "step": 5553
    },
    {
      "epoch": 0.40749470363271184,
      "grad_norm": 1.7859000566112047,
      "learning_rate": 6.705757887462312e-06,
      "loss": 0.7817,
      "step": 5554
    },
    {
      "epoch": 0.40756807322285094,
      "grad_norm": 1.4807451995629992,
      "learning_rate": 6.704640924344663e-06,
      "loss": 0.8422,
      "step": 5555
    },
    {
      "epoch": 0.4076414428129901,
      "grad_norm": 1.5100424227608442,
      "learning_rate": 6.7035238649617166e-06,
      "loss": 0.8377,
      "step": 5556
    },
    {
      "epoch": 0.40771481240312923,
      "grad_norm": 1.5321997391999624,
      "learning_rate": 6.70240670937656e-06,
      "loss": 0.7196,
      "step": 5557
    },
    {
      "epoch": 0.40778818199326833,
      "grad_norm": 0.5775986742711966,
      "learning_rate": 6.701289457652279e-06,
      "loss": 0.5214,
      "step": 5558
    },
    {
      "epoch": 0.4078615515834075,
      "grad_norm": 1.4677592461468858,
      "learning_rate": 6.700172109851971e-06,
      "loss": 0.7558,
      "step": 5559
    },
    {
      "epoch": 0.40793492117354657,
      "grad_norm": 0.5642200672134139,
      "learning_rate": 6.699054666038734e-06,
      "loss": 0.5027,
      "step": 5560
    },
    {
      "epoch": 0.40793492117354657,
      "lm_loss": 0.8719,
      "step": 5560,
      "vm_loss": 0.0441
    },
    {
      "epoch": 0.40793492117354657,
      "lm_loss": 0.2703,
      "step": 5560,
      "vm_loss": 0.129
    },
    {
      "epoch": 0.40793492117354657,
      "lm_loss": 0.6455,
      "step": 5560,
      "vm_loss": 0.123
    },
    {
      "epoch": 0.40793492117354657,
      "lm_loss": 0.9479,
      "step": 5560,
      "vm_loss": 0.1748
    },
    {
      "epoch": 0.40793492117354657,
      "lm_loss": 0.6868,
      "step": 5560,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.40793492117354657,
      "lm_loss": 0.7829,
      "step": 5560,
      "vm_loss": 0.0504
    },
    {
      "epoch": 0.40793492117354657,
      "lm_loss": 0.1755,
      "step": 5560,
      "vm_loss": 0.1192
    },
    {
      "epoch": 0.40793492117354657,
      "lm_loss": 0.7611,
      "step": 5560,
      "vm_loss": 0.1269
    },
    {
      "epoch": 0.4080082907636857,
      "grad_norm": 1.6805557031107656,
      "learning_rate": 6.6979371262756695e-06,
      "loss": 0.9204,
      "step": 5561
    },
    {
      "epoch": 0.40808166035382487,
      "grad_norm": 1.4652191377601285,
      "learning_rate": 6.696819490625893e-06,
      "loss": 0.6943,
      "step": 5562
    },
    {
      "epoch": 0.40815502994396397,
      "grad_norm": 1.4079455643711005,
      "learning_rate": 6.695701759152518e-06,
      "loss": 0.7333,
      "step": 5563
    },
    {
      "epoch": 0.4082283995341031,
      "grad_norm": 1.514001572774026,
      "learning_rate": 6.694583931918664e-06,
      "loss": 0.859,
      "step": 5564
    },
    {
      "epoch": 0.4083017691242422,
      "grad_norm": 1.4246253278116374,
      "learning_rate": 6.693466008987459e-06,
      "loss": 0.9177,
      "step": 5565
    },
    {
      "epoch": 0.40837513871438136,
      "grad_norm": 1.6397246138364725,
      "learning_rate": 6.692347990422034e-06,
      "loss": 0.8657,
      "step": 5566
    },
    {
      "epoch": 0.4084485083045205,
      "grad_norm": 1.530432526512802,
      "learning_rate": 6.691229876285529e-06,
      "loss": 0.8685,
      "step": 5567
    },
    {
      "epoch": 0.4085218778946596,
      "grad_norm": 1.4441015275679936,
      "learning_rate": 6.690111666641084e-06,
      "loss": 0.7517,
      "step": 5568
    },
    {
      "epoch": 0.4085218778946596,
      "lm_loss": 1.4986,
      "step": 5568,
      "vm_loss": 0.0309
    },
    {
      "epoch": 0.4085218778946596,
      "lm_loss": 0.5562,
      "step": 5568,
      "vm_loss": 0.0616
    },
    {
      "epoch": 0.4085218778946596,
      "lm_loss": 0.8432,
      "step": 5568,
      "vm_loss": 0.086
    },
    {
      "epoch": 0.4085218778946596,
      "lm_loss": 0.4933,
      "step": 5568,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.4085218778946596,
      "lm_loss": 0.8061,
      "step": 5568,
      "vm_loss": 0.1401
    },
    {
      "epoch": 0.4085218778946596,
      "lm_loss": 0.6414,
      "step": 5568,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.4085218778946596,
      "lm_loss": 0.7603,
      "step": 5568,
      "vm_loss": 0.099
    },
    {
      "epoch": 0.4085218778946596,
      "lm_loss": 0.1734,
      "step": 5568,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.40859524748479875,
      "grad_norm": 1.367113376264598,
      "learning_rate": 6.688993361551847e-06,
      "loss": 0.7074,
      "step": 5569
    },
    {
      "epoch": 0.40866861707493785,
      "grad_norm": 1.4830033339007112,
      "learning_rate": 6.687874961080972e-06,
      "loss": 0.936,
      "step": 5570
    },
    {
      "epoch": 0.408741986665077,
      "grad_norm": 1.3509524224695548,
      "learning_rate": 6.686756465291619e-06,
      "loss": 0.7644,
      "step": 5571
    },
    {
      "epoch": 0.40881535625521614,
      "grad_norm": 1.5753336153161712,
      "learning_rate": 6.68563787424695e-06,
      "loss": 0.8142,
      "step": 5572
    },
    {
      "epoch": 0.40888872584535524,
      "grad_norm": 0.6045124761504244,
      "learning_rate": 6.684519188010137e-06,
      "loss": 0.5121,
      "step": 5573
    },
    {
      "epoch": 0.4089620954354944,
      "grad_norm": 0.5967292917704781,
      "learning_rate": 6.683400406644353e-06,
      "loss": 0.5133,
      "step": 5574
    },
    {
      "epoch": 0.4090354650256335,
      "grad_norm": 0.5669729639978408,
      "learning_rate": 6.68228153021278e-06,
      "loss": 0.5244,
      "step": 5575
    },
    {
      "epoch": 0.40910883461577263,
      "grad_norm": 0.5655243006174476,
      "learning_rate": 6.6811625587786024e-06,
      "loss": 0.5355,
      "step": 5576
    },
    {
      "epoch": 0.40910883461577263,
      "lm_loss": 1.7233,
      "step": 5576,
      "vm_loss": 0.0576
    },
    {
      "epoch": 0.40910883461577263,
      "lm_loss": 0.3312,
      "step": 5576,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.40910883461577263,
      "lm_loss": 0.8398,
      "step": 5576,
      "vm_loss": 0.1008
    },
    {
      "epoch": 0.40910883461577263,
      "lm_loss": 0.1831,
      "step": 5576,
      "vm_loss": 0.0577
    },
    {
      "epoch": 0.40910883461577263,
      "lm_loss": 0.1993,
      "step": 5576,
      "vm_loss": 0.2258
    },
    {
      "epoch": 0.40910883461577263,
      "lm_loss": 1.642,
      "step": 5576,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.40910883461577263,
      "lm_loss": 0.0662,
      "step": 5576,
      "vm_loss": 0.0689
    },
    {
      "epoch": 0.40910883461577263,
      "lm_loss": 0.7163,
      "step": 5576,
      "vm_loss": 0.1983
    },
    {
      "epoch": 0.4091822042059118,
      "grad_norm": 1.4800391349411512,
      "learning_rate": 6.680043492405011e-06,
      "loss": 0.7555,
      "step": 5577
    },
    {
      "epoch": 0.4092555737960509,
      "grad_norm": 0.5695152232225453,
      "learning_rate": 6.6789243311552035e-06,
      "loss": 0.5255,
      "step": 5578
    },
    {
      "epoch": 0.40932894338619,
      "grad_norm": 1.7781708715390419,
      "learning_rate": 6.677805075092381e-06,
      "loss": 0.823,
      "step": 5579
    },
    {
      "epoch": 0.4094023129763291,
      "grad_norm": 1.4138984550733125,
      "learning_rate": 6.676685724279752e-06,
      "loss": 0.7042,
      "step": 5580
    },
    {
      "epoch": 0.40947568256646827,
      "grad_norm": 1.6463006482878164,
      "learning_rate": 6.675566278780527e-06,
      "loss": 0.8,
      "step": 5581
    },
    {
      "epoch": 0.40954905215660736,
      "grad_norm": 1.6471291569038606,
      "learning_rate": 6.674446738657926e-06,
      "loss": 0.8524,
      "step": 5582
    },
    {
      "epoch": 0.4096224217467465,
      "grad_norm": 0.6160681579617251,
      "learning_rate": 6.673327103975169e-06,
      "loss": 0.5217,
      "step": 5583
    },
    {
      "epoch": 0.40969579133688566,
      "grad_norm": 0.6207711657099898,
      "learning_rate": 6.672207374795488e-06,
      "loss": 0.5213,
      "step": 5584
    },
    {
      "epoch": 0.40969579133688566,
      "lm_loss": 1.6555,
      "step": 5584,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.40969579133688566,
      "lm_loss": 1.5756,
      "step": 5584,
      "vm_loss": 0.1398
    },
    {
      "epoch": 0.40969579133688566,
      "lm_loss": 1.0713,
      "step": 5584,
      "vm_loss": 0.254
    },
    {
      "epoch": 0.40969579133688566,
      "lm_loss": 0.6211,
      "step": 5584,
      "vm_loss": 0.2231
    },
    {
      "epoch": 0.40969579133688566,
      "lm_loss": 1.46,
      "step": 5584,
      "vm_loss": 0.101
    },
    {
      "epoch": 0.40969579133688566,
      "lm_loss": 0.8324,
      "step": 5584,
      "vm_loss": 0.1496
    },
    {
      "epoch": 0.40969579133688566,
      "lm_loss": 0.3209,
      "step": 5584,
      "vm_loss": 0.1019
    },
    {
      "epoch": 0.40969579133688566,
      "lm_loss": 1.1972,
      "step": 5584,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.40976916092702476,
      "grad_norm": 2.2545098621137316,
      "learning_rate": 6.671087551182114e-06,
      "loss": 0.8485,
      "step": 5585
    },
    {
      "epoch": 0.4098425305171639,
      "grad_norm": 1.4621308668249104,
      "learning_rate": 6.669967633198289e-06,
      "loss": 0.6736,
      "step": 5586
    },
    {
      "epoch": 0.409915900107303,
      "grad_norm": 0.6019514077784311,
      "learning_rate": 6.668847620907256e-06,
      "loss": 0.5082,
      "step": 5587
    },
    {
      "epoch": 0.40998926969744215,
      "grad_norm": 0.5868049048713939,
      "learning_rate": 6.667727514372264e-06,
      "loss": 0.5132,
      "step": 5588
    },
    {
      "epoch": 0.4100626392875813,
      "grad_norm": 0.570601543144629,
      "learning_rate": 6.66660731365657e-06,
      "loss": 0.517,
      "step": 5589
    },
    {
      "epoch": 0.4101360088777204,
      "grad_norm": 0.5789482639971602,
      "learning_rate": 6.665487018823435e-06,
      "loss": 0.5299,
      "step": 5590
    },
    {
      "epoch": 0.41020937846785954,
      "grad_norm": 1.9049671926213632,
      "learning_rate": 6.6643666299361205e-06,
      "loss": 0.8003,
      "step": 5591
    },
    {
      "epoch": 0.41028274805799864,
      "grad_norm": 2.095297094616019,
      "learning_rate": 6.663246147057902e-06,
      "loss": 0.91,
      "step": 5592
    },
    {
      "epoch": 0.41028274805799864,
      "lm_loss": 0.5847,
      "step": 5592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41028274805799864,
      "lm_loss": 0.5788,
      "step": 5592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41028274805799864,
      "lm_loss": 0.4407,
      "step": 5592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41028274805799864,
      "lm_loss": 0.6287,
      "step": 5592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41028274805799864,
      "lm_loss": 0.5642,
      "step": 5592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41028274805799864,
      "lm_loss": 0.5949,
      "step": 5592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41028274805799864,
      "lm_loss": 0.4247,
      "step": 5592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41028274805799864,
      "lm_loss": 0.7082,
      "step": 5592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4103561176481378,
      "grad_norm": 0.5690819023988323,
      "learning_rate": 6.6621255702520545e-06,
      "loss": 0.5147,
      "step": 5593
    },
    {
      "epoch": 0.41042948723827694,
      "grad_norm": 1.4638125757557325,
      "learning_rate": 6.661004899581861e-06,
      "loss": 0.7621,
      "step": 5594
    },
    {
      "epoch": 0.41050285682841603,
      "grad_norm": 1.3897443226520187,
      "learning_rate": 6.659884135110608e-06,
      "loss": 0.7081,
      "step": 5595
    },
    {
      "epoch": 0.4105762264185552,
      "grad_norm": 0.5658636111748476,
      "learning_rate": 6.658763276901585e-06,
      "loss": 0.5065,
      "step": 5596
    },
    {
      "epoch": 0.4106495960086943,
      "grad_norm": 0.5963289778138743,
      "learning_rate": 6.657642325018093e-06,
      "loss": 0.524,
      "step": 5597
    },
    {
      "epoch": 0.4107229655988334,
      "grad_norm": 1.9614056420463213,
      "learning_rate": 6.656521279523434e-06,
      "loss": 0.8866,
      "step": 5598
    },
    {
      "epoch": 0.4107963351889726,
      "grad_norm": 1.6813440343825157,
      "learning_rate": 6.655400140480916e-06,
      "loss": 0.9334,
      "step": 5599
    },
    {
      "epoch": 0.41086970477911167,
      "grad_norm": 1.5123759806244068,
      "learning_rate": 6.654278907953852e-06,
      "loss": 0.8437,
      "step": 5600
    },
    {
      "epoch": 0.41086970477911167,
      "lm_loss": 1.6814,
      "step": 5600,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.41086970477911167,
      "lm_loss": 0.1285,
      "step": 5600,
      "vm_loss": 0.0928
    },
    {
      "epoch": 0.41086970477911167,
      "lm_loss": 0.6897,
      "step": 5600,
      "vm_loss": 0.0634
    },
    {
      "epoch": 0.41086970477911167,
      "lm_loss": 0.6741,
      "step": 5600,
      "vm_loss": 0.1488
    },
    {
      "epoch": 0.41086970477911167,
      "lm_loss": 0.6936,
      "step": 5600,
      "vm_loss": 0.0699
    },
    {
      "epoch": 0.41086970477911167,
      "lm_loss": 0.6885,
      "step": 5600,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.41086970477911167,
      "lm_loss": 1.0119,
      "step": 5600,
      "vm_loss": 0.0602
    },
    {
      "epoch": 0.41086970477911167,
      "lm_loss": 1.1422,
      "step": 5600,
      "vm_loss": 0.1562
    },
    {
      "epoch": 0.4109430743692508,
      "grad_norm": 1.942145561086258,
      "learning_rate": 6.6531575820055605e-06,
      "loss": 0.8613,
      "step": 5601
    },
    {
      "epoch": 0.4110164439593899,
      "grad_norm": 0.5826794748054139,
      "learning_rate": 6.652036162699368e-06,
      "loss": 0.5065,
      "step": 5602
    },
    {
      "epoch": 0.41108981354952906,
      "grad_norm": 1.4366534813160696,
      "learning_rate": 6.650914650098602e-06,
      "loss": 0.6593,
      "step": 5603
    },
    {
      "epoch": 0.4111631831396682,
      "grad_norm": 1.4948041954294948,
      "learning_rate": 6.649793044266596e-06,
      "loss": 0.6984,
      "step": 5604
    },
    {
      "epoch": 0.4112365527298073,
      "grad_norm": 1.5665555246238871,
      "learning_rate": 6.648671345266693e-06,
      "loss": 0.8663,
      "step": 5605
    },
    {
      "epoch": 0.41130992231994645,
      "grad_norm": 0.5913659099702283,
      "learning_rate": 6.6475495531622356e-06,
      "loss": 0.4903,
      "step": 5606
    },
    {
      "epoch": 0.41138329191008555,
      "grad_norm": 1.4571749901150641,
      "learning_rate": 6.646427668016573e-06,
      "loss": 0.7924,
      "step": 5607
    },
    {
      "epoch": 0.4114566615002247,
      "grad_norm": 1.4714746851694998,
      "learning_rate": 6.6453056898930644e-06,
      "loss": 0.8332,
      "step": 5608
    },
    {
      "epoch": 0.4114566615002247,
      "lm_loss": 0.3088,
      "step": 5608,
      "vm_loss": 0.0949
    },
    {
      "epoch": 0.4114566615002247,
      "lm_loss": 0.4338,
      "step": 5608,
      "vm_loss": 0.2168
    },
    {
      "epoch": 0.4114566615002247,
      "lm_loss": 0.2274,
      "step": 5608,
      "vm_loss": 0.1262
    },
    {
      "epoch": 0.4114566615002247,
      "lm_loss": 0.7638,
      "step": 5608,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.4114566615002247,
      "lm_loss": 0.8864,
      "step": 5608,
      "vm_loss": 0.1923
    },
    {
      "epoch": 0.4114566615002247,
      "lm_loss": 1.5207,
      "step": 5608,
      "vm_loss": 0.2087
    },
    {
      "epoch": 0.4114566615002247,
      "lm_loss": 0.7393,
      "step": 5608,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.4114566615002247,
      "lm_loss": 0.3218,
      "step": 5608,
      "vm_loss": 0.0876
    },
    {
      "epoch": 0.41153003109036385,
      "grad_norm": 1.4645153944956895,
      "learning_rate": 6.64418361885507e-06,
      "loss": 0.805,
      "step": 5609
    },
    {
      "epoch": 0.41160340068050294,
      "grad_norm": 1.3570561582763943,
      "learning_rate": 6.643061454965953e-06,
      "loss": 0.6585,
      "step": 5610
    },
    {
      "epoch": 0.4116767702706421,
      "grad_norm": 1.6393887883352127,
      "learning_rate": 6.641939198289088e-06,
      "loss": 0.8177,
      "step": 5611
    },
    {
      "epoch": 0.4117501398607812,
      "grad_norm": 0.575620091716497,
      "learning_rate": 6.640816848887849e-06,
      "loss": 0.5137,
      "step": 5612
    },
    {
      "epoch": 0.41182350945092033,
      "grad_norm": 1.434508047201879,
      "learning_rate": 6.6396944068256195e-06,
      "loss": 0.7126,
      "step": 5613
    },
    {
      "epoch": 0.4118968790410595,
      "grad_norm": 1.267156152950263,
      "learning_rate": 6.638571872165788e-06,
      "loss": 0.6231,
      "step": 5614
    },
    {
      "epoch": 0.4119702486311986,
      "grad_norm": 1.6992958369970108,
      "learning_rate": 6.637449244971742e-06,
      "loss": 0.8759,
      "step": 5615
    },
    {
      "epoch": 0.4120436182213377,
      "grad_norm": 1.5812155381546653,
      "learning_rate": 6.636326525306884e-06,
      "loss": 0.8518,
      "step": 5616
    },
    {
      "epoch": 0.4120436182213377,
      "lm_loss": 1.6153,
      "step": 5616,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.4120436182213377,
      "lm_loss": 1.4357,
      "step": 5616,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.4120436182213377,
      "lm_loss": 0.0444,
      "step": 5616,
      "vm_loss": 0.1262
    },
    {
      "epoch": 0.4120436182213377,
      "lm_loss": 0.4382,
      "step": 5616,
      "vm_loss": 0.1033
    },
    {
      "epoch": 0.4120436182213377,
      "lm_loss": 0.065,
      "step": 5616,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.4120436182213377,
      "lm_loss": 0.117,
      "step": 5616,
      "vm_loss": 0.0418
    },
    {
      "epoch": 0.4120436182213377,
      "lm_loss": 0.7984,
      "step": 5616,
      "vm_loss": 0.1917
    },
    {
      "epoch": 0.4120436182213377,
      "lm_loss": 1.1788,
      "step": 5616,
      "vm_loss": 0.1335
    },
    {
      "epoch": 0.4121169878114768,
      "grad_norm": 1.7259892744879686,
      "learning_rate": 6.635203713234614e-06,
      "loss": 0.8109,
      "step": 5617
    },
    {
      "epoch": 0.41219035740161597,
      "grad_norm": 0.5786111171492677,
      "learning_rate": 6.634080808818342e-06,
      "loss": 0.5213,
      "step": 5618
    },
    {
      "epoch": 0.4122637269917551,
      "grad_norm": 0.5690139191755406,
      "learning_rate": 6.632957812121479e-06,
      "loss": 0.4921,
      "step": 5619
    },
    {
      "epoch": 0.4123370965818942,
      "grad_norm": 1.9145180493161726,
      "learning_rate": 6.631834723207444e-06,
      "loss": 0.8233,
      "step": 5620
    },
    {
      "epoch": 0.41241046617203336,
      "grad_norm": 1.4903955292670905,
      "learning_rate": 6.630711542139662e-06,
      "loss": 0.7545,
      "step": 5621
    },
    {
      "epoch": 0.41248383576217246,
      "grad_norm": 1.633211220206911,
      "learning_rate": 6.629588268981561e-06,
      "loss": 0.7664,
      "step": 5622
    },
    {
      "epoch": 0.4125572053523116,
      "grad_norm": 1.535463210645494,
      "learning_rate": 6.628464903796573e-06,
      "loss": 0.7667,
      "step": 5623
    },
    {
      "epoch": 0.4126305749424507,
      "grad_norm": 0.5843201413138616,
      "learning_rate": 6.627341446648141e-06,
      "loss": 0.5272,
      "step": 5624
    },
    {
      "epoch": 0.4126305749424507,
      "lm_loss": 1.1243,
      "step": 5624,
      "vm_loss": 0.1737
    },
    {
      "epoch": 0.4126305749424507,
      "lm_loss": 0.9441,
      "step": 5624,
      "vm_loss": 0.1666
    },
    {
      "epoch": 0.4126305749424507,
      "lm_loss": 0.2225,
      "step": 5624,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.4126305749424507,
      "lm_loss": 1.0932,
      "step": 5624,
      "vm_loss": 0.1976
    },
    {
      "epoch": 0.4126305749424507,
      "lm_loss": 0.4868,
      "step": 5624,
      "vm_loss": 0.1378
    },
    {
      "epoch": 0.4126305749424507,
      "lm_loss": 0.3402,
      "step": 5624,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.4126305749424507,
      "lm_loss": 1.103,
      "step": 5624,
      "vm_loss": 0.173
    },
    {
      "epoch": 0.4126305749424507,
      "lm_loss": 0.6797,
      "step": 5624,
      "vm_loss": 0.1766
    },
    {
      "epoch": 0.41270394453258985,
      "grad_norm": 2.6371414958440527,
      "learning_rate": 6.626217897599707e-06,
      "loss": 0.8841,
      "step": 5625
    },
    {
      "epoch": 0.412777314122729,
      "grad_norm": 1.443132231995119,
      "learning_rate": 6.62509425671472e-06,
      "loss": 0.6441,
      "step": 5626
    },
    {
      "epoch": 0.4128506837128681,
      "grad_norm": 1.5184215144404818,
      "learning_rate": 6.6239705240566375e-06,
      "loss": 0.8403,
      "step": 5627
    },
    {
      "epoch": 0.41292405330300724,
      "grad_norm": 1.745637916471724,
      "learning_rate": 6.622846699688916e-06,
      "loss": 0.7521,
      "step": 5628
    },
    {
      "epoch": 0.41299742289314634,
      "grad_norm": 1.5612651964778679,
      "learning_rate": 6.621722783675024e-06,
      "loss": 0.7009,
      "step": 5629
    },
    {
      "epoch": 0.4130707924832855,
      "grad_norm": 1.6711275441302105,
      "learning_rate": 6.620598776078431e-06,
      "loss": 0.8886,
      "step": 5630
    },
    {
      "epoch": 0.41314416207342464,
      "grad_norm": 1.6489383327419285,
      "learning_rate": 6.61947467696261e-06,
      "loss": 0.8744,
      "step": 5631
    },
    {
      "epoch": 0.41321753166356373,
      "grad_norm": 1.5760867105694645,
      "learning_rate": 6.618350486391044e-06,
      "loss": 0.7475,
      "step": 5632
    },
    {
      "epoch": 0.41321753166356373,
      "lm_loss": 0.4704,
      "step": 5632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41321753166356373,
      "lm_loss": 0.4767,
      "step": 5632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41321753166356373,
      "lm_loss": 0.536,
      "step": 5632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41321753166356373,
      "lm_loss": 0.6002,
      "step": 5632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41321753166356373,
      "lm_loss": 0.5035,
      "step": 5632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41321753166356373,
      "lm_loss": 0.4537,
      "step": 5632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41321753166356373,
      "lm_loss": 0.588,
      "step": 5632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41321753166356373,
      "lm_loss": 0.3745,
      "step": 5632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4132909012537029,
      "grad_norm": 0.5916931639325673,
      "learning_rate": 6.6172262044272186e-06,
      "loss": 0.5128,
      "step": 5633
    },
    {
      "epoch": 0.413364270843842,
      "grad_norm": 0.5804361192301242,
      "learning_rate": 6.616101831134624e-06,
      "loss": 0.5153,
      "step": 5634
    },
    {
      "epoch": 0.4134376404339811,
      "grad_norm": 1.6377973613788424,
      "learning_rate": 6.614977366576757e-06,
      "loss": 0.8295,
      "step": 5635
    },
    {
      "epoch": 0.4135110100241203,
      "grad_norm": 1.605264052117689,
      "learning_rate": 6.6138528108171175e-06,
      "loss": 0.755,
      "step": 5636
    },
    {
      "epoch": 0.41358437961425937,
      "grad_norm": 1.5157707469335204,
      "learning_rate": 6.612728163919215e-06,
      "loss": 0.7914,
      "step": 5637
    },
    {
      "epoch": 0.4136577492043985,
      "grad_norm": 1.4960880556305776,
      "learning_rate": 6.611603425946559e-06,
      "loss": 0.8032,
      "step": 5638
    },
    {
      "epoch": 0.4137311187945376,
      "grad_norm": 1.4844603819365692,
      "learning_rate": 6.610478596962665e-06,
      "loss": 0.8191,
      "step": 5639
    },
    {
      "epoch": 0.41380448838467676,
      "grad_norm": 1.4706104924279018,
      "learning_rate": 6.6093536770310575e-06,
      "loss": 0.8066,
      "step": 5640
    },
    {
      "epoch": 0.41380448838467676,
      "lm_loss": 0.4445,
      "step": 5640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41380448838467676,
      "lm_loss": 0.4141,
      "step": 5640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41380448838467676,
      "lm_loss": 0.3893,
      "step": 5640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41380448838467676,
      "lm_loss": 0.5153,
      "step": 5640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41380448838467676,
      "lm_loss": 0.4031,
      "step": 5640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41380448838467676,
      "lm_loss": 0.5518,
      "step": 5640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41380448838467676,
      "lm_loss": 0.6251,
      "step": 5640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41380448838467676,
      "lm_loss": 0.6835,
      "step": 5640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4138778579748159,
      "grad_norm": 0.619717743264924,
      "learning_rate": 6.608228666215262e-06,
      "loss": 0.5276,
      "step": 5641
    },
    {
      "epoch": 0.413951227564955,
      "grad_norm": 0.5689256213747136,
      "learning_rate": 6.60710356457881e-06,
      "loss": 0.5005,
      "step": 5642
    },
    {
      "epoch": 0.41402459715509415,
      "grad_norm": 1.6945007177422549,
      "learning_rate": 6.6059783721852426e-06,
      "loss": 0.7909,
      "step": 5643
    },
    {
      "epoch": 0.41409796674523325,
      "grad_norm": 1.5547826053125353,
      "learning_rate": 6.604853089098095e-06,
      "loss": 0.8485,
      "step": 5644
    },
    {
      "epoch": 0.4141713363353724,
      "grad_norm": 1.6227414254203751,
      "learning_rate": 6.603727715380923e-06,
      "loss": 0.7069,
      "step": 5645
    },
    {
      "epoch": 0.41424470592551155,
      "grad_norm": 2.2188945894869643,
      "learning_rate": 6.602602251097274e-06,
      "loss": 0.8948,
      "step": 5646
    },
    {
      "epoch": 0.41431807551565064,
      "grad_norm": 1.427691280259342,
      "learning_rate": 6.6014766963107065e-06,
      "loss": 0.8308,
      "step": 5647
    },
    {
      "epoch": 0.4143914451057898,
      "grad_norm": 1.5747429075130086,
      "learning_rate": 6.600351051084785e-06,
      "loss": 0.7734,
      "step": 5648
    },
    {
      "epoch": 0.4143914451057898,
      "lm_loss": 0.196,
      "step": 5648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4143914451057898,
      "lm_loss": 0.4172,
      "step": 5648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4143914451057898,
      "lm_loss": 0.4749,
      "step": 5648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4143914451057898,
      "lm_loss": 0.4797,
      "step": 5648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4143914451057898,
      "lm_loss": 0.4727,
      "step": 5648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4143914451057898,
      "lm_loss": 0.3905,
      "step": 5648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4143914451057898,
      "lm_loss": 0.5752,
      "step": 5648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4143914451057898,
      "lm_loss": 0.4998,
      "step": 5648,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4144648146959289,
      "grad_norm": 0.6715379239816894,
      "learning_rate": 6.599225315483076e-06,
      "loss": 0.5016,
      "step": 5649
    },
    {
      "epoch": 0.41453818428606803,
      "grad_norm": 1.4915352879013186,
      "learning_rate": 6.598099489569154e-06,
      "loss": 0.716,
      "step": 5650
    },
    {
      "epoch": 0.4146115538762072,
      "grad_norm": 1.6388353495072834,
      "learning_rate": 6.5969735734065956e-06,
      "loss": 0.9311,
      "step": 5651
    },
    {
      "epoch": 0.4146849234663463,
      "grad_norm": 1.5682166230401218,
      "learning_rate": 6.595847567058984e-06,
      "loss": 0.6994,
      "step": 5652
    },
    {
      "epoch": 0.4147582930564854,
      "grad_norm": 1.5861422377808387,
      "learning_rate": 6.594721470589908e-06,
      "loss": 0.8132,
      "step": 5653
    },
    {
      "epoch": 0.4148316626466245,
      "grad_norm": 1.3407576143523765,
      "learning_rate": 6.593595284062963e-06,
      "loss": 0.8398,
      "step": 5654
    },
    {
      "epoch": 0.41490503223676367,
      "grad_norm": 1.4627949070933506,
      "learning_rate": 6.592469007541744e-06,
      "loss": 0.801,
      "step": 5655
    },
    {
      "epoch": 0.4149784018269028,
      "grad_norm": 0.6229228183208255,
      "learning_rate": 6.591342641089857e-06,
      "loss": 0.5127,
      "step": 5656
    },
    {
      "epoch": 0.4149784018269028,
      "lm_loss": 0.4383,
      "step": 5656,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.4149784018269028,
      "lm_loss": 1.4169,
      "step": 5656,
      "vm_loss": 0.0732
    },
    {
      "epoch": 0.4149784018269028,
      "lm_loss": 0.2621,
      "step": 5656,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.4149784018269028,
      "lm_loss": 0.2562,
      "step": 5656,
      "vm_loss": 0.0996
    },
    {
      "epoch": 0.4149784018269028,
      "lm_loss": 0.2908,
      "step": 5656,
      "vm_loss": 0.1543
    },
    {
      "epoch": 0.4149784018269028,
      "lm_loss": 0.4831,
      "step": 5656,
      "vm_loss": 0.1622
    },
    {
      "epoch": 0.4149784018269028,
      "lm_loss": 0.8512,
      "step": 5656,
      "vm_loss": 0.2243
    },
    {
      "epoch": 0.4149784018269028,
      "lm_loss": 0.6104,
      "step": 5656,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.4150517714170419,
      "grad_norm": 1.4040437046668548,
      "learning_rate": 6.590216184770911e-06,
      "loss": 0.7924,
      "step": 5657
    },
    {
      "epoch": 0.41512514100718106,
      "grad_norm": 1.6147129423826352,
      "learning_rate": 6.589089638648518e-06,
      "loss": 0.8452,
      "step": 5658
    },
    {
      "epoch": 0.41519851059732016,
      "grad_norm": 0.5909601193037858,
      "learning_rate": 6.587963002786299e-06,
      "loss": 0.5171,
      "step": 5659
    },
    {
      "epoch": 0.4152718801874593,
      "grad_norm": 0.5849313851605853,
      "learning_rate": 6.586836277247874e-06,
      "loss": 0.5268,
      "step": 5660
    },
    {
      "epoch": 0.41534524977759846,
      "grad_norm": 1.3664159998164278,
      "learning_rate": 6.585709462096878e-06,
      "loss": 0.787,
      "step": 5661
    },
    {
      "epoch": 0.41541861936773755,
      "grad_norm": 1.6336425943847381,
      "learning_rate": 6.5845825573969406e-06,
      "loss": 0.7948,
      "step": 5662
    },
    {
      "epoch": 0.4154919889578767,
      "grad_norm": 1.46206103738428,
      "learning_rate": 6.5834555632117e-06,
      "loss": 0.745,
      "step": 5663
    },
    {
      "epoch": 0.4155653585480158,
      "grad_norm": 1.4476289941101566,
      "learning_rate": 6.5823284796048045e-06,
      "loss": 0.7675,
      "step": 5664
    },
    {
      "epoch": 0.4155653585480158,
      "lm_loss": 1.0354,
      "step": 5664,
      "vm_loss": 0.0945
    },
    {
      "epoch": 0.4155653585480158,
      "lm_loss": 0.2198,
      "step": 5664,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.4155653585480158,
      "lm_loss": 1.0983,
      "step": 5664,
      "vm_loss": 0.1614
    },
    {
      "epoch": 0.4155653585480158,
      "lm_loss": 0.1339,
      "step": 5664,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.4155653585480158,
      "lm_loss": 0.2007,
      "step": 5664,
      "vm_loss": 0.0987
    },
    {
      "epoch": 0.4155653585480158,
      "lm_loss": 0.4256,
      "step": 5664,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.4155653585480158,
      "lm_loss": 0.9274,
      "step": 5664,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.4155653585480158,
      "lm_loss": 0.7191,
      "step": 5664,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.41563872813815494,
      "grad_norm": 1.6425884747120965,
      "learning_rate": 6.581201306639901e-06,
      "loss": 0.7777,
      "step": 5665
    },
    {
      "epoch": 0.41571209772829404,
      "grad_norm": 1.571525856275321,
      "learning_rate": 6.580074044380643e-06,
      "loss": 0.8124,
      "step": 5666
    },
    {
      "epoch": 0.4157854673184332,
      "grad_norm": 1.409807034585023,
      "learning_rate": 6.578946692890692e-06,
      "loss": 0.7808,
      "step": 5667
    },
    {
      "epoch": 0.41585883690857234,
      "grad_norm": 0.6505562573232647,
      "learning_rate": 6.577819252233711e-06,
      "loss": 0.507,
      "step": 5668
    },
    {
      "epoch": 0.41593220649871143,
      "grad_norm": 1.5148576867067396,
      "learning_rate": 6.576691722473368e-06,
      "loss": 0.7649,
      "step": 5669
    },
    {
      "epoch": 0.4160055760888506,
      "grad_norm": 1.6014338960639294,
      "learning_rate": 6.57556410367334e-06,
      "loss": 0.7544,
      "step": 5670
    },
    {
      "epoch": 0.4160789456789897,
      "grad_norm": 1.5241965310536296,
      "learning_rate": 6.5744363958973045e-06,
      "loss": 0.8433,
      "step": 5671
    },
    {
      "epoch": 0.4161523152691288,
      "grad_norm": 1.4778034075597153,
      "learning_rate": 6.5733085992089476e-06,
      "loss": 0.854,
      "step": 5672
    },
    {
      "epoch": 0.4161523152691288,
      "lm_loss": 1.2722,
      "step": 5672,
      "vm_loss": 0.0469
    },
    {
      "epoch": 0.4161523152691288,
      "lm_loss": 0.6992,
      "step": 5672,
      "vm_loss": 0.1187
    },
    {
      "epoch": 0.4161523152691288,
      "lm_loss": 1.0101,
      "step": 5672,
      "vm_loss": 0.1048
    },
    {
      "epoch": 0.4161523152691288,
      "lm_loss": 0.2892,
      "step": 5672,
      "vm_loss": 0.102
    },
    {
      "epoch": 0.4161523152691288,
      "lm_loss": 0.406,
      "step": 5672,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.4161523152691288,
      "lm_loss": 0.1332,
      "step": 5672,
      "vm_loss": 0.2427
    },
    {
      "epoch": 0.4161523152691288,
      "lm_loss": 0.7907,
      "step": 5672,
      "vm_loss": 0.2579
    },
    {
      "epoch": 0.4161523152691288,
      "lm_loss": 1.1042,
      "step": 5672,
      "vm_loss": 0.1504
    },
    {
      "epoch": 0.416225684859268,
      "grad_norm": 1.617778736333022,
      "learning_rate": 6.572180713671958e-06,
      "loss": 0.8079,
      "step": 5673
    },
    {
      "epoch": 0.41629905444940707,
      "grad_norm": 1.472372962672688,
      "learning_rate": 6.571052739350029e-06,
      "loss": 0.7486,
      "step": 5674
    },
    {
      "epoch": 0.4163724240395462,
      "grad_norm": 1.4595114736229875,
      "learning_rate": 6.569924676306861e-06,
      "loss": 0.7149,
      "step": 5675
    },
    {
      "epoch": 0.4164457936296853,
      "grad_norm": 1.444395285750029,
      "learning_rate": 6.568796524606159e-06,
      "loss": 0.9453,
      "step": 5676
    },
    {
      "epoch": 0.41651916321982446,
      "grad_norm": 0.5899730708820288,
      "learning_rate": 6.567668284311633e-06,
      "loss": 0.5056,
      "step": 5677
    },
    {
      "epoch": 0.4165925328099636,
      "grad_norm": 1.4511167913271856,
      "learning_rate": 6.5665399554869964e-06,
      "loss": 0.822,
      "step": 5678
    },
    {
      "epoch": 0.4166659024001027,
      "grad_norm": 0.5538052652969746,
      "learning_rate": 6.5654115381959695e-06,
      "loss": 0.5087,
      "step": 5679
    },
    {
      "epoch": 0.41673927199024186,
      "grad_norm": 1.5761568260021268,
      "learning_rate": 6.5642830325022756e-06,
      "loss": 0.8731,
      "step": 5680
    },
    {
      "epoch": 0.41673927199024186,
      "lm_loss": 0.3617,
      "step": 5680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41673927199024186,
      "lm_loss": 0.6094,
      "step": 5680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41673927199024186,
      "lm_loss": 0.3708,
      "step": 5680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41673927199024186,
      "lm_loss": 0.4463,
      "step": 5680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41673927199024186,
      "lm_loss": 0.4555,
      "step": 5680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41673927199024186,
      "lm_loss": 0.6819,
      "step": 5680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41673927199024186,
      "lm_loss": 0.3753,
      "step": 5680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41673927199024186,
      "lm_loss": 0.7432,
      "step": 5680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41681264158038095,
      "grad_norm": 0.5693093871489189,
      "learning_rate": 6.563154438469645e-06,
      "loss": 0.516,
      "step": 5681
    },
    {
      "epoch": 0.4168860111705201,
      "grad_norm": 1.5521270962611782,
      "learning_rate": 6.562025756161812e-06,
      "loss": 0.8398,
      "step": 5682
    },
    {
      "epoch": 0.41695938076065925,
      "grad_norm": 1.391852017168658,
      "learning_rate": 6.5608969856425165e-06,
      "loss": 0.7004,
      "step": 5683
    },
    {
      "epoch": 0.41703275035079834,
      "grad_norm": 1.3441383692194018,
      "learning_rate": 6.559768126975503e-06,
      "loss": 0.729,
      "step": 5684
    },
    {
      "epoch": 0.4171061199409375,
      "grad_norm": 1.415494719229654,
      "learning_rate": 6.55863918022452e-06,
      "loss": 0.8165,
      "step": 5685
    },
    {
      "epoch": 0.4171794895310766,
      "grad_norm": 1.4639823065064597,
      "learning_rate": 6.557510145453323e-06,
      "loss": 0.8398,
      "step": 5686
    },
    {
      "epoch": 0.41725285912121574,
      "grad_norm": 1.2820169082477935,
      "learning_rate": 6.5563810227256705e-06,
      "loss": 0.859,
      "step": 5687
    },
    {
      "epoch": 0.4173262287113549,
      "grad_norm": 1.540537728884965,
      "learning_rate": 6.555251812105328e-06,
      "loss": 0.8124,
      "step": 5688
    },
    {
      "epoch": 0.4173262287113549,
      "lm_loss": 0.4303,
      "step": 5688,
      "vm_loss": 0.0323
    },
    {
      "epoch": 0.4173262287113549,
      "lm_loss": 0.3907,
      "step": 5688,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.4173262287113549,
      "lm_loss": 0.9064,
      "step": 5688,
      "vm_loss": 0.1251
    },
    {
      "epoch": 0.4173262287113549,
      "lm_loss": 0.9893,
      "step": 5688,
      "vm_loss": 0.1513
    },
    {
      "epoch": 0.4173262287113549,
      "lm_loss": 0.1793,
      "step": 5688,
      "vm_loss": 0.1717
    },
    {
      "epoch": 0.4173262287113549,
      "lm_loss": 0.378,
      "step": 5688,
      "vm_loss": 0.0813
    },
    {
      "epoch": 0.4173262287113549,
      "lm_loss": 0.3095,
      "step": 5688,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.4173262287113549,
      "lm_loss": 0.0948,
      "step": 5688,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.417399598301494,
      "grad_norm": 1.4923412106452738,
      "learning_rate": 6.554122513656065e-06,
      "loss": 0.7705,
      "step": 5689
    },
    {
      "epoch": 0.41747296789163313,
      "grad_norm": 0.625950402117795,
      "learning_rate": 6.552993127441654e-06,
      "loss": 0.5196,
      "step": 5690
    },
    {
      "epoch": 0.4175463374817722,
      "grad_norm": 1.5370125086582425,
      "learning_rate": 6.551863653525875e-06,
      "loss": 0.7775,
      "step": 5691
    },
    {
      "epoch": 0.4176197070719114,
      "grad_norm": 1.5673818616035862,
      "learning_rate": 6.550734091972512e-06,
      "loss": 0.7521,
      "step": 5692
    },
    {
      "epoch": 0.4176930766620505,
      "grad_norm": 1.411075066460446,
      "learning_rate": 6.549604442845354e-06,
      "loss": 0.7639,
      "step": 5693
    },
    {
      "epoch": 0.4177664462521896,
      "grad_norm": 0.5676533287438463,
      "learning_rate": 6.548474706208198e-06,
      "loss": 0.4994,
      "step": 5694
    },
    {
      "epoch": 0.41783981584232877,
      "grad_norm": 0.5744142944214187,
      "learning_rate": 6.547344882124837e-06,
      "loss": 0.5168,
      "step": 5695
    },
    {
      "epoch": 0.41791318543246786,
      "grad_norm": 1.4487445745491607,
      "learning_rate": 6.546214970659083e-06,
      "loss": 0.6956,
      "step": 5696
    },
    {
      "epoch": 0.41791318543246786,
      "lm_loss": 0.5959,
      "step": 5696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41791318543246786,
      "lm_loss": 0.5462,
      "step": 5696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41791318543246786,
      "lm_loss": 0.3041,
      "step": 5696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41791318543246786,
      "lm_loss": 0.5184,
      "step": 5696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41791318543246786,
      "lm_loss": 0.5302,
      "step": 5696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41791318543246786,
      "lm_loss": 0.4919,
      "step": 5696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41791318543246786,
      "lm_loss": 0.5454,
      "step": 5696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.41791318543246786,
      "lm_loss": 0.5842,
      "step": 5696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.417986555022607,
      "grad_norm": 0.5588149527605407,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.5036,
      "step": 5697
    },
    {
      "epoch": 0.41805992461274616,
      "grad_norm": 1.4813962264360616,
      "learning_rate": 6.5439548858356176e-06,
      "loss": 0.7911,
      "step": 5698
    },
    {
      "epoch": 0.41813329420288525,
      "grad_norm": 1.4800595062820292,
      "learning_rate": 6.542824712605542e-06,
      "loss": 0.8471,
      "step": 5699
    },
    {
      "epoch": 0.4182066637930244,
      "grad_norm": 1.734351447899208,
      "learning_rate": 6.541694452248335e-06,
      "loss": 0.7666,
      "step": 5700
    },
    {
      "epoch": 0.4182800333831635,
      "grad_norm": 1.377460397702389,
      "learning_rate": 6.540564104827824e-06,
      "loss": 0.8095,
      "step": 5701
    },
    {
      "epoch": 0.41835340297330265,
      "grad_norm": 1.598774023107293,
      "learning_rate": 6.539433670407842e-06,
      "loss": 0.8529,
      "step": 5702
    },
    {
      "epoch": 0.41842677256344174,
      "grad_norm": 1.4295246001500381,
      "learning_rate": 6.538303149052227e-06,
      "loss": 0.7363,
      "step": 5703
    },
    {
      "epoch": 0.4185001421535809,
      "grad_norm": 1.390525727158259,
      "learning_rate": 6.537172540824827e-06,
      "loss": 0.766,
      "step": 5704
    },
    {
      "epoch": 0.4185001421535809,
      "lm_loss": 1.3295,
      "step": 5704,
      "vm_loss": 0.0928
    },
    {
      "epoch": 0.4185001421535809,
      "lm_loss": 1.6443,
      "step": 5704,
      "vm_loss": 0.1536
    },
    {
      "epoch": 0.4185001421535809,
      "lm_loss": 0.1289,
      "step": 5704,
      "vm_loss": 0.1215
    },
    {
      "epoch": 0.4185001421535809,
      "lm_loss": 0.1052,
      "step": 5704,
      "vm_loss": 0.1596
    },
    {
      "epoch": 0.4185001421535809,
      "lm_loss": 0.4914,
      "step": 5704,
      "vm_loss": 0.1176
    },
    {
      "epoch": 0.4185001421535809,
      "lm_loss": 0.992,
      "step": 5704,
      "vm_loss": 0.1136
    },
    {
      "epoch": 0.4185001421535809,
      "lm_loss": 1.0283,
      "step": 5704,
      "vm_loss": 0.1804
    },
    {
      "epoch": 0.4185001421535809,
      "lm_loss": 0.5105,
      "step": 5704,
      "vm_loss": 0.1297
    },
    {
      "epoch": 0.41857351174372004,
      "grad_norm": 1.3167187601049328,
      "learning_rate": 6.536041845789487e-06,
      "loss": 0.778,
      "step": 5705
    },
    {
      "epoch": 0.41864688133385913,
      "grad_norm": 1.6605027335111697,
      "learning_rate": 6.534911064010056e-06,
      "loss": 0.7163,
      "step": 5706
    },
    {
      "epoch": 0.4187202509239983,
      "grad_norm": 1.539110863056966,
      "learning_rate": 6.5337801955504e-06,
      "loss": 0.7502,
      "step": 5707
    },
    {
      "epoch": 0.4187936205141374,
      "grad_norm": 0.6783127952720531,
      "learning_rate": 6.5326492404743755e-06,
      "loss": 0.512,
      "step": 5708
    },
    {
      "epoch": 0.4188669901042765,
      "grad_norm": 0.6471949404860918,
      "learning_rate": 6.531518198845854e-06,
      "loss": 0.492,
      "step": 5709
    },
    {
      "epoch": 0.4189403596944157,
      "grad_norm": 1.307766860594199,
      "learning_rate": 6.530387070728708e-06,
      "loss": 0.7957,
      "step": 5710
    },
    {
      "epoch": 0.41901372928455477,
      "grad_norm": 1.527563764978825,
      "learning_rate": 6.529255856186812e-06,
      "loss": 0.7054,
      "step": 5711
    },
    {
      "epoch": 0.4190870988746939,
      "grad_norm": 1.3376547731769857,
      "learning_rate": 6.528124555284052e-06,
      "loss": 0.732,
      "step": 5712
    },
    {
      "epoch": 0.4190870988746939,
      "lm_loss": 1.888,
      "step": 5712,
      "vm_loss": 0.0541
    },
    {
      "epoch": 0.4190870988746939,
      "lm_loss": 0.2761,
      "step": 5712,
      "vm_loss": 0.1353
    },
    {
      "epoch": 0.4190870988746939,
      "lm_loss": 1.0452,
      "step": 5712,
      "vm_loss": 0.1448
    },
    {
      "epoch": 0.4190870988746939,
      "lm_loss": 0.9346,
      "step": 5712,
      "vm_loss": 0.1501
    },
    {
      "epoch": 0.4190870988746939,
      "lm_loss": 0.8651,
      "step": 5712,
      "vm_loss": 0.148
    },
    {
      "epoch": 0.4190870988746939,
      "lm_loss": 0.8516,
      "step": 5712,
      "vm_loss": 0.1411
    },
    {
      "epoch": 0.4190870988746939,
      "lm_loss": 0.8689,
      "step": 5712,
      "vm_loss": 0.2277
    },
    {
      "epoch": 0.4190870988746939,
      "lm_loss": 0.3745,
      "step": 5712,
      "vm_loss": 0.089
    },
    {
      "epoch": 0.419160468464833,
      "grad_norm": 1.5136369225305109,
      "learning_rate": 6.526993168084315e-06,
      "loss": 0.7543,
      "step": 5713
    },
    {
      "epoch": 0.41923383805497216,
      "grad_norm": 1.4831961344120026,
      "learning_rate": 6.525861694651489e-06,
      "loss": 0.8492,
      "step": 5714
    },
    {
      "epoch": 0.4193072076451113,
      "grad_norm": 1.7119256114345367,
      "learning_rate": 6.524730135049477e-06,
      "loss": 0.8388,
      "step": 5715
    },
    {
      "epoch": 0.4193805772352504,
      "grad_norm": 1.4263705205709876,
      "learning_rate": 6.523598489342176e-06,
      "loss": 0.7291,
      "step": 5716
    },
    {
      "epoch": 0.41945394682538956,
      "grad_norm": 1.5709826132536042,
      "learning_rate": 6.522466757593498e-06,
      "loss": 0.8332,
      "step": 5717
    },
    {
      "epoch": 0.41952731641552865,
      "grad_norm": 1.3427032576019156,
      "learning_rate": 6.521334939867351e-06,
      "loss": 0.7799,
      "step": 5718
    },
    {
      "epoch": 0.4196006860056678,
      "grad_norm": 0.7890542515575315,
      "learning_rate": 6.520203036227651e-06,
      "loss": 0.521,
      "step": 5719
    },
    {
      "epoch": 0.41967405559580695,
      "grad_norm": 1.52621070400978,
      "learning_rate": 6.519071046738323e-06,
      "loss": 0.7355,
      "step": 5720
    },
    {
      "epoch": 0.41967405559580695,
      "lm_loss": 1.3991,
      "step": 5720,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.41967405559580695,
      "lm_loss": 0.91,
      "step": 5720,
      "vm_loss": 0.1637
    },
    {
      "epoch": 0.41967405559580695,
      "lm_loss": 0.4864,
      "step": 5720,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.41967405559580695,
      "lm_loss": 0.2196,
      "step": 5720,
      "vm_loss": 0.1289
    },
    {
      "epoch": 0.41967405559580695,
      "lm_loss": 0.5838,
      "step": 5720,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.41967405559580695,
      "lm_loss": 0.5489,
      "step": 5720,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.41967405559580695,
      "lm_loss": 0.1843,
      "step": 5720,
      "vm_loss": 0.0679
    },
    {
      "epoch": 0.41967405559580695,
      "lm_loss": 0.0965,
      "step": 5720,
      "vm_loss": 0.1568
    },
    {
      "epoch": 0.41974742518594604,
      "grad_norm": 1.4303673040880878,
      "learning_rate": 6.5179389714632905e-06,
      "loss": 0.7025,
      "step": 5721
    },
    {
      "epoch": 0.4198207947760852,
      "grad_norm": 1.5357639228669928,
      "learning_rate": 6.516806810466484e-06,
      "loss": 0.9089,
      "step": 5722
    },
    {
      "epoch": 0.4198941643662243,
      "grad_norm": 1.485572312473698,
      "learning_rate": 6.5156745638118416e-06,
      "loss": 0.7477,
      "step": 5723
    },
    {
      "epoch": 0.41996753395636344,
      "grad_norm": 1.4688960818481,
      "learning_rate": 6.514542231563302e-06,
      "loss": 0.7822,
      "step": 5724
    },
    {
      "epoch": 0.4200409035465026,
      "grad_norm": 1.4149680246365153,
      "learning_rate": 6.513409813784813e-06,
      "loss": 0.79,
      "step": 5725
    },
    {
      "epoch": 0.4201142731366417,
      "grad_norm": 1.4746265168165305,
      "learning_rate": 6.5122773105403235e-06,
      "loss": 0.8551,
      "step": 5726
    },
    {
      "epoch": 0.42018764272678083,
      "grad_norm": 1.56789339362449,
      "learning_rate": 6.511144721893789e-06,
      "loss": 0.784,
      "step": 5727
    },
    {
      "epoch": 0.4202610123169199,
      "grad_norm": 1.560410091762272,
      "learning_rate": 6.510012047909171e-06,
      "loss": 0.7259,
      "step": 5728
    },
    {
      "epoch": 0.4202610123169199,
      "lm_loss": 1.7895,
      "step": 5728,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.4202610123169199,
      "lm_loss": 0.44,
      "step": 5728,
      "vm_loss": 0.0695
    },
    {
      "epoch": 0.4202610123169199,
      "lm_loss": 0.82,
      "step": 5728,
      "vm_loss": 0.1426
    },
    {
      "epoch": 0.4202610123169199,
      "lm_loss": 0.687,
      "step": 5728,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.4202610123169199,
      "lm_loss": 0.2026,
      "step": 5728,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.4202610123169199,
      "lm_loss": 0.8277,
      "step": 5728,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.4202610123169199,
      "lm_loss": 0.1825,
      "step": 5728,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.4202610123169199,
      "lm_loss": 0.2336,
      "step": 5728,
      "vm_loss": 0.1234
    },
    {
      "epoch": 0.4203343819070591,
      "grad_norm": 1.442335720358996,
      "learning_rate": 6.508879288650431e-06,
      "loss": 0.7391,
      "step": 5729
    },
    {
      "epoch": 0.4204077514971982,
      "grad_norm": 0.5824842390747049,
      "learning_rate": 6.507746444181542e-06,
      "loss": 0.5162,
      "step": 5730
    },
    {
      "epoch": 0.4204811210873373,
      "grad_norm": 1.3843824022417606,
      "learning_rate": 6.506613514566478e-06,
      "loss": 0.6945,
      "step": 5731
    },
    {
      "epoch": 0.42055449067747647,
      "grad_norm": 1.5326146671181478,
      "learning_rate": 6.505480499869218e-06,
      "loss": 0.6758,
      "step": 5732
    },
    {
      "epoch": 0.42062786026761556,
      "grad_norm": 1.4444558803247,
      "learning_rate": 6.504347400153745e-06,
      "loss": 0.7654,
      "step": 5733
    },
    {
      "epoch": 0.4207012298577547,
      "grad_norm": 1.5494525711062677,
      "learning_rate": 6.503214215484048e-06,
      "loss": 0.8673,
      "step": 5734
    },
    {
      "epoch": 0.42077459944789386,
      "grad_norm": 1.5558737423569147,
      "learning_rate": 6.502080945924123e-06,
      "loss": 0.7562,
      "step": 5735
    },
    {
      "epoch": 0.42084796903803295,
      "grad_norm": 1.3638040716192124,
      "learning_rate": 6.500947591537967e-06,
      "loss": 0.7611,
      "step": 5736
    },
    {
      "epoch": 0.42084796903803295,
      "lm_loss": 1.0061,
      "step": 5736,
      "vm_loss": 0.0619
    },
    {
      "epoch": 0.42084796903803295,
      "lm_loss": 0.1545,
      "step": 5736,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.42084796903803295,
      "lm_loss": 0.1599,
      "step": 5736,
      "vm_loss": 0.0816
    },
    {
      "epoch": 0.42084796903803295,
      "lm_loss": 0.8294,
      "step": 5736,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.42084796903803295,
      "lm_loss": 0.2463,
      "step": 5736,
      "vm_loss": 0.1157
    },
    {
      "epoch": 0.42084796903803295,
      "lm_loss": 0.7769,
      "step": 5736,
      "vm_loss": 0.0747
    },
    {
      "epoch": 0.42084796903803295,
      "lm_loss": 0.1121,
      "step": 5736,
      "vm_loss": 0.0684
    },
    {
      "epoch": 0.42084796903803295,
      "lm_loss": 1.3518,
      "step": 5736,
      "vm_loss": 0.1553
    },
    {
      "epoch": 0.4209213386281721,
      "grad_norm": 1.5546032368560385,
      "learning_rate": 6.499814152389584e-06,
      "loss": 0.6912,
      "step": 5737
    },
    {
      "epoch": 0.4209947082183112,
      "grad_norm": 1.3841702297786709,
      "learning_rate": 6.498680628542979e-06,
      "loss": 0.7755,
      "step": 5738
    },
    {
      "epoch": 0.42106807780845035,
      "grad_norm": 1.5660784394927427,
      "learning_rate": 6.4975470200621705e-06,
      "loss": 0.8082,
      "step": 5739
    },
    {
      "epoch": 0.4211414473985895,
      "grad_norm": 1.4758373177673252,
      "learning_rate": 6.496413327011173e-06,
      "loss": 0.7491,
      "step": 5740
    },
    {
      "epoch": 0.4212148169887286,
      "grad_norm": 1.4036047033314203,
      "learning_rate": 6.495279549454007e-06,
      "loss": 0.7557,
      "step": 5741
    },
    {
      "epoch": 0.42128818657886774,
      "grad_norm": 1.6769535884548668,
      "learning_rate": 6.494145687454704e-06,
      "loss": 0.7907,
      "step": 5742
    },
    {
      "epoch": 0.42136155616900683,
      "grad_norm": 1.4746041324844739,
      "learning_rate": 6.493011741077292e-06,
      "loss": 0.8232,
      "step": 5743
    },
    {
      "epoch": 0.421434925759146,
      "grad_norm": 0.5834560298411924,
      "learning_rate": 6.49187771038581e-06,
      "loss": 0.5256,
      "step": 5744
    },
    {
      "epoch": 0.421434925759146,
      "lm_loss": 1.6628,
      "step": 5744,
      "vm_loss": 0.0445
    },
    {
      "epoch": 0.421434925759146,
      "lm_loss": 1.0247,
      "step": 5744,
      "vm_loss": 0.0673
    },
    {
      "epoch": 0.421434925759146,
      "lm_loss": 0.7415,
      "step": 5744,
      "vm_loss": 0.1133
    },
    {
      "epoch": 0.421434925759146,
      "lm_loss": 0.7402,
      "step": 5744,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.421434925759146,
      "lm_loss": 0.4489,
      "step": 5744,
      "vm_loss": 0.2191
    },
    {
      "epoch": 0.421434925759146,
      "lm_loss": 0.9016,
      "step": 5744,
      "vm_loss": 0.1629
    },
    {
      "epoch": 0.421434925759146,
      "lm_loss": 0.0742,
      "step": 5744,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.421434925759146,
      "lm_loss": 0.3698,
      "step": 5744,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.4215082953492851,
      "grad_norm": 1.6762548409540228,
      "learning_rate": 6.4907435954442995e-06,
      "loss": 0.7802,
      "step": 5745
    },
    {
      "epoch": 0.4215816649394242,
      "grad_norm": 1.4440911738378426,
      "learning_rate": 6.489609396316806e-06,
      "loss": 0.8076,
      "step": 5746
    },
    {
      "epoch": 0.4216550345295634,
      "grad_norm": 1.3625862050680264,
      "learning_rate": 6.488475113067383e-06,
      "loss": 0.8148,
      "step": 5747
    },
    {
      "epoch": 0.42172840411970247,
      "grad_norm": 0.5603883547110562,
      "learning_rate": 6.487340745760082e-06,
      "loss": 0.516,
      "step": 5748
    },
    {
      "epoch": 0.4218017737098416,
      "grad_norm": 1.5170055437677925,
      "learning_rate": 6.486206294458966e-06,
      "loss": 0.7813,
      "step": 5749
    },
    {
      "epoch": 0.4218751432999807,
      "grad_norm": 0.5410017682056439,
      "learning_rate": 6.485071759228101e-06,
      "loss": 0.4879,
      "step": 5750
    },
    {
      "epoch": 0.42194851289011986,
      "grad_norm": 1.289755894279864,
      "learning_rate": 6.483937140131554e-06,
      "loss": 0.7426,
      "step": 5751
    },
    {
      "epoch": 0.422021882480259,
      "grad_norm": 1.5121014262748202,
      "learning_rate": 6.4828024372334044e-06,
      "loss": 0.7181,
      "step": 5752
    },
    {
      "epoch": 0.422021882480259,
      "lm_loss": 1.6203,
      "step": 5752,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.422021882480259,
      "lm_loss": 1.3545,
      "step": 5752,
      "vm_loss": 0.0497
    },
    {
      "epoch": 0.422021882480259,
      "lm_loss": 1.0903,
      "step": 5752,
      "vm_loss": 0.116
    },
    {
      "epoch": 0.422021882480259,
      "lm_loss": 0.0638,
      "step": 5752,
      "vm_loss": 0.1337
    },
    {
      "epoch": 0.422021882480259,
      "lm_loss": 0.3863,
      "step": 5752,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.422021882480259,
      "lm_loss": 0.341,
      "step": 5752,
      "vm_loss": 0.1729
    },
    {
      "epoch": 0.422021882480259,
      "lm_loss": 0.3455,
      "step": 5752,
      "vm_loss": 0.0684
    },
    {
      "epoch": 0.422021882480259,
      "lm_loss": 1.179,
      "step": 5752,
      "vm_loss": 0.1885
    },
    {
      "epoch": 0.4220952520703981,
      "grad_norm": 1.525459362589203,
      "learning_rate": 6.481667650597727e-06,
      "loss": 0.8319,
      "step": 5753
    },
    {
      "epoch": 0.42216862166053726,
      "grad_norm": 1.4419969466306468,
      "learning_rate": 6.480532780288609e-06,
      "loss": 0.8347,
      "step": 5754
    },
    {
      "epoch": 0.42224199125067635,
      "grad_norm": 1.5686288330324523,
      "learning_rate": 6.479397826370139e-06,
      "loss": 0.8821,
      "step": 5755
    },
    {
      "epoch": 0.4223153608408155,
      "grad_norm": 1.5936518353222306,
      "learning_rate": 6.4782627889064095e-06,
      "loss": 0.7111,
      "step": 5756
    },
    {
      "epoch": 0.42238873043095465,
      "grad_norm": 1.4879985766671167,
      "learning_rate": 6.4771276679615195e-06,
      "loss": 0.7277,
      "step": 5757
    },
    {
      "epoch": 0.42246210002109374,
      "grad_norm": 1.578973342852665,
      "learning_rate": 6.475992463599572e-06,
      "loss": 0.7639,
      "step": 5758
    },
    {
      "epoch": 0.4225354696112329,
      "grad_norm": 0.6499762965528777,
      "learning_rate": 6.474857175884673e-06,
      "loss": 0.5162,
      "step": 5759
    },
    {
      "epoch": 0.422608839201372,
      "grad_norm": 0.6305521078294426,
      "learning_rate": 6.473721804880939e-06,
      "loss": 0.5032,
      "step": 5760
    },
    {
      "epoch": 0.422608839201372,
      "lm_loss": 0.8613,
      "step": 5760,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.422608839201372,
      "lm_loss": 0.9548,
      "step": 5760,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.422608839201372,
      "lm_loss": 0.3843,
      "step": 5760,
      "vm_loss": 0.1456
    },
    {
      "epoch": 0.422608839201372,
      "lm_loss": 0.8368,
      "step": 5760,
      "vm_loss": 0.1357
    },
    {
      "epoch": 0.422608839201372,
      "lm_loss": 1.1315,
      "step": 5760,
      "vm_loss": 0.0813
    },
    {
      "epoch": 0.422608839201372,
      "lm_loss": 0.229,
      "step": 5760,
      "vm_loss": 0.1899
    },
    {
      "epoch": 0.422608839201372,
      "lm_loss": 0.5462,
      "step": 5760,
      "vm_loss": 0.067
    },
    {
      "epoch": 0.422608839201372,
      "lm_loss": 1.0499,
      "step": 5760,
      "vm_loss": 0.209
    },
    {
      "epoch": 0.42268220879151114,
      "grad_norm": 1.4658488950052297,
      "learning_rate": 6.472586350652486e-06,
      "loss": 0.7621,
      "step": 5761
    },
    {
      "epoch": 0.4227555783816503,
      "grad_norm": 1.4029811657290079,
      "learning_rate": 6.471450813263433e-06,
      "loss": 0.7293,
      "step": 5762
    },
    {
      "epoch": 0.4228289479717894,
      "grad_norm": 1.3791399272788256,
      "learning_rate": 6.47031519277791e-06,
      "loss": 0.7965,
      "step": 5763
    },
    {
      "epoch": 0.42290231756192853,
      "grad_norm": 1.3516952416142587,
      "learning_rate": 6.469179489260047e-06,
      "loss": 0.7237,
      "step": 5764
    },
    {
      "epoch": 0.4229756871520676,
      "grad_norm": 1.5574350413539269,
      "learning_rate": 6.468043702773977e-06,
      "loss": 0.8529,
      "step": 5765
    },
    {
      "epoch": 0.4230490567422068,
      "grad_norm": 1.4781308989657187,
      "learning_rate": 6.466907833383846e-06,
      "loss": 0.9021,
      "step": 5766
    },
    {
      "epoch": 0.4231224263323459,
      "grad_norm": 1.33375687526108,
      "learning_rate": 6.465771881153797e-06,
      "loss": 0.8348,
      "step": 5767
    },
    {
      "epoch": 0.423195795922485,
      "grad_norm": 1.4764313175634949,
      "learning_rate": 6.46463584614798e-06,
      "loss": 0.8134,
      "step": 5768
    },
    {
      "epoch": 0.423195795922485,
      "lm_loss": 1.5827,
      "step": 5768,
      "vm_loss": 0.0745
    },
    {
      "epoch": 0.423195795922485,
      "lm_loss": 1.0406,
      "step": 5768,
      "vm_loss": 0.1708
    },
    {
      "epoch": 0.423195795922485,
      "lm_loss": 0.271,
      "step": 5768,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.423195795922485,
      "lm_loss": 0.073,
      "step": 5768,
      "vm_loss": 0.0741
    },
    {
      "epoch": 0.423195795922485,
      "lm_loss": 0.0971,
      "step": 5768,
      "vm_loss": 0.1008
    },
    {
      "epoch": 0.423195795922485,
      "lm_loss": 0.7368,
      "step": 5768,
      "vm_loss": 0.1049
    },
    {
      "epoch": 0.423195795922485,
      "lm_loss": 0.3187,
      "step": 5768,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.423195795922485,
      "lm_loss": 1.3538,
      "step": 5768,
      "vm_loss": 0.1376
    },
    {
      "epoch": 0.42326916551262417,
      "grad_norm": 1.629086581385912,
      "learning_rate": 6.463499728430549e-06,
      "loss": 0.734,
      "step": 5769
    },
    {
      "epoch": 0.42334253510276326,
      "grad_norm": 1.6886179606449665,
      "learning_rate": 6.462363528065663e-06,
      "loss": 0.8412,
      "step": 5770
    },
    {
      "epoch": 0.4234159046929024,
      "grad_norm": 1.491703271666337,
      "learning_rate": 6.461227245117488e-06,
      "loss": 0.7108,
      "step": 5771
    },
    {
      "epoch": 0.42348927428304156,
      "grad_norm": 1.4847945490130312,
      "learning_rate": 6.460090879650194e-06,
      "loss": 0.7665,
      "step": 5772
    },
    {
      "epoch": 0.42356264387318066,
      "grad_norm": 0.6639944679413318,
      "learning_rate": 6.4589544317279485e-06,
      "loss": 0.5165,
      "step": 5773
    },
    {
      "epoch": 0.4236360134633198,
      "grad_norm": 1.3670662134560516,
      "learning_rate": 6.457817901414936e-06,
      "loss": 0.7793,
      "step": 5774
    },
    {
      "epoch": 0.4237093830534589,
      "grad_norm": 0.586338479861058,
      "learning_rate": 6.456681288775335e-06,
      "loss": 0.5093,
      "step": 5775
    },
    {
      "epoch": 0.42378275264359805,
      "grad_norm": 1.6044894240464067,
      "learning_rate": 6.455544593873334e-06,
      "loss": 0.8712,
      "step": 5776
    },
    {
      "epoch": 0.42378275264359805,
      "lm_loss": 0.1775,
      "step": 5776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42378275264359805,
      "lm_loss": 0.4549,
      "step": 5776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42378275264359805,
      "lm_loss": 0.4611,
      "step": 5776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42378275264359805,
      "lm_loss": 0.4475,
      "step": 5776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42378275264359805,
      "lm_loss": 0.4204,
      "step": 5776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42378275264359805,
      "lm_loss": 0.4512,
      "step": 5776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42378275264359805,
      "lm_loss": 0.5182,
      "step": 5776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42378275264359805,
      "lm_loss": 0.5523,
      "step": 5776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4238561222337372,
      "grad_norm": 0.5723152346592878,
      "learning_rate": 6.454407816773126e-06,
      "loss": 0.4899,
      "step": 5777
    },
    {
      "epoch": 0.4239294918238763,
      "grad_norm": 1.6433327119572585,
      "learning_rate": 6.453270957538906e-06,
      "loss": 0.8753,
      "step": 5778
    },
    {
      "epoch": 0.42400286141401544,
      "grad_norm": 1.6246374746580865,
      "learning_rate": 6.4521340162348765e-06,
      "loss": 0.9093,
      "step": 5779
    },
    {
      "epoch": 0.42407623100415454,
      "grad_norm": 1.5147446315070652,
      "learning_rate": 6.450996992925244e-06,
      "loss": 0.8818,
      "step": 5780
    },
    {
      "epoch": 0.4241496005942937,
      "grad_norm": 1.6175183721626316,
      "learning_rate": 6.449859887674217e-06,
      "loss": 0.8298,
      "step": 5781
    },
    {
      "epoch": 0.42422297018443283,
      "grad_norm": 1.328305358082072,
      "learning_rate": 6.448722700546013e-06,
      "loss": 0.8191,
      "step": 5782
    },
    {
      "epoch": 0.42429633977457193,
      "grad_norm": 1.3604721389921366,
      "learning_rate": 6.447585431604848e-06,
      "loss": 0.6545,
      "step": 5783
    },
    {
      "epoch": 0.4243697093647111,
      "grad_norm": 1.5254888136696516,
      "learning_rate": 6.4464480809149495e-06,
      "loss": 0.8153,
      "step": 5784
    },
    {
      "epoch": 0.4243697093647111,
      "lm_loss": 1.8101,
      "step": 5784,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.4243697093647111,
      "lm_loss": 1.2215,
      "step": 5784,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.4243697093647111,
      "lm_loss": 0.4682,
      "step": 5784,
      "vm_loss": 0.3527
    },
    {
      "epoch": 0.4243697093647111,
      "lm_loss": 0.4043,
      "step": 5784,
      "vm_loss": 0.1692
    },
    {
      "epoch": 0.4243697093647111,
      "lm_loss": 1.1077,
      "step": 5784,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.4243697093647111,
      "lm_loss": 0.7775,
      "step": 5784,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.4243697093647111,
      "lm_loss": 0.2557,
      "step": 5784,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.4243697093647111,
      "lm_loss": 0.3423,
      "step": 5784,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.4244430789548502,
      "grad_norm": 1.4665709704146632,
      "learning_rate": 6.445310648540547e-06,
      "loss": 0.8205,
      "step": 5785
    },
    {
      "epoch": 0.4245164485449893,
      "grad_norm": 1.7656239444505561,
      "learning_rate": 6.4441731345458715e-06,
      "loss": 0.8207,
      "step": 5786
    },
    {
      "epoch": 0.4245898181351284,
      "grad_norm": 0.7821468622126044,
      "learning_rate": 6.443035538995164e-06,
      "loss": 0.5103,
      "step": 5787
    },
    {
      "epoch": 0.42466318772526757,
      "grad_norm": 1.72252835780511,
      "learning_rate": 6.4418978619526654e-06,
      "loss": 0.749,
      "step": 5788
    },
    {
      "epoch": 0.4247365573154067,
      "grad_norm": 1.600662102682278,
      "learning_rate": 6.4407601034826225e-06,
      "loss": 0.8138,
      "step": 5789
    },
    {
      "epoch": 0.4248099269055458,
      "grad_norm": 1.6477602962436766,
      "learning_rate": 6.43962226364929e-06,
      "loss": 0.821,
      "step": 5790
    },
    {
      "epoch": 0.42488329649568496,
      "grad_norm": 1.439230597339051,
      "learning_rate": 6.438484342516922e-06,
      "loss": 0.759,
      "step": 5791
    },
    {
      "epoch": 0.42495666608582405,
      "grad_norm": 1.544060595411164,
      "learning_rate": 6.437346340149782e-06,
      "loss": 0.9032,
      "step": 5792
    },
    {
      "epoch": 0.42495666608582405,
      "lm_loss": 1.339,
      "step": 5792,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.42495666608582405,
      "lm_loss": 1.0146,
      "step": 5792,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.42495666608582405,
      "lm_loss": 1.0523,
      "step": 5792,
      "vm_loss": 0.1761
    },
    {
      "epoch": 0.42495666608582405,
      "lm_loss": 0.5244,
      "step": 5792,
      "vm_loss": 0.0504
    },
    {
      "epoch": 0.42495666608582405,
      "lm_loss": 0.5703,
      "step": 5792,
      "vm_loss": 0.1016
    },
    {
      "epoch": 0.42495666608582405,
      "lm_loss": 0.112,
      "step": 5792,
      "vm_loss": 0.1215
    },
    {
      "epoch": 0.42495666608582405,
      "lm_loss": 0.6267,
      "step": 5792,
      "vm_loss": 0.1332
    },
    {
      "epoch": 0.42495666608582405,
      "lm_loss": 0.7691,
      "step": 5792,
      "vm_loss": 0.0974
    },
    {
      "epoch": 0.4250300356759632,
      "grad_norm": 1.4063310520003074,
      "learning_rate": 6.436208256612134e-06,
      "loss": 0.7835,
      "step": 5793
    },
    {
      "epoch": 0.42510340526610235,
      "grad_norm": 1.5924432337484766,
      "learning_rate": 6.43507009196825e-06,
      "loss": 0.9378,
      "step": 5794
    },
    {
      "epoch": 0.42517677485624145,
      "grad_norm": 1.5132019939211392,
      "learning_rate": 6.433931846282403e-06,
      "loss": 0.6746,
      "step": 5795
    },
    {
      "epoch": 0.4252501444463806,
      "grad_norm": 0.7706517492037968,
      "learning_rate": 6.432793519618874e-06,
      "loss": 0.4954,
      "step": 5796
    },
    {
      "epoch": 0.4253235140365197,
      "grad_norm": 1.6121180950286136,
      "learning_rate": 6.4316551120419455e-06,
      "loss": 0.7641,
      "step": 5797
    },
    {
      "epoch": 0.42539688362665884,
      "grad_norm": 1.5597165544287528,
      "learning_rate": 6.4305166236159085e-06,
      "loss": 0.7618,
      "step": 5798
    },
    {
      "epoch": 0.425470253216798,
      "grad_norm": 1.505988459473516,
      "learning_rate": 6.429378054405055e-06,
      "loss": 0.8104,
      "step": 5799
    },
    {
      "epoch": 0.4255436228069371,
      "grad_norm": 0.5726195710389922,
      "learning_rate": 6.428239404473684e-06,
      "loss": 0.5005,
      "step": 5800
    },
    {
      "epoch": 0.4255436228069371,
      "lm_loss": 0.9517,
      "step": 5800,
      "vm_loss": 0.0536
    },
    {
      "epoch": 0.4255436228069371,
      "lm_loss": 0.7702,
      "step": 5800,
      "vm_loss": 0.1552
    },
    {
      "epoch": 0.4255436228069371,
      "lm_loss": 0.8969,
      "step": 5800,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.4255436228069371,
      "lm_loss": 0.0361,
      "step": 5800,
      "vm_loss": 0.1511
    },
    {
      "epoch": 0.4255436228069371,
      "lm_loss": 0.8507,
      "step": 5800,
      "vm_loss": 0.1523
    },
    {
      "epoch": 0.4255436228069371,
      "lm_loss": 1.0135,
      "step": 5800,
      "vm_loss": 0.1182
    },
    {
      "epoch": 0.4255436228069371,
      "lm_loss": 0.2399,
      "step": 5800,
      "vm_loss": 0.1541
    },
    {
      "epoch": 0.4255436228069371,
      "lm_loss": 0.0747,
      "step": 5800,
      "vm_loss": 0.0823
    },
    {
      "epoch": 0.42561699239707623,
      "grad_norm": 1.350420934423983,
      "learning_rate": 6.427100673886096e-06,
      "loss": 0.8272,
      "step": 5801
    },
    {
      "epoch": 0.4256903619872153,
      "grad_norm": 1.516195140018435,
      "learning_rate": 6.425961862706598e-06,
      "loss": 0.7721,
      "step": 5802
    },
    {
      "epoch": 0.4257637315773545,
      "grad_norm": 1.7665516643421846,
      "learning_rate": 6.424822970999502e-06,
      "loss": 0.794,
      "step": 5803
    },
    {
      "epoch": 0.4258371011674936,
      "grad_norm": 1.6349277667605415,
      "learning_rate": 6.4236839988291264e-06,
      "loss": 0.9106,
      "step": 5804
    },
    {
      "epoch": 0.4259104707576327,
      "grad_norm": 1.5992206026411515,
      "learning_rate": 6.422544946259788e-06,
      "loss": 0.7983,
      "step": 5805
    },
    {
      "epoch": 0.42598384034777187,
      "grad_norm": 1.3420465016801142,
      "learning_rate": 6.421405813355815e-06,
      "loss": 0.7553,
      "step": 5806
    },
    {
      "epoch": 0.42605720993791096,
      "grad_norm": 1.6783311338749904,
      "learning_rate": 6.420266600181534e-06,
      "loss": 0.8386,
      "step": 5807
    },
    {
      "epoch": 0.4261305795280501,
      "grad_norm": 1.5874876665043,
      "learning_rate": 6.419127306801282e-06,
      "loss": 0.8397,
      "step": 5808
    },
    {
      "epoch": 0.4261305795280501,
      "lm_loss": 0.6106,
      "step": 5808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4261305795280501,
      "lm_loss": 0.5681,
      "step": 5808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4261305795280501,
      "lm_loss": 0.7074,
      "step": 5808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4261305795280501,
      "lm_loss": 0.4467,
      "step": 5808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4261305795280501,
      "lm_loss": 0.4541,
      "step": 5808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4261305795280501,
      "lm_loss": 0.4295,
      "step": 5808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4261305795280501,
      "lm_loss": 0.2706,
      "step": 5808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4261305795280501,
      "lm_loss": 0.74,
      "step": 5808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42620394911818926,
      "grad_norm": 0.5993810226245675,
      "learning_rate": 6.417987933279397e-06,
      "loss": 0.5097,
      "step": 5809
    },
    {
      "epoch": 0.42627731870832836,
      "grad_norm": 1.5826131386772035,
      "learning_rate": 6.41684847968022e-06,
      "loss": 0.8967,
      "step": 5810
    },
    {
      "epoch": 0.4263506882984675,
      "grad_norm": 1.544272281720512,
      "learning_rate": 6.415708946068102e-06,
      "loss": 0.7679,
      "step": 5811
    },
    {
      "epoch": 0.4264240578886066,
      "grad_norm": 0.561235008395034,
      "learning_rate": 6.4145693325073944e-06,
      "loss": 0.5123,
      "step": 5812
    },
    {
      "epoch": 0.42649742747874575,
      "grad_norm": 1.4789274933328704,
      "learning_rate": 6.413429639062452e-06,
      "loss": 0.8179,
      "step": 5813
    },
    {
      "epoch": 0.4265707970688849,
      "grad_norm": 0.5556603756068103,
      "learning_rate": 6.412289865797639e-06,
      "loss": 0.481,
      "step": 5814
    },
    {
      "epoch": 0.426644166659024,
      "grad_norm": 1.6805502574188387,
      "learning_rate": 6.411150012777319e-06,
      "loss": 0.7735,
      "step": 5815
    },
    {
      "epoch": 0.42671753624916314,
      "grad_norm": 1.8381271657121814,
      "learning_rate": 6.410010080065862e-06,
      "loss": 0.779,
      "step": 5816
    },
    {
      "epoch": 0.42671753624916314,
      "lm_loss": 0.6989,
      "step": 5816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42671753624916314,
      "lm_loss": 0.5432,
      "step": 5816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42671753624916314,
      "lm_loss": 0.5929,
      "step": 5816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42671753624916314,
      "lm_loss": 0.5629,
      "step": 5816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42671753624916314,
      "lm_loss": 0.4414,
      "step": 5816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42671753624916314,
      "lm_loss": 0.2717,
      "step": 5816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42671753624916314,
      "lm_loss": 0.4631,
      "step": 5816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42671753624916314,
      "lm_loss": 0.5193,
      "step": 5816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42679090583930224,
      "grad_norm": 0.5838924203070963,
      "learning_rate": 6.408870067727645e-06,
      "loss": 0.5149,
      "step": 5817
    },
    {
      "epoch": 0.4268642754294414,
      "grad_norm": 1.3967953624328535,
      "learning_rate": 6.407729975827046e-06,
      "loss": 0.8224,
      "step": 5818
    },
    {
      "epoch": 0.42693764501958054,
      "grad_norm": 1.4932650192088024,
      "learning_rate": 6.406589804428449e-06,
      "loss": 0.7479,
      "step": 5819
    },
    {
      "epoch": 0.42701101460971963,
      "grad_norm": 0.5810999119215376,
      "learning_rate": 6.405449553596244e-06,
      "loss": 0.511,
      "step": 5820
    },
    {
      "epoch": 0.4270843841998588,
      "grad_norm": 1.617880856258499,
      "learning_rate": 6.40430922339482e-06,
      "loss": 0.9032,
      "step": 5821
    },
    {
      "epoch": 0.4271577537899979,
      "grad_norm": 1.3434781828784068,
      "learning_rate": 6.4031688138885784e-06,
      "loss": 0.6875,
      "step": 5822
    },
    {
      "epoch": 0.427231123380137,
      "grad_norm": 0.557367160409066,
      "learning_rate": 6.402028325141918e-06,
      "loss": 0.5065,
      "step": 5823
    },
    {
      "epoch": 0.4273044929702761,
      "grad_norm": 0.5450994498508736,
      "learning_rate": 6.4008877572192455e-06,
      "loss": 0.5061,
      "step": 5824
    },
    {
      "epoch": 0.4273044929702761,
      "lm_loss": 0.21,
      "step": 5824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4273044929702761,
      "lm_loss": 0.5462,
      "step": 5824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4273044929702761,
      "lm_loss": 0.5015,
      "step": 5824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4273044929702761,
      "lm_loss": 0.4584,
      "step": 5824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4273044929702761,
      "lm_loss": 0.5598,
      "step": 5824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4273044929702761,
      "lm_loss": 0.4626,
      "step": 5824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4273044929702761,
      "lm_loss": 0.5042,
      "step": 5824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4273044929702761,
      "lm_loss": 0.5049,
      "step": 5824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.42737786256041527,
      "grad_norm": 0.548362129997532,
      "learning_rate": 6.399747110184973e-06,
      "loss": 0.4921,
      "step": 5825
    },
    {
      "epoch": 0.4274512321505544,
      "grad_norm": 1.5734701006245382,
      "learning_rate": 6.398606384103514e-06,
      "loss": 0.7562,
      "step": 5826
    },
    {
      "epoch": 0.4275246017406935,
      "grad_norm": 1.5650637716923872,
      "learning_rate": 6.3974655790392895e-06,
      "loss": 0.8328,
      "step": 5827
    },
    {
      "epoch": 0.42759797133083266,
      "grad_norm": 0.536184709882291,
      "learning_rate": 6.396324695056723e-06,
      "loss": 0.491,
      "step": 5828
    },
    {
      "epoch": 0.42767134092097175,
      "grad_norm": 0.5498677089399534,
      "learning_rate": 6.395183732220242e-06,
      "loss": 0.5212,
      "step": 5829
    },
    {
      "epoch": 0.4277447105111109,
      "grad_norm": 1.5604923725277036,
      "learning_rate": 6.394042690594282e-06,
      "loss": 0.6786,
      "step": 5830
    },
    {
      "epoch": 0.42781808010125005,
      "grad_norm": 1.5047899391897384,
      "learning_rate": 6.392901570243279e-06,
      "loss": 0.6597,
      "step": 5831
    },
    {
      "epoch": 0.42789144969138915,
      "grad_norm": 1.413538323017969,
      "learning_rate": 6.3917603712316735e-06,
      "loss": 0.8724,
      "step": 5832
    },
    {
      "epoch": 0.42789144969138915,
      "lm_loss": 1.3591,
      "step": 5832,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.42789144969138915,
      "lm_loss": 0.8603,
      "step": 5832,
      "vm_loss": 0.1398
    },
    {
      "epoch": 0.42789144969138915,
      "lm_loss": 0.1646,
      "step": 5832,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.42789144969138915,
      "lm_loss": 0.8011,
      "step": 5832,
      "vm_loss": 0.1845
    },
    {
      "epoch": 0.42789144969138915,
      "lm_loss": 0.6599,
      "step": 5832,
      "vm_loss": 0.1848
    },
    {
      "epoch": 0.42789144969138915,
      "lm_loss": 1.1227,
      "step": 5832,
      "vm_loss": 0.1218
    },
    {
      "epoch": 0.42789144969138915,
      "lm_loss": 0.1484,
      "step": 5832,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.42789144969138915,
      "lm_loss": 1.002,
      "step": 5832,
      "vm_loss": 0.0279
    },
    {
      "epoch": 0.4279648192815283,
      "grad_norm": 1.5858485898196457,
      "learning_rate": 6.3906190936239134e-06,
      "loss": 0.7154,
      "step": 5833
    },
    {
      "epoch": 0.4280381888716674,
      "grad_norm": 1.5502177576134162,
      "learning_rate": 6.38947773748445e-06,
      "loss": 0.7173,
      "step": 5834
    },
    {
      "epoch": 0.42811155846180654,
      "grad_norm": 1.6590673850655306,
      "learning_rate": 6.388336302877738e-06,
      "loss": 0.9255,
      "step": 5835
    },
    {
      "epoch": 0.4281849280519457,
      "grad_norm": 0.6333226261887737,
      "learning_rate": 6.387194789868236e-06,
      "loss": 0.5198,
      "step": 5836
    },
    {
      "epoch": 0.4282582976420848,
      "grad_norm": 0.6315632158872286,
      "learning_rate": 6.386053198520409e-06,
      "loss": 0.513,
      "step": 5837
    },
    {
      "epoch": 0.42833166723222393,
      "grad_norm": 0.588283669064189,
      "learning_rate": 6.384911528898725e-06,
      "loss": 0.5081,
      "step": 5838
    },
    {
      "epoch": 0.42840503682236303,
      "grad_norm": 1.649022156933939,
      "learning_rate": 6.3837697810676595e-06,
      "loss": 0.7497,
      "step": 5839
    },
    {
      "epoch": 0.4284784064125022,
      "grad_norm": 1.4604363359654833,
      "learning_rate": 6.382627955091685e-06,
      "loss": 0.8037,
      "step": 5840
    },
    {
      "epoch": 0.4284784064125022,
      "lm_loss": 1.3345,
      "step": 5840,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.4284784064125022,
      "lm_loss": 0.368,
      "step": 5840,
      "vm_loss": 0.1368
    },
    {
      "epoch": 0.4284784064125022,
      "lm_loss": 1.2551,
      "step": 5840,
      "vm_loss": 0.0648
    },
    {
      "epoch": 0.4284784064125022,
      "lm_loss": 1.2554,
      "step": 5840,
      "vm_loss": 0.0756
    },
    {
      "epoch": 0.4284784064125022,
      "lm_loss": 0.288,
      "step": 5840,
      "vm_loss": 0.1234
    },
    {
      "epoch": 0.4284784064125022,
      "lm_loss": 0.7413,
      "step": 5840,
      "vm_loss": 0.1059
    },
    {
      "epoch": 0.4284784064125022,
      "lm_loss": 0.1631,
      "step": 5840,
      "vm_loss": 0.0954
    },
    {
      "epoch": 0.4284784064125022,
      "lm_loss": 0.7669,
      "step": 5840,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.4285517760026413,
      "grad_norm": 1.5370888509599556,
      "learning_rate": 6.3814860510352875e-06,
      "loss": 0.7751,
      "step": 5841
    },
    {
      "epoch": 0.4286251455927804,
      "grad_norm": 1.4959433609283037,
      "learning_rate": 6.380344068962952e-06,
      "loss": 0.7696,
      "step": 5842
    },
    {
      "epoch": 0.42869851518291957,
      "grad_norm": 1.4214739504821747,
      "learning_rate": 6.379202008939166e-06,
      "loss": 0.8376,
      "step": 5843
    },
    {
      "epoch": 0.42877188477305866,
      "grad_norm": 1.3291892742020668,
      "learning_rate": 6.378059871028428e-06,
      "loss": 0.7402,
      "step": 5844
    },
    {
      "epoch": 0.4288452543631978,
      "grad_norm": 1.3578193071286069,
      "learning_rate": 6.376917655295236e-06,
      "loss": 0.7309,
      "step": 5845
    },
    {
      "epoch": 0.42891862395333696,
      "grad_norm": 1.4222385464180163,
      "learning_rate": 6.375775361804096e-06,
      "loss": 0.7647,
      "step": 5846
    },
    {
      "epoch": 0.42899199354347606,
      "grad_norm": 1.9076417501019811,
      "learning_rate": 6.3746329906195135e-06,
      "loss": 0.8866,
      "step": 5847
    },
    {
      "epoch": 0.4290653631336152,
      "grad_norm": 1.5234578647409562,
      "learning_rate": 6.373490541806001e-06,
      "loss": 0.8532,
      "step": 5848
    },
    {
      "epoch": 0.4290653631336152,
      "lm_loss": 1.3267,
      "step": 5848,
      "vm_loss": 0.0358
    },
    {
      "epoch": 0.4290653631336152,
      "lm_loss": 0.3958,
      "step": 5848,
      "vm_loss": 0.1416
    },
    {
      "epoch": 0.4290653631336152,
      "lm_loss": 0.8944,
      "step": 5848,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.4290653631336152,
      "lm_loss": 0.339,
      "step": 5848,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.4290653631336152,
      "lm_loss": 0.6982,
      "step": 5848,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.4290653631336152,
      "lm_loss": 0.1028,
      "step": 5848,
      "vm_loss": 0.0562
    },
    {
      "epoch": 0.4290653631336152,
      "lm_loss": 0.0281,
      "step": 5848,
      "vm_loss": 0.11
    },
    {
      "epoch": 0.4290653631336152,
      "lm_loss": 0.4557,
      "step": 5848,
      "vm_loss": 0.0562
    },
    {
      "epoch": 0.4291387327237543,
      "grad_norm": 1.491735141367403,
      "learning_rate": 6.372348015428077e-06,
      "loss": 0.7873,
      "step": 5849
    },
    {
      "epoch": 0.42921210231389345,
      "grad_norm": 1.4238192846001982,
      "learning_rate": 6.371205411550262e-06,
      "loss": 0.8043,
      "step": 5850
    },
    {
      "epoch": 0.4292854719040326,
      "grad_norm": 1.4587531557082956,
      "learning_rate": 6.370062730237081e-06,
      "loss": 0.7642,
      "step": 5851
    },
    {
      "epoch": 0.4293588414941717,
      "grad_norm": 1.434070990831217,
      "learning_rate": 6.368919971553066e-06,
      "loss": 0.7409,
      "step": 5852
    },
    {
      "epoch": 0.42943221108431084,
      "grad_norm": 1.8492559768430776,
      "learning_rate": 6.367777135562748e-06,
      "loss": 0.7703,
      "step": 5853
    },
    {
      "epoch": 0.42950558067444994,
      "grad_norm": 1.7329908108536176,
      "learning_rate": 6.36663422233067e-06,
      "loss": 0.8279,
      "step": 5854
    },
    {
      "epoch": 0.4295789502645891,
      "grad_norm": 1.4513292199691785,
      "learning_rate": 6.365491231921373e-06,
      "loss": 0.79,
      "step": 5855
    },
    {
      "epoch": 0.42965231985472824,
      "grad_norm": 1.568301342993142,
      "learning_rate": 6.364348164399403e-06,
      "loss": 0.7863,
      "step": 5856
    },
    {
      "epoch": 0.42965231985472824,
      "lm_loss": 1.1852,
      "step": 5856,
      "vm_loss": 0.0853
    },
    {
      "epoch": 0.42965231985472824,
      "lm_loss": 0.1193,
      "step": 5856,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.42965231985472824,
      "lm_loss": 0.3518,
      "step": 5856,
      "vm_loss": 0.0468
    },
    {
      "epoch": 0.42965231985472824,
      "lm_loss": 0.36,
      "step": 5856,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.42965231985472824,
      "lm_loss": 0.9317,
      "step": 5856,
      "vm_loss": 0.1048
    },
    {
      "epoch": 0.42965231985472824,
      "lm_loss": 0.3208,
      "step": 5856,
      "vm_loss": 0.1408
    },
    {
      "epoch": 0.42965231985472824,
      "lm_loss": 0.4261,
      "step": 5856,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.42965231985472824,
      "lm_loss": 1.2399,
      "step": 5856,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.42972568944486733,
      "grad_norm": 1.6445514446700562,
      "learning_rate": 6.363205019829315e-06,
      "loss": 0.7667,
      "step": 5857
    },
    {
      "epoch": 0.4297990590350065,
      "grad_norm": 1.388243689443935,
      "learning_rate": 6.362061798275663e-06,
      "loss": 0.758,
      "step": 5858
    },
    {
      "epoch": 0.4298724286251456,
      "grad_norm": 0.971273244179852,
      "learning_rate": 6.360918499803008e-06,
      "loss": 0.503,
      "step": 5859
    },
    {
      "epoch": 0.4299457982152847,
      "grad_norm": 1.734578472656531,
      "learning_rate": 6.359775124475915e-06,
      "loss": 0.8161,
      "step": 5860
    },
    {
      "epoch": 0.4300191678054239,
      "grad_norm": 1.6083403862322636,
      "learning_rate": 6.358631672358952e-06,
      "loss": 0.7423,
      "step": 5861
    },
    {
      "epoch": 0.43009253739556297,
      "grad_norm": 1.4986218764581303,
      "learning_rate": 6.3574881435166964e-06,
      "loss": 0.7806,
      "step": 5862
    },
    {
      "epoch": 0.4301659069857021,
      "grad_norm": 1.6380296650275445,
      "learning_rate": 6.356344538013724e-06,
      "loss": 0.7293,
      "step": 5863
    },
    {
      "epoch": 0.4302392765758412,
      "grad_norm": 1.615345347225004,
      "learning_rate": 6.355200855914613e-06,
      "loss": 0.7982,
      "step": 5864
    },
    {
      "epoch": 0.4302392765758412,
      "lm_loss": 0.7495,
      "step": 5864,
      "vm_loss": 0.0976
    },
    {
      "epoch": 0.4302392765758412,
      "lm_loss": 0.3842,
      "step": 5864,
      "vm_loss": 0.1581
    },
    {
      "epoch": 0.4302392765758412,
      "lm_loss": 1.1007,
      "step": 5864,
      "vm_loss": 0.1471
    },
    {
      "epoch": 0.4302392765758412,
      "lm_loss": 0.2206,
      "step": 5864,
      "vm_loss": 0.0958
    },
    {
      "epoch": 0.4302392765758412,
      "lm_loss": 0.7461,
      "step": 5864,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.4302392765758412,
      "lm_loss": 0.8834,
      "step": 5864,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.4302392765758412,
      "lm_loss": 0.839,
      "step": 5864,
      "vm_loss": 0.1722
    },
    {
      "epoch": 0.4302392765758412,
      "lm_loss": 0.1215,
      "step": 5864,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.43031264616598036,
      "grad_norm": 1.4409023286424916,
      "learning_rate": 6.354057097283956e-06,
      "loss": 0.758,
      "step": 5865
    },
    {
      "epoch": 0.43038601575611946,
      "grad_norm": 1.680192387109628,
      "learning_rate": 6.3529132621863394e-06,
      "loss": 0.8258,
      "step": 5866
    },
    {
      "epoch": 0.4304593853462586,
      "grad_norm": 1.6114062956849535,
      "learning_rate": 6.351769350686363e-06,
      "loss": 0.8081,
      "step": 5867
    },
    {
      "epoch": 0.43053275493639775,
      "grad_norm": 1.7128319110001105,
      "learning_rate": 6.350625362848624e-06,
      "loss": 0.8563,
      "step": 5868
    },
    {
      "epoch": 0.43060612452653685,
      "grad_norm": 0.7681492932804156,
      "learning_rate": 6.349481298737723e-06,
      "loss": 0.5002,
      "step": 5869
    },
    {
      "epoch": 0.430679494116676,
      "grad_norm": 0.6749807448732492,
      "learning_rate": 6.348337158418274e-06,
      "loss": 0.511,
      "step": 5870
    },
    {
      "epoch": 0.4307528637068151,
      "grad_norm": 0.5918763429731236,
      "learning_rate": 6.347192941954886e-06,
      "loss": 0.4969,
      "step": 5871
    },
    {
      "epoch": 0.43082623329695424,
      "grad_norm": 1.609844098134766,
      "learning_rate": 6.346048649412175e-06,
      "loss": 0.8456,
      "step": 5872
    },
    {
      "epoch": 0.43082623329695424,
      "lm_loss": 0.1948,
      "step": 5872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43082623329695424,
      "lm_loss": 0.5821,
      "step": 5872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43082623329695424,
      "lm_loss": 0.5838,
      "step": 5872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43082623329695424,
      "lm_loss": 0.4689,
      "step": 5872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43082623329695424,
      "lm_loss": 0.357,
      "step": 5872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43082623329695424,
      "lm_loss": 0.3959,
      "step": 5872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43082623329695424,
      "lm_loss": 0.4442,
      "step": 5872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43082623329695424,
      "lm_loss": 0.5947,
      "step": 5872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4308996028870934,
      "grad_norm": 0.6009972397500709,
      "learning_rate": 6.344904280854764e-06,
      "loss": 0.5026,
      "step": 5873
    },
    {
      "epoch": 0.4309729724772325,
      "grad_norm": 1.725152769188187,
      "learning_rate": 6.343759836347277e-06,
      "loss": 0.7421,
      "step": 5874
    },
    {
      "epoch": 0.43104634206737164,
      "grad_norm": 1.7096729273112337,
      "learning_rate": 6.342615315954346e-06,
      "loss": 0.8237,
      "step": 5875
    },
    {
      "epoch": 0.43111971165751073,
      "grad_norm": 1.6805530323567381,
      "learning_rate": 6.341470719740604e-06,
      "loss": 0.7402,
      "step": 5876
    },
    {
      "epoch": 0.4311930812476499,
      "grad_norm": 1.9197963073506938,
      "learning_rate": 6.340326047770684e-06,
      "loss": 0.7635,
      "step": 5877
    },
    {
      "epoch": 0.43126645083778903,
      "grad_norm": 1.3756315695418608,
      "learning_rate": 6.3391813001092365e-06,
      "loss": 0.816,
      "step": 5878
    },
    {
      "epoch": 0.4313398204279281,
      "grad_norm": 0.7905762362367066,
      "learning_rate": 6.3380364768209036e-06,
      "loss": 0.4992,
      "step": 5879
    },
    {
      "epoch": 0.43141319001806727,
      "grad_norm": 1.6217491045782282,
      "learning_rate": 6.336891577970337e-06,
      "loss": 0.808,
      "step": 5880
    },
    {
      "epoch": 0.43141319001806727,
      "lm_loss": 1.3153,
      "step": 5880,
      "vm_loss": 0.0779
    },
    {
      "epoch": 0.43141319001806727,
      "lm_loss": 0.0117,
      "step": 5880,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.43141319001806727,
      "lm_loss": 0.1957,
      "step": 5880,
      "vm_loss": 0.1465
    },
    {
      "epoch": 0.43141319001806727,
      "lm_loss": 0.8322,
      "step": 5880,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.43141319001806727,
      "lm_loss": 0.0903,
      "step": 5880,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.43141319001806727,
      "lm_loss": 0.0602,
      "step": 5880,
      "vm_loss": 0.0513
    },
    {
      "epoch": 0.43141319001806727,
      "lm_loss": 0.2599,
      "step": 5880,
      "vm_loss": 0.0946
    },
    {
      "epoch": 0.43141319001806727,
      "lm_loss": 0.8104,
      "step": 5880,
      "vm_loss": 0.1764
    },
    {
      "epoch": 0.43148655960820637,
      "grad_norm": 1.504675393457046,
      "learning_rate": 6.335746603622192e-06,
      "loss": 0.6871,
      "step": 5881
    },
    {
      "epoch": 0.4315599291983455,
      "grad_norm": 1.595644573422536,
      "learning_rate": 6.334601553841129e-06,
      "loss": 0.7924,
      "step": 5882
    },
    {
      "epoch": 0.43163329878848467,
      "grad_norm": 1.6390083870521748,
      "learning_rate": 6.333456428691811e-06,
      "loss": 0.7885,
      "step": 5883
    },
    {
      "epoch": 0.43170666837862376,
      "grad_norm": 0.5760208520742024,
      "learning_rate": 6.3323112282389065e-06,
      "loss": 0.4996,
      "step": 5884
    },
    {
      "epoch": 0.4317800379687629,
      "grad_norm": 0.5420178719117744,
      "learning_rate": 6.331165952547087e-06,
      "loss": 0.4988,
      "step": 5885
    },
    {
      "epoch": 0.431853407558902,
      "grad_norm": 1.5019276732875342,
      "learning_rate": 6.33002060168103e-06,
      "loss": 0.744,
      "step": 5886
    },
    {
      "epoch": 0.43192677714904115,
      "grad_norm": 1.585304976008122,
      "learning_rate": 6.328875175705416e-06,
      "loss": 0.8586,
      "step": 5887
    },
    {
      "epoch": 0.4320001467391803,
      "grad_norm": 0.5772662231142824,
      "learning_rate": 6.327729674684928e-06,
      "loss": 0.5118,
      "step": 5888
    },
    {
      "epoch": 0.4320001467391803,
      "lm_loss": 0.9554,
      "step": 5888,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.4320001467391803,
      "lm_loss": 1.4803,
      "step": 5888,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.4320001467391803,
      "lm_loss": 0.2015,
      "step": 5888,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.4320001467391803,
      "lm_loss": 0.8152,
      "step": 5888,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.4320001467391803,
      "lm_loss": 0.2045,
      "step": 5888,
      "vm_loss": 0.1342
    },
    {
      "epoch": 0.4320001467391803,
      "lm_loss": 0.1972,
      "step": 5888,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.4320001467391803,
      "lm_loss": 0.9455,
      "step": 5888,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.4320001467391803,
      "lm_loss": 1.2749,
      "step": 5888,
      "vm_loss": 0.0903
    },
    {
      "epoch": 0.4320735163293194,
      "grad_norm": 1.6038930960046298,
      "learning_rate": 6.32658409868426e-06,
      "loss": 0.9257,
      "step": 5889
    },
    {
      "epoch": 0.43214688591945855,
      "grad_norm": 0.5711509017322176,
      "learning_rate": 6.325438447768102e-06,
      "loss": 0.5094,
      "step": 5890
    },
    {
      "epoch": 0.43222025550959764,
      "grad_norm": 1.4588267673071496,
      "learning_rate": 6.3242927220011505e-06,
      "loss": 0.6875,
      "step": 5891
    },
    {
      "epoch": 0.4322936250997368,
      "grad_norm": 1.4755001395743115,
      "learning_rate": 6.323146921448111e-06,
      "loss": 0.8392,
      "step": 5892
    },
    {
      "epoch": 0.43236699468987594,
      "grad_norm": 0.5873378780153479,
      "learning_rate": 6.322001046173688e-06,
      "loss": 0.5241,
      "step": 5893
    },
    {
      "epoch": 0.43244036428001503,
      "grad_norm": 1.7543243234213142,
      "learning_rate": 6.320855096242592e-06,
      "loss": 0.8076,
      "step": 5894
    },
    {
      "epoch": 0.4325137338701542,
      "grad_norm": 1.7183326373050962,
      "learning_rate": 6.319709071719539e-06,
      "loss": 0.829,
      "step": 5895
    },
    {
      "epoch": 0.4325871034602933,
      "grad_norm": 1.498361850375941,
      "learning_rate": 6.318562972669244e-06,
      "loss": 0.87,
      "step": 5896
    },
    {
      "epoch": 0.4325871034602933,
      "lm_loss": 1.3831,
      "step": 5896,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.4325871034602933,
      "lm_loss": 0.6835,
      "step": 5896,
      "vm_loss": 0.0768
    },
    {
      "epoch": 0.4325871034602933,
      "lm_loss": 0.0575,
      "step": 5896,
      "vm_loss": 0.1146
    },
    {
      "epoch": 0.4325871034602933,
      "lm_loss": 0.0222,
      "step": 5896,
      "vm_loss": 0.1458
    },
    {
      "epoch": 0.4325871034602933,
      "lm_loss": 0.9635,
      "step": 5896,
      "vm_loss": 0.1165
    },
    {
      "epoch": 0.4325871034602933,
      "lm_loss": 0.0466,
      "step": 5896,
      "vm_loss": 0.0733
    },
    {
      "epoch": 0.4325871034602933,
      "lm_loss": 0.8387,
      "step": 5896,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.4325871034602933,
      "lm_loss": 0.6126,
      "step": 5896,
      "vm_loss": 0.1286
    },
    {
      "epoch": 0.4326604730504324,
      "grad_norm": 1.4857197446546542,
      "learning_rate": 6.317416799156434e-06,
      "loss": 0.7484,
      "step": 5897
    },
    {
      "epoch": 0.4327338426405716,
      "grad_norm": 0.5783277161872306,
      "learning_rate": 6.316270551245836e-06,
      "loss": 0.4992,
      "step": 5898
    },
    {
      "epoch": 0.43280721223071067,
      "grad_norm": 0.5832128616182517,
      "learning_rate": 6.315124229002177e-06,
      "loss": 0.5023,
      "step": 5899
    },
    {
      "epoch": 0.4328805818208498,
      "grad_norm": 1.4011166261620622,
      "learning_rate": 6.313977832490199e-06,
      "loss": 0.8061,
      "step": 5900
    },
    {
      "epoch": 0.4329539514109889,
      "grad_norm": 0.5529058796533428,
      "learning_rate": 6.312831361774636e-06,
      "loss": 0.4987,
      "step": 5901
    },
    {
      "epoch": 0.43302732100112806,
      "grad_norm": 0.5501790859445177,
      "learning_rate": 6.311684816920237e-06,
      "loss": 0.5135,
      "step": 5902
    },
    {
      "epoch": 0.4331006905912672,
      "grad_norm": 1.771149006391132,
      "learning_rate": 6.310538197991748e-06,
      "loss": 0.8626,
      "step": 5903
    },
    {
      "epoch": 0.4331740601814063,
      "grad_norm": 1.4919981707460739,
      "learning_rate": 6.309391505053921e-06,
      "loss": 0.717,
      "step": 5904
    },
    {
      "epoch": 0.4331740601814063,
      "lm_loss": 1.0607,
      "step": 5904,
      "vm_loss": 0.0677
    },
    {
      "epoch": 0.4331740601814063,
      "lm_loss": 1.4084,
      "step": 5904,
      "vm_loss": 0.0838
    },
    {
      "epoch": 0.4331740601814063,
      "lm_loss": 1.0612,
      "step": 5904,
      "vm_loss": 0.2121
    },
    {
      "epoch": 0.4331740601814063,
      "lm_loss": 1.0034,
      "step": 5904,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.4331740601814063,
      "lm_loss": 0.0772,
      "step": 5904,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.4331740601814063,
      "lm_loss": 0.1976,
      "step": 5904,
      "vm_loss": 0.1527
    },
    {
      "epoch": 0.4331740601814063,
      "lm_loss": 0.0019,
      "step": 5904,
      "vm_loss": 0.0603
    },
    {
      "epoch": 0.4331740601814063,
      "lm_loss": 0.1051,
      "step": 5904,
      "vm_loss": 0.1157
    },
    {
      "epoch": 0.43324742977154546,
      "grad_norm": 1.439341927446661,
      "learning_rate": 6.308244738171513e-06,
      "loss": 0.7692,
      "step": 5905
    },
    {
      "epoch": 0.43332079936168455,
      "grad_norm": 1.4573660522689154,
      "learning_rate": 6.307097897409286e-06,
      "loss": 0.7086,
      "step": 5906
    },
    {
      "epoch": 0.4333941689518237,
      "grad_norm": 1.5094458793864158,
      "learning_rate": 6.305950982832001e-06,
      "loss": 0.7683,
      "step": 5907
    },
    {
      "epoch": 0.4334675385419628,
      "grad_norm": 1.4984677305719134,
      "learning_rate": 6.304803994504432e-06,
      "loss": 0.7906,
      "step": 5908
    },
    {
      "epoch": 0.43354090813210194,
      "grad_norm": 1.4876771550092964,
      "learning_rate": 6.303656932491349e-06,
      "loss": 0.7126,
      "step": 5909
    },
    {
      "epoch": 0.4336142777222411,
      "grad_norm": 0.6492191503611447,
      "learning_rate": 6.3025097968575325e-06,
      "loss": 0.5242,
      "step": 5910
    },
    {
      "epoch": 0.4336876473123802,
      "grad_norm": 1.3527394586245611,
      "learning_rate": 6.301362587667761e-06,
      "loss": 0.6999,
      "step": 5911
    },
    {
      "epoch": 0.43376101690251934,
      "grad_norm": 1.6347910168570683,
      "learning_rate": 6.3002153049868195e-06,
      "loss": 0.7049,
      "step": 5912
    },
    {
      "epoch": 0.43376101690251934,
      "lm_loss": 0.7615,
      "step": 5912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43376101690251934,
      "lm_loss": 0.5211,
      "step": 5912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43376101690251934,
      "lm_loss": 0.4855,
      "step": 5912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43376101690251934,
      "lm_loss": 0.3548,
      "step": 5912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43376101690251934,
      "lm_loss": 0.541,
      "step": 5912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43376101690251934,
      "lm_loss": 0.4562,
      "step": 5912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43376101690251934,
      "lm_loss": 0.4474,
      "step": 5912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43376101690251934,
      "lm_loss": 0.6115,
      "step": 5912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43383438649265843,
      "grad_norm": 0.5925598876051045,
      "learning_rate": 6.299067948879503e-06,
      "loss": 0.5054,
      "step": 5913
    },
    {
      "epoch": 0.4339077560827976,
      "grad_norm": 1.3905535446545227,
      "learning_rate": 6.297920519410601e-06,
      "loss": 0.826,
      "step": 5914
    },
    {
      "epoch": 0.43398112567293673,
      "grad_norm": 1.3899078642521916,
      "learning_rate": 6.296773016644912e-06,
      "loss": 0.7337,
      "step": 5915
    },
    {
      "epoch": 0.4340544952630758,
      "grad_norm": 1.342645939474354,
      "learning_rate": 6.29562544064724e-06,
      "loss": 0.7349,
      "step": 5916
    },
    {
      "epoch": 0.434127864853215,
      "grad_norm": 0.5726692849050897,
      "learning_rate": 6.294477791482392e-06,
      "loss": 0.4978,
      "step": 5917
    },
    {
      "epoch": 0.43420123444335407,
      "grad_norm": 0.5867459067581882,
      "learning_rate": 6.293330069215175e-06,
      "loss": 0.5129,
      "step": 5918
    },
    {
      "epoch": 0.4342746040334932,
      "grad_norm": 1.588925859944895,
      "learning_rate": 6.292182273910409e-06,
      "loss": 0.8311,
      "step": 5919
    },
    {
      "epoch": 0.43434797362363237,
      "grad_norm": 1.4772539656602413,
      "learning_rate": 6.291034405632908e-06,
      "loss": 0.754,
      "step": 5920
    },
    {
      "epoch": 0.43434797362363237,
      "lm_loss": 1.4631,
      "step": 5920,
      "vm_loss": 0.0345
    },
    {
      "epoch": 0.43434797362363237,
      "lm_loss": 0.0892,
      "step": 5920,
      "vm_loss": 0.0483
    },
    {
      "epoch": 0.43434797362363237,
      "lm_loss": 1.0222,
      "step": 5920,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.43434797362363237,
      "lm_loss": 0.5706,
      "step": 5920,
      "vm_loss": 0.1281
    },
    {
      "epoch": 0.43434797362363237,
      "lm_loss": 0.1012,
      "step": 5920,
      "vm_loss": 0.0528
    },
    {
      "epoch": 0.43434797362363237,
      "lm_loss": 0.4866,
      "step": 5920,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.43434797362363237,
      "lm_loss": 0.5699,
      "step": 5920,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.43434797362363237,
      "lm_loss": 0.2215,
      "step": 5920,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.43442134321377146,
      "grad_norm": 1.6404190013588273,
      "learning_rate": 6.289886464447497e-06,
      "loss": 0.6761,
      "step": 5921
    },
    {
      "epoch": 0.4344947128039106,
      "grad_norm": 1.6040113862862628,
      "learning_rate": 6.288738450419004e-06,
      "loss": 0.7514,
      "step": 5922
    },
    {
      "epoch": 0.4345680823940497,
      "grad_norm": 1.872454584819234,
      "learning_rate": 6.287590363612258e-06,
      "loss": 0.8872,
      "step": 5923
    },
    {
      "epoch": 0.43464145198418885,
      "grad_norm": 1.5454981886067551,
      "learning_rate": 6.286442204092097e-06,
      "loss": 0.8142,
      "step": 5924
    },
    {
      "epoch": 0.434714821574328,
      "grad_norm": 0.6022966700069162,
      "learning_rate": 6.285293971923358e-06,
      "loss": 0.5231,
      "step": 5925
    },
    {
      "epoch": 0.4347881911644671,
      "grad_norm": 1.683782751282965,
      "learning_rate": 6.284145667170886e-06,
      "loss": 0.8166,
      "step": 5926
    },
    {
      "epoch": 0.43486156075460625,
      "grad_norm": 1.396780880261043,
      "learning_rate": 6.2829972898995274e-06,
      "loss": 0.7979,
      "step": 5927
    },
    {
      "epoch": 0.43493493034474534,
      "grad_norm": 0.5691358491650914,
      "learning_rate": 6.281848840174134e-06,
      "loss": 0.5186,
      "step": 5928
    },
    {
      "epoch": 0.43493493034474534,
      "lm_loss": 1.4347,
      "step": 5928,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.43493493034474534,
      "lm_loss": 0.8425,
      "step": 5928,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.43493493034474534,
      "lm_loss": 0.3069,
      "step": 5928,
      "vm_loss": 0.061
    },
    {
      "epoch": 0.43493493034474534,
      "lm_loss": 0.7018,
      "step": 5928,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.43493493034474534,
      "lm_loss": 1.106,
      "step": 5928,
      "vm_loss": 0.1151
    },
    {
      "epoch": 0.43493493034474534,
      "lm_loss": 1.02,
      "step": 5928,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.43493493034474534,
      "lm_loss": 0.0932,
      "step": 5928,
      "vm_loss": 0.1218
    },
    {
      "epoch": 0.43493493034474534,
      "lm_loss": 0.3361,
      "step": 5928,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.4350082999348845,
      "grad_norm": 1.5804900494017409,
      "learning_rate": 6.280700318059563e-06,
      "loss": 0.8145,
      "step": 5929
    },
    {
      "epoch": 0.43508166952502364,
      "grad_norm": 1.4213559450629782,
      "learning_rate": 6.279551723620674e-06,
      "loss": 0.7309,
      "step": 5930
    },
    {
      "epoch": 0.43515503911516273,
      "grad_norm": 1.3675271409927505,
      "learning_rate": 6.27840305692233e-06,
      "loss": 0.8129,
      "step": 5931
    },
    {
      "epoch": 0.4352284087053019,
      "grad_norm": 1.6291550179597072,
      "learning_rate": 6.2772543180294e-06,
      "loss": 0.7538,
      "step": 5932
    },
    {
      "epoch": 0.435301778295441,
      "grad_norm": 0.5688324799462386,
      "learning_rate": 6.276105507006753e-06,
      "loss": 0.4964,
      "step": 5933
    },
    {
      "epoch": 0.4353751478855801,
      "grad_norm": 0.5687835748105751,
      "learning_rate": 6.27495662391927e-06,
      "loss": 0.5168,
      "step": 5934
    },
    {
      "epoch": 0.4354485174757193,
      "grad_norm": 1.4547209905559884,
      "learning_rate": 6.27380766883183e-06,
      "loss": 0.8383,
      "step": 5935
    },
    {
      "epoch": 0.43552188706585837,
      "grad_norm": 0.54493033464031,
      "learning_rate": 6.272658641809314e-06,
      "loss": 0.5111,
      "step": 5936
    },
    {
      "epoch": 0.43552188706585837,
      "lm_loss": 1.2932,
      "step": 5936,
      "vm_loss": 0.0517
    },
    {
      "epoch": 0.43552188706585837,
      "lm_loss": 0.0201,
      "step": 5936,
      "vm_loss": 0.0815
    },
    {
      "epoch": 0.43552188706585837,
      "lm_loss": 0.784,
      "step": 5936,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.43552188706585837,
      "lm_loss": 0.2615,
      "step": 5936,
      "vm_loss": 0.1051
    },
    {
      "epoch": 0.43552188706585837,
      "lm_loss": 0.0755,
      "step": 5936,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.43552188706585837,
      "lm_loss": 0.1509,
      "step": 5936,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.43552188706585837,
      "lm_loss": 0.1543,
      "step": 5936,
      "vm_loss": 0.0557
    },
    {
      "epoch": 0.43552188706585837,
      "lm_loss": 0.1137,
      "step": 5936,
      "vm_loss": 0.1505
    },
    {
      "epoch": 0.4355952566559975,
      "grad_norm": 1.5793029344739888,
      "learning_rate": 6.271509542916614e-06,
      "loss": 0.677,
      "step": 5937
    },
    {
      "epoch": 0.4356686262461366,
      "grad_norm": 1.4001774985480975,
      "learning_rate": 6.270360372218621e-06,
      "loss": 0.7398,
      "step": 5938
    },
    {
      "epoch": 0.43574199583627576,
      "grad_norm": 1.5219084849918407,
      "learning_rate": 6.269211129780232e-06,
      "loss": 0.7862,
      "step": 5939
    },
    {
      "epoch": 0.4358153654264149,
      "grad_norm": 1.5587651842233685,
      "learning_rate": 6.2680618156663475e-06,
      "loss": 0.815,
      "step": 5940
    },
    {
      "epoch": 0.435888735016554,
      "grad_norm": 1.5576034776743943,
      "learning_rate": 6.266912429941871e-06,
      "loss": 0.849,
      "step": 5941
    },
    {
      "epoch": 0.43596210460669316,
      "grad_norm": 1.4629387273069436,
      "learning_rate": 6.2657629726717125e-06,
      "loss": 0.7289,
      "step": 5942
    },
    {
      "epoch": 0.43603547419683225,
      "grad_norm": 1.4243715088111748,
      "learning_rate": 6.2646134439207865e-06,
      "loss": 0.8695,
      "step": 5943
    },
    {
      "epoch": 0.4361088437869714,
      "grad_norm": 1.391477843126629,
      "learning_rate": 6.263463843754004e-06,
      "loss": 0.7674,
      "step": 5944
    },
    {
      "epoch": 0.4361088437869714,
      "lm_loss": 0.468,
      "step": 5944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4361088437869714,
      "lm_loss": 0.5415,
      "step": 5944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4361088437869714,
      "lm_loss": 0.4016,
      "step": 5944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4361088437869714,
      "lm_loss": 0.5144,
      "step": 5944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4361088437869714,
      "lm_loss": 0.3614,
      "step": 5944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4361088437869714,
      "lm_loss": 0.6908,
      "step": 5944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4361088437869714,
      "lm_loss": 0.5612,
      "step": 5944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4361088437869714,
      "lm_loss": 0.4879,
      "step": 5944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4361822133771105,
      "grad_norm": 0.6187358194903539,
      "learning_rate": 6.262314172236292e-06,
      "loss": 0.5092,
      "step": 5945
    },
    {
      "epoch": 0.43625558296724964,
      "grad_norm": 1.3701444441884347,
      "learning_rate": 6.261164429432571e-06,
      "loss": 0.8048,
      "step": 5946
    },
    {
      "epoch": 0.4363289525573888,
      "grad_norm": 1.65242180034065,
      "learning_rate": 6.260014615407772e-06,
      "loss": 0.8449,
      "step": 5947
    },
    {
      "epoch": 0.4364023221475279,
      "grad_norm": 1.4038912080187815,
      "learning_rate": 6.258864730226828e-06,
      "loss": 0.7815,
      "step": 5948
    },
    {
      "epoch": 0.43647569173766704,
      "grad_norm": 1.6161984974743577,
      "learning_rate": 6.257714773954674e-06,
      "loss": 0.6736,
      "step": 5949
    },
    {
      "epoch": 0.43654906132780613,
      "grad_norm": 1.3470689066313803,
      "learning_rate": 6.2565647466562525e-06,
      "loss": 0.6783,
      "step": 5950
    },
    {
      "epoch": 0.4366224309179453,
      "grad_norm": 1.436883123807872,
      "learning_rate": 6.25541464839651e-06,
      "loss": 0.737,
      "step": 5951
    },
    {
      "epoch": 0.43669580050808443,
      "grad_norm": 1.4502792398781916,
      "learning_rate": 6.25426447924039e-06,
      "loss": 0.8708,
      "step": 5952
    },
    {
      "epoch": 0.43669580050808443,
      "lm_loss": 1.299,
      "step": 5952,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.43669580050808443,
      "lm_loss": 1.392,
      "step": 5952,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.43669580050808443,
      "lm_loss": 0.1778,
      "step": 5952,
      "vm_loss": 0.0899
    },
    {
      "epoch": 0.43669580050808443,
      "lm_loss": 1.0331,
      "step": 5952,
      "vm_loss": 0.2557
    },
    {
      "epoch": 0.43669580050808443,
      "lm_loss": 0.3311,
      "step": 5952,
      "vm_loss": 0.1651
    },
    {
      "epoch": 0.43669580050808443,
      "lm_loss": 1.2101,
      "step": 5952,
      "vm_loss": 0.1685
    },
    {
      "epoch": 0.43669580050808443,
      "lm_loss": 0.1619,
      "step": 5952,
      "vm_loss": 0.0622
    },
    {
      "epoch": 0.43669580050808443,
      "lm_loss": 0.5285,
      "step": 5952,
      "vm_loss": 0.1233
    },
    {
      "epoch": 0.4367691700982235,
      "grad_norm": 1.4364770351931322,
      "learning_rate": 6.2531142392528495e-06,
      "loss": 0.842,
      "step": 5953
    },
    {
      "epoch": 0.4368425396883627,
      "grad_norm": 0.6610300470474259,
      "learning_rate": 6.2519639284988465e-06,
      "loss": 0.5155,
      "step": 5954
    },
    {
      "epoch": 0.43691590927850177,
      "grad_norm": 1.622091918749489,
      "learning_rate": 6.250813547043337e-06,
      "loss": 0.876,
      "step": 5955
    },
    {
      "epoch": 0.4369892788686409,
      "grad_norm": 1.6036627546508277,
      "learning_rate": 6.2496630949512915e-06,
      "loss": 0.7512,
      "step": 5956
    },
    {
      "epoch": 0.43706264845878007,
      "grad_norm": 1.4157645317875578,
      "learning_rate": 6.248512572287675e-06,
      "loss": 0.6401,
      "step": 5957
    },
    {
      "epoch": 0.43713601804891916,
      "grad_norm": 1.311368011881798,
      "learning_rate": 6.247361979117462e-06,
      "loss": 0.7082,
      "step": 5958
    },
    {
      "epoch": 0.4372093876390583,
      "grad_norm": 1.5169659967396063,
      "learning_rate": 6.2462113155056305e-06,
      "loss": 0.7314,
      "step": 5959
    },
    {
      "epoch": 0.4372827572291974,
      "grad_norm": 0.5796611919228519,
      "learning_rate": 6.2450605815171574e-06,
      "loss": 0.4998,
      "step": 5960
    },
    {
      "epoch": 0.4372827572291974,
      "lm_loss": 0.547,
      "step": 5960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4372827572291974,
      "lm_loss": 0.5782,
      "step": 5960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4372827572291974,
      "lm_loss": 0.5919,
      "step": 5960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4372827572291974,
      "lm_loss": 0.3375,
      "step": 5960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4372827572291974,
      "lm_loss": 0.4562,
      "step": 5960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4372827572291974,
      "lm_loss": 0.484,
      "step": 5960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4372827572291974,
      "lm_loss": 0.4957,
      "step": 5960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4372827572291974,
      "lm_loss": 0.4109,
      "step": 5960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43735612681933655,
      "grad_norm": 0.5759523207080246,
      "learning_rate": 6.243909777217031e-06,
      "loss": 0.4951,
      "step": 5961
    },
    {
      "epoch": 0.4374294964094757,
      "grad_norm": 0.5681221363295758,
      "learning_rate": 6.242758902670239e-06,
      "loss": 0.5047,
      "step": 5962
    },
    {
      "epoch": 0.4375028659996148,
      "grad_norm": 2.181643881398939,
      "learning_rate": 6.241607957941774e-06,
      "loss": 0.8053,
      "step": 5963
    },
    {
      "epoch": 0.43757623558975395,
      "grad_norm": 0.535082426702751,
      "learning_rate": 6.240456943096633e-06,
      "loss": 0.4821,
      "step": 5964
    },
    {
      "epoch": 0.43764960517989304,
      "grad_norm": 0.5411110739148124,
      "learning_rate": 6.239305858199816e-06,
      "loss": 0.4908,
      "step": 5965
    },
    {
      "epoch": 0.4377229747700322,
      "grad_norm": 0.5604398720873198,
      "learning_rate": 6.238154703316328e-06,
      "loss": 0.5093,
      "step": 5966
    },
    {
      "epoch": 0.43779634436017134,
      "grad_norm": 0.5530123514428239,
      "learning_rate": 6.237003478511179e-06,
      "loss": 0.4982,
      "step": 5967
    },
    {
      "epoch": 0.43786971395031044,
      "grad_norm": 1.3587508625199993,
      "learning_rate": 6.235852183849379e-06,
      "loss": 0.6844,
      "step": 5968
    },
    {
      "epoch": 0.43786971395031044,
      "lm_loss": 0.5939,
      "step": 5968,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43786971395031044,
      "lm_loss": 0.562,
      "step": 5968,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43786971395031044,
      "lm_loss": 0.4399,
      "step": 5968,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43786971395031044,
      "lm_loss": 0.4717,
      "step": 5968,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43786971395031044,
      "lm_loss": 0.6316,
      "step": 5968,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43786971395031044,
      "lm_loss": 0.3388,
      "step": 5968,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43786971395031044,
      "lm_loss": 0.5202,
      "step": 5968,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43786971395031044,
      "lm_loss": 0.5015,
      "step": 5968,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4379430835404496,
      "grad_norm": 0.5480777962071991,
      "learning_rate": 6.234700819395946e-06,
      "loss": 0.5186,
      "step": 5969
    },
    {
      "epoch": 0.4380164531305887,
      "grad_norm": 1.452036242648023,
      "learning_rate": 6.2335493852159e-06,
      "loss": 0.7688,
      "step": 5970
    },
    {
      "epoch": 0.43808982272072783,
      "grad_norm": 1.5018972796455687,
      "learning_rate": 6.232397881374266e-06,
      "loss": 0.8418,
      "step": 5971
    },
    {
      "epoch": 0.438163192310867,
      "grad_norm": 0.5595166626556645,
      "learning_rate": 6.231246307936071e-06,
      "loss": 0.5249,
      "step": 5972
    },
    {
      "epoch": 0.43823656190100607,
      "grad_norm": 1.4139343963490612,
      "learning_rate": 6.2300946649663485e-06,
      "loss": 0.7871,
      "step": 5973
    },
    {
      "epoch": 0.4383099314911452,
      "grad_norm": 0.5711633762003588,
      "learning_rate": 6.228942952530132e-06,
      "loss": 0.5048,
      "step": 5974
    },
    {
      "epoch": 0.4383833010812843,
      "grad_norm": 0.5700644199711683,
      "learning_rate": 6.227791170692465e-06,
      "loss": 0.5179,
      "step": 5975
    },
    {
      "epoch": 0.43845667067142347,
      "grad_norm": 1.4356324537086786,
      "learning_rate": 6.226639319518389e-06,
      "loss": 0.774,
      "step": 5976
    },
    {
      "epoch": 0.43845667067142347,
      "lm_loss": 1.8578,
      "step": 5976,
      "vm_loss": 0.0422
    },
    {
      "epoch": 0.43845667067142347,
      "lm_loss": 0.0672,
      "step": 5976,
      "vm_loss": 0.1334
    },
    {
      "epoch": 0.43845667067142347,
      "lm_loss": 0.5702,
      "step": 5976,
      "vm_loss": 0.1603
    },
    {
      "epoch": 0.43845667067142347,
      "lm_loss": 0.1912,
      "step": 5976,
      "vm_loss": 0.0801
    },
    {
      "epoch": 0.43845667067142347,
      "lm_loss": 1.2104,
      "step": 5976,
      "vm_loss": 0.0656
    },
    {
      "epoch": 0.43845667067142347,
      "lm_loss": 0.6597,
      "step": 5976,
      "vm_loss": 0.2563
    },
    {
      "epoch": 0.43845667067142347,
      "lm_loss": 0.0316,
      "step": 5976,
      "vm_loss": 0.08
    },
    {
      "epoch": 0.43845667067142347,
      "lm_loss": 0.3989,
      "step": 5976,
      "vm_loss": 0.1455
    },
    {
      "epoch": 0.4385300402615626,
      "grad_norm": 1.4049380227066601,
      "learning_rate": 6.225487399072955e-06,
      "loss": 0.8008,
      "step": 5977
    },
    {
      "epoch": 0.4386034098517017,
      "grad_norm": 1.4311642918529475,
      "learning_rate": 6.2243354094212116e-06,
      "loss": 0.9067,
      "step": 5978
    },
    {
      "epoch": 0.43867677944184086,
      "grad_norm": 0.5535699931307233,
      "learning_rate": 6.223183350628215e-06,
      "loss": 0.4915,
      "step": 5979
    },
    {
      "epoch": 0.43875014903197995,
      "grad_norm": 1.480257672786072,
      "learning_rate": 6.222031222759026e-06,
      "loss": 0.7931,
      "step": 5980
    },
    {
      "epoch": 0.4388235186221191,
      "grad_norm": 1.504609205827271,
      "learning_rate": 6.220879025878709e-06,
      "loss": 0.7423,
      "step": 5981
    },
    {
      "epoch": 0.43889688821225825,
      "grad_norm": 0.5586186288841418,
      "learning_rate": 6.219726760052327e-06,
      "loss": 0.5208,
      "step": 5982
    },
    {
      "epoch": 0.43897025780239735,
      "grad_norm": 1.4933186420168139,
      "learning_rate": 6.218574425344957e-06,
      "loss": 0.7948,
      "step": 5983
    },
    {
      "epoch": 0.4390436273925365,
      "grad_norm": 1.585826457916395,
      "learning_rate": 6.217422021821669e-06,
      "loss": 0.8553,
      "step": 5984
    },
    {
      "epoch": 0.4390436273925365,
      "lm_loss": 0.6526,
      "step": 5984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4390436273925365,
      "lm_loss": 0.3947,
      "step": 5984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4390436273925365,
      "lm_loss": 0.5328,
      "step": 5984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4390436273925365,
      "lm_loss": 0.4729,
      "step": 5984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4390436273925365,
      "lm_loss": 0.4879,
      "step": 5984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4390436273925365,
      "lm_loss": 0.3892,
      "step": 5984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4390436273925365,
      "lm_loss": 0.5586,
      "step": 5984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4390436273925365,
      "lm_loss": 0.5089,
      "step": 5984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4391169969826756,
      "grad_norm": 0.5517711362523113,
      "learning_rate": 6.216269549547546e-06,
      "loss": 0.515,
      "step": 5985
    },
    {
      "epoch": 0.43919036657281474,
      "grad_norm": 1.5837108090576106,
      "learning_rate": 6.21511700858767e-06,
      "loss": 0.8718,
      "step": 5986
    },
    {
      "epoch": 0.43926373616295383,
      "grad_norm": 1.5571469935101732,
      "learning_rate": 6.213964399007126e-06,
      "loss": 0.7034,
      "step": 5987
    },
    {
      "epoch": 0.439337105753093,
      "grad_norm": 1.345517544279733,
      "learning_rate": 6.212811720871007e-06,
      "loss": 0.7236,
      "step": 5988
    },
    {
      "epoch": 0.43941047534323213,
      "grad_norm": 0.5718555287505039,
      "learning_rate": 6.211658974244407e-06,
      "loss": 0.5025,
      "step": 5989
    },
    {
      "epoch": 0.4394838449333712,
      "grad_norm": 1.3490199071475024,
      "learning_rate": 6.210506159192422e-06,
      "loss": 0.7373,
      "step": 5990
    },
    {
      "epoch": 0.4395572145235104,
      "grad_norm": 1.459384188498971,
      "learning_rate": 6.209353275780159e-06,
      "loss": 0.7269,
      "step": 5991
    },
    {
      "epoch": 0.43963058411364947,
      "grad_norm": 1.6136849626691876,
      "learning_rate": 6.2082003240727195e-06,
      "loss": 0.8483,
      "step": 5992
    },
    {
      "epoch": 0.43963058411364947,
      "lm_loss": 0.5747,
      "step": 5992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43963058411364947,
      "lm_loss": 0.41,
      "step": 5992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43963058411364947,
      "lm_loss": 0.4987,
      "step": 5992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43963058411364947,
      "lm_loss": 0.4484,
      "step": 5992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43963058411364947,
      "lm_loss": 0.4632,
      "step": 5992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43963058411364947,
      "lm_loss": 0.4373,
      "step": 5992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43963058411364947,
      "lm_loss": 0.4016,
      "step": 5992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.43963058411364947,
      "lm_loss": 0.5483,
      "step": 5992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4397039537037886,
      "grad_norm": 0.5520597895814556,
      "learning_rate": 6.207047304135217e-06,
      "loss": 0.5017,
      "step": 5993
    },
    {
      "epoch": 0.43977732329392777,
      "grad_norm": 0.548752961515001,
      "learning_rate": 6.205894216032764e-06,
      "loss": 0.4958,
      "step": 5994
    },
    {
      "epoch": 0.43985069288406686,
      "grad_norm": 0.57523092084607,
      "learning_rate": 6.204741059830477e-06,
      "loss": 0.5061,
      "step": 5995
    },
    {
      "epoch": 0.439924062474206,
      "grad_norm": 1.5488334287748882,
      "learning_rate": 6.20358783559348e-06,
      "loss": 0.7524,
      "step": 5996
    },
    {
      "epoch": 0.4399974320643451,
      "grad_norm": 1.49955926311639,
      "learning_rate": 6.202434543386897e-06,
      "loss": 0.702,
      "step": 5997
    },
    {
      "epoch": 0.44007080165448426,
      "grad_norm": 1.4299562725310666,
      "learning_rate": 6.201281183275857e-06,
      "loss": 0.7943,
      "step": 5998
    },
    {
      "epoch": 0.4401441712446234,
      "grad_norm": 1.6526605006049364,
      "learning_rate": 6.2001277553254946e-06,
      "loss": 0.7617,
      "step": 5999
    },
    {
      "epoch": 0.4402175408347625,
      "grad_norm": 1.387628917206402,
      "learning_rate": 6.198974259600946e-06,
      "loss": 0.7608,
      "step": 6000
    },
    {
      "epoch": 0.4402175408347625,
      "lm_loss": 1.5633,
      "step": 6000,
      "vm_loss": 0.0551
    },
    {
      "epoch": 0.4402175408347625,
      "lm_loss": 0.8149,
      "step": 6000,
      "vm_loss": 0.0896
    },
    {
      "epoch": 0.4402175408347625,
      "lm_loss": 1.4913,
      "step": 6000,
      "vm_loss": 0.1584
    },
    {
      "epoch": 0.4402175408347625,
      "lm_loss": 0.5418,
      "step": 6000,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.4402175408347625,
      "lm_loss": 0.4528,
      "step": 6000,
      "vm_loss": 0.099
    },
    {
      "epoch": 0.4402175408347625,
      "lm_loss": 0.9284,
      "step": 6000,
      "vm_loss": 0.2008
    },
    {
      "epoch": 0.4402175408347625,
      "lm_loss": 0.9271,
      "step": 6000,
      "vm_loss": 0.1301
    },
    {
      "epoch": 0.4402175408347625,
      "lm_loss": 0.3966,
      "step": 6000,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.44029091042490165,
      "grad_norm": 1.3613114145433218,
      "learning_rate": 6.197820696167351e-06,
      "loss": 0.8189,
      "step": 6001
    },
    {
      "epoch": 0.44036428001504074,
      "grad_norm": 0.5859042595822892,
      "learning_rate": 6.196667065089854e-06,
      "loss": 0.5073,
      "step": 6002
    },
    {
      "epoch": 0.4404376496051799,
      "grad_norm": 1.389547879438999,
      "learning_rate": 6.195513366433605e-06,
      "loss": 0.7287,
      "step": 6003
    },
    {
      "epoch": 0.44051101919531904,
      "grad_norm": 1.5711942045874352,
      "learning_rate": 6.194359600263757e-06,
      "loss": 0.6803,
      "step": 6004
    },
    {
      "epoch": 0.44058438878545814,
      "grad_norm": 0.5644364732421411,
      "learning_rate": 6.193205766645464e-06,
      "loss": 0.5141,
      "step": 6005
    },
    {
      "epoch": 0.4406577583755973,
      "grad_norm": 1.457400646263075,
      "learning_rate": 6.192051865643886e-06,
      "loss": 0.7771,
      "step": 6006
    },
    {
      "epoch": 0.4407311279657364,
      "grad_norm": 1.7325609559994068,
      "learning_rate": 6.190897897324187e-06,
      "loss": 0.7549,
      "step": 6007
    },
    {
      "epoch": 0.44080449755587553,
      "grad_norm": 1.558626009553066,
      "learning_rate": 6.189743861751535e-06,
      "loss": 0.694,
      "step": 6008
    },
    {
      "epoch": 0.44080449755587553,
      "lm_loss": 0.6507,
      "step": 6008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44080449755587553,
      "lm_loss": 0.42,
      "step": 6008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44080449755587553,
      "lm_loss": 0.3685,
      "step": 6008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44080449755587553,
      "lm_loss": 0.4234,
      "step": 6008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44080449755587553,
      "lm_loss": 0.526,
      "step": 6008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44080449755587553,
      "lm_loss": 0.5445,
      "step": 6008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44080449755587553,
      "lm_loss": 0.3946,
      "step": 6008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44080449755587553,
      "lm_loss": 0.5506,
      "step": 6008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4408778671460147,
      "grad_norm": 0.5743916252675594,
      "learning_rate": 6.1885897589911e-06,
      "loss": 0.5064,
      "step": 6009
    },
    {
      "epoch": 0.4409512367361538,
      "grad_norm": 0.5689894412562034,
      "learning_rate": 6.18743558910806e-06,
      "loss": 0.5054,
      "step": 6010
    },
    {
      "epoch": 0.4410246063262929,
      "grad_norm": 1.5650743553661832,
      "learning_rate": 6.186281352167588e-06,
      "loss": 0.7635,
      "step": 6011
    },
    {
      "epoch": 0.441097975916432,
      "grad_norm": 1.4982502627334564,
      "learning_rate": 6.185127048234872e-06,
      "loss": 0.8171,
      "step": 6012
    },
    {
      "epoch": 0.44117134550657117,
      "grad_norm": 1.4259036860443413,
      "learning_rate": 6.183972677375099e-06,
      "loss": 0.8218,
      "step": 6013
    },
    {
      "epoch": 0.4412447150967103,
      "grad_norm": 0.5835711911195032,
      "learning_rate": 6.182818239653453e-06,
      "loss": 0.5183,
      "step": 6014
    },
    {
      "epoch": 0.4413180846868494,
      "grad_norm": 1.4400024938719862,
      "learning_rate": 6.181663735135134e-06,
      "loss": 0.7755,
      "step": 6015
    },
    {
      "epoch": 0.44139145427698856,
      "grad_norm": 1.573586835236665,
      "learning_rate": 6.1805091638853365e-06,
      "loss": 0.7998,
      "step": 6016
    },
    {
      "epoch": 0.44139145427698856,
      "lm_loss": 1.3625,
      "step": 6016,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.44139145427698856,
      "lm_loss": 1.297,
      "step": 6016,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.44139145427698856,
      "lm_loss": 0.0953,
      "step": 6016,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.44139145427698856,
      "lm_loss": 0.8196,
      "step": 6016,
      "vm_loss": 0.0874
    },
    {
      "epoch": 0.44139145427698856,
      "lm_loss": 0.6009,
      "step": 6016,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.44139145427698856,
      "lm_loss": 0.9056,
      "step": 6016,
      "vm_loss": 0.1394
    },
    {
      "epoch": 0.44139145427698856,
      "lm_loss": 0.4558,
      "step": 6016,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.44139145427698856,
      "lm_loss": 0.2126,
      "step": 6016,
      "vm_loss": 0.1427
    },
    {
      "epoch": 0.44146482386712765,
      "grad_norm": 1.6465545693801478,
      "learning_rate": 6.179354525969263e-06,
      "loss": 0.9524,
      "step": 6017
    },
    {
      "epoch": 0.4415381934572668,
      "grad_norm": 1.4481168039356165,
      "learning_rate": 6.17819982145212e-06,
      "loss": 0.842,
      "step": 6018
    },
    {
      "epoch": 0.44161156304740595,
      "grad_norm": 1.5778871179889713,
      "learning_rate": 6.177045050399114e-06,
      "loss": 0.7898,
      "step": 6019
    },
    {
      "epoch": 0.44168493263754505,
      "grad_norm": 0.5747537579965898,
      "learning_rate": 6.17589021287546e-06,
      "loss": 0.5171,
      "step": 6020
    },
    {
      "epoch": 0.4417583022276842,
      "grad_norm": 1.2777012343181926,
      "learning_rate": 6.174735308946374e-06,
      "loss": 0.695,
      "step": 6021
    },
    {
      "epoch": 0.4418316718178233,
      "grad_norm": 0.5579619519718978,
      "learning_rate": 6.173580338677074e-06,
      "loss": 0.518,
      "step": 6022
    },
    {
      "epoch": 0.44190504140796244,
      "grad_norm": 1.4382791580850127,
      "learning_rate": 6.1724253021327886e-06,
      "loss": 0.7653,
      "step": 6023
    },
    {
      "epoch": 0.4419784109981016,
      "grad_norm": 1.4647070924372996,
      "learning_rate": 6.17127019937874e-06,
      "loss": 0.7521,
      "step": 6024
    },
    {
      "epoch": 0.4419784109981016,
      "lm_loss": 0.5724,
      "step": 6024,
      "vm_loss": 0.0708
    },
    {
      "epoch": 0.4419784109981016,
      "lm_loss": 0.7404,
      "step": 6024,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.4419784109981016,
      "lm_loss": 0.7761,
      "step": 6024,
      "vm_loss": 0.1341
    },
    {
      "epoch": 0.4419784109981016,
      "lm_loss": 0.8081,
      "step": 6024,
      "vm_loss": 0.1408
    },
    {
      "epoch": 0.4419784109981016,
      "lm_loss": 0.3488,
      "step": 6024,
      "vm_loss": 0.1369
    },
    {
      "epoch": 0.4419784109981016,
      "lm_loss": 0.0889,
      "step": 6024,
      "vm_loss": 0.1541
    },
    {
      "epoch": 0.4419784109981016,
      "lm_loss": 0.8224,
      "step": 6024,
      "vm_loss": 0.0747
    },
    {
      "epoch": 0.4419784109981016,
      "lm_loss": 0.6184,
      "step": 6024,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.4420517805882407,
      "grad_norm": 1.489570953095038,
      "learning_rate": 6.170115030480165e-06,
      "loss": 0.6855,
      "step": 6025
    },
    {
      "epoch": 0.44212515017837983,
      "grad_norm": 1.666823709745141,
      "learning_rate": 6.168959795502296e-06,
      "loss": 0.7507,
      "step": 6026
    },
    {
      "epoch": 0.4421985197685189,
      "grad_norm": 1.6025297509532828,
      "learning_rate": 6.167804494510372e-06,
      "loss": 0.7476,
      "step": 6027
    },
    {
      "epoch": 0.4422718893586581,
      "grad_norm": 1.5185932468024885,
      "learning_rate": 6.166649127569636e-06,
      "loss": 0.8151,
      "step": 6028
    },
    {
      "epoch": 0.44234525894879717,
      "grad_norm": 0.5983342301564972,
      "learning_rate": 6.1654936947453355e-06,
      "loss": 0.5062,
      "step": 6029
    },
    {
      "epoch": 0.4424186285389363,
      "grad_norm": 1.5116949683260292,
      "learning_rate": 6.164338196102719e-06,
      "loss": 0.8032,
      "step": 6030
    },
    {
      "epoch": 0.44249199812907547,
      "grad_norm": 1.702962299348481,
      "learning_rate": 6.163182631707042e-06,
      "loss": 0.8699,
      "step": 6031
    },
    {
      "epoch": 0.44256536771921456,
      "grad_norm": 1.456883265883933,
      "learning_rate": 6.16202700162356e-06,
      "loss": 0.7732,
      "step": 6032
    },
    {
      "epoch": 0.44256536771921456,
      "lm_loss": 0.935,
      "step": 6032,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.44256536771921456,
      "lm_loss": 1.4192,
      "step": 6032,
      "vm_loss": 0.0656
    },
    {
      "epoch": 0.44256536771921456,
      "lm_loss": 0.1423,
      "step": 6032,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.44256536771921456,
      "lm_loss": 0.0914,
      "step": 6032,
      "vm_loss": 0.1202
    },
    {
      "epoch": 0.44256536771921456,
      "lm_loss": 0.3361,
      "step": 6032,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.44256536771921456,
      "lm_loss": 0.934,
      "step": 6032,
      "vm_loss": 0.126
    },
    {
      "epoch": 0.44256536771921456,
      "lm_loss": 0.35,
      "step": 6032,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.44256536771921456,
      "lm_loss": 0.4052,
      "step": 6032,
      "vm_loss": 0.0962
    },
    {
      "epoch": 0.4426387373093537,
      "grad_norm": 1.436347098326306,
      "learning_rate": 6.160871305917536e-06,
      "loss": 0.7659,
      "step": 6033
    },
    {
      "epoch": 0.4427121068994928,
      "grad_norm": 1.612479894369656,
      "learning_rate": 6.159715544654235e-06,
      "loss": 0.7905,
      "step": 6034
    },
    {
      "epoch": 0.44278547648963196,
      "grad_norm": 1.5048967873773305,
      "learning_rate": 6.158559717898924e-06,
      "loss": 0.8356,
      "step": 6035
    },
    {
      "epoch": 0.4428588460797711,
      "grad_norm": 0.5798269987277579,
      "learning_rate": 6.157403825716878e-06,
      "loss": 0.5198,
      "step": 6036
    },
    {
      "epoch": 0.4429322156699102,
      "grad_norm": 1.6687435020898416,
      "learning_rate": 6.156247868173372e-06,
      "loss": 0.8889,
      "step": 6037
    },
    {
      "epoch": 0.44300558526004935,
      "grad_norm": 1.4974418468262065,
      "learning_rate": 6.155091845333682e-06,
      "loss": 0.8223,
      "step": 6038
    },
    {
      "epoch": 0.44307895485018844,
      "grad_norm": 1.7389868970930094,
      "learning_rate": 6.153935757263098e-06,
      "loss": 0.8577,
      "step": 6039
    },
    {
      "epoch": 0.4431523244403276,
      "grad_norm": 1.6449457388273339,
      "learning_rate": 6.152779604026905e-06,
      "loss": 0.8067,
      "step": 6040
    },
    {
      "epoch": 0.4431523244403276,
      "lm_loss": 0.6508,
      "step": 6040,
      "vm_loss": 0.0515
    },
    {
      "epoch": 0.4431523244403276,
      "lm_loss": 0.3528,
      "step": 6040,
      "vm_loss": 0.0587
    },
    {
      "epoch": 0.4431523244403276,
      "lm_loss": 0.3034,
      "step": 6040,
      "vm_loss": 0.1094
    },
    {
      "epoch": 0.4431523244403276,
      "lm_loss": 0.5065,
      "step": 6040,
      "vm_loss": 0.0903
    },
    {
      "epoch": 0.4431523244403276,
      "lm_loss": 0.062,
      "step": 6040,
      "vm_loss": 0.1324
    },
    {
      "epoch": 0.4431523244403276,
      "lm_loss": 1.1571,
      "step": 6040,
      "vm_loss": 0.1057
    },
    {
      "epoch": 0.4431523244403276,
      "lm_loss": 0.0116,
      "step": 6040,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.4431523244403276,
      "lm_loss": 1.1557,
      "step": 6040,
      "vm_loss": 0.1596
    },
    {
      "epoch": 0.44322569403046674,
      "grad_norm": 1.4851324009871234,
      "learning_rate": 6.1516233856903904e-06,
      "loss": 0.742,
      "step": 6041
    },
    {
      "epoch": 0.44329906362060584,
      "grad_norm": 0.5879858926882385,
      "learning_rate": 6.150467102318853e-06,
      "loss": 0.4998,
      "step": 6042
    },
    {
      "epoch": 0.443372433210745,
      "grad_norm": 1.3899137459380628,
      "learning_rate": 6.149310753977589e-06,
      "loss": 0.7787,
      "step": 6043
    },
    {
      "epoch": 0.4434458028008841,
      "grad_norm": 1.5429819523300148,
      "learning_rate": 6.1481543407319e-06,
      "loss": 0.7218,
      "step": 6044
    },
    {
      "epoch": 0.44351917239102323,
      "grad_norm": 1.5186871764349312,
      "learning_rate": 6.146997862647092e-06,
      "loss": 0.8353,
      "step": 6045
    },
    {
      "epoch": 0.4435925419811624,
      "grad_norm": 1.2906366463968268,
      "learning_rate": 6.145841319788475e-06,
      "loss": 0.7973,
      "step": 6046
    },
    {
      "epoch": 0.4436659115713015,
      "grad_norm": 1.4297367297920633,
      "learning_rate": 6.1446847122213604e-06,
      "loss": 0.6909,
      "step": 6047
    },
    {
      "epoch": 0.4437392811614406,
      "grad_norm": 1.4731039984565726,
      "learning_rate": 6.143528040011066e-06,
      "loss": 0.7864,
      "step": 6048
    },
    {
      "epoch": 0.4437392811614406,
      "lm_loss": 1.361,
      "step": 6048,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.4437392811614406,
      "lm_loss": 0.2849,
      "step": 6048,
      "vm_loss": 0.1691
    },
    {
      "epoch": 0.4437392811614406,
      "lm_loss": 0.4157,
      "step": 6048,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.4437392811614406,
      "lm_loss": 0.6921,
      "step": 6048,
      "vm_loss": 0.0407
    },
    {
      "epoch": 0.4437392811614406,
      "lm_loss": 0.7842,
      "step": 6048,
      "vm_loss": 0.2584
    },
    {
      "epoch": 0.4437392811614406,
      "lm_loss": 1.2662,
      "step": 6048,
      "vm_loss": 0.2577
    },
    {
      "epoch": 0.4437392811614406,
      "lm_loss": 0.7815,
      "step": 6048,
      "vm_loss": 0.1103
    },
    {
      "epoch": 0.4437392811614406,
      "lm_loss": 1.9371,
      "step": 6048,
      "vm_loss": 0.1331
    },
    {
      "epoch": 0.4438126507515797,
      "grad_norm": 1.7779500486793147,
      "learning_rate": 6.142371303222909e-06,
      "loss": 0.8315,
      "step": 6049
    },
    {
      "epoch": 0.44388602034171887,
      "grad_norm": 1.3809367418121665,
      "learning_rate": 6.141214501922218e-06,
      "loss": 0.7305,
      "step": 6050
    },
    {
      "epoch": 0.443959389931858,
      "grad_norm": 0.640623033658782,
      "learning_rate": 6.140057636174317e-06,
      "loss": 0.5244,
      "step": 6051
    },
    {
      "epoch": 0.4440327595219971,
      "grad_norm": 1.5534302731458474,
      "learning_rate": 6.138900706044538e-06,
      "loss": 0.7663,
      "step": 6052
    },
    {
      "epoch": 0.44410612911213626,
      "grad_norm": 1.3721907720163036,
      "learning_rate": 6.1377437115982166e-06,
      "loss": 0.7959,
      "step": 6053
    },
    {
      "epoch": 0.44417949870227535,
      "grad_norm": 1.5828098031195543,
      "learning_rate": 6.136586652900689e-06,
      "loss": 0.8646,
      "step": 6054
    },
    {
      "epoch": 0.4442528682924145,
      "grad_norm": 1.5536520410921275,
      "learning_rate": 6.135429530017298e-06,
      "loss": 0.8007,
      "step": 6055
    },
    {
      "epoch": 0.44432623788255365,
      "grad_norm": 0.5537196379875642,
      "learning_rate": 6.134272343013391e-06,
      "loss": 0.482,
      "step": 6056
    },
    {
      "epoch": 0.44432623788255365,
      "lm_loss": 1.3412,
      "step": 6056,
      "vm_loss": 0.0526
    },
    {
      "epoch": 0.44432623788255365,
      "lm_loss": 1.0089,
      "step": 6056,
      "vm_loss": 0.1518
    },
    {
      "epoch": 0.44432623788255365,
      "lm_loss": 0.4991,
      "step": 6056,
      "vm_loss": 0.1099
    },
    {
      "epoch": 0.44432623788255365,
      "lm_loss": 1.0433,
      "step": 6056,
      "vm_loss": 0.1493
    },
    {
      "epoch": 0.44432623788255365,
      "lm_loss": 0.2945,
      "step": 6056,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.44432623788255365,
      "lm_loss": 0.5186,
      "step": 6056,
      "vm_loss": 0.1969
    },
    {
      "epoch": 0.44432623788255365,
      "lm_loss": 0.6894,
      "step": 6056,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.44432623788255365,
      "lm_loss": 1.5132,
      "step": 6056,
      "vm_loss": 0.0736
    },
    {
      "epoch": 0.44439960747269275,
      "grad_norm": 1.7619665250976653,
      "learning_rate": 6.133115091954315e-06,
      "loss": 0.7982,
      "step": 6057
    },
    {
      "epoch": 0.4444729770628319,
      "grad_norm": 0.5731378761282613,
      "learning_rate": 6.131957776905424e-06,
      "loss": 0.5122,
      "step": 6058
    },
    {
      "epoch": 0.444546346652971,
      "grad_norm": 1.5961222903139072,
      "learning_rate": 6.130800397932073e-06,
      "loss": 0.8671,
      "step": 6059
    },
    {
      "epoch": 0.44461971624311014,
      "grad_norm": 1.6057881348625134,
      "learning_rate": 6.129642955099625e-06,
      "loss": 0.785,
      "step": 6060
    },
    {
      "epoch": 0.4446930858332493,
      "grad_norm": 0.5713824975517143,
      "learning_rate": 6.128485448473441e-06,
      "loss": 0.5128,
      "step": 6061
    },
    {
      "epoch": 0.4447664554233884,
      "grad_norm": 1.5853780795234165,
      "learning_rate": 6.127327878118888e-06,
      "loss": 0.7619,
      "step": 6062
    },
    {
      "epoch": 0.44483982501352753,
      "grad_norm": 0.5665924122939168,
      "learning_rate": 6.126170244101339e-06,
      "loss": 0.5016,
      "step": 6063
    },
    {
      "epoch": 0.44491319460366663,
      "grad_norm": 0.5581805920756812,
      "learning_rate": 6.125012546486168e-06,
      "loss": 0.5111,
      "step": 6064
    },
    {
      "epoch": 0.44491319460366663,
      "lm_loss": 0.6563,
      "step": 6064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44491319460366663,
      "lm_loss": 0.4846,
      "step": 6064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44491319460366663,
      "lm_loss": 0.5926,
      "step": 6064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44491319460366663,
      "lm_loss": 0.6017,
      "step": 6064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44491319460366663,
      "lm_loss": 0.5444,
      "step": 6064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44491319460366663,
      "lm_loss": 0.5575,
      "step": 6064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44491319460366663,
      "lm_loss": 0.5114,
      "step": 6064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44491319460366663,
      "lm_loss": 0.4834,
      "step": 6064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4449865641938058,
      "grad_norm": 0.5403412537399401,
      "learning_rate": 6.123854785338752e-06,
      "loss": 0.5056,
      "step": 6065
    },
    {
      "epoch": 0.44505993378394487,
      "grad_norm": 1.545968606725279,
      "learning_rate": 6.122696960724472e-06,
      "loss": 0.8263,
      "step": 6066
    },
    {
      "epoch": 0.445133303374084,
      "grad_norm": 0.5567601370237952,
      "learning_rate": 6.121539072708717e-06,
      "loss": 0.5225,
      "step": 6067
    },
    {
      "epoch": 0.44520667296422317,
      "grad_norm": 0.5790206267119166,
      "learning_rate": 6.12038112135687e-06,
      "loss": 0.5153,
      "step": 6068
    },
    {
      "epoch": 0.44528004255436227,
      "grad_norm": 1.4290479780741947,
      "learning_rate": 6.119223106734328e-06,
      "loss": 0.777,
      "step": 6069
    },
    {
      "epoch": 0.4453534121445014,
      "grad_norm": 0.5651745285067318,
      "learning_rate": 6.118065028906485e-06,
      "loss": 0.4988,
      "step": 6070
    },
    {
      "epoch": 0.4454267817346405,
      "grad_norm": 0.5501906969032181,
      "learning_rate": 6.11690688793874e-06,
      "loss": 0.5099,
      "step": 6071
    },
    {
      "epoch": 0.44550015132477966,
      "grad_norm": 1.5658409066567789,
      "learning_rate": 6.1157486838965e-06,
      "loss": 0.9288,
      "step": 6072
    },
    {
      "epoch": 0.44550015132477966,
      "lm_loss": 1.5743,
      "step": 6072,
      "vm_loss": 0.0713
    },
    {
      "epoch": 0.44550015132477966,
      "lm_loss": 1.3118,
      "step": 6072,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.44550015132477966,
      "lm_loss": 0.3176,
      "step": 6072,
      "vm_loss": 0.1317
    },
    {
      "epoch": 0.44550015132477966,
      "lm_loss": 0.8683,
      "step": 6072,
      "vm_loss": 0.1168
    },
    {
      "epoch": 0.44550015132477966,
      "lm_loss": 1.0156,
      "step": 6072,
      "vm_loss": 0.1767
    },
    {
      "epoch": 0.44550015132477966,
      "lm_loss": 0.8172,
      "step": 6072,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.44550015132477966,
      "lm_loss": 0.1515,
      "step": 6072,
      "vm_loss": 0.0744
    },
    {
      "epoch": 0.44550015132477966,
      "lm_loss": 1.3215,
      "step": 6072,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.4455735209149188,
      "grad_norm": 1.7255548811546295,
      "learning_rate": 6.114590416845167e-06,
      "loss": 0.8333,
      "step": 6073
    },
    {
      "epoch": 0.4456468905050579,
      "grad_norm": 0.565743984317679,
      "learning_rate": 6.113432086850153e-06,
      "loss": 0.507,
      "step": 6074
    },
    {
      "epoch": 0.44572026009519705,
      "grad_norm": 0.5727207060233865,
      "learning_rate": 6.112273693976871e-06,
      "loss": 0.5115,
      "step": 6075
    },
    {
      "epoch": 0.44579362968533615,
      "grad_norm": 1.6578892944794894,
      "learning_rate": 6.111115238290739e-06,
      "loss": 0.7626,
      "step": 6076
    },
    {
      "epoch": 0.4458669992754753,
      "grad_norm": 0.546396444552591,
      "learning_rate": 6.10995671985718e-06,
      "loss": 0.5037,
      "step": 6077
    },
    {
      "epoch": 0.44594036886561444,
      "grad_norm": 0.5511495267876322,
      "learning_rate": 6.108798138741613e-06,
      "loss": 0.5023,
      "step": 6078
    },
    {
      "epoch": 0.44601373845575354,
      "grad_norm": 1.6174480087507637,
      "learning_rate": 6.107639495009472e-06,
      "loss": 0.8209,
      "step": 6079
    },
    {
      "epoch": 0.4460871080458927,
      "grad_norm": 1.5052631437458994,
      "learning_rate": 6.106480788726184e-06,
      "loss": 0.7344,
      "step": 6080
    },
    {
      "epoch": 0.4460871080458927,
      "lm_loss": 1.0651,
      "step": 6080,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.4460871080458927,
      "lm_loss": 0.6628,
      "step": 6080,
      "vm_loss": 0.1128
    },
    {
      "epoch": 0.4460871080458927,
      "lm_loss": 0.1605,
      "step": 6080,
      "vm_loss": 0.1036
    },
    {
      "epoch": 0.4460871080458927,
      "lm_loss": 1.0608,
      "step": 6080,
      "vm_loss": 0.2145
    },
    {
      "epoch": 0.4460871080458927,
      "lm_loss": 1.0402,
      "step": 6080,
      "vm_loss": 0.1626
    },
    {
      "epoch": 0.4460871080458927,
      "lm_loss": 0.7256,
      "step": 6080,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.4460871080458927,
      "lm_loss": 0.762,
      "step": 6080,
      "vm_loss": 0.1394
    },
    {
      "epoch": 0.4460871080458927,
      "lm_loss": 0.6729,
      "step": 6080,
      "vm_loss": 0.1856
    },
    {
      "epoch": 0.4461604776360318,
      "grad_norm": 1.505750755794014,
      "learning_rate": 6.105322019957187e-06,
      "loss": 0.8458,
      "step": 6081
    },
    {
      "epoch": 0.44623384722617093,
      "grad_norm": 1.4698249299044703,
      "learning_rate": 6.104163188767919e-06,
      "loss": 0.7912,
      "step": 6082
    },
    {
      "epoch": 0.4463072168163101,
      "grad_norm": 0.5910697375623335,
      "learning_rate": 6.103004295223819e-06,
      "loss": 0.4936,
      "step": 6083
    },
    {
      "epoch": 0.4463805864064492,
      "grad_norm": 1.4804332943818275,
      "learning_rate": 6.101845339390336e-06,
      "loss": 0.707,
      "step": 6084
    },
    {
      "epoch": 0.4464539559965883,
      "grad_norm": 0.5649211562406019,
      "learning_rate": 6.100686321332918e-06,
      "loss": 0.4918,
      "step": 6085
    },
    {
      "epoch": 0.4465273255867274,
      "grad_norm": 1.3912250069078231,
      "learning_rate": 6.099527241117017e-06,
      "loss": 0.7557,
      "step": 6086
    },
    {
      "epoch": 0.44660069517686657,
      "grad_norm": 1.584522126717808,
      "learning_rate": 6.098368098808091e-06,
      "loss": 0.7632,
      "step": 6087
    },
    {
      "epoch": 0.4466740647670057,
      "grad_norm": 0.5386515695923542,
      "learning_rate": 6.0972088944715976e-06,
      "loss": 0.5049,
      "step": 6088
    },
    {
      "epoch": 0.4466740647670057,
      "lm_loss": 1.4688,
      "step": 6088,
      "vm_loss": 0.0567
    },
    {
      "epoch": 0.4466740647670057,
      "lm_loss": 0.9255,
      "step": 6088,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.4466740647670057,
      "lm_loss": 1.1963,
      "step": 6088,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.4466740647670057,
      "lm_loss": 0.24,
      "step": 6088,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.4466740647670057,
      "lm_loss": 0.5071,
      "step": 6088,
      "vm_loss": 0.0692
    },
    {
      "epoch": 0.4466740647670057,
      "lm_loss": 0.9656,
      "step": 6088,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0.4466740647670057,
      "lm_loss": 0.9802,
      "step": 6088,
      "vm_loss": 0.1521
    },
    {
      "epoch": 0.4466740647670057,
      "lm_loss": 0.5341,
      "step": 6088,
      "vm_loss": 0.1766
    },
    {
      "epoch": 0.4467474343571448,
      "grad_norm": 1.468078402577876,
      "learning_rate": 6.0960496281729995e-06,
      "loss": 0.8109,
      "step": 6089
    },
    {
      "epoch": 0.44682080394728396,
      "grad_norm": 0.5664247898240906,
      "learning_rate": 6.094890299977767e-06,
      "loss": 0.5168,
      "step": 6090
    },
    {
      "epoch": 0.44689417353742306,
      "grad_norm": 1.4810240340182763,
      "learning_rate": 6.093730909951365e-06,
      "loss": 0.7042,
      "step": 6091
    },
    {
      "epoch": 0.4469675431275622,
      "grad_norm": 1.4788012594358098,
      "learning_rate": 6.092571458159273e-06,
      "loss": 0.8202,
      "step": 6092
    },
    {
      "epoch": 0.44704091271770136,
      "grad_norm": 1.3966266312819082,
      "learning_rate": 6.091411944666964e-06,
      "loss": 0.8562,
      "step": 6093
    },
    {
      "epoch": 0.44711428230784045,
      "grad_norm": 1.4410028097226069,
      "learning_rate": 6.0902523695399186e-06,
      "loss": 0.8078,
      "step": 6094
    },
    {
      "epoch": 0.4471876518979796,
      "grad_norm": 0.5652493906433329,
      "learning_rate": 6.089092732843622e-06,
      "loss": 0.4804,
      "step": 6095
    },
    {
      "epoch": 0.4472610214881187,
      "grad_norm": 0.5792633906836019,
      "learning_rate": 6.087933034643563e-06,
      "loss": 0.5075,
      "step": 6096
    },
    {
      "epoch": 0.4472610214881187,
      "lm_loss": 1.1358,
      "step": 6096,
      "vm_loss": 0.0678
    },
    {
      "epoch": 0.4472610214881187,
      "lm_loss": 0.7494,
      "step": 6096,
      "vm_loss": 0.1635
    },
    {
      "epoch": 0.4472610214881187,
      "lm_loss": 1.059,
      "step": 6096,
      "vm_loss": 0.154
    },
    {
      "epoch": 0.4472610214881187,
      "lm_loss": 0.0575,
      "step": 6096,
      "vm_loss": 0.0682
    },
    {
      "epoch": 0.4472610214881187,
      "lm_loss": 0.0517,
      "step": 6096,
      "vm_loss": 0.1716
    },
    {
      "epoch": 0.4472610214881187,
      "lm_loss": 0.7199,
      "step": 6096,
      "vm_loss": 0.1618
    },
    {
      "epoch": 0.4472610214881187,
      "lm_loss": 0.1954,
      "step": 6096,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.4472610214881187,
      "lm_loss": 0.5998,
      "step": 6096,
      "vm_loss": 0.197
    },
    {
      "epoch": 0.44733439107825784,
      "grad_norm": 1.5326801905996015,
      "learning_rate": 6.08677327500523e-06,
      "loss": 0.8666,
      "step": 6097
    },
    {
      "epoch": 0.447407760668397,
      "grad_norm": 1.6840641168041912,
      "learning_rate": 6.0856134539941204e-06,
      "loss": 0.785,
      "step": 6098
    },
    {
      "epoch": 0.4474811302585361,
      "grad_norm": 1.5701416703796385,
      "learning_rate": 6.084453571675728e-06,
      "loss": 0.7088,
      "step": 6099
    },
    {
      "epoch": 0.44755449984867524,
      "grad_norm": 1.4787884499419404,
      "learning_rate": 6.08329362811556e-06,
      "loss": 0.7398,
      "step": 6100
    },
    {
      "epoch": 0.44762786943881433,
      "grad_norm": 0.6910559813846194,
      "learning_rate": 6.082133623379116e-06,
      "loss": 0.4937,
      "step": 6101
    },
    {
      "epoch": 0.4477012390289535,
      "grad_norm": 1.6781070961425466,
      "learning_rate": 6.080973557531907e-06,
      "loss": 0.7257,
      "step": 6102
    },
    {
      "epoch": 0.44777460861909263,
      "grad_norm": 0.6061035349067742,
      "learning_rate": 6.079813430639444e-06,
      "loss": 0.4834,
      "step": 6103
    },
    {
      "epoch": 0.4478479782092317,
      "grad_norm": 1.6118854740028705,
      "learning_rate": 6.078653242767243e-06,
      "loss": 0.7365,
      "step": 6104
    },
    {
      "epoch": 0.4478479782092317,
      "lm_loss": 1.4935,
      "step": 6104,
      "vm_loss": 0.0362
    },
    {
      "epoch": 0.4478479782092317,
      "lm_loss": 1.4579,
      "step": 6104,
      "vm_loss": 0.1287
    },
    {
      "epoch": 0.4478479782092317,
      "lm_loss": 0.027,
      "step": 6104,
      "vm_loss": 0.1889
    },
    {
      "epoch": 0.4478479782092317,
      "lm_loss": 0.2191,
      "step": 6104,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.4478479782092317,
      "lm_loss": 0.5254,
      "step": 6104,
      "vm_loss": 0.1788
    },
    {
      "epoch": 0.4478479782092317,
      "lm_loss": 0.0653,
      "step": 6104,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.4478479782092317,
      "lm_loss": 0.6275,
      "step": 6104,
      "vm_loss": 0.189
    },
    {
      "epoch": 0.4478479782092317,
      "lm_loss": 0.9528,
      "step": 6104,
      "vm_loss": 0.1307
    },
    {
      "epoch": 0.4479213477993709,
      "grad_norm": 1.3414206046481885,
      "learning_rate": 6.077492993980821e-06,
      "loss": 0.7392,
      "step": 6105
    },
    {
      "epoch": 0.44799471738950997,
      "grad_norm": 0.5580159580969651,
      "learning_rate": 6.076332684345703e-06,
      "loss": 0.5135,
      "step": 6106
    },
    {
      "epoch": 0.4480680869796491,
      "grad_norm": 1.4728781028926092,
      "learning_rate": 6.075172313927411e-06,
      "loss": 0.7916,
      "step": 6107
    },
    {
      "epoch": 0.4481414565697882,
      "grad_norm": 1.5413292933612577,
      "learning_rate": 6.074011882791476e-06,
      "loss": 0.7059,
      "step": 6108
    },
    {
      "epoch": 0.44821482615992736,
      "grad_norm": 1.5572079107653902,
      "learning_rate": 6.072851391003432e-06,
      "loss": 0.7796,
      "step": 6109
    },
    {
      "epoch": 0.4482881957500665,
      "grad_norm": 1.4339519716090616,
      "learning_rate": 6.071690838628811e-06,
      "loss": 0.7384,
      "step": 6110
    },
    {
      "epoch": 0.4483615653402056,
      "grad_norm": 1.4830045355421364,
      "learning_rate": 6.070530225733156e-06,
      "loss": 0.7553,
      "step": 6111
    },
    {
      "epoch": 0.44843493493034475,
      "grad_norm": 1.5843409553812586,
      "learning_rate": 6.069369552382007e-06,
      "loss": 0.7923,
      "step": 6112
    },
    {
      "epoch": 0.44843493493034475,
      "lm_loss": 0.3842,
      "step": 6112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44843493493034475,
      "lm_loss": 0.551,
      "step": 6112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44843493493034475,
      "lm_loss": 0.6548,
      "step": 6112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44843493493034475,
      "lm_loss": 0.5292,
      "step": 6112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44843493493034475,
      "lm_loss": 0.4242,
      "step": 6112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44843493493034475,
      "lm_loss": 0.6704,
      "step": 6112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44843493493034475,
      "lm_loss": 0.5917,
      "step": 6112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44843493493034475,
      "lm_loss": 0.5689,
      "step": 6112,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.44850830452048385,
      "grad_norm": 0.6995906892715851,
      "learning_rate": 6.068208818640912e-06,
      "loss": 0.4986,
      "step": 6113
    },
    {
      "epoch": 0.448581674110623,
      "grad_norm": 1.4696001899998166,
      "learning_rate": 6.067048024575419e-06,
      "loss": 0.7305,
      "step": 6114
    },
    {
      "epoch": 0.44865504370076215,
      "grad_norm": 1.448229603404823,
      "learning_rate": 6.065887170251081e-06,
      "loss": 0.8773,
      "step": 6115
    },
    {
      "epoch": 0.44872841329090124,
      "grad_norm": 1.5473052101882094,
      "learning_rate": 6.064726255733455e-06,
      "loss": 0.7946,
      "step": 6116
    },
    {
      "epoch": 0.4488017828810404,
      "grad_norm": 0.5570136642758278,
      "learning_rate": 6.063565281088102e-06,
      "loss": 0.4849,
      "step": 6117
    },
    {
      "epoch": 0.4488751524711795,
      "grad_norm": 1.560162862734287,
      "learning_rate": 6.062404246380582e-06,
      "loss": 0.7271,
      "step": 6118
    },
    {
      "epoch": 0.44894852206131863,
      "grad_norm": 1.432775270103786,
      "learning_rate": 6.061243151676465e-06,
      "loss": 0.7828,
      "step": 6119
    },
    {
      "epoch": 0.4490218916514578,
      "grad_norm": 1.4226957310920592,
      "learning_rate": 6.060081997041318e-06,
      "loss": 0.6795,
      "step": 6120
    },
    {
      "epoch": 0.4490218916514578,
      "lm_loss": 0.4376,
      "step": 6120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4490218916514578,
      "lm_loss": 0.5119,
      "step": 6120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4490218916514578,
      "lm_loss": 0.4911,
      "step": 6120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4490218916514578,
      "lm_loss": 0.6338,
      "step": 6120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4490218916514578,
      "lm_loss": 0.5999,
      "step": 6120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4490218916514578,
      "lm_loss": 0.5639,
      "step": 6120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4490218916514578,
      "lm_loss": 0.4913,
      "step": 6120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4490218916514578,
      "lm_loss": 0.4112,
      "step": 6120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4490952612415969,
      "grad_norm": 0.5908616735843708,
      "learning_rate": 6.058920782540714e-06,
      "loss": 0.5287,
      "step": 6121
    },
    {
      "epoch": 0.449168630831736,
      "grad_norm": 1.5532374785190068,
      "learning_rate": 6.057759508240234e-06,
      "loss": 0.7843,
      "step": 6122
    },
    {
      "epoch": 0.4492420004218751,
      "grad_norm": 1.4793690356552043,
      "learning_rate": 6.056598174205454e-06,
      "loss": 0.6913,
      "step": 6123
    },
    {
      "epoch": 0.44931537001201427,
      "grad_norm": 1.4500859865575597,
      "learning_rate": 6.055436780501957e-06,
      "loss": 0.727,
      "step": 6124
    },
    {
      "epoch": 0.4493887396021534,
      "grad_norm": 1.4286402601699708,
      "learning_rate": 6.054275327195332e-06,
      "loss": 0.8518,
      "step": 6125
    },
    {
      "epoch": 0.4494621091922925,
      "grad_norm": 0.5653038460458955,
      "learning_rate": 6.053113814351169e-06,
      "loss": 0.505,
      "step": 6126
    },
    {
      "epoch": 0.44953547878243166,
      "grad_norm": 1.432801275040697,
      "learning_rate": 6.05195224203506e-06,
      "loss": 0.7555,
      "step": 6127
    },
    {
      "epoch": 0.44960884837257076,
      "grad_norm": 0.5691998707209652,
      "learning_rate": 6.0507906103126045e-06,
      "loss": 0.4937,
      "step": 6128
    },
    {
      "epoch": 0.44960884837257076,
      "lm_loss": 1.4423,
      "step": 6128,
      "vm_loss": 0.0431
    },
    {
      "epoch": 0.44960884837257076,
      "lm_loss": 1.5392,
      "step": 6128,
      "vm_loss": 0.1238
    },
    {
      "epoch": 0.44960884837257076,
      "lm_loss": 0.684,
      "step": 6128,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.44960884837257076,
      "lm_loss": 0.705,
      "step": 6128,
      "vm_loss": 0.1742
    },
    {
      "epoch": 0.44960884837257076,
      "lm_loss": 0.821,
      "step": 6128,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.44960884837257076,
      "lm_loss": 0.3979,
      "step": 6128,
      "vm_loss": 0.0888
    },
    {
      "epoch": 0.44960884837257076,
      "lm_loss": 0.073,
      "step": 6128,
      "vm_loss": 0.1383
    },
    {
      "epoch": 0.44960884837257076,
      "lm_loss": 0.9597,
      "step": 6128,
      "vm_loss": 0.1896
    },
    {
      "epoch": 0.4496822179627099,
      "grad_norm": 1.568777807078527,
      "learning_rate": 6.0496289192494e-06,
      "loss": 0.8018,
      "step": 6129
    },
    {
      "epoch": 0.44975558755284906,
      "grad_norm": 0.5640001052459653,
      "learning_rate": 6.048467168911051e-06,
      "loss": 0.4835,
      "step": 6130
    },
    {
      "epoch": 0.44982895714298815,
      "grad_norm": 1.4997944588127308,
      "learning_rate": 6.047305359363165e-06,
      "loss": 0.6838,
      "step": 6131
    },
    {
      "epoch": 0.4499023267331273,
      "grad_norm": 1.5147576356775292,
      "learning_rate": 6.04614349067135e-06,
      "loss": 0.8084,
      "step": 6132
    },
    {
      "epoch": 0.4499756963232664,
      "grad_norm": 0.5838671544991579,
      "learning_rate": 6.044981562901223e-06,
      "loss": 0.5059,
      "step": 6133
    },
    {
      "epoch": 0.45004906591340554,
      "grad_norm": 1.4173856581181041,
      "learning_rate": 6.043819576118399e-06,
      "loss": 0.7211,
      "step": 6134
    },
    {
      "epoch": 0.4501224355035447,
      "grad_norm": 1.5068655039927548,
      "learning_rate": 6.042657530388498e-06,
      "loss": 0.762,
      "step": 6135
    },
    {
      "epoch": 0.4501958050936838,
      "grad_norm": 1.5340383873685794,
      "learning_rate": 6.0414954257771454e-06,
      "loss": 0.7425,
      "step": 6136
    },
    {
      "epoch": 0.4501958050936838,
      "lm_loss": 1.1468,
      "step": 6136,
      "vm_loss": 0.0414
    },
    {
      "epoch": 0.4501958050936838,
      "lm_loss": 0.4028,
      "step": 6136,
      "vm_loss": 0.0463
    },
    {
      "epoch": 0.4501958050936838,
      "lm_loss": 0.1811,
      "step": 6136,
      "vm_loss": 0.08
    },
    {
      "epoch": 0.4501958050936838,
      "lm_loss": 0.3074,
      "step": 6136,
      "vm_loss": 0.1903
    },
    {
      "epoch": 0.4501958050936838,
      "lm_loss": 0.6175,
      "step": 6136,
      "vm_loss": 0.0826
    },
    {
      "epoch": 0.4501958050936838,
      "lm_loss": 0.0956,
      "step": 6136,
      "vm_loss": 0.2119
    },
    {
      "epoch": 0.4501958050936838,
      "lm_loss": 0.7947,
      "step": 6136,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.4501958050936838,
      "lm_loss": 0.5162,
      "step": 6136,
      "vm_loss": 0.1357
    },
    {
      "epoch": 0.45026917468382294,
      "grad_norm": 1.5814152935578478,
      "learning_rate": 6.040333262349965e-06,
      "loss": 0.7704,
      "step": 6137
    },
    {
      "epoch": 0.45034254427396203,
      "grad_norm": 1.546584092416967,
      "learning_rate": 6.039171040172591e-06,
      "loss": 0.732,
      "step": 6138
    },
    {
      "epoch": 0.4504159138641012,
      "grad_norm": 1.6820292784248705,
      "learning_rate": 6.038008759310654e-06,
      "loss": 0.8244,
      "step": 6139
    },
    {
      "epoch": 0.45048928345424033,
      "grad_norm": 1.4749215299358736,
      "learning_rate": 6.036846419829792e-06,
      "loss": 0.8398,
      "step": 6140
    },
    {
      "epoch": 0.4505626530443794,
      "grad_norm": 0.6317648871002508,
      "learning_rate": 6.0356840217956435e-06,
      "loss": 0.4989,
      "step": 6141
    },
    {
      "epoch": 0.4506360226345186,
      "grad_norm": 1.2679837620494327,
      "learning_rate": 6.034521565273856e-06,
      "loss": 0.842,
      "step": 6142
    },
    {
      "epoch": 0.45070939222465767,
      "grad_norm": 1.5555933702838383,
      "learning_rate": 6.033359050330071e-06,
      "loss": 0.8856,
      "step": 6143
    },
    {
      "epoch": 0.4507827618147968,
      "grad_norm": 1.3637630736500173,
      "learning_rate": 6.032196477029944e-06,
      "loss": 0.7416,
      "step": 6144
    },
    {
      "epoch": 0.4507827618147968,
      "lm_loss": 1.1772,
      "step": 6144,
      "vm_loss": 0.0331
    },
    {
      "epoch": 0.4507827618147968,
      "lm_loss": 1.8187,
      "step": 6144,
      "vm_loss": 0.1615
    },
    {
      "epoch": 0.4507827618147968,
      "lm_loss": 0.0543,
      "step": 6144,
      "vm_loss": 0.066
    },
    {
      "epoch": 0.4507827618147968,
      "lm_loss": 1.5795,
      "step": 6144,
      "vm_loss": 0.1848
    },
    {
      "epoch": 0.4507827618147968,
      "lm_loss": 0.028,
      "step": 6144,
      "vm_loss": 0.0453
    },
    {
      "epoch": 0.4507827618147968,
      "lm_loss": 1.4309,
      "step": 6144,
      "vm_loss": 0.092
    },
    {
      "epoch": 0.4507827618147968,
      "lm_loss": 0.1552,
      "step": 6144,
      "vm_loss": 0.1931
    },
    {
      "epoch": 0.4507827618147968,
      "lm_loss": 0.9553,
      "step": 6144,
      "vm_loss": 0.1567
    },
    {
      "epoch": 0.4508561314049359,
      "grad_norm": 1.4008899366006673,
      "learning_rate": 6.031033845439122e-06,
      "loss": 0.746,
      "step": 6145
    },
    {
      "epoch": 0.45092950099507506,
      "grad_norm": 1.5046526904597506,
      "learning_rate": 6.029871155623269e-06,
      "loss": 0.7725,
      "step": 6146
    },
    {
      "epoch": 0.4510028705852142,
      "grad_norm": 0.5707208232536315,
      "learning_rate": 6.0287084076480404e-06,
      "loss": 0.5011,
      "step": 6147
    },
    {
      "epoch": 0.4510762401753533,
      "grad_norm": 0.6159658197868358,
      "learning_rate": 6.0275456015791e-06,
      "loss": 0.5174,
      "step": 6148
    },
    {
      "epoch": 0.45114960976549245,
      "grad_norm": 0.5926254712115961,
      "learning_rate": 6.026382737482116e-06,
      "loss": 0.5043,
      "step": 6149
    },
    {
      "epoch": 0.45122297935563155,
      "grad_norm": 1.6096590577888292,
      "learning_rate": 6.0252198154227574e-06,
      "loss": 0.8319,
      "step": 6150
    },
    {
      "epoch": 0.4512963489457707,
      "grad_norm": 1.4991832451127804,
      "learning_rate": 6.024056835466694e-06,
      "loss": 0.7907,
      "step": 6151
    },
    {
      "epoch": 0.45136971853590985,
      "grad_norm": 1.5622779763771,
      "learning_rate": 6.022893797679607e-06,
      "loss": 0.7869,
      "step": 6152
    },
    {
      "epoch": 0.45136971853590985,
      "lm_loss": 1.6363,
      "step": 6152,
      "vm_loss": 0.0706
    },
    {
      "epoch": 0.45136971853590985,
      "lm_loss": 0.0574,
      "step": 6152,
      "vm_loss": 0.1202
    },
    {
      "epoch": 0.45136971853590985,
      "lm_loss": 0.3386,
      "step": 6152,
      "vm_loss": 0.1297
    },
    {
      "epoch": 0.45136971853590985,
      "lm_loss": 0.8231,
      "step": 6152,
      "vm_loss": 0.1633
    },
    {
      "epoch": 0.45136971853590985,
      "lm_loss": 0.5479,
      "step": 6152,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.45136971853590985,
      "lm_loss": 0.5093,
      "step": 6152,
      "vm_loss": 0.1693
    },
    {
      "epoch": 0.45136971853590985,
      "lm_loss": 0.0343,
      "step": 6152,
      "vm_loss": 0.1619
    },
    {
      "epoch": 0.45136971853590985,
      "lm_loss": 0.3112,
      "step": 6152,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.45144308812604894,
      "grad_norm": 1.396648282645587,
      "learning_rate": 6.021730702127173e-06,
      "loss": 0.8409,
      "step": 6153
    },
    {
      "epoch": 0.4515164577161881,
      "grad_norm": 1.6947219466347991,
      "learning_rate": 6.020567548875078e-06,
      "loss": 0.8154,
      "step": 6154
    },
    {
      "epoch": 0.4515898273063272,
      "grad_norm": 0.6881570353113735,
      "learning_rate": 6.019404337989005e-06,
      "loss": 0.5022,
      "step": 6155
    },
    {
      "epoch": 0.45166319689646633,
      "grad_norm": 0.6674054293689451,
      "learning_rate": 6.018241069534645e-06,
      "loss": 0.5038,
      "step": 6156
    },
    {
      "epoch": 0.4517365664866055,
      "grad_norm": 0.5784818213281939,
      "learning_rate": 6.017077743577691e-06,
      "loss": 0.508,
      "step": 6157
    },
    {
      "epoch": 0.4518099360767446,
      "grad_norm": 1.4974470888693856,
      "learning_rate": 6.015914360183836e-06,
      "loss": 0.8208,
      "step": 6158
    },
    {
      "epoch": 0.45188330566688373,
      "grad_norm": 1.4557584999019684,
      "learning_rate": 6.0147509194187815e-06,
      "loss": 0.7639,
      "step": 6159
    },
    {
      "epoch": 0.4519566752570228,
      "grad_norm": 1.5123767055340505,
      "learning_rate": 6.01358742134823e-06,
      "loss": 0.7823,
      "step": 6160
    },
    {
      "epoch": 0.4519566752570228,
      "lm_loss": 1.3273,
      "step": 6160,
      "vm_loss": 0.0677
    },
    {
      "epoch": 0.4519566752570228,
      "lm_loss": 1.2279,
      "step": 6160,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.4519566752570228,
      "lm_loss": 0.5862,
      "step": 6160,
      "vm_loss": 0.1352
    },
    {
      "epoch": 0.4519566752570228,
      "lm_loss": 1.3164,
      "step": 6160,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.4519566752570228,
      "lm_loss": 0.8606,
      "step": 6160,
      "vm_loss": 0.2523
    },
    {
      "epoch": 0.4519566752570228,
      "lm_loss": 0.4251,
      "step": 6160,
      "vm_loss": 0.1605
    },
    {
      "epoch": 0.4519566752570228,
      "lm_loss": 0.9236,
      "step": 6160,
      "vm_loss": 0.1293
    },
    {
      "epoch": 0.4519566752570228,
      "lm_loss": 1.1895,
      "step": 6160,
      "vm_loss": 0.1974
    },
    {
      "epoch": 0.45203004484716197,
      "grad_norm": 1.6943484626657603,
      "learning_rate": 6.012423866037886e-06,
      "loss": 0.7984,
      "step": 6161
    },
    {
      "epoch": 0.4521034144373011,
      "grad_norm": 1.6709652398317374,
      "learning_rate": 6.0112602535534616e-06,
      "loss": 0.8263,
      "step": 6162
    },
    {
      "epoch": 0.4521767840274402,
      "grad_norm": 1.5577869930406347,
      "learning_rate": 6.010096583960664e-06,
      "loss": 0.8842,
      "step": 6163
    },
    {
      "epoch": 0.45225015361757936,
      "grad_norm": 0.9180780511706537,
      "learning_rate": 6.008932857325212e-06,
      "loss": 0.5126,
      "step": 6164
    },
    {
      "epoch": 0.45232352320771846,
      "grad_norm": 1.5511362058457037,
      "learning_rate": 6.007769073712822e-06,
      "loss": 0.8695,
      "step": 6165
    },
    {
      "epoch": 0.4523968927978576,
      "grad_norm": 1.3053041594727455,
      "learning_rate": 6.006605233189217e-06,
      "loss": 0.757,
      "step": 6166
    },
    {
      "epoch": 0.45247026238799676,
      "grad_norm": 0.6013580027151822,
      "learning_rate": 6.005441335820122e-06,
      "loss": 0.4951,
      "step": 6167
    },
    {
      "epoch": 0.45254363197813585,
      "grad_norm": 1.5218727046088103,
      "learning_rate": 6.004277381671264e-06,
      "loss": 0.701,
      "step": 6168
    },
    {
      "epoch": 0.45254363197813585,
      "lm_loss": 1.6779,
      "step": 6168,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.45254363197813585,
      "lm_loss": 1.3503,
      "step": 6168,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.45254363197813585,
      "lm_loss": 0.3412,
      "step": 6168,
      "vm_loss": 0.107
    },
    {
      "epoch": 0.45254363197813585,
      "lm_loss": 0.7753,
      "step": 6168,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.45254363197813585,
      "lm_loss": 1.0151,
      "step": 6168,
      "vm_loss": 0.228
    },
    {
      "epoch": 0.45254363197813585,
      "lm_loss": 1.0033,
      "step": 6168,
      "vm_loss": 0.1049
    },
    {
      "epoch": 0.45254363197813585,
      "lm_loss": 0.8453,
      "step": 6168,
      "vm_loss": 0.123
    },
    {
      "epoch": 0.45254363197813585,
      "lm_loss": 0.8819,
      "step": 6168,
      "vm_loss": 0.1296
    },
    {
      "epoch": 0.452617001568275,
      "grad_norm": 1.59050846824812,
      "learning_rate": 6.003113370808375e-06,
      "loss": 0.9417,
      "step": 6169
    },
    {
      "epoch": 0.4526903711584141,
      "grad_norm": 1.620974496231062,
      "learning_rate": 6.001949303297191e-06,
      "loss": 0.7677,
      "step": 6170
    },
    {
      "epoch": 0.45276374074855325,
      "grad_norm": 0.7325874507229488,
      "learning_rate": 6.000785179203449e-06,
      "loss": 0.5185,
      "step": 6171
    },
    {
      "epoch": 0.4528371103386924,
      "grad_norm": 1.4019178232812806,
      "learning_rate": 5.999620998592888e-06,
      "loss": 0.8016,
      "step": 6172
    },
    {
      "epoch": 0.4529104799288315,
      "grad_norm": 1.4614432877668833,
      "learning_rate": 5.998456761531255e-06,
      "loss": 0.7497,
      "step": 6173
    },
    {
      "epoch": 0.45298384951897064,
      "grad_norm": 0.6659848696960153,
      "learning_rate": 5.9972924680842945e-06,
      "loss": 0.4934,
      "step": 6174
    },
    {
      "epoch": 0.45305721910910973,
      "grad_norm": 0.6191592866047662,
      "learning_rate": 5.99612811831776e-06,
      "loss": 0.4944,
      "step": 6175
    },
    {
      "epoch": 0.4531305886992489,
      "grad_norm": 1.805062775485718,
      "learning_rate": 5.994963712297403e-06,
      "loss": 0.845,
      "step": 6176
    },
    {
      "epoch": 0.4531305886992489,
      "lm_loss": 1.4691,
      "step": 6176,
      "vm_loss": 0.102
    },
    {
      "epoch": 0.4531305886992489,
      "lm_loss": 1.5056,
      "step": 6176,
      "vm_loss": 0.0625
    },
    {
      "epoch": 0.4531305886992489,
      "lm_loss": 1.3583,
      "step": 6176,
      "vm_loss": 0.1482
    },
    {
      "epoch": 0.4531305886992489,
      "lm_loss": 0.0287,
      "step": 6176,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.4531305886992489,
      "lm_loss": 0.3413,
      "step": 6176,
      "vm_loss": 0.1327
    },
    {
      "epoch": 0.4531305886992489,
      "lm_loss": 0.5332,
      "step": 6176,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.4531305886992489,
      "lm_loss": 0.8451,
      "step": 6176,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.4531305886992489,
      "lm_loss": 0.1361,
      "step": 6176,
      "vm_loss": 0.0453
    },
    {
      "epoch": 0.45320395828938803,
      "grad_norm": 1.6336790672246784,
      "learning_rate": 5.9937992500889805e-06,
      "loss": 0.7846,
      "step": 6177
    },
    {
      "epoch": 0.4532773278795271,
      "grad_norm": 1.4394579552953604,
      "learning_rate": 5.992634731758255e-06,
      "loss": 0.7543,
      "step": 6178
    },
    {
      "epoch": 0.4533506974696663,
      "grad_norm": 1.6528778668898212,
      "learning_rate": 5.991470157370988e-06,
      "loss": 0.7692,
      "step": 6179
    },
    {
      "epoch": 0.45342406705980537,
      "grad_norm": 0.7539278458891947,
      "learning_rate": 5.990305526992944e-06,
      "loss": 0.5146,
      "step": 6180
    },
    {
      "epoch": 0.4534974366499445,
      "grad_norm": 1.7164979919813796,
      "learning_rate": 5.9891408406898955e-06,
      "loss": 0.8089,
      "step": 6181
    },
    {
      "epoch": 0.45357080624008367,
      "grad_norm": 1.4490095721191503,
      "learning_rate": 5.987976098527614e-06,
      "loss": 0.7699,
      "step": 6182
    },
    {
      "epoch": 0.45364417583022276,
      "grad_norm": 0.6494751535489776,
      "learning_rate": 5.986811300571875e-06,
      "loss": 0.5098,
      "step": 6183
    },
    {
      "epoch": 0.4537175454203619,
      "grad_norm": 0.5868407802152681,
      "learning_rate": 5.98564644688846e-06,
      "loss": 0.5032,
      "step": 6184
    },
    {
      "epoch": 0.4537175454203619,
      "lm_loss": 1.5874,
      "step": 6184,
      "vm_loss": 0.1009
    },
    {
      "epoch": 0.4537175454203619,
      "lm_loss": 1.3029,
      "step": 6184,
      "vm_loss": 0.1121
    },
    {
      "epoch": 0.4537175454203619,
      "lm_loss": 0.8822,
      "step": 6184,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.4537175454203619,
      "lm_loss": 0.1328,
      "step": 6184,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.4537175454203619,
      "lm_loss": 0.1382,
      "step": 6184,
      "vm_loss": 0.1289
    },
    {
      "epoch": 0.4537175454203619,
      "lm_loss": 0.078,
      "step": 6184,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.4537175454203619,
      "lm_loss": 0.2907,
      "step": 6184,
      "vm_loss": 0.0568
    },
    {
      "epoch": 0.4537175454203619,
      "lm_loss": 1.1472,
      "step": 6184,
      "vm_loss": 0.1384
    },
    {
      "epoch": 0.453790915010501,
      "grad_norm": 1.7313465635695993,
      "learning_rate": 5.9844815375431475e-06,
      "loss": 0.8157,
      "step": 6185
    },
    {
      "epoch": 0.45386428460064016,
      "grad_norm": 1.7211851319915805,
      "learning_rate": 5.983316572601725e-06,
      "loss": 0.7738,
      "step": 6186
    },
    {
      "epoch": 0.45393765419077925,
      "grad_norm": 1.612479335971223,
      "learning_rate": 5.982151552129981e-06,
      "loss": 0.7466,
      "step": 6187
    },
    {
      "epoch": 0.4540110237809184,
      "grad_norm": 0.6621633841058219,
      "learning_rate": 5.980986476193705e-06,
      "loss": 0.5097,
      "step": 6188
    },
    {
      "epoch": 0.45408439337105755,
      "grad_norm": 1.5636607443473152,
      "learning_rate": 5.979821344858695e-06,
      "loss": 0.7446,
      "step": 6189
    },
    {
      "epoch": 0.45415776296119664,
      "grad_norm": 0.6386036712754091,
      "learning_rate": 5.978656158190748e-06,
      "loss": 0.5164,
      "step": 6190
    },
    {
      "epoch": 0.4542311325513358,
      "grad_norm": 1.5748263101554734,
      "learning_rate": 5.977490916255661e-06,
      "loss": 0.8042,
      "step": 6191
    },
    {
      "epoch": 0.4543045021414749,
      "grad_norm": 1.294658912075087,
      "learning_rate": 5.976325619119243e-06,
      "loss": 0.7542,
      "step": 6192
    },
    {
      "epoch": 0.4543045021414749,
      "lm_loss": 0.5195,
      "step": 6192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4543045021414749,
      "lm_loss": 0.6004,
      "step": 6192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4543045021414749,
      "lm_loss": 0.3865,
      "step": 6192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4543045021414749,
      "lm_loss": 0.5449,
      "step": 6192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4543045021414749,
      "lm_loss": 0.5852,
      "step": 6192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4543045021414749,
      "lm_loss": 0.4099,
      "step": 6192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4543045021414749,
      "lm_loss": 0.5235,
      "step": 6192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4543045021414749,
      "lm_loss": 0.5134,
      "step": 6192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45437787173161404,
      "grad_norm": 0.6059603794830678,
      "learning_rate": 5.975160266847299e-06,
      "loss": 0.5119,
      "step": 6193
    },
    {
      "epoch": 0.4544512413217532,
      "grad_norm": 1.6760200749446834,
      "learning_rate": 5.9739948595056405e-06,
      "loss": 0.8779,
      "step": 6194
    },
    {
      "epoch": 0.4545246109118923,
      "grad_norm": 1.5752300710815739,
      "learning_rate": 5.97282939716008e-06,
      "loss": 0.7924,
      "step": 6195
    },
    {
      "epoch": 0.45459798050203143,
      "grad_norm": 1.7985509403490607,
      "learning_rate": 5.971663879876433e-06,
      "loss": 0.7665,
      "step": 6196
    },
    {
      "epoch": 0.4546713500921705,
      "grad_norm": 0.626515308144315,
      "learning_rate": 5.970498307720523e-06,
      "loss": 0.5025,
      "step": 6197
    },
    {
      "epoch": 0.4547447196823097,
      "grad_norm": 0.5897054301134134,
      "learning_rate": 5.96933268075817e-06,
      "loss": 0.5011,
      "step": 6198
    },
    {
      "epoch": 0.4548180892724488,
      "grad_norm": 1.7371411804706944,
      "learning_rate": 5.968166999055197e-06,
      "loss": 0.7753,
      "step": 6199
    },
    {
      "epoch": 0.4548914588625879,
      "grad_norm": 0.5648862203010664,
      "learning_rate": 5.967001262677439e-06,
      "loss": 0.5012,
      "step": 6200
    },
    {
      "epoch": 0.4548914588625879,
      "lm_loss": 1.1938,
      "step": 6200,
      "vm_loss": 0.0496
    },
    {
      "epoch": 0.4548914588625879,
      "lm_loss": 0.6144,
      "step": 6200,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.4548914588625879,
      "lm_loss": 0.1633,
      "step": 6200,
      "vm_loss": 0.148
    },
    {
      "epoch": 0.4548914588625879,
      "lm_loss": 0.0968,
      "step": 6200,
      "vm_loss": 0.1612
    },
    {
      "epoch": 0.4548914588625879,
      "lm_loss": 0.6921,
      "step": 6200,
      "vm_loss": 0.1701
    },
    {
      "epoch": 0.4548914588625879,
      "lm_loss": 0.0467,
      "step": 6200,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.4548914588625879,
      "lm_loss": 0.4201,
      "step": 6200,
      "vm_loss": 0.0585
    },
    {
      "epoch": 0.4548914588625879,
      "lm_loss": 0.0793,
      "step": 6200,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.45496482845272707,
      "grad_norm": 1.4813959009905335,
      "learning_rate": 5.965835471690723e-06,
      "loss": 0.6959,
      "step": 6201
    },
    {
      "epoch": 0.45503819804286616,
      "grad_norm": 0.5866214025960738,
      "learning_rate": 5.9646696261608875e-06,
      "loss": 0.5226,
      "step": 6202
    },
    {
      "epoch": 0.4551115676330053,
      "grad_norm": 1.4076791917075064,
      "learning_rate": 5.963503726153768e-06,
      "loss": 0.708,
      "step": 6203
    },
    {
      "epoch": 0.45518493722314446,
      "grad_norm": 1.848564029961764,
      "learning_rate": 5.962337771735207e-06,
      "loss": 0.7754,
      "step": 6204
    },
    {
      "epoch": 0.45525830681328355,
      "grad_norm": 0.5859482320504242,
      "learning_rate": 5.96117176297105e-06,
      "loss": 0.514,
      "step": 6205
    },
    {
      "epoch": 0.4553316764034227,
      "grad_norm": 1.2860136567177878,
      "learning_rate": 5.960005699927143e-06,
      "loss": 0.8052,
      "step": 6206
    },
    {
      "epoch": 0.4554050459935618,
      "grad_norm": 1.52032494675685,
      "learning_rate": 5.958839582669336e-06,
      "loss": 0.8766,
      "step": 6207
    },
    {
      "epoch": 0.45547841558370095,
      "grad_norm": 0.564046271063497,
      "learning_rate": 5.957673411263484e-06,
      "loss": 0.4978,
      "step": 6208
    },
    {
      "epoch": 0.45547841558370095,
      "lm_loss": 1.8746,
      "step": 6208,
      "vm_loss": 0.0626
    },
    {
      "epoch": 0.45547841558370095,
      "lm_loss": 1.4854,
      "step": 6208,
      "vm_loss": 0.046
    },
    {
      "epoch": 0.45547841558370095,
      "lm_loss": 0.2722,
      "step": 6208,
      "vm_loss": 0.089
    },
    {
      "epoch": 0.45547841558370095,
      "lm_loss": 1.0748,
      "step": 6208,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.45547841558370095,
      "lm_loss": 0.7722,
      "step": 6208,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.45547841558370095,
      "lm_loss": 0.4363,
      "step": 6208,
      "vm_loss": 0.1312
    },
    {
      "epoch": 0.45547841558370095,
      "lm_loss": 0.618,
      "step": 6208,
      "vm_loss": 0.1842
    },
    {
      "epoch": 0.45547841558370095,
      "lm_loss": 1.2121,
      "step": 6208,
      "vm_loss": 0.1761
    },
    {
      "epoch": 0.4555517851738401,
      "grad_norm": 1.7033087919521503,
      "learning_rate": 5.956507185775441e-06,
      "loss": 0.7939,
      "step": 6209
    },
    {
      "epoch": 0.4556251547639792,
      "grad_norm": 1.7038129272373181,
      "learning_rate": 5.9553409062710695e-06,
      "loss": 0.7576,
      "step": 6210
    },
    {
      "epoch": 0.45569852435411834,
      "grad_norm": 1.6756947564563853,
      "learning_rate": 5.9541745728162325e-06,
      "loss": 0.8374,
      "step": 6211
    },
    {
      "epoch": 0.45577189394425743,
      "grad_norm": 0.5663809848125472,
      "learning_rate": 5.953008185476793e-06,
      "loss": 0.5088,
      "step": 6212
    },
    {
      "epoch": 0.4558452635343966,
      "grad_norm": 1.3982274141733877,
      "learning_rate": 5.951841744318622e-06,
      "loss": 0.6891,
      "step": 6213
    },
    {
      "epoch": 0.45591863312453573,
      "grad_norm": 1.5425936950044543,
      "learning_rate": 5.9506752494075905e-06,
      "loss": 0.7761,
      "step": 6214
    },
    {
      "epoch": 0.4559920027146748,
      "grad_norm": 0.573157241855304,
      "learning_rate": 5.949508700809572e-06,
      "loss": 0.5102,
      "step": 6215
    },
    {
      "epoch": 0.456065372304814,
      "grad_norm": 1.6037508740244526,
      "learning_rate": 5.9483420985904475e-06,
      "loss": 0.8711,
      "step": 6216
    },
    {
      "epoch": 0.456065372304814,
      "lm_loss": 1.2021,
      "step": 6216,
      "vm_loss": 0.0517
    },
    {
      "epoch": 0.456065372304814,
      "lm_loss": 0.8234,
      "step": 6216,
      "vm_loss": 0.1333
    },
    {
      "epoch": 0.456065372304814,
      "lm_loss": 0.9925,
      "step": 6216,
      "vm_loss": 0.1332
    },
    {
      "epoch": 0.456065372304814,
      "lm_loss": 0.6052,
      "step": 6216,
      "vm_loss": 0.1661
    },
    {
      "epoch": 0.456065372304814,
      "lm_loss": 0.9222,
      "step": 6216,
      "vm_loss": 0.1578
    },
    {
      "epoch": 0.456065372304814,
      "lm_loss": 0.1093,
      "step": 6216,
      "vm_loss": 0.1049
    },
    {
      "epoch": 0.456065372304814,
      "lm_loss": 0.5259,
      "step": 6216,
      "vm_loss": 0.1887
    },
    {
      "epoch": 0.456065372304814,
      "lm_loss": 0.4974,
      "step": 6216,
      "vm_loss": 0.0439
    },
    {
      "epoch": 0.45613874189495307,
      "grad_norm": 1.501124467686493,
      "learning_rate": 5.947175442816095e-06,
      "loss": 0.8057,
      "step": 6217
    },
    {
      "epoch": 0.4562121114850922,
      "grad_norm": 1.6441810523537854,
      "learning_rate": 5.9460087335524e-06,
      "loss": 0.8059,
      "step": 6218
    },
    {
      "epoch": 0.45628548107523137,
      "grad_norm": 0.5855764912054624,
      "learning_rate": 5.944841970865249e-06,
      "loss": 0.4999,
      "step": 6219
    },
    {
      "epoch": 0.45635885066537046,
      "grad_norm": 1.5245085382700296,
      "learning_rate": 5.943675154820532e-06,
      "loss": 0.7207,
      "step": 6220
    },
    {
      "epoch": 0.4564322202555096,
      "grad_norm": 1.5614430640703203,
      "learning_rate": 5.942508285484142e-06,
      "loss": 0.8401,
      "step": 6221
    },
    {
      "epoch": 0.4565055898456487,
      "grad_norm": 1.5413192313932467,
      "learning_rate": 5.941341362921975e-06,
      "loss": 0.7946,
      "step": 6222
    },
    {
      "epoch": 0.45657895943578786,
      "grad_norm": 1.5610020442436563,
      "learning_rate": 5.940174387199931e-06,
      "loss": 0.8136,
      "step": 6223
    },
    {
      "epoch": 0.456652329025927,
      "grad_norm": 1.3728672004216074,
      "learning_rate": 5.93900735838391e-06,
      "loss": 0.7972,
      "step": 6224
    },
    {
      "epoch": 0.456652329025927,
      "lm_loss": 0.4007,
      "step": 6224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.456652329025927,
      "lm_loss": 0.4369,
      "step": 6224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.456652329025927,
      "lm_loss": 0.626,
      "step": 6224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.456652329025927,
      "lm_loss": 0.4618,
      "step": 6224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.456652329025927,
      "lm_loss": 0.629,
      "step": 6224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.456652329025927,
      "lm_loss": 0.5486,
      "step": 6224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.456652329025927,
      "lm_loss": 0.5251,
      "step": 6224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.456652329025927,
      "lm_loss": 0.5566,
      "step": 6224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4567256986160661,
      "grad_norm": 0.5729741905214883,
      "learning_rate": 5.937840276539818e-06,
      "loss": 0.4967,
      "step": 6225
    },
    {
      "epoch": 0.45679906820620525,
      "grad_norm": 1.4186931741350512,
      "learning_rate": 5.936673141733563e-06,
      "loss": 0.8236,
      "step": 6226
    },
    {
      "epoch": 0.45687243779634434,
      "grad_norm": 0.5812203031687342,
      "learning_rate": 5.935505954031057e-06,
      "loss": 0.5109,
      "step": 6227
    },
    {
      "epoch": 0.4569458073864835,
      "grad_norm": 1.3147101988333818,
      "learning_rate": 5.9343387134982115e-06,
      "loss": 0.7133,
      "step": 6228
    },
    {
      "epoch": 0.4570191769766226,
      "grad_norm": 1.6905748775337686,
      "learning_rate": 5.933171420200946e-06,
      "loss": 0.8228,
      "step": 6229
    },
    {
      "epoch": 0.45709254656676174,
      "grad_norm": 1.6347121186969145,
      "learning_rate": 5.93200407420518e-06,
      "loss": 0.8333,
      "step": 6230
    },
    {
      "epoch": 0.4571659161569009,
      "grad_norm": 0.5689761749825938,
      "learning_rate": 5.930836675576836e-06,
      "loss": 0.5215,
      "step": 6231
    },
    {
      "epoch": 0.45723928574704,
      "grad_norm": 1.4048349188614593,
      "learning_rate": 5.92966922438184e-06,
      "loss": 0.7154,
      "step": 6232
    },
    {
      "epoch": 0.45723928574704,
      "lm_loss": 0.5303,
      "step": 6232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45723928574704,
      "lm_loss": 0.675,
      "step": 6232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45723928574704,
      "lm_loss": 0.4371,
      "step": 6232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45723928574704,
      "lm_loss": 0.4618,
      "step": 6232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45723928574704,
      "lm_loss": 0.4914,
      "step": 6232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45723928574704,
      "lm_loss": 0.6163,
      "step": 6232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45723928574704,
      "lm_loss": 0.584,
      "step": 6232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45723928574704,
      "lm_loss": 0.4121,
      "step": 6232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45731265533717913,
      "grad_norm": 0.561715460112472,
      "learning_rate": 5.928501720686118e-06,
      "loss": 0.5169,
      "step": 6233
    },
    {
      "epoch": 0.4573860249273182,
      "grad_norm": 1.4502884833426797,
      "learning_rate": 5.927334164555607e-06,
      "loss": 0.7569,
      "step": 6234
    },
    {
      "epoch": 0.4574593945174574,
      "grad_norm": 1.4224932865908104,
      "learning_rate": 5.926166556056239e-06,
      "loss": 0.8313,
      "step": 6235
    },
    {
      "epoch": 0.4575327641075965,
      "grad_norm": 1.5356836823992257,
      "learning_rate": 5.92499889525395e-06,
      "loss": 0.7812,
      "step": 6236
    },
    {
      "epoch": 0.4576061336977356,
      "grad_norm": 1.365666642996389,
      "learning_rate": 5.9238311822146845e-06,
      "loss": 0.7475,
      "step": 6237
    },
    {
      "epoch": 0.45767950328787477,
      "grad_norm": 0.5720707430858204,
      "learning_rate": 5.922663417004384e-06,
      "loss": 0.5071,
      "step": 6238
    },
    {
      "epoch": 0.45775287287801386,
      "grad_norm": 0.5774341900714597,
      "learning_rate": 5.921495599688994e-06,
      "loss": 0.522,
      "step": 6239
    },
    {
      "epoch": 0.457826242468153,
      "grad_norm": 1.6168652352968824,
      "learning_rate": 5.920327730334467e-06,
      "loss": 0.7797,
      "step": 6240
    },
    {
      "epoch": 0.457826242468153,
      "lm_loss": 0.5162,
      "step": 6240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.457826242468153,
      "lm_loss": 0.4953,
      "step": 6240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.457826242468153,
      "lm_loss": 0.5293,
      "step": 6240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.457826242468153,
      "lm_loss": 0.4864,
      "step": 6240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.457826242468153,
      "lm_loss": 0.5794,
      "step": 6240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.457826242468153,
      "lm_loss": 0.3394,
      "step": 6240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.457826242468153,
      "lm_loss": 0.4818,
      "step": 6240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.457826242468153,
      "lm_loss": 0.6277,
      "step": 6240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45789961205829216,
      "grad_norm": 0.5608029887637812,
      "learning_rate": 5.919159809006753e-06,
      "loss": 0.4917,
      "step": 6241
    },
    {
      "epoch": 0.45797298164843125,
      "grad_norm": 1.5252736108254985,
      "learning_rate": 5.917991835771809e-06,
      "loss": 0.844,
      "step": 6242
    },
    {
      "epoch": 0.4580463512385704,
      "grad_norm": 1.358931978596048,
      "learning_rate": 5.916823810695593e-06,
      "loss": 0.8226,
      "step": 6243
    },
    {
      "epoch": 0.4581197208287095,
      "grad_norm": 1.5165202261044766,
      "learning_rate": 5.915655733844064e-06,
      "loss": 0.8349,
      "step": 6244
    },
    {
      "epoch": 0.45819309041884865,
      "grad_norm": 1.480479645834584,
      "learning_rate": 5.91448760528319e-06,
      "loss": 0.7458,
      "step": 6245
    },
    {
      "epoch": 0.4582664600089878,
      "grad_norm": 1.5009790091331605,
      "learning_rate": 5.913319425078937e-06,
      "loss": 0.7438,
      "step": 6246
    },
    {
      "epoch": 0.4583398295991269,
      "grad_norm": 1.3813158322846126,
      "learning_rate": 5.912151193297272e-06,
      "loss": 0.7841,
      "step": 6247
    },
    {
      "epoch": 0.45841319918926604,
      "grad_norm": 1.593768383974533,
      "learning_rate": 5.9109829100041715e-06,
      "loss": 0.826,
      "step": 6248
    },
    {
      "epoch": 0.45841319918926604,
      "lm_loss": 0.6291,
      "step": 6248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45841319918926604,
      "lm_loss": 0.4115,
      "step": 6248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45841319918926604,
      "lm_loss": 0.4781,
      "step": 6248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45841319918926604,
      "lm_loss": 0.5608,
      "step": 6248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45841319918926604,
      "lm_loss": 0.6631,
      "step": 6248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45841319918926604,
      "lm_loss": 0.4653,
      "step": 6248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45841319918926604,
      "lm_loss": 0.4754,
      "step": 6248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45841319918926604,
      "lm_loss": 0.5118,
      "step": 6248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45848656877940513,
      "grad_norm": 0.5636614069786605,
      "learning_rate": 5.909814575265609e-06,
      "loss": 0.4937,
      "step": 6249
    },
    {
      "epoch": 0.4585599383695443,
      "grad_norm": 1.4325547699975014,
      "learning_rate": 5.908646189147565e-06,
      "loss": 0.7401,
      "step": 6250
    },
    {
      "epoch": 0.45863330795968343,
      "grad_norm": 1.4147378409354558,
      "learning_rate": 5.907477751716022e-06,
      "loss": 0.7059,
      "step": 6251
    },
    {
      "epoch": 0.45870667754982253,
      "grad_norm": 1.4558195624873187,
      "learning_rate": 5.906309263036961e-06,
      "loss": 0.6934,
      "step": 6252
    },
    {
      "epoch": 0.4587800471399617,
      "grad_norm": 1.606912755458329,
      "learning_rate": 5.905140723176373e-06,
      "loss": 0.7825,
      "step": 6253
    },
    {
      "epoch": 0.45885341673010077,
      "grad_norm": 0.5742259559414922,
      "learning_rate": 5.903972132200245e-06,
      "loss": 0.5023,
      "step": 6254
    },
    {
      "epoch": 0.4589267863202399,
      "grad_norm": 0.570644677888578,
      "learning_rate": 5.902803490174574e-06,
      "loss": 0.5119,
      "step": 6255
    },
    {
      "epoch": 0.45900015591037907,
      "grad_norm": 0.544303545153188,
      "learning_rate": 5.901634797165352e-06,
      "loss": 0.5123,
      "step": 6256
    },
    {
      "epoch": 0.45900015591037907,
      "lm_loss": 0.6798,
      "step": 6256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45900015591037907,
      "lm_loss": 0.5284,
      "step": 6256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45900015591037907,
      "lm_loss": 0.4666,
      "step": 6256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45900015591037907,
      "lm_loss": 0.3559,
      "step": 6256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45900015591037907,
      "lm_loss": 0.6391,
      "step": 6256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45900015591037907,
      "lm_loss": 0.3925,
      "step": 6256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45900015591037907,
      "lm_loss": 0.5976,
      "step": 6256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45900015591037907,
      "lm_loss": 0.5392,
      "step": 6256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.45907352550051816,
      "grad_norm": 0.5618381405963279,
      "learning_rate": 5.9004660532385805e-06,
      "loss": 0.4964,
      "step": 6257
    },
    {
      "epoch": 0.4591468950906573,
      "grad_norm": 1.6383019496465434,
      "learning_rate": 5.899297258460263e-06,
      "loss": 0.8656,
      "step": 6258
    },
    {
      "epoch": 0.4592202646807964,
      "grad_norm": 0.5760700397369853,
      "learning_rate": 5.898128412896401e-06,
      "loss": 0.5139,
      "step": 6259
    },
    {
      "epoch": 0.45929363427093556,
      "grad_norm": 1.3574666214155147,
      "learning_rate": 5.896959516613003e-06,
      "loss": 0.7407,
      "step": 6260
    },
    {
      "epoch": 0.4593670038610747,
      "grad_norm": 1.5439641355772389,
      "learning_rate": 5.895790569676081e-06,
      "loss": 0.7448,
      "step": 6261
    },
    {
      "epoch": 0.4594403734512138,
      "grad_norm": 0.565391431828189,
      "learning_rate": 5.894621572151646e-06,
      "loss": 0.5103,
      "step": 6262
    },
    {
      "epoch": 0.45951374304135295,
      "grad_norm": 1.6332061202205457,
      "learning_rate": 5.893452524105715e-06,
      "loss": 0.7674,
      "step": 6263
    },
    {
      "epoch": 0.45958711263149205,
      "grad_norm": 0.5813967361998403,
      "learning_rate": 5.892283425604308e-06,
      "loss": 0.5111,
      "step": 6264
    },
    {
      "epoch": 0.45958711263149205,
      "lm_loss": 1.3112,
      "step": 6264,
      "vm_loss": 0.0411
    },
    {
      "epoch": 0.45958711263149205,
      "lm_loss": 1.464,
      "step": 6264,
      "vm_loss": 0.1391
    },
    {
      "epoch": 0.45958711263149205,
      "lm_loss": 0.3236,
      "step": 6264,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.45958711263149205,
      "lm_loss": 0.1986,
      "step": 6264,
      "vm_loss": 0.2604
    },
    {
      "epoch": 0.45958711263149205,
      "lm_loss": 0.4684,
      "step": 6264,
      "vm_loss": 0.1549
    },
    {
      "epoch": 0.45958711263149205,
      "lm_loss": 0.6522,
      "step": 6264,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.45958711263149205,
      "lm_loss": 0.9909,
      "step": 6264,
      "vm_loss": 0.1698
    },
    {
      "epoch": 0.45958711263149205,
      "lm_loss": 0.1679,
      "step": 6264,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.4596604822216312,
      "grad_norm": 1.4905306297094285,
      "learning_rate": 5.8911142767134465e-06,
      "loss": 0.9033,
      "step": 6265
    },
    {
      "epoch": 0.4597338518117703,
      "grad_norm": 1.5455074817774053,
      "learning_rate": 5.889945077499155e-06,
      "loss": 0.7918,
      "step": 6266
    },
    {
      "epoch": 0.45980722140190944,
      "grad_norm": 1.533013816543743,
      "learning_rate": 5.888775828027461e-06,
      "loss": 0.8136,
      "step": 6267
    },
    {
      "epoch": 0.4598805909920486,
      "grad_norm": 1.5675243064407571,
      "learning_rate": 5.887606528364394e-06,
      "loss": 0.77,
      "step": 6268
    },
    {
      "epoch": 0.4599539605821877,
      "grad_norm": 1.313785844184174,
      "learning_rate": 5.88643717857599e-06,
      "loss": 0.7916,
      "step": 6269
    },
    {
      "epoch": 0.46002733017232683,
      "grad_norm": 1.5505724582098905,
      "learning_rate": 5.885267778728282e-06,
      "loss": 0.7826,
      "step": 6270
    },
    {
      "epoch": 0.4601006997624659,
      "grad_norm": 1.4300404084596128,
      "learning_rate": 5.88409832888731e-06,
      "loss": 0.7944,
      "step": 6271
    },
    {
      "epoch": 0.4601740693526051,
      "grad_norm": 1.5033626622254095,
      "learning_rate": 5.882928829119117e-06,
      "loss": 0.7007,
      "step": 6272
    },
    {
      "epoch": 0.4601740693526051,
      "lm_loss": 0.5131,
      "step": 6272,
      "vm_loss": 0.098
    },
    {
      "epoch": 0.4601740693526051,
      "lm_loss": 0.0288,
      "step": 6272,
      "vm_loss": 0.2154
    },
    {
      "epoch": 0.4601740693526051,
      "lm_loss": 0.0488,
      "step": 6272,
      "vm_loss": 0.0679
    },
    {
      "epoch": 0.4601740693526051,
      "lm_loss": 0.9197,
      "step": 6272,
      "vm_loss": 0.1511
    },
    {
      "epoch": 0.4601740693526051,
      "lm_loss": 0.3155,
      "step": 6272,
      "vm_loss": 0.123
    },
    {
      "epoch": 0.4601740693526051,
      "lm_loss": 0.7732,
      "step": 6272,
      "vm_loss": 0.0948
    },
    {
      "epoch": 0.4601740693526051,
      "lm_loss": 0.0057,
      "step": 6272,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.4601740693526051,
      "lm_loss": 0.9401,
      "step": 6272,
      "vm_loss": 0.0953
    },
    {
      "epoch": 0.4602474389427442,
      "grad_norm": 1.4930601909562762,
      "learning_rate": 5.881759279489746e-06,
      "loss": 0.6945,
      "step": 6273
    },
    {
      "epoch": 0.4603208085328833,
      "grad_norm": 0.6169022088186785,
      "learning_rate": 5.880589680065244e-06,
      "loss": 0.5012,
      "step": 6274
    },
    {
      "epoch": 0.46039417812302247,
      "grad_norm": 1.5397744039649222,
      "learning_rate": 5.879420030911663e-06,
      "loss": 0.8047,
      "step": 6275
    },
    {
      "epoch": 0.46046754771316156,
      "grad_norm": 1.4975586713189868,
      "learning_rate": 5.878250332095055e-06,
      "loss": 0.7384,
      "step": 6276
    },
    {
      "epoch": 0.4605409173033007,
      "grad_norm": 1.4362846921807144,
      "learning_rate": 5.877080583681476e-06,
      "loss": 0.743,
      "step": 6277
    },
    {
      "epoch": 0.46061428689343986,
      "grad_norm": 0.5752082962050865,
      "learning_rate": 5.8759107857369834e-06,
      "loss": 0.5061,
      "step": 6278
    },
    {
      "epoch": 0.46068765648357896,
      "grad_norm": 1.2926706343004604,
      "learning_rate": 5.87474093832764e-06,
      "loss": 0.6833,
      "step": 6279
    },
    {
      "epoch": 0.4607610260737181,
      "grad_norm": 0.6070036344353728,
      "learning_rate": 5.87357104151951e-06,
      "loss": 0.5214,
      "step": 6280
    },
    {
      "epoch": 0.4607610260737181,
      "lm_loss": 0.5373,
      "step": 6280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4607610260737181,
      "lm_loss": 0.563,
      "step": 6280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4607610260737181,
      "lm_loss": 0.351,
      "step": 6280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4607610260737181,
      "lm_loss": 0.3958,
      "step": 6280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4607610260737181,
      "lm_loss": 0.6116,
      "step": 6280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4607610260737181,
      "lm_loss": 0.4298,
      "step": 6280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4607610260737181,
      "lm_loss": 0.5223,
      "step": 6280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4607610260737181,
      "lm_loss": 0.5134,
      "step": 6280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4608343956638572,
      "grad_norm": 0.5689470388220815,
      "learning_rate": 5.87240109537866e-06,
      "loss": 0.5025,
      "step": 6281
    },
    {
      "epoch": 0.46090776525399635,
      "grad_norm": 0.572359742661527,
      "learning_rate": 5.871231099971159e-06,
      "loss": 0.5069,
      "step": 6282
    },
    {
      "epoch": 0.4609811348441355,
      "grad_norm": 1.6575679153434804,
      "learning_rate": 5.870061055363081e-06,
      "loss": 0.9378,
      "step": 6283
    },
    {
      "epoch": 0.4610545044342746,
      "grad_norm": 1.552044780720716,
      "learning_rate": 5.8688909616204995e-06,
      "loss": 0.8371,
      "step": 6284
    },
    {
      "epoch": 0.46112787402441374,
      "grad_norm": 1.5092935204039506,
      "learning_rate": 5.867720818809495e-06,
      "loss": 0.799,
      "step": 6285
    },
    {
      "epoch": 0.46120124361455284,
      "grad_norm": 1.409607852682989,
      "learning_rate": 5.866550626996146e-06,
      "loss": 0.7327,
      "step": 6286
    },
    {
      "epoch": 0.461274613204692,
      "grad_norm": 1.6908457974608997,
      "learning_rate": 5.865380386246536e-06,
      "loss": 0.8737,
      "step": 6287
    },
    {
      "epoch": 0.46134798279483114,
      "grad_norm": 1.440365209835466,
      "learning_rate": 5.864210096626754e-06,
      "loss": 0.7154,
      "step": 6288
    },
    {
      "epoch": 0.46134798279483114,
      "lm_loss": 1.5133,
      "step": 6288,
      "vm_loss": 0.0485
    },
    {
      "epoch": 0.46134798279483114,
      "lm_loss": 0.026,
      "step": 6288,
      "vm_loss": 0.126
    },
    {
      "epoch": 0.46134798279483114,
      "lm_loss": 0.6915,
      "step": 6288,
      "vm_loss": 0.1088
    },
    {
      "epoch": 0.46134798279483114,
      "lm_loss": 0.1097,
      "step": 6288,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.46134798279483114,
      "lm_loss": 0.5532,
      "step": 6288,
      "vm_loss": 0.1313
    },
    {
      "epoch": 0.46134798279483114,
      "lm_loss": 0.3909,
      "step": 6288,
      "vm_loss": 0.0978
    },
    {
      "epoch": 0.46134798279483114,
      "lm_loss": 0.4415,
      "step": 6288,
      "vm_loss": 0.0919
    },
    {
      "epoch": 0.46134798279483114,
      "lm_loss": 0.7202,
      "step": 6288,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.46142135238497023,
      "grad_norm": 1.4109493970597093,
      "learning_rate": 5.863039758202889e-06,
      "loss": 0.6661,
      "step": 6289
    },
    {
      "epoch": 0.4614947219751094,
      "grad_norm": 1.6543019216641957,
      "learning_rate": 5.861869371041029e-06,
      "loss": 0.8039,
      "step": 6290
    },
    {
      "epoch": 0.4615680915652485,
      "grad_norm": 1.6003704892114041,
      "learning_rate": 5.860698935207273e-06,
      "loss": 0.8356,
      "step": 6291
    },
    {
      "epoch": 0.4616414611553876,
      "grad_norm": 0.7025238067629697,
      "learning_rate": 5.859528450767715e-06,
      "loss": 0.5187,
      "step": 6292
    },
    {
      "epoch": 0.46171483074552677,
      "grad_norm": 1.645300823287918,
      "learning_rate": 5.8583579177884575e-06,
      "loss": 0.8686,
      "step": 6293
    },
    {
      "epoch": 0.46178820033566587,
      "grad_norm": 0.595290904054176,
      "learning_rate": 5.857187336335603e-06,
      "loss": 0.4996,
      "step": 6294
    },
    {
      "epoch": 0.461861569925805,
      "grad_norm": 1.531837358194319,
      "learning_rate": 5.856016706475257e-06,
      "loss": 0.7846,
      "step": 6295
    },
    {
      "epoch": 0.4619349395159441,
      "grad_norm": 0.5699133915861945,
      "learning_rate": 5.854846028273527e-06,
      "loss": 0.5073,
      "step": 6296
    },
    {
      "epoch": 0.4619349395159441,
      "lm_loss": 1.268,
      "step": 6296,
      "vm_loss": 0.0464
    },
    {
      "epoch": 0.4619349395159441,
      "lm_loss": 1.0314,
      "step": 6296,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.4619349395159441,
      "lm_loss": 0.8447,
      "step": 6296,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.4619349395159441,
      "lm_loss": 0.3071,
      "step": 6296,
      "vm_loss": 0.1798
    },
    {
      "epoch": 0.4619349395159441,
      "lm_loss": 0.1887,
      "step": 6296,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.4619349395159441,
      "lm_loss": 0.0662,
      "step": 6296,
      "vm_loss": 0.0756
    },
    {
      "epoch": 0.4619349395159441,
      "lm_loss": 0.6547,
      "step": 6296,
      "vm_loss": 0.1821
    },
    {
      "epoch": 0.4619349395159441,
      "lm_loss": 0.5131,
      "step": 6296,
      "vm_loss": 0.1809
    },
    {
      "epoch": 0.46200830910608326,
      "grad_norm": 1.4919488857544045,
      "learning_rate": 5.8536753017965265e-06,
      "loss": 0.8582,
      "step": 6297
    },
    {
      "epoch": 0.4620816786962224,
      "grad_norm": 1.5235107702106685,
      "learning_rate": 5.852504527110365e-06,
      "loss": 0.8091,
      "step": 6298
    },
    {
      "epoch": 0.4621550482863615,
      "grad_norm": 1.4739116304573763,
      "learning_rate": 5.851333704281164e-06,
      "loss": 0.7541,
      "step": 6299
    },
    {
      "epoch": 0.46222841787650065,
      "grad_norm": 1.6295034549240264,
      "learning_rate": 5.85016283337504e-06,
      "loss": 0.8707,
      "step": 6300
    },
    {
      "epoch": 0.46230178746663975,
      "grad_norm": 1.699997387260962,
      "learning_rate": 5.848991914458114e-06,
      "loss": 0.7582,
      "step": 6301
    },
    {
      "epoch": 0.4623751570567789,
      "grad_norm": 1.7260066207402993,
      "learning_rate": 5.847820947596513e-06,
      "loss": 0.8499,
      "step": 6302
    },
    {
      "epoch": 0.46244852664691805,
      "grad_norm": 0.7575926842472378,
      "learning_rate": 5.846649932856363e-06,
      "loss": 0.516,
      "step": 6303
    },
    {
      "epoch": 0.46252189623705714,
      "grad_norm": 0.6556610040282661,
      "learning_rate": 5.845478870303796e-06,
      "loss": 0.5022,
      "step": 6304
    },
    {
      "epoch": 0.46252189623705714,
      "lm_loss": 1.7719,
      "step": 6304,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.46252189623705714,
      "lm_loss": 1.0469,
      "step": 6304,
      "vm_loss": 0.2251
    },
    {
      "epoch": 0.46252189623705714,
      "lm_loss": 0.2972,
      "step": 6304,
      "vm_loss": 0.1296
    },
    {
      "epoch": 0.46252189623705714,
      "lm_loss": 1.0793,
      "step": 6304,
      "vm_loss": 0.2655
    },
    {
      "epoch": 0.46252189623705714,
      "lm_loss": 0.7172,
      "step": 6304,
      "vm_loss": 0.1821
    },
    {
      "epoch": 0.46252189623705714,
      "lm_loss": 0.4606,
      "step": 6304,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.46252189623705714,
      "lm_loss": 0.1577,
      "step": 6304,
      "vm_loss": 0.0852
    },
    {
      "epoch": 0.46252189623705714,
      "lm_loss": 1.0688,
      "step": 6304,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.4625952658271963,
      "grad_norm": 1.472180221391768,
      "learning_rate": 5.844307760004943e-06,
      "loss": 0.8275,
      "step": 6305
    },
    {
      "epoch": 0.4626686354173354,
      "grad_norm": 1.539609962654012,
      "learning_rate": 5.843136602025938e-06,
      "loss": 0.7857,
      "step": 6306
    },
    {
      "epoch": 0.46274200500747453,
      "grad_norm": 1.5253189695336293,
      "learning_rate": 5.841965396432924e-06,
      "loss": 0.8571,
      "step": 6307
    },
    {
      "epoch": 0.4628153745976136,
      "grad_norm": 1.669034036590229,
      "learning_rate": 5.840794143292037e-06,
      "loss": 0.7737,
      "step": 6308
    },
    {
      "epoch": 0.4628887441877528,
      "grad_norm": 1.5664863895673122,
      "learning_rate": 5.839622842669423e-06,
      "loss": 0.781,
      "step": 6309
    },
    {
      "epoch": 0.4629621137778919,
      "grad_norm": 1.3995295055548234,
      "learning_rate": 5.838451494631227e-06,
      "loss": 0.703,
      "step": 6310
    },
    {
      "epoch": 0.463035483368031,
      "grad_norm": 0.8252832786102781,
      "learning_rate": 5.8372800992436e-06,
      "loss": 0.4996,
      "step": 6311
    },
    {
      "epoch": 0.46310885295817017,
      "grad_norm": 1.6896648999125214,
      "learning_rate": 5.8361086565726925e-06,
      "loss": 0.8733,
      "step": 6312
    },
    {
      "epoch": 0.46310885295817017,
      "lm_loss": 0.4666,
      "step": 6312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.46310885295817017,
      "lm_loss": 0.5372,
      "step": 6312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.46310885295817017,
      "lm_loss": 0.5009,
      "step": 6312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.46310885295817017,
      "lm_loss": 0.4653,
      "step": 6312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.46310885295817017,
      "lm_loss": 0.6192,
      "step": 6312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.46310885295817017,
      "lm_loss": 0.6653,
      "step": 6312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.46310885295817017,
      "lm_loss": 0.533,
      "step": 6312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.46310885295817017,
      "lm_loss": 0.4902,
      "step": 6312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.46318222254830926,
      "grad_norm": 0.6820836592415548,
      "learning_rate": 5.834937166684658e-06,
      "loss": 0.5024,
      "step": 6313
    },
    {
      "epoch": 0.4632555921384484,
      "grad_norm": 1.378039949994506,
      "learning_rate": 5.833765629645652e-06,
      "loss": 0.8206,
      "step": 6314
    },
    {
      "epoch": 0.46332896172858756,
      "grad_norm": 1.5600353010754546,
      "learning_rate": 5.8325940455218385e-06,
      "loss": 0.7549,
      "step": 6315
    },
    {
      "epoch": 0.46340233131872666,
      "grad_norm": 0.6138725368664997,
      "learning_rate": 5.831422414379376e-06,
      "loss": 0.5131,
      "step": 6316
    },
    {
      "epoch": 0.4634757009088658,
      "grad_norm": 0.652091611218367,
      "learning_rate": 5.830250736284433e-06,
      "loss": 0.512,
      "step": 6317
    },
    {
      "epoch": 0.4635490704990049,
      "grad_norm": 1.81538529309147,
      "learning_rate": 5.8290790113031734e-06,
      "loss": 0.7959,
      "step": 6318
    },
    {
      "epoch": 0.46362244008914405,
      "grad_norm": 0.6054426674523413,
      "learning_rate": 5.827907239501769e-06,
      "loss": 0.4767,
      "step": 6319
    },
    {
      "epoch": 0.4636958096792832,
      "grad_norm": 1.5052278108573218,
      "learning_rate": 5.826735420946392e-06,
      "loss": 0.7234,
      "step": 6320
    },
    {
      "epoch": 0.4636958096792832,
      "lm_loss": 0.4852,
      "step": 6320,
      "vm_loss": 0.1031
    },
    {
      "epoch": 0.4636958096792832,
      "lm_loss": 0.149,
      "step": 6320,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.4636958096792832,
      "lm_loss": 0.7063,
      "step": 6320,
      "vm_loss": 0.1656
    },
    {
      "epoch": 0.4636958096792832,
      "lm_loss": 0.7679,
      "step": 6320,
      "vm_loss": 0.1273
    },
    {
      "epoch": 0.4636958096792832,
      "lm_loss": 0.3788,
      "step": 6320,
      "vm_loss": 0.1065
    },
    {
      "epoch": 0.4636958096792832,
      "lm_loss": 0.4927,
      "step": 6320,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.4636958096792832,
      "lm_loss": 0.8994,
      "step": 6320,
      "vm_loss": 0.1908
    },
    {
      "epoch": 0.4636958096792832,
      "lm_loss": 0.2709,
      "step": 6320,
      "vm_loss": 0.1358
    },
    {
      "epoch": 0.4637691792694223,
      "grad_norm": 1.5253191649114701,
      "learning_rate": 5.82556355570322e-06,
      "loss": 0.7465,
      "step": 6321
    },
    {
      "epoch": 0.46384254885956144,
      "grad_norm": 1.4793894311670484,
      "learning_rate": 5.824391643838427e-06,
      "loss": 0.8388,
      "step": 6322
    },
    {
      "epoch": 0.46391591844970054,
      "grad_norm": 1.4640754972750376,
      "learning_rate": 5.823219685418198e-06,
      "loss": 0.6877,
      "step": 6323
    },
    {
      "epoch": 0.4639892880398397,
      "grad_norm": 1.6475209260200812,
      "learning_rate": 5.822047680508713e-06,
      "loss": 0.7935,
      "step": 6324
    },
    {
      "epoch": 0.46406265762997884,
      "grad_norm": 0.724820920053229,
      "learning_rate": 5.820875629176162e-06,
      "loss": 0.5113,
      "step": 6325
    },
    {
      "epoch": 0.46413602722011793,
      "grad_norm": 1.513306736771702,
      "learning_rate": 5.819703531486731e-06,
      "loss": 0.7119,
      "step": 6326
    },
    {
      "epoch": 0.4642093968102571,
      "grad_norm": 1.4627752822699467,
      "learning_rate": 5.8185313875066095e-06,
      "loss": 0.705,
      "step": 6327
    },
    {
      "epoch": 0.4642827664003962,
      "grad_norm": 1.5110228438068334,
      "learning_rate": 5.817359197301996e-06,
      "loss": 0.8486,
      "step": 6328
    },
    {
      "epoch": 0.4642827664003962,
      "lm_loss": 1.7438,
      "step": 6328,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.4642827664003962,
      "lm_loss": 0.0849,
      "step": 6328,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.4642827664003962,
      "lm_loss": 0.2817,
      "step": 6328,
      "vm_loss": 0.1144
    },
    {
      "epoch": 0.4642827664003962,
      "lm_loss": 0.218,
      "step": 6328,
      "vm_loss": 0.0798
    },
    {
      "epoch": 0.4642827664003962,
      "lm_loss": 1.4309,
      "step": 6328,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.4642827664003962,
      "lm_loss": 0.4493,
      "step": 6328,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.4642827664003962,
      "lm_loss": 1.2834,
      "step": 6328,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.4642827664003962,
      "lm_loss": 0.933,
      "step": 6328,
      "vm_loss": 0.1867
    },
    {
      "epoch": 0.4643561359905353,
      "grad_norm": 1.5072786615747167,
      "learning_rate": 5.816186960939084e-06,
      "loss": 0.7546,
      "step": 6329
    },
    {
      "epoch": 0.4644295055806745,
      "grad_norm": 1.496444966422105,
      "learning_rate": 5.815014678484071e-06,
      "loss": 0.7817,
      "step": 6330
    },
    {
      "epoch": 0.46450287517081357,
      "grad_norm": 1.7407042724558623,
      "learning_rate": 5.813842350003163e-06,
      "loss": 0.8073,
      "step": 6331
    },
    {
      "epoch": 0.4645762447609527,
      "grad_norm": 1.528725086328376,
      "learning_rate": 5.812669975562561e-06,
      "loss": 0.8374,
      "step": 6332
    },
    {
      "epoch": 0.4646496143510918,
      "grad_norm": 0.5911035263555057,
      "learning_rate": 5.8114975552284735e-06,
      "loss": 0.488,
      "step": 6333
    },
    {
      "epoch": 0.46472298394123096,
      "grad_norm": 1.4666937585093294,
      "learning_rate": 5.8103250890671095e-06,
      "loss": 0.7821,
      "step": 6334
    },
    {
      "epoch": 0.4647963535313701,
      "grad_norm": 0.5743307140098757,
      "learning_rate": 5.80915257714468e-06,
      "loss": 0.5106,
      "step": 6335
    },
    {
      "epoch": 0.4648697231215092,
      "grad_norm": 1.3780766763708794,
      "learning_rate": 5.8079800195274014e-06,
      "loss": 0.7451,
      "step": 6336
    },
    {
      "epoch": 0.4648697231215092,
      "lm_loss": 1.4005,
      "step": 6336,
      "vm_loss": 0.0465
    },
    {
      "epoch": 0.4648697231215092,
      "lm_loss": 1.5106,
      "step": 6336,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.4648697231215092,
      "lm_loss": 1.1056,
      "step": 6336,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.4648697231215092,
      "lm_loss": 0.9459,
      "step": 6336,
      "vm_loss": 0.1614
    },
    {
      "epoch": 0.4648697231215092,
      "lm_loss": 0.7764,
      "step": 6336,
      "vm_loss": 0.1275
    },
    {
      "epoch": 0.4648697231215092,
      "lm_loss": 0.8057,
      "step": 6336,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.4648697231215092,
      "lm_loss": 0.3851,
      "step": 6336,
      "vm_loss": 0.1427
    },
    {
      "epoch": 0.4648697231215092,
      "lm_loss": 0.3716,
      "step": 6336,
      "vm_loss": 0.1068
    },
    {
      "epoch": 0.46494309271164835,
      "grad_norm": 1.5009232173823182,
      "learning_rate": 5.80680741628149e-06,
      "loss": 0.7985,
      "step": 6337
    },
    {
      "epoch": 0.46501646230178745,
      "grad_norm": 1.4014768402464706,
      "learning_rate": 5.8056347674731665e-06,
      "loss": 0.7863,
      "step": 6338
    },
    {
      "epoch": 0.4650898318919266,
      "grad_norm": 0.6186176591750736,
      "learning_rate": 5.804462073168652e-06,
      "loss": 0.5195,
      "step": 6339
    },
    {
      "epoch": 0.46516320148206575,
      "grad_norm": 1.3516555344133476,
      "learning_rate": 5.803289333434172e-06,
      "loss": 0.7655,
      "step": 6340
    },
    {
      "epoch": 0.46523657107220484,
      "grad_norm": 0.555066213318832,
      "learning_rate": 5.802116548335953e-06,
      "loss": 0.5026,
      "step": 6341
    },
    {
      "epoch": 0.465309940662344,
      "grad_norm": 1.4014945754066062,
      "learning_rate": 5.8009437179402285e-06,
      "loss": 0.8295,
      "step": 6342
    },
    {
      "epoch": 0.4653833102524831,
      "grad_norm": 1.4382904193612556,
      "learning_rate": 5.799770842313226e-06,
      "loss": 0.6583,
      "step": 6343
    },
    {
      "epoch": 0.46545667984262223,
      "grad_norm": 1.4258530165920404,
      "learning_rate": 5.7985979215211855e-06,
      "loss": 0.8071,
      "step": 6344
    },
    {
      "epoch": 0.46545667984262223,
      "lm_loss": 1.0066,
      "step": 6344,
      "vm_loss": 0.0665
    },
    {
      "epoch": 0.46545667984262223,
      "lm_loss": 0.1599,
      "step": 6344,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.46545667984262223,
      "lm_loss": 0.6215,
      "step": 6344,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.46545667984262223,
      "lm_loss": 0.9873,
      "step": 6344,
      "vm_loss": 0.0723
    },
    {
      "epoch": 0.46545667984262223,
      "lm_loss": 0.7695,
      "step": 6344,
      "vm_loss": 0.162
    },
    {
      "epoch": 0.46545667984262223,
      "lm_loss": 0.4798,
      "step": 6344,
      "vm_loss": 0.163
    },
    {
      "epoch": 0.46545667984262223,
      "lm_loss": 0.3523,
      "step": 6344,
      "vm_loss": 0.1166
    },
    {
      "epoch": 0.46545667984262223,
      "lm_loss": 1.1491,
      "step": 6344,
      "vm_loss": 0.115
    },
    {
      "epoch": 0.4655300494327614,
      "grad_norm": 1.5363065074156157,
      "learning_rate": 5.797424955630342e-06,
      "loss": 0.7296,
      "step": 6345
    },
    {
      "epoch": 0.4656034190229005,
      "grad_norm": 0.5945619170394215,
      "learning_rate": 5.7962519447069354e-06,
      "loss": 0.5019,
      "step": 6346
    },
    {
      "epoch": 0.4656767886130396,
      "grad_norm": 1.5164097630815498,
      "learning_rate": 5.795078888817212e-06,
      "loss": 0.7715,
      "step": 6347
    },
    {
      "epoch": 0.4657501582031787,
      "grad_norm": 0.5994413685862731,
      "learning_rate": 5.793905788027413e-06,
      "loss": 0.5121,
      "step": 6348
    },
    {
      "epoch": 0.46582352779331787,
      "grad_norm": 1.326280818001915,
      "learning_rate": 5.7927326424037875e-06,
      "loss": 0.7527,
      "step": 6349
    },
    {
      "epoch": 0.46589689738345696,
      "grad_norm": 1.5447899448761073,
      "learning_rate": 5.791559452012588e-06,
      "loss": 0.8271,
      "step": 6350
    },
    {
      "epoch": 0.4659702669735961,
      "grad_norm": 1.6704490539089794,
      "learning_rate": 5.7903862169200654e-06,
      "loss": 0.7748,
      "step": 6351
    },
    {
      "epoch": 0.46604363656373526,
      "grad_norm": 1.5074829658491617,
      "learning_rate": 5.789212937192476e-06,
      "loss": 0.8503,
      "step": 6352
    },
    {
      "epoch": 0.46604363656373526,
      "lm_loss": 0.7135,
      "step": 6352,
      "vm_loss": 0.0218
    },
    {
      "epoch": 0.46604363656373526,
      "lm_loss": 0.3923,
      "step": 6352,
      "vm_loss": 0.0838
    },
    {
      "epoch": 0.46604363656373526,
      "lm_loss": 0.8704,
      "step": 6352,
      "vm_loss": 0.0833
    },
    {
      "epoch": 0.46604363656373526,
      "lm_loss": 0.6431,
      "step": 6352,
      "vm_loss": 0.1189
    },
    {
      "epoch": 0.46604363656373526,
      "lm_loss": 0.3021,
      "step": 6352,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.46604363656373526,
      "lm_loss": 0.9051,
      "step": 6352,
      "vm_loss": 0.1115
    },
    {
      "epoch": 0.46604363656373526,
      "lm_loss": 0.041,
      "step": 6352,
      "vm_loss": 0.0539
    },
    {
      "epoch": 0.46604363656373526,
      "lm_loss": 0.8579,
      "step": 6352,
      "vm_loss": 0.1239
    },
    {
      "epoch": 0.46611700615387436,
      "grad_norm": 1.731366520469205,
      "learning_rate": 5.788039612896077e-06,
      "loss": 0.8685,
      "step": 6353
    },
    {
      "epoch": 0.4661903757440135,
      "grad_norm": 1.6042438433915518,
      "learning_rate": 5.786866244097129e-06,
      "loss": 0.8755,
      "step": 6354
    },
    {
      "epoch": 0.4662637453341526,
      "grad_norm": 0.6324036507178867,
      "learning_rate": 5.7856928308618975e-06,
      "loss": 0.5271,
      "step": 6355
    },
    {
      "epoch": 0.46633711492429175,
      "grad_norm": 1.7873631694506111,
      "learning_rate": 5.784519373256645e-06,
      "loss": 0.7365,
      "step": 6356
    },
    {
      "epoch": 0.4664104845144309,
      "grad_norm": 1.722630251700892,
      "learning_rate": 5.7833458713476395e-06,
      "loss": 0.7581,
      "step": 6357
    },
    {
      "epoch": 0.46648385410457,
      "grad_norm": 1.4452879005178867,
      "learning_rate": 5.782172325201155e-06,
      "loss": 0.8217,
      "step": 6358
    },
    {
      "epoch": 0.46655722369470914,
      "grad_norm": 1.538507227699513,
      "learning_rate": 5.7809987348834605e-06,
      "loss": 0.9468,
      "step": 6359
    },
    {
      "epoch": 0.46663059328484824,
      "grad_norm": 1.4198861751738543,
      "learning_rate": 5.779825100460834e-06,
      "loss": 0.8909,
      "step": 6360
    },
    {
      "epoch": 0.46663059328484824,
      "lm_loss": 1.4698,
      "step": 6360,
      "vm_loss": 0.1077
    },
    {
      "epoch": 0.46663059328484824,
      "lm_loss": 0.2486,
      "step": 6360,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.46663059328484824,
      "lm_loss": 0.1778,
      "step": 6360,
      "vm_loss": 0.1695
    },
    {
      "epoch": 0.46663059328484824,
      "lm_loss": 0.007,
      "step": 6360,
      "vm_loss": 0.1289
    },
    {
      "epoch": 0.46663059328484824,
      "lm_loss": 0.0161,
      "step": 6360,
      "vm_loss": 0.1206
    },
    {
      "epoch": 0.46663059328484824,
      "lm_loss": 1.1052,
      "step": 6360,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.46663059328484824,
      "lm_loss": 0.6712,
      "step": 6360,
      "vm_loss": 0.1746
    },
    {
      "epoch": 0.46663059328484824,
      "lm_loss": 0.7995,
      "step": 6360,
      "vm_loss": 0.1803
    },
    {
      "epoch": 0.4667039628749874,
      "grad_norm": 1.580864633811075,
      "learning_rate": 5.778651421999554e-06,
      "loss": 0.7908,
      "step": 6361
    },
    {
      "epoch": 0.46677733246512654,
      "grad_norm": 1.3048273828714176,
      "learning_rate": 5.7774776995658975e-06,
      "loss": 0.6398,
      "step": 6362
    },
    {
      "epoch": 0.46685070205526563,
      "grad_norm": 1.5625532586073394,
      "learning_rate": 5.776303933226152e-06,
      "loss": 0.7889,
      "step": 6363
    },
    {
      "epoch": 0.4669240716454048,
      "grad_norm": 1.580992720427033,
      "learning_rate": 5.7751301230466005e-06,
      "loss": 0.6574,
      "step": 6364
    },
    {
      "epoch": 0.4669974412355439,
      "grad_norm": 1.4532366762638154,
      "learning_rate": 5.77395626909353e-06,
      "loss": 0.806,
      "step": 6365
    },
    {
      "epoch": 0.467070810825683,
      "grad_norm": 1.543069138482282,
      "learning_rate": 5.772782371433234e-06,
      "loss": 0.8738,
      "step": 6366
    },
    {
      "epoch": 0.4671441804158222,
      "grad_norm": 1.4283385113256057,
      "learning_rate": 5.771608430132003e-06,
      "loss": 0.7194,
      "step": 6367
    },
    {
      "epoch": 0.46721755000596127,
      "grad_norm": 0.6047718418503362,
      "learning_rate": 5.770434445256134e-06,
      "loss": 0.5143,
      "step": 6368
    },
    {
      "epoch": 0.46721755000596127,
      "lm_loss": 0.5533,
      "step": 6368,
      "vm_loss": 0.0958
    },
    {
      "epoch": 0.46721755000596127,
      "lm_loss": 0.5529,
      "step": 6368,
      "vm_loss": 0.0864
    },
    {
      "epoch": 0.46721755000596127,
      "lm_loss": 1.2205,
      "step": 6368,
      "vm_loss": 0.1548
    },
    {
      "epoch": 0.46721755000596127,
      "lm_loss": 0.0866,
      "step": 6368,
      "vm_loss": 0.1209
    },
    {
      "epoch": 0.46721755000596127,
      "lm_loss": 0.7093,
      "step": 6368,
      "vm_loss": 0.1688
    },
    {
      "epoch": 0.46721755000596127,
      "lm_loss": 0.6135,
      "step": 6368,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.46721755000596127,
      "lm_loss": 0.0845,
      "step": 6368,
      "vm_loss": 0.1796
    },
    {
      "epoch": 0.46721755000596127,
      "lm_loss": 1.0999,
      "step": 6368,
      "vm_loss": 0.2263
    },
    {
      "epoch": 0.4672909195961004,
      "grad_norm": 1.6437509933535879,
      "learning_rate": 5.7692604168719225e-06,
      "loss": 0.6864,
      "step": 6369
    },
    {
      "epoch": 0.4673642891862395,
      "grad_norm": 1.4757908552895143,
      "learning_rate": 5.76808634504567e-06,
      "loss": 0.6909,
      "step": 6370
    },
    {
      "epoch": 0.46743765877637866,
      "grad_norm": 1.6551025966507822,
      "learning_rate": 5.766912229843681e-06,
      "loss": 0.687,
      "step": 6371
    },
    {
      "epoch": 0.4675110283665178,
      "grad_norm": 1.434906299480205,
      "learning_rate": 5.765738071332258e-06,
      "loss": 0.782,
      "step": 6372
    },
    {
      "epoch": 0.4675843979566569,
      "grad_norm": 1.4914269290805067,
      "learning_rate": 5.76456386957771e-06,
      "loss": 0.8057,
      "step": 6373
    },
    {
      "epoch": 0.46765776754679605,
      "grad_norm": 0.5685630567375374,
      "learning_rate": 5.763389624646347e-06,
      "loss": 0.5122,
      "step": 6374
    },
    {
      "epoch": 0.46773113713693515,
      "grad_norm": 1.583942124412074,
      "learning_rate": 5.762215336604482e-06,
      "loss": 0.7182,
      "step": 6375
    },
    {
      "epoch": 0.4678045067270743,
      "grad_norm": 1.497929916130373,
      "learning_rate": 5.761041005518428e-06,
      "loss": 0.7236,
      "step": 6376
    },
    {
      "epoch": 0.4678045067270743,
      "lm_loss": 1.283,
      "step": 6376,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.4678045067270743,
      "lm_loss": 1.3672,
      "step": 6376,
      "vm_loss": 0.0981
    },
    {
      "epoch": 0.4678045067270743,
      "lm_loss": 0.3041,
      "step": 6376,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.4678045067270743,
      "lm_loss": 0.9678,
      "step": 6376,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.4678045067270743,
      "lm_loss": 0.253,
      "step": 6376,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.4678045067270743,
      "lm_loss": 0.4247,
      "step": 6376,
      "vm_loss": 0.1118
    },
    {
      "epoch": 0.4678045067270743,
      "lm_loss": 0.5165,
      "step": 6376,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.4678045067270743,
      "lm_loss": 1.029,
      "step": 6376,
      "vm_loss": 0.2395
    },
    {
      "epoch": 0.46787787631721345,
      "grad_norm": 1.419539604119444,
      "learning_rate": 5.759866631454505e-06,
      "loss": 0.8048,
      "step": 6377
    },
    {
      "epoch": 0.46795124590735254,
      "grad_norm": 1.5035704868493271,
      "learning_rate": 5.758692214479031e-06,
      "loss": 0.7086,
      "step": 6378
    },
    {
      "epoch": 0.4680246154974917,
      "grad_norm": 0.5614755308424185,
      "learning_rate": 5.7575177546583294e-06,
      "loss": 0.4963,
      "step": 6379
    },
    {
      "epoch": 0.4680979850876308,
      "grad_norm": 0.5469968200134021,
      "learning_rate": 5.756343252058724e-06,
      "loss": 0.5054,
      "step": 6380
    },
    {
      "epoch": 0.46817135467776994,
      "grad_norm": 0.5517649963440315,
      "learning_rate": 5.755168706746542e-06,
      "loss": 0.4972,
      "step": 6381
    },
    {
      "epoch": 0.4682447242679091,
      "grad_norm": 1.6648460435094805,
      "learning_rate": 5.753994118788113e-06,
      "loss": 0.8232,
      "step": 6382
    },
    {
      "epoch": 0.4683180938580482,
      "grad_norm": 1.3823399149222644,
      "learning_rate": 5.752819488249768e-06,
      "loss": 0.8713,
      "step": 6383
    },
    {
      "epoch": 0.46839146344818733,
      "grad_norm": 0.5670670198366419,
      "learning_rate": 5.751644815197843e-06,
      "loss": 0.5111,
      "step": 6384
    },
    {
      "epoch": 0.46839146344818733,
      "lm_loss": 1.3658,
      "step": 6384,
      "vm_loss": 0.0485
    },
    {
      "epoch": 0.46839146344818733,
      "lm_loss": 0.886,
      "step": 6384,
      "vm_loss": 0.0732
    },
    {
      "epoch": 0.46839146344818733,
      "lm_loss": 0.2742,
      "step": 6384,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.46839146344818733,
      "lm_loss": 0.1422,
      "step": 6384,
      "vm_loss": 0.0725
    },
    {
      "epoch": 0.46839146344818733,
      "lm_loss": 0.2422,
      "step": 6384,
      "vm_loss": 0.1637
    },
    {
      "epoch": 0.46839146344818733,
      "lm_loss": 1.2873,
      "step": 6384,
      "vm_loss": 0.1182
    },
    {
      "epoch": 0.46839146344818733,
      "lm_loss": 0.7949,
      "step": 6384,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.46839146344818733,
      "lm_loss": 1.1637,
      "step": 6384,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.4684648330383264,
      "grad_norm": 1.5114910523896472,
      "learning_rate": 5.7504700996986754e-06,
      "loss": 0.6962,
      "step": 6385
    },
    {
      "epoch": 0.46853820262846557,
      "grad_norm": 1.5815822299584417,
      "learning_rate": 5.749295341818599e-06,
      "loss": 0.734,
      "step": 6386
    },
    {
      "epoch": 0.46861157221860467,
      "grad_norm": 0.592625126996898,
      "learning_rate": 5.748120541623962e-06,
      "loss": 0.4948,
      "step": 6387
    },
    {
      "epoch": 0.4686849418087438,
      "grad_norm": 1.5499207490223272,
      "learning_rate": 5.746945699181105e-06,
      "loss": 0.7706,
      "step": 6388
    },
    {
      "epoch": 0.46875831139888297,
      "grad_norm": 1.6959220544859392,
      "learning_rate": 5.745770814556373e-06,
      "loss": 0.7619,
      "step": 6389
    },
    {
      "epoch": 0.46883168098902206,
      "grad_norm": 1.449057099101908,
      "learning_rate": 5.744595887816118e-06,
      "loss": 0.7258,
      "step": 6390
    },
    {
      "epoch": 0.4689050505791612,
      "grad_norm": 0.5482838305238797,
      "learning_rate": 5.743420919026688e-06,
      "loss": 0.507,
      "step": 6391
    },
    {
      "epoch": 0.4689784201693003,
      "grad_norm": 1.7281893362462593,
      "learning_rate": 5.742245908254438e-06,
      "loss": 0.7795,
      "step": 6392
    },
    {
      "epoch": 0.4689784201693003,
      "lm_loss": 0.6779,
      "step": 6392,
      "vm_loss": 0.0353
    },
    {
      "epoch": 0.4689784201693003,
      "lm_loss": 0.118,
      "step": 6392,
      "vm_loss": 0.1038
    },
    {
      "epoch": 0.4689784201693003,
      "lm_loss": 0.0508,
      "step": 6392,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.4689784201693003,
      "lm_loss": 0.1685,
      "step": 6392,
      "vm_loss": 0.146
    },
    {
      "epoch": 0.4689784201693003,
      "lm_loss": 0.6951,
      "step": 6392,
      "vm_loss": 0.0852
    },
    {
      "epoch": 0.4689784201693003,
      "lm_loss": 0.9894,
      "step": 6392,
      "vm_loss": 0.1365
    },
    {
      "epoch": 0.4689784201693003,
      "lm_loss": 0.6778,
      "step": 6392,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.4689784201693003,
      "lm_loss": 0.093,
      "step": 6392,
      "vm_loss": 0.0852
    },
    {
      "epoch": 0.46905178975943945,
      "grad_norm": 1.4372022459429603,
      "learning_rate": 5.741070855565725e-06,
      "loss": 0.7163,
      "step": 6393
    },
    {
      "epoch": 0.4691251593495786,
      "grad_norm": 1.4698669582283916,
      "learning_rate": 5.739895761026903e-06,
      "loss": 0.8478,
      "step": 6394
    },
    {
      "epoch": 0.4691985289397177,
      "grad_norm": 1.359055833006295,
      "learning_rate": 5.738720624704336e-06,
      "loss": 0.8096,
      "step": 6395
    },
    {
      "epoch": 0.46927189852985685,
      "grad_norm": 1.5704595154053587,
      "learning_rate": 5.737545446664387e-06,
      "loss": 0.7792,
      "step": 6396
    },
    {
      "epoch": 0.46934526811999594,
      "grad_norm": 1.392281278771516,
      "learning_rate": 5.736370226973418e-06,
      "loss": 0.7218,
      "step": 6397
    },
    {
      "epoch": 0.4694186377101351,
      "grad_norm": 1.6784377237610537,
      "learning_rate": 5.735194965697801e-06,
      "loss": 0.8085,
      "step": 6398
    },
    {
      "epoch": 0.46949200730027424,
      "grad_norm": 1.5722478140225171,
      "learning_rate": 5.734019662903901e-06,
      "loss": 0.8767,
      "step": 6399
    },
    {
      "epoch": 0.46956537689041333,
      "grad_norm": 1.4614679069155945,
      "learning_rate": 5.732844318658095e-06,
      "loss": 0.7308,
      "step": 6400
    },
    {
      "epoch": 0.46956537689041333,
      "lm_loss": 1.3141,
      "step": 6400,
      "vm_loss": 0.0981
    },
    {
      "epoch": 0.46956537689041333,
      "lm_loss": 1.0695,
      "step": 6400,
      "vm_loss": 0.2502
    },
    {
      "epoch": 0.46956537689041333,
      "lm_loss": 0.2498,
      "step": 6400,
      "vm_loss": 0.145
    },
    {
      "epoch": 0.46956537689041333,
      "lm_loss": 0.6432,
      "step": 6400,
      "vm_loss": 0.0672
    },
    {
      "epoch": 0.46956537689041333,
      "lm_loss": 0.1504,
      "step": 6400,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.46956537689041333,
      "lm_loss": 0.5788,
      "step": 6400,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.46956537689041333,
      "lm_loss": 0.8656,
      "step": 6400,
      "vm_loss": 0.2229
    },
    {
      "epoch": 0.46956537689041333,
      "lm_loss": 0.0224,
      "step": 6400,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.4696387464805525,
      "grad_norm": 1.4368847131256837,
      "learning_rate": 5.731668933026755e-06,
      "loss": 0.7906,
      "step": 6401
    },
    {
      "epoch": 0.4697121160706916,
      "grad_norm": 1.5775430591742081,
      "learning_rate": 5.730493506076258e-06,
      "loss": 0.8334,
      "step": 6402
    },
    {
      "epoch": 0.4697854856608307,
      "grad_norm": 1.7984690809210562,
      "learning_rate": 5.729318037872984e-06,
      "loss": 0.8493,
      "step": 6403
    },
    {
      "epoch": 0.4698588552509699,
      "grad_norm": 1.3957589561369255,
      "learning_rate": 5.728142528483315e-06,
      "loss": 0.7105,
      "step": 6404
    },
    {
      "epoch": 0.46993222484110897,
      "grad_norm": 1.6412877969666375,
      "learning_rate": 5.7269669779736325e-06,
      "loss": 0.8578,
      "step": 6405
    },
    {
      "epoch": 0.4700055944312481,
      "grad_norm": 1.4211575501579012,
      "learning_rate": 5.725791386410327e-06,
      "loss": 0.8546,
      "step": 6406
    },
    {
      "epoch": 0.4700789640213872,
      "grad_norm": 1.4991360130963758,
      "learning_rate": 5.724615753859782e-06,
      "loss": 0.8219,
      "step": 6407
    },
    {
      "epoch": 0.47015233361152636,
      "grad_norm": 1.5739520982151431,
      "learning_rate": 5.7234400803883915e-06,
      "loss": 0.8337,
      "step": 6408
    },
    {
      "epoch": 0.47015233361152636,
      "lm_loss": 0.5597,
      "step": 6408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47015233361152636,
      "lm_loss": 0.4684,
      "step": 6408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47015233361152636,
      "lm_loss": 0.4118,
      "step": 6408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47015233361152636,
      "lm_loss": 0.5592,
      "step": 6408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47015233361152636,
      "lm_loss": 0.4862,
      "step": 6408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47015233361152636,
      "lm_loss": 0.5083,
      "step": 6408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47015233361152636,
      "lm_loss": 0.3897,
      "step": 6408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47015233361152636,
      "lm_loss": 0.5268,
      "step": 6408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4702257032016655,
      "grad_norm": 0.6383072168548247,
      "learning_rate": 5.722264366062549e-06,
      "loss": 0.4819,
      "step": 6409
    },
    {
      "epoch": 0.4702990727918046,
      "grad_norm": 1.465952404022964,
      "learning_rate": 5.721088610948647e-06,
      "loss": 0.6416,
      "step": 6410
    },
    {
      "epoch": 0.47037244238194376,
      "grad_norm": 1.46434970486023,
      "learning_rate": 5.719912815113088e-06,
      "loss": 0.7838,
      "step": 6411
    },
    {
      "epoch": 0.47044581197208285,
      "grad_norm": 1.4336945094426397,
      "learning_rate": 5.7187369786222694e-06,
      "loss": 0.7436,
      "step": 6412
    },
    {
      "epoch": 0.470519181562222,
      "grad_norm": 1.4939840016425343,
      "learning_rate": 5.717561101542591e-06,
      "loss": 0.7672,
      "step": 6413
    },
    {
      "epoch": 0.47059255115236115,
      "grad_norm": 1.4761637860306869,
      "learning_rate": 5.7163851839404625e-06,
      "loss": 0.8153,
      "step": 6414
    },
    {
      "epoch": 0.47066592074250024,
      "grad_norm": 0.5623591491478426,
      "learning_rate": 5.715209225882288e-06,
      "loss": 0.5168,
      "step": 6415
    },
    {
      "epoch": 0.4707392903326394,
      "grad_norm": 0.5607451098893687,
      "learning_rate": 5.7140332274344764e-06,
      "loss": 0.4978,
      "step": 6416
    },
    {
      "epoch": 0.4707392903326394,
      "lm_loss": 0.9238,
      "step": 6416,
      "vm_loss": 0.0617
    },
    {
      "epoch": 0.4707392903326394,
      "lm_loss": 0.5836,
      "step": 6416,
      "vm_loss": 0.0966
    },
    {
      "epoch": 0.4707392903326394,
      "lm_loss": 0.817,
      "step": 6416,
      "vm_loss": 0.0635
    },
    {
      "epoch": 0.4707392903326394,
      "lm_loss": 0.3166,
      "step": 6416,
      "vm_loss": 0.0532
    },
    {
      "epoch": 0.4707392903326394,
      "lm_loss": 0.8661,
      "step": 6416,
      "vm_loss": 0.1381
    },
    {
      "epoch": 0.4707392903326394,
      "lm_loss": 0.211,
      "step": 6416,
      "vm_loss": 0.0561
    },
    {
      "epoch": 0.4707392903326394,
      "lm_loss": 0.1511,
      "step": 6416,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.4707392903326394,
      "lm_loss": 1.0387,
      "step": 6416,
      "vm_loss": 0.103
    },
    {
      "epoch": 0.4708126599227785,
      "grad_norm": 1.4165485538022233,
      "learning_rate": 5.712857188663442e-06,
      "loss": 0.7757,
      "step": 6417
    },
    {
      "epoch": 0.47088602951291764,
      "grad_norm": 1.6046920740363944,
      "learning_rate": 5.711681109635595e-06,
      "loss": 0.7619,
      "step": 6418
    },
    {
      "epoch": 0.4709593991030568,
      "grad_norm": 1.5375365138948311,
      "learning_rate": 5.710504990417355e-06,
      "loss": 0.8259,
      "step": 6419
    },
    {
      "epoch": 0.4710327686931959,
      "grad_norm": 1.5095672915258949,
      "learning_rate": 5.7093288310751375e-06,
      "loss": 0.7896,
      "step": 6420
    },
    {
      "epoch": 0.47110613828333503,
      "grad_norm": 1.6037797478609563,
      "learning_rate": 5.708152631675364e-06,
      "loss": 0.8255,
      "step": 6421
    },
    {
      "epoch": 0.4711795078734741,
      "grad_norm": 1.4237373782911489,
      "learning_rate": 5.706976392284459e-06,
      "loss": 0.799,
      "step": 6422
    },
    {
      "epoch": 0.4712528774636133,
      "grad_norm": 0.5873408663635518,
      "learning_rate": 5.705800112968846e-06,
      "loss": 0.5152,
      "step": 6423
    },
    {
      "epoch": 0.4713262470537524,
      "grad_norm": 1.6130521169854188,
      "learning_rate": 5.704623793794951e-06,
      "loss": 0.8092,
      "step": 6424
    },
    {
      "epoch": 0.4713262470537524,
      "lm_loss": 1.1271,
      "step": 6424,
      "vm_loss": 0.0824
    },
    {
      "epoch": 0.4713262470537524,
      "lm_loss": 0.9572,
      "step": 6424,
      "vm_loss": 0.0959
    },
    {
      "epoch": 0.4713262470537524,
      "lm_loss": 0.269,
      "step": 6424,
      "vm_loss": 0.1768
    },
    {
      "epoch": 0.4713262470537524,
      "lm_loss": 0.4121,
      "step": 6424,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.4713262470537524,
      "lm_loss": 0.7676,
      "step": 6424,
      "vm_loss": 0.1548
    },
    {
      "epoch": 0.4713262470537524,
      "lm_loss": 1.2068,
      "step": 6424,
      "vm_loss": 0.2449
    },
    {
      "epoch": 0.4713262470537524,
      "lm_loss": 0.2762,
      "step": 6424,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.4713262470537524,
      "lm_loss": 0.438,
      "step": 6424,
      "vm_loss": 0.1443
    },
    {
      "epoch": 0.4713996166438915,
      "grad_norm": 1.5520219911367332,
      "learning_rate": 5.7034474348292065e-06,
      "loss": 0.7328,
      "step": 6425
    },
    {
      "epoch": 0.47147298623403067,
      "grad_norm": 1.4697569849862362,
      "learning_rate": 5.702271036138043e-06,
      "loss": 0.7495,
      "step": 6426
    },
    {
      "epoch": 0.47154635582416976,
      "grad_norm": 1.4616247188961409,
      "learning_rate": 5.701094597787896e-06,
      "loss": 0.8171,
      "step": 6427
    },
    {
      "epoch": 0.4716197254143089,
      "grad_norm": 1.4456479685381334,
      "learning_rate": 5.6999181198452e-06,
      "loss": 0.6921,
      "step": 6428
    },
    {
      "epoch": 0.471693095004448,
      "grad_norm": 0.5559083037985598,
      "learning_rate": 5.698741602376395e-06,
      "loss": 0.5147,
      "step": 6429
    },
    {
      "epoch": 0.47176646459458715,
      "grad_norm": 1.5793215286576328,
      "learning_rate": 5.697565045447922e-06,
      "loss": 0.7878,
      "step": 6430
    },
    {
      "epoch": 0.4718398341847263,
      "grad_norm": 1.4399469658795165,
      "learning_rate": 5.6963884491262224e-06,
      "loss": 0.7899,
      "step": 6431
    },
    {
      "epoch": 0.4719132037748654,
      "grad_norm": 1.3920887632392855,
      "learning_rate": 5.6952118134777415e-06,
      "loss": 0.7464,
      "step": 6432
    },
    {
      "epoch": 0.4719132037748654,
      "lm_loss": 1.2275,
      "step": 6432,
      "vm_loss": 0.0337
    },
    {
      "epoch": 0.4719132037748654,
      "lm_loss": 0.9962,
      "step": 6432,
      "vm_loss": 0.1784
    },
    {
      "epoch": 0.4719132037748654,
      "lm_loss": 0.6634,
      "step": 6432,
      "vm_loss": 0.1588
    },
    {
      "epoch": 0.4719132037748654,
      "lm_loss": 0.224,
      "step": 6432,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.4719132037748654,
      "lm_loss": 1.2247,
      "step": 6432,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.4719132037748654,
      "lm_loss": 0.3042,
      "step": 6432,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.4719132037748654,
      "lm_loss": 0.1541,
      "step": 6432,
      "vm_loss": 0.0889
    },
    {
      "epoch": 0.4719132037748654,
      "lm_loss": 0.2279,
      "step": 6432,
      "vm_loss": 0.2111
    },
    {
      "epoch": 0.47198657336500455,
      "grad_norm": 1.3903625272389324,
      "learning_rate": 5.694035138568929e-06,
      "loss": 0.7122,
      "step": 6433
    },
    {
      "epoch": 0.47205994295514364,
      "grad_norm": 0.5627669104705421,
      "learning_rate": 5.692858424466233e-06,
      "loss": 0.4998,
      "step": 6434
    },
    {
      "epoch": 0.4721333125452828,
      "grad_norm": 0.5715269215290631,
      "learning_rate": 5.691681671236106e-06,
      "loss": 0.4975,
      "step": 6435
    },
    {
      "epoch": 0.47220668213542194,
      "grad_norm": 1.439867664776347,
      "learning_rate": 5.690504878945003e-06,
      "loss": 0.8341,
      "step": 6436
    },
    {
      "epoch": 0.47228005172556103,
      "grad_norm": 1.4521562489509283,
      "learning_rate": 5.689328047659378e-06,
      "loss": 0.8597,
      "step": 6437
    },
    {
      "epoch": 0.4723534213157002,
      "grad_norm": 1.3391458705588868,
      "learning_rate": 5.688151177445693e-06,
      "loss": 0.7557,
      "step": 6438
    },
    {
      "epoch": 0.4724267909058393,
      "grad_norm": 1.3863801033305052,
      "learning_rate": 5.686974268370405e-06,
      "loss": 0.7826,
      "step": 6439
    },
    {
      "epoch": 0.4725001604959784,
      "grad_norm": 0.5625945891543601,
      "learning_rate": 5.685797320499979e-06,
      "loss": 0.5122,
      "step": 6440
    },
    {
      "epoch": 0.4725001604959784,
      "lm_loss": 1.2132,
      "step": 6440,
      "vm_loss": 0.0351
    },
    {
      "epoch": 0.4725001604959784,
      "lm_loss": 1.3127,
      "step": 6440,
      "vm_loss": 0.0584
    },
    {
      "epoch": 0.4725001604959784,
      "lm_loss": 0.4787,
      "step": 6440,
      "vm_loss": 0.0548
    },
    {
      "epoch": 0.4725001604959784,
      "lm_loss": 0.1411,
      "step": 6440,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.4725001604959784,
      "lm_loss": 0.6114,
      "step": 6440,
      "vm_loss": 0.0967
    },
    {
      "epoch": 0.4725001604959784,
      "lm_loss": 0.076,
      "step": 6440,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.4725001604959784,
      "lm_loss": 0.1111,
      "step": 6440,
      "vm_loss": 0.2016
    },
    {
      "epoch": 0.4725001604959784,
      "lm_loss": 1.1052,
      "step": 6440,
      "vm_loss": 0.2062
    },
    {
      "epoch": 0.4725735300861176,
      "grad_norm": 1.4018760729703559,
      "learning_rate": 5.684620333900881e-06,
      "loss": 0.8074,
      "step": 6441
    },
    {
      "epoch": 0.47264689967625667,
      "grad_norm": 0.5419573335735519,
      "learning_rate": 5.683443308639578e-06,
      "loss": 0.4911,
      "step": 6442
    },
    {
      "epoch": 0.4727202692663958,
      "grad_norm": 0.5418493908632018,
      "learning_rate": 5.682266244782537e-06,
      "loss": 0.4838,
      "step": 6443
    },
    {
      "epoch": 0.4727936388565349,
      "grad_norm": 1.6204309499689562,
      "learning_rate": 5.681089142396232e-06,
      "loss": 0.7534,
      "step": 6444
    },
    {
      "epoch": 0.47286700844667406,
      "grad_norm": 0.5544330004728135,
      "learning_rate": 5.679912001547136e-06,
      "loss": 0.5171,
      "step": 6445
    },
    {
      "epoch": 0.4729403780368132,
      "grad_norm": 1.459981809558742,
      "learning_rate": 5.6787348223017265e-06,
      "loss": 0.7019,
      "step": 6446
    },
    {
      "epoch": 0.4730137476269523,
      "grad_norm": 1.5466342145044873,
      "learning_rate": 5.677557604726481e-06,
      "loss": 0.8309,
      "step": 6447
    },
    {
      "epoch": 0.47308711721709146,
      "grad_norm": 1.5752771356766133,
      "learning_rate": 5.6763803488878776e-06,
      "loss": 0.861,
      "step": 6448
    },
    {
      "epoch": 0.47308711721709146,
      "lm_loss": 0.2471,
      "step": 6448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47308711721709146,
      "lm_loss": 0.3853,
      "step": 6448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47308711721709146,
      "lm_loss": 0.5761,
      "step": 6448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47308711721709146,
      "lm_loss": 0.5198,
      "step": 6448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47308711721709146,
      "lm_loss": 0.5715,
      "step": 6448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47308711721709146,
      "lm_loss": 0.5687,
      "step": 6448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47308711721709146,
      "lm_loss": 0.479,
      "step": 6448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47308711721709146,
      "lm_loss": 0.5797,
      "step": 6448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47316048680723055,
      "grad_norm": 0.5906756620597771,
      "learning_rate": 5.675203054852403e-06,
      "loss": 0.5213,
      "step": 6449
    },
    {
      "epoch": 0.4732338563973697,
      "grad_norm": 1.594480273701282,
      "learning_rate": 5.674025722686541e-06,
      "loss": 0.8361,
      "step": 6450
    },
    {
      "epoch": 0.47330722598750885,
      "grad_norm": 1.3886284903881405,
      "learning_rate": 5.672848352456774e-06,
      "loss": 0.7308,
      "step": 6451
    },
    {
      "epoch": 0.47338059557764794,
      "grad_norm": 1.3795467977446436,
      "learning_rate": 5.671670944229595e-06,
      "loss": 0.7398,
      "step": 6452
    },
    {
      "epoch": 0.4734539651677871,
      "grad_norm": 1.4419523400223484,
      "learning_rate": 5.670493498071494e-06,
      "loss": 0.8132,
      "step": 6453
    },
    {
      "epoch": 0.4735273347579262,
      "grad_norm": 0.5820727002775014,
      "learning_rate": 5.669316014048967e-06,
      "loss": 0.5049,
      "step": 6454
    },
    {
      "epoch": 0.47360070434806534,
      "grad_norm": 1.3529678281451927,
      "learning_rate": 5.668138492228505e-06,
      "loss": 0.7581,
      "step": 6455
    },
    {
      "epoch": 0.4736740739382045,
      "grad_norm": 0.559730776713294,
      "learning_rate": 5.666960932676609e-06,
      "loss": 0.5113,
      "step": 6456
    },
    {
      "epoch": 0.4736740739382045,
      "lm_loss": 0.7527,
      "step": 6456,
      "vm_loss": 0.0459
    },
    {
      "epoch": 0.4736740739382045,
      "lm_loss": 1.6951,
      "step": 6456,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.4736740739382045,
      "lm_loss": 0.4586,
      "step": 6456,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.4736740739382045,
      "lm_loss": 0.1408,
      "step": 6456,
      "vm_loss": 0.0869
    },
    {
      "epoch": 0.4736740739382045,
      "lm_loss": 0.2909,
      "step": 6456,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.4736740739382045,
      "lm_loss": 1.0892,
      "step": 6456,
      "vm_loss": 0.1584
    },
    {
      "epoch": 0.4736740739382045,
      "lm_loss": 0.7802,
      "step": 6456,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.4736740739382045,
      "lm_loss": 0.734,
      "step": 6456,
      "vm_loss": 0.2156
    },
    {
      "epoch": 0.4737474435283436,
      "grad_norm": 1.236068892680813,
      "learning_rate": 5.665783335459778e-06,
      "loss": 0.7174,
      "step": 6457
    },
    {
      "epoch": 0.47382081311848273,
      "grad_norm": 0.548000779076189,
      "learning_rate": 5.664605700644512e-06,
      "loss": 0.4926,
      "step": 6458
    },
    {
      "epoch": 0.4738941827086218,
      "grad_norm": 1.638191731308887,
      "learning_rate": 5.6634280282973165e-06,
      "loss": 0.7368,
      "step": 6459
    },
    {
      "epoch": 0.473967552298761,
      "grad_norm": 1.4526940158721886,
      "learning_rate": 5.662250318484698e-06,
      "loss": 0.7709,
      "step": 6460
    },
    {
      "epoch": 0.4740409218889001,
      "grad_norm": 1.520405225333553,
      "learning_rate": 5.661072571273162e-06,
      "loss": 0.7154,
      "step": 6461
    },
    {
      "epoch": 0.4741142914790392,
      "grad_norm": 0.5509104727881782,
      "learning_rate": 5.659894786729223e-06,
      "loss": 0.4938,
      "step": 6462
    },
    {
      "epoch": 0.47418766106917837,
      "grad_norm": 1.4478394286188692,
      "learning_rate": 5.6587169649193906e-06,
      "loss": 0.838,
      "step": 6463
    },
    {
      "epoch": 0.47426103065931746,
      "grad_norm": 1.2394034822879747,
      "learning_rate": 5.657539105910178e-06,
      "loss": 0.6448,
      "step": 6464
    },
    {
      "epoch": 0.47426103065931746,
      "lm_loss": 0.6951,
      "step": 6464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47426103065931746,
      "lm_loss": 0.6128,
      "step": 6464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47426103065931746,
      "lm_loss": 0.4969,
      "step": 6464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47426103065931746,
      "lm_loss": 0.574,
      "step": 6464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47426103065931746,
      "lm_loss": 0.4121,
      "step": 6464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47426103065931746,
      "lm_loss": 0.4172,
      "step": 6464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47426103065931746,
      "lm_loss": 0.5663,
      "step": 6464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47426103065931746,
      "lm_loss": 0.5388,
      "step": 6464,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4743344002494566,
      "grad_norm": 0.5791448123693377,
      "learning_rate": 5.656361209768106e-06,
      "loss": 0.5096,
      "step": 6465
    },
    {
      "epoch": 0.47440776983959576,
      "grad_norm": 0.57158200955057,
      "learning_rate": 5.655183276559689e-06,
      "loss": 0.506,
      "step": 6466
    },
    {
      "epoch": 0.47448113942973486,
      "grad_norm": 0.5742967771748733,
      "learning_rate": 5.6540053063514526e-06,
      "loss": 0.5006,
      "step": 6467
    },
    {
      "epoch": 0.474554509019874,
      "grad_norm": 1.5868310641172376,
      "learning_rate": 5.6528272992099154e-06,
      "loss": 0.8467,
      "step": 6468
    },
    {
      "epoch": 0.4746278786100131,
      "grad_norm": 1.3675046167702483,
      "learning_rate": 5.651649255201603e-06,
      "loss": 0.767,
      "step": 6469
    },
    {
      "epoch": 0.47470124820015225,
      "grad_norm": 1.579369557360913,
      "learning_rate": 5.6504711743930455e-06,
      "loss": 0.7726,
      "step": 6470
    },
    {
      "epoch": 0.47477461779029134,
      "grad_norm": 0.5815946471416922,
      "learning_rate": 5.649293056850769e-06,
      "loss": 0.5074,
      "step": 6471
    },
    {
      "epoch": 0.4748479873804305,
      "grad_norm": 1.5867144909248785,
      "learning_rate": 5.648114902641304e-06,
      "loss": 0.7183,
      "step": 6472
    },
    {
      "epoch": 0.4748479873804305,
      "lm_loss": 0.7079,
      "step": 6472,
      "vm_loss": 0.0422
    },
    {
      "epoch": 0.4748479873804305,
      "lm_loss": 0.0552,
      "step": 6472,
      "vm_loss": 0.1455
    },
    {
      "epoch": 0.4748479873804305,
      "lm_loss": 0.9944,
      "step": 6472,
      "vm_loss": 0.3048
    },
    {
      "epoch": 0.4748479873804305,
      "lm_loss": 0.4074,
      "step": 6472,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.4748479873804305,
      "lm_loss": 0.2852,
      "step": 6472,
      "vm_loss": 0.1416
    },
    {
      "epoch": 0.4748479873804305,
      "lm_loss": 0.2819,
      "step": 6472,
      "vm_loss": 0.1801
    },
    {
      "epoch": 0.4748479873804305,
      "lm_loss": 0.614,
      "step": 6472,
      "vm_loss": 0.2087
    },
    {
      "epoch": 0.4748479873804305,
      "lm_loss": 0.3631,
      "step": 6472,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.47492135697056964,
      "grad_norm": 1.5261687357180074,
      "learning_rate": 5.646936711831187e-06,
      "loss": 0.759,
      "step": 6473
    },
    {
      "epoch": 0.47499472656070874,
      "grad_norm": 0.5810675942711218,
      "learning_rate": 5.645758484486952e-06,
      "loss": 0.5112,
      "step": 6474
    },
    {
      "epoch": 0.4750680961508479,
      "grad_norm": 0.5723949692207371,
      "learning_rate": 5.644580220675135e-06,
      "loss": 0.5103,
      "step": 6475
    },
    {
      "epoch": 0.475141465740987,
      "grad_norm": 0.55042302037016,
      "learning_rate": 5.643401920462278e-06,
      "loss": 0.4901,
      "step": 6476
    },
    {
      "epoch": 0.47521483533112613,
      "grad_norm": 1.7697317047770544,
      "learning_rate": 5.64222358391492e-06,
      "loss": 0.8167,
      "step": 6477
    },
    {
      "epoch": 0.4752882049212653,
      "grad_norm": 0.5626155257681357,
      "learning_rate": 5.641045211099607e-06,
      "loss": 0.5045,
      "step": 6478
    },
    {
      "epoch": 0.4753615745114044,
      "grad_norm": 1.4672899541658826,
      "learning_rate": 5.639866802082883e-06,
      "loss": 0.7791,
      "step": 6479
    },
    {
      "epoch": 0.4754349441015435,
      "grad_norm": 1.603157340042364,
      "learning_rate": 5.6386883569312945e-06,
      "loss": 0.9156,
      "step": 6480
    },
    {
      "epoch": 0.4754349441015435,
      "lm_loss": 1.5258,
      "step": 6480,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.4754349441015435,
      "lm_loss": 1.0792,
      "step": 6480,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.4754349441015435,
      "lm_loss": 0.7076,
      "step": 6480,
      "vm_loss": 0.214
    },
    {
      "epoch": 0.4754349441015435,
      "lm_loss": 0.7,
      "step": 6480,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.4754349441015435,
      "lm_loss": 0.6981,
      "step": 6480,
      "vm_loss": 0.1744
    },
    {
      "epoch": 0.4754349441015435,
      "lm_loss": 0.0928,
      "step": 6480,
      "vm_loss": 0.0707
    },
    {
      "epoch": 0.4754349441015435,
      "lm_loss": 0.0833,
      "step": 6480,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.4754349441015435,
      "lm_loss": 0.7998,
      "step": 6480,
      "vm_loss": 0.092
    },
    {
      "epoch": 0.4755083136916826,
      "grad_norm": 1.540848753502155,
      "learning_rate": 5.637509875711395e-06,
      "loss": 0.8178,
      "step": 6481
    },
    {
      "epoch": 0.47558168328182177,
      "grad_norm": 1.4526466196944863,
      "learning_rate": 5.636331358489733e-06,
      "loss": 0.7092,
      "step": 6482
    },
    {
      "epoch": 0.4756550528719609,
      "grad_norm": 0.6484479745945839,
      "learning_rate": 5.635152805332864e-06,
      "loss": 0.5297,
      "step": 6483
    },
    {
      "epoch": 0.4757284224621,
      "grad_norm": 0.5910646135336937,
      "learning_rate": 5.633974216307343e-06,
      "loss": 0.5126,
      "step": 6484
    },
    {
      "epoch": 0.47580179205223916,
      "grad_norm": 0.5599965127493022,
      "learning_rate": 5.6327955914797276e-06,
      "loss": 0.5073,
      "step": 6485
    },
    {
      "epoch": 0.47587516164237825,
      "grad_norm": 1.5368254373994528,
      "learning_rate": 5.631616930916578e-06,
      "loss": 0.873,
      "step": 6486
    },
    {
      "epoch": 0.4759485312325174,
      "grad_norm": 1.7403925963512612,
      "learning_rate": 5.630438234684457e-06,
      "loss": 0.8611,
      "step": 6487
    },
    {
      "epoch": 0.47602190082265655,
      "grad_norm": 1.4717090687906538,
      "learning_rate": 5.629259502849925e-06,
      "loss": 0.7113,
      "step": 6488
    },
    {
      "epoch": 0.47602190082265655,
      "lm_loss": 1.3887,
      "step": 6488,
      "vm_loss": 0.0461
    },
    {
      "epoch": 0.47602190082265655,
      "lm_loss": 1.3565,
      "step": 6488,
      "vm_loss": 0.1511
    },
    {
      "epoch": 0.47602190082265655,
      "lm_loss": 1.3384,
      "step": 6488,
      "vm_loss": 0.2207
    },
    {
      "epoch": 0.47602190082265655,
      "lm_loss": 0.95,
      "step": 6488,
      "vm_loss": 0.1419
    },
    {
      "epoch": 0.47602190082265655,
      "lm_loss": 0.5715,
      "step": 6488,
      "vm_loss": 0.1268
    },
    {
      "epoch": 0.47602190082265655,
      "lm_loss": 0.6204,
      "step": 6488,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.47602190082265655,
      "lm_loss": 0.4796,
      "step": 6488,
      "vm_loss": 0.1921
    },
    {
      "epoch": 0.47602190082265655,
      "lm_loss": 1.1892,
      "step": 6488,
      "vm_loss": 0.1415
    },
    {
      "epoch": 0.47609527041279565,
      "grad_norm": 1.498700226565949,
      "learning_rate": 5.628080735479553e-06,
      "loss": 0.8152,
      "step": 6489
    },
    {
      "epoch": 0.4761686400029348,
      "grad_norm": 1.4277940558476256,
      "learning_rate": 5.626901932639906e-06,
      "loss": 0.7225,
      "step": 6490
    },
    {
      "epoch": 0.4762420095930739,
      "grad_norm": 1.5209031857552826,
      "learning_rate": 5.625723094397555e-06,
      "loss": 0.8511,
      "step": 6491
    },
    {
      "epoch": 0.47631537918321304,
      "grad_norm": 1.5868918523393551,
      "learning_rate": 5.624544220819071e-06,
      "loss": 0.8663,
      "step": 6492
    },
    {
      "epoch": 0.4763887487733522,
      "grad_norm": 0.8819854480172885,
      "learning_rate": 5.623365311971026e-06,
      "loss": 0.5182,
      "step": 6493
    },
    {
      "epoch": 0.4764621183634913,
      "grad_norm": 1.4653637229060392,
      "learning_rate": 5.622186367920001e-06,
      "loss": 0.6819,
      "step": 6494
    },
    {
      "epoch": 0.47653548795363043,
      "grad_norm": 0.6926457778371566,
      "learning_rate": 5.62100738873257e-06,
      "loss": 0.5079,
      "step": 6495
    },
    {
      "epoch": 0.4766088575437695,
      "grad_norm": 0.5783019169751589,
      "learning_rate": 5.619828374475313e-06,
      "loss": 0.5039,
      "step": 6496
    },
    {
      "epoch": 0.4766088575437695,
      "lm_loss": 0.4836,
      "step": 6496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4766088575437695,
      "lm_loss": 0.5646,
      "step": 6496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4766088575437695,
      "lm_loss": 0.4563,
      "step": 6496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4766088575437695,
      "lm_loss": 0.4918,
      "step": 6496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4766088575437695,
      "lm_loss": 0.5571,
      "step": 6496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4766088575437695,
      "lm_loss": 0.3931,
      "step": 6496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4766088575437695,
      "lm_loss": 0.5771,
      "step": 6496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4766088575437695,
      "lm_loss": 0.4926,
      "step": 6496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4766822271339087,
      "grad_norm": 0.5544028018749487,
      "learning_rate": 5.6186493252148135e-06,
      "loss": 0.4902,
      "step": 6497
    },
    {
      "epoch": 0.4767555967240478,
      "grad_norm": 1.692078737433901,
      "learning_rate": 5.6174702410176555e-06,
      "loss": 0.8318,
      "step": 6498
    },
    {
      "epoch": 0.4768289663141869,
      "grad_norm": 1.5104239641149915,
      "learning_rate": 5.616291121950421e-06,
      "loss": 0.7895,
      "step": 6499
    },
    {
      "epoch": 0.47690233590432607,
      "grad_norm": 0.7709086359372956,
      "learning_rate": 5.615111968079704e-06,
      "loss": 0.5104,
      "step": 6500
    },
    {
      "epoch": 0.47697570549446516,
      "grad_norm": 0.7425566972558851,
      "learning_rate": 5.613932779472088e-06,
      "loss": 0.5053,
      "step": 6501
    },
    {
      "epoch": 0.4770490750846043,
      "grad_norm": 1.5844789106820711,
      "learning_rate": 5.612753556194169e-06,
      "loss": 0.7742,
      "step": 6502
    },
    {
      "epoch": 0.47712244467474346,
      "grad_norm": 0.5830696754002456,
      "learning_rate": 5.61157429831254e-06,
      "loss": 0.5004,
      "step": 6503
    },
    {
      "epoch": 0.47719581426488256,
      "grad_norm": 1.4858914255254876,
      "learning_rate": 5.610395005893794e-06,
      "loss": 0.7457,
      "step": 6504
    },
    {
      "epoch": 0.47719581426488256,
      "lm_loss": 0.8458,
      "step": 6504,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.47719581426488256,
      "lm_loss": 0.954,
      "step": 6504,
      "vm_loss": 0.0991
    },
    {
      "epoch": 0.47719581426488256,
      "lm_loss": 0.7597,
      "step": 6504,
      "vm_loss": 0.0813
    },
    {
      "epoch": 0.47719581426488256,
      "lm_loss": 0.0655,
      "step": 6504,
      "vm_loss": 0.1872
    },
    {
      "epoch": 0.47719581426488256,
      "lm_loss": 0.3674,
      "step": 6504,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.47719581426488256,
      "lm_loss": 1.0379,
      "step": 6504,
      "vm_loss": 0.1318
    },
    {
      "epoch": 0.47719581426488256,
      "lm_loss": 0.263,
      "step": 6504,
      "vm_loss": 0.1217
    },
    {
      "epoch": 0.47719581426488256,
      "lm_loss": 0.375,
      "step": 6504,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.4772691838550217,
      "grad_norm": 1.8435578442093725,
      "learning_rate": 5.609215679004532e-06,
      "loss": 0.7086,
      "step": 6505
    },
    {
      "epoch": 0.4773425534451608,
      "grad_norm": 0.6769937399109747,
      "learning_rate": 5.608036317711352e-06,
      "loss": 0.5008,
      "step": 6506
    },
    {
      "epoch": 0.47741592303529995,
      "grad_norm": 1.4083125249314918,
      "learning_rate": 5.606856922080853e-06,
      "loss": 0.8786,
      "step": 6507
    },
    {
      "epoch": 0.47748929262543904,
      "grad_norm": 0.6601164223330687,
      "learning_rate": 5.6056774921796435e-06,
      "loss": 0.5119,
      "step": 6508
    },
    {
      "epoch": 0.4775626622155782,
      "grad_norm": 1.5167332157711406,
      "learning_rate": 5.604498028074323e-06,
      "loss": 0.8824,
      "step": 6509
    },
    {
      "epoch": 0.47763603180571734,
      "grad_norm": 1.560490290635277,
      "learning_rate": 5.603318529831505e-06,
      "loss": 0.7666,
      "step": 6510
    },
    {
      "epoch": 0.47770940139585644,
      "grad_norm": 0.5744249731160186,
      "learning_rate": 5.602138997517794e-06,
      "loss": 0.5089,
      "step": 6511
    },
    {
      "epoch": 0.4777827709859956,
      "grad_norm": 1.525860141693532,
      "learning_rate": 5.6009594311998025e-06,
      "loss": 0.7459,
      "step": 6512
    },
    {
      "epoch": 0.4777827709859956,
      "lm_loss": 0.5662,
      "step": 6512,
      "vm_loss": 0.0407
    },
    {
      "epoch": 0.4777827709859956,
      "lm_loss": 1.4152,
      "step": 6512,
      "vm_loss": 0.1404
    },
    {
      "epoch": 0.4777827709859956,
      "lm_loss": 0.9676,
      "step": 6512,
      "vm_loss": 0.1682
    },
    {
      "epoch": 0.4777827709859956,
      "lm_loss": 0.7766,
      "step": 6512,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.4777827709859956,
      "lm_loss": 0.4605,
      "step": 6512,
      "vm_loss": 0.1059
    },
    {
      "epoch": 0.4777827709859956,
      "lm_loss": 0.6699,
      "step": 6512,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.4777827709859956,
      "lm_loss": 0.5001,
      "step": 6512,
      "vm_loss": 0.1593
    },
    {
      "epoch": 0.4777827709859956,
      "lm_loss": 0.3929,
      "step": 6512,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.4778561405761347,
      "grad_norm": 1.5587779226959557,
      "learning_rate": 5.599779830944145e-06,
      "loss": 0.8206,
      "step": 6513
    },
    {
      "epoch": 0.47792951016627383,
      "grad_norm": 0.609140174208269,
      "learning_rate": 5.598600196817435e-06,
      "loss": 0.4951,
      "step": 6514
    },
    {
      "epoch": 0.478002879756413,
      "grad_norm": 1.3986500465621245,
      "learning_rate": 5.597420528886289e-06,
      "loss": 0.7878,
      "step": 6515
    },
    {
      "epoch": 0.4780762493465521,
      "grad_norm": 0.6088810938419346,
      "learning_rate": 5.596240827217325e-06,
      "loss": 0.5005,
      "step": 6516
    },
    {
      "epoch": 0.4781496189366912,
      "grad_norm": 1.4726603442349946,
      "learning_rate": 5.5950610918771666e-06,
      "loss": 0.817,
      "step": 6517
    },
    {
      "epoch": 0.4782229885268303,
      "grad_norm": 1.4572757898681736,
      "learning_rate": 5.593881322932433e-06,
      "loss": 0.849,
      "step": 6518
    },
    {
      "epoch": 0.47829635811696947,
      "grad_norm": 1.4089269292331525,
      "learning_rate": 5.592701520449751e-06,
      "loss": 0.787,
      "step": 6519
    },
    {
      "epoch": 0.4783697277071086,
      "grad_norm": 1.4100078968447678,
      "learning_rate": 5.591521684495746e-06,
      "loss": 0.8006,
      "step": 6520
    },
    {
      "epoch": 0.4783697277071086,
      "lm_loss": 1.3476,
      "step": 6520,
      "vm_loss": 0.0947
    },
    {
      "epoch": 0.4783697277071086,
      "lm_loss": 0.2716,
      "step": 6520,
      "vm_loss": 0.1241
    },
    {
      "epoch": 0.4783697277071086,
      "lm_loss": 0.2238,
      "step": 6520,
      "vm_loss": 0.0846
    },
    {
      "epoch": 0.4783697277071086,
      "lm_loss": 0.1772,
      "step": 6520,
      "vm_loss": 0.1153
    },
    {
      "epoch": 0.4783697277071086,
      "lm_loss": 0.4196,
      "step": 6520,
      "vm_loss": 0.076
    },
    {
      "epoch": 0.4783697277071086,
      "lm_loss": 0.1044,
      "step": 6520,
      "vm_loss": 0.0371
    },
    {
      "epoch": 0.4783697277071086,
      "lm_loss": 0.4477,
      "step": 6520,
      "vm_loss": 0.0681
    },
    {
      "epoch": 0.4783697277071086,
      "lm_loss": 0.8253,
      "step": 6520,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.4784430972972477,
      "grad_norm": 1.5337575824118215,
      "learning_rate": 5.5903418151370485e-06,
      "loss": 0.7577,
      "step": 6521
    },
    {
      "epoch": 0.47851646688738686,
      "grad_norm": 0.6049406458978437,
      "learning_rate": 5.5891619124402866e-06,
      "loss": 0.5027,
      "step": 6522
    },
    {
      "epoch": 0.47858983647752595,
      "grad_norm": 1.41001732824024,
      "learning_rate": 5.58798197647209e-06,
      "loss": 0.7369,
      "step": 6523
    },
    {
      "epoch": 0.4786632060676651,
      "grad_norm": 1.5729996771114017,
      "learning_rate": 5.586802007299097e-06,
      "loss": 0.7958,
      "step": 6524
    },
    {
      "epoch": 0.47873657565780425,
      "grad_norm": 1.3547282666399423,
      "learning_rate": 5.58562200498794e-06,
      "loss": 0.726,
      "step": 6525
    },
    {
      "epoch": 0.47880994524794335,
      "grad_norm": 1.5605139712433627,
      "learning_rate": 5.584441969605259e-06,
      "loss": 0.8663,
      "step": 6526
    },
    {
      "epoch": 0.4788833148380825,
      "grad_norm": 1.72288658844512,
      "learning_rate": 5.5832619012176936e-06,
      "loss": 0.8261,
      "step": 6527
    },
    {
      "epoch": 0.4789566844282216,
      "grad_norm": 0.5813395244759645,
      "learning_rate": 5.582081799891882e-06,
      "loss": 0.499,
      "step": 6528
    },
    {
      "epoch": 0.4789566844282216,
      "lm_loss": 0.6143,
      "step": 6528,
      "vm_loss": 0.0346
    },
    {
      "epoch": 0.4789566844282216,
      "lm_loss": 0.8809,
      "step": 6528,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.4789566844282216,
      "lm_loss": 0.1322,
      "step": 6528,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.4789566844282216,
      "lm_loss": 0.872,
      "step": 6528,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.4789566844282216,
      "lm_loss": 0.6623,
      "step": 6528,
      "vm_loss": 0.1424
    },
    {
      "epoch": 0.4789566844282216,
      "lm_loss": 0.1893,
      "step": 6528,
      "vm_loss": 0.1936
    },
    {
      "epoch": 0.4789566844282216,
      "lm_loss": 1.3306,
      "step": 6528,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.4789566844282216,
      "lm_loss": 0.2312,
      "step": 6528,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.47903005401836074,
      "grad_norm": 1.6350631920120613,
      "learning_rate": 5.580901665694471e-06,
      "loss": 0.6607,
      "step": 6529
    },
    {
      "epoch": 0.4791034236084999,
      "grad_norm": 1.4941416471335,
      "learning_rate": 5.5797214986921055e-06,
      "loss": 0.8518,
      "step": 6530
    },
    {
      "epoch": 0.479176793198639,
      "grad_norm": 0.5662137073489679,
      "learning_rate": 5.5785412989514295e-06,
      "loss": 0.5073,
      "step": 6531
    },
    {
      "epoch": 0.47925016278877813,
      "grad_norm": 0.5689384264514408,
      "learning_rate": 5.577361066539095e-06,
      "loss": 0.4996,
      "step": 6532
    },
    {
      "epoch": 0.4793235323789172,
      "grad_norm": 1.6172607021910959,
      "learning_rate": 5.57618080152175e-06,
      "loss": 0.7915,
      "step": 6533
    },
    {
      "epoch": 0.4793969019690564,
      "grad_norm": 0.5563111672319885,
      "learning_rate": 5.575000503966049e-06,
      "loss": 0.5038,
      "step": 6534
    },
    {
      "epoch": 0.4794702715591955,
      "grad_norm": 1.43099443699254,
      "learning_rate": 5.5738201739386465e-06,
      "loss": 0.8118,
      "step": 6535
    },
    {
      "epoch": 0.4795436411493346,
      "grad_norm": 1.3841219027216796,
      "learning_rate": 5.572639811506195e-06,
      "loss": 0.703,
      "step": 6536
    },
    {
      "epoch": 0.4795436411493346,
      "lm_loss": 0.6507,
      "step": 6536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4795436411493346,
      "lm_loss": 0.5511,
      "step": 6536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4795436411493346,
      "lm_loss": 0.3457,
      "step": 6536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4795436411493346,
      "lm_loss": 0.6057,
      "step": 6536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4795436411493346,
      "lm_loss": 0.4659,
      "step": 6536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4795436411493346,
      "lm_loss": 0.4922,
      "step": 6536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4795436411493346,
      "lm_loss": 0.5405,
      "step": 6536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4795436411493346,
      "lm_loss": 0.4067,
      "step": 6536,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.47961701073947377,
      "grad_norm": 0.5310217314165491,
      "learning_rate": 5.571459416735358e-06,
      "loss": 0.5148,
      "step": 6537
    },
    {
      "epoch": 0.47969038032961286,
      "grad_norm": 1.466951682345421,
      "learning_rate": 5.570278989692791e-06,
      "loss": 0.6844,
      "step": 6538
    },
    {
      "epoch": 0.479763749919752,
      "grad_norm": 0.5911126020070118,
      "learning_rate": 5.5690985304451575e-06,
      "loss": 0.5247,
      "step": 6539
    },
    {
      "epoch": 0.47983711950989116,
      "grad_norm": 0.5895729797540052,
      "learning_rate": 5.56791803905912e-06,
      "loss": 0.5211,
      "step": 6540
    },
    {
      "epoch": 0.47991048910003026,
      "grad_norm": 1.488728365567483,
      "learning_rate": 5.5667375156013445e-06,
      "loss": 0.8528,
      "step": 6541
    },
    {
      "epoch": 0.4799838586901694,
      "grad_norm": 1.5117493419877868,
      "learning_rate": 5.5655569601384995e-06,
      "loss": 0.781,
      "step": 6542
    },
    {
      "epoch": 0.4800572282803085,
      "grad_norm": 1.4391294338424496,
      "learning_rate": 5.5643763727372514e-06,
      "loss": 0.7803,
      "step": 6543
    },
    {
      "epoch": 0.48013059787044765,
      "grad_norm": 1.4547454808223468,
      "learning_rate": 5.563195753464271e-06,
      "loss": 0.7367,
      "step": 6544
    },
    {
      "epoch": 0.48013059787044765,
      "lm_loss": 1.1081,
      "step": 6544,
      "vm_loss": 0.0617
    },
    {
      "epoch": 0.48013059787044765,
      "lm_loss": 0.6729,
      "step": 6544,
      "vm_loss": 0.1616
    },
    {
      "epoch": 0.48013059787044765,
      "lm_loss": 0.3854,
      "step": 6544,
      "vm_loss": 0.1308
    },
    {
      "epoch": 0.48013059787044765,
      "lm_loss": 0.1838,
      "step": 6544,
      "vm_loss": 0.1012
    },
    {
      "epoch": 0.48013059787044765,
      "lm_loss": 0.8723,
      "step": 6544,
      "vm_loss": 0.1216
    },
    {
      "epoch": 0.48013059787044765,
      "lm_loss": 0.8739,
      "step": 6544,
      "vm_loss": 0.1032
    },
    {
      "epoch": 0.48013059787044765,
      "lm_loss": 0.4411,
      "step": 6544,
      "vm_loss": 0.2366
    },
    {
      "epoch": 0.48013059787044765,
      "lm_loss": 0.8732,
      "step": 6544,
      "vm_loss": 0.1572
    },
    {
      "epoch": 0.4802039674605868,
      "grad_norm": 1.4595399044062987,
      "learning_rate": 5.562015102386234e-06,
      "loss": 0.7947,
      "step": 6545
    },
    {
      "epoch": 0.4802773370507259,
      "grad_norm": 1.4361555977529943,
      "learning_rate": 5.560834419569811e-06,
      "loss": 0.8613,
      "step": 6546
    },
    {
      "epoch": 0.48035070664086504,
      "grad_norm": 1.4159549001954688,
      "learning_rate": 5.559653705081679e-06,
      "loss": 0.7707,
      "step": 6547
    },
    {
      "epoch": 0.48042407623100414,
      "grad_norm": 1.4750586040331022,
      "learning_rate": 5.558472958988517e-06,
      "loss": 0.7856,
      "step": 6548
    },
    {
      "epoch": 0.4804974458211433,
      "grad_norm": 0.6449733382487802,
      "learning_rate": 5.557292181357003e-06,
      "loss": 0.4982,
      "step": 6549
    },
    {
      "epoch": 0.4805708154112824,
      "grad_norm": 1.548284143609847,
      "learning_rate": 5.5561113722538196e-06,
      "loss": 0.7433,
      "step": 6550
    },
    {
      "epoch": 0.48064418500142153,
      "grad_norm": 1.507930259150619,
      "learning_rate": 5.5549305317456514e-06,
      "loss": 0.8998,
      "step": 6551
    },
    {
      "epoch": 0.4807175545915607,
      "grad_norm": 1.5507034123108696,
      "learning_rate": 5.553749659899179e-06,
      "loss": 0.7846,
      "step": 6552
    },
    {
      "epoch": 0.4807175545915607,
      "lm_loss": 1.5359,
      "step": 6552,
      "vm_loss": 0.025
    },
    {
      "epoch": 0.4807175545915607,
      "lm_loss": 1.0926,
      "step": 6552,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.4807175545915607,
      "lm_loss": 0.7156,
      "step": 6552,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.4807175545915607,
      "lm_loss": 0.0693,
      "step": 6552,
      "vm_loss": 0.1212
    },
    {
      "epoch": 0.4807175545915607,
      "lm_loss": 0.8697,
      "step": 6552,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.4807175545915607,
      "lm_loss": 0.8316,
      "step": 6552,
      "vm_loss": 0.2093
    },
    {
      "epoch": 0.4807175545915607,
      "lm_loss": 1.1198,
      "step": 6552,
      "vm_loss": 0.1466
    },
    {
      "epoch": 0.4807175545915607,
      "lm_loss": 0.7405,
      "step": 6552,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.4807909241816998,
      "grad_norm": 1.389851726507744,
      "learning_rate": 5.5525687567810946e-06,
      "loss": 0.8485,
      "step": 6553
    },
    {
      "epoch": 0.4808642937718389,
      "grad_norm": 1.3890693704534447,
      "learning_rate": 5.551387822458084e-06,
      "loss": 0.734,
      "step": 6554
    },
    {
      "epoch": 0.480937663361978,
      "grad_norm": 0.5612407099622013,
      "learning_rate": 5.550206856996836e-06,
      "loss": 0.5064,
      "step": 6555
    },
    {
      "epoch": 0.48101103295211717,
      "grad_norm": 1.4319263798222288,
      "learning_rate": 5.549025860464044e-06,
      "loss": 0.8342,
      "step": 6556
    },
    {
      "epoch": 0.4810844025422563,
      "grad_norm": 1.5852721605472908,
      "learning_rate": 5.547844832926404e-06,
      "loss": 0.6912,
      "step": 6557
    },
    {
      "epoch": 0.4811577721323954,
      "grad_norm": 0.5612747683141034,
      "learning_rate": 5.546663774450608e-06,
      "loss": 0.4974,
      "step": 6558
    },
    {
      "epoch": 0.48123114172253456,
      "grad_norm": 2.0107624193643523,
      "learning_rate": 5.545482685103356e-06,
      "loss": 0.8516,
      "step": 6559
    },
    {
      "epoch": 0.48130451131267366,
      "grad_norm": 1.4785176815549996,
      "learning_rate": 5.544301564951344e-06,
      "loss": 0.815,
      "step": 6560
    },
    {
      "epoch": 0.48130451131267366,
      "lm_loss": 0.6129,
      "step": 6560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48130451131267366,
      "lm_loss": 0.4964,
      "step": 6560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48130451131267366,
      "lm_loss": 0.5431,
      "step": 6560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48130451131267366,
      "lm_loss": 0.3662,
      "step": 6560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48130451131267366,
      "lm_loss": 0.5563,
      "step": 6560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48130451131267366,
      "lm_loss": 0.5367,
      "step": 6560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48130451131267366,
      "lm_loss": 0.5527,
      "step": 6560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48130451131267366,
      "lm_loss": 0.4891,
      "step": 6560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4813778809028128,
      "grad_norm": 0.58968920827695,
      "learning_rate": 5.543120414061277e-06,
      "loss": 0.5128,
      "step": 6561
    },
    {
      "epoch": 0.48145125049295195,
      "grad_norm": 1.5828384014815673,
      "learning_rate": 5.541939232499855e-06,
      "loss": 0.8866,
      "step": 6562
    },
    {
      "epoch": 0.48152462008309105,
      "grad_norm": 1.4281978207438182,
      "learning_rate": 5.540758020333781e-06,
      "loss": 0.8144,
      "step": 6563
    },
    {
      "epoch": 0.4815979896732302,
      "grad_norm": 1.4195576822704132,
      "learning_rate": 5.5395767776297624e-06,
      "loss": 0.7552,
      "step": 6564
    },
    {
      "epoch": 0.4816713592633693,
      "grad_norm": 1.7054808411650084,
      "learning_rate": 5.538395504454509e-06,
      "loss": 0.9038,
      "step": 6565
    },
    {
      "epoch": 0.48174472885350844,
      "grad_norm": 0.568595769669594,
      "learning_rate": 5.537214200874726e-06,
      "loss": 0.4933,
      "step": 6566
    },
    {
      "epoch": 0.4818180984436476,
      "grad_norm": 1.4151570164184717,
      "learning_rate": 5.536032866957127e-06,
      "loss": 0.798,
      "step": 6567
    },
    {
      "epoch": 0.4818914680337867,
      "grad_norm": 0.5871157781064962,
      "learning_rate": 5.534851502768425e-06,
      "loss": 0.502,
      "step": 6568
    },
    {
      "epoch": 0.4818914680337867,
      "lm_loss": 0.7484,
      "step": 6568,
      "vm_loss": 0.069
    },
    {
      "epoch": 0.4818914680337867,
      "lm_loss": 1.024,
      "step": 6568,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.4818914680337867,
      "lm_loss": 0.0695,
      "step": 6568,
      "vm_loss": 0.2175
    },
    {
      "epoch": 0.4818914680337867,
      "lm_loss": 0.4975,
      "step": 6568,
      "vm_loss": 0.1094
    },
    {
      "epoch": 0.4818914680337867,
      "lm_loss": 0.776,
      "step": 6568,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.4818914680337867,
      "lm_loss": 0.1379,
      "step": 6568,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.4818914680337867,
      "lm_loss": 0.0584,
      "step": 6568,
      "vm_loss": 0.0753
    },
    {
      "epoch": 0.4818914680337867,
      "lm_loss": 0.3136,
      "step": 6568,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.48196483762392583,
      "grad_norm": 1.5871202172892294,
      "learning_rate": 5.533670108375334e-06,
      "loss": 0.8131,
      "step": 6569
    },
    {
      "epoch": 0.48203820721406493,
      "grad_norm": 0.5841369127731569,
      "learning_rate": 5.532488683844571e-06,
      "loss": 0.5114,
      "step": 6570
    },
    {
      "epoch": 0.4821115768042041,
      "grad_norm": 1.502162761722712,
      "learning_rate": 5.5313072292428526e-06,
      "loss": 0.9284,
      "step": 6571
    },
    {
      "epoch": 0.48218494639434323,
      "grad_norm": 1.353249962654486,
      "learning_rate": 5.530125744636899e-06,
      "loss": 0.7631,
      "step": 6572
    },
    {
      "epoch": 0.4822583159844823,
      "grad_norm": 1.5363628186555713,
      "learning_rate": 5.528944230093434e-06,
      "loss": 0.8531,
      "step": 6573
    },
    {
      "epoch": 0.48233168557462147,
      "grad_norm": 0.5807009099688749,
      "learning_rate": 5.5277626856791755e-06,
      "loss": 0.5022,
      "step": 6574
    },
    {
      "epoch": 0.48240505516476057,
      "grad_norm": 0.5868282958239782,
      "learning_rate": 5.526581111460852e-06,
      "loss": 0.5139,
      "step": 6575
    },
    {
      "epoch": 0.4824784247548997,
      "grad_norm": 1.509865124974825,
      "learning_rate": 5.525399507505189e-06,
      "loss": 0.8295,
      "step": 6576
    },
    {
      "epoch": 0.4824784247548997,
      "lm_loss": 0.6295,
      "step": 6576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4824784247548997,
      "lm_loss": 0.4974,
      "step": 6576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4824784247548997,
      "lm_loss": 0.511,
      "step": 6576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4824784247548997,
      "lm_loss": 0.6036,
      "step": 6576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4824784247548997,
      "lm_loss": 0.5057,
      "step": 6576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4824784247548997,
      "lm_loss": 0.5544,
      "step": 6576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4824784247548997,
      "lm_loss": 0.4929,
      "step": 6576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4824784247548997,
      "lm_loss": 0.5722,
      "step": 6576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48255179434503886,
      "grad_norm": 0.5858443036798164,
      "learning_rate": 5.524217873878916e-06,
      "loss": 0.5186,
      "step": 6577
    },
    {
      "epoch": 0.48262516393517796,
      "grad_norm": 0.5792063934679113,
      "learning_rate": 5.52303621064876e-06,
      "loss": 0.4989,
      "step": 6578
    },
    {
      "epoch": 0.4826985335253171,
      "grad_norm": 1.374983769111393,
      "learning_rate": 5.521854517881454e-06,
      "loss": 0.7607,
      "step": 6579
    },
    {
      "epoch": 0.4827719031154562,
      "grad_norm": 1.5355595364203216,
      "learning_rate": 5.5206727956437315e-06,
      "loss": 0.7454,
      "step": 6580
    },
    {
      "epoch": 0.48284527270559535,
      "grad_norm": 1.5918011563788936,
      "learning_rate": 5.519491044002327e-06,
      "loss": 0.8357,
      "step": 6581
    },
    {
      "epoch": 0.4829186422957345,
      "grad_norm": 1.6501591588108115,
      "learning_rate": 5.518309263023976e-06,
      "loss": 0.728,
      "step": 6582
    },
    {
      "epoch": 0.4829920118858736,
      "grad_norm": 1.4860552792865314,
      "learning_rate": 5.517127452775418e-06,
      "loss": 0.743,
      "step": 6583
    },
    {
      "epoch": 0.48306538147601275,
      "grad_norm": 0.5582032690070188,
      "learning_rate": 5.5159456133233914e-06,
      "loss": 0.4916,
      "step": 6584
    },
    {
      "epoch": 0.48306538147601275,
      "lm_loss": 1.2818,
      "step": 6584,
      "vm_loss": 0.1147
    },
    {
      "epoch": 0.48306538147601275,
      "lm_loss": 1.2853,
      "step": 6584,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.48306538147601275,
      "lm_loss": 0.8037,
      "step": 6584,
      "vm_loss": 0.2088
    },
    {
      "epoch": 0.48306538147601275,
      "lm_loss": 0.8252,
      "step": 6584,
      "vm_loss": 0.2482
    },
    {
      "epoch": 0.48306538147601275,
      "lm_loss": 0.4881,
      "step": 6584,
      "vm_loss": 0.1998
    },
    {
      "epoch": 0.48306538147601275,
      "lm_loss": 0.4087,
      "step": 6584,
      "vm_loss": 0.1671
    },
    {
      "epoch": 0.48306538147601275,
      "lm_loss": 1.1157,
      "step": 6584,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.48306538147601275,
      "lm_loss": 0.5984,
      "step": 6584,
      "vm_loss": 0.0824
    },
    {
      "epoch": 0.48313875106615184,
      "grad_norm": 1.3762476318763552,
      "learning_rate": 5.514763744734639e-06,
      "loss": 0.8256,
      "step": 6585
    },
    {
      "epoch": 0.483212120656291,
      "grad_norm": 0.6028450256807459,
      "learning_rate": 5.513581847075904e-06,
      "loss": 0.5004,
      "step": 6586
    },
    {
      "epoch": 0.48328549024643014,
      "grad_norm": 0.5773888294720083,
      "learning_rate": 5.512399920413928e-06,
      "loss": 0.5,
      "step": 6587
    },
    {
      "epoch": 0.48335885983656923,
      "grad_norm": 1.5149816138293053,
      "learning_rate": 5.511217964815461e-06,
      "loss": 0.8159,
      "step": 6588
    },
    {
      "epoch": 0.4834322294267084,
      "grad_norm": 1.4871392784062119,
      "learning_rate": 5.510035980347249e-06,
      "loss": 0.8629,
      "step": 6589
    },
    {
      "epoch": 0.4835055990168475,
      "grad_norm": 1.5369516795102767,
      "learning_rate": 5.508853967076042e-06,
      "loss": 0.7982,
      "step": 6590
    },
    {
      "epoch": 0.4835789686069866,
      "grad_norm": 0.5662852848275295,
      "learning_rate": 5.507671925068591e-06,
      "loss": 0.4939,
      "step": 6591
    },
    {
      "epoch": 0.4836523381971257,
      "grad_norm": 0.5967575525260893,
      "learning_rate": 5.50648985439165e-06,
      "loss": 0.5084,
      "step": 6592
    },
    {
      "epoch": 0.4836523381971257,
      "lm_loss": 1.1299,
      "step": 6592,
      "vm_loss": 0.1018
    },
    {
      "epoch": 0.4836523381971257,
      "lm_loss": 0.5326,
      "step": 6592,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.4836523381971257,
      "lm_loss": 0.4315,
      "step": 6592,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.4836523381971257,
      "lm_loss": 0.7712,
      "step": 6592,
      "vm_loss": 0.0469
    },
    {
      "epoch": 0.4836523381971257,
      "lm_loss": 0.7342,
      "step": 6592,
      "vm_loss": 0.0627
    },
    {
      "epoch": 0.4836523381971257,
      "lm_loss": 0.1716,
      "step": 6592,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.4836523381971257,
      "lm_loss": 0.9363,
      "step": 6592,
      "vm_loss": 0.1567
    },
    {
      "epoch": 0.4836523381971257,
      "lm_loss": 1.2157,
      "step": 6592,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.48372570778726487,
      "grad_norm": 1.4259244538354197,
      "learning_rate": 5.505307755111973e-06,
      "loss": 0.7197,
      "step": 6593
    },
    {
      "epoch": 0.483799077377404,
      "grad_norm": 0.5733027254727581,
      "learning_rate": 5.504125627296316e-06,
      "loss": 0.4826,
      "step": 6594
    },
    {
      "epoch": 0.4838724469675431,
      "grad_norm": 1.475744700041454,
      "learning_rate": 5.502943471011435e-06,
      "loss": 0.803,
      "step": 6595
    },
    {
      "epoch": 0.48394581655768226,
      "grad_norm": 1.4034590757214023,
      "learning_rate": 5.501761286324091e-06,
      "loss": 0.7335,
      "step": 6596
    },
    {
      "epoch": 0.48401918614782136,
      "grad_norm": 1.57797452282632,
      "learning_rate": 5.500579073301046e-06,
      "loss": 0.7731,
      "step": 6597
    },
    {
      "epoch": 0.4840925557379605,
      "grad_norm": 1.5270991688241253,
      "learning_rate": 5.499396832009059e-06,
      "loss": 0.786,
      "step": 6598
    },
    {
      "epoch": 0.48416592532809966,
      "grad_norm": 0.6346302387344075,
      "learning_rate": 5.498214562514896e-06,
      "loss": 0.5039,
      "step": 6599
    },
    {
      "epoch": 0.48423929491823875,
      "grad_norm": 0.6361967808580146,
      "learning_rate": 5.497032264885325e-06,
      "loss": 0.4748,
      "step": 6600
    },
    {
      "epoch": 0.48423929491823875,
      "lm_loss": 0.4663,
      "step": 6600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48423929491823875,
      "lm_loss": 0.5704,
      "step": 6600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48423929491823875,
      "lm_loss": 0.4031,
      "step": 6600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48423929491823875,
      "lm_loss": 0.4788,
      "step": 6600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48423929491823875,
      "lm_loss": 0.5011,
      "step": 6600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48423929491823875,
      "lm_loss": 0.5107,
      "step": 6600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48423929491823875,
      "lm_loss": 0.4344,
      "step": 6600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48423929491823875,
      "lm_loss": 0.4057,
      "step": 6600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4843126645083779,
      "grad_norm": 0.5758409311731059,
      "learning_rate": 5.495849939187109e-06,
      "loss": 0.5063,
      "step": 6601
    },
    {
      "epoch": 0.484386034098517,
      "grad_norm": 0.555438942162049,
      "learning_rate": 5.494667585487022e-06,
      "loss": 0.4974,
      "step": 6602
    },
    {
      "epoch": 0.48445940368865614,
      "grad_norm": 1.8253041056998043,
      "learning_rate": 5.493485203851827e-06,
      "loss": 0.7701,
      "step": 6603
    },
    {
      "epoch": 0.4845327732787953,
      "grad_norm": 1.6136100626599035,
      "learning_rate": 5.492302794348303e-06,
      "loss": 0.7485,
      "step": 6604
    },
    {
      "epoch": 0.4846061428689344,
      "grad_norm": 1.362414750040477,
      "learning_rate": 5.491120357043221e-06,
      "loss": 0.6297,
      "step": 6605
    },
    {
      "epoch": 0.48467951245907354,
      "grad_norm": 1.4468607179149346,
      "learning_rate": 5.489937892003357e-06,
      "loss": 0.7129,
      "step": 6606
    },
    {
      "epoch": 0.48475288204921263,
      "grad_norm": 1.5599315189003449,
      "learning_rate": 5.488755399295487e-06,
      "loss": 0.8772,
      "step": 6607
    },
    {
      "epoch": 0.4848262516393518,
      "grad_norm": 1.5439278045830165,
      "learning_rate": 5.487572878986387e-06,
      "loss": 0.8221,
      "step": 6608
    },
    {
      "epoch": 0.4848262516393518,
      "lm_loss": 1.862,
      "step": 6608,
      "vm_loss": 0.1233
    },
    {
      "epoch": 0.4848262516393518,
      "lm_loss": 0.0477,
      "step": 6608,
      "vm_loss": 0.1069
    },
    {
      "epoch": 0.4848262516393518,
      "lm_loss": 0.1667,
      "step": 6608,
      "vm_loss": 0.0562
    },
    {
      "epoch": 0.4848262516393518,
      "lm_loss": 0.4026,
      "step": 6608,
      "vm_loss": 0.1
    },
    {
      "epoch": 0.4848262516393518,
      "lm_loss": 1.0379,
      "step": 6608,
      "vm_loss": 0.1866
    },
    {
      "epoch": 0.4848262516393518,
      "lm_loss": 0.5226,
      "step": 6608,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.4848262516393518,
      "lm_loss": 0.6107,
      "step": 6608,
      "vm_loss": 0.1
    },
    {
      "epoch": 0.4848262516393518,
      "lm_loss": 1.0741,
      "step": 6608,
      "vm_loss": 0.1584
    },
    {
      "epoch": 0.48489962122949093,
      "grad_norm": 1.5685395839087426,
      "learning_rate": 5.486390331142841e-06,
      "loss": 0.7687,
      "step": 6609
    },
    {
      "epoch": 0.48497299081963,
      "grad_norm": 1.5130011469934128,
      "learning_rate": 5.485207755831629e-06,
      "loss": 0.7234,
      "step": 6610
    },
    {
      "epoch": 0.4850463604097692,
      "grad_norm": 0.95386742501132,
      "learning_rate": 5.484025153119531e-06,
      "loss": 0.5161,
      "step": 6611
    },
    {
      "epoch": 0.48511972999990827,
      "grad_norm": 1.4982455622986506,
      "learning_rate": 5.482842523073335e-06,
      "loss": 0.7756,
      "step": 6612
    },
    {
      "epoch": 0.4851930995900474,
      "grad_norm": 0.6557922637620556,
      "learning_rate": 5.481659865759827e-06,
      "loss": 0.4921,
      "step": 6613
    },
    {
      "epoch": 0.48526646918018657,
      "grad_norm": 0.5696037830097862,
      "learning_rate": 5.480477181245792e-06,
      "loss": 0.5125,
      "step": 6614
    },
    {
      "epoch": 0.48533983877032566,
      "grad_norm": 0.5682326333556291,
      "learning_rate": 5.479294469598021e-06,
      "loss": 0.4908,
      "step": 6615
    },
    {
      "epoch": 0.4854132083604648,
      "grad_norm": 1.7507825929808,
      "learning_rate": 5.478111730883305e-06,
      "loss": 0.8267,
      "step": 6616
    },
    {
      "epoch": 0.4854132083604648,
      "lm_loss": 1.2608,
      "step": 6616,
      "vm_loss": 0.0459
    },
    {
      "epoch": 0.4854132083604648,
      "lm_loss": 0.8312,
      "step": 6616,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.4854132083604648,
      "lm_loss": 0.4929,
      "step": 6616,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.4854132083604648,
      "lm_loss": 0.0827,
      "step": 6616,
      "vm_loss": 0.1365
    },
    {
      "epoch": 0.4854132083604648,
      "lm_loss": 0.9425,
      "step": 6616,
      "vm_loss": 0.0659
    },
    {
      "epoch": 0.4854132083604648,
      "lm_loss": 0.6992,
      "step": 6616,
      "vm_loss": 0.0539
    },
    {
      "epoch": 0.4854132083604648,
      "lm_loss": 0.2993,
      "step": 6616,
      "vm_loss": 0.0753
    },
    {
      "epoch": 0.4854132083604648,
      "lm_loss": 1.1764,
      "step": 6616,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.4854865779506039,
      "grad_norm": 1.4955759080917483,
      "learning_rate": 5.476928965168435e-06,
      "loss": 0.7568,
      "step": 6617
    },
    {
      "epoch": 0.48555994754074305,
      "grad_norm": 1.5706414282098848,
      "learning_rate": 5.475746172520205e-06,
      "loss": 0.8037,
      "step": 6618
    },
    {
      "epoch": 0.4856333171308822,
      "grad_norm": 1.7304013586733507,
      "learning_rate": 5.474563353005411e-06,
      "loss": 0.7916,
      "step": 6619
    },
    {
      "epoch": 0.4857066867210213,
      "grad_norm": 1.4108444419606163,
      "learning_rate": 5.47338050669085e-06,
      "loss": 0.8619,
      "step": 6620
    },
    {
      "epoch": 0.48578005631116045,
      "grad_norm": 0.9712539585152197,
      "learning_rate": 5.472197633643319e-06,
      "loss": 0.5029,
      "step": 6621
    },
    {
      "epoch": 0.48585342590129954,
      "grad_norm": 1.6512425710319873,
      "learning_rate": 5.471014733929617e-06,
      "loss": 0.7593,
      "step": 6622
    },
    {
      "epoch": 0.4859267954914387,
      "grad_norm": 1.6394279325078494,
      "learning_rate": 5.469831807616548e-06,
      "loss": 0.7723,
      "step": 6623
    },
    {
      "epoch": 0.48600016508157784,
      "grad_norm": 1.4583127590465668,
      "learning_rate": 5.468648854770912e-06,
      "loss": 0.7936,
      "step": 6624
    },
    {
      "epoch": 0.48600016508157784,
      "lm_loss": 0.3835,
      "step": 6624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48600016508157784,
      "lm_loss": 0.3238,
      "step": 6624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48600016508157784,
      "lm_loss": 0.4827,
      "step": 6624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48600016508157784,
      "lm_loss": 0.5389,
      "step": 6624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48600016508157784,
      "lm_loss": 0.3355,
      "step": 6624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48600016508157784,
      "lm_loss": 0.3655,
      "step": 6624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48600016508157784,
      "lm_loss": 0.5189,
      "step": 6624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48600016508157784,
      "lm_loss": 0.4226,
      "step": 6624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48607353467171693,
      "grad_norm": 0.6082572807597854,
      "learning_rate": 5.467465875459516e-06,
      "loss": 0.4977,
      "step": 6625
    },
    {
      "epoch": 0.4861469042618561,
      "grad_norm": 1.6062672979801007,
      "learning_rate": 5.4662828697491635e-06,
      "loss": 0.7698,
      "step": 6626
    },
    {
      "epoch": 0.4862202738519952,
      "grad_norm": 1.5908473355078179,
      "learning_rate": 5.465099837706662e-06,
      "loss": 0.8496,
      "step": 6627
    },
    {
      "epoch": 0.4862936434421343,
      "grad_norm": 1.6121207382991827,
      "learning_rate": 5.463916779398822e-06,
      "loss": 0.7556,
      "step": 6628
    },
    {
      "epoch": 0.4863670130322734,
      "grad_norm": 0.6432267710735352,
      "learning_rate": 5.462733694892452e-06,
      "loss": 0.502,
      "step": 6629
    },
    {
      "epoch": 0.48644038262241257,
      "grad_norm": 1.5082216691588057,
      "learning_rate": 5.461550584254365e-06,
      "loss": 0.8688,
      "step": 6630
    },
    {
      "epoch": 0.4865137522125517,
      "grad_norm": 1.4915538831794661,
      "learning_rate": 5.460367447551373e-06,
      "loss": 0.7823,
      "step": 6631
    },
    {
      "epoch": 0.4865871218026908,
      "grad_norm": 1.8998894200810685,
      "learning_rate": 5.459184284850291e-06,
      "loss": 0.8043,
      "step": 6632
    },
    {
      "epoch": 0.4865871218026908,
      "lm_loss": 1.4911,
      "step": 6632,
      "vm_loss": 0.0534
    },
    {
      "epoch": 0.4865871218026908,
      "lm_loss": 0.3861,
      "step": 6632,
      "vm_loss": 0.2449
    },
    {
      "epoch": 0.4865871218026908,
      "lm_loss": 0.6485,
      "step": 6632,
      "vm_loss": 0.1625
    },
    {
      "epoch": 0.4865871218026908,
      "lm_loss": 0.7809,
      "step": 6632,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.4865871218026908,
      "lm_loss": 0.6331,
      "step": 6632,
      "vm_loss": 0.1677
    },
    {
      "epoch": 0.4865871218026908,
      "lm_loss": 1.3365,
      "step": 6632,
      "vm_loss": 0.1313
    },
    {
      "epoch": 0.4865871218026908,
      "lm_loss": 0.8192,
      "step": 6632,
      "vm_loss": 0.0819
    },
    {
      "epoch": 0.4865871218026908,
      "lm_loss": 0.0548,
      "step": 6632,
      "vm_loss": 0.1138
    },
    {
      "epoch": 0.48666049139282996,
      "grad_norm": 1.4905632586213522,
      "learning_rate": 5.4580010962179355e-06,
      "loss": 0.8296,
      "step": 6633
    },
    {
      "epoch": 0.48673386098296906,
      "grad_norm": 0.6615375770996751,
      "learning_rate": 5.456817881721125e-06,
      "loss": 0.5021,
      "step": 6634
    },
    {
      "epoch": 0.4868072305731082,
      "grad_norm": 1.580425991182132,
      "learning_rate": 5.455634641426675e-06,
      "loss": 0.7677,
      "step": 6635
    },
    {
      "epoch": 0.48688060016324736,
      "grad_norm": 1.4114529917555445,
      "learning_rate": 5.454451375401411e-06,
      "loss": 0.7441,
      "step": 6636
    },
    {
      "epoch": 0.48695396975338645,
      "grad_norm": 0.5934835046626904,
      "learning_rate": 5.453268083712151e-06,
      "loss": 0.521,
      "step": 6637
    },
    {
      "epoch": 0.4870273393435256,
      "grad_norm": 1.7226413840765522,
      "learning_rate": 5.4520847664257195e-06,
      "loss": 0.7195,
      "step": 6638
    },
    {
      "epoch": 0.4871007089336647,
      "grad_norm": 1.4626805621053875,
      "learning_rate": 5.450901423608942e-06,
      "loss": 0.7877,
      "step": 6639
    },
    {
      "epoch": 0.48717407852380384,
      "grad_norm": 1.6327408959327867,
      "learning_rate": 5.449718055328643e-06,
      "loss": 0.8557,
      "step": 6640
    },
    {
      "epoch": 0.48717407852380384,
      "lm_loss": 0.2185,
      "step": 6640,
      "vm_loss": 0.0265
    },
    {
      "epoch": 0.48717407852380384,
      "lm_loss": 0.2552,
      "step": 6640,
      "vm_loss": 0.0387
    },
    {
      "epoch": 0.48717407852380384,
      "lm_loss": 0.84,
      "step": 6640,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.48717407852380384,
      "lm_loss": 0.8939,
      "step": 6640,
      "vm_loss": 0.1425
    },
    {
      "epoch": 0.48717407852380384,
      "lm_loss": 0.0513,
      "step": 6640,
      "vm_loss": 0.1143
    },
    {
      "epoch": 0.48717407852380384,
      "lm_loss": 0.5983,
      "step": 6640,
      "vm_loss": 0.2094
    },
    {
      "epoch": 0.48717407852380384,
      "lm_loss": 1.4602,
      "step": 6640,
      "vm_loss": 0.1023
    },
    {
      "epoch": 0.48717407852380384,
      "lm_loss": 0.514,
      "step": 6640,
      "vm_loss": 0.1612
    },
    {
      "epoch": 0.487247448113943,
      "grad_norm": 1.6519714420848903,
      "learning_rate": 5.4485346616516545e-06,
      "loss": 0.7569,
      "step": 6641
    },
    {
      "epoch": 0.4873208177040821,
      "grad_norm": 0.5951261668803771,
      "learning_rate": 5.447351242644802e-06,
      "loss": 0.4963,
      "step": 6642
    },
    {
      "epoch": 0.48739418729422124,
      "grad_norm": 1.7385988872823346,
      "learning_rate": 5.446167798374916e-06,
      "loss": 0.7407,
      "step": 6643
    },
    {
      "epoch": 0.48746755688436033,
      "grad_norm": 1.4875608731322076,
      "learning_rate": 5.444984328908831e-06,
      "loss": 0.7637,
      "step": 6644
    },
    {
      "epoch": 0.4875409264744995,
      "grad_norm": 0.5861131448723075,
      "learning_rate": 5.443800834313378e-06,
      "loss": 0.4988,
      "step": 6645
    },
    {
      "epoch": 0.48761429606463863,
      "grad_norm": 0.5772132217932248,
      "learning_rate": 5.442617314655392e-06,
      "loss": 0.5097,
      "step": 6646
    },
    {
      "epoch": 0.4876876656547777,
      "grad_norm": 1.48235998197757,
      "learning_rate": 5.4414337700017114e-06,
      "loss": 0.6717,
      "step": 6647
    },
    {
      "epoch": 0.4877610352449169,
      "grad_norm": 1.5491528753510972,
      "learning_rate": 5.440250200419173e-06,
      "loss": 0.8015,
      "step": 6648
    },
    {
      "epoch": 0.4877610352449169,
      "lm_loss": 1.5769,
      "step": 6648,
      "vm_loss": 0.062
    },
    {
      "epoch": 0.4877610352449169,
      "lm_loss": 0.272,
      "step": 6648,
      "vm_loss": 0.1052
    },
    {
      "epoch": 0.4877610352449169,
      "lm_loss": 0.1564,
      "step": 6648,
      "vm_loss": 0.1165
    },
    {
      "epoch": 0.4877610352449169,
      "lm_loss": 0.4265,
      "step": 6648,
      "vm_loss": 0.0446
    },
    {
      "epoch": 0.4877610352449169,
      "lm_loss": 0.8317,
      "step": 6648,
      "vm_loss": 0.1416
    },
    {
      "epoch": 0.4877610352449169,
      "lm_loss": 0.4904,
      "step": 6648,
      "vm_loss": 0.1741
    },
    {
      "epoch": 0.4877610352449169,
      "lm_loss": 0.2572,
      "step": 6648,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.4877610352449169,
      "lm_loss": 0.0152,
      "step": 6648,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.48783440483505597,
      "grad_norm": 1.42946419829725,
      "learning_rate": 5.439066605974615e-06,
      "loss": 0.6498,
      "step": 6649
    },
    {
      "epoch": 0.4879077744251951,
      "grad_norm": 1.6407450102463488,
      "learning_rate": 5.437882986734878e-06,
      "loss": 0.7391,
      "step": 6650
    },
    {
      "epoch": 0.48798114401533427,
      "grad_norm": 1.543038732514842,
      "learning_rate": 5.436699342766803e-06,
      "loss": 0.7152,
      "step": 6651
    },
    {
      "epoch": 0.48805451360547336,
      "grad_norm": 1.5425621380337877,
      "learning_rate": 5.435515674137237e-06,
      "loss": 0.8466,
      "step": 6652
    },
    {
      "epoch": 0.4881278831956125,
      "grad_norm": 1.434733644377049,
      "learning_rate": 5.434331980913022e-06,
      "loss": 0.7259,
      "step": 6653
    },
    {
      "epoch": 0.4882012527857516,
      "grad_norm": 0.6514668497802978,
      "learning_rate": 5.433148263161003e-06,
      "loss": 0.4765,
      "step": 6654
    },
    {
      "epoch": 0.48827462237589075,
      "grad_norm": 1.4083787156644827,
      "learning_rate": 5.431964520948031e-06,
      "loss": 0.7939,
      "step": 6655
    },
    {
      "epoch": 0.4883479919660299,
      "grad_norm": 0.5957914186672794,
      "learning_rate": 5.430780754340952e-06,
      "loss": 0.5118,
      "step": 6656
    },
    {
      "epoch": 0.4883479919660299,
      "lm_loss": 1.5099,
      "step": 6656,
      "vm_loss": 0.0655
    },
    {
      "epoch": 0.4883479919660299,
      "lm_loss": 0.6782,
      "step": 6656,
      "vm_loss": 0.0476
    },
    {
      "epoch": 0.4883479919660299,
      "lm_loss": 0.9811,
      "step": 6656,
      "vm_loss": 0.1713
    },
    {
      "epoch": 0.4883479919660299,
      "lm_loss": 1.0988,
      "step": 6656,
      "vm_loss": 0.0682
    },
    {
      "epoch": 0.4883479919660299,
      "lm_loss": 0.6907,
      "step": 6656,
      "vm_loss": 0.1509
    },
    {
      "epoch": 0.4883479919660299,
      "lm_loss": 0.9164,
      "step": 6656,
      "vm_loss": 0.0513
    },
    {
      "epoch": 0.4883479919660299,
      "lm_loss": 0.1714,
      "step": 6656,
      "vm_loss": 0.1033
    },
    {
      "epoch": 0.4883479919660299,
      "lm_loss": 0.831,
      "step": 6656,
      "vm_loss": 0.1585
    },
    {
      "epoch": 0.488421361556169,
      "grad_norm": 1.370923770875658,
      "learning_rate": 5.429596963406616e-06,
      "loss": 0.7787,
      "step": 6657
    },
    {
      "epoch": 0.48849473114630815,
      "grad_norm": 1.493298034516456,
      "learning_rate": 5.428413148211876e-06,
      "loss": 0.6612,
      "step": 6658
    },
    {
      "epoch": 0.48856810073644724,
      "grad_norm": 1.5412373727650637,
      "learning_rate": 5.427229308823585e-06,
      "loss": 0.7727,
      "step": 6659
    },
    {
      "epoch": 0.4886414703265864,
      "grad_norm": 1.6764073156230177,
      "learning_rate": 5.426045445308598e-06,
      "loss": 0.8045,
      "step": 6660
    },
    {
      "epoch": 0.48871483991672554,
      "grad_norm": 1.4716815200911544,
      "learning_rate": 5.424861557733768e-06,
      "loss": 0.8139,
      "step": 6661
    },
    {
      "epoch": 0.48878820950686463,
      "grad_norm": 0.6134206302626308,
      "learning_rate": 5.423677646165953e-06,
      "loss": 0.5168,
      "step": 6662
    },
    {
      "epoch": 0.4888615790970038,
      "grad_norm": 1.5948312814890975,
      "learning_rate": 5.4224937106720135e-06,
      "loss": 0.7712,
      "step": 6663
    },
    {
      "epoch": 0.4889349486871429,
      "grad_norm": 1.6458790240185235,
      "learning_rate": 5.421309751318808e-06,
      "loss": 0.8054,
      "step": 6664
    },
    {
      "epoch": 0.4889349486871429,
      "lm_loss": 0.6228,
      "step": 6664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4889349486871429,
      "lm_loss": 0.547,
      "step": 6664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4889349486871429,
      "lm_loss": 0.358,
      "step": 6664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4889349486871429,
      "lm_loss": 0.3732,
      "step": 6664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4889349486871429,
      "lm_loss": 0.3402,
      "step": 6664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4889349486871429,
      "lm_loss": 0.617,
      "step": 6664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4889349486871429,
      "lm_loss": 0.4556,
      "step": 6664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4889349486871429,
      "lm_loss": 0.4201,
      "step": 6664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48900831827728203,
      "grad_norm": 0.6046815932260715,
      "learning_rate": 5.4201257681731965e-06,
      "loss": 0.5153,
      "step": 6665
    },
    {
      "epoch": 0.4890816878674212,
      "grad_norm": 1.5345040104165442,
      "learning_rate": 5.4189417613020436e-06,
      "loss": 0.7555,
      "step": 6666
    },
    {
      "epoch": 0.48915505745756027,
      "grad_norm": 1.655776451057564,
      "learning_rate": 5.417757730772211e-06,
      "loss": 0.7643,
      "step": 6667
    },
    {
      "epoch": 0.4892284270476994,
      "grad_norm": 0.5821783790487319,
      "learning_rate": 5.416573676650565e-06,
      "loss": 0.4926,
      "step": 6668
    },
    {
      "epoch": 0.4893017966378385,
      "grad_norm": 0.5700462554852712,
      "learning_rate": 5.415389599003972e-06,
      "loss": 0.5056,
      "step": 6669
    },
    {
      "epoch": 0.48937516622797766,
      "grad_norm": 0.5601993090308836,
      "learning_rate": 5.414205497899299e-06,
      "loss": 0.5043,
      "step": 6670
    },
    {
      "epoch": 0.48944853581811676,
      "grad_norm": 1.6295884054297918,
      "learning_rate": 5.413021373403416e-06,
      "loss": 0.7146,
      "step": 6671
    },
    {
      "epoch": 0.4895219054082559,
      "grad_norm": 0.5832224083986843,
      "learning_rate": 5.411837225583195e-06,
      "loss": 0.5044,
      "step": 6672
    },
    {
      "epoch": 0.4895219054082559,
      "lm_loss": 0.5322,
      "step": 6672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4895219054082559,
      "lm_loss": 0.4261,
      "step": 6672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4895219054082559,
      "lm_loss": 0.3043,
      "step": 6672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4895219054082559,
      "lm_loss": 0.4491,
      "step": 6672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4895219054082559,
      "lm_loss": 0.5101,
      "step": 6672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4895219054082559,
      "lm_loss": 0.4068,
      "step": 6672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4895219054082559,
      "lm_loss": 0.5881,
      "step": 6672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4895219054082559,
      "lm_loss": 0.5312,
      "step": 6672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.48959527499839506,
      "grad_norm": 0.5626717243110818,
      "learning_rate": 5.410653054505503e-06,
      "loss": 0.4932,
      "step": 6673
    },
    {
      "epoch": 0.48966864458853415,
      "grad_norm": 1.3345559053753768,
      "learning_rate": 5.409468860237219e-06,
      "loss": 0.8083,
      "step": 6674
    },
    {
      "epoch": 0.4897420141786733,
      "grad_norm": 0.5540963537530661,
      "learning_rate": 5.408284642845213e-06,
      "loss": 0.4857,
      "step": 6675
    },
    {
      "epoch": 0.4898153837688124,
      "grad_norm": 1.617843531680942,
      "learning_rate": 5.407100402396361e-06,
      "loss": 0.7648,
      "step": 6676
    },
    {
      "epoch": 0.48988875335895155,
      "grad_norm": 1.6267995675924578,
      "learning_rate": 5.405916138957543e-06,
      "loss": 0.6779,
      "step": 6677
    },
    {
      "epoch": 0.4899621229490907,
      "grad_norm": 1.5216056288693103,
      "learning_rate": 5.404731852595634e-06,
      "loss": 0.8605,
      "step": 6678
    },
    {
      "epoch": 0.4900354925392298,
      "grad_norm": 1.7415642037784471,
      "learning_rate": 5.403547543377516e-06,
      "loss": 0.8409,
      "step": 6679
    },
    {
      "epoch": 0.49010886212936894,
      "grad_norm": 1.366977705244464,
      "learning_rate": 5.402363211370069e-06,
      "loss": 0.7779,
      "step": 6680
    },
    {
      "epoch": 0.49010886212936894,
      "lm_loss": 1.2554,
      "step": 6680,
      "vm_loss": 0.0558
    },
    {
      "epoch": 0.49010886212936894,
      "lm_loss": 1.2425,
      "step": 6680,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.49010886212936894,
      "lm_loss": 0.4332,
      "step": 6680,
      "vm_loss": 0.0459
    },
    {
      "epoch": 0.49010886212936894,
      "lm_loss": 0.1384,
      "step": 6680,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.49010886212936894,
      "lm_loss": 0.4068,
      "step": 6680,
      "vm_loss": 0.1616
    },
    {
      "epoch": 0.49010886212936894,
      "lm_loss": 0.9822,
      "step": 6680,
      "vm_loss": 0.1261
    },
    {
      "epoch": 0.49010886212936894,
      "lm_loss": 0.0769,
      "step": 6680,
      "vm_loss": 0.1935
    },
    {
      "epoch": 0.49010886212936894,
      "lm_loss": 0.5166,
      "step": 6680,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.49018223171950803,
      "grad_norm": 1.434538195363218,
      "learning_rate": 5.401178856640174e-06,
      "loss": 0.8112,
      "step": 6681
    },
    {
      "epoch": 0.4902556013096472,
      "grad_norm": 0.6492342095397929,
      "learning_rate": 5.399994479254718e-06,
      "loss": 0.5092,
      "step": 6682
    },
    {
      "epoch": 0.49032897089978633,
      "grad_norm": 0.6009289704919963,
      "learning_rate": 5.398810079280579e-06,
      "loss": 0.4972,
      "step": 6683
    },
    {
      "epoch": 0.4904023404899254,
      "grad_norm": 1.638716054222247,
      "learning_rate": 5.397625656784651e-06,
      "loss": 0.7507,
      "step": 6684
    },
    {
      "epoch": 0.4904757100800646,
      "grad_norm": 1.573906449641084,
      "learning_rate": 5.396441211833816e-06,
      "loss": 0.7556,
      "step": 6685
    },
    {
      "epoch": 0.49054907967020367,
      "grad_norm": 1.4921181887025008,
      "learning_rate": 5.395256744494963e-06,
      "loss": 0.7716,
      "step": 6686
    },
    {
      "epoch": 0.4906224492603428,
      "grad_norm": 1.6446784648954458,
      "learning_rate": 5.394072254834984e-06,
      "loss": 0.7762,
      "step": 6687
    },
    {
      "epoch": 0.49069581885048197,
      "grad_norm": 0.570686245098428,
      "learning_rate": 5.392887742920768e-06,
      "loss": 0.4864,
      "step": 6688
    },
    {
      "epoch": 0.49069581885048197,
      "lm_loss": 1.2888,
      "step": 6688,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.49069581885048197,
      "lm_loss": 0.1917,
      "step": 6688,
      "vm_loss": 0.1074
    },
    {
      "epoch": 0.49069581885048197,
      "lm_loss": 0.7556,
      "step": 6688,
      "vm_loss": 0.1021
    },
    {
      "epoch": 0.49069581885048197,
      "lm_loss": 0.7353,
      "step": 6688,
      "vm_loss": 0.0819
    },
    {
      "epoch": 0.49069581885048197,
      "lm_loss": 0.6138,
      "step": 6688,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.49069581885048197,
      "lm_loss": 0.5091,
      "step": 6688,
      "vm_loss": 0.1141
    },
    {
      "epoch": 0.49069581885048197,
      "lm_loss": 0.5609,
      "step": 6688,
      "vm_loss": 0.0872
    },
    {
      "epoch": 0.49069581885048197,
      "lm_loss": 0.1737,
      "step": 6688,
      "vm_loss": 0.1023
    },
    {
      "epoch": 0.49076918844062106,
      "grad_norm": 1.4629694941487716,
      "learning_rate": 5.391703208819209e-06,
      "loss": 0.7172,
      "step": 6689
    },
    {
      "epoch": 0.4908425580307602,
      "grad_norm": 1.7315546737635363,
      "learning_rate": 5.3905186525971995e-06,
      "loss": 0.8069,
      "step": 6690
    },
    {
      "epoch": 0.4909159276208993,
      "grad_norm": 0.6252322657251858,
      "learning_rate": 5.389334074321634e-06,
      "loss": 0.5055,
      "step": 6691
    },
    {
      "epoch": 0.49098929721103846,
      "grad_norm": 1.4546566625643547,
      "learning_rate": 5.38814947405941e-06,
      "loss": 0.7574,
      "step": 6692
    },
    {
      "epoch": 0.4910626668011776,
      "grad_norm": 0.5864670554799416,
      "learning_rate": 5.386964851877424e-06,
      "loss": 0.5168,
      "step": 6693
    },
    {
      "epoch": 0.4911360363913167,
      "grad_norm": 1.4961923458361137,
      "learning_rate": 5.385780207842573e-06,
      "loss": 0.7801,
      "step": 6694
    },
    {
      "epoch": 0.49120940598145585,
      "grad_norm": 1.4980993053864564,
      "learning_rate": 5.38459554202176e-06,
      "loss": 0.8327,
      "step": 6695
    },
    {
      "epoch": 0.49128277557159494,
      "grad_norm": 1.3313939301180187,
      "learning_rate": 5.3834108544818844e-06,
      "loss": 0.7634,
      "step": 6696
    },
    {
      "epoch": 0.49128277557159494,
      "lm_loss": 1.375,
      "step": 6696,
      "vm_loss": 0.0357
    },
    {
      "epoch": 0.49128277557159494,
      "lm_loss": 0.0811,
      "step": 6696,
      "vm_loss": 0.0686
    },
    {
      "epoch": 0.49128277557159494,
      "lm_loss": 0.9697,
      "step": 6696,
      "vm_loss": 0.328
    },
    {
      "epoch": 0.49128277557159494,
      "lm_loss": 0.6174,
      "step": 6696,
      "vm_loss": 0.2397
    },
    {
      "epoch": 0.49128277557159494,
      "lm_loss": 0.745,
      "step": 6696,
      "vm_loss": 0.2051
    },
    {
      "epoch": 0.49128277557159494,
      "lm_loss": 0.4128,
      "step": 6696,
      "vm_loss": 0.1
    },
    {
      "epoch": 0.49128277557159494,
      "lm_loss": 0.6507,
      "step": 6696,
      "vm_loss": 0.1621
    },
    {
      "epoch": 0.49128277557159494,
      "lm_loss": 0.3962,
      "step": 6696,
      "vm_loss": 0.1104
    },
    {
      "epoch": 0.4913561451617341,
      "grad_norm": 1.2996275676345388,
      "learning_rate": 5.382226145289847e-06,
      "loss": 0.7708,
      "step": 6697
    },
    {
      "epoch": 0.49142951475187324,
      "grad_norm": 1.3199737994074414,
      "learning_rate": 5.3810414145125535e-06,
      "loss": 0.7479,
      "step": 6698
    },
    {
      "epoch": 0.49150288434201234,
      "grad_norm": 1.4215327885802325,
      "learning_rate": 5.379856662216907e-06,
      "loss": 0.7382,
      "step": 6699
    },
    {
      "epoch": 0.4915762539321515,
      "grad_norm": 1.5474998566835199,
      "learning_rate": 5.378671888469816e-06,
      "loss": 0.6977,
      "step": 6700
    },
    {
      "epoch": 0.4916496235222906,
      "grad_norm": 1.7228767728421324,
      "learning_rate": 5.377487093338184e-06,
      "loss": 0.8355,
      "step": 6701
    },
    {
      "epoch": 0.49172299311242973,
      "grad_norm": 0.6503495018942067,
      "learning_rate": 5.37630227688892e-06,
      "loss": 0.5064,
      "step": 6702
    },
    {
      "epoch": 0.4917963627025689,
      "grad_norm": 0.617318648547202,
      "learning_rate": 5.375117439188936e-06,
      "loss": 0.5204,
      "step": 6703
    },
    {
      "epoch": 0.491869732292708,
      "grad_norm": 0.5870029070643638,
      "learning_rate": 5.373932580305142e-06,
      "loss": 0.4928,
      "step": 6704
    },
    {
      "epoch": 0.491869732292708,
      "lm_loss": 1.495,
      "step": 6704,
      "vm_loss": 0.088
    },
    {
      "epoch": 0.491869732292708,
      "lm_loss": 0.6852,
      "step": 6704,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.491869732292708,
      "lm_loss": 0.1702,
      "step": 6704,
      "vm_loss": 0.1231
    },
    {
      "epoch": 0.491869732292708,
      "lm_loss": 0.2978,
      "step": 6704,
      "vm_loss": 0.1898
    },
    {
      "epoch": 0.491869732292708,
      "lm_loss": 0.7581,
      "step": 6704,
      "vm_loss": 0.1239
    },
    {
      "epoch": 0.491869732292708,
      "lm_loss": 0.2579,
      "step": 6704,
      "vm_loss": 0.1415
    },
    {
      "epoch": 0.491869732292708,
      "lm_loss": 0.4794,
      "step": 6704,
      "vm_loss": 0.1204
    },
    {
      "epoch": 0.491869732292708,
      "lm_loss": 0.6653,
      "step": 6704,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.4919431018828471,
      "grad_norm": 1.5862266934757558,
      "learning_rate": 5.372747700304447e-06,
      "loss": 0.7359,
      "step": 6705
    },
    {
      "epoch": 0.4920164714729862,
      "grad_norm": 1.7172966661425784,
      "learning_rate": 5.3715627992537675e-06,
      "loss": 0.7754,
      "step": 6706
    },
    {
      "epoch": 0.49208984106312537,
      "grad_norm": 0.6454115599026475,
      "learning_rate": 5.370377877220016e-06,
      "loss": 0.5172,
      "step": 6707
    },
    {
      "epoch": 0.4921632106532645,
      "grad_norm": 1.5493079330880464,
      "learning_rate": 5.36919293427011e-06,
      "loss": 0.8407,
      "step": 6708
    },
    {
      "epoch": 0.4922365802434036,
      "grad_norm": 0.6783899389017495,
      "learning_rate": 5.368007970470964e-06,
      "loss": 0.5105,
      "step": 6709
    },
    {
      "epoch": 0.49230994983354276,
      "grad_norm": 1.537523435563137,
      "learning_rate": 5.366822985889496e-06,
      "loss": 0.8045,
      "step": 6710
    },
    {
      "epoch": 0.49238331942368185,
      "grad_norm": 1.4589481048520416,
      "learning_rate": 5.365637980592626e-06,
      "loss": 0.7479,
      "step": 6711
    },
    {
      "epoch": 0.492456689013821,
      "grad_norm": 0.6046076619331908,
      "learning_rate": 5.364452954647275e-06,
      "loss": 0.4936,
      "step": 6712
    },
    {
      "epoch": 0.492456689013821,
      "lm_loss": 1.0884,
      "step": 6712,
      "vm_loss": 0.1364
    },
    {
      "epoch": 0.492456689013821,
      "lm_loss": 1.4835,
      "step": 6712,
      "vm_loss": 0.0646
    },
    {
      "epoch": 0.492456689013821,
      "lm_loss": 0.7488,
      "step": 6712,
      "vm_loss": 0.0546
    },
    {
      "epoch": 0.492456689013821,
      "lm_loss": 0.6878,
      "step": 6712,
      "vm_loss": 0.1219
    },
    {
      "epoch": 0.492456689013821,
      "lm_loss": 1.5289,
      "step": 6712,
      "vm_loss": 0.1437
    },
    {
      "epoch": 0.492456689013821,
      "lm_loss": 0.3508,
      "step": 6712,
      "vm_loss": 0.0996
    },
    {
      "epoch": 0.492456689013821,
      "lm_loss": 0.7261,
      "step": 6712,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.492456689013821,
      "lm_loss": 0.5761,
      "step": 6712,
      "vm_loss": 0.094
    },
    {
      "epoch": 0.4925300586039601,
      "grad_norm": 1.5562466547873095,
      "learning_rate": 5.363267908120361e-06,
      "loss": 0.8354,
      "step": 6713
    },
    {
      "epoch": 0.49260342819409925,
      "grad_norm": 1.6100670997982465,
      "learning_rate": 5.362082841078809e-06,
      "loss": 0.7676,
      "step": 6714
    },
    {
      "epoch": 0.4926767977842384,
      "grad_norm": 0.5882748563089776,
      "learning_rate": 5.360897753589542e-06,
      "loss": 0.4991,
      "step": 6715
    },
    {
      "epoch": 0.4927501673743775,
      "grad_norm": 1.5499368198378034,
      "learning_rate": 5.359712645719486e-06,
      "loss": 0.8346,
      "step": 6716
    },
    {
      "epoch": 0.49282353696451664,
      "grad_norm": 1.6639412613239866,
      "learning_rate": 5.3585275175355664e-06,
      "loss": 0.7817,
      "step": 6717
    },
    {
      "epoch": 0.49289690655465573,
      "grad_norm": 1.6161667200590453,
      "learning_rate": 5.357342369104709e-06,
      "loss": 0.7651,
      "step": 6718
    },
    {
      "epoch": 0.4929702761447949,
      "grad_norm": 0.6662515926620959,
      "learning_rate": 5.356157200493843e-06,
      "loss": 0.5163,
      "step": 6719
    },
    {
      "epoch": 0.49304364573493403,
      "grad_norm": 1.4995545652365823,
      "learning_rate": 5.3549720117699e-06,
      "loss": 0.729,
      "step": 6720
    },
    {
      "epoch": 0.49304364573493403,
      "lm_loss": 1.2822,
      "step": 6720,
      "vm_loss": 0.0762
    },
    {
      "epoch": 0.49304364573493403,
      "lm_loss": 2.3262,
      "step": 6720,
      "vm_loss": 0.1593
    },
    {
      "epoch": 0.49304364573493403,
      "lm_loss": 0.4209,
      "step": 6720,
      "vm_loss": 0.095
    },
    {
      "epoch": 0.49304364573493403,
      "lm_loss": 0.0881,
      "step": 6720,
      "vm_loss": 0.1585
    },
    {
      "epoch": 0.49304364573493403,
      "lm_loss": 0.8452,
      "step": 6720,
      "vm_loss": 0.1738
    },
    {
      "epoch": 0.49304364573493403,
      "lm_loss": 1.3805,
      "step": 6720,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.49304364573493403,
      "lm_loss": 0.6902,
      "step": 6720,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.49304364573493403,
      "lm_loss": 1.2137,
      "step": 6720,
      "vm_loss": 0.097
    },
    {
      "epoch": 0.4931170153250731,
      "grad_norm": 1.5161679064114728,
      "learning_rate": 5.353786802999805e-06,
      "loss": 0.8721,
      "step": 6721
    },
    {
      "epoch": 0.4931903849152123,
      "grad_norm": 1.456206289871433,
      "learning_rate": 5.352601574250496e-06,
      "loss": 0.7886,
      "step": 6722
    },
    {
      "epoch": 0.49326375450535137,
      "grad_norm": 1.639395679143153,
      "learning_rate": 5.351416325588902e-06,
      "loss": 0.7054,
      "step": 6723
    },
    {
      "epoch": 0.4933371240954905,
      "grad_norm": 1.5582792695260155,
      "learning_rate": 5.350231057081957e-06,
      "loss": 0.7268,
      "step": 6724
    },
    {
      "epoch": 0.49341049368562967,
      "grad_norm": 1.566263410742173,
      "learning_rate": 5.349045768796598e-06,
      "loss": 0.6743,
      "step": 6725
    },
    {
      "epoch": 0.49348386327576876,
      "grad_norm": 1.7484814502231485,
      "learning_rate": 5.3478604607997586e-06,
      "loss": 0.8035,
      "step": 6726
    },
    {
      "epoch": 0.4935572328659079,
      "grad_norm": 1.5202940568038985,
      "learning_rate": 5.3466751331583785e-06,
      "loss": 0.8384,
      "step": 6727
    },
    {
      "epoch": 0.493630602456047,
      "grad_norm": 0.5938129901561989,
      "learning_rate": 5.345489785939396e-06,
      "loss": 0.4842,
      "step": 6728
    },
    {
      "epoch": 0.493630602456047,
      "lm_loss": 1.533,
      "step": 6728,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.493630602456047,
      "lm_loss": 0.6602,
      "step": 6728,
      "vm_loss": 0.1137
    },
    {
      "epoch": 0.493630602456047,
      "lm_loss": 0.512,
      "step": 6728,
      "vm_loss": 0.1099
    },
    {
      "epoch": 0.493630602456047,
      "lm_loss": 0.4922,
      "step": 6728,
      "vm_loss": 0.1217
    },
    {
      "epoch": 0.493630602456047,
      "lm_loss": 0.4586,
      "step": 6728,
      "vm_loss": 0.0509
    },
    {
      "epoch": 0.493630602456047,
      "lm_loss": 0.4307,
      "step": 6728,
      "vm_loss": 0.0727
    },
    {
      "epoch": 0.493630602456047,
      "lm_loss": 1.3301,
      "step": 6728,
      "vm_loss": 0.2105
    },
    {
      "epoch": 0.493630602456047,
      "lm_loss": 0.8453,
      "step": 6728,
      "vm_loss": 0.2256
    },
    {
      "epoch": 0.49370397204618616,
      "grad_norm": 1.8052344687214998,
      "learning_rate": 5.344304419209748e-06,
      "loss": 0.7625,
      "step": 6729
    },
    {
      "epoch": 0.4937773416363253,
      "grad_norm": 1.8342209417307789,
      "learning_rate": 5.343119033036378e-06,
      "loss": 0.7937,
      "step": 6730
    },
    {
      "epoch": 0.4938507112264644,
      "grad_norm": 1.4895529156758072,
      "learning_rate": 5.341933627486226e-06,
      "loss": 0.782,
      "step": 6731
    },
    {
      "epoch": 0.49392408081660355,
      "grad_norm": 0.5893250906309462,
      "learning_rate": 5.340748202626236e-06,
      "loss": 0.5126,
      "step": 6732
    },
    {
      "epoch": 0.49399745040674264,
      "grad_norm": 1.796308340377831,
      "learning_rate": 5.33956275852335e-06,
      "loss": 0.8151,
      "step": 6733
    },
    {
      "epoch": 0.4940708199968818,
      "grad_norm": 0.5887366593378801,
      "learning_rate": 5.338377295244515e-06,
      "loss": 0.4861,
      "step": 6734
    },
    {
      "epoch": 0.49414418958702094,
      "grad_norm": 1.7410909348116397,
      "learning_rate": 5.337191812856675e-06,
      "loss": 0.7789,
      "step": 6735
    },
    {
      "epoch": 0.49421755917716004,
      "grad_norm": 1.796087280452216,
      "learning_rate": 5.33600631142678e-06,
      "loss": 0.7984,
      "step": 6736
    },
    {
      "epoch": 0.49421755917716004,
      "lm_loss": 1.7699,
      "step": 6736,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.49421755917716004,
      "lm_loss": 1.0009,
      "step": 6736,
      "vm_loss": 0.1147
    },
    {
      "epoch": 0.49421755917716004,
      "lm_loss": 0.9961,
      "step": 6736,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.49421755917716004,
      "lm_loss": 0.5316,
      "step": 6736,
      "vm_loss": 0.2341
    },
    {
      "epoch": 0.49421755917716004,
      "lm_loss": 0.3134,
      "step": 6736,
      "vm_loss": 0.2149
    },
    {
      "epoch": 0.49421755917716004,
      "lm_loss": 1.1382,
      "step": 6736,
      "vm_loss": 0.1297
    },
    {
      "epoch": 0.49421755917716004,
      "lm_loss": 0.2712,
      "step": 6736,
      "vm_loss": 0.1331
    },
    {
      "epoch": 0.49421755917716004,
      "lm_loss": 0.5003,
      "step": 6736,
      "vm_loss": 0.0419
    },
    {
      "epoch": 0.4942909287672992,
      "grad_norm": 1.5899620538725443,
      "learning_rate": 5.3348207910217745e-06,
      "loss": 0.8321,
      "step": 6737
    },
    {
      "epoch": 0.4943642983574383,
      "grad_norm": 1.7020358692201438,
      "learning_rate": 5.333635251708611e-06,
      "loss": 0.8748,
      "step": 6738
    },
    {
      "epoch": 0.49443766794757743,
      "grad_norm": 1.5643231077280875,
      "learning_rate": 5.332449693554239e-06,
      "loss": 0.8239,
      "step": 6739
    },
    {
      "epoch": 0.4945110375377166,
      "grad_norm": 1.5027329430113048,
      "learning_rate": 5.331264116625608e-06,
      "loss": 0.6991,
      "step": 6740
    },
    {
      "epoch": 0.4945844071278557,
      "grad_norm": 0.6313523028134835,
      "learning_rate": 5.330078520989673e-06,
      "loss": 0.5252,
      "step": 6741
    },
    {
      "epoch": 0.4946577767179948,
      "grad_norm": 0.600191514257226,
      "learning_rate": 5.328892906713387e-06,
      "loss": 0.4689,
      "step": 6742
    },
    {
      "epoch": 0.4947311463081339,
      "grad_norm": 0.5822686796329174,
      "learning_rate": 5.327707273863703e-06,
      "loss": 0.5043,
      "step": 6743
    },
    {
      "epoch": 0.49480451589827307,
      "grad_norm": 1.6567930660734864,
      "learning_rate": 5.326521622507579e-06,
      "loss": 0.7841,
      "step": 6744
    },
    {
      "epoch": 0.49480451589827307,
      "lm_loss": 0.652,
      "step": 6744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49480451589827307,
      "lm_loss": 0.4608,
      "step": 6744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49480451589827307,
      "lm_loss": 0.5014,
      "step": 6744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49480451589827307,
      "lm_loss": 0.4404,
      "step": 6744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49480451589827307,
      "lm_loss": 0.3695,
      "step": 6744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49480451589827307,
      "lm_loss": 0.5198,
      "step": 6744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49480451589827307,
      "lm_loss": 0.5376,
      "step": 6744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49480451589827307,
      "lm_loss": 0.5758,
      "step": 6744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4948778854884122,
      "grad_norm": 0.5735769746913484,
      "learning_rate": 5.325335952711969e-06,
      "loss": 0.489,
      "step": 6745
    },
    {
      "epoch": 0.4949512550785513,
      "grad_norm": 1.812221408886913,
      "learning_rate": 5.324150264543832e-06,
      "loss": 0.8241,
      "step": 6746
    },
    {
      "epoch": 0.49502462466869046,
      "grad_norm": 1.4486601936993662,
      "learning_rate": 5.3229645580701285e-06,
      "loss": 0.7887,
      "step": 6747
    },
    {
      "epoch": 0.49509799425882955,
      "grad_norm": 0.60158936299387,
      "learning_rate": 5.321778833357815e-06,
      "loss": 0.4959,
      "step": 6748
    },
    {
      "epoch": 0.4951713638489687,
      "grad_norm": 1.625420824989434,
      "learning_rate": 5.3205930904738544e-06,
      "loss": 0.7534,
      "step": 6749
    },
    {
      "epoch": 0.4952447334391078,
      "grad_norm": 1.5669270678316167,
      "learning_rate": 5.319407329485208e-06,
      "loss": 0.8148,
      "step": 6750
    },
    {
      "epoch": 0.49531810302924695,
      "grad_norm": 1.7064866534260132,
      "learning_rate": 5.318221550458838e-06,
      "loss": 0.8576,
      "step": 6751
    },
    {
      "epoch": 0.4953914726193861,
      "grad_norm": 1.631383999483615,
      "learning_rate": 5.31703575346171e-06,
      "loss": 0.8228,
      "step": 6752
    },
    {
      "epoch": 0.4953914726193861,
      "lm_loss": 1.5146,
      "step": 6752,
      "vm_loss": 0.0314
    },
    {
      "epoch": 0.4953914726193861,
      "lm_loss": 0.4923,
      "step": 6752,
      "vm_loss": 0.0907
    },
    {
      "epoch": 0.4953914726193861,
      "lm_loss": 0.5995,
      "step": 6752,
      "vm_loss": 0.076
    },
    {
      "epoch": 0.4953914726193861,
      "lm_loss": 0.0978,
      "step": 6752,
      "vm_loss": 0.1486
    },
    {
      "epoch": 0.4953914726193861,
      "lm_loss": 0.1844,
      "step": 6752,
      "vm_loss": 0.139
    },
    {
      "epoch": 0.4953914726193861,
      "lm_loss": 1.3684,
      "step": 6752,
      "vm_loss": 0.1885
    },
    {
      "epoch": 0.4953914726193861,
      "lm_loss": 0.7266,
      "step": 6752,
      "vm_loss": 0.0713
    },
    {
      "epoch": 0.4953914726193861,
      "lm_loss": 0.1423,
      "step": 6752,
      "vm_loss": 0.0639
    },
    {
      "epoch": 0.4954648422095252,
      "grad_norm": 1.6012154136632777,
      "learning_rate": 5.315849938560787e-06,
      "loss": 0.6972,
      "step": 6753
    },
    {
      "epoch": 0.49553821179966434,
      "grad_norm": 0.7120173774578378,
      "learning_rate": 5.314664105823036e-06,
      "loss": 0.5151,
      "step": 6754
    },
    {
      "epoch": 0.49561158138980344,
      "grad_norm": 1.5465135990687628,
      "learning_rate": 5.313478255315425e-06,
      "loss": 0.7673,
      "step": 6755
    },
    {
      "epoch": 0.4956849509799426,
      "grad_norm": 1.6659359843936021,
      "learning_rate": 5.312292387104919e-06,
      "loss": 0.7844,
      "step": 6756
    },
    {
      "epoch": 0.49575832057008173,
      "grad_norm": 0.6311753649608747,
      "learning_rate": 5.31110650125849e-06,
      "loss": 0.4653,
      "step": 6757
    },
    {
      "epoch": 0.49583169016022083,
      "grad_norm": 0.6188285716488288,
      "learning_rate": 5.309920597843106e-06,
      "loss": 0.5199,
      "step": 6758
    },
    {
      "epoch": 0.49590505975036,
      "grad_norm": 0.5781716494795237,
      "learning_rate": 5.308734676925739e-06,
      "loss": 0.5066,
      "step": 6759
    },
    {
      "epoch": 0.49597842934049907,
      "grad_norm": 1.4114216514518123,
      "learning_rate": 5.307548738573359e-06,
      "loss": 0.7128,
      "step": 6760
    },
    {
      "epoch": 0.49597842934049907,
      "lm_loss": 0.3873,
      "step": 6760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49597842934049907,
      "lm_loss": 0.4344,
      "step": 6760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49597842934049907,
      "lm_loss": 0.5858,
      "step": 6760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49597842934049907,
      "lm_loss": 0.3896,
      "step": 6760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49597842934049907,
      "lm_loss": 0.5924,
      "step": 6760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49597842934049907,
      "lm_loss": 0.6044,
      "step": 6760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49597842934049907,
      "lm_loss": 0.3683,
      "step": 6760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49597842934049907,
      "lm_loss": 0.3682,
      "step": 6760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4960517989306382,
      "grad_norm": 0.5837882387841388,
      "learning_rate": 5.30636278285294e-06,
      "loss": 0.5034,
      "step": 6761
    },
    {
      "epoch": 0.49612516852077737,
      "grad_norm": 1.487780935749942,
      "learning_rate": 5.305176809831456e-06,
      "loss": 0.8233,
      "step": 6762
    },
    {
      "epoch": 0.49619853811091647,
      "grad_norm": 0.6074742118081828,
      "learning_rate": 5.303990819575882e-06,
      "loss": 0.5089,
      "step": 6763
    },
    {
      "epoch": 0.4962719077010556,
      "grad_norm": 1.486858094809543,
      "learning_rate": 5.302804812153194e-06,
      "loss": 0.8125,
      "step": 6764
    },
    {
      "epoch": 0.4963452772911947,
      "grad_norm": 1.5172892485927247,
      "learning_rate": 5.3016187876303685e-06,
      "loss": 0.8204,
      "step": 6765
    },
    {
      "epoch": 0.49641864688133386,
      "grad_norm": 0.5862983269783575,
      "learning_rate": 5.300432746074381e-06,
      "loss": 0.5152,
      "step": 6766
    },
    {
      "epoch": 0.496492016471473,
      "grad_norm": 1.4719849464616623,
      "learning_rate": 5.299246687552214e-06,
      "loss": 0.7379,
      "step": 6767
    },
    {
      "epoch": 0.4965653860616121,
      "grad_norm": 1.3356284561564036,
      "learning_rate": 5.298060612130846e-06,
      "loss": 0.789,
      "step": 6768
    },
    {
      "epoch": 0.4965653860616121,
      "lm_loss": 0.1926,
      "step": 6768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4965653860616121,
      "lm_loss": 0.4591,
      "step": 6768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4965653860616121,
      "lm_loss": 0.5396,
      "step": 6768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4965653860616121,
      "lm_loss": 0.3158,
      "step": 6768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4965653860616121,
      "lm_loss": 0.515,
      "step": 6768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4965653860616121,
      "lm_loss": 0.6114,
      "step": 6768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4965653860616121,
      "lm_loss": 0.4915,
      "step": 6768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4965653860616121,
      "lm_loss": 0.4406,
      "step": 6768,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49663875565175125,
      "grad_norm": 0.5918015567748366,
      "learning_rate": 5.296874519877256e-06,
      "loss": 0.4939,
      "step": 6769
    },
    {
      "epoch": 0.49671212524189035,
      "grad_norm": 0.5671909874932777,
      "learning_rate": 5.295688410858427e-06,
      "loss": 0.4859,
      "step": 6770
    },
    {
      "epoch": 0.4967854948320295,
      "grad_norm": 1.5459394385220464,
      "learning_rate": 5.294502285141343e-06,
      "loss": 0.8209,
      "step": 6771
    },
    {
      "epoch": 0.49685886442216864,
      "grad_norm": 1.54862613807464,
      "learning_rate": 5.293316142792982e-06,
      "loss": 0.7514,
      "step": 6772
    },
    {
      "epoch": 0.49693223401230774,
      "grad_norm": 1.6306586359947333,
      "learning_rate": 5.2921299838803355e-06,
      "loss": 0.8275,
      "step": 6773
    },
    {
      "epoch": 0.4970056036024469,
      "grad_norm": 0.5506172843396014,
      "learning_rate": 5.290943808470382e-06,
      "loss": 0.5043,
      "step": 6774
    },
    {
      "epoch": 0.497078973192586,
      "grad_norm": 1.5328567787187517,
      "learning_rate": 5.289757616630115e-06,
      "loss": 0.7968,
      "step": 6775
    },
    {
      "epoch": 0.49715234278272513,
      "grad_norm": 1.2937735007398525,
      "learning_rate": 5.288571408426516e-06,
      "loss": 0.8223,
      "step": 6776
    },
    {
      "epoch": 0.49715234278272513,
      "lm_loss": 1.2892,
      "step": 6776,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.49715234278272513,
      "lm_loss": 1.2334,
      "step": 6776,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.49715234278272513,
      "lm_loss": 0.0698,
      "step": 6776,
      "vm_loss": 0.0515
    },
    {
      "epoch": 0.49715234278272513,
      "lm_loss": 0.1296,
      "step": 6776,
      "vm_loss": 0.1926
    },
    {
      "epoch": 0.49715234278272513,
      "lm_loss": 0.2138,
      "step": 6776,
      "vm_loss": 0.1634
    },
    {
      "epoch": 0.49715234278272513,
      "lm_loss": 0.1253,
      "step": 6776,
      "vm_loss": 0.078
    },
    {
      "epoch": 0.49715234278272513,
      "lm_loss": 0.2755,
      "step": 6776,
      "vm_loss": 0.0369
    },
    {
      "epoch": 0.49715234278272513,
      "lm_loss": 0.2261,
      "step": 6776,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.4972257123728643,
      "grad_norm": 1.5084809105215047,
      "learning_rate": 5.287385183926573e-06,
      "loss": 0.7069,
      "step": 6777
    },
    {
      "epoch": 0.4972990819630034,
      "grad_norm": 1.5059103618456966,
      "learning_rate": 5.2861989431972795e-06,
      "loss": 0.7606,
      "step": 6778
    },
    {
      "epoch": 0.4973724515531425,
      "grad_norm": 0.6013056873732517,
      "learning_rate": 5.285012686305623e-06,
      "loss": 0.508,
      "step": 6779
    },
    {
      "epoch": 0.4974458211432816,
      "grad_norm": 0.5609730386213719,
      "learning_rate": 5.283826413318592e-06,
      "loss": 0.4818,
      "step": 6780
    },
    {
      "epoch": 0.49751919073342077,
      "grad_norm": 0.5746913786874673,
      "learning_rate": 5.282640124303183e-06,
      "loss": 0.505,
      "step": 6781
    },
    {
      "epoch": 0.4975925603235599,
      "grad_norm": 1.5004708302706538,
      "learning_rate": 5.281453819326384e-06,
      "loss": 0.7738,
      "step": 6782
    },
    {
      "epoch": 0.497665929913699,
      "grad_norm": 1.4995683726296278,
      "learning_rate": 5.280267498455193e-06,
      "loss": 0.7784,
      "step": 6783
    },
    {
      "epoch": 0.49773929950383816,
      "grad_norm": 1.595894637094435,
      "learning_rate": 5.279081161756602e-06,
      "loss": 0.8019,
      "step": 6784
    },
    {
      "epoch": 0.49773929950383816,
      "lm_loss": 1.1239,
      "step": 6784,
      "vm_loss": 0.0399
    },
    {
      "epoch": 0.49773929950383816,
      "lm_loss": 0.1107,
      "step": 6784,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.49773929950383816,
      "lm_loss": 0.4774,
      "step": 6784,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.49773929950383816,
      "lm_loss": 0.4588,
      "step": 6784,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.49773929950383816,
      "lm_loss": 0.908,
      "step": 6784,
      "vm_loss": 0.1655
    },
    {
      "epoch": 0.49773929950383816,
      "lm_loss": 0.2185,
      "step": 6784,
      "vm_loss": 0.0772
    },
    {
      "epoch": 0.49773929950383816,
      "lm_loss": 1.2957,
      "step": 6784,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.49773929950383816,
      "lm_loss": 0.2141,
      "step": 6784,
      "vm_loss": 0.1477
    },
    {
      "epoch": 0.49781266909397726,
      "grad_norm": 1.4857827922426032,
      "learning_rate": 5.277894809297604e-06,
      "loss": 0.7619,
      "step": 6785
    },
    {
      "epoch": 0.4978860386841164,
      "grad_norm": 1.6156796783652378,
      "learning_rate": 5.2767084411452e-06,
      "loss": 0.7545,
      "step": 6786
    },
    {
      "epoch": 0.49795940827425555,
      "grad_norm": 1.4115217731834269,
      "learning_rate": 5.275522057366387e-06,
      "loss": 0.8549,
      "step": 6787
    },
    {
      "epoch": 0.49803277786439465,
      "grad_norm": 1.530232023098878,
      "learning_rate": 5.274335658028158e-06,
      "loss": 0.7927,
      "step": 6788
    },
    {
      "epoch": 0.4981061474545338,
      "grad_norm": 0.6231581925334851,
      "learning_rate": 5.273149243197517e-06,
      "loss": 0.4831,
      "step": 6789
    },
    {
      "epoch": 0.4981795170446729,
      "grad_norm": 1.3181160440896795,
      "learning_rate": 5.271962812941459e-06,
      "loss": 0.7109,
      "step": 6790
    },
    {
      "epoch": 0.49825288663481204,
      "grad_norm": 1.5904691678395044,
      "learning_rate": 5.270776367326991e-06,
      "loss": 0.7486,
      "step": 6791
    },
    {
      "epoch": 0.49832625622495114,
      "grad_norm": 0.6141716461662492,
      "learning_rate": 5.269589906421108e-06,
      "loss": 0.5013,
      "step": 6792
    },
    {
      "epoch": 0.49832625622495114,
      "lm_loss": 0.918,
      "step": 6792,
      "vm_loss": 0.0561
    },
    {
      "epoch": 0.49832625622495114,
      "lm_loss": 0.1194,
      "step": 6792,
      "vm_loss": 0.0816
    },
    {
      "epoch": 0.49832625622495114,
      "lm_loss": 0.3744,
      "step": 6792,
      "vm_loss": 0.1309
    },
    {
      "epoch": 0.49832625622495114,
      "lm_loss": 0.4605,
      "step": 6792,
      "vm_loss": 0.0493
    },
    {
      "epoch": 0.49832625622495114,
      "lm_loss": 0.8918,
      "step": 6792,
      "vm_loss": 0.2548
    },
    {
      "epoch": 0.49832625622495114,
      "lm_loss": 1.0175,
      "step": 6792,
      "vm_loss": 0.0447
    },
    {
      "epoch": 0.49832625622495114,
      "lm_loss": 0.157,
      "step": 6792,
      "vm_loss": 0.1237
    },
    {
      "epoch": 0.49832625622495114,
      "lm_loss": 0.7138,
      "step": 6792,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.4983996258150903,
      "grad_norm": 1.5106024777192548,
      "learning_rate": 5.268403430290817e-06,
      "loss": 0.7296,
      "step": 6793
    },
    {
      "epoch": 0.49847299540522944,
      "grad_norm": 1.4330803927423081,
      "learning_rate": 5.2672169390031195e-06,
      "loss": 0.7082,
      "step": 6794
    },
    {
      "epoch": 0.49854636499536853,
      "grad_norm": 1.548056071093078,
      "learning_rate": 5.26603043262502e-06,
      "loss": 0.7445,
      "step": 6795
    },
    {
      "epoch": 0.4986197345855077,
      "grad_norm": 1.4059003754364785,
      "learning_rate": 5.2648439112235215e-06,
      "loss": 0.778,
      "step": 6796
    },
    {
      "epoch": 0.4986931041756468,
      "grad_norm": 0.5862002348166877,
      "learning_rate": 5.263657374865634e-06,
      "loss": 0.5082,
      "step": 6797
    },
    {
      "epoch": 0.4987664737657859,
      "grad_norm": 1.3263536883306988,
      "learning_rate": 5.26247082361836e-06,
      "loss": 0.664,
      "step": 6798
    },
    {
      "epoch": 0.49883984335592507,
      "grad_norm": 1.49225503433573,
      "learning_rate": 5.261284257548708e-06,
      "loss": 0.85,
      "step": 6799
    },
    {
      "epoch": 0.49891321294606417,
      "grad_norm": 1.5333250541513332,
      "learning_rate": 5.2600976767236875e-06,
      "loss": 0.7773,
      "step": 6800
    },
    {
      "epoch": 0.49891321294606417,
      "lm_loss": 0.48,
      "step": 6800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49891321294606417,
      "lm_loss": 0.4925,
      "step": 6800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49891321294606417,
      "lm_loss": 0.6689,
      "step": 6800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49891321294606417,
      "lm_loss": 0.4991,
      "step": 6800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49891321294606417,
      "lm_loss": 0.5888,
      "step": 6800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49891321294606417,
      "lm_loss": 0.5225,
      "step": 6800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49891321294606417,
      "lm_loss": 0.5524,
      "step": 6800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.49891321294606417,
      "lm_loss": 0.4895,
      "step": 6800,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.4989865825362033,
      "grad_norm": 0.5959685798572553,
      "learning_rate": 5.258911081210307e-06,
      "loss": 0.5089,
      "step": 6801
    },
    {
      "epoch": 0.4990599521263424,
      "grad_norm": 1.4670633002561875,
      "learning_rate": 5.257724471075577e-06,
      "loss": 0.7159,
      "step": 6802
    },
    {
      "epoch": 0.49913332171648156,
      "grad_norm": 1.5981034298444625,
      "learning_rate": 5.2565378463865094e-06,
      "loss": 0.7466,
      "step": 6803
    },
    {
      "epoch": 0.4992066913066207,
      "grad_norm": 0.5694718700143753,
      "learning_rate": 5.255351207210111e-06,
      "loss": 0.486,
      "step": 6804
    },
    {
      "epoch": 0.4992800608967598,
      "grad_norm": 0.5609897786251173,
      "learning_rate": 5.2541645536134e-06,
      "loss": 0.488,
      "step": 6805
    },
    {
      "epoch": 0.49935343048689895,
      "grad_norm": 0.5571678142736942,
      "learning_rate": 5.2529778856633865e-06,
      "loss": 0.5131,
      "step": 6806
    },
    {
      "epoch": 0.49942680007703805,
      "grad_norm": 0.5599094854018475,
      "learning_rate": 5.251791203427085e-06,
      "loss": 0.4846,
      "step": 6807
    },
    {
      "epoch": 0.4995001696671772,
      "grad_norm": 1.6519497954666689,
      "learning_rate": 5.250604506971511e-06,
      "loss": 0.8188,
      "step": 6808
    },
    {
      "epoch": 0.4995001696671772,
      "lm_loss": 1.4185,
      "step": 6808,
      "vm_loss": 0.0427
    },
    {
      "epoch": 0.4995001696671772,
      "lm_loss": 0.3679,
      "step": 6808,
      "vm_loss": 0.1229
    },
    {
      "epoch": 0.4995001696671772,
      "lm_loss": 1.1047,
      "step": 6808,
      "vm_loss": 0.2255
    },
    {
      "epoch": 0.4995001696671772,
      "lm_loss": 0.5656,
      "step": 6808,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.4995001696671772,
      "lm_loss": 0.1995,
      "step": 6808,
      "vm_loss": 0.1481
    },
    {
      "epoch": 0.4995001696671772,
      "lm_loss": 0.5531,
      "step": 6808,
      "vm_loss": 0.2466
    },
    {
      "epoch": 0.4995001696671772,
      "lm_loss": 0.4578,
      "step": 6808,
      "vm_loss": 0.1248
    },
    {
      "epoch": 0.4995001696671772,
      "lm_loss": 0.6632,
      "step": 6808,
      "vm_loss": 0.1458
    },
    {
      "epoch": 0.49957353925731635,
      "grad_norm": 1.7715377570582884,
      "learning_rate": 5.2494177963636785e-06,
      "loss": 0.7284,
      "step": 6809
    },
    {
      "epoch": 0.49964690884745544,
      "grad_norm": 1.5474892653104046,
      "learning_rate": 5.248231071670608e-06,
      "loss": 0.7997,
      "step": 6810
    },
    {
      "epoch": 0.4997202784375946,
      "grad_norm": 1.553354466353105,
      "learning_rate": 5.2470443329593125e-06,
      "loss": 0.7913,
      "step": 6811
    },
    {
      "epoch": 0.4997936480277337,
      "grad_norm": 1.5814840320047991,
      "learning_rate": 5.245857580296811e-06,
      "loss": 0.8268,
      "step": 6812
    },
    {
      "epoch": 0.49986701761787283,
      "grad_norm": 0.6152043811225927,
      "learning_rate": 5.244670813750125e-06,
      "loss": 0.5097,
      "step": 6813
    },
    {
      "epoch": 0.499940387208012,
      "grad_norm": 1.4865099930657233,
      "learning_rate": 5.2434840333862714e-06,
      "loss": 0.728,
      "step": 6814
    },
    {
      "epoch": 0.5000137567981511,
      "grad_norm": 1.606592699004546,
      "learning_rate": 5.2422972392722716e-06,
      "loss": 0.8726,
      "step": 6815
    },
    {
      "epoch": 0.5000871263882902,
      "grad_norm": 1.3291196721942016,
      "learning_rate": 5.241110431475146e-06,
      "loss": 0.7004,
      "step": 6816
    },
    {
      "epoch": 0.5000871263882902,
      "lm_loss": 1.3158,
      "step": 6816,
      "vm_loss": 0.0745
    },
    {
      "epoch": 0.5000871263882902,
      "lm_loss": 1.2791,
      "step": 6816,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.5000871263882902,
      "lm_loss": 0.77,
      "step": 6816,
      "vm_loss": 0.2089
    },
    {
      "epoch": 0.5000871263882902,
      "lm_loss": 0.8042,
      "step": 6816,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.5000871263882902,
      "lm_loss": 0.8741,
      "step": 6816,
      "vm_loss": 0.0711
    },
    {
      "epoch": 0.5000871263882902,
      "lm_loss": 0.6347,
      "step": 6816,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.5000871263882902,
      "lm_loss": 0.2969,
      "step": 6816,
      "vm_loss": 0.1598
    },
    {
      "epoch": 0.5000871263882902,
      "lm_loss": 0.8863,
      "step": 6816,
      "vm_loss": 0.162
    },
    {
      "epoch": 0.5001604959784294,
      "grad_norm": 1.4149792069528526,
      "learning_rate": 5.239923610061918e-06,
      "loss": 0.7526,
      "step": 6817
    },
    {
      "epoch": 0.5002338655685684,
      "grad_norm": 1.6129361707755654,
      "learning_rate": 5.238736775099609e-06,
      "loss": 0.6988,
      "step": 6818
    },
    {
      "epoch": 0.5003072351587076,
      "grad_norm": 0.6029035812577935,
      "learning_rate": 5.237549926655243e-06,
      "loss": 0.5125,
      "step": 6819
    },
    {
      "epoch": 0.5003806047488467,
      "grad_norm": 1.6382349594799952,
      "learning_rate": 5.2363630647958444e-06,
      "loss": 0.7879,
      "step": 6820
    },
    {
      "epoch": 0.5004539743389859,
      "grad_norm": 1.4449250181958715,
      "learning_rate": 5.235176189588439e-06,
      "loss": 0.7596,
      "step": 6821
    },
    {
      "epoch": 0.500527343929125,
      "grad_norm": 1.3743525108447854,
      "learning_rate": 5.233989301100052e-06,
      "loss": 0.7851,
      "step": 6822
    },
    {
      "epoch": 0.500600713519264,
      "grad_norm": 1.4566175319199253,
      "learning_rate": 5.232802399397708e-06,
      "loss": 0.9001,
      "step": 6823
    },
    {
      "epoch": 0.5006740831094032,
      "grad_norm": 0.5592348930409775,
      "learning_rate": 5.231615484548438e-06,
      "loss": 0.4904,
      "step": 6824
    },
    {
      "epoch": 0.5006740831094032,
      "lm_loss": 1.2186,
      "step": 6824,
      "vm_loss": 0.1055
    },
    {
      "epoch": 0.5006740831094032,
      "lm_loss": 0.1285,
      "step": 6824,
      "vm_loss": 0.0747
    },
    {
      "epoch": 0.5006740831094032,
      "lm_loss": 0.1042,
      "step": 6824,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.5006740831094032,
      "lm_loss": 0.0796,
      "step": 6824,
      "vm_loss": 0.071
    },
    {
      "epoch": 0.5006740831094032,
      "lm_loss": 0.8605,
      "step": 6824,
      "vm_loss": 0.1468
    },
    {
      "epoch": 0.5006740831094032,
      "lm_loss": 0.7311,
      "step": 6824,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.5006740831094032,
      "lm_loss": 0.6551,
      "step": 6824,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.5006740831094032,
      "lm_loss": 0.7963,
      "step": 6824,
      "vm_loss": 0.1548
    },
    {
      "epoch": 0.5007474526995424,
      "grad_norm": 1.4367052002147565,
      "learning_rate": 5.230428556619267e-06,
      "loss": 0.5974,
      "step": 6825
    },
    {
      "epoch": 0.5008208222896815,
      "grad_norm": 1.7091787900810038,
      "learning_rate": 5.229241615677224e-06,
      "loss": 0.7684,
      "step": 6826
    },
    {
      "epoch": 0.5008941918798206,
      "grad_norm": 1.4693971300441304,
      "learning_rate": 5.22805466178934e-06,
      "loss": 0.7465,
      "step": 6827
    },
    {
      "epoch": 0.5009675614699597,
      "grad_norm": 1.962567851793101,
      "learning_rate": 5.226867695022644e-06,
      "loss": 0.8023,
      "step": 6828
    },
    {
      "epoch": 0.5010409310600988,
      "grad_norm": 1.404746532567973,
      "learning_rate": 5.225680715444168e-06,
      "loss": 0.7727,
      "step": 6829
    },
    {
      "epoch": 0.501114300650238,
      "grad_norm": 1.43204142898031,
      "learning_rate": 5.2244937231209435e-06,
      "loss": 0.7175,
      "step": 6830
    },
    {
      "epoch": 0.5011876702403771,
      "grad_norm": 0.5569558583196753,
      "learning_rate": 5.223306718120001e-06,
      "loss": 0.4897,
      "step": 6831
    },
    {
      "epoch": 0.5012610398305163,
      "grad_norm": 1.7643697825017703,
      "learning_rate": 5.2221197005083776e-06,
      "loss": 0.7185,
      "step": 6832
    },
    {
      "epoch": 0.5012610398305163,
      "lm_loss": 1.4197,
      "step": 6832,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.5012610398305163,
      "lm_loss": 1.2406,
      "step": 6832,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.5012610398305163,
      "lm_loss": 0.1466,
      "step": 6832,
      "vm_loss": 0.239
    },
    {
      "epoch": 0.5012610398305163,
      "lm_loss": 0.4391,
      "step": 6832,
      "vm_loss": 0.1343
    },
    {
      "epoch": 0.5012610398305163,
      "lm_loss": 0.1794,
      "step": 6832,
      "vm_loss": 0.1753
    },
    {
      "epoch": 0.5012610398305163,
      "lm_loss": 0.3277,
      "step": 6832,
      "vm_loss": 0.0955
    },
    {
      "epoch": 0.5012610398305163,
      "lm_loss": 0.0656,
      "step": 6832,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.5012610398305163,
      "lm_loss": 0.7822,
      "step": 6832,
      "vm_loss": 0.1197
    },
    {
      "epoch": 0.5013344094206553,
      "grad_norm": 1.8685861811548141,
      "learning_rate": 5.220932670353103e-06,
      "loss": 0.7645,
      "step": 6833
    },
    {
      "epoch": 0.5014077790107945,
      "grad_norm": 1.5561412544922355,
      "learning_rate": 5.2197456277212145e-06,
      "loss": 0.7201,
      "step": 6834
    },
    {
      "epoch": 0.5014811486009336,
      "grad_norm": 1.5170681517003113,
      "learning_rate": 5.218558572679745e-06,
      "loss": 0.6723,
      "step": 6835
    },
    {
      "epoch": 0.5015545181910728,
      "grad_norm": 0.5710200805473724,
      "learning_rate": 5.217371505295731e-06,
      "loss": 0.4883,
      "step": 6836
    },
    {
      "epoch": 0.5016278877812119,
      "grad_norm": 0.5762457902054314,
      "learning_rate": 5.216184425636211e-06,
      "loss": 0.4905,
      "step": 6837
    },
    {
      "epoch": 0.501701257371351,
      "grad_norm": 0.5781177676383519,
      "learning_rate": 5.21499733376822e-06,
      "loss": 0.51,
      "step": 6838
    },
    {
      "epoch": 0.5017746269614901,
      "grad_norm": 1.7895407885986236,
      "learning_rate": 5.213810229758799e-06,
      "loss": 0.8004,
      "step": 6839
    },
    {
      "epoch": 0.5018479965516293,
      "grad_norm": 1.7358714388886463,
      "learning_rate": 5.212623113674984e-06,
      "loss": 0.7372,
      "step": 6840
    },
    {
      "epoch": 0.5018479965516293,
      "lm_loss": 1.4086,
      "step": 6840,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.5018479965516293,
      "lm_loss": 0.7443,
      "step": 6840,
      "vm_loss": 0.1066
    },
    {
      "epoch": 0.5018479965516293,
      "lm_loss": 0.1086,
      "step": 6840,
      "vm_loss": 0.1933
    },
    {
      "epoch": 0.5018479965516293,
      "lm_loss": 0.664,
      "step": 6840,
      "vm_loss": 0.1166
    },
    {
      "epoch": 0.5018479965516293,
      "lm_loss": 0.1014,
      "step": 6840,
      "vm_loss": 0.0763
    },
    {
      "epoch": 0.5018479965516293,
      "lm_loss": 1.1258,
      "step": 6840,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.5018479965516293,
      "lm_loss": 0.8751,
      "step": 6840,
      "vm_loss": 0.1826
    },
    {
      "epoch": 0.5018479965516293,
      "lm_loss": 0.6545,
      "step": 6840,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.5019213661417684,
      "grad_norm": 1.5746321617933552,
      "learning_rate": 5.211435985583813e-06,
      "loss": 0.6778,
      "step": 6841
    },
    {
      "epoch": 0.5019947357319076,
      "grad_norm": 1.41901611565775,
      "learning_rate": 5.210248845552331e-06,
      "loss": 0.7326,
      "step": 6842
    },
    {
      "epoch": 0.5020681053220466,
      "grad_norm": 1.5581184587170236,
      "learning_rate": 5.209061693647577e-06,
      "loss": 0.9458,
      "step": 6843
    },
    {
      "epoch": 0.5021414749121857,
      "grad_norm": 1.5523466665837464,
      "learning_rate": 5.207874529936589e-06,
      "loss": 0.7761,
      "step": 6844
    },
    {
      "epoch": 0.5022148445023249,
      "grad_norm": 1.5334933100684305,
      "learning_rate": 5.206687354486413e-06,
      "loss": 0.7617,
      "step": 6845
    },
    {
      "epoch": 0.502288214092464,
      "grad_norm": 1.620413154963537,
      "learning_rate": 5.205500167364089e-06,
      "loss": 0.7706,
      "step": 6846
    },
    {
      "epoch": 0.5023615836826032,
      "grad_norm": 0.6564742443364573,
      "learning_rate": 5.204312968636661e-06,
      "loss": 0.5096,
      "step": 6847
    },
    {
      "epoch": 0.5024349532727422,
      "grad_norm": 0.5961906264595178,
      "learning_rate": 5.203125758371176e-06,
      "loss": 0.4831,
      "step": 6848
    },
    {
      "epoch": 0.5024349532727422,
      "lm_loss": 1.4764,
      "step": 6848,
      "vm_loss": 0.0524
    },
    {
      "epoch": 0.5024349532727422,
      "lm_loss": 1.329,
      "step": 6848,
      "vm_loss": 0.0537
    },
    {
      "epoch": 0.5024349532727422,
      "lm_loss": 0.425,
      "step": 6848,
      "vm_loss": 0.0402
    },
    {
      "epoch": 0.5024349532727422,
      "lm_loss": 0.7809,
      "step": 6848,
      "vm_loss": 0.224
    },
    {
      "epoch": 0.5024349532727422,
      "lm_loss": 0.9259,
      "step": 6848,
      "vm_loss": 0.1339
    },
    {
      "epoch": 0.5024349532727422,
      "lm_loss": 0.7864,
      "step": 6848,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.5024349532727422,
      "lm_loss": 0.0511,
      "step": 6848,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.5024349532727422,
      "lm_loss": 0.5371,
      "step": 6848,
      "vm_loss": 0.126
    },
    {
      "epoch": 0.5025083228628814,
      "grad_norm": 1.508918202412179,
      "learning_rate": 5.201938536634674e-06,
      "loss": 0.7913,
      "step": 6849
    },
    {
      "epoch": 0.5025816924530205,
      "grad_norm": 1.4921044746924796,
      "learning_rate": 5.200751303494205e-06,
      "loss": 0.7105,
      "step": 6850
    },
    {
      "epoch": 0.5026550620431597,
      "grad_norm": 1.547007941630039,
      "learning_rate": 5.199564059016811e-06,
      "loss": 0.6977,
      "step": 6851
    },
    {
      "epoch": 0.5027284316332988,
      "grad_norm": 1.5400491756377501,
      "learning_rate": 5.19837680326954e-06,
      "loss": 0.7066,
      "step": 6852
    },
    {
      "epoch": 0.5028018012234379,
      "grad_norm": 1.6349763627221254,
      "learning_rate": 5.197189536319442e-06,
      "loss": 0.7753,
      "step": 6853
    },
    {
      "epoch": 0.502875170813577,
      "grad_norm": 1.6296752853217251,
      "learning_rate": 5.196002258233561e-06,
      "loss": 0.6908,
      "step": 6854
    },
    {
      "epoch": 0.5029485404037162,
      "grad_norm": 1.4086815908706858,
      "learning_rate": 5.194814969078947e-06,
      "loss": 0.7339,
      "step": 6855
    },
    {
      "epoch": 0.5030219099938553,
      "grad_norm": 0.6071292532196406,
      "learning_rate": 5.19362766892265e-06,
      "loss": 0.4931,
      "step": 6856
    },
    {
      "epoch": 0.5030219099938553,
      "lm_loss": 1.3695,
      "step": 6856,
      "vm_loss": 0.0959
    },
    {
      "epoch": 0.5030219099938553,
      "lm_loss": 0.2188,
      "step": 6856,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.5030219099938553,
      "lm_loss": 0.1854,
      "step": 6856,
      "vm_loss": 0.175
    },
    {
      "epoch": 0.5030219099938553,
      "lm_loss": 1.0225,
      "step": 6856,
      "vm_loss": 0.2723
    },
    {
      "epoch": 0.5030219099938553,
      "lm_loss": 0.9881,
      "step": 6856,
      "vm_loss": 0.1334
    },
    {
      "epoch": 0.5030219099938553,
      "lm_loss": 0.404,
      "step": 6856,
      "vm_loss": 0.1565
    },
    {
      "epoch": 0.5030219099938553,
      "lm_loss": 0.4245,
      "step": 6856,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.5030219099938553,
      "lm_loss": 0.1227,
      "step": 6856,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.5030952795839945,
      "grad_norm": 1.5585918098664309,
      "learning_rate": 5.192440357831719e-06,
      "loss": 0.7546,
      "step": 6857
    },
    {
      "epoch": 0.5031686491741335,
      "grad_norm": 1.5574452162508043,
      "learning_rate": 5.191253035873203e-06,
      "loss": 0.7363,
      "step": 6858
    },
    {
      "epoch": 0.5032420187642727,
      "grad_norm": 1.5687839951862272,
      "learning_rate": 5.190065703114157e-06,
      "loss": 0.7173,
      "step": 6859
    },
    {
      "epoch": 0.5033153883544118,
      "grad_norm": 1.639907905548904,
      "learning_rate": 5.188878359621628e-06,
      "loss": 0.699,
      "step": 6860
    },
    {
      "epoch": 0.503388757944551,
      "grad_norm": 1.5812234578611815,
      "learning_rate": 5.18769100546267e-06,
      "loss": 0.7597,
      "step": 6861
    },
    {
      "epoch": 0.5034621275346901,
      "grad_norm": 0.5473632706008492,
      "learning_rate": 5.186503640704338e-06,
      "loss": 0.5011,
      "step": 6862
    },
    {
      "epoch": 0.5035354971248291,
      "grad_norm": 0.5544895123868075,
      "learning_rate": 5.185316265413681e-06,
      "loss": 0.5101,
      "step": 6863
    },
    {
      "epoch": 0.5036088667149683,
      "grad_norm": 1.5734887563300448,
      "learning_rate": 5.184128879657757e-06,
      "loss": 0.8054,
      "step": 6864
    },
    {
      "epoch": 0.5036088667149683,
      "lm_loss": 0.5187,
      "step": 6864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5036088667149683,
      "lm_loss": 0.1784,
      "step": 6864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5036088667149683,
      "lm_loss": 0.6174,
      "step": 6864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5036088667149683,
      "lm_loss": 0.3203,
      "step": 6864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5036088667149683,
      "lm_loss": 0.4796,
      "step": 6864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5036088667149683,
      "lm_loss": 0.5474,
      "step": 6864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5036088667149683,
      "lm_loss": 0.4895,
      "step": 6864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5036088667149683,
      "lm_loss": 0.4186,
      "step": 6864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5036822363051074,
      "grad_norm": 0.5553676515281135,
      "learning_rate": 5.1829414835036185e-06,
      "loss": 0.4968,
      "step": 6865
    },
    {
      "epoch": 0.5037556058952466,
      "grad_norm": 0.541417904436069,
      "learning_rate": 5.1817540770183224e-06,
      "loss": 0.4848,
      "step": 6866
    },
    {
      "epoch": 0.5038289754853857,
      "grad_norm": 1.658983867602852,
      "learning_rate": 5.180566660268924e-06,
      "loss": 0.8197,
      "step": 6867
    },
    {
      "epoch": 0.5039023450755248,
      "grad_norm": 1.6475475463305849,
      "learning_rate": 5.179379233322478e-06,
      "loss": 0.8184,
      "step": 6868
    },
    {
      "epoch": 0.5039757146656639,
      "grad_norm": 1.3545430989109102,
      "learning_rate": 5.178191796246043e-06,
      "loss": 0.795,
      "step": 6869
    },
    {
      "epoch": 0.5040490842558031,
      "grad_norm": 1.4580170323455397,
      "learning_rate": 5.1770043491066755e-06,
      "loss": 0.7382,
      "step": 6870
    },
    {
      "epoch": 0.5041224538459422,
      "grad_norm": 1.451394023533848,
      "learning_rate": 5.175816891971433e-06,
      "loss": 0.8509,
      "step": 6871
    },
    {
      "epoch": 0.5041958234360814,
      "grad_norm": 1.2662397040479674,
      "learning_rate": 5.174629424907378e-06,
      "loss": 0.728,
      "step": 6872
    },
    {
      "epoch": 0.5041958234360814,
      "lm_loss": 1.2258,
      "step": 6872,
      "vm_loss": 0.0408
    },
    {
      "epoch": 0.5041958234360814,
      "lm_loss": 1.3924,
      "step": 6872,
      "vm_loss": 0.1707
    },
    {
      "epoch": 0.5041958234360814,
      "lm_loss": 0.1015,
      "step": 6872,
      "vm_loss": 0.0308
    },
    {
      "epoch": 0.5041958234360814,
      "lm_loss": 0.2843,
      "step": 6872,
      "vm_loss": 0.1265
    },
    {
      "epoch": 0.5041958234360814,
      "lm_loss": 0.0965,
      "step": 6872,
      "vm_loss": 0.2004
    },
    {
      "epoch": 0.5041958234360814,
      "lm_loss": 0.4904,
      "step": 6872,
      "vm_loss": 0.1449
    },
    {
      "epoch": 0.5041958234360814,
      "lm_loss": 0.7992,
      "step": 6872,
      "vm_loss": 0.163
    },
    {
      "epoch": 0.5041958234360814,
      "lm_loss": 1.2401,
      "step": 6872,
      "vm_loss": 0.2017
    },
    {
      "epoch": 0.5042691930262204,
      "grad_norm": 1.475910577836288,
      "learning_rate": 5.173441947981566e-06,
      "loss": 0.7442,
      "step": 6873
    },
    {
      "epoch": 0.5043425626163596,
      "grad_norm": 1.497875342844954,
      "learning_rate": 5.172254461261056e-06,
      "loss": 0.7892,
      "step": 6874
    },
    {
      "epoch": 0.5044159322064987,
      "grad_norm": 1.5804820735964336,
      "learning_rate": 5.171066964812911e-06,
      "loss": 0.7115,
      "step": 6875
    },
    {
      "epoch": 0.5044893017966379,
      "grad_norm": 1.444933974683728,
      "learning_rate": 5.16987945870419e-06,
      "loss": 0.7809,
      "step": 6876
    },
    {
      "epoch": 0.504562671386777,
      "grad_norm": 0.6521390390580231,
      "learning_rate": 5.168691943001957e-06,
      "loss": 0.4985,
      "step": 6877
    },
    {
      "epoch": 0.5046360409769161,
      "grad_norm": 0.6330654449147997,
      "learning_rate": 5.167504417773271e-06,
      "loss": 0.5006,
      "step": 6878
    },
    {
      "epoch": 0.5047094105670552,
      "grad_norm": 1.3676143421210443,
      "learning_rate": 5.166316883085196e-06,
      "loss": 0.7836,
      "step": 6879
    },
    {
      "epoch": 0.5047827801571944,
      "grad_norm": 0.5674139763871814,
      "learning_rate": 5.165129339004796e-06,
      "loss": 0.5041,
      "step": 6880
    },
    {
      "epoch": 0.5047827801571944,
      "lm_loss": 0.138,
      "step": 6880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5047827801571944,
      "lm_loss": 0.541,
      "step": 6880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5047827801571944,
      "lm_loss": 0.4902,
      "step": 6880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5047827801571944,
      "lm_loss": 0.5732,
      "step": 6880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5047827801571944,
      "lm_loss": 0.4086,
      "step": 6880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5047827801571944,
      "lm_loss": 0.4,
      "step": 6880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5047827801571944,
      "lm_loss": 0.4132,
      "step": 6880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5047827801571944,
      "lm_loss": 0.4329,
      "step": 6880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5048561497473335,
      "grad_norm": 0.5611733263127356,
      "learning_rate": 5.163941785599133e-06,
      "loss": 0.496,
      "step": 6881
    },
    {
      "epoch": 0.5049295193374727,
      "grad_norm": 1.726774141889632,
      "learning_rate": 5.162754222935269e-06,
      "loss": 0.9122,
      "step": 6882
    },
    {
      "epoch": 0.5050028889276117,
      "grad_norm": 0.6130519048564905,
      "learning_rate": 5.161566651080273e-06,
      "loss": 0.5181,
      "step": 6883
    },
    {
      "epoch": 0.5050762585177508,
      "grad_norm": 1.4932959076417334,
      "learning_rate": 5.160379070101207e-06,
      "loss": 0.8775,
      "step": 6884
    },
    {
      "epoch": 0.50514962810789,
      "grad_norm": 0.5775990834165757,
      "learning_rate": 5.159191480065139e-06,
      "loss": 0.4997,
      "step": 6885
    },
    {
      "epoch": 0.5052229976980291,
      "grad_norm": 1.500478492970901,
      "learning_rate": 5.158003881039133e-06,
      "loss": 0.7097,
      "step": 6886
    },
    {
      "epoch": 0.5052963672881683,
      "grad_norm": 1.4348280058427476,
      "learning_rate": 5.156816273090256e-06,
      "loss": 0.7939,
      "step": 6887
    },
    {
      "epoch": 0.5053697368783073,
      "grad_norm": 0.5767824866868886,
      "learning_rate": 5.155628656285577e-06,
      "loss": 0.51,
      "step": 6888
    },
    {
      "epoch": 0.5053697368783073,
      "lm_loss": 1.7034,
      "step": 6888,
      "vm_loss": 0.067
    },
    {
      "epoch": 0.5053697368783073,
      "lm_loss": 0.1004,
      "step": 6888,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.5053697368783073,
      "lm_loss": 0.9639,
      "step": 6888,
      "vm_loss": 0.1223
    },
    {
      "epoch": 0.5053697368783073,
      "lm_loss": 0.8166,
      "step": 6888,
      "vm_loss": 0.1635
    },
    {
      "epoch": 0.5053697368783073,
      "lm_loss": 0.5869,
      "step": 6888,
      "vm_loss": 0.1236
    },
    {
      "epoch": 0.5053697368783073,
      "lm_loss": 0.8151,
      "step": 6888,
      "vm_loss": 0.0947
    },
    {
      "epoch": 0.5053697368783073,
      "lm_loss": 0.0513,
      "step": 6888,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.5053697368783073,
      "lm_loss": 0.125,
      "step": 6888,
      "vm_loss": 0.0928
    },
    {
      "epoch": 0.5054431064684465,
      "grad_norm": 1.4290435912739137,
      "learning_rate": 5.154441030692162e-06,
      "loss": 0.7281,
      "step": 6889
    },
    {
      "epoch": 0.5055164760585856,
      "grad_norm": 1.3672390131229784,
      "learning_rate": 5.153253396377079e-06,
      "loss": 0.6849,
      "step": 6890
    },
    {
      "epoch": 0.5055898456487248,
      "grad_norm": 0.6003698356833169,
      "learning_rate": 5.152065753407398e-06,
      "loss": 0.4906,
      "step": 6891
    },
    {
      "epoch": 0.5056632152388638,
      "grad_norm": 1.3807969289034132,
      "learning_rate": 5.150878101850187e-06,
      "loss": 0.7356,
      "step": 6892
    },
    {
      "epoch": 0.505736584829003,
      "grad_norm": 1.533012418996563,
      "learning_rate": 5.149690441772515e-06,
      "loss": 0.7953,
      "step": 6893
    },
    {
      "epoch": 0.5058099544191421,
      "grad_norm": 0.5594448824785659,
      "learning_rate": 5.148502773241454e-06,
      "loss": 0.496,
      "step": 6894
    },
    {
      "epoch": 0.5058833240092813,
      "grad_norm": 1.339338352865733,
      "learning_rate": 5.147315096324073e-06,
      "loss": 0.7759,
      "step": 6895
    },
    {
      "epoch": 0.5059566935994204,
      "grad_norm": 1.5880355428818538,
      "learning_rate": 5.1461274110874445e-06,
      "loss": 0.7794,
      "step": 6896
    },
    {
      "epoch": 0.5059566935994204,
      "lm_loss": 1.2288,
      "step": 6896,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.5059566935994204,
      "lm_loss": 0.9424,
      "step": 6896,
      "vm_loss": 0.1635
    },
    {
      "epoch": 0.5059566935994204,
      "lm_loss": 0.7777,
      "step": 6896,
      "vm_loss": 0.1703
    },
    {
      "epoch": 0.5059566935994204,
      "lm_loss": 0.4421,
      "step": 6896,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.5059566935994204,
      "lm_loss": 0.7552,
      "step": 6896,
      "vm_loss": 0.1288
    },
    {
      "epoch": 0.5059566935994204,
      "lm_loss": 0.4391,
      "step": 6896,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.5059566935994204,
      "lm_loss": 0.6031,
      "step": 6896,
      "vm_loss": 0.058
    },
    {
      "epoch": 0.5059566935994204,
      "lm_loss": 0.2275,
      "step": 6896,
      "vm_loss": 0.0857
    },
    {
      "epoch": 0.5060300631895595,
      "grad_norm": 1.3813843572940685,
      "learning_rate": 5.144939717598639e-06,
      "loss": 0.7586,
      "step": 6897
    },
    {
      "epoch": 0.5061034327796986,
      "grad_norm": 1.6038550745340816,
      "learning_rate": 5.143752015924728e-06,
      "loss": 0.8356,
      "step": 6898
    },
    {
      "epoch": 0.5061768023698378,
      "grad_norm": 1.714010703598806,
      "learning_rate": 5.1425643061327856e-06,
      "loss": 0.879,
      "step": 6899
    },
    {
      "epoch": 0.5062501719599769,
      "grad_norm": 1.5011756912934375,
      "learning_rate": 5.141376588289885e-06,
      "loss": 0.7903,
      "step": 6900
    },
    {
      "epoch": 0.506323541550116,
      "grad_norm": 0.6104254995028718,
      "learning_rate": 5.140188862463096e-06,
      "loss": 0.5263,
      "step": 6901
    },
    {
      "epoch": 0.5063969111402551,
      "grad_norm": 1.423525464449633,
      "learning_rate": 5.139001128719495e-06,
      "loss": 0.7737,
      "step": 6902
    },
    {
      "epoch": 0.5064702807303942,
      "grad_norm": 1.5090816387630561,
      "learning_rate": 5.137813387126156e-06,
      "loss": 0.7648,
      "step": 6903
    },
    {
      "epoch": 0.5065436503205334,
      "grad_norm": 1.5306020180163276,
      "learning_rate": 5.136625637750156e-06,
      "loss": 0.6645,
      "step": 6904
    },
    {
      "epoch": 0.5065436503205334,
      "lm_loss": 3.0157,
      "step": 6904,
      "vm_loss": 0.049
    },
    {
      "epoch": 0.5065436503205334,
      "lm_loss": 1.0239,
      "step": 6904,
      "vm_loss": 0.0369
    },
    {
      "epoch": 0.5065436503205334,
      "lm_loss": 0.2677,
      "step": 6904,
      "vm_loss": 0.1579
    },
    {
      "epoch": 0.5065436503205334,
      "lm_loss": 0.1949,
      "step": 6904,
      "vm_loss": 0.0775
    },
    {
      "epoch": 0.5065436503205334,
      "lm_loss": 0.478,
      "step": 6904,
      "vm_loss": 0.1536
    },
    {
      "epoch": 0.5065436503205334,
      "lm_loss": 0.7935,
      "step": 6904,
      "vm_loss": 0.1237
    },
    {
      "epoch": 0.5065436503205334,
      "lm_loss": 0.7885,
      "step": 6904,
      "vm_loss": 0.1633
    },
    {
      "epoch": 0.5065436503205334,
      "lm_loss": 0.639,
      "step": 6904,
      "vm_loss": 0.1887
    },
    {
      "epoch": 0.5066170199106725,
      "grad_norm": 1.4158822942379639,
      "learning_rate": 5.135437880658566e-06,
      "loss": 0.8064,
      "step": 6905
    },
    {
      "epoch": 0.5066903895008117,
      "grad_norm": 1.491903157538395,
      "learning_rate": 5.134250115918462e-06,
      "loss": 0.7551,
      "step": 6906
    },
    {
      "epoch": 0.5067637590909507,
      "grad_norm": 1.619728340709564,
      "learning_rate": 5.133062343596925e-06,
      "loss": 0.7696,
      "step": 6907
    },
    {
      "epoch": 0.5068371286810899,
      "grad_norm": 1.5146667509369376,
      "learning_rate": 5.131874563761025e-06,
      "loss": 0.7475,
      "step": 6908
    },
    {
      "epoch": 0.506910498271229,
      "grad_norm": 0.5848296427082302,
      "learning_rate": 5.1306867764778445e-06,
      "loss": 0.5067,
      "step": 6909
    },
    {
      "epoch": 0.5069838678613682,
      "grad_norm": 1.5776688290083736,
      "learning_rate": 5.129498981814457e-06,
      "loss": 0.7867,
      "step": 6910
    },
    {
      "epoch": 0.5070572374515073,
      "grad_norm": 1.3810822397775429,
      "learning_rate": 5.128311179837939e-06,
      "loss": 0.6944,
      "step": 6911
    },
    {
      "epoch": 0.5071306070416464,
      "grad_norm": 0.55302720434226,
      "learning_rate": 5.127123370615373e-06,
      "loss": 0.4945,
      "step": 6912
    },
    {
      "epoch": 0.5071306070416464,
      "lm_loss": 0.5263,
      "step": 6912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5071306070416464,
      "lm_loss": 0.6057,
      "step": 6912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5071306070416464,
      "lm_loss": 0.4567,
      "step": 6912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5071306070416464,
      "lm_loss": 0.5068,
      "step": 6912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5071306070416464,
      "lm_loss": 0.5519,
      "step": 6912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5071306070416464,
      "lm_loss": 0.5143,
      "step": 6912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5071306070416464,
      "lm_loss": 0.5881,
      "step": 6912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5071306070416464,
      "lm_loss": 0.5183,
      "step": 6912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5072039766317855,
      "grad_norm": 0.539132075442193,
      "learning_rate": 5.1259355542138355e-06,
      "loss": 0.504,
      "step": 6913
    },
    {
      "epoch": 0.5072773462219247,
      "grad_norm": 1.5156363306968375,
      "learning_rate": 5.124747730700406e-06,
      "loss": 0.8624,
      "step": 6914
    },
    {
      "epoch": 0.5073507158120638,
      "grad_norm": 1.6708649338902897,
      "learning_rate": 5.123559900142162e-06,
      "loss": 0.8385,
      "step": 6915
    },
    {
      "epoch": 0.507424085402203,
      "grad_norm": 1.6762405063813455,
      "learning_rate": 5.122372062606183e-06,
      "loss": 0.8542,
      "step": 6916
    },
    {
      "epoch": 0.507497454992342,
      "grad_norm": 1.4865341610366407,
      "learning_rate": 5.121184218159554e-06,
      "loss": 0.7476,
      "step": 6917
    },
    {
      "epoch": 0.5075708245824812,
      "grad_norm": 0.5614107396203354,
      "learning_rate": 5.119996366869351e-06,
      "loss": 0.5023,
      "step": 6918
    },
    {
      "epoch": 0.5076441941726203,
      "grad_norm": 1.4941483074024577,
      "learning_rate": 5.118808508802654e-06,
      "loss": 0.8572,
      "step": 6919
    },
    {
      "epoch": 0.5077175637627594,
      "grad_norm": 1.5576715655437243,
      "learning_rate": 5.117620644026548e-06,
      "loss": 0.7419,
      "step": 6920
    },
    {
      "epoch": 0.5077175637627594,
      "lm_loss": 0.4525,
      "step": 6920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5077175637627594,
      "lm_loss": 0.5117,
      "step": 6920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5077175637627594,
      "lm_loss": 0.5514,
      "step": 6920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5077175637627594,
      "lm_loss": 0.5246,
      "step": 6920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5077175637627594,
      "lm_loss": 0.5664,
      "step": 6920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5077175637627594,
      "lm_loss": 0.5189,
      "step": 6920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5077175637627594,
      "lm_loss": 0.514,
      "step": 6920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5077175637627594,
      "lm_loss": 0.3426,
      "step": 6920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5077909333528986,
      "grad_norm": 0.5655169621635942,
      "learning_rate": 5.1164327726081135e-06,
      "loss": 0.4924,
      "step": 6921
    },
    {
      "epoch": 0.5078643029430376,
      "grad_norm": 1.50326405216214,
      "learning_rate": 5.115244894614429e-06,
      "loss": 0.7845,
      "step": 6922
    },
    {
      "epoch": 0.5079376725331768,
      "grad_norm": 1.586108151402205,
      "learning_rate": 5.114057010112583e-06,
      "loss": 0.8132,
      "step": 6923
    },
    {
      "epoch": 0.5080110421233159,
      "grad_norm": 0.5536595176932755,
      "learning_rate": 5.112869119169654e-06,
      "loss": 0.5118,
      "step": 6924
    },
    {
      "epoch": 0.5080844117134551,
      "grad_norm": 1.5614799181564902,
      "learning_rate": 5.111681221852727e-06,
      "loss": 0.8254,
      "step": 6925
    },
    {
      "epoch": 0.5081577813035942,
      "grad_norm": 1.4162622007616688,
      "learning_rate": 5.110493318228885e-06,
      "loss": 0.8721,
      "step": 6926
    },
    {
      "epoch": 0.5082311508937333,
      "grad_norm": 1.679772139228737,
      "learning_rate": 5.109305408365211e-06,
      "loss": 0.7828,
      "step": 6927
    },
    {
      "epoch": 0.5083045204838724,
      "grad_norm": 1.352521915854706,
      "learning_rate": 5.108117492328789e-06,
      "loss": 0.7956,
      "step": 6928
    },
    {
      "epoch": 0.5083045204838724,
      "lm_loss": 0.3097,
      "step": 6928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5083045204838724,
      "lm_loss": 0.6016,
      "step": 6928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5083045204838724,
      "lm_loss": 0.5437,
      "step": 6928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5083045204838724,
      "lm_loss": 0.4385,
      "step": 6928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5083045204838724,
      "lm_loss": 0.3799,
      "step": 6928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5083045204838724,
      "lm_loss": 0.5234,
      "step": 6928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5083045204838724,
      "lm_loss": 0.5772,
      "step": 6928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5083045204838724,
      "lm_loss": 0.4889,
      "step": 6928,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5083778900740116,
      "grad_norm": 0.5574773940116238,
      "learning_rate": 5.106929570186706e-06,
      "loss": 0.4972,
      "step": 6929
    },
    {
      "epoch": 0.5084512596641507,
      "grad_norm": 1.3091429560339032,
      "learning_rate": 5.105741642006045e-06,
      "loss": 0.7532,
      "step": 6930
    },
    {
      "epoch": 0.5085246292542899,
      "grad_norm": 1.5554965104493859,
      "learning_rate": 5.104553707853891e-06,
      "loss": 0.7886,
      "step": 6931
    },
    {
      "epoch": 0.5085979988444289,
      "grad_norm": 1.4173024261557732,
      "learning_rate": 5.10336576779733e-06,
      "loss": 0.6751,
      "step": 6932
    },
    {
      "epoch": 0.5086713684345681,
      "grad_norm": 0.5561999214151477,
      "learning_rate": 5.102177821903449e-06,
      "loss": 0.5097,
      "step": 6933
    },
    {
      "epoch": 0.5087447380247072,
      "grad_norm": 1.3745511864258566,
      "learning_rate": 5.100989870239333e-06,
      "loss": 0.7541,
      "step": 6934
    },
    {
      "epoch": 0.5088181076148464,
      "grad_norm": 1.5989095340212212,
      "learning_rate": 5.099801912872069e-06,
      "loss": 0.8184,
      "step": 6935
    },
    {
      "epoch": 0.5088914772049855,
      "grad_norm": 1.387104636964763,
      "learning_rate": 5.0986139498687435e-06,
      "loss": 0.7082,
      "step": 6936
    },
    {
      "epoch": 0.5088914772049855,
      "lm_loss": 1.3951,
      "step": 6936,
      "vm_loss": 0.051
    },
    {
      "epoch": 0.5088914772049855,
      "lm_loss": 0.289,
      "step": 6936,
      "vm_loss": 0.1108
    },
    {
      "epoch": 0.5088914772049855,
      "lm_loss": 0.2235,
      "step": 6936,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.5088914772049855,
      "lm_loss": 0.2893,
      "step": 6936,
      "vm_loss": 0.1894
    },
    {
      "epoch": 0.5088914772049855,
      "lm_loss": 0.0593,
      "step": 6936,
      "vm_loss": 0.1592
    },
    {
      "epoch": 0.5088914772049855,
      "lm_loss": 0.3891,
      "step": 6936,
      "vm_loss": 0.1781
    },
    {
      "epoch": 0.5088914772049855,
      "lm_loss": 0.281,
      "step": 6936,
      "vm_loss": 0.0732
    },
    {
      "epoch": 0.5088914772049855,
      "lm_loss": 0.6502,
      "step": 6936,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.5089648467951245,
      "grad_norm": 1.3082011779185034,
      "learning_rate": 5.097425981296445e-06,
      "loss": 0.695,
      "step": 6937
    },
    {
      "epoch": 0.5090382163852637,
      "grad_norm": 1.5683814843084976,
      "learning_rate": 5.096238007222258e-06,
      "loss": 0.8321,
      "step": 6938
    },
    {
      "epoch": 0.5091115859754028,
      "grad_norm": 1.4507071197264012,
      "learning_rate": 5.095050027713275e-06,
      "loss": 0.7923,
      "step": 6939
    },
    {
      "epoch": 0.509184955565542,
      "grad_norm": 1.4793604851319222,
      "learning_rate": 5.0938620428365795e-06,
      "loss": 0.689,
      "step": 6940
    },
    {
      "epoch": 0.5092583251556811,
      "grad_norm": 1.6798863580829664,
      "learning_rate": 5.092674052659263e-06,
      "loss": 0.7198,
      "step": 6941
    },
    {
      "epoch": 0.5093316947458202,
      "grad_norm": 1.525104700385417,
      "learning_rate": 5.091486057248414e-06,
      "loss": 0.7322,
      "step": 6942
    },
    {
      "epoch": 0.5094050643359593,
      "grad_norm": 1.427510963326217,
      "learning_rate": 5.090298056671119e-06,
      "loss": 0.8082,
      "step": 6943
    },
    {
      "epoch": 0.5094784339260985,
      "grad_norm": 1.3904654057765313,
      "learning_rate": 5.08911005099447e-06,
      "loss": 0.7282,
      "step": 6944
    },
    {
      "epoch": 0.5094784339260985,
      "lm_loss": 0.2577,
      "step": 6944,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.5094784339260985,
      "lm_loss": 1.5623,
      "step": 6944,
      "vm_loss": 0.19
    },
    {
      "epoch": 0.5094784339260985,
      "lm_loss": 0.5724,
      "step": 6944,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.5094784339260985,
      "lm_loss": 0.7765,
      "step": 6944,
      "vm_loss": 0.3537
    },
    {
      "epoch": 0.5094784339260985,
      "lm_loss": 0.7224,
      "step": 6944,
      "vm_loss": 0.121
    },
    {
      "epoch": 0.5094784339260985,
      "lm_loss": 0.0123,
      "step": 6944,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.5094784339260985,
      "lm_loss": 0.7079,
      "step": 6944,
      "vm_loss": 0.1227
    },
    {
      "epoch": 0.5094784339260985,
      "lm_loss": 0.8704,
      "step": 6944,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.5095518035162376,
      "grad_norm": 1.4486873928670643,
      "learning_rate": 5.087922040285556e-06,
      "loss": 0.8051,
      "step": 6945
    },
    {
      "epoch": 0.5096251731063768,
      "grad_norm": 1.586816723843425,
      "learning_rate": 5.086734024611466e-06,
      "loss": 0.7877,
      "step": 6946
    },
    {
      "epoch": 0.5096985426965158,
      "grad_norm": 1.4667590764309864,
      "learning_rate": 5.085546004039292e-06,
      "loss": 0.7769,
      "step": 6947
    },
    {
      "epoch": 0.509771912286655,
      "grad_norm": 1.267628939977866,
      "learning_rate": 5.084357978636123e-06,
      "loss": 0.7305,
      "step": 6948
    },
    {
      "epoch": 0.5098452818767941,
      "grad_norm": 1.3228331945770033,
      "learning_rate": 5.083169948469049e-06,
      "loss": 0.7292,
      "step": 6949
    },
    {
      "epoch": 0.5099186514669333,
      "grad_norm": 1.5080509565166509,
      "learning_rate": 5.081981913605163e-06,
      "loss": 0.7388,
      "step": 6950
    },
    {
      "epoch": 0.5099920210570724,
      "grad_norm": 1.2868623390424523,
      "learning_rate": 5.0807938741115535e-06,
      "loss": 0.6883,
      "step": 6951
    },
    {
      "epoch": 0.5100653906472115,
      "grad_norm": 0.5895883100932992,
      "learning_rate": 5.079605830055315e-06,
      "loss": 0.5079,
      "step": 6952
    },
    {
      "epoch": 0.5100653906472115,
      "lm_loss": 1.0691,
      "step": 6952,
      "vm_loss": 0.0277
    },
    {
      "epoch": 0.5100653906472115,
      "lm_loss": 0.9156,
      "step": 6952,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.5100653906472115,
      "lm_loss": 0.3549,
      "step": 6952,
      "vm_loss": 0.2253
    },
    {
      "epoch": 0.5100653906472115,
      "lm_loss": 0.5115,
      "step": 6952,
      "vm_loss": 0.0695
    },
    {
      "epoch": 0.5100653906472115,
      "lm_loss": 1.0798,
      "step": 6952,
      "vm_loss": 0.1408
    },
    {
      "epoch": 0.5100653906472115,
      "lm_loss": 0.193,
      "step": 6952,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.5100653906472115,
      "lm_loss": 0.3485,
      "step": 6952,
      "vm_loss": 0.0479
    },
    {
      "epoch": 0.5100653906472115,
      "lm_loss": 0.8245,
      "step": 6952,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.5101387602373506,
      "grad_norm": 1.4326335623820705,
      "learning_rate": 5.0784177815035395e-06,
      "loss": 0.7948,
      "step": 6953
    },
    {
      "epoch": 0.5102121298274898,
      "grad_norm": 1.4502916577176963,
      "learning_rate": 5.077229728523316e-06,
      "loss": 0.7707,
      "step": 6954
    },
    {
      "epoch": 0.5102854994176289,
      "grad_norm": 0.5354594187764957,
      "learning_rate": 5.07604167118174e-06,
      "loss": 0.5014,
      "step": 6955
    },
    {
      "epoch": 0.5103588690077681,
      "grad_norm": 0.5636327204365498,
      "learning_rate": 5.074853609545901e-06,
      "loss": 0.4783,
      "step": 6956
    },
    {
      "epoch": 0.5104322385979071,
      "grad_norm": 1.488832203444095,
      "learning_rate": 5.073665543682893e-06,
      "loss": 0.719,
      "step": 6957
    },
    {
      "epoch": 0.5105056081880462,
      "grad_norm": 1.4080067702579586,
      "learning_rate": 5.072477473659812e-06,
      "loss": 0.8359,
      "step": 6958
    },
    {
      "epoch": 0.5105789777781854,
      "grad_norm": 1.3149372664798211,
      "learning_rate": 5.071289399543745e-06,
      "loss": 0.7485,
      "step": 6959
    },
    {
      "epoch": 0.5106523473683245,
      "grad_norm": 1.4083809155399158,
      "learning_rate": 5.070101321401792e-06,
      "loss": 0.7951,
      "step": 6960
    },
    {
      "epoch": 0.5106523473683245,
      "lm_loss": 1.5831,
      "step": 6960,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.5106523473683245,
      "lm_loss": 0.9244,
      "step": 6960,
      "vm_loss": 0.1243
    },
    {
      "epoch": 0.5106523473683245,
      "lm_loss": 0.7969,
      "step": 6960,
      "vm_loss": 0.1354
    },
    {
      "epoch": 0.5106523473683245,
      "lm_loss": 0.7533,
      "step": 6960,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.5106523473683245,
      "lm_loss": 0.5641,
      "step": 6960,
      "vm_loss": 0.2354
    },
    {
      "epoch": 0.5106523473683245,
      "lm_loss": 0.6057,
      "step": 6960,
      "vm_loss": 0.1168
    },
    {
      "epoch": 0.5106523473683245,
      "lm_loss": 0.2693,
      "step": 6960,
      "vm_loss": 0.0925
    },
    {
      "epoch": 0.5106523473683245,
      "lm_loss": 0.0696,
      "step": 6960,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.5107257169584637,
      "grad_norm": 1.3538489916897816,
      "learning_rate": 5.0689132393010424e-06,
      "loss": 0.8582,
      "step": 6961
    },
    {
      "epoch": 0.5107990865486027,
      "grad_norm": 1.5511761862294762,
      "learning_rate": 5.067725153308592e-06,
      "loss": 0.8808,
      "step": 6962
    },
    {
      "epoch": 0.5108724561387419,
      "grad_norm": 1.4127182417714845,
      "learning_rate": 5.066537063491533e-06,
      "loss": 0.7459,
      "step": 6963
    },
    {
      "epoch": 0.510945825728881,
      "grad_norm": 1.4544997315413055,
      "learning_rate": 5.065348969916964e-06,
      "loss": 0.7675,
      "step": 6964
    },
    {
      "epoch": 0.5110191953190202,
      "grad_norm": 0.6393861252088376,
      "learning_rate": 5.064160872651974e-06,
      "loss": 0.508,
      "step": 6965
    },
    {
      "epoch": 0.5110925649091593,
      "grad_norm": 1.3724817251374368,
      "learning_rate": 5.062972771763662e-06,
      "loss": 0.7335,
      "step": 6966
    },
    {
      "epoch": 0.5111659344992984,
      "grad_norm": 1.5483548571845702,
      "learning_rate": 5.0617846673191204e-06,
      "loss": 0.7607,
      "step": 6967
    },
    {
      "epoch": 0.5112393040894375,
      "grad_norm": 0.5370508995533156,
      "learning_rate": 5.060596559385448e-06,
      "loss": 0.4873,
      "step": 6968
    },
    {
      "epoch": 0.5112393040894375,
      "lm_loss": 1.4387,
      "step": 6968,
      "vm_loss": 0.0594
    },
    {
      "epoch": 0.5112393040894375,
      "lm_loss": 0.2442,
      "step": 6968,
      "vm_loss": 0.1313
    },
    {
      "epoch": 0.5112393040894375,
      "lm_loss": 0.575,
      "step": 6968,
      "vm_loss": 0.087
    },
    {
      "epoch": 0.5112393040894375,
      "lm_loss": 0.1313,
      "step": 6968,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.5112393040894375,
      "lm_loss": 0.3766,
      "step": 6968,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.5112393040894375,
      "lm_loss": 1.1673,
      "step": 6968,
      "vm_loss": 0.1512
    },
    {
      "epoch": 0.5112393040894375,
      "lm_loss": 0.6708,
      "step": 6968,
      "vm_loss": 0.1335
    },
    {
      "epoch": 0.5112393040894375,
      "lm_loss": 0.1839,
      "step": 6968,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.5113126736795767,
      "grad_norm": 1.7212989334230187,
      "learning_rate": 5.059408448029737e-06,
      "loss": 0.864,
      "step": 6969
    },
    {
      "epoch": 0.5113860432697158,
      "grad_norm": 1.2659169569511828,
      "learning_rate": 5.0582203333190825e-06,
      "loss": 0.6686,
      "step": 6970
    },
    {
      "epoch": 0.5114594128598549,
      "grad_norm": 0.587139101424457,
      "learning_rate": 5.057032215320583e-06,
      "loss": 0.4984,
      "step": 6971
    },
    {
      "epoch": 0.511532782449994,
      "grad_norm": 0.5597352101843737,
      "learning_rate": 5.055844094101333e-06,
      "loss": 0.5048,
      "step": 6972
    },
    {
      "epoch": 0.5116061520401332,
      "grad_norm": 0.5736209647542256,
      "learning_rate": 5.054655969728427e-06,
      "loss": 0.5013,
      "step": 6973
    },
    {
      "epoch": 0.5116795216302723,
      "grad_norm": 0.5549615207853754,
      "learning_rate": 5.053467842268965e-06,
      "loss": 0.5126,
      "step": 6974
    },
    {
      "epoch": 0.5117528912204115,
      "grad_norm": 0.5584938238835957,
      "learning_rate": 5.052279711790042e-06,
      "loss": 0.4895,
      "step": 6975
    },
    {
      "epoch": 0.5118262608105505,
      "grad_norm": 1.5794169751175795,
      "learning_rate": 5.051091578358752e-06,
      "loss": 0.8046,
      "step": 6976
    },
    {
      "epoch": 0.5118262608105505,
      "lm_loss": 1.0762,
      "step": 6976,
      "vm_loss": 0.1248
    },
    {
      "epoch": 0.5118262608105505,
      "lm_loss": 1.2901,
      "step": 6976,
      "vm_loss": 0.1196
    },
    {
      "epoch": 0.5118262608105505,
      "lm_loss": 1.8545,
      "step": 6976,
      "vm_loss": 0.2024
    },
    {
      "epoch": 0.5118262608105505,
      "lm_loss": 0.1311,
      "step": 6976,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.5118262608105505,
      "lm_loss": 0.2678,
      "step": 6976,
      "vm_loss": 0.1375
    },
    {
      "epoch": 0.5118262608105505,
      "lm_loss": 0.836,
      "step": 6976,
      "vm_loss": 0.1219
    },
    {
      "epoch": 0.5118262608105505,
      "lm_loss": 0.0817,
      "step": 6976,
      "vm_loss": 0.1697
    },
    {
      "epoch": 0.5118262608105505,
      "lm_loss": 0.1877,
      "step": 6976,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.5118996304006896,
      "grad_norm": 1.4891268719786235,
      "learning_rate": 5.049903442042196e-06,
      "loss": 0.848,
      "step": 6977
    },
    {
      "epoch": 0.5119729999908288,
      "grad_norm": 1.4189442750098262,
      "learning_rate": 5.0487153029074665e-06,
      "loss": 0.6659,
      "step": 6978
    },
    {
      "epoch": 0.5120463695809679,
      "grad_norm": 1.6085707061255543,
      "learning_rate": 5.0475271610216655e-06,
      "loss": 0.7689,
      "step": 6979
    },
    {
      "epoch": 0.5121197391711071,
      "grad_norm": 1.6200771398635752,
      "learning_rate": 5.046339016451887e-06,
      "loss": 0.8508,
      "step": 6980
    },
    {
      "epoch": 0.5121931087612461,
      "grad_norm": 1.5632954141044424,
      "learning_rate": 5.04515086926523e-06,
      "loss": 0.804,
      "step": 6981
    },
    {
      "epoch": 0.5122664783513853,
      "grad_norm": 1.6646778117696632,
      "learning_rate": 5.043962719528794e-06,
      "loss": 0.7572,
      "step": 6982
    },
    {
      "epoch": 0.5123398479415244,
      "grad_norm": 1.4559288103939436,
      "learning_rate": 5.042774567309669e-06,
      "loss": 0.7937,
      "step": 6983
    },
    {
      "epoch": 0.5124132175316636,
      "grad_norm": 1.5167137007394513,
      "learning_rate": 5.041586412674963e-06,
      "loss": 0.7248,
      "step": 6984
    },
    {
      "epoch": 0.5124132175316636,
      "lm_loss": 0.597,
      "step": 6984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5124132175316636,
      "lm_loss": 0.365,
      "step": 6984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5124132175316636,
      "lm_loss": 0.6187,
      "step": 6984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5124132175316636,
      "lm_loss": 0.3869,
      "step": 6984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5124132175316636,
      "lm_loss": 0.4937,
      "step": 6984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5124132175316636,
      "lm_loss": 0.4694,
      "step": 6984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5124132175316636,
      "lm_loss": 0.6367,
      "step": 6984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5124132175316636,
      "lm_loss": 0.378,
      "step": 6984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5124865871218027,
      "grad_norm": 0.627524485184807,
      "learning_rate": 5.040398255691768e-06,
      "loss": 0.5013,
      "step": 6985
    },
    {
      "epoch": 0.5125599567119418,
      "grad_norm": 0.6250395290442946,
      "learning_rate": 5.039210096427181e-06,
      "loss": 0.4994,
      "step": 6986
    },
    {
      "epoch": 0.5126333263020809,
      "grad_norm": 0.5876088476455967,
      "learning_rate": 5.038021934948305e-06,
      "loss": 0.4936,
      "step": 6987
    },
    {
      "epoch": 0.5127066958922201,
      "grad_norm": 1.2851703040959417,
      "learning_rate": 5.036833771322235e-06,
      "loss": 0.7924,
      "step": 6988
    },
    {
      "epoch": 0.5127800654823592,
      "grad_norm": 1.656018518029307,
      "learning_rate": 5.0356456056160715e-06,
      "loss": 0.7761,
      "step": 6989
    },
    {
      "epoch": 0.5128534350724984,
      "grad_norm": 1.4793959361524223,
      "learning_rate": 5.034457437896913e-06,
      "loss": 0.8023,
      "step": 6990
    },
    {
      "epoch": 0.5129268046626374,
      "grad_norm": 1.6456652304320538,
      "learning_rate": 5.033269268231856e-06,
      "loss": 0.7316,
      "step": 6991
    },
    {
      "epoch": 0.5130001742527766,
      "grad_norm": 1.5921029760346852,
      "learning_rate": 5.032081096688003e-06,
      "loss": 0.8809,
      "step": 6992
    },
    {
      "epoch": 0.5130001742527766,
      "lm_loss": 0.6375,
      "step": 6992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5130001742527766,
      "lm_loss": 0.5583,
      "step": 6992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5130001742527766,
      "lm_loss": 0.5601,
      "step": 6992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5130001742527766,
      "lm_loss": 0.5143,
      "step": 6992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5130001742527766,
      "lm_loss": 0.5727,
      "step": 6992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5130001742527766,
      "lm_loss": 0.4944,
      "step": 6992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5130001742527766,
      "lm_loss": 0.5701,
      "step": 6992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5130001742527766,
      "lm_loss": 0.4918,
      "step": 6992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5130735438429157,
      "grad_norm": 0.6190839786689337,
      "learning_rate": 5.03089292333245e-06,
      "loss": 0.5185,
      "step": 6993
    },
    {
      "epoch": 0.5131469134330549,
      "grad_norm": 1.6219159563291947,
      "learning_rate": 5.029704748232297e-06,
      "loss": 0.8873,
      "step": 6994
    },
    {
      "epoch": 0.513220283023194,
      "grad_norm": 1.4326167536093093,
      "learning_rate": 5.0285165714546435e-06,
      "loss": 0.7434,
      "step": 6995
    },
    {
      "epoch": 0.513293652613333,
      "grad_norm": 1.7152490523701214,
      "learning_rate": 5.027328393066589e-06,
      "loss": 0.8634,
      "step": 6996
    },
    {
      "epoch": 0.5133670222034722,
      "grad_norm": 1.5833323018343937,
      "learning_rate": 5.026140213135231e-06,
      "loss": 0.779,
      "step": 6997
    },
    {
      "epoch": 0.5134403917936113,
      "grad_norm": 1.3672119396486926,
      "learning_rate": 5.024952031727672e-06,
      "loss": 0.7917,
      "step": 6998
    },
    {
      "epoch": 0.5135137613837505,
      "grad_norm": 1.750841122686095,
      "learning_rate": 5.023763848911009e-06,
      "loss": 0.8481,
      "step": 6999
    },
    {
      "epoch": 0.5135871309738896,
      "grad_norm": 1.5601411513762748,
      "learning_rate": 5.022575664752343e-06,
      "loss": 0.8308,
      "step": 7000
    },
    {
      "epoch": 0.5135871309738896,
      "lm_loss": 0.2729,
      "step": 7000,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.5135871309738896,
      "lm_loss": 0.1639,
      "step": 7000,
      "vm_loss": 0.1297
    },
    {
      "epoch": 0.5135871309738896,
      "lm_loss": 0.6575,
      "step": 7000,
      "vm_loss": 0.1155
    },
    {
      "epoch": 0.5135871309738896,
      "lm_loss": 0.5773,
      "step": 7000,
      "vm_loss": 0.1466
    },
    {
      "epoch": 0.5135871309738896,
      "lm_loss": 0.6956,
      "step": 7000,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.5135871309738896,
      "lm_loss": 0.8389,
      "step": 7000,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.5135871309738896,
      "lm_loss": 0.3243,
      "step": 7000,
      "vm_loss": 0.0441
    },
    {
      "epoch": 0.5135871309738896,
      "lm_loss": 0.4452,
      "step": 7000,
      "vm_loss": 0.0736
    },
    {
      "epoch": 0.5136605005640287,
      "grad_norm": 1.4069081696631909,
      "learning_rate": 5.021387479318773e-06,
      "loss": 0.7218,
      "step": 7001
    },
    {
      "epoch": 0.5137338701541678,
      "grad_norm": 1.58227575284914,
      "learning_rate": 5.020199292677399e-06,
      "loss": 0.774,
      "step": 7002
    },
    {
      "epoch": 0.513807239744307,
      "grad_norm": 1.6031455997670871,
      "learning_rate": 5.0190111048953225e-06,
      "loss": 0.8754,
      "step": 7003
    },
    {
      "epoch": 0.5138806093344461,
      "grad_norm": 1.5527901342635024,
      "learning_rate": 5.0178229160396405e-06,
      "loss": 0.7359,
      "step": 7004
    },
    {
      "epoch": 0.5139539789245853,
      "grad_norm": 0.6203387155907104,
      "learning_rate": 5.016634726177454e-06,
      "loss": 0.4932,
      "step": 7005
    },
    {
      "epoch": 0.5140273485147243,
      "grad_norm": 1.3247534430969257,
      "learning_rate": 5.015446535375864e-06,
      "loss": 0.5964,
      "step": 7006
    },
    {
      "epoch": 0.5141007181048635,
      "grad_norm": 0.5861332472817447,
      "learning_rate": 5.014258343701969e-06,
      "loss": 0.4996,
      "step": 7007
    },
    {
      "epoch": 0.5141740876950026,
      "grad_norm": 1.4409188006205476,
      "learning_rate": 5.013070151222871e-06,
      "loss": 0.6929,
      "step": 7008
    },
    {
      "epoch": 0.5141740876950026,
      "lm_loss": 1.3035,
      "step": 7008,
      "vm_loss": 0.0758
    },
    {
      "epoch": 0.5141740876950026,
      "lm_loss": 0.0944,
      "step": 7008,
      "vm_loss": 0.039
    },
    {
      "epoch": 0.5141740876950026,
      "lm_loss": 0.0707,
      "step": 7008,
      "vm_loss": 0.0995
    },
    {
      "epoch": 0.5141740876950026,
      "lm_loss": 0.1885,
      "step": 7008,
      "vm_loss": 0.1623
    },
    {
      "epoch": 0.5141740876950026,
      "lm_loss": 0.3732,
      "step": 7008,
      "vm_loss": 0.087
    },
    {
      "epoch": 0.5141740876950026,
      "lm_loss": 1.2821,
      "step": 7008,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.5141740876950026,
      "lm_loss": 0.7627,
      "step": 7008,
      "vm_loss": 0.2086
    },
    {
      "epoch": 0.5141740876950026,
      "lm_loss": 0.4771,
      "step": 7008,
      "vm_loss": 0.1263
    },
    {
      "epoch": 0.5142474572851418,
      "grad_norm": 1.4511110681923098,
      "learning_rate": 5.0118819580056686e-06,
      "loss": 0.6701,
      "step": 7009
    },
    {
      "epoch": 0.5143208268752809,
      "grad_norm": 0.5854399855018029,
      "learning_rate": 5.010693764117462e-06,
      "loss": 0.5049,
      "step": 7010
    },
    {
      "epoch": 0.51439419646542,
      "grad_norm": 1.556964658131642,
      "learning_rate": 5.0095055696253545e-06,
      "loss": 0.7415,
      "step": 7011
    },
    {
      "epoch": 0.5144675660555591,
      "grad_norm": 1.4293132210905974,
      "learning_rate": 5.008317374596442e-06,
      "loss": 0.6905,
      "step": 7012
    },
    {
      "epoch": 0.5145409356456982,
      "grad_norm": 1.4039676821488491,
      "learning_rate": 5.007129179097827e-06,
      "loss": 0.6776,
      "step": 7013
    },
    {
      "epoch": 0.5146143052358374,
      "grad_norm": 1.561161035316636,
      "learning_rate": 5.00594098319661e-06,
      "loss": 0.753,
      "step": 7014
    },
    {
      "epoch": 0.5146876748259765,
      "grad_norm": 1.4271879772076064,
      "learning_rate": 5.00475278695989e-06,
      "loss": 0.6965,
      "step": 7015
    },
    {
      "epoch": 0.5147610444161156,
      "grad_norm": 1.4831099645978434,
      "learning_rate": 5.003564590454769e-06,
      "loss": 0.8827,
      "step": 7016
    },
    {
      "epoch": 0.5147610444161156,
      "lm_loss": 1.626,
      "step": 7016,
      "vm_loss": 0.0416
    },
    {
      "epoch": 0.5147610444161156,
      "lm_loss": 1.2439,
      "step": 7016,
      "vm_loss": 0.2251
    },
    {
      "epoch": 0.5147610444161156,
      "lm_loss": 0.0568,
      "step": 7016,
      "vm_loss": 0.1967
    },
    {
      "epoch": 0.5147610444161156,
      "lm_loss": 0.0483,
      "step": 7016,
      "vm_loss": 0.1101
    },
    {
      "epoch": 0.5147610444161156,
      "lm_loss": 0.7008,
      "step": 7016,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.5147610444161156,
      "lm_loss": 0.7739,
      "step": 7016,
      "vm_loss": 0.1578
    },
    {
      "epoch": 0.5147610444161156,
      "lm_loss": 0.982,
      "step": 7016,
      "vm_loss": 0.1152
    },
    {
      "epoch": 0.5147610444161156,
      "lm_loss": 0.2618,
      "step": 7016,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.5148344140062547,
      "grad_norm": 1.3205269086016056,
      "learning_rate": 5.002376393748347e-06,
      "loss": 0.8045,
      "step": 7017
    },
    {
      "epoch": 0.5149077835963939,
      "grad_norm": 1.4687299159655052,
      "learning_rate": 5.001188196907723e-06,
      "loss": 0.7946,
      "step": 7018
    },
    {
      "epoch": 0.514981153186533,
      "grad_norm": 1.3850118334973514,
      "learning_rate": 5e-06,
      "loss": 0.7475,
      "step": 7019
    },
    {
      "epoch": 0.5150545227766722,
      "grad_norm": 1.6014962359742335,
      "learning_rate": 4.9988118030922775e-06,
      "loss": 0.7937,
      "step": 7020
    },
    {
      "epoch": 0.5151278923668112,
      "grad_norm": 1.558196843175089,
      "learning_rate": 4.997623606251654e-06,
      "loss": 0.7744,
      "step": 7021
    },
    {
      "epoch": 0.5152012619569504,
      "grad_norm": 1.6218540568410702,
      "learning_rate": 4.996435409545232e-06,
      "loss": 0.774,
      "step": 7022
    },
    {
      "epoch": 0.5152746315470895,
      "grad_norm": 1.5104615390108487,
      "learning_rate": 4.995247213040111e-06,
      "loss": 0.7694,
      "step": 7023
    },
    {
      "epoch": 0.5153480011372287,
      "grad_norm": 1.4896130478525877,
      "learning_rate": 4.994059016803392e-06,
      "loss": 0.8161,
      "step": 7024
    },
    {
      "epoch": 0.5153480011372287,
      "lm_loss": 2.0428,
      "step": 7024,
      "vm_loss": 0.0376
    },
    {
      "epoch": 0.5153480011372287,
      "lm_loss": 1.1638,
      "step": 7024,
      "vm_loss": 0.1156
    },
    {
      "epoch": 0.5153480011372287,
      "lm_loss": 0.7564,
      "step": 7024,
      "vm_loss": 0.2401
    },
    {
      "epoch": 0.5153480011372287,
      "lm_loss": 0.0793,
      "step": 7024,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.5153480011372287,
      "lm_loss": 0.7692,
      "step": 7024,
      "vm_loss": 0.1866
    },
    {
      "epoch": 0.5153480011372287,
      "lm_loss": 0.3514,
      "step": 7024,
      "vm_loss": 0.1814
    },
    {
      "epoch": 0.5153480011372287,
      "lm_loss": 1.1343,
      "step": 7024,
      "vm_loss": 0.1012
    },
    {
      "epoch": 0.5153480011372287,
      "lm_loss": 0.78,
      "step": 7024,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.5154213707273678,
      "grad_norm": 1.4444570584830496,
      "learning_rate": 4.9928708209021745e-06,
      "loss": 0.8363,
      "step": 7025
    },
    {
      "epoch": 0.5154947403175069,
      "grad_norm": 1.5456541927358782,
      "learning_rate": 4.991682625403559e-06,
      "loss": 0.841,
      "step": 7026
    },
    {
      "epoch": 0.515568109907646,
      "grad_norm": 0.648127264814638,
      "learning_rate": 4.990494430374647e-06,
      "loss": 0.4945,
      "step": 7027
    },
    {
      "epoch": 0.5156414794977852,
      "grad_norm": 1.4833761102095246,
      "learning_rate": 4.9893062358825385e-06,
      "loss": 0.8372,
      "step": 7028
    },
    {
      "epoch": 0.5157148490879243,
      "grad_norm": 1.3867030529751267,
      "learning_rate": 4.988118041994332e-06,
      "loss": 0.7544,
      "step": 7029
    },
    {
      "epoch": 0.5157882186780635,
      "grad_norm": 0.6027733551112661,
      "learning_rate": 4.98692984877713e-06,
      "loss": 0.5063,
      "step": 7030
    },
    {
      "epoch": 0.5158615882682025,
      "grad_norm": 1.6240289056635413,
      "learning_rate": 4.985741656298033e-06,
      "loss": 0.7296,
      "step": 7031
    },
    {
      "epoch": 0.5159349578583416,
      "grad_norm": 1.6959645994373422,
      "learning_rate": 4.984553464624137e-06,
      "loss": 0.7963,
      "step": 7032
    },
    {
      "epoch": 0.5159349578583416,
      "lm_loss": 0.702,
      "step": 7032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5159349578583416,
      "lm_loss": 0.337,
      "step": 7032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5159349578583416,
      "lm_loss": 0.5004,
      "step": 7032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5159349578583416,
      "lm_loss": 0.3869,
      "step": 7032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5159349578583416,
      "lm_loss": 0.5261,
      "step": 7032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5159349578583416,
      "lm_loss": 0.3588,
      "step": 7032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5159349578583416,
      "lm_loss": 0.6039,
      "step": 7032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5159349578583416,
      "lm_loss": 0.475,
      "step": 7032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5160083274484808,
      "grad_norm": 0.5542121164969732,
      "learning_rate": 4.983365273822547e-06,
      "loss": 0.4974,
      "step": 7033
    },
    {
      "epoch": 0.51608169703862,
      "grad_norm": 1.4285464314729366,
      "learning_rate": 4.98217708396036e-06,
      "loss": 0.7613,
      "step": 7034
    },
    {
      "epoch": 0.5161550666287591,
      "grad_norm": 0.5834306999103281,
      "learning_rate": 4.980988895104679e-06,
      "loss": 0.5089,
      "step": 7035
    },
    {
      "epoch": 0.5162284362188981,
      "grad_norm": 1.4624396788867275,
      "learning_rate": 4.979800707322602e-06,
      "loss": 0.7085,
      "step": 7036
    },
    {
      "epoch": 0.5163018058090373,
      "grad_norm": 1.4654458265054449,
      "learning_rate": 4.978612520681228e-06,
      "loss": 0.7846,
      "step": 7037
    },
    {
      "epoch": 0.5163751753991764,
      "grad_norm": 1.3531587425988725,
      "learning_rate": 4.977424335247658e-06,
      "loss": 0.7925,
      "step": 7038
    },
    {
      "epoch": 0.5164485449893156,
      "grad_norm": 1.748623605908743,
      "learning_rate": 4.976236151088993e-06,
      "loss": 0.8117,
      "step": 7039
    },
    {
      "epoch": 0.5165219145794547,
      "grad_norm": 1.4471299723023432,
      "learning_rate": 4.97504796827233e-06,
      "loss": 0.7225,
      "step": 7040
    },
    {
      "epoch": 0.5165219145794547,
      "lm_loss": 1.1538,
      "step": 7040,
      "vm_loss": 0.0549
    },
    {
      "epoch": 0.5165219145794547,
      "lm_loss": 0.9862,
      "step": 7040,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.5165219145794547,
      "lm_loss": 1.0244,
      "step": 7040,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.5165219145794547,
      "lm_loss": 0.5565,
      "step": 7040,
      "vm_loss": 0.169
    },
    {
      "epoch": 0.5165219145794547,
      "lm_loss": 0.6825,
      "step": 7040,
      "vm_loss": 0.1218
    },
    {
      "epoch": 0.5165219145794547,
      "lm_loss": 0.8245,
      "step": 7040,
      "vm_loss": 0.0524
    },
    {
      "epoch": 0.5165219145794547,
      "lm_loss": 0.2889,
      "step": 7040,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.5165219145794547,
      "lm_loss": 0.9734,
      "step": 7040,
      "vm_loss": 0.0497
    },
    {
      "epoch": 0.5165952841695938,
      "grad_norm": 1.4216642419254393,
      "learning_rate": 4.97385978686477e-06,
      "loss": 0.8243,
      "step": 7041
    },
    {
      "epoch": 0.5166686537597329,
      "grad_norm": 1.6287421505827235,
      "learning_rate": 4.972671606933412e-06,
      "loss": 0.6791,
      "step": 7042
    },
    {
      "epoch": 0.5167420233498721,
      "grad_norm": 0.6200466267173184,
      "learning_rate": 4.971483428545358e-06,
      "loss": 0.5035,
      "step": 7043
    },
    {
      "epoch": 0.5168153929400112,
      "grad_norm": 0.6015028908742833,
      "learning_rate": 4.970295251767705e-06,
      "loss": 0.4946,
      "step": 7044
    },
    {
      "epoch": 0.5168887625301504,
      "grad_norm": 1.568570195740546,
      "learning_rate": 4.969107076667551e-06,
      "loss": 0.8822,
      "step": 7045
    },
    {
      "epoch": 0.5169621321202894,
      "grad_norm": 1.5393203838912939,
      "learning_rate": 4.967918903311999e-06,
      "loss": 0.8319,
      "step": 7046
    },
    {
      "epoch": 0.5170355017104286,
      "grad_norm": 1.469628527417006,
      "learning_rate": 4.966730731768145e-06,
      "loss": 0.7622,
      "step": 7047
    },
    {
      "epoch": 0.5171088713005677,
      "grad_norm": 1.478371200473877,
      "learning_rate": 4.965542562103088e-06,
      "loss": 0.7182,
      "step": 7048
    },
    {
      "epoch": 0.5171088713005677,
      "lm_loss": 0.8725,
      "step": 7048,
      "vm_loss": 0.0574
    },
    {
      "epoch": 0.5171088713005677,
      "lm_loss": 0.8436,
      "step": 7048,
      "vm_loss": 0.0608
    },
    {
      "epoch": 0.5171088713005677,
      "lm_loss": 0.6016,
      "step": 7048,
      "vm_loss": 0.1412
    },
    {
      "epoch": 0.5171088713005677,
      "lm_loss": 0.6685,
      "step": 7048,
      "vm_loss": 0.1701
    },
    {
      "epoch": 0.5171088713005677,
      "lm_loss": 1.1387,
      "step": 7048,
      "vm_loss": 0.1352
    },
    {
      "epoch": 0.5171088713005677,
      "lm_loss": 0.4762,
      "step": 7048,
      "vm_loss": 0.1871
    },
    {
      "epoch": 0.5171088713005677,
      "lm_loss": 0.1537,
      "step": 7048,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.5171088713005677,
      "lm_loss": 0.3162,
      "step": 7048,
      "vm_loss": 0.1196
    },
    {
      "epoch": 0.5171822408907069,
      "grad_norm": 1.5737728779291116,
      "learning_rate": 4.964354394383929e-06,
      "loss": 0.7856,
      "step": 7049
    },
    {
      "epoch": 0.517255610480846,
      "grad_norm": 1.3429312001317473,
      "learning_rate": 4.963166228677766e-06,
      "loss": 0.7341,
      "step": 7050
    },
    {
      "epoch": 0.517328980070985,
      "grad_norm": 1.3855486996008677,
      "learning_rate": 4.961978065051697e-06,
      "loss": 0.7486,
      "step": 7051
    },
    {
      "epoch": 0.5174023496611242,
      "grad_norm": 1.4725107152518417,
      "learning_rate": 4.96078990357282e-06,
      "loss": 0.8029,
      "step": 7052
    },
    {
      "epoch": 0.5174757192512633,
      "grad_norm": 1.4847671419057265,
      "learning_rate": 4.959601744308235e-06,
      "loss": 0.6852,
      "step": 7053
    },
    {
      "epoch": 0.5175490888414025,
      "grad_norm": 1.539518235737695,
      "learning_rate": 4.95841358732504e-06,
      "loss": 0.7545,
      "step": 7054
    },
    {
      "epoch": 0.5176224584315415,
      "grad_norm": 1.5819593995177879,
      "learning_rate": 4.9572254326903315e-06,
      "loss": 0.6894,
      "step": 7055
    },
    {
      "epoch": 0.5176958280216807,
      "grad_norm": 1.547007014049594,
      "learning_rate": 4.956037280471209e-06,
      "loss": 0.8327,
      "step": 7056
    },
    {
      "epoch": 0.5176958280216807,
      "lm_loss": 0.4242,
      "step": 7056,
      "vm_loss": 0.0424
    },
    {
      "epoch": 0.5176958280216807,
      "lm_loss": 0.847,
      "step": 7056,
      "vm_loss": 0.1467
    },
    {
      "epoch": 0.5176958280216807,
      "lm_loss": 0.3882,
      "step": 7056,
      "vm_loss": 0.1197
    },
    {
      "epoch": 0.5176958280216807,
      "lm_loss": 0.6901,
      "step": 7056,
      "vm_loss": 0.1268
    },
    {
      "epoch": 0.5176958280216807,
      "lm_loss": 0.2719,
      "step": 7056,
      "vm_loss": 0.1591
    },
    {
      "epoch": 0.5176958280216807,
      "lm_loss": 0.581,
      "step": 7056,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.5176958280216807,
      "lm_loss": 0.0352,
      "step": 7056,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.5176958280216807,
      "lm_loss": 0.887,
      "step": 7056,
      "vm_loss": 0.1388
    },
    {
      "epoch": 0.5177691976118198,
      "grad_norm": 1.5056221692147078,
      "learning_rate": 4.95484913073477e-06,
      "loss": 0.6926,
      "step": 7057
    },
    {
      "epoch": 0.517842567201959,
      "grad_norm": 1.578196746521194,
      "learning_rate": 4.9536609835481135e-06,
      "loss": 0.7771,
      "step": 7058
    },
    {
      "epoch": 0.5179159367920981,
      "grad_norm": 1.4041789487859624,
      "learning_rate": 4.952472838978335e-06,
      "loss": 0.702,
      "step": 7059
    },
    {
      "epoch": 0.5179893063822372,
      "grad_norm": 1.4696036575288136,
      "learning_rate": 4.9512846970925335e-06,
      "loss": 0.8856,
      "step": 7060
    },
    {
      "epoch": 0.5180626759723763,
      "grad_norm": 1.4059544647746693,
      "learning_rate": 4.950096557957805e-06,
      "loss": 0.7384,
      "step": 7061
    },
    {
      "epoch": 0.5181360455625155,
      "grad_norm": 1.6691492896814415,
      "learning_rate": 4.948908421641249e-06,
      "loss": 0.8456,
      "step": 7062
    },
    {
      "epoch": 0.5182094151526546,
      "grad_norm": 1.4964636646013219,
      "learning_rate": 4.94772028820996e-06,
      "loss": 0.7955,
      "step": 7063
    },
    {
      "epoch": 0.5182827847427938,
      "grad_norm": 1.552949418670589,
      "learning_rate": 4.946532157731036e-06,
      "loss": 0.7794,
      "step": 7064
    },
    {
      "epoch": 0.5182827847427938,
      "lm_loss": 0.5256,
      "step": 7064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5182827847427938,
      "lm_loss": 0.4066,
      "step": 7064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5182827847427938,
      "lm_loss": 0.4639,
      "step": 7064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5182827847427938,
      "lm_loss": 0.5379,
      "step": 7064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5182827847427938,
      "lm_loss": 0.435,
      "step": 7064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5182827847427938,
      "lm_loss": 0.3522,
      "step": 7064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5182827847427938,
      "lm_loss": 0.4506,
      "step": 7064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5182827847427938,
      "lm_loss": 0.6435,
      "step": 7064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5183561543329328,
      "grad_norm": 0.7756414197853606,
      "learning_rate": 4.945344030271573e-06,
      "loss": 0.4755,
      "step": 7065
    },
    {
      "epoch": 0.518429523923072,
      "grad_norm": 1.5328581925404232,
      "learning_rate": 4.9441559058986686e-06,
      "loss": 0.7092,
      "step": 7066
    },
    {
      "epoch": 0.5185028935132111,
      "grad_norm": 0.6312029197362923,
      "learning_rate": 4.942967784679417e-06,
      "loss": 0.5097,
      "step": 7067
    },
    {
      "epoch": 0.5185762631033503,
      "grad_norm": 1.5596232225265894,
      "learning_rate": 4.9417796666809175e-06,
      "loss": 0.7888,
      "step": 7068
    },
    {
      "epoch": 0.5186496326934894,
      "grad_norm": 1.555103218494338,
      "learning_rate": 4.940591551970264e-06,
      "loss": 0.7708,
      "step": 7069
    },
    {
      "epoch": 0.5187230022836284,
      "grad_norm": 1.3139510367518135,
      "learning_rate": 4.939403440614552e-06,
      "loss": 0.7587,
      "step": 7070
    },
    {
      "epoch": 0.5187963718737676,
      "grad_norm": 0.6245041105196363,
      "learning_rate": 4.938215332680879e-06,
      "loss": 0.4896,
      "step": 7071
    },
    {
      "epoch": 0.5188697414639067,
      "grad_norm": 0.6522974482728905,
      "learning_rate": 4.937027228236338e-06,
      "loss": 0.4934,
      "step": 7072
    },
    {
      "epoch": 0.5188697414639067,
      "lm_loss": 1.1904,
      "step": 7072,
      "vm_loss": 0.0757
    },
    {
      "epoch": 0.5188697414639067,
      "lm_loss": 0.1275,
      "step": 7072,
      "vm_loss": 0.0581
    },
    {
      "epoch": 0.5188697414639067,
      "lm_loss": 0.7131,
      "step": 7072,
      "vm_loss": 0.1392
    },
    {
      "epoch": 0.5188697414639067,
      "lm_loss": 0.652,
      "step": 7072,
      "vm_loss": 0.102
    },
    {
      "epoch": 0.5188697414639067,
      "lm_loss": 0.8746,
      "step": 7072,
      "vm_loss": 0.2721
    },
    {
      "epoch": 0.5188697414639067,
      "lm_loss": 0.2679,
      "step": 7072,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.5188697414639067,
      "lm_loss": 0.9833,
      "step": 7072,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.5188697414639067,
      "lm_loss": 0.9247,
      "step": 7072,
      "vm_loss": 0.1243
    },
    {
      "epoch": 0.5189431110540459,
      "grad_norm": 1.5012634677368388,
      "learning_rate": 4.935839127348027e-06,
      "loss": 0.69,
      "step": 7073
    },
    {
      "epoch": 0.519016480644185,
      "grad_norm": 0.5810035919388974,
      "learning_rate": 4.934651030083038e-06,
      "loss": 0.4905,
      "step": 7074
    },
    {
      "epoch": 0.5190898502343241,
      "grad_norm": 0.5673330991610926,
      "learning_rate": 4.933462936508467e-06,
      "loss": 0.4978,
      "step": 7075
    },
    {
      "epoch": 0.5191632198244632,
      "grad_norm": 1.5587405653702622,
      "learning_rate": 4.932274846691409e-06,
      "loss": 0.835,
      "step": 7076
    },
    {
      "epoch": 0.5192365894146024,
      "grad_norm": 1.5029588045569409,
      "learning_rate": 4.931086760698959e-06,
      "loss": 0.7432,
      "step": 7077
    },
    {
      "epoch": 0.5193099590047415,
      "grad_norm": 0.6239754627942385,
      "learning_rate": 4.929898678598209e-06,
      "loss": 0.5017,
      "step": 7078
    },
    {
      "epoch": 0.5193833285948807,
      "grad_norm": 1.502637996258077,
      "learning_rate": 4.928710600456255e-06,
      "loss": 0.7729,
      "step": 7079
    },
    {
      "epoch": 0.5194566981850197,
      "grad_norm": 0.6233487175820376,
      "learning_rate": 4.927522526340189e-06,
      "loss": 0.4995,
      "step": 7080
    },
    {
      "epoch": 0.5194566981850197,
      "lm_loss": 0.4742,
      "step": 7080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5194566981850197,
      "lm_loss": 0.4227,
      "step": 7080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5194566981850197,
      "lm_loss": 0.3981,
      "step": 7080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5194566981850197,
      "lm_loss": 0.5348,
      "step": 7080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5194566981850197,
      "lm_loss": 0.5255,
      "step": 7080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5194566981850197,
      "lm_loss": 0.6616,
      "step": 7080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5194566981850197,
      "lm_loss": 0.4791,
      "step": 7080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5194566981850197,
      "lm_loss": 0.5688,
      "step": 7080,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5195300677751589,
      "grad_norm": 0.59691197464915,
      "learning_rate": 4.9263344563171065e-06,
      "loss": 0.515,
      "step": 7081
    },
    {
      "epoch": 0.519603437365298,
      "grad_norm": 1.5946317352180266,
      "learning_rate": 4.9251463904541e-06,
      "loss": 0.7538,
      "step": 7082
    },
    {
      "epoch": 0.5196768069554372,
      "grad_norm": 1.5223115603674353,
      "learning_rate": 4.923958328818261e-06,
      "loss": 0.8275,
      "step": 7083
    },
    {
      "epoch": 0.5197501765455763,
      "grad_norm": 1.4973478527518307,
      "learning_rate": 4.922770271476685e-06,
      "loss": 0.7071,
      "step": 7084
    },
    {
      "epoch": 0.5198235461357154,
      "grad_norm": 1.4490859464054353,
      "learning_rate": 4.921582218496461e-06,
      "loss": 0.7513,
      "step": 7085
    },
    {
      "epoch": 0.5198969157258545,
      "grad_norm": 0.6038587385210137,
      "learning_rate": 4.920394169944685e-06,
      "loss": 0.5083,
      "step": 7086
    },
    {
      "epoch": 0.5199702853159937,
      "grad_norm": 0.6458464184056385,
      "learning_rate": 4.9192061258884464e-06,
      "loss": 0.5096,
      "step": 7087
    },
    {
      "epoch": 0.5200436549061328,
      "grad_norm": 1.5968683747233756,
      "learning_rate": 4.9180180863948376e-06,
      "loss": 0.7825,
      "step": 7088
    },
    {
      "epoch": 0.5200436549061328,
      "lm_loss": 0.4962,
      "step": 7088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5200436549061328,
      "lm_loss": 0.4655,
      "step": 7088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5200436549061328,
      "lm_loss": 0.3649,
      "step": 7088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5200436549061328,
      "lm_loss": 0.4744,
      "step": 7088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5200436549061328,
      "lm_loss": 0.5557,
      "step": 7088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5200436549061328,
      "lm_loss": 0.386,
      "step": 7088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5200436549061328,
      "lm_loss": 0.5831,
      "step": 7088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5200436549061328,
      "lm_loss": 0.5896,
      "step": 7088,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.520117024496272,
      "grad_norm": 0.5700311920658893,
      "learning_rate": 4.9168300515309515e-06,
      "loss": 0.4883,
      "step": 7089
    },
    {
      "epoch": 0.520190394086411,
      "grad_norm": 1.5459344565314286,
      "learning_rate": 4.915642021363879e-06,
      "loss": 0.7436,
      "step": 7090
    },
    {
      "epoch": 0.5202637636765501,
      "grad_norm": 1.588345671054082,
      "learning_rate": 4.914453995960708e-06,
      "loss": 0.7217,
      "step": 7091
    },
    {
      "epoch": 0.5203371332666893,
      "grad_norm": 1.6078185690154119,
      "learning_rate": 4.913265975388534e-06,
      "loss": 0.8255,
      "step": 7092
    },
    {
      "epoch": 0.5204105028568284,
      "grad_norm": 1.356979054637838,
      "learning_rate": 4.912077959714445e-06,
      "loss": 0.7204,
      "step": 7093
    },
    {
      "epoch": 0.5204838724469676,
      "grad_norm": 0.6040337001034257,
      "learning_rate": 4.91088994900553e-06,
      "loss": 0.4972,
      "step": 7094
    },
    {
      "epoch": 0.5205572420371066,
      "grad_norm": 1.605872038112786,
      "learning_rate": 4.909701943328883e-06,
      "loss": 0.7572,
      "step": 7095
    },
    {
      "epoch": 0.5206306116272458,
      "grad_norm": 1.5192149966260737,
      "learning_rate": 4.908513942751589e-06,
      "loss": 0.7841,
      "step": 7096
    },
    {
      "epoch": 0.5206306116272458,
      "lm_loss": 0.3667,
      "step": 7096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5206306116272458,
      "lm_loss": 0.3448,
      "step": 7096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5206306116272458,
      "lm_loss": 0.4528,
      "step": 7096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5206306116272458,
      "lm_loss": 0.5742,
      "step": 7096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5206306116272458,
      "lm_loss": 0.4965,
      "step": 7096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5206306116272458,
      "lm_loss": 0.4111,
      "step": 7096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5206306116272458,
      "lm_loss": 0.5387,
      "step": 7096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5206306116272458,
      "lm_loss": 0.5206,
      "step": 7096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5207039812173849,
      "grad_norm": 0.6143564240172247,
      "learning_rate": 4.907325947340739e-06,
      "loss": 0.4915,
      "step": 7097
    },
    {
      "epoch": 0.5207773508075241,
      "grad_norm": 0.571246444178335,
      "learning_rate": 4.906137957163422e-06,
      "loss": 0.4916,
      "step": 7098
    },
    {
      "epoch": 0.5208507203976632,
      "grad_norm": 1.5641034284510795,
      "learning_rate": 4.904949972286728e-06,
      "loss": 0.7165,
      "step": 7099
    },
    {
      "epoch": 0.5209240899878023,
      "grad_norm": 1.5165390643913124,
      "learning_rate": 4.9037619927777445e-06,
      "loss": 0.807,
      "step": 7100
    },
    {
      "epoch": 0.5209974595779414,
      "grad_norm": 1.7643869798093483,
      "learning_rate": 4.902574018703557e-06,
      "loss": 0.7532,
      "step": 7101
    },
    {
      "epoch": 0.5210708291680806,
      "grad_norm": 0.5708758723738245,
      "learning_rate": 4.901386050131259e-06,
      "loss": 0.5047,
      "step": 7102
    },
    {
      "epoch": 0.5211441987582197,
      "grad_norm": 1.5131949760291452,
      "learning_rate": 4.900198087127933e-06,
      "loss": 0.7353,
      "step": 7103
    },
    {
      "epoch": 0.5212175683483589,
      "grad_norm": 1.57181261899714,
      "learning_rate": 4.899010129760669e-06,
      "loss": 0.7641,
      "step": 7104
    },
    {
      "epoch": 0.5212175683483589,
      "lm_loss": 1.3546,
      "step": 7104,
      "vm_loss": 0.0784
    },
    {
      "epoch": 0.5212175683483589,
      "lm_loss": 0.1445,
      "step": 7104,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.5212175683483589,
      "lm_loss": 0.8003,
      "step": 7104,
      "vm_loss": 0.0601
    },
    {
      "epoch": 0.5212175683483589,
      "lm_loss": 0.1837,
      "step": 7104,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.5212175683483589,
      "lm_loss": 0.7833,
      "step": 7104,
      "vm_loss": 0.1642
    },
    {
      "epoch": 0.5212175683483589,
      "lm_loss": 0.0188,
      "step": 7104,
      "vm_loss": 0.1387
    },
    {
      "epoch": 0.5212175683483589,
      "lm_loss": 0.2207,
      "step": 7104,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.5212175683483589,
      "lm_loss": 0.616,
      "step": 7104,
      "vm_loss": 0.2616
    },
    {
      "epoch": 0.5212909379384979,
      "grad_norm": 1.4412373627036021,
      "learning_rate": 4.897822178096554e-06,
      "loss": 0.6594,
      "step": 7105
    },
    {
      "epoch": 0.521364307528637,
      "grad_norm": 1.5707654032551392,
      "learning_rate": 4.896634232202672e-06,
      "loss": 0.8038,
      "step": 7106
    },
    {
      "epoch": 0.5214376771187762,
      "grad_norm": 1.504754454099281,
      "learning_rate": 4.895446292146112e-06,
      "loss": 0.7496,
      "step": 7107
    },
    {
      "epoch": 0.5215110467089153,
      "grad_norm": 1.624255343218917,
      "learning_rate": 4.894258357993959e-06,
      "loss": 0.7418,
      "step": 7108
    },
    {
      "epoch": 0.5215844162990545,
      "grad_norm": 1.5791949073724778,
      "learning_rate": 4.8930704298132965e-06,
      "loss": 0.8269,
      "step": 7109
    },
    {
      "epoch": 0.5216577858891935,
      "grad_norm": 1.4999613056288965,
      "learning_rate": 4.891882507671213e-06,
      "loss": 0.7081,
      "step": 7110
    },
    {
      "epoch": 0.5217311554793327,
      "grad_norm": 0.6173717743005215,
      "learning_rate": 4.890694591634791e-06,
      "loss": 0.4872,
      "step": 7111
    },
    {
      "epoch": 0.5218045250694718,
      "grad_norm": 1.9389214112633084,
      "learning_rate": 4.889506681771118e-06,
      "loss": 0.8311,
      "step": 7112
    },
    {
      "epoch": 0.5218045250694718,
      "lm_loss": 0.2931,
      "step": 7112,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.5218045250694718,
      "lm_loss": 1.3193,
      "step": 7112,
      "vm_loss": 0.1166
    },
    {
      "epoch": 0.5218045250694718,
      "lm_loss": 0.7971,
      "step": 7112,
      "vm_loss": 0.1668
    },
    {
      "epoch": 0.5218045250694718,
      "lm_loss": 0.4936,
      "step": 7112,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.5218045250694718,
      "lm_loss": 1.0218,
      "step": 7112,
      "vm_loss": 0.1489
    },
    {
      "epoch": 0.5218045250694718,
      "lm_loss": 0.4614,
      "step": 7112,
      "vm_loss": 0.0791
    },
    {
      "epoch": 0.5218045250694718,
      "lm_loss": 0.0088,
      "step": 7112,
      "vm_loss": 0.1055
    },
    {
      "epoch": 0.5218045250694718,
      "lm_loss": 0.5361,
      "step": 7112,
      "vm_loss": 0.1955
    },
    {
      "epoch": 0.521877894659611,
      "grad_norm": 1.5975262632685554,
      "learning_rate": 4.888318778147276e-06,
      "loss": 0.7869,
      "step": 7113
    },
    {
      "epoch": 0.5219512642497501,
      "grad_norm": 1.932670752007397,
      "learning_rate": 4.887130880830348e-06,
      "loss": 0.7696,
      "step": 7114
    },
    {
      "epoch": 0.5220246338398892,
      "grad_norm": 1.4332632555477536,
      "learning_rate": 4.885942989887419e-06,
      "loss": 0.8177,
      "step": 7115
    },
    {
      "epoch": 0.5220980034300283,
      "grad_norm": 0.5727618528620753,
      "learning_rate": 4.884755105385572e-06,
      "loss": 0.4814,
      "step": 7116
    },
    {
      "epoch": 0.5221713730201675,
      "grad_norm": 0.5763735548849297,
      "learning_rate": 4.883567227391889e-06,
      "loss": 0.4951,
      "step": 7117
    },
    {
      "epoch": 0.5222447426103066,
      "grad_norm": 1.7124082713341136,
      "learning_rate": 4.882379355973454e-06,
      "loss": 0.726,
      "step": 7118
    },
    {
      "epoch": 0.5223181122004458,
      "grad_norm": 1.4134653001162854,
      "learning_rate": 4.881191491197348e-06,
      "loss": 0.7594,
      "step": 7119
    },
    {
      "epoch": 0.5223914817905848,
      "grad_norm": 1.4656156723998546,
      "learning_rate": 4.880003633130652e-06,
      "loss": 0.6979,
      "step": 7120
    },
    {
      "epoch": 0.5223914817905848,
      "lm_loss": 1.4649,
      "step": 7120,
      "vm_loss": 0.0241
    },
    {
      "epoch": 0.5223914817905848,
      "lm_loss": 0.5504,
      "step": 7120,
      "vm_loss": 0.0733
    },
    {
      "epoch": 0.5223914817905848,
      "lm_loss": 0.9751,
      "step": 7120,
      "vm_loss": 0.1609
    },
    {
      "epoch": 0.5223914817905848,
      "lm_loss": 0.9813,
      "step": 7120,
      "vm_loss": 0.1545
    },
    {
      "epoch": 0.5223914817905848,
      "lm_loss": 0.3131,
      "step": 7120,
      "vm_loss": 0.2319
    },
    {
      "epoch": 0.5223914817905848,
      "lm_loss": 0.6057,
      "step": 7120,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.5223914817905848,
      "lm_loss": 0.0221,
      "step": 7120,
      "vm_loss": 0.112
    },
    {
      "epoch": 0.5223914817905848,
      "lm_loss": 0.4391,
      "step": 7120,
      "vm_loss": 0.1581
    },
    {
      "epoch": 0.522464851380724,
      "grad_norm": 1.6153137314880244,
      "learning_rate": 4.878815781840449e-06,
      "loss": 0.7695,
      "step": 7121
    },
    {
      "epoch": 0.5225382209708631,
      "grad_norm": 1.4046695507258184,
      "learning_rate": 4.8776279373938175e-06,
      "loss": 0.7598,
      "step": 7122
    },
    {
      "epoch": 0.5226115905610023,
      "grad_norm": 1.5351506040568041,
      "learning_rate": 4.8764400998578406e-06,
      "loss": 0.8346,
      "step": 7123
    },
    {
      "epoch": 0.5226849601511414,
      "grad_norm": 1.5919628340685947,
      "learning_rate": 4.875252269299597e-06,
      "loss": 0.7991,
      "step": 7124
    },
    {
      "epoch": 0.5227583297412804,
      "grad_norm": 0.5920338853381678,
      "learning_rate": 4.874064445786166e-06,
      "loss": 0.4753,
      "step": 7125
    },
    {
      "epoch": 0.5228316993314196,
      "grad_norm": 1.5876387756483865,
      "learning_rate": 4.8728766293846286e-06,
      "loss": 0.7967,
      "step": 7126
    },
    {
      "epoch": 0.5229050689215587,
      "grad_norm": 1.4011885448680275,
      "learning_rate": 4.871688820162063e-06,
      "loss": 0.6926,
      "step": 7127
    },
    {
      "epoch": 0.5229784385116979,
      "grad_norm": 1.4415630593939763,
      "learning_rate": 4.870501018185546e-06,
      "loss": 0.762,
      "step": 7128
    },
    {
      "epoch": 0.5229784385116979,
      "lm_loss": 0.6851,
      "step": 7128,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.5229784385116979,
      "lm_loss": 0.0857,
      "step": 7128,
      "vm_loss": 0.0778
    },
    {
      "epoch": 0.5229784385116979,
      "lm_loss": 0.6726,
      "step": 7128,
      "vm_loss": 0.2058
    },
    {
      "epoch": 0.5229784385116979,
      "lm_loss": 0.7537,
      "step": 7128,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.5229784385116979,
      "lm_loss": 0.4157,
      "step": 7128,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.5229784385116979,
      "lm_loss": 0.0737,
      "step": 7128,
      "vm_loss": 0.1509
    },
    {
      "epoch": 0.5229784385116979,
      "lm_loss": 0.6327,
      "step": 7128,
      "vm_loss": 0.0809
    },
    {
      "epoch": 0.5229784385116979,
      "lm_loss": 0.5697,
      "step": 7128,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.523051808101837,
      "grad_norm": 1.4306925385101739,
      "learning_rate": 4.869313223522159e-06,
      "loss": 0.7118,
      "step": 7129
    },
    {
      "epoch": 0.5231251776919761,
      "grad_norm": 1.4938402611556851,
      "learning_rate": 4.8681254362389755e-06,
      "loss": 0.7926,
      "step": 7130
    },
    {
      "epoch": 0.5231985472821152,
      "grad_norm": 0.591912604271817,
      "learning_rate": 4.866937656403078e-06,
      "loss": 0.5064,
      "step": 7131
    },
    {
      "epoch": 0.5232719168722544,
      "grad_norm": 1.4901859850798111,
      "learning_rate": 4.865749884081539e-06,
      "loss": 0.7616,
      "step": 7132
    },
    {
      "epoch": 0.5233452864623935,
      "grad_norm": 1.5138324854853689,
      "learning_rate": 4.864562119341436e-06,
      "loss": 0.7093,
      "step": 7133
    },
    {
      "epoch": 0.5234186560525326,
      "grad_norm": 1.4148788521800144,
      "learning_rate": 4.863374362249847e-06,
      "loss": 0.7958,
      "step": 7134
    },
    {
      "epoch": 0.5234920256426717,
      "grad_norm": 1.45527942336591,
      "learning_rate": 4.862186612873845e-06,
      "loss": 0.7214,
      "step": 7135
    },
    {
      "epoch": 0.5235653952328109,
      "grad_norm": 1.5353138876651182,
      "learning_rate": 4.860998871280506e-06,
      "loss": 0.7954,
      "step": 7136
    },
    {
      "epoch": 0.5235653952328109,
      "lm_loss": 0.5738,
      "step": 7136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5235653952328109,
      "lm_loss": 0.4566,
      "step": 7136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5235653952328109,
      "lm_loss": 0.4973,
      "step": 7136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5235653952328109,
      "lm_loss": 0.5044,
      "step": 7136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5235653952328109,
      "lm_loss": 0.4035,
      "step": 7136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5235653952328109,
      "lm_loss": 0.5431,
      "step": 7136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5235653952328109,
      "lm_loss": 0.6429,
      "step": 7136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5235653952328109,
      "lm_loss": 0.4905,
      "step": 7136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.52363876482295,
      "grad_norm": 0.5500626912451801,
      "learning_rate": 4.859811137536906e-06,
      "loss": 0.4791,
      "step": 7137
    },
    {
      "epoch": 0.5237121344130892,
      "grad_norm": 1.6654087288834665,
      "learning_rate": 4.8586234117101175e-06,
      "loss": 0.7535,
      "step": 7138
    },
    {
      "epoch": 0.5237855040032282,
      "grad_norm": 1.5785657463932405,
      "learning_rate": 4.857435693867215e-06,
      "loss": 0.8934,
      "step": 7139
    },
    {
      "epoch": 0.5238588735933674,
      "grad_norm": 0.5623167052505934,
      "learning_rate": 4.856247984075273e-06,
      "loss": 0.4938,
      "step": 7140
    },
    {
      "epoch": 0.5239322431835065,
      "grad_norm": 0.5485005757506453,
      "learning_rate": 4.855060282401362e-06,
      "loss": 0.4982,
      "step": 7141
    },
    {
      "epoch": 0.5240056127736457,
      "grad_norm": 1.7014053337617994,
      "learning_rate": 4.853872588912556e-06,
      "loss": 0.7393,
      "step": 7142
    },
    {
      "epoch": 0.5240789823637848,
      "grad_norm": 1.7677533941415884,
      "learning_rate": 4.852684903675928e-06,
      "loss": 0.6824,
      "step": 7143
    },
    {
      "epoch": 0.5241523519539238,
      "grad_norm": 1.4848007584248746,
      "learning_rate": 4.851497226758547e-06,
      "loss": 0.7155,
      "step": 7144
    },
    {
      "epoch": 0.5241523519539238,
      "lm_loss": 1.2482,
      "step": 7144,
      "vm_loss": 0.0339
    },
    {
      "epoch": 0.5241523519539238,
      "lm_loss": 1.0562,
      "step": 7144,
      "vm_loss": 0.166
    },
    {
      "epoch": 0.5241523519539238,
      "lm_loss": 1.2461,
      "step": 7144,
      "vm_loss": 0.1983
    },
    {
      "epoch": 0.5241523519539238,
      "lm_loss": 0.269,
      "step": 7144,
      "vm_loss": 0.0604
    },
    {
      "epoch": 0.5241523519539238,
      "lm_loss": 0.5164,
      "step": 7144,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.5241523519539238,
      "lm_loss": 0.2304,
      "step": 7144,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.5241523519539238,
      "lm_loss": 0.5426,
      "step": 7144,
      "vm_loss": 0.2172
    },
    {
      "epoch": 0.5241523519539238,
      "lm_loss": 1.1895,
      "step": 7144,
      "vm_loss": 0.2109
    },
    {
      "epoch": 0.524225721544063,
      "grad_norm": 1.3905237883205988,
      "learning_rate": 4.850309558227486e-06,
      "loss": 0.7259,
      "step": 7145
    },
    {
      "epoch": 0.5242990911342021,
      "grad_norm": 1.258141661083344,
      "learning_rate": 4.849121898149816e-06,
      "loss": 0.7778,
      "step": 7146
    },
    {
      "epoch": 0.5243724607243413,
      "grad_norm": 1.6067471399084345,
      "learning_rate": 4.8479342465926034e-06,
      "loss": 0.7502,
      "step": 7147
    },
    {
      "epoch": 0.5244458303144804,
      "grad_norm": 1.5779560490402134,
      "learning_rate": 4.846746603622923e-06,
      "loss": 0.8916,
      "step": 7148
    },
    {
      "epoch": 0.5245191999046195,
      "grad_norm": 1.51495711452843,
      "learning_rate": 4.845558969307839e-06,
      "loss": 0.812,
      "step": 7149
    },
    {
      "epoch": 0.5245925694947586,
      "grad_norm": 1.5417170049692026,
      "learning_rate": 4.8443713437144245e-06,
      "loss": 0.9507,
      "step": 7150
    },
    {
      "epoch": 0.5246659390848978,
      "grad_norm": 1.6793600263183026,
      "learning_rate": 4.843183726909746e-06,
      "loss": 0.8148,
      "step": 7151
    },
    {
      "epoch": 0.5247393086750369,
      "grad_norm": 0.6303918116942755,
      "learning_rate": 4.841996118960868e-06,
      "loss": 0.493,
      "step": 7152
    },
    {
      "epoch": 0.5247393086750369,
      "lm_loss": 0.4376,
      "step": 7152,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.5247393086750369,
      "lm_loss": 0.5141,
      "step": 7152,
      "vm_loss": 0.1049
    },
    {
      "epoch": 0.5247393086750369,
      "lm_loss": 0.1646,
      "step": 7152,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.5247393086750369,
      "lm_loss": 0.4072,
      "step": 7152,
      "vm_loss": 0.1937
    },
    {
      "epoch": 0.5247393086750369,
      "lm_loss": 0.4838,
      "step": 7152,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.5247393086750369,
      "lm_loss": 0.3456,
      "step": 7152,
      "vm_loss": 0.0428
    },
    {
      "epoch": 0.5247393086750369,
      "lm_loss": 0.5958,
      "step": 7152,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.5247393086750369,
      "lm_loss": 0.7269,
      "step": 7152,
      "vm_loss": 0.1472
    },
    {
      "epoch": 0.5248126782651761,
      "grad_norm": 1.5399138455559926,
      "learning_rate": 4.840808519934863e-06,
      "loss": 0.7178,
      "step": 7153
    },
    {
      "epoch": 0.5248860478553151,
      "grad_norm": 1.4145922197056526,
      "learning_rate": 4.839620929898795e-06,
      "loss": 0.8413,
      "step": 7154
    },
    {
      "epoch": 0.5249594174454543,
      "grad_norm": 0.5902019441758964,
      "learning_rate": 4.838433348919729e-06,
      "loss": 0.5032,
      "step": 7155
    },
    {
      "epoch": 0.5250327870355934,
      "grad_norm": 1.6175251176664,
      "learning_rate": 4.837245777064732e-06,
      "loss": 0.8302,
      "step": 7156
    },
    {
      "epoch": 0.5251061566257326,
      "grad_norm": 1.4507394433126146,
      "learning_rate": 4.83605821440087e-06,
      "loss": 0.866,
      "step": 7157
    },
    {
      "epoch": 0.5251795262158717,
      "grad_norm": 1.482476753842686,
      "learning_rate": 4.834870660995206e-06,
      "loss": 0.7302,
      "step": 7158
    },
    {
      "epoch": 0.5252528958060108,
      "grad_norm": 0.5383088602849514,
      "learning_rate": 4.833683116914805e-06,
      "loss": 0.4676,
      "step": 7159
    },
    {
      "epoch": 0.5253262653961499,
      "grad_norm": 1.4459197381667994,
      "learning_rate": 4.83249558222673e-06,
      "loss": 0.7739,
      "step": 7160
    },
    {
      "epoch": 0.5253262653961499,
      "lm_loss": 0.5095,
      "step": 7160,
      "vm_loss": 0.0423
    },
    {
      "epoch": 0.5253262653961499,
      "lm_loss": 0.9557,
      "step": 7160,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.5253262653961499,
      "lm_loss": 0.9043,
      "step": 7160,
      "vm_loss": 0.1276
    },
    {
      "epoch": 0.5253262653961499,
      "lm_loss": 0.369,
      "step": 7160,
      "vm_loss": 0.105
    },
    {
      "epoch": 0.5253262653961499,
      "lm_loss": 0.6343,
      "step": 7160,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.5253262653961499,
      "lm_loss": 0.356,
      "step": 7160,
      "vm_loss": 0.1353
    },
    {
      "epoch": 0.5253262653961499,
      "lm_loss": 0.0706,
      "step": 7160,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.5253262653961499,
      "lm_loss": 1.448,
      "step": 7160,
      "vm_loss": 0.2027
    },
    {
      "epoch": 0.525399634986289,
      "grad_norm": 1.372533312166267,
      "learning_rate": 4.831308056998044e-06,
      "loss": 0.7459,
      "step": 7161
    },
    {
      "epoch": 0.5254730045764282,
      "grad_norm": 0.5741437888214891,
      "learning_rate": 4.830120541295811e-06,
      "loss": 0.4954,
      "step": 7162
    },
    {
      "epoch": 0.5255463741665674,
      "grad_norm": 1.3488681498201025,
      "learning_rate": 4.82893303518709e-06,
      "loss": 0.7829,
      "step": 7163
    },
    {
      "epoch": 0.5256197437567064,
      "grad_norm": 1.6894482717279329,
      "learning_rate": 4.827745538738946e-06,
      "loss": 0.7807,
      "step": 7164
    },
    {
      "epoch": 0.5256931133468455,
      "grad_norm": 0.5853047222188675,
      "learning_rate": 4.826558052018436e-06,
      "loss": 0.5066,
      "step": 7165
    },
    {
      "epoch": 0.5257664829369847,
      "grad_norm": 1.597970560280684,
      "learning_rate": 4.825370575092624e-06,
      "loss": 0.7927,
      "step": 7166
    },
    {
      "epoch": 0.5258398525271238,
      "grad_norm": 1.7300171158738198,
      "learning_rate": 4.8241831080285675e-06,
      "loss": 0.8145,
      "step": 7167
    },
    {
      "epoch": 0.525913222117263,
      "grad_norm": 1.4423901768288716,
      "learning_rate": 4.822995650893325e-06,
      "loss": 0.7349,
      "step": 7168
    },
    {
      "epoch": 0.525913222117263,
      "lm_loss": 2.1426,
      "step": 7168,
      "vm_loss": 0.0529
    },
    {
      "epoch": 0.525913222117263,
      "lm_loss": 1.1925,
      "step": 7168,
      "vm_loss": 0.1368
    },
    {
      "epoch": 0.525913222117263,
      "lm_loss": 1.1672,
      "step": 7168,
      "vm_loss": 0.2131
    },
    {
      "epoch": 0.525913222117263,
      "lm_loss": 0.7071,
      "step": 7168,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.525913222117263,
      "lm_loss": 0.7369,
      "step": 7168,
      "vm_loss": 0.1524
    },
    {
      "epoch": 0.525913222117263,
      "lm_loss": 0.7551,
      "step": 7168,
      "vm_loss": 0.1614
    },
    {
      "epoch": 0.525913222117263,
      "lm_loss": 0.481,
      "step": 7168,
      "vm_loss": 0.1028
    },
    {
      "epoch": 0.525913222117263,
      "lm_loss": 0.6163,
      "step": 7168,
      "vm_loss": 0.1185
    },
    {
      "epoch": 0.525986591707402,
      "grad_norm": 1.4747957582048459,
      "learning_rate": 4.821808203753959e-06,
      "loss": 0.8106,
      "step": 7169
    },
    {
      "epoch": 0.5260599612975412,
      "grad_norm": 1.523074342508824,
      "learning_rate": 4.820620766677524e-06,
      "loss": 0.8078,
      "step": 7170
    },
    {
      "epoch": 0.5261333308876803,
      "grad_norm": 0.5719449909432157,
      "learning_rate": 4.819433339731077e-06,
      "loss": 0.4926,
      "step": 7171
    },
    {
      "epoch": 0.5262067004778195,
      "grad_norm": 0.5859533991182191,
      "learning_rate": 4.818245922981678e-06,
      "loss": 0.4966,
      "step": 7172
    },
    {
      "epoch": 0.5262800700679586,
      "grad_norm": 1.5036309081846886,
      "learning_rate": 4.817058516496382e-06,
      "loss": 0.8016,
      "step": 7173
    },
    {
      "epoch": 0.5263534396580977,
      "grad_norm": 0.5638808624653431,
      "learning_rate": 4.815871120342244e-06,
      "loss": 0.4683,
      "step": 7174
    },
    {
      "epoch": 0.5264268092482368,
      "grad_norm": 1.5176818786879285,
      "learning_rate": 4.81468373458632e-06,
      "loss": 0.8486,
      "step": 7175
    },
    {
      "epoch": 0.526500178838376,
      "grad_norm": 1.696556035315358,
      "learning_rate": 4.813496359295664e-06,
      "loss": 0.8105,
      "step": 7176
    },
    {
      "epoch": 0.526500178838376,
      "lm_loss": 2.4308,
      "step": 7176,
      "vm_loss": 0.0805
    },
    {
      "epoch": 0.526500178838376,
      "lm_loss": 0.81,
      "step": 7176,
      "vm_loss": 0.0439
    },
    {
      "epoch": 0.526500178838376,
      "lm_loss": 0.8462,
      "step": 7176,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.526500178838376,
      "lm_loss": 0.5387,
      "step": 7176,
      "vm_loss": 0.0913
    },
    {
      "epoch": 0.526500178838376,
      "lm_loss": 0.7083,
      "step": 7176,
      "vm_loss": 0.0412
    },
    {
      "epoch": 0.526500178838376,
      "lm_loss": 0.6128,
      "step": 7176,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.526500178838376,
      "lm_loss": 0.3276,
      "step": 7176,
      "vm_loss": 0.0884
    },
    {
      "epoch": 0.526500178838376,
      "lm_loss": 0.4399,
      "step": 7176,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.5265735484285151,
      "grad_norm": 1.7162878362832832,
      "learning_rate": 4.812308994537331e-06,
      "loss": 0.7242,
      "step": 7177
    },
    {
      "epoch": 0.5266469180186543,
      "grad_norm": 1.6624078611057223,
      "learning_rate": 4.811121640378375e-06,
      "loss": 0.7239,
      "step": 7178
    },
    {
      "epoch": 0.5267202876087933,
      "grad_norm": 1.668986562979207,
      "learning_rate": 4.809934296885845e-06,
      "loss": 0.94,
      "step": 7179
    },
    {
      "epoch": 0.5267936571989325,
      "grad_norm": 1.4140949815666704,
      "learning_rate": 4.808746964126798e-06,
      "loss": 0.722,
      "step": 7180
    },
    {
      "epoch": 0.5268670267890716,
      "grad_norm": 1.7793582373313404,
      "learning_rate": 4.807559642168284e-06,
      "loss": 0.936,
      "step": 7181
    },
    {
      "epoch": 0.5269403963792108,
      "grad_norm": 1.7030232446431106,
      "learning_rate": 4.806372331077351e-06,
      "loss": 0.6755,
      "step": 7182
    },
    {
      "epoch": 0.5270137659693499,
      "grad_norm": 1.382451415212444,
      "learning_rate": 4.805185030921055e-06,
      "loss": 0.7801,
      "step": 7183
    },
    {
      "epoch": 0.5270871355594889,
      "grad_norm": 0.6502057767612445,
      "learning_rate": 4.8039977417664406e-06,
      "loss": 0.5098,
      "step": 7184
    },
    {
      "epoch": 0.5270871355594889,
      "lm_loss": 1.4441,
      "step": 7184,
      "vm_loss": 0.0639
    },
    {
      "epoch": 0.5270871355594889,
      "lm_loss": 0.242,
      "step": 7184,
      "vm_loss": 0.1284
    },
    {
      "epoch": 0.5270871355594889,
      "lm_loss": 0.1306,
      "step": 7184,
      "vm_loss": 0.0757
    },
    {
      "epoch": 0.5270871355594889,
      "lm_loss": 0.1973,
      "step": 7184,
      "vm_loss": 0.0411
    },
    {
      "epoch": 0.5270871355594889,
      "lm_loss": 0.3473,
      "step": 7184,
      "vm_loss": 0.1555
    },
    {
      "epoch": 0.5270871355594889,
      "lm_loss": 0.1683,
      "step": 7184,
      "vm_loss": 0.1539
    },
    {
      "epoch": 0.5270871355594889,
      "lm_loss": 1.059,
      "step": 7184,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.5270871355594889,
      "lm_loss": 0.5043,
      "step": 7184,
      "vm_loss": 0.126
    },
    {
      "epoch": 0.5271605051496281,
      "grad_norm": 1.5450833278349803,
      "learning_rate": 4.802810463680559e-06,
      "loss": 0.7492,
      "step": 7185
    },
    {
      "epoch": 0.5272338747397672,
      "grad_norm": 1.9635683700576314,
      "learning_rate": 4.801623196730461e-06,
      "loss": 0.7339,
      "step": 7186
    },
    {
      "epoch": 0.5273072443299064,
      "grad_norm": 1.6479988959581253,
      "learning_rate": 4.80043594098319e-06,
      "loss": 0.8461,
      "step": 7187
    },
    {
      "epoch": 0.5273806139200455,
      "grad_norm": 1.674843427042842,
      "learning_rate": 4.7992486965057974e-06,
      "loss": 0.9149,
      "step": 7188
    },
    {
      "epoch": 0.5274539835101846,
      "grad_norm": 0.6078568514513962,
      "learning_rate": 4.798061463365327e-06,
      "loss": 0.4985,
      "step": 7189
    },
    {
      "epoch": 0.5275273531003237,
      "grad_norm": 1.7195557799501038,
      "learning_rate": 4.796874241628826e-06,
      "loss": 0.8063,
      "step": 7190
    },
    {
      "epoch": 0.5276007226904629,
      "grad_norm": 0.584935858273491,
      "learning_rate": 4.7956870313633405e-06,
      "loss": 0.5115,
      "step": 7191
    },
    {
      "epoch": 0.527674092280602,
      "grad_norm": 1.3954286187171836,
      "learning_rate": 4.794499832635912e-06,
      "loss": 0.7947,
      "step": 7192
    },
    {
      "epoch": 0.527674092280602,
      "lm_loss": 0.5688,
      "step": 7192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.527674092280602,
      "lm_loss": 0.4876,
      "step": 7192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.527674092280602,
      "lm_loss": 0.5347,
      "step": 7192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.527674092280602,
      "lm_loss": 0.5988,
      "step": 7192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.527674092280602,
      "lm_loss": 0.4229,
      "step": 7192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.527674092280602,
      "lm_loss": 0.7318,
      "step": 7192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.527674092280602,
      "lm_loss": 0.5278,
      "step": 7192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.527674092280602,
      "lm_loss": 0.4967,
      "step": 7192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5277474618707412,
      "grad_norm": 0.5790490447716382,
      "learning_rate": 4.7933126455135895e-06,
      "loss": 0.5089,
      "step": 7193
    },
    {
      "epoch": 0.5278208314608802,
      "grad_norm": 1.516049855087921,
      "learning_rate": 4.792125470063413e-06,
      "loss": 0.856,
      "step": 7194
    },
    {
      "epoch": 0.5278942010510194,
      "grad_norm": 1.5841733716861137,
      "learning_rate": 4.790938306352425e-06,
      "loss": 0.8204,
      "step": 7195
    },
    {
      "epoch": 0.5279675706411585,
      "grad_norm": 1.5303699790500034,
      "learning_rate": 4.7897511544476695e-06,
      "loss": 0.73,
      "step": 7196
    },
    {
      "epoch": 0.5280409402312977,
      "grad_norm": 1.4826461181254509,
      "learning_rate": 4.788564014416187e-06,
      "loss": 0.7459,
      "step": 7197
    },
    {
      "epoch": 0.5281143098214368,
      "grad_norm": 1.5672206884988926,
      "learning_rate": 4.787376886325018e-06,
      "loss": 0.741,
      "step": 7198
    },
    {
      "epoch": 0.5281876794115758,
      "grad_norm": 0.6280610582137109,
      "learning_rate": 4.786189770241203e-06,
      "loss": 0.4782,
      "step": 7199
    },
    {
      "epoch": 0.528261049001715,
      "grad_norm": 1.8710992165476164,
      "learning_rate": 4.785002666231781e-06,
      "loss": 0.7493,
      "step": 7200
    },
    {
      "epoch": 0.528261049001715,
      "lm_loss": 1.416,
      "step": 7200,
      "vm_loss": 0.0642
    },
    {
      "epoch": 0.528261049001715,
      "lm_loss": 0.5411,
      "step": 7200,
      "vm_loss": 0.0856
    },
    {
      "epoch": 0.528261049001715,
      "lm_loss": 0.3753,
      "step": 7200,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.528261049001715,
      "lm_loss": 1.1004,
      "step": 7200,
      "vm_loss": 0.1768
    },
    {
      "epoch": 0.528261049001715,
      "lm_loss": 0.1141,
      "step": 7200,
      "vm_loss": 0.0946
    },
    {
      "epoch": 0.528261049001715,
      "lm_loss": 0.0991,
      "step": 7200,
      "vm_loss": 0.1198
    },
    {
      "epoch": 0.528261049001715,
      "lm_loss": 0.6914,
      "step": 7200,
      "vm_loss": 0.0929
    },
    {
      "epoch": 0.528261049001715,
      "lm_loss": 0.2846,
      "step": 7200,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.5283344185918541,
      "grad_norm": 1.5094099119198336,
      "learning_rate": 4.7838155743637894e-06,
      "loss": 0.666,
      "step": 7201
    },
    {
      "epoch": 0.5284077881819933,
      "grad_norm": 1.4186251918074937,
      "learning_rate": 4.78262849470427e-06,
      "loss": 0.8169,
      "step": 7202
    },
    {
      "epoch": 0.5284811577721324,
      "grad_norm": 1.4163978607727579,
      "learning_rate": 4.7814414273202565e-06,
      "loss": 0.7548,
      "step": 7203
    },
    {
      "epoch": 0.5285545273622715,
      "grad_norm": 0.619417419994739,
      "learning_rate": 4.780254372278787e-06,
      "loss": 0.5078,
      "step": 7204
    },
    {
      "epoch": 0.5286278969524106,
      "grad_norm": 1.4491900389778853,
      "learning_rate": 4.779067329646899e-06,
      "loss": 0.7386,
      "step": 7205
    },
    {
      "epoch": 0.5287012665425498,
      "grad_norm": 0.5477087580858803,
      "learning_rate": 4.777880299491623e-06,
      "loss": 0.4779,
      "step": 7206
    },
    {
      "epoch": 0.5287746361326889,
      "grad_norm": 1.4342020416959058,
      "learning_rate": 4.776693281879999e-06,
      "loss": 0.7109,
      "step": 7207
    },
    {
      "epoch": 0.5288480057228281,
      "grad_norm": 1.5118887929092701,
      "learning_rate": 4.775506276879058e-06,
      "loss": 0.7297,
      "step": 7208
    },
    {
      "epoch": 0.5288480057228281,
      "lm_loss": 1.4359,
      "step": 7208,
      "vm_loss": 0.0369
    },
    {
      "epoch": 0.5288480057228281,
      "lm_loss": 0.4075,
      "step": 7208,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.5288480057228281,
      "lm_loss": 0.7449,
      "step": 7208,
      "vm_loss": 0.1095
    },
    {
      "epoch": 0.5288480057228281,
      "lm_loss": 1.5204,
      "step": 7208,
      "vm_loss": 0.1005
    },
    {
      "epoch": 0.5288480057228281,
      "lm_loss": 0.1238,
      "step": 7208,
      "vm_loss": 0.1115
    },
    {
      "epoch": 0.5288480057228281,
      "lm_loss": 0.5778,
      "step": 7208,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.5288480057228281,
      "lm_loss": 0.0974,
      "step": 7208,
      "vm_loss": 0.086
    },
    {
      "epoch": 0.5288480057228281,
      "lm_loss": 0.6901,
      "step": 7208,
      "vm_loss": 0.1261
    },
    {
      "epoch": 0.5289213753129671,
      "grad_norm": 1.5620079089874979,
      "learning_rate": 4.774319284555833e-06,
      "loss": 0.8186,
      "step": 7209
    },
    {
      "epoch": 0.5289947449031063,
      "grad_norm": 0.5685711374785971,
      "learning_rate": 4.773132304977356e-06,
      "loss": 0.48,
      "step": 7210
    },
    {
      "epoch": 0.5290681144932454,
      "grad_norm": 1.3332349044205005,
      "learning_rate": 4.77194533821066e-06,
      "loss": 0.6689,
      "step": 7211
    },
    {
      "epoch": 0.5291414840833846,
      "grad_norm": 0.5872670950058875,
      "learning_rate": 4.770758384322777e-06,
      "loss": 0.5176,
      "step": 7212
    },
    {
      "epoch": 0.5292148536735236,
      "grad_norm": 1.5532688776141699,
      "learning_rate": 4.7695714433807344e-06,
      "loss": 0.7923,
      "step": 7213
    },
    {
      "epoch": 0.5292882232636628,
      "grad_norm": 0.554225157269317,
      "learning_rate": 4.768384515451563e-06,
      "loss": 0.471,
      "step": 7214
    },
    {
      "epoch": 0.5293615928538019,
      "grad_norm": 0.5709657212651489,
      "learning_rate": 4.7671976006022926e-06,
      "loss": 0.4915,
      "step": 7215
    },
    {
      "epoch": 0.5294349624439411,
      "grad_norm": 1.3478285700775439,
      "learning_rate": 4.766010698899949e-06,
      "loss": 0.8596,
      "step": 7216
    },
    {
      "epoch": 0.5294349624439411,
      "lm_loss": 1.1806,
      "step": 7216,
      "vm_loss": 0.0191
    },
    {
      "epoch": 0.5294349624439411,
      "lm_loss": 1.2878,
      "step": 7216,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.5294349624439411,
      "lm_loss": 0.8752,
      "step": 7216,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.5294349624439411,
      "lm_loss": 0.8666,
      "step": 7216,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.5294349624439411,
      "lm_loss": 0.0688,
      "step": 7216,
      "vm_loss": 0.2256
    },
    {
      "epoch": 0.5294349624439411,
      "lm_loss": 1.0536,
      "step": 7216,
      "vm_loss": 0.2402
    },
    {
      "epoch": 0.5294349624439411,
      "lm_loss": 0.8239,
      "step": 7216,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.5294349624439411,
      "lm_loss": 0.7825,
      "step": 7216,
      "vm_loss": 0.1571
    },
    {
      "epoch": 0.5295083320340802,
      "grad_norm": 1.5135835640828539,
      "learning_rate": 4.764823810411561e-06,
      "loss": 0.832,
      "step": 7217
    },
    {
      "epoch": 0.5295817016242192,
      "grad_norm": 1.421066521405291,
      "learning_rate": 4.7636369352041555e-06,
      "loss": 0.8447,
      "step": 7218
    },
    {
      "epoch": 0.5296550712143584,
      "grad_norm": 1.4054876279381292,
      "learning_rate": 4.762450073344756e-06,
      "loss": 0.6369,
      "step": 7219
    },
    {
      "epoch": 0.5297284408044975,
      "grad_norm": 0.5932464974153955,
      "learning_rate": 4.761263224900392e-06,
      "loss": 0.4914,
      "step": 7220
    },
    {
      "epoch": 0.5298018103946367,
      "grad_norm": 1.6331092982451896,
      "learning_rate": 4.760076389938083e-06,
      "loss": 0.7654,
      "step": 7221
    },
    {
      "epoch": 0.5298751799847758,
      "grad_norm": 1.5016263571673913,
      "learning_rate": 4.758889568524854e-06,
      "loss": 0.7351,
      "step": 7222
    },
    {
      "epoch": 0.5299485495749149,
      "grad_norm": 1.4143631467454325,
      "learning_rate": 4.757702760727729e-06,
      "loss": 0.752,
      "step": 7223
    },
    {
      "epoch": 0.530021919165054,
      "grad_norm": 1.353213958352184,
      "learning_rate": 4.756515966613729e-06,
      "loss": 0.7088,
      "step": 7224
    },
    {
      "epoch": 0.530021919165054,
      "lm_loss": 0.9678,
      "step": 7224,
      "vm_loss": 0.0798
    },
    {
      "epoch": 0.530021919165054,
      "lm_loss": 1.1,
      "step": 7224,
      "vm_loss": 0.1651
    },
    {
      "epoch": 0.530021919165054,
      "lm_loss": 0.2511,
      "step": 7224,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.530021919165054,
      "lm_loss": 0.6581,
      "step": 7224,
      "vm_loss": 0.2147
    },
    {
      "epoch": 0.530021919165054,
      "lm_loss": 0.3815,
      "step": 7224,
      "vm_loss": 0.2454
    },
    {
      "epoch": 0.530021919165054,
      "lm_loss": 0.7358,
      "step": 7224,
      "vm_loss": 0.1513
    },
    {
      "epoch": 0.530021919165054,
      "lm_loss": 0.751,
      "step": 7224,
      "vm_loss": 0.2351
    },
    {
      "epoch": 0.530021919165054,
      "lm_loss": 0.616,
      "step": 7224,
      "vm_loss": 0.0898
    },
    {
      "epoch": 0.5300952887551932,
      "grad_norm": 1.6931467968070693,
      "learning_rate": 4.755329186249875e-06,
      "loss": 0.8323,
      "step": 7225
    },
    {
      "epoch": 0.5301686583453323,
      "grad_norm": 1.4216615877047507,
      "learning_rate": 4.754142419703189e-06,
      "loss": 0.7423,
      "step": 7226
    },
    {
      "epoch": 0.5302420279354715,
      "grad_norm": 1.377440567993094,
      "learning_rate": 4.752955667040689e-06,
      "loss": 0.7978,
      "step": 7227
    },
    {
      "epoch": 0.5303153975256105,
      "grad_norm": 0.5492596324016088,
      "learning_rate": 4.751768928329392e-06,
      "loss": 0.4962,
      "step": 7228
    },
    {
      "epoch": 0.5303887671157497,
      "grad_norm": 0.5621137560418414,
      "learning_rate": 4.7505822036363214e-06,
      "loss": 0.5166,
      "step": 7229
    },
    {
      "epoch": 0.5304621367058888,
      "grad_norm": 0.5586407155680937,
      "learning_rate": 4.749395493028489e-06,
      "loss": 0.5029,
      "step": 7230
    },
    {
      "epoch": 0.530535506296028,
      "grad_norm": 1.6077784457171755,
      "learning_rate": 4.748208796572916e-06,
      "loss": 0.8806,
      "step": 7231
    },
    {
      "epoch": 0.5306088758861671,
      "grad_norm": 0.5595890265792843,
      "learning_rate": 4.747022114336615e-06,
      "loss": 0.5068,
      "step": 7232
    },
    {
      "epoch": 0.5306088758861671,
      "lm_loss": 0.6853,
      "step": 7232,
      "vm_loss": 0.0701
    },
    {
      "epoch": 0.5306088758861671,
      "lm_loss": 1.4195,
      "step": 7232,
      "vm_loss": 0.1333
    },
    {
      "epoch": 0.5306088758861671,
      "lm_loss": 0.552,
      "step": 7232,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.5306088758861671,
      "lm_loss": 0.1804,
      "step": 7232,
      "vm_loss": 0.1614
    },
    {
      "epoch": 0.5306088758861671,
      "lm_loss": 0.9186,
      "step": 7232,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.5306088758861671,
      "lm_loss": 0.6805,
      "step": 7232,
      "vm_loss": 0.0837
    },
    {
      "epoch": 0.5306088758861671,
      "lm_loss": 0.9587,
      "step": 7232,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.5306088758861671,
      "lm_loss": 0.0039,
      "step": 7232,
      "vm_loss": 0.0659
    },
    {
      "epoch": 0.5306822454763062,
      "grad_norm": 1.3027782478897778,
      "learning_rate": 4.745835446386601e-06,
      "loss": 0.7285,
      "step": 7233
    },
    {
      "epoch": 0.5307556150664453,
      "grad_norm": 0.5624212773633088,
      "learning_rate": 4.744648792789889e-06,
      "loss": 0.4996,
      "step": 7234
    },
    {
      "epoch": 0.5308289846565845,
      "grad_norm": 1.5500731228554328,
      "learning_rate": 4.743462153613493e-06,
      "loss": 0.6441,
      "step": 7235
    },
    {
      "epoch": 0.5309023542467236,
      "grad_norm": 1.4255362092555577,
      "learning_rate": 4.742275528924423e-06,
      "loss": 0.7177,
      "step": 7236
    },
    {
      "epoch": 0.5309757238368628,
      "grad_norm": 0.5588789400041317,
      "learning_rate": 4.741088918789693e-06,
      "loss": 0.5122,
      "step": 7237
    },
    {
      "epoch": 0.5310490934270018,
      "grad_norm": 1.5098297463857318,
      "learning_rate": 4.739902323276312e-06,
      "loss": 0.8303,
      "step": 7238
    },
    {
      "epoch": 0.5311224630171409,
      "grad_norm": 1.4849977741094857,
      "learning_rate": 4.738715742451292e-06,
      "loss": 0.7691,
      "step": 7239
    },
    {
      "epoch": 0.5311958326072801,
      "grad_norm": 0.5878778093195122,
      "learning_rate": 4.737529176381642e-06,
      "loss": 0.5031,
      "step": 7240
    },
    {
      "epoch": 0.5311958326072801,
      "lm_loss": 0.2018,
      "step": 7240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5311958326072801,
      "lm_loss": 0.4369,
      "step": 7240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5311958326072801,
      "lm_loss": 0.5123,
      "step": 7240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5311958326072801,
      "lm_loss": 0.517,
      "step": 7240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5311958326072801,
      "lm_loss": 0.55,
      "step": 7240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5311958326072801,
      "lm_loss": 0.4822,
      "step": 7240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5311958326072801,
      "lm_loss": 0.4558,
      "step": 7240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5311958326072801,
      "lm_loss": 0.3412,
      "step": 7240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5312692021974192,
      "grad_norm": 0.5622047115314056,
      "learning_rate": 4.736342625134367e-06,
      "loss": 0.4912,
      "step": 7241
    },
    {
      "epoch": 0.5313425717875584,
      "grad_norm": 1.7197316697256653,
      "learning_rate": 4.7351560887764784e-06,
      "loss": 0.7886,
      "step": 7242
    },
    {
      "epoch": 0.5314159413776974,
      "grad_norm": 1.532763544062784,
      "learning_rate": 4.733969567374982e-06,
      "loss": 0.7239,
      "step": 7243
    },
    {
      "epoch": 0.5314893109678366,
      "grad_norm": 1.580727316723974,
      "learning_rate": 4.7327830609968805e-06,
      "loss": 0.8965,
      "step": 7244
    },
    {
      "epoch": 0.5315626805579757,
      "grad_norm": 0.5643515923726333,
      "learning_rate": 4.7315965697091846e-06,
      "loss": 0.4827,
      "step": 7245
    },
    {
      "epoch": 0.5316360501481149,
      "grad_norm": 0.5705902432333818,
      "learning_rate": 4.730410093578893e-06,
      "loss": 0.5212,
      "step": 7246
    },
    {
      "epoch": 0.531709419738254,
      "grad_norm": 0.5344771989448088,
      "learning_rate": 4.729223632673012e-06,
      "loss": 0.4787,
      "step": 7247
    },
    {
      "epoch": 0.5317827893283931,
      "grad_norm": 0.5866165682505294,
      "learning_rate": 4.728037187058542e-06,
      "loss": 0.4808,
      "step": 7248
    },
    {
      "epoch": 0.5317827893283931,
      "lm_loss": 1.408,
      "step": 7248,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.5317827893283931,
      "lm_loss": 0.8055,
      "step": 7248,
      "vm_loss": 0.0907
    },
    {
      "epoch": 0.5317827893283931,
      "lm_loss": 0.7793,
      "step": 7248,
      "vm_loss": 0.2055
    },
    {
      "epoch": 0.5317827893283931,
      "lm_loss": 0.7687,
      "step": 7248,
      "vm_loss": 0.1353
    },
    {
      "epoch": 0.5317827893283931,
      "lm_loss": 0.1381,
      "step": 7248,
      "vm_loss": 0.1055
    },
    {
      "epoch": 0.5317827893283931,
      "lm_loss": 0.9039,
      "step": 7248,
      "vm_loss": 0.1364
    },
    {
      "epoch": 0.5317827893283931,
      "lm_loss": 0.085,
      "step": 7248,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.5317827893283931,
      "lm_loss": 0.3241,
      "step": 7248,
      "vm_loss": 0.0952
    },
    {
      "epoch": 0.5318561589185322,
      "grad_norm": 1.4908932559427233,
      "learning_rate": 4.726850756802486e-06,
      "loss": 0.7288,
      "step": 7249
    },
    {
      "epoch": 0.5319295285086714,
      "grad_norm": 1.4321685883343103,
      "learning_rate": 4.725664341971845e-06,
      "loss": 0.8092,
      "step": 7250
    },
    {
      "epoch": 0.5320028980988105,
      "grad_norm": 0.5620243228890485,
      "learning_rate": 4.724477942633616e-06,
      "loss": 0.4769,
      "step": 7251
    },
    {
      "epoch": 0.5320762676889497,
      "grad_norm": 0.560030455928591,
      "learning_rate": 4.723291558854801e-06,
      "loss": 0.4834,
      "step": 7252
    },
    {
      "epoch": 0.5321496372790887,
      "grad_norm": 1.9271462186433799,
      "learning_rate": 4.722105190702397e-06,
      "loss": 0.6923,
      "step": 7253
    },
    {
      "epoch": 0.5322230068692279,
      "grad_norm": 1.5708866997410638,
      "learning_rate": 4.720918838243402e-06,
      "loss": 0.7851,
      "step": 7254
    },
    {
      "epoch": 0.532296376459367,
      "grad_norm": 1.6546448548063208,
      "learning_rate": 4.71973250154481e-06,
      "loss": 0.7533,
      "step": 7255
    },
    {
      "epoch": 0.5323697460495062,
      "grad_norm": 1.549551480213155,
      "learning_rate": 4.718546180673617e-06,
      "loss": 0.8085,
      "step": 7256
    },
    {
      "epoch": 0.5323697460495062,
      "lm_loss": 0.4543,
      "step": 7256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5323697460495062,
      "lm_loss": 0.4444,
      "step": 7256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5323697460495062,
      "lm_loss": 0.4648,
      "step": 7256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5323697460495062,
      "lm_loss": 0.5009,
      "step": 7256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5323697460495062,
      "lm_loss": 0.5106,
      "step": 7256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5323697460495062,
      "lm_loss": 0.471,
      "step": 7256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5323697460495062,
      "lm_loss": 0.5851,
      "step": 7256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5323697460495062,
      "lm_loss": 0.5482,
      "step": 7256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5324431156396453,
      "grad_norm": 0.5693844958406986,
      "learning_rate": 4.7173598756968194e-06,
      "loss": 0.4989,
      "step": 7257
    },
    {
      "epoch": 0.5325164852297843,
      "grad_norm": 1.6587088219003536,
      "learning_rate": 4.7161735866814104e-06,
      "loss": 0.7219,
      "step": 7258
    },
    {
      "epoch": 0.5325898548199235,
      "grad_norm": 1.5773025692639455,
      "learning_rate": 4.71498731369438e-06,
      "loss": 0.6833,
      "step": 7259
    },
    {
      "epoch": 0.5326632244100626,
      "grad_norm": 0.5747457431218964,
      "learning_rate": 4.713801056802722e-06,
      "loss": 0.4985,
      "step": 7260
    },
    {
      "epoch": 0.5327365940002018,
      "grad_norm": 1.4868390136244736,
      "learning_rate": 4.712614816073428e-06,
      "loss": 0.8712,
      "step": 7261
    },
    {
      "epoch": 0.5328099635903409,
      "grad_norm": 1.4363288750134129,
      "learning_rate": 4.711428591573487e-06,
      "loss": 0.7569,
      "step": 7262
    },
    {
      "epoch": 0.53288333318048,
      "grad_norm": 1.5700940833339345,
      "learning_rate": 4.710242383369889e-06,
      "loss": 0.7923,
      "step": 7263
    },
    {
      "epoch": 0.5329567027706191,
      "grad_norm": 1.444686474868235,
      "learning_rate": 4.709056191529619e-06,
      "loss": 0.7969,
      "step": 7264
    },
    {
      "epoch": 0.5329567027706191,
      "lm_loss": 0.4492,
      "step": 7264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5329567027706191,
      "lm_loss": 0.6045,
      "step": 7264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5329567027706191,
      "lm_loss": 0.532,
      "step": 7264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5329567027706191,
      "lm_loss": 0.3515,
      "step": 7264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5329567027706191,
      "lm_loss": 0.4638,
      "step": 7264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5329567027706191,
      "lm_loss": 0.5074,
      "step": 7264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5329567027706191,
      "lm_loss": 0.5269,
      "step": 7264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5329567027706191,
      "lm_loss": 0.4474,
      "step": 7264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5330300723607583,
      "grad_norm": 0.5657177882106488,
      "learning_rate": 4.707870016119667e-06,
      "loss": 0.493,
      "step": 7265
    },
    {
      "epoch": 0.5331034419508974,
      "grad_norm": 1.4243225259970769,
      "learning_rate": 4.7066838572070195e-06,
      "loss": 0.7991,
      "step": 7266
    },
    {
      "epoch": 0.5331768115410366,
      "grad_norm": 0.5564756498551047,
      "learning_rate": 4.70549771485866e-06,
      "loss": 0.5044,
      "step": 7267
    },
    {
      "epoch": 0.5332501811311756,
      "grad_norm": 0.5521478514110338,
      "learning_rate": 4.704311589141574e-06,
      "loss": 0.4775,
      "step": 7268
    },
    {
      "epoch": 0.5333235507213148,
      "grad_norm": 1.6999193607514609,
      "learning_rate": 4.703125480122747e-06,
      "loss": 0.8392,
      "step": 7269
    },
    {
      "epoch": 0.5333969203114539,
      "grad_norm": 1.432666132394978,
      "learning_rate": 4.701939387869156e-06,
      "loss": 0.6942,
      "step": 7270
    },
    {
      "epoch": 0.5334702899015931,
      "grad_norm": 1.4549174075323585,
      "learning_rate": 4.700753312447787e-06,
      "loss": 0.7501,
      "step": 7271
    },
    {
      "epoch": 0.5335436594917322,
      "grad_norm": 1.476401212085915,
      "learning_rate": 4.69956725392562e-06,
      "loss": 0.8455,
      "step": 7272
    },
    {
      "epoch": 0.5335436594917322,
      "lm_loss": 0.5373,
      "step": 7272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5335436594917322,
      "lm_loss": 0.4895,
      "step": 7272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5335436594917322,
      "lm_loss": 0.5481,
      "step": 7272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5335436594917322,
      "lm_loss": 0.4452,
      "step": 7272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5335436594917322,
      "lm_loss": 0.4837,
      "step": 7272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5335436594917322,
      "lm_loss": 0.4842,
      "step": 7272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5335436594917322,
      "lm_loss": 0.4139,
      "step": 7272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5335436594917322,
      "lm_loss": 0.3978,
      "step": 7272,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5336170290818713,
      "grad_norm": 0.5589382238679901,
      "learning_rate": 4.698381212369635e-06,
      "loss": 0.4917,
      "step": 7273
    },
    {
      "epoch": 0.5336903986720104,
      "grad_norm": 1.5095547253859058,
      "learning_rate": 4.6971951878468094e-06,
      "loss": 0.7913,
      "step": 7274
    },
    {
      "epoch": 0.5337637682621496,
      "grad_norm": 0.5727330572830222,
      "learning_rate": 4.696009180424121e-06,
      "loss": 0.4971,
      "step": 7275
    },
    {
      "epoch": 0.5338371378522887,
      "grad_norm": 1.3282089469895098,
      "learning_rate": 4.694823190168547e-06,
      "loss": 0.6908,
      "step": 7276
    },
    {
      "epoch": 0.5339105074424279,
      "grad_norm": 1.6634333572248003,
      "learning_rate": 4.693637217147063e-06,
      "loss": 0.7482,
      "step": 7277
    },
    {
      "epoch": 0.5339838770325669,
      "grad_norm": 0.5532205493081481,
      "learning_rate": 4.692451261426643e-06,
      "loss": 0.5033,
      "step": 7278
    },
    {
      "epoch": 0.534057246622706,
      "grad_norm": 1.437558679360917,
      "learning_rate": 4.6912653230742644e-06,
      "loss": 0.7435,
      "step": 7279
    },
    {
      "epoch": 0.5341306162128452,
      "grad_norm": 1.5685155279948158,
      "learning_rate": 4.690079402156896e-06,
      "loss": 0.7781,
      "step": 7280
    },
    {
      "epoch": 0.5341306162128452,
      "lm_loss": 1.5367,
      "step": 7280,
      "vm_loss": 0.0576
    },
    {
      "epoch": 0.5341306162128452,
      "lm_loss": 0.5561,
      "step": 7280,
      "vm_loss": 0.1307
    },
    {
      "epoch": 0.5341306162128452,
      "lm_loss": 0.0582,
      "step": 7280,
      "vm_loss": 0.0893
    },
    {
      "epoch": 0.5341306162128452,
      "lm_loss": 0.7025,
      "step": 7280,
      "vm_loss": 0.1429
    },
    {
      "epoch": 0.5341306162128452,
      "lm_loss": 0.1685,
      "step": 7280,
      "vm_loss": 0.0325
    },
    {
      "epoch": 0.5341306162128452,
      "lm_loss": 0.1382,
      "step": 7280,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.5341306162128452,
      "lm_loss": 0.8871,
      "step": 7280,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.5341306162128452,
      "lm_loss": 0.1703,
      "step": 7280,
      "vm_loss": 0.0991
    },
    {
      "epoch": 0.5342039858029843,
      "grad_norm": 1.4983985393822876,
      "learning_rate": 4.688893498741511e-06,
      "loss": 0.7458,
      "step": 7281
    },
    {
      "epoch": 0.5342773553931235,
      "grad_norm": 0.5492972493670124,
      "learning_rate": 4.687707612895082e-06,
      "loss": 0.4948,
      "step": 7282
    },
    {
      "epoch": 0.5343507249832625,
      "grad_norm": 1.3203477461867896,
      "learning_rate": 4.686521744684576e-06,
      "loss": 0.7543,
      "step": 7283
    },
    {
      "epoch": 0.5344240945734017,
      "grad_norm": 0.5799146421126042,
      "learning_rate": 4.6853358941769645e-06,
      "loss": 0.4831,
      "step": 7284
    },
    {
      "epoch": 0.5344974641635408,
      "grad_norm": 1.447752479618919,
      "learning_rate": 4.6841500614392145e-06,
      "loss": 0.8287,
      "step": 7285
    },
    {
      "epoch": 0.53457083375368,
      "grad_norm": 1.9196226515320667,
      "learning_rate": 4.682964246538291e-06,
      "loss": 0.7692,
      "step": 7286
    },
    {
      "epoch": 0.5346442033438191,
      "grad_norm": 1.6453596568228697,
      "learning_rate": 4.681778449541163e-06,
      "loss": 0.8962,
      "step": 7287
    },
    {
      "epoch": 0.5347175729339582,
      "grad_norm": 1.3588993962973182,
      "learning_rate": 4.680592670514793e-06,
      "loss": 0.7825,
      "step": 7288
    },
    {
      "epoch": 0.5347175729339582,
      "lm_loss": 1.5323,
      "step": 7288,
      "vm_loss": 0.1318
    },
    {
      "epoch": 0.5347175729339582,
      "lm_loss": 0.0113,
      "step": 7288,
      "vm_loss": 0.1135
    },
    {
      "epoch": 0.5347175729339582,
      "lm_loss": 0.1719,
      "step": 7288,
      "vm_loss": 0.1812
    },
    {
      "epoch": 0.5347175729339582,
      "lm_loss": 0.1103,
      "step": 7288,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.5347175729339582,
      "lm_loss": 0.8219,
      "step": 7288,
      "vm_loss": 0.2547
    },
    {
      "epoch": 0.5347175729339582,
      "lm_loss": 0.8872,
      "step": 7288,
      "vm_loss": 0.0964
    },
    {
      "epoch": 0.5347175729339582,
      "lm_loss": 0.1835,
      "step": 7288,
      "vm_loss": 0.1739
    },
    {
      "epoch": 0.5347175729339582,
      "lm_loss": 0.2431,
      "step": 7288,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.5347909425240973,
      "grad_norm": 1.3760280555348117,
      "learning_rate": 4.679406909526147e-06,
      "loss": 0.7344,
      "step": 7289
    },
    {
      "epoch": 0.5348643121142365,
      "grad_norm": 1.3353070099386821,
      "learning_rate": 4.678221166642188e-06,
      "loss": 0.7384,
      "step": 7290
    },
    {
      "epoch": 0.5349376817043756,
      "grad_norm": 1.6815846539615684,
      "learning_rate": 4.677035441929873e-06,
      "loss": 0.7903,
      "step": 7291
    },
    {
      "epoch": 0.5350110512945148,
      "grad_norm": 1.5915559568049353,
      "learning_rate": 4.6758497354561695e-06,
      "loss": 0.7111,
      "step": 7292
    },
    {
      "epoch": 0.5350844208846538,
      "grad_norm": 1.5945169910215728,
      "learning_rate": 4.674664047288034e-06,
      "loss": 0.7589,
      "step": 7293
    },
    {
      "epoch": 0.535157790474793,
      "grad_norm": 1.5000536443410613,
      "learning_rate": 4.673478377492423e-06,
      "loss": 0.8955,
      "step": 7294
    },
    {
      "epoch": 0.5352311600649321,
      "grad_norm": 0.6272043555788853,
      "learning_rate": 4.672292726136298e-06,
      "loss": 0.5021,
      "step": 7295
    },
    {
      "epoch": 0.5353045296550712,
      "grad_norm": 1.641279822837606,
      "learning_rate": 4.671107093286615e-06,
      "loss": 0.75,
      "step": 7296
    },
    {
      "epoch": 0.5353045296550712,
      "lm_loss": 0.6171,
      "step": 7296,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.5353045296550712,
      "lm_loss": 1.5006,
      "step": 7296,
      "vm_loss": 0.1507
    },
    {
      "epoch": 0.5353045296550712,
      "lm_loss": 0.8888,
      "step": 7296,
      "vm_loss": 0.101
    },
    {
      "epoch": 0.5353045296550712,
      "lm_loss": 0.9547,
      "step": 7296,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.5353045296550712,
      "lm_loss": 1.3576,
      "step": 7296,
      "vm_loss": 0.2171
    },
    {
      "epoch": 0.5353045296550712,
      "lm_loss": 0.0526,
      "step": 7296,
      "vm_loss": 0.1077
    },
    {
      "epoch": 0.5353045296550712,
      "lm_loss": 0.1859,
      "step": 7296,
      "vm_loss": 0.1245
    },
    {
      "epoch": 0.5353045296550712,
      "lm_loss": 0.426,
      "step": 7296,
      "vm_loss": 0.05
    },
    {
      "epoch": 0.5353778992452103,
      "grad_norm": 1.6502988080181258,
      "learning_rate": 4.669921479010328e-06,
      "loss": 0.7687,
      "step": 7297
    },
    {
      "epoch": 0.5354512688353494,
      "grad_norm": 1.470068730186851,
      "learning_rate": 4.668735883374393e-06,
      "loss": 0.704,
      "step": 7298
    },
    {
      "epoch": 0.5355246384254886,
      "grad_norm": 1.3905923647360896,
      "learning_rate": 4.667550306445762e-06,
      "loss": 0.7959,
      "step": 7299
    },
    {
      "epoch": 0.5355980080156277,
      "grad_norm": 0.5422284390794935,
      "learning_rate": 4.66636474829139e-06,
      "loss": 0.4835,
      "step": 7300
    },
    {
      "epoch": 0.5356713776057669,
      "grad_norm": 0.5559502742079856,
      "learning_rate": 4.665179208978226e-06,
      "loss": 0.4887,
      "step": 7301
    },
    {
      "epoch": 0.5357447471959059,
      "grad_norm": 1.601036873993621,
      "learning_rate": 4.663993688573222e-06,
      "loss": 0.798,
      "step": 7302
    },
    {
      "epoch": 0.5358181167860451,
      "grad_norm": 0.5601590917526734,
      "learning_rate": 4.662808187143326e-06,
      "loss": 0.4944,
      "step": 7303
    },
    {
      "epoch": 0.5358914863761842,
      "grad_norm": 1.589126756658386,
      "learning_rate": 4.661622704755487e-06,
      "loss": 0.7894,
      "step": 7304
    },
    {
      "epoch": 0.5358914863761842,
      "lm_loss": 0.5497,
      "step": 7304,
      "vm_loss": 0.0505
    },
    {
      "epoch": 0.5358914863761842,
      "lm_loss": 0.7325,
      "step": 7304,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.5358914863761842,
      "lm_loss": 0.2848,
      "step": 7304,
      "vm_loss": 0.1074
    },
    {
      "epoch": 0.5358914863761842,
      "lm_loss": 0.7356,
      "step": 7304,
      "vm_loss": 0.1855
    },
    {
      "epoch": 0.5358914863761842,
      "lm_loss": 0.5304,
      "step": 7304,
      "vm_loss": 0.0747
    },
    {
      "epoch": 0.5358914863761842,
      "lm_loss": 0.8728,
      "step": 7304,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.5358914863761842,
      "lm_loss": 1.2373,
      "step": 7304,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.5358914863761842,
      "lm_loss": 0.912,
      "step": 7304,
      "vm_loss": 0.1059
    },
    {
      "epoch": 0.5359648559663234,
      "grad_norm": 1.4908259449696553,
      "learning_rate": 4.660437241476651e-06,
      "loss": 0.8796,
      "step": 7305
    },
    {
      "epoch": 0.5360382255564625,
      "grad_norm": 1.511654669080476,
      "learning_rate": 4.659251797373766e-06,
      "loss": 0.7582,
      "step": 7306
    },
    {
      "epoch": 0.5361115951466016,
      "grad_norm": 0.609777072823099,
      "learning_rate": 4.6580663725137745e-06,
      "loss": 0.4888,
      "step": 7307
    },
    {
      "epoch": 0.5361849647367407,
      "grad_norm": 0.5770944403775701,
      "learning_rate": 4.656880966963623e-06,
      "loss": 0.4981,
      "step": 7308
    },
    {
      "epoch": 0.5362583343268799,
      "grad_norm": 0.5490217380259442,
      "learning_rate": 4.655695580790254e-06,
      "loss": 0.4728,
      "step": 7309
    },
    {
      "epoch": 0.536331703917019,
      "grad_norm": 1.5586758443414663,
      "learning_rate": 4.654510214060605e-06,
      "loss": 0.7274,
      "step": 7310
    },
    {
      "epoch": 0.5364050735071582,
      "grad_norm": 1.3780397406375986,
      "learning_rate": 4.653324866841622e-06,
      "loss": 0.7306,
      "step": 7311
    },
    {
      "epoch": 0.5364784430972972,
      "grad_norm": 1.6582425927410087,
      "learning_rate": 4.652139539200242e-06,
      "loss": 0.7747,
      "step": 7312
    },
    {
      "epoch": 0.5364784430972972,
      "lm_loss": 0.4018,
      "step": 7312,
      "vm_loss": 0.0731
    },
    {
      "epoch": 0.5364784430972972,
      "lm_loss": 0.3576,
      "step": 7312,
      "vm_loss": 0.1086
    },
    {
      "epoch": 0.5364784430972972,
      "lm_loss": 0.4777,
      "step": 7312,
      "vm_loss": 0.137
    },
    {
      "epoch": 0.5364784430972972,
      "lm_loss": 0.3029,
      "step": 7312,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.5364784430972972,
      "lm_loss": 0.5798,
      "step": 7312,
      "vm_loss": 0.0805
    },
    {
      "epoch": 0.5364784430972972,
      "lm_loss": 0.1712,
      "step": 7312,
      "vm_loss": 0.0801
    },
    {
      "epoch": 0.5364784430972972,
      "lm_loss": 0.0219,
      "step": 7312,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.5364784430972972,
      "lm_loss": 0.5748,
      "step": 7312,
      "vm_loss": 0.1093
    },
    {
      "epoch": 0.5365518126874363,
      "grad_norm": 1.471111887985226,
      "learning_rate": 4.650954231203404e-06,
      "loss": 0.6285,
      "step": 7313
    },
    {
      "epoch": 0.5366251822775755,
      "grad_norm": 1.5797768810543065,
      "learning_rate": 4.649768942918044e-06,
      "loss": 0.7793,
      "step": 7314
    },
    {
      "epoch": 0.5366985518677146,
      "grad_norm": 0.6261734307632835,
      "learning_rate": 4.648583674411099e-06,
      "loss": 0.5022,
      "step": 7315
    },
    {
      "epoch": 0.5367719214578538,
      "grad_norm": 1.6156668654495203,
      "learning_rate": 4.647398425749505e-06,
      "loss": 0.7486,
      "step": 7316
    },
    {
      "epoch": 0.5368452910479928,
      "grad_norm": 0.5953564999917258,
      "learning_rate": 4.646213197000196e-06,
      "loss": 0.4803,
      "step": 7317
    },
    {
      "epoch": 0.536918660638132,
      "grad_norm": 0.5843335044303062,
      "learning_rate": 4.645027988230102e-06,
      "loss": 0.4958,
      "step": 7318
    },
    {
      "epoch": 0.5369920302282711,
      "grad_norm": 0.5338440068669855,
      "learning_rate": 4.6438427995061575e-06,
      "loss": 0.4793,
      "step": 7319
    },
    {
      "epoch": 0.5370653998184103,
      "grad_norm": 1.6441780843383287,
      "learning_rate": 4.642657630895293e-06,
      "loss": 0.8628,
      "step": 7320
    },
    {
      "epoch": 0.5370653998184103,
      "lm_loss": 0.4802,
      "step": 7320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5370653998184103,
      "lm_loss": 0.4896,
      "step": 7320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5370653998184103,
      "lm_loss": 0.4315,
      "step": 7320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5370653998184103,
      "lm_loss": 0.4567,
      "step": 7320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5370653998184103,
      "lm_loss": 0.6076,
      "step": 7320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5370653998184103,
      "lm_loss": 0.4473,
      "step": 7320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5370653998184103,
      "lm_loss": 0.3869,
      "step": 7320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5370653998184103,
      "lm_loss": 0.689,
      "step": 7320,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5371387694085494,
      "grad_norm": 0.5735767910551197,
      "learning_rate": 4.641472482464434e-06,
      "loss": 0.4871,
      "step": 7321
    },
    {
      "epoch": 0.5372121389986885,
      "grad_norm": 1.469881511606471,
      "learning_rate": 4.640287354280515e-06,
      "loss": 0.7007,
      "step": 7322
    },
    {
      "epoch": 0.5372855085888276,
      "grad_norm": 1.396813917127261,
      "learning_rate": 4.6391022464104595e-06,
      "loss": 0.7453,
      "step": 7323
    },
    {
      "epoch": 0.5373588781789668,
      "grad_norm": 1.3976500423523142,
      "learning_rate": 4.637917158921192e-06,
      "loss": 0.72,
      "step": 7324
    },
    {
      "epoch": 0.5374322477691059,
      "grad_norm": 1.5024294800090436,
      "learning_rate": 4.636732091879641e-06,
      "loss": 0.8244,
      "step": 7325
    },
    {
      "epoch": 0.5375056173592451,
      "grad_norm": 1.4554629284436293,
      "learning_rate": 4.635547045352728e-06,
      "loss": 0.7819,
      "step": 7326
    },
    {
      "epoch": 0.5375789869493841,
      "grad_norm": 1.4945068587597785,
      "learning_rate": 4.634362019407375e-06,
      "loss": 0.7278,
      "step": 7327
    },
    {
      "epoch": 0.5376523565395233,
      "grad_norm": 0.7123134536694642,
      "learning_rate": 4.633177014110505e-06,
      "loss": 0.5139,
      "step": 7328
    },
    {
      "epoch": 0.5376523565395233,
      "lm_loss": 1.3745,
      "step": 7328,
      "vm_loss": 0.0524
    },
    {
      "epoch": 0.5376523565395233,
      "lm_loss": 0.1574,
      "step": 7328,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.5376523565395233,
      "lm_loss": 0.7626,
      "step": 7328,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.5376523565395233,
      "lm_loss": 0.8774,
      "step": 7328,
      "vm_loss": 0.2035
    },
    {
      "epoch": 0.5376523565395233,
      "lm_loss": 0.601,
      "step": 7328,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.5376523565395233,
      "lm_loss": 0.2817,
      "step": 7328,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.5376523565395233,
      "lm_loss": 0.069,
      "step": 7328,
      "vm_loss": 0.121
    },
    {
      "epoch": 0.5376523565395233,
      "lm_loss": 0.5724,
      "step": 7328,
      "vm_loss": 0.0815
    },
    {
      "epoch": 0.5377257261296624,
      "grad_norm": 1.6306511678065758,
      "learning_rate": 4.631992029529037e-06,
      "loss": 0.7353,
      "step": 7329
    },
    {
      "epoch": 0.5377990957198016,
      "grad_norm": 1.6609104205194964,
      "learning_rate": 4.630807065729891e-06,
      "loss": 0.7034,
      "step": 7330
    },
    {
      "epoch": 0.5378724653099407,
      "grad_norm": 1.5384252277102661,
      "learning_rate": 4.6296221227799845e-06,
      "loss": 0.7899,
      "step": 7331
    },
    {
      "epoch": 0.5379458349000797,
      "grad_norm": 1.5888949685606968,
      "learning_rate": 4.628437200746233e-06,
      "loss": 0.8474,
      "step": 7332
    },
    {
      "epoch": 0.5380192044902189,
      "grad_norm": 1.4621181797404335,
      "learning_rate": 4.627252299695553e-06,
      "loss": 0.8481,
      "step": 7333
    },
    {
      "epoch": 0.538092574080358,
      "grad_norm": 0.588801124141456,
      "learning_rate": 4.626067419694859e-06,
      "loss": 0.5221,
      "step": 7334
    },
    {
      "epoch": 0.5381659436704972,
      "grad_norm": 1.43862465187047,
      "learning_rate": 4.624882560811065e-06,
      "loss": 0.753,
      "step": 7335
    },
    {
      "epoch": 0.5382393132606363,
      "grad_norm": 1.4550989672345376,
      "learning_rate": 4.623697723111081e-06,
      "loss": 0.7774,
      "step": 7336
    },
    {
      "epoch": 0.5382393132606363,
      "lm_loss": 1.9668,
      "step": 7336,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.5382393132606363,
      "lm_loss": 0.0684,
      "step": 7336,
      "vm_loss": 0.0551
    },
    {
      "epoch": 0.5382393132606363,
      "lm_loss": 0.3155,
      "step": 7336,
      "vm_loss": 0.2021
    },
    {
      "epoch": 0.5382393132606363,
      "lm_loss": 0.7766,
      "step": 7336,
      "vm_loss": 0.1768
    },
    {
      "epoch": 0.5382393132606363,
      "lm_loss": 0.7171,
      "step": 7336,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.5382393132606363,
      "lm_loss": 0.5145,
      "step": 7336,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.5382393132606363,
      "lm_loss": 0.1358,
      "step": 7336,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.5382393132606363,
      "lm_loss": 0.3656,
      "step": 7336,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.5383126828507754,
      "grad_norm": 1.6648625560448351,
      "learning_rate": 4.6225129066618176e-06,
      "loss": 0.7913,
      "step": 7337
    },
    {
      "epoch": 0.5383860524409145,
      "grad_norm": 1.4304180801707778,
      "learning_rate": 4.621328111530187e-06,
      "loss": 0.75,
      "step": 7338
    },
    {
      "epoch": 0.5384594220310537,
      "grad_norm": 1.4198375406260841,
      "learning_rate": 4.620143337783095e-06,
      "loss": 0.7395,
      "step": 7339
    },
    {
      "epoch": 0.5385327916211928,
      "grad_norm": 1.416673973860984,
      "learning_rate": 4.618958585487447e-06,
      "loss": 0.7204,
      "step": 7340
    },
    {
      "epoch": 0.538606161211332,
      "grad_norm": 0.598401026747327,
      "learning_rate": 4.617773854710154e-06,
      "loss": 0.4879,
      "step": 7341
    },
    {
      "epoch": 0.538679530801471,
      "grad_norm": 0.580766556069825,
      "learning_rate": 4.616589145518116e-06,
      "loss": 0.4923,
      "step": 7342
    },
    {
      "epoch": 0.5387529003916102,
      "grad_norm": 0.5696487787647475,
      "learning_rate": 4.615404457978241e-06,
      "loss": 0.4909,
      "step": 7343
    },
    {
      "epoch": 0.5388262699817493,
      "grad_norm": 0.5657564220052475,
      "learning_rate": 4.6142197921574274e-06,
      "loss": 0.5139,
      "step": 7344
    },
    {
      "epoch": 0.5388262699817493,
      "lm_loss": 0.7265,
      "step": 7344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5388262699817493,
      "lm_loss": 0.4209,
      "step": 7344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5388262699817493,
      "lm_loss": 0.4949,
      "step": 7344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5388262699817493,
      "lm_loss": 0.397,
      "step": 7344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5388262699817493,
      "lm_loss": 0.4836,
      "step": 7344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5388262699817493,
      "lm_loss": 0.544,
      "step": 7344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5388262699817493,
      "lm_loss": 0.443,
      "step": 7344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5388262699817493,
      "lm_loss": 0.4831,
      "step": 7344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5388996395718885,
      "grad_norm": 0.5796055818593407,
      "learning_rate": 4.613035148122577e-06,
      "loss": 0.4996,
      "step": 7345
    },
    {
      "epoch": 0.5389730091620276,
      "grad_norm": 1.615217490224694,
      "learning_rate": 4.611850525940592e-06,
      "loss": 0.7692,
      "step": 7346
    },
    {
      "epoch": 0.5390463787521667,
      "grad_norm": 1.6175607122032232,
      "learning_rate": 4.610665925678367e-06,
      "loss": 0.8633,
      "step": 7347
    },
    {
      "epoch": 0.5391197483423058,
      "grad_norm": 0.6048126177694145,
      "learning_rate": 4.609481347402801e-06,
      "loss": 0.512,
      "step": 7348
    },
    {
      "epoch": 0.539193117932445,
      "grad_norm": 1.7025265712841906,
      "learning_rate": 4.608296791180793e-06,
      "loss": 0.7473,
      "step": 7349
    },
    {
      "epoch": 0.5392664875225841,
      "grad_norm": 0.5802518784453011,
      "learning_rate": 4.607112257079233e-06,
      "loss": 0.4971,
      "step": 7350
    },
    {
      "epoch": 0.5393398571127233,
      "grad_norm": 1.4952690133302187,
      "learning_rate": 4.605927745165018e-06,
      "loss": 0.7324,
      "step": 7351
    },
    {
      "epoch": 0.5394132267028623,
      "grad_norm": 1.732902744661392,
      "learning_rate": 4.604743255505038e-06,
      "loss": 0.7605,
      "step": 7352
    },
    {
      "epoch": 0.5394132267028623,
      "lm_loss": 1.321,
      "step": 7352,
      "vm_loss": 0.0572
    },
    {
      "epoch": 0.5394132267028623,
      "lm_loss": 0.0565,
      "step": 7352,
      "vm_loss": 0.0716
    },
    {
      "epoch": 0.5394132267028623,
      "lm_loss": 0.1505,
      "step": 7352,
      "vm_loss": 0.0924
    },
    {
      "epoch": 0.5394132267028623,
      "lm_loss": 0.0379,
      "step": 7352,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.5394132267028623,
      "lm_loss": 0.7058,
      "step": 7352,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.5394132267028623,
      "lm_loss": 0.4647,
      "step": 7352,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.5394132267028623,
      "lm_loss": 0.8258,
      "step": 7352,
      "vm_loss": 0.1257
    },
    {
      "epoch": 0.5394132267028623,
      "lm_loss": 0.4742,
      "step": 7352,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.5394865962930014,
      "grad_norm": 1.4222495448334374,
      "learning_rate": 4.603558788166186e-06,
      "loss": 0.7263,
      "step": 7353
    },
    {
      "epoch": 0.5395599658831406,
      "grad_norm": 0.5900897506100581,
      "learning_rate": 4.6023743432153505e-06,
      "loss": 0.4958,
      "step": 7354
    },
    {
      "epoch": 0.5396333354732797,
      "grad_norm": 0.5662610495318402,
      "learning_rate": 4.6011899207194215e-06,
      "loss": 0.4963,
      "step": 7355
    },
    {
      "epoch": 0.5397067050634189,
      "grad_norm": 0.5672092571837101,
      "learning_rate": 4.6000055207452845e-06,
      "loss": 0.486,
      "step": 7356
    },
    {
      "epoch": 0.5397800746535579,
      "grad_norm": 1.4256639932312638,
      "learning_rate": 4.5988211433598264e-06,
      "loss": 0.6859,
      "step": 7357
    },
    {
      "epoch": 0.5398534442436971,
      "grad_norm": 1.6372393076295564,
      "learning_rate": 4.597636788629932e-06,
      "loss": 0.7273,
      "step": 7358
    },
    {
      "epoch": 0.5399268138338362,
      "grad_norm": 1.363052879231041,
      "learning_rate": 4.596452456622484e-06,
      "loss": 0.7693,
      "step": 7359
    },
    {
      "epoch": 0.5400001834239754,
      "grad_norm": 0.5703467610428774,
      "learning_rate": 4.5952681474043655e-06,
      "loss": 0.4852,
      "step": 7360
    },
    {
      "epoch": 0.5400001834239754,
      "lm_loss": 1.2706,
      "step": 7360,
      "vm_loss": 0.0419
    },
    {
      "epoch": 0.5400001834239754,
      "lm_loss": 0.0516,
      "step": 7360,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.5400001834239754,
      "lm_loss": 0.8191,
      "step": 7360,
      "vm_loss": 0.1993
    },
    {
      "epoch": 0.5400001834239754,
      "lm_loss": 0.0251,
      "step": 7360,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.5400001834239754,
      "lm_loss": 0.3673,
      "step": 7360,
      "vm_loss": 0.0591
    },
    {
      "epoch": 0.5400001834239754,
      "lm_loss": 0.1951,
      "step": 7360,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.5400001834239754,
      "lm_loss": 0.3274,
      "step": 7360,
      "vm_loss": 0.1721
    },
    {
      "epoch": 0.5400001834239754,
      "lm_loss": 0.0824,
      "step": 7360,
      "vm_loss": 0.0724
    },
    {
      "epoch": 0.5400735530141145,
      "grad_norm": 1.4791531220874363,
      "learning_rate": 4.594083861042457e-06,
      "loss": 0.7455,
      "step": 7361
    },
    {
      "epoch": 0.5401469226042536,
      "grad_norm": 1.457146043460671,
      "learning_rate": 4.592899597603639e-06,
      "loss": 0.7424,
      "step": 7362
    },
    {
      "epoch": 0.5402202921943927,
      "grad_norm": 1.468022098319688,
      "learning_rate": 4.591715357154789e-06,
      "loss": 0.7584,
      "step": 7363
    },
    {
      "epoch": 0.5402936617845319,
      "grad_norm": 0.5646181209365895,
      "learning_rate": 4.590531139762782e-06,
      "loss": 0.4809,
      "step": 7364
    },
    {
      "epoch": 0.540367031374671,
      "grad_norm": 0.5747340870088478,
      "learning_rate": 4.589346945494497e-06,
      "loss": 0.4728,
      "step": 7365
    },
    {
      "epoch": 0.5404404009648102,
      "grad_norm": 1.592534810433137,
      "learning_rate": 4.5881627744168065e-06,
      "loss": 0.7738,
      "step": 7366
    },
    {
      "epoch": 0.5405137705549492,
      "grad_norm": 1.6084770132424466,
      "learning_rate": 4.5869786265965836e-06,
      "loss": 0.8075,
      "step": 7367
    },
    {
      "epoch": 0.5405871401450884,
      "grad_norm": 1.6584228156375393,
      "learning_rate": 4.585794502100701e-06,
      "loss": 0.8875,
      "step": 7368
    },
    {
      "epoch": 0.5405871401450884,
      "lm_loss": 1.4657,
      "step": 7368,
      "vm_loss": 0.0486
    },
    {
      "epoch": 0.5405871401450884,
      "lm_loss": 1.0658,
      "step": 7368,
      "vm_loss": 0.107
    },
    {
      "epoch": 0.5405871401450884,
      "lm_loss": 0.1915,
      "step": 7368,
      "vm_loss": 0.1423
    },
    {
      "epoch": 0.5405871401450884,
      "lm_loss": 0.9076,
      "step": 7368,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.5405871401450884,
      "lm_loss": 0.0539,
      "step": 7368,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.5405871401450884,
      "lm_loss": 0.0352,
      "step": 7368,
      "vm_loss": 0.1493
    },
    {
      "epoch": 0.5405871401450884,
      "lm_loss": 0.1602,
      "step": 7368,
      "vm_loss": 0.0635
    },
    {
      "epoch": 0.5405871401450884,
      "lm_loss": 0.1303,
      "step": 7368,
      "vm_loss": 0.1256
    },
    {
      "epoch": 0.5406605097352275,
      "grad_norm": 1.550822629381221,
      "learning_rate": 4.584610400996028e-06,
      "loss": 0.7477,
      "step": 7369
    },
    {
      "epoch": 0.5407338793253667,
      "grad_norm": 1.4575589420077355,
      "learning_rate": 4.583426323349435e-06,
      "loss": 0.7372,
      "step": 7370
    },
    {
      "epoch": 0.5408072489155058,
      "grad_norm": 1.4144818542692943,
      "learning_rate": 4.58224226922779e-06,
      "loss": 0.7596,
      "step": 7371
    },
    {
      "epoch": 0.5408806185056448,
      "grad_norm": 1.5245705079666485,
      "learning_rate": 4.581058238697957e-06,
      "loss": 0.7437,
      "step": 7372
    },
    {
      "epoch": 0.540953988095784,
      "grad_norm": 1.5080459496018044,
      "learning_rate": 4.579874231826804e-06,
      "loss": 0.8385,
      "step": 7373
    },
    {
      "epoch": 0.5410273576859231,
      "grad_norm": 1.2903737362954089,
      "learning_rate": 4.578690248681193e-06,
      "loss": 0.7056,
      "step": 7374
    },
    {
      "epoch": 0.5411007272760623,
      "grad_norm": 0.6192077435841844,
      "learning_rate": 4.5775062893279865e-06,
      "loss": 0.523,
      "step": 7375
    },
    {
      "epoch": 0.5411740968662013,
      "grad_norm": 1.5725420455135404,
      "learning_rate": 4.576322353834048e-06,
      "loss": 0.7709,
      "step": 7376
    },
    {
      "epoch": 0.5411740968662013,
      "lm_loss": 1.1039,
      "step": 7376,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.5411740968662013,
      "lm_loss": 0.1991,
      "step": 7376,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.5411740968662013,
      "lm_loss": 0.4326,
      "step": 7376,
      "vm_loss": 0.1746
    },
    {
      "epoch": 0.5411740968662013,
      "lm_loss": 0.7276,
      "step": 7376,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.5411740968662013,
      "lm_loss": 1.1819,
      "step": 7376,
      "vm_loss": 0.182
    },
    {
      "epoch": 0.5411740968662013,
      "lm_loss": 0.4587,
      "step": 7376,
      "vm_loss": 0.1345
    },
    {
      "epoch": 0.5411740968662013,
      "lm_loss": 0.6446,
      "step": 7376,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.5411740968662013,
      "lm_loss": 0.152,
      "step": 7376,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.5412474664563405,
      "grad_norm": 1.5831141356220695,
      "learning_rate": 4.5751384422662326e-06,
      "loss": 0.8028,
      "step": 7377
    },
    {
      "epoch": 0.5413208360464796,
      "grad_norm": 1.5006033173369129,
      "learning_rate": 4.573954554691403e-06,
      "loss": 0.8011,
      "step": 7378
    },
    {
      "epoch": 0.5413942056366188,
      "grad_norm": 1.3993179376002167,
      "learning_rate": 4.572770691176415e-06,
      "loss": 0.759,
      "step": 7379
    },
    {
      "epoch": 0.5414675752267579,
      "grad_norm": 1.479412900477901,
      "learning_rate": 4.571586851788124e-06,
      "loss": 0.7714,
      "step": 7380
    },
    {
      "epoch": 0.541540944816897,
      "grad_norm": 1.4542869712328852,
      "learning_rate": 4.570403036593385e-06,
      "loss": 0.8229,
      "step": 7381
    },
    {
      "epoch": 0.5416143144070361,
      "grad_norm": 1.523866741722856,
      "learning_rate": 4.56921924565905e-06,
      "loss": 0.7963,
      "step": 7382
    },
    {
      "epoch": 0.5416876839971753,
      "grad_norm": 1.3965634449452955,
      "learning_rate": 4.5680354790519695e-06,
      "loss": 0.6895,
      "step": 7383
    },
    {
      "epoch": 0.5417610535873144,
      "grad_norm": 1.6605701401134467,
      "learning_rate": 4.566851736838997e-06,
      "loss": 0.6773,
      "step": 7384
    },
    {
      "epoch": 0.5417610535873144,
      "lm_loss": 1.448,
      "step": 7384,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.5417610535873144,
      "lm_loss": 0.5648,
      "step": 7384,
      "vm_loss": 0.1489
    },
    {
      "epoch": 0.5417610535873144,
      "lm_loss": 0.5166,
      "step": 7384,
      "vm_loss": 0.1558
    },
    {
      "epoch": 0.5417610535873144,
      "lm_loss": 0.144,
      "step": 7384,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.5417610535873144,
      "lm_loss": 0.2061,
      "step": 7384,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.5417610535873144,
      "lm_loss": 0.0084,
      "step": 7384,
      "vm_loss": 0.182
    },
    {
      "epoch": 0.5417610535873144,
      "lm_loss": 0.6673,
      "step": 7384,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.5417610535873144,
      "lm_loss": 1.2555,
      "step": 7384,
      "vm_loss": 0.1244
    },
    {
      "epoch": 0.5418344231774536,
      "grad_norm": 1.4813161101330223,
      "learning_rate": 4.565668019086979e-06,
      "loss": 0.7195,
      "step": 7385
    },
    {
      "epoch": 0.5419077927675926,
      "grad_norm": 1.5319646864364613,
      "learning_rate": 4.564484325862763e-06,
      "loss": 0.7944,
      "step": 7386
    },
    {
      "epoch": 0.5419811623577317,
      "grad_norm": 0.5903975424696363,
      "learning_rate": 4.563300657233197e-06,
      "loss": 0.4787,
      "step": 7387
    },
    {
      "epoch": 0.5420545319478709,
      "grad_norm": 1.4639898324527885,
      "learning_rate": 4.562117013265123e-06,
      "loss": 0.8944,
      "step": 7388
    },
    {
      "epoch": 0.54212790153801,
      "grad_norm": 1.4999665215430062,
      "learning_rate": 4.560933394025386e-06,
      "loss": 0.7647,
      "step": 7389
    },
    {
      "epoch": 0.5422012711281492,
      "grad_norm": 1.5609705403328817,
      "learning_rate": 4.559749799580828e-06,
      "loss": 0.7033,
      "step": 7390
    },
    {
      "epoch": 0.5422746407182882,
      "grad_norm": 1.4756126584976064,
      "learning_rate": 4.5585662299982885e-06,
      "loss": 0.7861,
      "step": 7391
    },
    {
      "epoch": 0.5423480103084274,
      "grad_norm": 1.454346326135342,
      "learning_rate": 4.557382685344608e-06,
      "loss": 0.8249,
      "step": 7392
    },
    {
      "epoch": 0.5423480103084274,
      "lm_loss": 0.2558,
      "step": 7392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5423480103084274,
      "lm_loss": 0.5861,
      "step": 7392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5423480103084274,
      "lm_loss": 0.5295,
      "step": 7392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5423480103084274,
      "lm_loss": 0.6558,
      "step": 7392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5423480103084274,
      "lm_loss": 0.3972,
      "step": 7392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5423480103084274,
      "lm_loss": 0.4755,
      "step": 7392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5423480103084274,
      "lm_loss": 0.5425,
      "step": 7392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5423480103084274,
      "lm_loss": 0.4753,
      "step": 7392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5424213798985665,
      "grad_norm": 0.5532693163861536,
      "learning_rate": 4.5561991656866235e-06,
      "loss": 0.4852,
      "step": 7393
    },
    {
      "epoch": 0.5424947494887057,
      "grad_norm": 1.5214613109686206,
      "learning_rate": 4.55501567109117e-06,
      "loss": 0.7574,
      "step": 7394
    },
    {
      "epoch": 0.5425681190788448,
      "grad_norm": 1.5112548398008505,
      "learning_rate": 4.5538322016250855e-06,
      "loss": 0.7664,
      "step": 7395
    },
    {
      "epoch": 0.5426414886689839,
      "grad_norm": 1.6046730572831052,
      "learning_rate": 4.552648757355201e-06,
      "loss": 0.8639,
      "step": 7396
    },
    {
      "epoch": 0.542714858259123,
      "grad_norm": 1.6075230081106884,
      "learning_rate": 4.551465338348348e-06,
      "loss": 0.8273,
      "step": 7397
    },
    {
      "epoch": 0.5427882278492622,
      "grad_norm": 1.7442152199431602,
      "learning_rate": 4.5502819446713575e-06,
      "loss": 0.7649,
      "step": 7398
    },
    {
      "epoch": 0.5428615974394013,
      "grad_norm": 1.5573393891280956,
      "learning_rate": 4.549098576391061e-06,
      "loss": 0.7368,
      "step": 7399
    },
    {
      "epoch": 0.5429349670295405,
      "grad_norm": 1.3742644250069176,
      "learning_rate": 4.547915233574284e-06,
      "loss": 0.8028,
      "step": 7400
    },
    {
      "epoch": 0.5429349670295405,
      "lm_loss": 1.2888,
      "step": 7400,
      "vm_loss": 0.116
    },
    {
      "epoch": 0.5429349670295405,
      "lm_loss": 0.1698,
      "step": 7400,
      "vm_loss": 0.1456
    },
    {
      "epoch": 0.5429349670295405,
      "lm_loss": 0.7916,
      "step": 7400,
      "vm_loss": 0.0952
    },
    {
      "epoch": 0.5429349670295405,
      "lm_loss": 1.1666,
      "step": 7400,
      "vm_loss": 0.2257
    },
    {
      "epoch": 0.5429349670295405,
      "lm_loss": 0.1083,
      "step": 7400,
      "vm_loss": 0.0682
    },
    {
      "epoch": 0.5429349670295405,
      "lm_loss": 0.1598,
      "step": 7400,
      "vm_loss": 0.0582
    },
    {
      "epoch": 0.5429349670295405,
      "lm_loss": 0.1042,
      "step": 7400,
      "vm_loss": 0.1827
    },
    {
      "epoch": 0.5429349670295405,
      "lm_loss": 1.0309,
      "step": 7400,
      "vm_loss": 0.1784
    },
    {
      "epoch": 0.5430083366196795,
      "grad_norm": 1.415220851828527,
      "learning_rate": 4.5467319162878525e-06,
      "loss": 0.6902,
      "step": 7401
    },
    {
      "epoch": 0.5430817062098187,
      "grad_norm": 1.7217238211135981,
      "learning_rate": 4.545548624598592e-06,
      "loss": 0.8837,
      "step": 7402
    },
    {
      "epoch": 0.5431550757999578,
      "grad_norm": 1.6406902209855792,
      "learning_rate": 4.544365358573327e-06,
      "loss": 0.8524,
      "step": 7403
    },
    {
      "epoch": 0.543228445390097,
      "grad_norm": 1.5589914913312686,
      "learning_rate": 4.5431821182788785e-06,
      "loss": 0.8255,
      "step": 7404
    },
    {
      "epoch": 0.5433018149802361,
      "grad_norm": 0.6026297092044691,
      "learning_rate": 4.541998903782067e-06,
      "loss": 0.4819,
      "step": 7405
    },
    {
      "epoch": 0.5433751845703751,
      "grad_norm": 1.4482567678972902,
      "learning_rate": 4.540815715149711e-06,
      "loss": 0.7238,
      "step": 7406
    },
    {
      "epoch": 0.5434485541605143,
      "grad_norm": 1.4632621008011222,
      "learning_rate": 4.539632552448629e-06,
      "loss": 0.7633,
      "step": 7407
    },
    {
      "epoch": 0.5435219237506534,
      "grad_norm": 0.5435327598357034,
      "learning_rate": 4.538449415745638e-06,
      "loss": 0.4913,
      "step": 7408
    },
    {
      "epoch": 0.5435219237506534,
      "lm_loss": 0.5939,
      "step": 7408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5435219237506534,
      "lm_loss": 0.421,
      "step": 7408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5435219237506534,
      "lm_loss": 0.3982,
      "step": 7408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5435219237506534,
      "lm_loss": 0.5011,
      "step": 7408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5435219237506534,
      "lm_loss": 0.5488,
      "step": 7408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5435219237506534,
      "lm_loss": 0.5328,
      "step": 7408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5435219237506534,
      "lm_loss": 0.4313,
      "step": 7408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5435219237506534,
      "lm_loss": 0.3612,
      "step": 7408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5435952933407926,
      "grad_norm": 0.5650284283420152,
      "learning_rate": 4.537266305107549e-06,
      "loss": 0.4821,
      "step": 7409
    },
    {
      "epoch": 0.5436686629309317,
      "grad_norm": 1.5669424202847917,
      "learning_rate": 4.53608322060118e-06,
      "loss": 0.7864,
      "step": 7410
    },
    {
      "epoch": 0.5437420325210708,
      "grad_norm": 1.4882502298713014,
      "learning_rate": 4.534900162293339e-06,
      "loss": 0.8698,
      "step": 7411
    },
    {
      "epoch": 0.5438154021112099,
      "grad_norm": 0.5679269674469848,
      "learning_rate": 4.533717130250839e-06,
      "loss": 0.5159,
      "step": 7412
    },
    {
      "epoch": 0.5438887717013491,
      "grad_norm": 0.5590659274637838,
      "learning_rate": 4.5325341245404866e-06,
      "loss": 0.4887,
      "step": 7413
    },
    {
      "epoch": 0.5439621412914882,
      "grad_norm": 0.5494900870643947,
      "learning_rate": 4.53135114522909e-06,
      "loss": 0.4877,
      "step": 7414
    },
    {
      "epoch": 0.5440355108816274,
      "grad_norm": 0.55972017726915,
      "learning_rate": 4.530168192383455e-06,
      "loss": 0.5012,
      "step": 7415
    },
    {
      "epoch": 0.5441088804717664,
      "grad_norm": 1.6651418551804862,
      "learning_rate": 4.528985266070386e-06,
      "loss": 0.8277,
      "step": 7416
    },
    {
      "epoch": 0.5441088804717664,
      "lm_loss": 1.4004,
      "step": 7416,
      "vm_loss": 0.0748
    },
    {
      "epoch": 0.5441088804717664,
      "lm_loss": 1.2154,
      "step": 7416,
      "vm_loss": 0.191
    },
    {
      "epoch": 0.5441088804717664,
      "lm_loss": 0.7356,
      "step": 7416,
      "vm_loss": 0.1563
    },
    {
      "epoch": 0.5441088804717664,
      "lm_loss": 0.2872,
      "step": 7416,
      "vm_loss": 0.1844
    },
    {
      "epoch": 0.5441088804717664,
      "lm_loss": 0.7911,
      "step": 7416,
      "vm_loss": 0.1755
    },
    {
      "epoch": 0.5441088804717664,
      "lm_loss": 0.2885,
      "step": 7416,
      "vm_loss": 0.0819
    },
    {
      "epoch": 0.5441088804717664,
      "lm_loss": 0.1738,
      "step": 7416,
      "vm_loss": 0.1141
    },
    {
      "epoch": 0.5441088804717664,
      "lm_loss": 0.3605,
      "step": 7416,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.5441822500619056,
      "grad_norm": 1.523388732338769,
      "learning_rate": 4.527802366356684e-06,
      "loss": 0.7792,
      "step": 7417
    },
    {
      "epoch": 0.5442556196520447,
      "grad_norm": 1.7190114439974755,
      "learning_rate": 4.526619493309153e-06,
      "loss": 0.8673,
      "step": 7418
    },
    {
      "epoch": 0.5443289892421839,
      "grad_norm": 1.4731462017383699,
      "learning_rate": 4.525436646994591e-06,
      "loss": 0.621,
      "step": 7419
    },
    {
      "epoch": 0.544402358832323,
      "grad_norm": 0.5732857523921807,
      "learning_rate": 4.524253827479796e-06,
      "loss": 0.499,
      "step": 7420
    },
    {
      "epoch": 0.5444757284224621,
      "grad_norm": 2.0235700449743614,
      "learning_rate": 4.5230710348315675e-06,
      "loss": 0.8119,
      "step": 7421
    },
    {
      "epoch": 0.5445490980126012,
      "grad_norm": 1.623131889878724,
      "learning_rate": 4.521888269116697e-06,
      "loss": 0.7894,
      "step": 7422
    },
    {
      "epoch": 0.5446224676027404,
      "grad_norm": 1.442514539395991,
      "learning_rate": 4.520705530401982e-06,
      "loss": 0.7773,
      "step": 7423
    },
    {
      "epoch": 0.5446958371928795,
      "grad_norm": 1.6625058685150689,
      "learning_rate": 4.519522818754211e-06,
      "loss": 0.7456,
      "step": 7424
    },
    {
      "epoch": 0.5446958371928795,
      "lm_loss": 1.6597,
      "step": 7424,
      "vm_loss": 0.1107
    },
    {
      "epoch": 0.5446958371928795,
      "lm_loss": 0.2677,
      "step": 7424,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.5446958371928795,
      "lm_loss": 0.0508,
      "step": 7424,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.5446958371928795,
      "lm_loss": 1.0499,
      "step": 7424,
      "vm_loss": 0.0744
    },
    {
      "epoch": 0.5446958371928795,
      "lm_loss": 0.8003,
      "step": 7424,
      "vm_loss": 0.1618
    },
    {
      "epoch": 0.5446958371928795,
      "lm_loss": 0.6483,
      "step": 7424,
      "vm_loss": 0.067
    },
    {
      "epoch": 0.5446958371928795,
      "lm_loss": 0.0751,
      "step": 7424,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.5446958371928795,
      "lm_loss": 1.2095,
      "step": 7424,
      "vm_loss": 0.1874
    },
    {
      "epoch": 0.5447692067830187,
      "grad_norm": 1.9125530677073928,
      "learning_rate": 4.518340134240176e-06,
      "loss": 0.8012,
      "step": 7425
    },
    {
      "epoch": 0.5448425763731577,
      "grad_norm": 1.634608140757,
      "learning_rate": 4.517157476926668e-06,
      "loss": 0.6701,
      "step": 7426
    },
    {
      "epoch": 0.5449159459632968,
      "grad_norm": 1.6237310223051151,
      "learning_rate": 4.515974846880472e-06,
      "loss": 0.7359,
      "step": 7427
    },
    {
      "epoch": 0.544989315553436,
      "grad_norm": 0.6378151539448247,
      "learning_rate": 4.514792244168375e-06,
      "loss": 0.4931,
      "step": 7428
    },
    {
      "epoch": 0.5450626851435751,
      "grad_norm": 1.53514375949066,
      "learning_rate": 4.513609668857162e-06,
      "loss": 0.7309,
      "step": 7429
    },
    {
      "epoch": 0.5451360547337143,
      "grad_norm": 0.6155624189007198,
      "learning_rate": 4.512427121013614e-06,
      "loss": 0.5107,
      "step": 7430
    },
    {
      "epoch": 0.5452094243238533,
      "grad_norm": 1.6548854483884055,
      "learning_rate": 4.5112446007045155e-06,
      "loss": 0.82,
      "step": 7431
    },
    {
      "epoch": 0.5452827939139925,
      "grad_norm": 1.4769535126515898,
      "learning_rate": 4.510062107996645e-06,
      "loss": 0.8285,
      "step": 7432
    },
    {
      "epoch": 0.5452827939139925,
      "lm_loss": 0.3886,
      "step": 7432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5452827939139925,
      "lm_loss": 0.4828,
      "step": 7432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5452827939139925,
      "lm_loss": 0.4838,
      "step": 7432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5452827939139925,
      "lm_loss": 0.3265,
      "step": 7432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5452827939139925,
      "lm_loss": 0.6256,
      "step": 7432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5452827939139925,
      "lm_loss": 0.569,
      "step": 7432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5452827939139925,
      "lm_loss": 0.4612,
      "step": 7432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5452827939139925,
      "lm_loss": 0.3766,
      "step": 7432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5453561635041316,
      "grad_norm": 0.5567390611101302,
      "learning_rate": 4.508879642956779e-06,
      "loss": 0.4871,
      "step": 7433
    },
    {
      "epoch": 0.5454295330942708,
      "grad_norm": 1.654954170446982,
      "learning_rate": 4.507697205651698e-06,
      "loss": 0.7727,
      "step": 7434
    },
    {
      "epoch": 0.5455029026844099,
      "grad_norm": 1.4079265076738854,
      "learning_rate": 4.506514796148174e-06,
      "loss": 0.7113,
      "step": 7435
    },
    {
      "epoch": 0.545576272274549,
      "grad_norm": 1.5507672145384177,
      "learning_rate": 4.505332414512981e-06,
      "loss": 0.8623,
      "step": 7436
    },
    {
      "epoch": 0.5456496418646881,
      "grad_norm": 0.5634022877791761,
      "learning_rate": 4.504150060812892e-06,
      "loss": 0.5041,
      "step": 7437
    },
    {
      "epoch": 0.5457230114548273,
      "grad_norm": 1.4489644641797266,
      "learning_rate": 4.502967735114676e-06,
      "loss": 0.7851,
      "step": 7438
    },
    {
      "epoch": 0.5457963810449664,
      "grad_norm": 1.4036587685214086,
      "learning_rate": 4.5017854374851045e-06,
      "loss": 0.7933,
      "step": 7439
    },
    {
      "epoch": 0.5458697506351056,
      "grad_norm": 1.5973454021652458,
      "learning_rate": 4.500603167990943e-06,
      "loss": 0.7337,
      "step": 7440
    },
    {
      "epoch": 0.5458697506351056,
      "lm_loss": 1.4789,
      "step": 7440,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.5458697506351056,
      "lm_loss": 0.1344,
      "step": 7440,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.5458697506351056,
      "lm_loss": 0.0099,
      "step": 7440,
      "vm_loss": 0.1185
    },
    {
      "epoch": 0.5458697506351056,
      "lm_loss": 1.0021,
      "step": 7440,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.5458697506351056,
      "lm_loss": 0.9838,
      "step": 7440,
      "vm_loss": 0.0599
    },
    {
      "epoch": 0.5458697506351056,
      "lm_loss": 1.0884,
      "step": 7440,
      "vm_loss": 0.1798
    },
    {
      "epoch": 0.5458697506351056,
      "lm_loss": 0.138,
      "step": 7440,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.5458697506351056,
      "lm_loss": 0.3096,
      "step": 7440,
      "vm_loss": 0.1364
    },
    {
      "epoch": 0.5459431202252446,
      "grad_norm": 1.5489590222071932,
      "learning_rate": 4.499420926698956e-06,
      "loss": 0.764,
      "step": 7441
    },
    {
      "epoch": 0.5460164898153838,
      "grad_norm": 0.5710621118909817,
      "learning_rate": 4.49823871367591e-06,
      "loss": 0.484,
      "step": 7442
    },
    {
      "epoch": 0.5460898594055229,
      "grad_norm": 1.5362521383089789,
      "learning_rate": 4.4970565289885675e-06,
      "loss": 0.7409,
      "step": 7443
    },
    {
      "epoch": 0.546163228995662,
      "grad_norm": 1.4470902167399111,
      "learning_rate": 4.4958743727036856e-06,
      "loss": 0.829,
      "step": 7444
    },
    {
      "epoch": 0.5462365985858012,
      "grad_norm": 1.39292915562801,
      "learning_rate": 4.494692244888028e-06,
      "loss": 0.7565,
      "step": 7445
    },
    {
      "epoch": 0.5463099681759402,
      "grad_norm": 1.579571176004589,
      "learning_rate": 4.493510145608351e-06,
      "loss": 0.792,
      "step": 7446
    },
    {
      "epoch": 0.5463833377660794,
      "grad_norm": 1.5925018667070834,
      "learning_rate": 4.4923280749314094e-06,
      "loss": 0.665,
      "step": 7447
    },
    {
      "epoch": 0.5464567073562185,
      "grad_norm": 1.5096893745119926,
      "learning_rate": 4.491146032923959e-06,
      "loss": 0.7389,
      "step": 7448
    },
    {
      "epoch": 0.5464567073562185,
      "lm_loss": 0.9674,
      "step": 7448,
      "vm_loss": 0.0501
    },
    {
      "epoch": 0.5464567073562185,
      "lm_loss": 1.5605,
      "step": 7448,
      "vm_loss": 0.1683
    },
    {
      "epoch": 0.5464567073562185,
      "lm_loss": 0.4383,
      "step": 7448,
      "vm_loss": 0.121
    },
    {
      "epoch": 0.5464567073562185,
      "lm_loss": 0.4502,
      "step": 7448,
      "vm_loss": 0.1608
    },
    {
      "epoch": 0.5464567073562185,
      "lm_loss": 0.0983,
      "step": 7448,
      "vm_loss": 0.1281
    },
    {
      "epoch": 0.5464567073562185,
      "lm_loss": 0.7371,
      "step": 7448,
      "vm_loss": 0.1049
    },
    {
      "epoch": 0.5464567073562185,
      "lm_loss": 0.4973,
      "step": 7448,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.5464567073562185,
      "lm_loss": 1.0922,
      "step": 7448,
      "vm_loss": 0.1504
    },
    {
      "epoch": 0.5465300769463577,
      "grad_norm": 1.3104122675517331,
      "learning_rate": 4.489964019652752e-06,
      "loss": 0.7642,
      "step": 7449
    },
    {
      "epoch": 0.5466034465364968,
      "grad_norm": 0.5914904579616953,
      "learning_rate": 4.488782035184541e-06,
      "loss": 0.5099,
      "step": 7450
    },
    {
      "epoch": 0.5466768161266359,
      "grad_norm": 1.54270440138173,
      "learning_rate": 4.487600079586074e-06,
      "loss": 0.7298,
      "step": 7451
    },
    {
      "epoch": 0.546750185716775,
      "grad_norm": 0.5795591582393729,
      "learning_rate": 4.486418152924098e-06,
      "loss": 0.4967,
      "step": 7452
    },
    {
      "epoch": 0.5468235553069142,
      "grad_norm": 1.5196571298962414,
      "learning_rate": 4.4852362552653625e-06,
      "loss": 0.683,
      "step": 7453
    },
    {
      "epoch": 0.5468969248970533,
      "grad_norm": 1.5801565727810554,
      "learning_rate": 4.48405438667661e-06,
      "loss": 0.8529,
      "step": 7454
    },
    {
      "epoch": 0.5469702944871924,
      "grad_norm": 1.5082168331371155,
      "learning_rate": 4.4828725472245835e-06,
      "loss": 0.7129,
      "step": 7455
    },
    {
      "epoch": 0.5470436640773315,
      "grad_norm": 0.5658341106938723,
      "learning_rate": 4.481690736976025e-06,
      "loss": 0.4892,
      "step": 7456
    },
    {
      "epoch": 0.5470436640773315,
      "lm_loss": 0.5241,
      "step": 7456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5470436640773315,
      "lm_loss": 0.5379,
      "step": 7456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5470436640773315,
      "lm_loss": 0.4234,
      "step": 7456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5470436640773315,
      "lm_loss": 0.4577,
      "step": 7456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5470436640773315,
      "lm_loss": 0.4617,
      "step": 7456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5470436640773315,
      "lm_loss": 0.4986,
      "step": 7456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5470436640773315,
      "lm_loss": 0.5528,
      "step": 7456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5470436640773315,
      "lm_loss": 0.5094,
      "step": 7456,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5471170336674707,
      "grad_norm": 0.5613874051933744,
      "learning_rate": 4.4805089559976745e-06,
      "loss": 0.4812,
      "step": 7457
    },
    {
      "epoch": 0.5471904032576098,
      "grad_norm": 1.4695447348943111,
      "learning_rate": 4.479327204356269e-06,
      "loss": 0.8248,
      "step": 7458
    },
    {
      "epoch": 0.547263772847749,
      "grad_norm": 1.6336648494759305,
      "learning_rate": 4.478145482118547e-06,
      "loss": 0.7986,
      "step": 7459
    },
    {
      "epoch": 0.547337142437888,
      "grad_norm": 1.4476624691010562,
      "learning_rate": 4.476963789351241e-06,
      "loss": 0.7903,
      "step": 7460
    },
    {
      "epoch": 0.5474105120280272,
      "grad_norm": 1.4468392849977962,
      "learning_rate": 4.475782126121086e-06,
      "loss": 0.7243,
      "step": 7461
    },
    {
      "epoch": 0.5474838816181663,
      "grad_norm": 0.5545141570071183,
      "learning_rate": 4.474600492494813e-06,
      "loss": 0.5081,
      "step": 7462
    },
    {
      "epoch": 0.5475572512083055,
      "grad_norm": 0.5560997444182263,
      "learning_rate": 4.473418888539149e-06,
      "loss": 0.4852,
      "step": 7463
    },
    {
      "epoch": 0.5476306207984446,
      "grad_norm": 1.6601846102892899,
      "learning_rate": 4.472237314320826e-06,
      "loss": 0.8648,
      "step": 7464
    },
    {
      "epoch": 0.5476306207984446,
      "lm_loss": 1.0217,
      "step": 7464,
      "vm_loss": 0.0891
    },
    {
      "epoch": 0.5476306207984446,
      "lm_loss": 1.3866,
      "step": 7464,
      "vm_loss": 0.1136
    },
    {
      "epoch": 0.5476306207984446,
      "lm_loss": 0.8006,
      "step": 7464,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.5476306207984446,
      "lm_loss": 0.8287,
      "step": 7464,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.5476306207984446,
      "lm_loss": 0.2374,
      "step": 7464,
      "vm_loss": 0.0486
    },
    {
      "epoch": 0.5476306207984446,
      "lm_loss": 0.4083,
      "step": 7464,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.5476306207984446,
      "lm_loss": 0.5033,
      "step": 7464,
      "vm_loss": 0.1498
    },
    {
      "epoch": 0.5476306207984446,
      "lm_loss": 0.0527,
      "step": 7464,
      "vm_loss": 0.1444
    },
    {
      "epoch": 0.5477039903885836,
      "grad_norm": 1.846128701342836,
      "learning_rate": 4.471055769906569e-06,
      "loss": 0.7456,
      "step": 7465
    },
    {
      "epoch": 0.5477773599787228,
      "grad_norm": 1.3974689343569309,
      "learning_rate": 4.469874255363102e-06,
      "loss": 0.7148,
      "step": 7466
    },
    {
      "epoch": 0.5478507295688619,
      "grad_norm": 1.4688523316642748,
      "learning_rate": 4.468692770757149e-06,
      "loss": 0.7477,
      "step": 7467
    },
    {
      "epoch": 0.5479240991590011,
      "grad_norm": 1.8280060653061565,
      "learning_rate": 4.4675113161554305e-06,
      "loss": 0.7793,
      "step": 7468
    },
    {
      "epoch": 0.5479974687491402,
      "grad_norm": 0.5352205333514872,
      "learning_rate": 4.4663298916246665e-06,
      "loss": 0.4857,
      "step": 7469
    },
    {
      "epoch": 0.5480708383392793,
      "grad_norm": 1.4031413583291656,
      "learning_rate": 4.465148497231576e-06,
      "loss": 0.6935,
      "step": 7470
    },
    {
      "epoch": 0.5481442079294184,
      "grad_norm": 1.4787072919168556,
      "learning_rate": 4.4639671330428735e-06,
      "loss": 0.6534,
      "step": 7471
    },
    {
      "epoch": 0.5482175775195576,
      "grad_norm": 0.5473839763154466,
      "learning_rate": 4.462785799125275e-06,
      "loss": 0.4838,
      "step": 7472
    },
    {
      "epoch": 0.5482175775195576,
      "lm_loss": 1.1715,
      "step": 7472,
      "vm_loss": 0.0702
    },
    {
      "epoch": 0.5482175775195576,
      "lm_loss": 1.2486,
      "step": 7472,
      "vm_loss": 0.1685
    },
    {
      "epoch": 0.5482175775195576,
      "lm_loss": 0.4674,
      "step": 7472,
      "vm_loss": 0.112
    },
    {
      "epoch": 0.5482175775195576,
      "lm_loss": 1.0256,
      "step": 7472,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.5482175775195576,
      "lm_loss": 0.5775,
      "step": 7472,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.5482175775195576,
      "lm_loss": 0.8855,
      "step": 7472,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.5482175775195576,
      "lm_loss": 0.3611,
      "step": 7472,
      "vm_loss": 0.1466
    },
    {
      "epoch": 0.5482175775195576,
      "lm_loss": 0.5664,
      "step": 7472,
      "vm_loss": 0.0728
    },
    {
      "epoch": 0.5482909471096967,
      "grad_norm": 1.4338199501483448,
      "learning_rate": 4.461604495545493e-06,
      "loss": 0.8202,
      "step": 7473
    },
    {
      "epoch": 0.5483643166998359,
      "grad_norm": 1.4040203010718626,
      "learning_rate": 4.460423222370238e-06,
      "loss": 0.7085,
      "step": 7474
    },
    {
      "epoch": 0.5484376862899749,
      "grad_norm": 1.6555807414100834,
      "learning_rate": 4.4592419796662206e-06,
      "loss": 0.6884,
      "step": 7475
    },
    {
      "epoch": 0.5485110558801141,
      "grad_norm": 1.6738761992504314,
      "learning_rate": 4.458060767500147e-06,
      "loss": 0.7085,
      "step": 7476
    },
    {
      "epoch": 0.5485844254702532,
      "grad_norm": 1.615415989269743,
      "learning_rate": 4.456879585938725e-06,
      "loss": 0.7427,
      "step": 7477
    },
    {
      "epoch": 0.5486577950603924,
      "grad_norm": 1.5770324943992937,
      "learning_rate": 4.455698435048657e-06,
      "loss": 0.7392,
      "step": 7478
    },
    {
      "epoch": 0.5487311646505315,
      "grad_norm": 1.5848553665407745,
      "learning_rate": 4.4545173148966456e-06,
      "loss": 0.7095,
      "step": 7479
    },
    {
      "epoch": 0.5488045342406705,
      "grad_norm": 0.5787785046741265,
      "learning_rate": 4.453336225549393e-06,
      "loss": 0.4849,
      "step": 7480
    },
    {
      "epoch": 0.5488045342406705,
      "lm_loss": 0.3685,
      "step": 7480,
      "vm_loss": 0.0527
    },
    {
      "epoch": 0.5488045342406705,
      "lm_loss": 1.3213,
      "step": 7480,
      "vm_loss": 0.1995
    },
    {
      "epoch": 0.5488045342406705,
      "lm_loss": 0.0553,
      "step": 7480,
      "vm_loss": 0.1161
    },
    {
      "epoch": 0.5488045342406705,
      "lm_loss": 1.314,
      "step": 7480,
      "vm_loss": 0.1579
    },
    {
      "epoch": 0.5488045342406705,
      "lm_loss": 0.2068,
      "step": 7480,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.5488045342406705,
      "lm_loss": 0.2959,
      "step": 7480,
      "vm_loss": 0.1368
    },
    {
      "epoch": 0.5488045342406705,
      "lm_loss": 0.3375,
      "step": 7480,
      "vm_loss": 0.1235
    },
    {
      "epoch": 0.5488045342406705,
      "lm_loss": 0.8557,
      "step": 7480,
      "vm_loss": 0.0716
    },
    {
      "epoch": 0.5488779038308097,
      "grad_norm": 1.398963621221079,
      "learning_rate": 4.452155167073598e-06,
      "loss": 0.7716,
      "step": 7481
    },
    {
      "epoch": 0.5489512734209488,
      "grad_norm": 1.6437992561155634,
      "learning_rate": 4.450974139535956e-06,
      "loss": 0.8319,
      "step": 7482
    },
    {
      "epoch": 0.549024643011088,
      "grad_norm": 1.7762455190524076,
      "learning_rate": 4.449793143003166e-06,
      "loss": 0.699,
      "step": 7483
    },
    {
      "epoch": 0.5490980126012271,
      "grad_norm": 1.393723093688801,
      "learning_rate": 4.448612177541918e-06,
      "loss": 0.6925,
      "step": 7484
    },
    {
      "epoch": 0.5491713821913662,
      "grad_norm": 1.5012060065523722,
      "learning_rate": 4.447431243218906e-06,
      "loss": 0.7605,
      "step": 7485
    },
    {
      "epoch": 0.5492447517815053,
      "grad_norm": 1.355500689839531,
      "learning_rate": 4.4462503401008214e-06,
      "loss": 0.7798,
      "step": 7486
    },
    {
      "epoch": 0.5493181213716445,
      "grad_norm": 1.4751134718254422,
      "learning_rate": 4.44506946825435e-06,
      "loss": 0.7663,
      "step": 7487
    },
    {
      "epoch": 0.5493914909617836,
      "grad_norm": 1.2648242307210837,
      "learning_rate": 4.443888627746181e-06,
      "loss": 0.612,
      "step": 7488
    },
    {
      "epoch": 0.5493914909617836,
      "lm_loss": 0.0805,
      "step": 7488,
      "vm_loss": 0.0701
    },
    {
      "epoch": 0.5493914909617836,
      "lm_loss": 1.48,
      "step": 7488,
      "vm_loss": 0.1709
    },
    {
      "epoch": 0.5493914909617836,
      "lm_loss": 0.0197,
      "step": 7488,
      "vm_loss": 0.0408
    },
    {
      "epoch": 0.5493914909617836,
      "lm_loss": 0.288,
      "step": 7488,
      "vm_loss": 0.0907
    },
    {
      "epoch": 0.5493914909617836,
      "lm_loss": 0.3132,
      "step": 7488,
      "vm_loss": 0.1513
    },
    {
      "epoch": 0.5493914909617836,
      "lm_loss": 0.2551,
      "step": 7488,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.5493914909617836,
      "lm_loss": 0.0689,
      "step": 7488,
      "vm_loss": 0.0719
    },
    {
      "epoch": 0.5493914909617836,
      "lm_loss": 0.0925,
      "step": 7488,
      "vm_loss": 0.1259
    },
    {
      "epoch": 0.5494648605519228,
      "grad_norm": 1.4602702092945656,
      "learning_rate": 4.442707818642999e-06,
      "loss": 0.7092,
      "step": 7489
    },
    {
      "epoch": 0.5495382301420618,
      "grad_norm": 0.6053422116303303,
      "learning_rate": 4.441527041011485e-06,
      "loss": 0.4881,
      "step": 7490
    },
    {
      "epoch": 0.549611599732201,
      "grad_norm": 1.4700717843681637,
      "learning_rate": 4.440346294918323e-06,
      "loss": 0.7833,
      "step": 7491
    },
    {
      "epoch": 0.5496849693223401,
      "grad_norm": 1.4171505838493081,
      "learning_rate": 4.43916558043019e-06,
      "loss": 0.7154,
      "step": 7492
    },
    {
      "epoch": 0.5497583389124793,
      "grad_norm": 1.7635429791149575,
      "learning_rate": 4.437984897613768e-06,
      "loss": 0.7581,
      "step": 7493
    },
    {
      "epoch": 0.5498317085026184,
      "grad_norm": 1.412352127990516,
      "learning_rate": 4.43680424653573e-06,
      "loss": 0.7341,
      "step": 7494
    },
    {
      "epoch": 0.5499050780927575,
      "grad_norm": 0.5821728229620828,
      "learning_rate": 4.435623627262749e-06,
      "loss": 0.5206,
      "step": 7495
    },
    {
      "epoch": 0.5499784476828966,
      "grad_norm": 2.187688110751218,
      "learning_rate": 4.434443039861502e-06,
      "loss": 0.7444,
      "step": 7496
    },
    {
      "epoch": 0.5499784476828966,
      "lm_loss": 1.2061,
      "step": 7496,
      "vm_loss": 0.0572
    },
    {
      "epoch": 0.5499784476828966,
      "lm_loss": 0.0842,
      "step": 7496,
      "vm_loss": 0.059
    },
    {
      "epoch": 0.5499784476828966,
      "lm_loss": 0.9466,
      "step": 7496,
      "vm_loss": 0.0869
    },
    {
      "epoch": 0.5499784476828966,
      "lm_loss": 0.0254,
      "step": 7496,
      "vm_loss": 0.1734
    },
    {
      "epoch": 0.5499784476828966,
      "lm_loss": 1.4862,
      "step": 7496,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.5499784476828966,
      "lm_loss": 0.7348,
      "step": 7496,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.5499784476828966,
      "lm_loss": 0.5383,
      "step": 7496,
      "vm_loss": 0.0914
    },
    {
      "epoch": 0.5499784476828966,
      "lm_loss": 0.153,
      "step": 7496,
      "vm_loss": 0.0826
    },
    {
      "epoch": 0.5500518172730358,
      "grad_norm": 1.5571021448321416,
      "learning_rate": 4.433262484398656e-06,
      "loss": 0.7327,
      "step": 7497
    },
    {
      "epoch": 0.5501251868631749,
      "grad_norm": 1.5009257998928365,
      "learning_rate": 4.432081960940881e-06,
      "loss": 0.7708,
      "step": 7498
    },
    {
      "epoch": 0.5501985564533141,
      "grad_norm": 1.4549228070951161,
      "learning_rate": 4.430901469554844e-06,
      "loss": 0.7154,
      "step": 7499
    },
    {
      "epoch": 0.5502719260434531,
      "grad_norm": 1.7837826721190277,
      "learning_rate": 4.42972101030721e-06,
      "loss": 0.7213,
      "step": 7500
    },
    {
      "epoch": 0.5503452956335922,
      "grad_norm": 0.5892750065509023,
      "learning_rate": 4.428540583264644e-06,
      "loss": 0.4849,
      "step": 7501
    },
    {
      "epoch": 0.5504186652237314,
      "grad_norm": 1.3820166024328298,
      "learning_rate": 4.427360188493806e-06,
      "loss": 0.7212,
      "step": 7502
    },
    {
      "epoch": 0.5504920348138705,
      "grad_norm": 0.5472691608927488,
      "learning_rate": 4.426179826061355e-06,
      "loss": 0.4896,
      "step": 7503
    },
    {
      "epoch": 0.5505654044040097,
      "grad_norm": 0.5586448649682555,
      "learning_rate": 4.424999496033952e-06,
      "loss": 0.5191,
      "step": 7504
    },
    {
      "epoch": 0.5505654044040097,
      "lm_loss": 1.2434,
      "step": 7504,
      "vm_loss": 0.1651
    },
    {
      "epoch": 0.5505654044040097,
      "lm_loss": 0.2299,
      "step": 7504,
      "vm_loss": 0.0959
    },
    {
      "epoch": 0.5505654044040097,
      "lm_loss": 0.96,
      "step": 7504,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.5505654044040097,
      "lm_loss": 0.2465,
      "step": 7504,
      "vm_loss": 0.0762
    },
    {
      "epoch": 0.5505654044040097,
      "lm_loss": 0.616,
      "step": 7504,
      "vm_loss": 0.091
    },
    {
      "epoch": 0.5505654044040097,
      "lm_loss": 0.1682,
      "step": 7504,
      "vm_loss": 0.1775
    },
    {
      "epoch": 0.5505654044040097,
      "lm_loss": 0.2386,
      "step": 7504,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.5505654044040097,
      "lm_loss": 0.2873,
      "step": 7504,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.5506387739941487,
      "grad_norm": 1.6736536257837322,
      "learning_rate": 4.423819198478251e-06,
      "loss": 0.6914,
      "step": 7505
    },
    {
      "epoch": 0.5507121435842879,
      "grad_norm": 1.828409299802325,
      "learning_rate": 4.422638933460906e-06,
      "loss": 0.8664,
      "step": 7506
    },
    {
      "epoch": 0.550785513174427,
      "grad_norm": 1.610287691029766,
      "learning_rate": 4.421458701048571e-06,
      "loss": 0.8586,
      "step": 7507
    },
    {
      "epoch": 0.5508588827645662,
      "grad_norm": 1.377168301143116,
      "learning_rate": 4.420278501307896e-06,
      "loss": 0.726,
      "step": 7508
    },
    {
      "epoch": 0.5509322523547053,
      "grad_norm": 0.557058977679327,
      "learning_rate": 4.419098334305529e-06,
      "loss": 0.4885,
      "step": 7509
    },
    {
      "epoch": 0.5510056219448444,
      "grad_norm": 1.4357963417589368,
      "learning_rate": 4.4179182001081174e-06,
      "loss": 0.7173,
      "step": 7510
    },
    {
      "epoch": 0.5510789915349835,
      "grad_norm": 1.4847720355736278,
      "learning_rate": 4.416738098782307e-06,
      "loss": 0.7561,
      "step": 7511
    },
    {
      "epoch": 0.5511523611251227,
      "grad_norm": 1.542186246300555,
      "learning_rate": 4.415558030394741e-06,
      "loss": 0.7101,
      "step": 7512
    },
    {
      "epoch": 0.5511523611251227,
      "lm_loss": 0.3028,
      "step": 7512,
      "vm_loss": 0.0525
    },
    {
      "epoch": 0.5511523611251227,
      "lm_loss": 1.2509,
      "step": 7512,
      "vm_loss": 0.0515
    },
    {
      "epoch": 0.5511523611251227,
      "lm_loss": 0.9947,
      "step": 7512,
      "vm_loss": 0.0684
    },
    {
      "epoch": 0.5511523611251227,
      "lm_loss": 0.1267,
      "step": 7512,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.5511523611251227,
      "lm_loss": 0.9291,
      "step": 7512,
      "vm_loss": 0.1069
    },
    {
      "epoch": 0.5511523611251227,
      "lm_loss": 0.7312,
      "step": 7512,
      "vm_loss": 0.0703
    },
    {
      "epoch": 0.5511523611251227,
      "lm_loss": 0.7721,
      "step": 7512,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.5511523611251227,
      "lm_loss": 0.4248,
      "step": 7512,
      "vm_loss": 0.106
    },
    {
      "epoch": 0.5512257307152618,
      "grad_norm": 1.4130278054855434,
      "learning_rate": 4.41437799501206e-06,
      "loss": 0.7655,
      "step": 7513
    },
    {
      "epoch": 0.551299100305401,
      "grad_norm": 1.532099805778081,
      "learning_rate": 4.4131979927009035e-06,
      "loss": 0.8122,
      "step": 7514
    },
    {
      "epoch": 0.55137246989554,
      "grad_norm": 1.4296258725093869,
      "learning_rate": 4.41201802352791e-06,
      "loss": 0.7267,
      "step": 7515
    },
    {
      "epoch": 0.5514458394856792,
      "grad_norm": 1.7352595099202237,
      "learning_rate": 4.410838087559716e-06,
      "loss": 0.7941,
      "step": 7516
    },
    {
      "epoch": 0.5515192090758183,
      "grad_norm": 0.5623276842483522,
      "learning_rate": 4.409658184862952e-06,
      "loss": 0.5036,
      "step": 7517
    },
    {
      "epoch": 0.5515925786659575,
      "grad_norm": 1.3542127456125677,
      "learning_rate": 4.4084783155042536e-06,
      "loss": 0.718,
      "step": 7518
    },
    {
      "epoch": 0.5516659482560966,
      "grad_norm": 1.4358242166801152,
      "learning_rate": 4.407298479550249e-06,
      "loss": 0.78,
      "step": 7519
    },
    {
      "epoch": 0.5517393178462356,
      "grad_norm": 1.3553248164111513,
      "learning_rate": 4.406118677067568e-06,
      "loss": 0.7322,
      "step": 7520
    },
    {
      "epoch": 0.5517393178462356,
      "lm_loss": 1.4705,
      "step": 7520,
      "vm_loss": 0.051
    },
    {
      "epoch": 0.5517393178462356,
      "lm_loss": 0.3914,
      "step": 7520,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.5517393178462356,
      "lm_loss": 0.089,
      "step": 7520,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.5517393178462356,
      "lm_loss": 0.4275,
      "step": 7520,
      "vm_loss": 0.1872
    },
    {
      "epoch": 0.5517393178462356,
      "lm_loss": 0.8002,
      "step": 7520,
      "vm_loss": 0.2134
    },
    {
      "epoch": 0.5517393178462356,
      "lm_loss": 0.623,
      "step": 7520,
      "vm_loss": 0.1315
    },
    {
      "epoch": 0.5517393178462356,
      "lm_loss": 0.2308,
      "step": 7520,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.5517393178462356,
      "lm_loss": 0.1491,
      "step": 7520,
      "vm_loss": 0.1337
    },
    {
      "epoch": 0.5518126874363748,
      "grad_norm": 1.5448078131718597,
      "learning_rate": 4.404938908122835e-06,
      "loss": 0.6981,
      "step": 7521
    },
    {
      "epoch": 0.5518860570265139,
      "grad_norm": 0.5736186512695487,
      "learning_rate": 4.403759172782676e-06,
      "loss": 0.5063,
      "step": 7522
    },
    {
      "epoch": 0.5519594266166531,
      "grad_norm": 0.547044639564267,
      "learning_rate": 4.402579471113712e-06,
      "loss": 0.4873,
      "step": 7523
    },
    {
      "epoch": 0.5520327962067922,
      "grad_norm": 1.465829307558746,
      "learning_rate": 4.401399803182566e-06,
      "loss": 0.7239,
      "step": 7524
    },
    {
      "epoch": 0.5521061657969313,
      "grad_norm": 1.4997164319262077,
      "learning_rate": 4.4002201690558555e-06,
      "loss": 0.7376,
      "step": 7525
    },
    {
      "epoch": 0.5521795353870704,
      "grad_norm": 1.376646998686704,
      "learning_rate": 4.3990405688001975e-06,
      "loss": 0.6021,
      "step": 7526
    },
    {
      "epoch": 0.5522529049772096,
      "grad_norm": 1.4962490896072174,
      "learning_rate": 4.397861002482205e-06,
      "loss": 0.7414,
      "step": 7527
    },
    {
      "epoch": 0.5523262745673487,
      "grad_norm": 1.3970399554825579,
      "learning_rate": 4.396681470168495e-06,
      "loss": 0.8063,
      "step": 7528
    },
    {
      "epoch": 0.5523262745673487,
      "lm_loss": 1.4544,
      "step": 7528,
      "vm_loss": 0.1151
    },
    {
      "epoch": 0.5523262745673487,
      "lm_loss": 0.3091,
      "step": 7528,
      "vm_loss": 0.0399
    },
    {
      "epoch": 0.5523262745673487,
      "lm_loss": 1.0341,
      "step": 7528,
      "vm_loss": 0.1223
    },
    {
      "epoch": 0.5523262745673487,
      "lm_loss": 0.9034,
      "step": 7528,
      "vm_loss": 0.1993
    },
    {
      "epoch": 0.5523262745673487,
      "lm_loss": 0.6144,
      "step": 7528,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.5523262745673487,
      "lm_loss": 0.0724,
      "step": 7528,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.5523262745673487,
      "lm_loss": 0.1234,
      "step": 7528,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.5523262745673487,
      "lm_loss": 0.6517,
      "step": 7528,
      "vm_loss": 0.1996
    },
    {
      "epoch": 0.5523996441574879,
      "grad_norm": 1.6565575811421918,
      "learning_rate": 4.395501971925677e-06,
      "loss": 0.7654,
      "step": 7529
    },
    {
      "epoch": 0.5524730137476269,
      "grad_norm": 0.5681706328706425,
      "learning_rate": 4.394322507820357e-06,
      "loss": 0.4915,
      "step": 7530
    },
    {
      "epoch": 0.5525463833377661,
      "grad_norm": 1.2990857450332534,
      "learning_rate": 4.393143077919147e-06,
      "loss": 0.7194,
      "step": 7531
    },
    {
      "epoch": 0.5526197529279052,
      "grad_norm": 1.629461764300725,
      "learning_rate": 4.39196368228865e-06,
      "loss": 0.6651,
      "step": 7532
    },
    {
      "epoch": 0.5526931225180444,
      "grad_norm": 1.5429112033858952,
      "learning_rate": 4.390784320995469e-06,
      "loss": 0.8037,
      "step": 7533
    },
    {
      "epoch": 0.5527664921081835,
      "grad_norm": 0.5664779407221556,
      "learning_rate": 4.3896049941062065e-06,
      "loss": 0.4978,
      "step": 7534
    },
    {
      "epoch": 0.5528398616983226,
      "grad_norm": 0.5654945059764018,
      "learning_rate": 4.388425701687461e-06,
      "loss": 0.5128,
      "step": 7535
    },
    {
      "epoch": 0.5529132312884617,
      "grad_norm": 1.511529774969569,
      "learning_rate": 4.387246443805831e-06,
      "loss": 0.7881,
      "step": 7536
    },
    {
      "epoch": 0.5529132312884617,
      "lm_loss": 1.2658,
      "step": 7536,
      "vm_loss": 0.0418
    },
    {
      "epoch": 0.5529132312884617,
      "lm_loss": 0.6154,
      "step": 7536,
      "vm_loss": 0.0463
    },
    {
      "epoch": 0.5529132312884617,
      "lm_loss": 0.2849,
      "step": 7536,
      "vm_loss": 0.157
    },
    {
      "epoch": 0.5529132312884617,
      "lm_loss": 0.6923,
      "step": 7536,
      "vm_loss": 0.0897
    },
    {
      "epoch": 0.5529132312884617,
      "lm_loss": 1.1597,
      "step": 7536,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.5529132312884617,
      "lm_loss": 0.0351,
      "step": 7536,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.5529132312884617,
      "lm_loss": 0.4609,
      "step": 7536,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.5529132312884617,
      "lm_loss": 0.1765,
      "step": 7536,
      "vm_loss": 0.0737
    },
    {
      "epoch": 0.5529866008786009,
      "grad_norm": 1.6281431921394007,
      "learning_rate": 4.386067220527912e-06,
      "loss": 0.7112,
      "step": 7537
    },
    {
      "epoch": 0.55305997046874,
      "grad_norm": 0.5623018746200918,
      "learning_rate": 4.384888031920297e-06,
      "loss": 0.4766,
      "step": 7538
    },
    {
      "epoch": 0.553133340058879,
      "grad_norm": 1.5562226413047724,
      "learning_rate": 4.383708878049579e-06,
      "loss": 0.7312,
      "step": 7539
    },
    {
      "epoch": 0.5532067096490182,
      "grad_norm": 1.507210684667211,
      "learning_rate": 4.382529758982346e-06,
      "loss": 0.7844,
      "step": 7540
    },
    {
      "epoch": 0.5532800792391573,
      "grad_norm": 1.582028243750117,
      "learning_rate": 4.3813506747851865e-06,
      "loss": 0.7946,
      "step": 7541
    },
    {
      "epoch": 0.5533534488292965,
      "grad_norm": 1.6214799562256692,
      "learning_rate": 4.380171625524688e-06,
      "loss": 0.7199,
      "step": 7542
    },
    {
      "epoch": 0.5534268184194356,
      "grad_norm": 1.3880364837508203,
      "learning_rate": 4.3789926112674315e-06,
      "loss": 0.6973,
      "step": 7543
    },
    {
      "epoch": 0.5535001880095747,
      "grad_norm": 0.5913375281090578,
      "learning_rate": 4.37781363208e-06,
      "loss": 0.4844,
      "step": 7544
    },
    {
      "epoch": 0.5535001880095747,
      "lm_loss": 0.8677,
      "step": 7544,
      "vm_loss": 0.0568
    },
    {
      "epoch": 0.5535001880095747,
      "lm_loss": 0.0059,
      "step": 7544,
      "vm_loss": 0.1115
    },
    {
      "epoch": 0.5535001880095747,
      "lm_loss": 0.2257,
      "step": 7544,
      "vm_loss": 0.0718
    },
    {
      "epoch": 0.5535001880095747,
      "lm_loss": 0.7481,
      "step": 7544,
      "vm_loss": 0.1537
    },
    {
      "epoch": 0.5535001880095747,
      "lm_loss": 0.5661,
      "step": 7544,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.5535001880095747,
      "lm_loss": 0.394,
      "step": 7544,
      "vm_loss": 0.1655
    },
    {
      "epoch": 0.5535001880095747,
      "lm_loss": 0.4717,
      "step": 7544,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.5535001880095747,
      "lm_loss": 0.6866,
      "step": 7544,
      "vm_loss": 0.1751
    },
    {
      "epoch": 0.5535735575997138,
      "grad_norm": 1.5124245913668384,
      "learning_rate": 4.376634688028975e-06,
      "loss": 0.7129,
      "step": 7545
    },
    {
      "epoch": 0.553646927189853,
      "grad_norm": 0.5591637113850817,
      "learning_rate": 4.375455779180932e-06,
      "loss": 0.5,
      "step": 7546
    },
    {
      "epoch": 0.5537202967799921,
      "grad_norm": 1.4016290970825123,
      "learning_rate": 4.374276905602448e-06,
      "loss": 0.7325,
      "step": 7547
    },
    {
      "epoch": 0.5537936663701313,
      "grad_norm": 1.480902125314391,
      "learning_rate": 4.373098067360096e-06,
      "loss": 0.773,
      "step": 7548
    },
    {
      "epoch": 0.5538670359602703,
      "grad_norm": 1.7393523775044994,
      "learning_rate": 4.371919264520449e-06,
      "loss": 0.8038,
      "step": 7549
    },
    {
      "epoch": 0.5539404055504095,
      "grad_norm": 1.5164060426153134,
      "learning_rate": 4.370740497150077e-06,
      "loss": 0.7731,
      "step": 7550
    },
    {
      "epoch": 0.5540137751405486,
      "grad_norm": 1.567236583639963,
      "learning_rate": 4.369561765315547e-06,
      "loss": 0.7105,
      "step": 7551
    },
    {
      "epoch": 0.5540871447306878,
      "grad_norm": 1.6474804722654013,
      "learning_rate": 4.368383069083425e-06,
      "loss": 0.8582,
      "step": 7552
    },
    {
      "epoch": 0.5540871447306878,
      "lm_loss": 0.4964,
      "step": 7552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5540871447306878,
      "lm_loss": 0.498,
      "step": 7552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5540871447306878,
      "lm_loss": 0.3509,
      "step": 7552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5540871447306878,
      "lm_loss": 0.4393,
      "step": 7552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5540871447306878,
      "lm_loss": 0.5162,
      "step": 7552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5540871447306878,
      "lm_loss": 0.6669,
      "step": 7552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5540871447306878,
      "lm_loss": 0.6818,
      "step": 7552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5540871447306878,
      "lm_loss": 0.5254,
      "step": 7552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5541605143208269,
      "grad_norm": 0.5812020280473158,
      "learning_rate": 4.367204408520275e-06,
      "loss": 0.4868,
      "step": 7553
    },
    {
      "epoch": 0.554233883910966,
      "grad_norm": 1.591983995700913,
      "learning_rate": 4.36602578369266e-06,
      "loss": 0.7588,
      "step": 7554
    },
    {
      "epoch": 0.5543072535011051,
      "grad_norm": 1.3982136489877623,
      "learning_rate": 4.364847194667139e-06,
      "loss": 0.8195,
      "step": 7555
    },
    {
      "epoch": 0.5543806230912443,
      "grad_norm": 1.4345000077687875,
      "learning_rate": 4.363668641510268e-06,
      "loss": 0.6963,
      "step": 7556
    },
    {
      "epoch": 0.5544539926813834,
      "grad_norm": 1.388757700224595,
      "learning_rate": 4.362490124288607e-06,
      "loss": 0.7452,
      "step": 7557
    },
    {
      "epoch": 0.5545273622715226,
      "grad_norm": 1.5637488923950662,
      "learning_rate": 4.361311643068707e-06,
      "loss": 0.8567,
      "step": 7558
    },
    {
      "epoch": 0.5546007318616616,
      "grad_norm": 1.4806787416304321,
      "learning_rate": 4.36013319791712e-06,
      "loss": 0.7308,
      "step": 7559
    },
    {
      "epoch": 0.5546741014518007,
      "grad_norm": 1.7079990011957369,
      "learning_rate": 4.358954788900396e-06,
      "loss": 0.8154,
      "step": 7560
    },
    {
      "epoch": 0.5546741014518007,
      "lm_loss": 0.4619,
      "step": 7560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5546741014518007,
      "lm_loss": 0.5303,
      "step": 7560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5546741014518007,
      "lm_loss": 0.6416,
      "step": 7560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5546741014518007,
      "lm_loss": 0.3854,
      "step": 7560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5546741014518007,
      "lm_loss": 0.5144,
      "step": 7560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5546741014518007,
      "lm_loss": 0.3956,
      "step": 7560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5546741014518007,
      "lm_loss": 0.3633,
      "step": 7560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5546741014518007,
      "lm_loss": 0.4094,
      "step": 7560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5547474710419399,
      "grad_norm": 0.5811553829755893,
      "learning_rate": 4.357776416085082e-06,
      "loss": 0.4734,
      "step": 7561
    },
    {
      "epoch": 0.554820840632079,
      "grad_norm": 0.5609961918339901,
      "learning_rate": 4.356598079537725e-06,
      "loss": 0.4876,
      "step": 7562
    },
    {
      "epoch": 0.5548942102222182,
      "grad_norm": 1.4701128597639723,
      "learning_rate": 4.355419779324867e-06,
      "loss": 0.679,
      "step": 7563
    },
    {
      "epoch": 0.5549675798123572,
      "grad_norm": 0.5755444677568354,
      "learning_rate": 4.35424151551305e-06,
      "loss": 0.4738,
      "step": 7564
    },
    {
      "epoch": 0.5550409494024964,
      "grad_norm": 1.3174736458134066,
      "learning_rate": 4.353063288168815e-06,
      "loss": 0.6712,
      "step": 7565
    },
    {
      "epoch": 0.5551143189926355,
      "grad_norm": 1.468197468687721,
      "learning_rate": 4.351885097358698e-06,
      "loss": 0.7633,
      "step": 7566
    },
    {
      "epoch": 0.5551876885827747,
      "grad_norm": 1.4714829946761245,
      "learning_rate": 4.3507069431492345e-06,
      "loss": 0.8685,
      "step": 7567
    },
    {
      "epoch": 0.5552610581729138,
      "grad_norm": 1.5955335162856918,
      "learning_rate": 4.349528825606958e-06,
      "loss": 0.7305,
      "step": 7568
    },
    {
      "epoch": 0.5552610581729138,
      "lm_loss": 1.7419,
      "step": 7568,
      "vm_loss": 0.0517
    },
    {
      "epoch": 0.5552610581729138,
      "lm_loss": 1.5586,
      "step": 7568,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.5552610581729138,
      "lm_loss": 0.255,
      "step": 7568,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.5552610581729138,
      "lm_loss": 1.1578,
      "step": 7568,
      "vm_loss": 0.1359
    },
    {
      "epoch": 0.5552610581729138,
      "lm_loss": 1.0484,
      "step": 7568,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.5552610581729138,
      "lm_loss": 0.6445,
      "step": 7568,
      "vm_loss": 0.2088
    },
    {
      "epoch": 0.5552610581729138,
      "lm_loss": 1.0397,
      "step": 7568,
      "vm_loss": 0.179
    },
    {
      "epoch": 0.5552610581729138,
      "lm_loss": 0.1468,
      "step": 7568,
      "vm_loss": 0.1665
    },
    {
      "epoch": 0.5553344277630529,
      "grad_norm": 1.3994584863881787,
      "learning_rate": 4.348350744798399e-06,
      "loss": 0.7959,
      "step": 7569
    },
    {
      "epoch": 0.555407797353192,
      "grad_norm": 0.634005269705482,
      "learning_rate": 4.347172700790087e-06,
      "loss": 0.4928,
      "step": 7570
    },
    {
      "epoch": 0.5554811669433312,
      "grad_norm": 0.628227902655267,
      "learning_rate": 4.34599469364855e-06,
      "loss": 0.4878,
      "step": 7571
    },
    {
      "epoch": 0.5555545365334703,
      "grad_norm": 1.7397765043864908,
      "learning_rate": 4.3448167234403115e-06,
      "loss": 0.8258,
      "step": 7572
    },
    {
      "epoch": 0.5556279061236095,
      "grad_norm": 1.6258068987067034,
      "learning_rate": 4.343638790231897e-06,
      "loss": 0.7048,
      "step": 7573
    },
    {
      "epoch": 0.5557012757137485,
      "grad_norm": 0.5640904565010563,
      "learning_rate": 4.342460894089824e-06,
      "loss": 0.5004,
      "step": 7574
    },
    {
      "epoch": 0.5557746453038876,
      "grad_norm": 1.5154527318170465,
      "learning_rate": 4.341283035080612e-06,
      "loss": 0.8051,
      "step": 7575
    },
    {
      "epoch": 0.5558480148940268,
      "grad_norm": 1.443505758582172,
      "learning_rate": 4.34010521327078e-06,
      "loss": 0.8228,
      "step": 7576
    },
    {
      "epoch": 0.5558480148940268,
      "lm_loss": 0.2056,
      "step": 7576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5558480148940268,
      "lm_loss": 0.414,
      "step": 7576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5558480148940268,
      "lm_loss": 0.4442,
      "step": 7576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5558480148940268,
      "lm_loss": 0.5695,
      "step": 7576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5558480148940268,
      "lm_loss": 0.4324,
      "step": 7576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5558480148940268,
      "lm_loss": 0.4396,
      "step": 7576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5558480148940268,
      "lm_loss": 0.5017,
      "step": 7576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5558480148940268,
      "lm_loss": 0.5509,
      "step": 7576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.555921384484166,
      "grad_norm": 0.6454960044161585,
      "learning_rate": 4.33892742872684e-06,
      "loss": 0.499,
      "step": 7577
    },
    {
      "epoch": 0.5559947540743051,
      "grad_norm": 0.6296379134078107,
      "learning_rate": 4.3377496815153045e-06,
      "loss": 0.5062,
      "step": 7578
    },
    {
      "epoch": 0.5560681236644441,
      "grad_norm": 0.5893093315856239,
      "learning_rate": 4.336571971702686e-06,
      "loss": 0.498,
      "step": 7579
    },
    {
      "epoch": 0.5561414932545833,
      "grad_norm": 0.5328161649261715,
      "learning_rate": 4.33539429935549e-06,
      "loss": 0.4719,
      "step": 7580
    },
    {
      "epoch": 0.5562148628447224,
      "grad_norm": 1.5488520563745436,
      "learning_rate": 4.334216664540224e-06,
      "loss": 0.7524,
      "step": 7581
    },
    {
      "epoch": 0.5562882324348616,
      "grad_norm": 0.5656263615226252,
      "learning_rate": 4.333039067323393e-06,
      "loss": 0.4959,
      "step": 7582
    },
    {
      "epoch": 0.5563616020250007,
      "grad_norm": 0.5647749811135254,
      "learning_rate": 4.331861507771496e-06,
      "loss": 0.4876,
      "step": 7583
    },
    {
      "epoch": 0.5564349716151398,
      "grad_norm": 0.5616482117717233,
      "learning_rate": 4.330683985951035e-06,
      "loss": 0.5021,
      "step": 7584
    },
    {
      "epoch": 0.5564349716151398,
      "lm_loss": 0.9678,
      "step": 7584,
      "vm_loss": 0.0323
    },
    {
      "epoch": 0.5564349716151398,
      "lm_loss": 0.0596,
      "step": 7584,
      "vm_loss": 0.0719
    },
    {
      "epoch": 0.5564349716151398,
      "lm_loss": 0.0672,
      "step": 7584,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.5564349716151398,
      "lm_loss": 0.405,
      "step": 7584,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.5564349716151398,
      "lm_loss": 0.0555,
      "step": 7584,
      "vm_loss": 0.0828
    },
    {
      "epoch": 0.5564349716151398,
      "lm_loss": 0.6738,
      "step": 7584,
      "vm_loss": 0.139
    },
    {
      "epoch": 0.5564349716151398,
      "lm_loss": 0.16,
      "step": 7584,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.5564349716151398,
      "lm_loss": 0.1431,
      "step": 7584,
      "vm_loss": 0.0757
    },
    {
      "epoch": 0.5565083412052789,
      "grad_norm": 1.577882071918463,
      "learning_rate": 4.329506501928507e-06,
      "loss": 0.7109,
      "step": 7585
    },
    {
      "epoch": 0.5565817107954181,
      "grad_norm": 1.8462950924301427,
      "learning_rate": 4.328329055770406e-06,
      "loss": 0.6987,
      "step": 7586
    },
    {
      "epoch": 0.5566550803855572,
      "grad_norm": 0.5458482473784257,
      "learning_rate": 4.327151647543228e-06,
      "loss": 0.4968,
      "step": 7587
    },
    {
      "epoch": 0.5567284499756964,
      "grad_norm": 1.4809549856516173,
      "learning_rate": 4.325974277313462e-06,
      "loss": 0.8158,
      "step": 7588
    },
    {
      "epoch": 0.5568018195658354,
      "grad_norm": 0.5561731332061655,
      "learning_rate": 4.324796945147598e-06,
      "loss": 0.5002,
      "step": 7589
    },
    {
      "epoch": 0.5568751891559746,
      "grad_norm": 1.51309418013956,
      "learning_rate": 4.323619651112123e-06,
      "loss": 0.8179,
      "step": 7590
    },
    {
      "epoch": 0.5569485587461137,
      "grad_norm": 0.5561655040147776,
      "learning_rate": 4.3224423952735205e-06,
      "loss": 0.4993,
      "step": 7591
    },
    {
      "epoch": 0.5570219283362529,
      "grad_norm": 1.420060098065319,
      "learning_rate": 4.321265177698274e-06,
      "loss": 0.8597,
      "step": 7592
    },
    {
      "epoch": 0.5570219283362529,
      "lm_loss": 0.5626,
      "step": 7592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5570219283362529,
      "lm_loss": 0.3944,
      "step": 7592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5570219283362529,
      "lm_loss": 0.5929,
      "step": 7592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5570219283362529,
      "lm_loss": 0.6621,
      "step": 7592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5570219283362529,
      "lm_loss": 0.6023,
      "step": 7592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5570219283362529,
      "lm_loss": 0.5156,
      "step": 7592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5570219283362529,
      "lm_loss": 0.5914,
      "step": 7592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5570219283362529,
      "lm_loss": 0.6865,
      "step": 7592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.557095297926392,
      "grad_norm": 0.5519978965107069,
      "learning_rate": 4.3200879984528656e-06,
      "loss": 0.4959,
      "step": 7593
    },
    {
      "epoch": 0.557168667516531,
      "grad_norm": 0.5687099308328395,
      "learning_rate": 4.3189108576037695e-06,
      "loss": 0.4862,
      "step": 7594
    },
    {
      "epoch": 0.5572420371066702,
      "grad_norm": 0.5661625692891528,
      "learning_rate": 4.317733755217465e-06,
      "loss": 0.4915,
      "step": 7595
    },
    {
      "epoch": 0.5573154066968093,
      "grad_norm": 0.54392135298693,
      "learning_rate": 4.316556691360424e-06,
      "loss": 0.4891,
      "step": 7596
    },
    {
      "epoch": 0.5573887762869485,
      "grad_norm": 0.5383701090974378,
      "learning_rate": 4.31537966609912e-06,
      "loss": 0.4989,
      "step": 7597
    },
    {
      "epoch": 0.5574621458770876,
      "grad_norm": 0.5335235475328743,
      "learning_rate": 4.3142026795000216e-06,
      "loss": 0.4912,
      "step": 7598
    },
    {
      "epoch": 0.5575355154672267,
      "grad_norm": 1.4895906732610689,
      "learning_rate": 4.313025731629596e-06,
      "loss": 0.7683,
      "step": 7599
    },
    {
      "epoch": 0.5576088850573658,
      "grad_norm": 0.5741712076736966,
      "learning_rate": 4.311848822554309e-06,
      "loss": 0.4984,
      "step": 7600
    },
    {
      "epoch": 0.5576088850573658,
      "lm_loss": 1.5351,
      "step": 7600,
      "vm_loss": 0.0429
    },
    {
      "epoch": 0.5576088850573658,
      "lm_loss": 0.1256,
      "step": 7600,
      "vm_loss": 0.0964
    },
    {
      "epoch": 0.5576088850573658,
      "lm_loss": 1.1314,
      "step": 7600,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.5576088850573658,
      "lm_loss": 1.5201,
      "step": 7600,
      "vm_loss": 0.2156
    },
    {
      "epoch": 0.5576088850573658,
      "lm_loss": 0.3375,
      "step": 7600,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.5576088850573658,
      "lm_loss": 1.0122,
      "step": 7600,
      "vm_loss": 0.1225
    },
    {
      "epoch": 0.5576088850573658,
      "lm_loss": 0.1519,
      "step": 7600,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.5576088850573658,
      "lm_loss": 0.873,
      "step": 7600,
      "vm_loss": 0.0513
    },
    {
      "epoch": 0.557682254647505,
      "grad_norm": 1.5865155373841993,
      "learning_rate": 4.310671952340623e-06,
      "loss": 0.7958,
      "step": 7601
    },
    {
      "epoch": 0.5577556242376441,
      "grad_norm": 0.557856795883546,
      "learning_rate": 4.309495121054998e-06,
      "loss": 0.4879,
      "step": 7602
    },
    {
      "epoch": 0.5578289938277833,
      "grad_norm": 1.6385920681780488,
      "learning_rate": 4.308318328763895e-06,
      "loss": 0.7085,
      "step": 7603
    },
    {
      "epoch": 0.5579023634179223,
      "grad_norm": 1.5673513089929174,
      "learning_rate": 4.307141575533768e-06,
      "loss": 0.7952,
      "step": 7604
    },
    {
      "epoch": 0.5579757330080615,
      "grad_norm": 0.5453568267688977,
      "learning_rate": 4.305964861431072e-06,
      "loss": 0.4978,
      "step": 7605
    },
    {
      "epoch": 0.5580491025982006,
      "grad_norm": 1.5235396703400776,
      "learning_rate": 4.304788186522259e-06,
      "loss": 0.7654,
      "step": 7606
    },
    {
      "epoch": 0.5581224721883398,
      "grad_norm": 1.3646044321796127,
      "learning_rate": 4.303611550873779e-06,
      "loss": 0.7309,
      "step": 7607
    },
    {
      "epoch": 0.5581958417784789,
      "grad_norm": 1.5252624888862325,
      "learning_rate": 4.30243495455208e-06,
      "loss": 0.6916,
      "step": 7608
    },
    {
      "epoch": 0.5581958417784789,
      "lm_loss": 0.5747,
      "step": 7608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5581958417784789,
      "lm_loss": 0.5265,
      "step": 7608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5581958417784789,
      "lm_loss": 0.5222,
      "step": 7608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5581958417784789,
      "lm_loss": 0.5099,
      "step": 7608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5581958417784789,
      "lm_loss": 0.3313,
      "step": 7608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5581958417784789,
      "lm_loss": 0.491,
      "step": 7608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5581958417784789,
      "lm_loss": 0.5594,
      "step": 7608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5581958417784789,
      "lm_loss": 0.5419,
      "step": 7608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.558269211368618,
      "grad_norm": 0.5529040042060528,
      "learning_rate": 4.301258397623606e-06,
      "loss": 0.4807,
      "step": 7609
    },
    {
      "epoch": 0.5583425809587571,
      "grad_norm": 1.556986815433188,
      "learning_rate": 4.3000818801548e-06,
      "loss": 0.7697,
      "step": 7610
    },
    {
      "epoch": 0.5584159505488963,
      "grad_norm": 1.3303073967754473,
      "learning_rate": 4.298905402212105e-06,
      "loss": 0.7894,
      "step": 7611
    },
    {
      "epoch": 0.5584893201390354,
      "grad_norm": 0.5686889001773942,
      "learning_rate": 4.297728963861958e-06,
      "loss": 0.4944,
      "step": 7612
    },
    {
      "epoch": 0.5585626897291746,
      "grad_norm": 1.5787748235543448,
      "learning_rate": 4.296552565170794e-06,
      "loss": 0.8116,
      "step": 7613
    },
    {
      "epoch": 0.5586360593193136,
      "grad_norm": 1.7656837032183226,
      "learning_rate": 4.2953762062050504e-06,
      "loss": 0.7919,
      "step": 7614
    },
    {
      "epoch": 0.5587094289094527,
      "grad_norm": 1.553652346956046,
      "learning_rate": 4.294199887031156e-06,
      "loss": 0.7995,
      "step": 7615
    },
    {
      "epoch": 0.5587827984995919,
      "grad_norm": 1.6266412831896697,
      "learning_rate": 4.293023607715543e-06,
      "loss": 0.7103,
      "step": 7616
    },
    {
      "epoch": 0.5587827984995919,
      "lm_loss": 0.8753,
      "step": 7616,
      "vm_loss": 0.0487
    },
    {
      "epoch": 0.5587827984995919,
      "lm_loss": 1.0614,
      "step": 7616,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.5587827984995919,
      "lm_loss": 0.5645,
      "step": 7616,
      "vm_loss": 0.0736
    },
    {
      "epoch": 0.5587827984995919,
      "lm_loss": 0.885,
      "step": 7616,
      "vm_loss": 0.112
    },
    {
      "epoch": 0.5587827984995919,
      "lm_loss": 0.6651,
      "step": 7616,
      "vm_loss": 0.1455
    },
    {
      "epoch": 0.5587827984995919,
      "lm_loss": 0.2252,
      "step": 7616,
      "vm_loss": 0.0679
    },
    {
      "epoch": 0.5587827984995919,
      "lm_loss": 0.3382,
      "step": 7616,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.5587827984995919,
      "lm_loss": 0.2587,
      "step": 7616,
      "vm_loss": 0.1454
    },
    {
      "epoch": 0.558856168089731,
      "grad_norm": 1.584392628954603,
      "learning_rate": 4.291847368324637e-06,
      "loss": 0.8366,
      "step": 7617
    },
    {
      "epoch": 0.5589295376798701,
      "grad_norm": 1.4814916730214847,
      "learning_rate": 4.290671168924863e-06,
      "loss": 0.7348,
      "step": 7618
    },
    {
      "epoch": 0.5590029072700092,
      "grad_norm": 1.474259835294666,
      "learning_rate": 4.289495009582647e-06,
      "loss": 0.7195,
      "step": 7619
    },
    {
      "epoch": 0.5590762768601484,
      "grad_norm": 0.5667198053581471,
      "learning_rate": 4.288318890364407e-06,
      "loss": 0.4874,
      "step": 7620
    },
    {
      "epoch": 0.5591496464502875,
      "grad_norm": 1.5585331309163022,
      "learning_rate": 4.28714281133656e-06,
      "loss": 0.7671,
      "step": 7621
    },
    {
      "epoch": 0.5592230160404267,
      "grad_norm": 1.5957603793927204,
      "learning_rate": 4.285966772565524e-06,
      "loss": 0.775,
      "step": 7622
    },
    {
      "epoch": 0.5592963856305657,
      "grad_norm": 1.520725889570173,
      "learning_rate": 4.284790774117713e-06,
      "loss": 0.8503,
      "step": 7623
    },
    {
      "epoch": 0.5593697552207049,
      "grad_norm": 1.4708711780005357,
      "learning_rate": 4.283614816059539e-06,
      "loss": 0.7124,
      "step": 7624
    },
    {
      "epoch": 0.5593697552207049,
      "lm_loss": 1.0849,
      "step": 7624,
      "vm_loss": 0.0653
    },
    {
      "epoch": 0.5593697552207049,
      "lm_loss": 1.484,
      "step": 7624,
      "vm_loss": 0.1144
    },
    {
      "epoch": 0.5593697552207049,
      "lm_loss": 0.7974,
      "step": 7624,
      "vm_loss": 0.195
    },
    {
      "epoch": 0.5593697552207049,
      "lm_loss": 0.3374,
      "step": 7624,
      "vm_loss": 0.1268
    },
    {
      "epoch": 0.5593697552207049,
      "lm_loss": 1.0019,
      "step": 7624,
      "vm_loss": 0.1772
    },
    {
      "epoch": 0.5593697552207049,
      "lm_loss": 0.7226,
      "step": 7624,
      "vm_loss": 0.1927
    },
    {
      "epoch": 0.5593697552207049,
      "lm_loss": 0.0157,
      "step": 7624,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.5593697552207049,
      "lm_loss": 0.3655,
      "step": 7624,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.559443124810844,
      "grad_norm": 1.4524276658077964,
      "learning_rate": 4.282438898457411e-06,
      "loss": 0.7759,
      "step": 7625
    },
    {
      "epoch": 0.5595164944009832,
      "grad_norm": 0.5561895676965475,
      "learning_rate": 4.281263021377733e-06,
      "loss": 0.478,
      "step": 7626
    },
    {
      "epoch": 0.5595898639911223,
      "grad_norm": 0.5675116705788538,
      "learning_rate": 4.280087184886914e-06,
      "loss": 0.4853,
      "step": 7627
    },
    {
      "epoch": 0.5596632335812614,
      "grad_norm": 1.607791894385385,
      "learning_rate": 4.278911389051354e-06,
      "loss": 0.7329,
      "step": 7628
    },
    {
      "epoch": 0.5597366031714005,
      "grad_norm": 1.2750062115228402,
      "learning_rate": 4.2777356339374526e-06,
      "loss": 0.6867,
      "step": 7629
    },
    {
      "epoch": 0.5598099727615397,
      "grad_norm": 1.8300259993591097,
      "learning_rate": 4.27655991961161e-06,
      "loss": 0.7661,
      "step": 7630
    },
    {
      "epoch": 0.5598833423516788,
      "grad_norm": 1.7692183115159361,
      "learning_rate": 4.27538424614022e-06,
      "loss": 0.7442,
      "step": 7631
    },
    {
      "epoch": 0.559956711941818,
      "grad_norm": 0.5729627908494814,
      "learning_rate": 4.274208613589675e-06,
      "loss": 0.5125,
      "step": 7632
    },
    {
      "epoch": 0.559956711941818,
      "lm_loss": 1.3828,
      "step": 7632,
      "vm_loss": 0.0778
    },
    {
      "epoch": 0.559956711941818,
      "lm_loss": 0.4678,
      "step": 7632,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.559956711941818,
      "lm_loss": 0.8178,
      "step": 7632,
      "vm_loss": 0.1975
    },
    {
      "epoch": 0.559956711941818,
      "lm_loss": 0.2067,
      "step": 7632,
      "vm_loss": 0.1162
    },
    {
      "epoch": 0.559956711941818,
      "lm_loss": 0.6055,
      "step": 7632,
      "vm_loss": 0.2153
    },
    {
      "epoch": 0.559956711941818,
      "lm_loss": 0.7318,
      "step": 7632,
      "vm_loss": 0.1112
    },
    {
      "epoch": 0.559956711941818,
      "lm_loss": 0.4988,
      "step": 7632,
      "vm_loss": 0.089
    },
    {
      "epoch": 0.559956711941818,
      "lm_loss": 0.9441,
      "step": 7632,
      "vm_loss": 0.0723
    },
    {
      "epoch": 0.560030081531957,
      "grad_norm": 1.466640544418271,
      "learning_rate": 4.273033022026368e-06,
      "loss": 0.7812,
      "step": 7633
    },
    {
      "epoch": 0.5601034511220961,
      "grad_norm": 0.578900166631519,
      "learning_rate": 4.271857471516686e-06,
      "loss": 0.48,
      "step": 7634
    },
    {
      "epoch": 0.5601768207122353,
      "grad_norm": 0.5439082653017003,
      "learning_rate": 4.270681962127017e-06,
      "loss": 0.4878,
      "step": 7635
    },
    {
      "epoch": 0.5602501903023744,
      "grad_norm": 0.562617022810886,
      "learning_rate": 4.269506493923744e-06,
      "loss": 0.5095,
      "step": 7636
    },
    {
      "epoch": 0.5603235598925136,
      "grad_norm": 0.5301392442610456,
      "learning_rate": 4.268331066973246e-06,
      "loss": 0.4675,
      "step": 7637
    },
    {
      "epoch": 0.5603969294826526,
      "grad_norm": 1.4885480582172632,
      "learning_rate": 4.267155681341907e-06,
      "loss": 0.7653,
      "step": 7638
    },
    {
      "epoch": 0.5604702990727918,
      "grad_norm": 1.6596858087315725,
      "learning_rate": 4.2659803370961e-06,
      "loss": 0.8233,
      "step": 7639
    },
    {
      "epoch": 0.5605436686629309,
      "grad_norm": 1.5836178742986824,
      "learning_rate": 4.264805034302201e-06,
      "loss": 0.7493,
      "step": 7640
    },
    {
      "epoch": 0.5605436686629309,
      "lm_loss": 0.6246,
      "step": 7640,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.5605436686629309,
      "lm_loss": 0.4968,
      "step": 7640,
      "vm_loss": 0.1169
    },
    {
      "epoch": 0.5605436686629309,
      "lm_loss": 0.5627,
      "step": 7640,
      "vm_loss": 0.0322
    },
    {
      "epoch": 0.5605436686629309,
      "lm_loss": 0.9135,
      "step": 7640,
      "vm_loss": 0.1996
    },
    {
      "epoch": 0.5605436686629309,
      "lm_loss": 0.6105,
      "step": 7640,
      "vm_loss": 0.0753
    },
    {
      "epoch": 0.5605436686629309,
      "lm_loss": 0.9917,
      "step": 7640,
      "vm_loss": 0.1059
    },
    {
      "epoch": 0.5605436686629309,
      "lm_loss": 0.6202,
      "step": 7640,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.5605436686629309,
      "lm_loss": 0.9101,
      "step": 7640,
      "vm_loss": 0.1556
    },
    {
      "epoch": 0.5606170382530701,
      "grad_norm": 1.4859258834288043,
      "learning_rate": 4.263629773026583e-06,
      "loss": 0.7987,
      "step": 7641
    },
    {
      "epoch": 0.5606904078432092,
      "grad_norm": 1.5988769678704586,
      "learning_rate": 4.262454553335615e-06,
      "loss": 0.779,
      "step": 7642
    },
    {
      "epoch": 0.5607637774333483,
      "grad_norm": 0.5979729608194172,
      "learning_rate": 4.261279375295665e-06,
      "loss": 0.4997,
      "step": 7643
    },
    {
      "epoch": 0.5608371470234874,
      "grad_norm": 1.6552161289387084,
      "learning_rate": 4.260104238973098e-06,
      "loss": 0.7622,
      "step": 7644
    },
    {
      "epoch": 0.5609105166136266,
      "grad_norm": 1.8044655605212343,
      "learning_rate": 4.258929144434277e-06,
      "loss": 0.835,
      "step": 7645
    },
    {
      "epoch": 0.5609838862037657,
      "grad_norm": 1.607215523570743,
      "learning_rate": 4.257754091745563e-06,
      "loss": 0.7831,
      "step": 7646
    },
    {
      "epoch": 0.5610572557939049,
      "grad_norm": 1.359938320606191,
      "learning_rate": 4.256579080973314e-06,
      "loss": 0.6575,
      "step": 7647
    },
    {
      "epoch": 0.5611306253840439,
      "grad_norm": 0.5643164757568241,
      "learning_rate": 4.255404112183884e-06,
      "loss": 0.4886,
      "step": 7648
    },
    {
      "epoch": 0.5611306253840439,
      "lm_loss": 1.3426,
      "step": 7648,
      "vm_loss": 0.1138
    },
    {
      "epoch": 0.5611306253840439,
      "lm_loss": 1.0709,
      "step": 7648,
      "vm_loss": 0.146
    },
    {
      "epoch": 0.5611306253840439,
      "lm_loss": 0.2953,
      "step": 7648,
      "vm_loss": 0.0889
    },
    {
      "epoch": 0.5611306253840439,
      "lm_loss": 0.0928,
      "step": 7648,
      "vm_loss": 0.0462
    },
    {
      "epoch": 0.5611306253840439,
      "lm_loss": 0.2219,
      "step": 7648,
      "vm_loss": 0.1176
    },
    {
      "epoch": 0.5611306253840439,
      "lm_loss": 0.6879,
      "step": 7648,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.5611306253840439,
      "lm_loss": 0.2718,
      "step": 7648,
      "vm_loss": 0.1188
    },
    {
      "epoch": 0.5611306253840439,
      "lm_loss": 0.0556,
      "step": 7648,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.561203994974183,
      "grad_norm": 1.608731578077661,
      "learning_rate": 4.254229185443628e-06,
      "loss": 0.7429,
      "step": 7649
    },
    {
      "epoch": 0.5612773645643222,
      "grad_norm": 1.6211074709289903,
      "learning_rate": 4.253054300818896e-06,
      "loss": 0.7774,
      "step": 7650
    },
    {
      "epoch": 0.5613507341544614,
      "grad_norm": 0.5685992630659833,
      "learning_rate": 4.251879458376039e-06,
      "loss": 0.5016,
      "step": 7651
    },
    {
      "epoch": 0.5614241037446005,
      "grad_norm": 1.475957170013621,
      "learning_rate": 4.250704658181402e-06,
      "loss": 0.7842,
      "step": 7652
    },
    {
      "epoch": 0.5614974733347395,
      "grad_norm": 1.6100731162585318,
      "learning_rate": 4.249529900301327e-06,
      "loss": 0.817,
      "step": 7653
    },
    {
      "epoch": 0.5615708429248787,
      "grad_norm": 1.6000787780749122,
      "learning_rate": 4.248355184802158e-06,
      "loss": 0.8843,
      "step": 7654
    },
    {
      "epoch": 0.5616442125150178,
      "grad_norm": 1.3059211892894047,
      "learning_rate": 4.247180511750233e-06,
      "loss": 0.6319,
      "step": 7655
    },
    {
      "epoch": 0.561717582105157,
      "grad_norm": 1.6416773329753385,
      "learning_rate": 4.246005881211888e-06,
      "loss": 0.8745,
      "step": 7656
    },
    {
      "epoch": 0.561717582105157,
      "lm_loss": 1.3489,
      "step": 7656,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.561717582105157,
      "lm_loss": 0.58,
      "step": 7656,
      "vm_loss": 0.2038
    },
    {
      "epoch": 0.561717582105157,
      "lm_loss": 1.2094,
      "step": 7656,
      "vm_loss": 0.0944
    },
    {
      "epoch": 0.561717582105157,
      "lm_loss": 1.023,
      "step": 7656,
      "vm_loss": 0.1401
    },
    {
      "epoch": 0.561717582105157,
      "lm_loss": 0.6187,
      "step": 7656,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.561717582105157,
      "lm_loss": 0.0415,
      "step": 7656,
      "vm_loss": 0.1287
    },
    {
      "epoch": 0.561717582105157,
      "lm_loss": 2.1136,
      "step": 7656,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.561717582105157,
      "lm_loss": 0.9066,
      "step": 7656,
      "vm_loss": 0.1444
    },
    {
      "epoch": 0.5617909516952961,
      "grad_norm": 1.549596704844764,
      "learning_rate": 4.24483129325346e-06,
      "loss": 0.792,
      "step": 7657
    },
    {
      "epoch": 0.5618643212854352,
      "grad_norm": 1.4654740253329963,
      "learning_rate": 4.243656747941278e-06,
      "loss": 0.7847,
      "step": 7658
    },
    {
      "epoch": 0.5619376908755743,
      "grad_norm": 1.590195162548893,
      "learning_rate": 4.242482245341671e-06,
      "loss": 0.7577,
      "step": 7659
    },
    {
      "epoch": 0.5620110604657135,
      "grad_norm": 1.4557187783573902,
      "learning_rate": 4.2413077855209694e-06,
      "loss": 0.735,
      "step": 7660
    },
    {
      "epoch": 0.5620844300558526,
      "grad_norm": 0.6189645077907732,
      "learning_rate": 4.2401333685454945e-06,
      "loss": 0.4823,
      "step": 7661
    },
    {
      "epoch": 0.5621577996459918,
      "grad_norm": 0.6065517905171733,
      "learning_rate": 4.238958994481572e-06,
      "loss": 0.5,
      "step": 7662
    },
    {
      "epoch": 0.5622311692361308,
      "grad_norm": 1.3914061998378857,
      "learning_rate": 4.237784663395519e-06,
      "loss": 0.7279,
      "step": 7663
    },
    {
      "epoch": 0.56230453882627,
      "grad_norm": 1.5744144188756608,
      "learning_rate": 4.236610375353653e-06,
      "loss": 0.7133,
      "step": 7664
    },
    {
      "epoch": 0.56230453882627,
      "lm_loss": 0.5905,
      "step": 7664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.56230453882627,
      "lm_loss": 0.4991,
      "step": 7664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.56230453882627,
      "lm_loss": 0.5026,
      "step": 7664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.56230453882627,
      "lm_loss": 0.4576,
      "step": 7664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.56230453882627,
      "lm_loss": 0.4098,
      "step": 7664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.56230453882627,
      "lm_loss": 0.335,
      "step": 7664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.56230453882627,
      "lm_loss": 0.5252,
      "step": 7664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.56230453882627,
      "lm_loss": 0.3065,
      "step": 7664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5623779084164091,
      "grad_norm": 0.5679574071207718,
      "learning_rate": 4.2354361304222905e-06,
      "loss": 0.501,
      "step": 7665
    },
    {
      "epoch": 0.5624512780065483,
      "grad_norm": 1.7141424336924507,
      "learning_rate": 4.2342619286677426e-06,
      "loss": 0.7403,
      "step": 7666
    },
    {
      "epoch": 0.5625246475966874,
      "grad_norm": 0.5552289704454302,
      "learning_rate": 4.23308777015632e-06,
      "loss": 0.5024,
      "step": 7667
    },
    {
      "epoch": 0.5625980171868264,
      "grad_norm": 1.429026172115712,
      "learning_rate": 4.231913654954331e-06,
      "loss": 0.7067,
      "step": 7668
    },
    {
      "epoch": 0.5626713867769656,
      "grad_norm": 1.727587981913772,
      "learning_rate": 4.230739583128078e-06,
      "loss": 0.7745,
      "step": 7669
    },
    {
      "epoch": 0.5627447563671047,
      "grad_norm": 0.5786417160125895,
      "learning_rate": 4.229565554743867e-06,
      "loss": 0.4979,
      "step": 7670
    },
    {
      "epoch": 0.5628181259572439,
      "grad_norm": 0.5846820121506994,
      "learning_rate": 4.228391569867998e-06,
      "loss": 0.5078,
      "step": 7671
    },
    {
      "epoch": 0.562891495547383,
      "grad_norm": 1.54445487120507,
      "learning_rate": 4.227217628566766e-06,
      "loss": 0.7777,
      "step": 7672
    },
    {
      "epoch": 0.562891495547383,
      "lm_loss": 1.52,
      "step": 7672,
      "vm_loss": 0.049
    },
    {
      "epoch": 0.562891495547383,
      "lm_loss": 0.5845,
      "step": 7672,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.562891495547383,
      "lm_loss": 1.0351,
      "step": 7672,
      "vm_loss": 0.2091
    },
    {
      "epoch": 0.562891495547383,
      "lm_loss": 0.712,
      "step": 7672,
      "vm_loss": 0.1528
    },
    {
      "epoch": 0.562891495547383,
      "lm_loss": 0.2809,
      "step": 7672,
      "vm_loss": 0.0924
    },
    {
      "epoch": 0.562891495547383,
      "lm_loss": 0.7131,
      "step": 7672,
      "vm_loss": 0.0726
    },
    {
      "epoch": 0.562891495547383,
      "lm_loss": 0.2163,
      "step": 7672,
      "vm_loss": 0.1049
    },
    {
      "epoch": 0.562891495547383,
      "lm_loss": 0.4326,
      "step": 7672,
      "vm_loss": 0.1573
    },
    {
      "epoch": 0.5629648651375221,
      "grad_norm": 1.5701590499198679,
      "learning_rate": 4.22604373090647e-06,
      "loss": 0.7394,
      "step": 7673
    },
    {
      "epoch": 0.5630382347276612,
      "grad_norm": 1.5439152035400476,
      "learning_rate": 4.224869876953401e-06,
      "loss": 0.7547,
      "step": 7674
    },
    {
      "epoch": 0.5631116043178004,
      "grad_norm": 1.6596930169890598,
      "learning_rate": 4.223696066773848e-06,
      "loss": 0.7346,
      "step": 7675
    },
    {
      "epoch": 0.5631849739079395,
      "grad_norm": 1.6747843892791026,
      "learning_rate": 4.222522300434102e-06,
      "loss": 0.843,
      "step": 7676
    },
    {
      "epoch": 0.5632583434980787,
      "grad_norm": 1.6194513425761545,
      "learning_rate": 4.221348578000447e-06,
      "loss": 0.7217,
      "step": 7677
    },
    {
      "epoch": 0.5633317130882177,
      "grad_norm": 0.6179991876023561,
      "learning_rate": 4.220174899539166e-06,
      "loss": 0.4985,
      "step": 7678
    },
    {
      "epoch": 0.5634050826783569,
      "grad_norm": 1.4975720137834567,
      "learning_rate": 4.2190012651165395e-06,
      "loss": 0.7663,
      "step": 7679
    },
    {
      "epoch": 0.563478452268496,
      "grad_norm": 1.612046708398536,
      "learning_rate": 4.217827674798845e-06,
      "loss": 0.7845,
      "step": 7680
    },
    {
      "epoch": 0.563478452268496,
      "lm_loss": 1.7589,
      "step": 7680,
      "vm_loss": 0.1369
    },
    {
      "epoch": 0.563478452268496,
      "lm_loss": 0.2459,
      "step": 7680,
      "vm_loss": 0.1151
    },
    {
      "epoch": 0.563478452268496,
      "lm_loss": 0.5862,
      "step": 7680,
      "vm_loss": 0.121
    },
    {
      "epoch": 0.563478452268496,
      "lm_loss": 1.0286,
      "step": 7680,
      "vm_loss": 0.1654
    },
    {
      "epoch": 0.563478452268496,
      "lm_loss": 0.505,
      "step": 7680,
      "vm_loss": 0.1275
    },
    {
      "epoch": 0.563478452268496,
      "lm_loss": 0.1099,
      "step": 7680,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.563478452268496,
      "lm_loss": 0.5671,
      "step": 7680,
      "vm_loss": 0.2201
    },
    {
      "epoch": 0.563478452268496,
      "lm_loss": 0.807,
      "step": 7680,
      "vm_loss": 0.1798
    },
    {
      "epoch": 0.5635518218586352,
      "grad_norm": 1.6602165674857567,
      "learning_rate": 4.21665412865236e-06,
      "loss": 0.7213,
      "step": 7681
    },
    {
      "epoch": 0.5636251914487743,
      "grad_norm": 0.5734349224905199,
      "learning_rate": 4.2154806267433565e-06,
      "loss": 0.503,
      "step": 7682
    },
    {
      "epoch": 0.5636985610389134,
      "grad_norm": 0.5526036558032504,
      "learning_rate": 4.214307169138103e-06,
      "loss": 0.5026,
      "step": 7683
    },
    {
      "epoch": 0.5637719306290525,
      "grad_norm": 1.609996880505302,
      "learning_rate": 4.21313375590287e-06,
      "loss": 0.7791,
      "step": 7684
    },
    {
      "epoch": 0.5638453002191917,
      "grad_norm": 1.5043034539473,
      "learning_rate": 4.211960387103924e-06,
      "loss": 0.7603,
      "step": 7685
    },
    {
      "epoch": 0.5639186698093308,
      "grad_norm": 1.4875922041983403,
      "learning_rate": 4.210787062807525e-06,
      "loss": 0.8199,
      "step": 7686
    },
    {
      "epoch": 0.56399203939947,
      "grad_norm": 1.4855380327884964,
      "learning_rate": 4.209613783079935e-06,
      "loss": 0.6783,
      "step": 7687
    },
    {
      "epoch": 0.564065408989609,
      "grad_norm": 0.5915645328543334,
      "learning_rate": 4.208440547987413e-06,
      "loss": 0.4776,
      "step": 7688
    },
    {
      "epoch": 0.564065408989609,
      "lm_loss": 1.6046,
      "step": 7688,
      "vm_loss": 0.1258
    },
    {
      "epoch": 0.564065408989609,
      "lm_loss": 0.0977,
      "step": 7688,
      "vm_loss": 0.0858
    },
    {
      "epoch": 0.564065408989609,
      "lm_loss": 0.1714,
      "step": 7688,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.564065408989609,
      "lm_loss": 0.417,
      "step": 7688,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.564065408989609,
      "lm_loss": 0.0964,
      "step": 7688,
      "vm_loss": 0.0924
    },
    {
      "epoch": 0.564065408989609,
      "lm_loss": 0.8675,
      "step": 7688,
      "vm_loss": 0.2205
    },
    {
      "epoch": 0.564065408989609,
      "lm_loss": 0.4895,
      "step": 7688,
      "vm_loss": 0.1355
    },
    {
      "epoch": 0.564065408989609,
      "lm_loss": 0.0805,
      "step": 7688,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.5641387785797481,
      "grad_norm": 1.8861596814710386,
      "learning_rate": 4.2072673575962125e-06,
      "loss": 0.7339,
      "step": 7689
    },
    {
      "epoch": 0.5642121481698873,
      "grad_norm": 0.60584108871633,
      "learning_rate": 4.206094211972588e-06,
      "loss": 0.4918,
      "step": 7690
    },
    {
      "epoch": 0.5642855177600264,
      "grad_norm": 0.573803889342263,
      "learning_rate": 4.204921111182789e-06,
      "loss": 0.4831,
      "step": 7691
    },
    {
      "epoch": 0.5643588873501656,
      "grad_norm": 1.6342238015490125,
      "learning_rate": 4.2037480552930645e-06,
      "loss": 0.7326,
      "step": 7692
    },
    {
      "epoch": 0.5644322569403046,
      "grad_norm": 0.5342270779609447,
      "learning_rate": 4.2025750443696595e-06,
      "loss": 0.4977,
      "step": 7693
    },
    {
      "epoch": 0.5645056265304438,
      "grad_norm": 1.444544544179472,
      "learning_rate": 4.201402078478815e-06,
      "loss": 0.7824,
      "step": 7694
    },
    {
      "epoch": 0.5645789961205829,
      "grad_norm": 1.4467166773955162,
      "learning_rate": 4.200229157686774e-06,
      "loss": 0.7543,
      "step": 7695
    },
    {
      "epoch": 0.5646523657107221,
      "grad_norm": 1.4548242104670204,
      "learning_rate": 4.199056282059775e-06,
      "loss": 0.8054,
      "step": 7696
    },
    {
      "epoch": 0.5646523657107221,
      "lm_loss": 1.1216,
      "step": 7696,
      "vm_loss": 0.1391
    },
    {
      "epoch": 0.5646523657107221,
      "lm_loss": 1.3687,
      "step": 7696,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.5646523657107221,
      "lm_loss": 0.0201,
      "step": 7696,
      "vm_loss": 0.0612
    },
    {
      "epoch": 0.5646523657107221,
      "lm_loss": 0.5111,
      "step": 7696,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.5646523657107221,
      "lm_loss": 1.0752,
      "step": 7696,
      "vm_loss": 0.1985
    },
    {
      "epoch": 0.5646523657107221,
      "lm_loss": 0.5311,
      "step": 7696,
      "vm_loss": 0.064
    },
    {
      "epoch": 0.5646523657107221,
      "lm_loss": 0.7681,
      "step": 7696,
      "vm_loss": 0.1426
    },
    {
      "epoch": 0.5646523657107221,
      "lm_loss": 0.4946,
      "step": 7696,
      "vm_loss": 0.1384
    },
    {
      "epoch": 0.5647257353008611,
      "grad_norm": 1.4574588931159356,
      "learning_rate": 4.1978834516640496e-06,
      "loss": 0.8211,
      "step": 7697
    },
    {
      "epoch": 0.5647991048910003,
      "grad_norm": 1.5955488455265092,
      "learning_rate": 4.19671066656583e-06,
      "loss": 0.7913,
      "step": 7698
    },
    {
      "epoch": 0.5648724744811394,
      "grad_norm": 1.5033958003323193,
      "learning_rate": 4.195537926831351e-06,
      "loss": 0.8137,
      "step": 7699
    },
    {
      "epoch": 0.5649458440712786,
      "grad_norm": 1.3449385660644317,
      "learning_rate": 4.194365232526836e-06,
      "loss": 0.6268,
      "step": 7700
    },
    {
      "epoch": 0.5650192136614177,
      "grad_norm": 1.4103542995613452,
      "learning_rate": 4.193192583718512e-06,
      "loss": 0.7181,
      "step": 7701
    },
    {
      "epoch": 0.5650925832515568,
      "grad_norm": 1.4571760484395577,
      "learning_rate": 4.192019980472601e-06,
      "loss": 0.733,
      "step": 7702
    },
    {
      "epoch": 0.5651659528416959,
      "grad_norm": 1.4808516837480727,
      "learning_rate": 4.190847422855321e-06,
      "loss": 0.6954,
      "step": 7703
    },
    {
      "epoch": 0.5652393224318351,
      "grad_norm": 1.3962446172680714,
      "learning_rate": 4.189674910932893e-06,
      "loss": 0.7251,
      "step": 7704
    },
    {
      "epoch": 0.5652393224318351,
      "lm_loss": 1.48,
      "step": 7704,
      "vm_loss": 0.0247
    },
    {
      "epoch": 0.5652393224318351,
      "lm_loss": 0.147,
      "step": 7704,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0.5652393224318351,
      "lm_loss": 0.8604,
      "step": 7704,
      "vm_loss": 0.1204
    },
    {
      "epoch": 0.5652393224318351,
      "lm_loss": 0.9387,
      "step": 7704,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.5652393224318351,
      "lm_loss": 0.1254,
      "step": 7704,
      "vm_loss": 0.0575
    },
    {
      "epoch": 0.5652393224318351,
      "lm_loss": 1.0311,
      "step": 7704,
      "vm_loss": 0.1661
    },
    {
      "epoch": 0.5652393224318351,
      "lm_loss": 0.9688,
      "step": 7704,
      "vm_loss": 0.1414
    },
    {
      "epoch": 0.5652393224318351,
      "lm_loss": 0.9897,
      "step": 7704,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.5653126920219742,
      "grad_norm": 1.4924871923336576,
      "learning_rate": 4.188502444771529e-06,
      "loss": 0.7142,
      "step": 7705
    },
    {
      "epoch": 0.5653860616121134,
      "grad_norm": 1.5127590909775117,
      "learning_rate": 4.18733002443744e-06,
      "loss": 0.8303,
      "step": 7706
    },
    {
      "epoch": 0.5654594312022524,
      "grad_norm": 0.8049210904648421,
      "learning_rate": 4.186157649996839e-06,
      "loss": 0.4908,
      "step": 7707
    },
    {
      "epoch": 0.5655328007923915,
      "grad_norm": 0.708024150621005,
      "learning_rate": 4.1849853215159305e-06,
      "loss": 0.4795,
      "step": 7708
    },
    {
      "epoch": 0.5656061703825307,
      "grad_norm": 0.6045742433662088,
      "learning_rate": 4.183813039060919e-06,
      "loss": 0.4901,
      "step": 7709
    },
    {
      "epoch": 0.5656795399726698,
      "grad_norm": 1.5660049681194013,
      "learning_rate": 4.182640802698007e-06,
      "loss": 0.7332,
      "step": 7710
    },
    {
      "epoch": 0.565752909562809,
      "grad_norm": 1.4244157519587184,
      "learning_rate": 4.181468612493391e-06,
      "loss": 0.7372,
      "step": 7711
    },
    {
      "epoch": 0.565826279152948,
      "grad_norm": 1.5382526901748919,
      "learning_rate": 4.180296468513272e-06,
      "loss": 0.7919,
      "step": 7712
    },
    {
      "epoch": 0.565826279152948,
      "lm_loss": 1.2609,
      "step": 7712,
      "vm_loss": 0.0562
    },
    {
      "epoch": 0.565826279152948,
      "lm_loss": 1.0389,
      "step": 7712,
      "vm_loss": 0.1459
    },
    {
      "epoch": 0.565826279152948,
      "lm_loss": 0.9301,
      "step": 7712,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.565826279152948,
      "lm_loss": 0.6942,
      "step": 7712,
      "vm_loss": 0.1533
    },
    {
      "epoch": 0.565826279152948,
      "lm_loss": 1.1408,
      "step": 7712,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.565826279152948,
      "lm_loss": 0.0332,
      "step": 7712,
      "vm_loss": 0.0945
    },
    {
      "epoch": 0.565826279152948,
      "lm_loss": 0.0673,
      "step": 7712,
      "vm_loss": 0.0899
    },
    {
      "epoch": 0.565826279152948,
      "lm_loss": 0.5931,
      "step": 7712,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.5658996487430872,
      "grad_norm": 1.4674923824629202,
      "learning_rate": 4.179124370823841e-06,
      "loss": 0.7232,
      "step": 7713
    },
    {
      "epoch": 0.5659730183332263,
      "grad_norm": 0.9051656555259255,
      "learning_rate": 4.177952319491288e-06,
      "loss": 0.4961,
      "step": 7714
    },
    {
      "epoch": 0.5660463879233655,
      "grad_norm": 1.93377260501924,
      "learning_rate": 4.176780314581805e-06,
      "loss": 0.729,
      "step": 7715
    },
    {
      "epoch": 0.5661197575135046,
      "grad_norm": 0.8903257174937835,
      "learning_rate": 4.1756083561615755e-06,
      "loss": 0.5095,
      "step": 7716
    },
    {
      "epoch": 0.5661931271036437,
      "grad_norm": 1.3917656682557593,
      "learning_rate": 4.174436444296783e-06,
      "loss": 0.7817,
      "step": 7717
    },
    {
      "epoch": 0.5662664966937828,
      "grad_norm": 1.4315099387228467,
      "learning_rate": 4.173264579053611e-06,
      "loss": 0.7372,
      "step": 7718
    },
    {
      "epoch": 0.566339866283922,
      "grad_norm": 1.534905798718037,
      "learning_rate": 4.1720927604982335e-06,
      "loss": 0.7953,
      "step": 7719
    },
    {
      "epoch": 0.5664132358740611,
      "grad_norm": 1.4151607956026626,
      "learning_rate": 4.170920988696829e-06,
      "loss": 0.7218,
      "step": 7720
    },
    {
      "epoch": 0.5664132358740611,
      "lm_loss": 0.5225,
      "step": 7720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5664132358740611,
      "lm_loss": 0.5461,
      "step": 7720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5664132358740611,
      "lm_loss": 0.6085,
      "step": 7720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5664132358740611,
      "lm_loss": 0.4258,
      "step": 7720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5664132358740611,
      "lm_loss": 0.5337,
      "step": 7720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5664132358740611,
      "lm_loss": 0.4759,
      "step": 7720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5664132358740611,
      "lm_loss": 0.3613,
      "step": 7720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5664132358740611,
      "lm_loss": 0.5425,
      "step": 7720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5664866054642003,
      "grad_norm": 0.6171863246549969,
      "learning_rate": 4.16974926371557e-06,
      "loss": 0.4893,
      "step": 7721
    },
    {
      "epoch": 0.5665599750543393,
      "grad_norm": 1.4602409240100183,
      "learning_rate": 4.168577585620625e-06,
      "loss": 0.7949,
      "step": 7722
    },
    {
      "epoch": 0.5666333446444785,
      "grad_norm": 1.6326641737074907,
      "learning_rate": 4.167405954478164e-06,
      "loss": 0.7237,
      "step": 7723
    },
    {
      "epoch": 0.5667067142346176,
      "grad_norm": 1.3510052050117611,
      "learning_rate": 4.1662343703543494e-06,
      "loss": 0.7263,
      "step": 7724
    },
    {
      "epoch": 0.5667800838247568,
      "grad_norm": 1.3702190328565502,
      "learning_rate": 4.165062833315345e-06,
      "loss": 0.7634,
      "step": 7725
    },
    {
      "epoch": 0.5668534534148959,
      "grad_norm": 1.333592045619754,
      "learning_rate": 4.163891343427311e-06,
      "loss": 0.6761,
      "step": 7726
    },
    {
      "epoch": 0.5669268230050349,
      "grad_norm": 0.679918950963214,
      "learning_rate": 4.162719900756403e-06,
      "loss": 0.4916,
      "step": 7727
    },
    {
      "epoch": 0.5670001925951741,
      "grad_norm": 1.550994341976868,
      "learning_rate": 4.161548505368774e-06,
      "loss": 0.7514,
      "step": 7728
    },
    {
      "epoch": 0.5670001925951741,
      "lm_loss": 0.6289,
      "step": 7728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5670001925951741,
      "lm_loss": 0.5799,
      "step": 7728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5670001925951741,
      "lm_loss": 0.3932,
      "step": 7728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5670001925951741,
      "lm_loss": 0.3973,
      "step": 7728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5670001925951741,
      "lm_loss": 0.5506,
      "step": 7728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5670001925951741,
      "lm_loss": 0.5457,
      "step": 7728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5670001925951741,
      "lm_loss": 0.5035,
      "step": 7728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5670001925951741,
      "lm_loss": 0.4257,
      "step": 7728,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5670735621853132,
      "grad_norm": 0.5990421812388809,
      "learning_rate": 4.160377157330579e-06,
      "loss": 0.4937,
      "step": 7729
    },
    {
      "epoch": 0.5671469317754524,
      "grad_norm": 1.6965578892305495,
      "learning_rate": 4.159205856707965e-06,
      "loss": 0.7837,
      "step": 7730
    },
    {
      "epoch": 0.5672203013655915,
      "grad_norm": 1.3675216364509883,
      "learning_rate": 4.1580346035670785e-06,
      "loss": 0.7387,
      "step": 7731
    },
    {
      "epoch": 0.5672936709557306,
      "grad_norm": 1.4393249487474016,
      "learning_rate": 4.156863397974063e-06,
      "loss": 0.8118,
      "step": 7732
    },
    {
      "epoch": 0.5673670405458697,
      "grad_norm": 1.5806834041553377,
      "learning_rate": 4.15569223999506e-06,
      "loss": 0.6891,
      "step": 7733
    },
    {
      "epoch": 0.5674404101360089,
      "grad_norm": 1.5073459802539746,
      "learning_rate": 4.1545211296962065e-06,
      "loss": 0.8285,
      "step": 7734
    },
    {
      "epoch": 0.567513779726148,
      "grad_norm": 1.5414098878163576,
      "learning_rate": 4.153350067143638e-06,
      "loss": 0.7235,
      "step": 7735
    },
    {
      "epoch": 0.5675871493162872,
      "grad_norm": 1.519245369462195,
      "learning_rate": 4.152179052403488e-06,
      "loss": 0.6552,
      "step": 7736
    },
    {
      "epoch": 0.5675871493162872,
      "lm_loss": 1.7087,
      "step": 7736,
      "vm_loss": 0.0574
    },
    {
      "epoch": 0.5675871493162872,
      "lm_loss": 0.2362,
      "step": 7736,
      "vm_loss": 0.0499
    },
    {
      "epoch": 0.5675871493162872,
      "lm_loss": 0.9572,
      "step": 7736,
      "vm_loss": 0.1086
    },
    {
      "epoch": 0.5675871493162872,
      "lm_loss": 0.8709,
      "step": 7736,
      "vm_loss": 0.1796
    },
    {
      "epoch": 0.5675871493162872,
      "lm_loss": 0.9641,
      "step": 7736,
      "vm_loss": 0.2622
    },
    {
      "epoch": 0.5675871493162872,
      "lm_loss": 0.2611,
      "step": 7736,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.5675871493162872,
      "lm_loss": 0.7393,
      "step": 7736,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.5675871493162872,
      "lm_loss": 0.4987,
      "step": 7736,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.5676605189064262,
      "grad_norm": 1.5173428636565174,
      "learning_rate": 4.151008085541888e-06,
      "loss": 0.7857,
      "step": 7737
    },
    {
      "epoch": 0.5677338884965654,
      "grad_norm": 1.5414091807006283,
      "learning_rate": 4.149837166624961e-06,
      "loss": 0.7594,
      "step": 7738
    },
    {
      "epoch": 0.5678072580867045,
      "grad_norm": 0.7242281836217004,
      "learning_rate": 4.148666295718837e-06,
      "loss": 0.4842,
      "step": 7739
    },
    {
      "epoch": 0.5678806276768437,
      "grad_norm": 1.4817554104785706,
      "learning_rate": 4.147495472889636e-06,
      "loss": 0.694,
      "step": 7740
    },
    {
      "epoch": 0.5679539972669828,
      "grad_norm": 0.6403325613231934,
      "learning_rate": 4.146324698203475e-06,
      "loss": 0.4956,
      "step": 7741
    },
    {
      "epoch": 0.5680273668571219,
      "grad_norm": 1.417038318453763,
      "learning_rate": 4.145153971726474e-06,
      "loss": 0.7238,
      "step": 7742
    },
    {
      "epoch": 0.568100736447261,
      "grad_norm": 1.3912526383013173,
      "learning_rate": 4.1439832935247444e-06,
      "loss": 0.7527,
      "step": 7743
    },
    {
      "epoch": 0.5681741060374002,
      "grad_norm": 1.392926012246489,
      "learning_rate": 4.142812663664398e-06,
      "loss": 0.7053,
      "step": 7744
    },
    {
      "epoch": 0.5681741060374002,
      "lm_loss": 0.3959,
      "step": 7744,
      "vm_loss": 0.0753
    },
    {
      "epoch": 0.5681741060374002,
      "lm_loss": 1.6034,
      "step": 7744,
      "vm_loss": 0.1039
    },
    {
      "epoch": 0.5681741060374002,
      "lm_loss": 1.1967,
      "step": 7744,
      "vm_loss": 0.2391
    },
    {
      "epoch": 0.5681741060374002,
      "lm_loss": 1.3853,
      "step": 7744,
      "vm_loss": 0.226
    },
    {
      "epoch": 0.5681741060374002,
      "lm_loss": 0.0451,
      "step": 7744,
      "vm_loss": 0.1495
    },
    {
      "epoch": 0.5681741060374002,
      "lm_loss": 0.4729,
      "step": 7744,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.5681741060374002,
      "lm_loss": 0.3742,
      "step": 7744,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.5681741060374002,
      "lm_loss": 0.8,
      "step": 7744,
      "vm_loss": 0.1531
    },
    {
      "epoch": 0.5682474756275393,
      "grad_norm": 1.5794712217031281,
      "learning_rate": 4.141642082211543e-06,
      "loss": 0.8844,
      "step": 7745
    },
    {
      "epoch": 0.5683208452176784,
      "grad_norm": 0.5823895466353342,
      "learning_rate": 4.140471549232286e-06,
      "loss": 0.4916,
      "step": 7746
    },
    {
      "epoch": 0.5683942148078175,
      "grad_norm": 1.3748923702514388,
      "learning_rate": 4.139301064792729e-06,
      "loss": 0.7811,
      "step": 7747
    },
    {
      "epoch": 0.5684675843979566,
      "grad_norm": 1.6134509229629017,
      "learning_rate": 4.1381306289589725e-06,
      "loss": 0.9009,
      "step": 7748
    },
    {
      "epoch": 0.5685409539880958,
      "grad_norm": 1.4771635210565102,
      "learning_rate": 4.136960241797113e-06,
      "loss": 0.7824,
      "step": 7749
    },
    {
      "epoch": 0.5686143235782349,
      "grad_norm": 1.6652925790104578,
      "learning_rate": 4.1357899033732465e-06,
      "loss": 0.9714,
      "step": 7750
    },
    {
      "epoch": 0.5686876931683741,
      "grad_norm": 1.5225757204656591,
      "learning_rate": 4.134619613753465e-06,
      "loss": 0.7518,
      "step": 7751
    },
    {
      "epoch": 0.5687610627585131,
      "grad_norm": 1.320589609756874,
      "learning_rate": 4.133449373003855e-06,
      "loss": 0.682,
      "step": 7752
    },
    {
      "epoch": 0.5687610627585131,
      "lm_loss": 0.5314,
      "step": 7752,
      "vm_loss": 0.0527
    },
    {
      "epoch": 0.5687610627585131,
      "lm_loss": 0.2941,
      "step": 7752,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.5687610627585131,
      "lm_loss": 0.1104,
      "step": 7752,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.5687610627585131,
      "lm_loss": 0.1068,
      "step": 7752,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.5687610627585131,
      "lm_loss": 0.1007,
      "step": 7752,
      "vm_loss": 0.1742
    },
    {
      "epoch": 0.5687610627585131,
      "lm_loss": 0.3282,
      "step": 7752,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.5687610627585131,
      "lm_loss": 0.6989,
      "step": 7752,
      "vm_loss": 0.1712
    },
    {
      "epoch": 0.5687610627585131,
      "lm_loss": 0.6599,
      "step": 7752,
      "vm_loss": 0.1165
    },
    {
      "epoch": 0.5688344323486523,
      "grad_norm": 1.7244190961456927,
      "learning_rate": 4.132279181190507e-06,
      "loss": 0.8226,
      "step": 7753
    },
    {
      "epoch": 0.5689078019387914,
      "grad_norm": 1.53496728916527,
      "learning_rate": 4.131109038379502e-06,
      "loss": 0.7922,
      "step": 7754
    },
    {
      "epoch": 0.5689811715289306,
      "grad_norm": 1.5053667805773414,
      "learning_rate": 4.129938944636921e-06,
      "loss": 0.7801,
      "step": 7755
    },
    {
      "epoch": 0.5690545411190697,
      "grad_norm": 1.5363089671019543,
      "learning_rate": 4.128768900028843e-06,
      "loss": 0.8528,
      "step": 7756
    },
    {
      "epoch": 0.5691279107092088,
      "grad_norm": 1.5007881918063264,
      "learning_rate": 4.127598904621341e-06,
      "loss": 0.7612,
      "step": 7757
    },
    {
      "epoch": 0.5692012802993479,
      "grad_norm": 1.3624409331358933,
      "learning_rate": 4.1264289584804915e-06,
      "loss": 0.7757,
      "step": 7758
    },
    {
      "epoch": 0.5692746498894871,
      "grad_norm": 0.6617613416631173,
      "learning_rate": 4.125259061672362e-06,
      "loss": 0.48,
      "step": 7759
    },
    {
      "epoch": 0.5693480194796262,
      "grad_norm": 0.6070238152218957,
      "learning_rate": 4.124089214263017e-06,
      "loss": 0.489,
      "step": 7760
    },
    {
      "epoch": 0.5693480194796262,
      "lm_loss": 0.5785,
      "step": 7760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5693480194796262,
      "lm_loss": 0.5137,
      "step": 7760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5693480194796262,
      "lm_loss": 0.6173,
      "step": 7760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5693480194796262,
      "lm_loss": 0.4837,
      "step": 7760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5693480194796262,
      "lm_loss": 0.4496,
      "step": 7760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5693480194796262,
      "lm_loss": 0.3731,
      "step": 7760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5693480194796262,
      "lm_loss": 0.5532,
      "step": 7760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5693480194796262,
      "lm_loss": 0.6044,
      "step": 7760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5694213890697654,
      "grad_norm": 0.58330653869569,
      "learning_rate": 4.122919416318526e-06,
      "loss": 0.5056,
      "step": 7761
    },
    {
      "epoch": 0.5694947586599044,
      "grad_norm": 1.3128172205910535,
      "learning_rate": 4.121749667904947e-06,
      "loss": 0.7053,
      "step": 7762
    },
    {
      "epoch": 0.5695681282500435,
      "grad_norm": 1.4282940636902086,
      "learning_rate": 4.120579969088338e-06,
      "loss": 0.6598,
      "step": 7763
    },
    {
      "epoch": 0.5696414978401827,
      "grad_norm": 1.381933977196133,
      "learning_rate": 4.119410319934757e-06,
      "loss": 0.788,
      "step": 7764
    },
    {
      "epoch": 0.5697148674303218,
      "grad_norm": 1.6323769082000887,
      "learning_rate": 4.118240720510255e-06,
      "loss": 0.7124,
      "step": 7765
    },
    {
      "epoch": 0.569788237020461,
      "grad_norm": 0.6820378012846079,
      "learning_rate": 4.117071170880885e-06,
      "loss": 0.4994,
      "step": 7766
    },
    {
      "epoch": 0.5698616066106,
      "grad_norm": 1.5794197359294386,
      "learning_rate": 4.1159016711126915e-06,
      "loss": 0.794,
      "step": 7767
    },
    {
      "epoch": 0.5699349762007392,
      "grad_norm": 1.594745030415688,
      "learning_rate": 4.114732221271719e-06,
      "loss": 0.7943,
      "step": 7768
    },
    {
      "epoch": 0.5699349762007392,
      "lm_loss": 0.457,
      "step": 7768,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.5699349762007392,
      "lm_loss": 0.9808,
      "step": 7768,
      "vm_loss": 0.1052
    },
    {
      "epoch": 0.5699349762007392,
      "lm_loss": 0.3865,
      "step": 7768,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.5699349762007392,
      "lm_loss": 0.3097,
      "step": 7768,
      "vm_loss": 0.1318
    },
    {
      "epoch": 0.5699349762007392,
      "lm_loss": 0.7799,
      "step": 7768,
      "vm_loss": 0.1782
    },
    {
      "epoch": 0.5699349762007392,
      "lm_loss": 0.068,
      "step": 7768,
      "vm_loss": 0.0491
    },
    {
      "epoch": 0.5699349762007392,
      "lm_loss": 0.0627,
      "step": 7768,
      "vm_loss": 0.1042
    },
    {
      "epoch": 0.5699349762007392,
      "lm_loss": 0.1693,
      "step": 7768,
      "vm_loss": 0.1619
    },
    {
      "epoch": 0.5700083457908783,
      "grad_norm": 1.3430550690417642,
      "learning_rate": 4.113562821424012e-06,
      "loss": 0.6625,
      "step": 7769
    },
    {
      "epoch": 0.5700817153810175,
      "grad_norm": 1.5917998042161539,
      "learning_rate": 4.112393471635607e-06,
      "loss": 0.7722,
      "step": 7770
    },
    {
      "epoch": 0.5701550849711566,
      "grad_norm": 1.4350001724380448,
      "learning_rate": 4.111224171972541e-06,
      "loss": 0.7492,
      "step": 7771
    },
    {
      "epoch": 0.5702284545612957,
      "grad_norm": 1.4620532352939812,
      "learning_rate": 4.110054922500847e-06,
      "loss": 0.8305,
      "step": 7772
    },
    {
      "epoch": 0.5703018241514348,
      "grad_norm": 0.6176849167835242,
      "learning_rate": 4.108885723286554e-06,
      "loss": 0.4876,
      "step": 7773
    },
    {
      "epoch": 0.570375193741574,
      "grad_norm": 1.5457906360732026,
      "learning_rate": 4.107716574395693e-06,
      "loss": 0.7922,
      "step": 7774
    },
    {
      "epoch": 0.5704485633317131,
      "grad_norm": 0.5908815493031089,
      "learning_rate": 4.106547475894286e-06,
      "loss": 0.502,
      "step": 7775
    },
    {
      "epoch": 0.5705219329218523,
      "grad_norm": 1.5338134683728117,
      "learning_rate": 4.1053784278483555e-06,
      "loss": 0.7487,
      "step": 7776
    },
    {
      "epoch": 0.5705219329218523,
      "lm_loss": 1.4103,
      "step": 7776,
      "vm_loss": 0.0761
    },
    {
      "epoch": 0.5705219329218523,
      "lm_loss": 0.0185,
      "step": 7776,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.5705219329218523,
      "lm_loss": 0.3777,
      "step": 7776,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.5705219329218523,
      "lm_loss": 0.9393,
      "step": 7776,
      "vm_loss": 0.1627
    },
    {
      "epoch": 0.5705219329218523,
      "lm_loss": 0.3633,
      "step": 7776,
      "vm_loss": 0.1315
    },
    {
      "epoch": 0.5705219329218523,
      "lm_loss": 0.3828,
      "step": 7776,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.5705219329218523,
      "lm_loss": 0.775,
      "step": 7776,
      "vm_loss": 0.2438
    },
    {
      "epoch": 0.5705219329218523,
      "lm_loss": 0.6189,
      "step": 7776,
      "vm_loss": 0.073
    },
    {
      "epoch": 0.5705953025119913,
      "grad_norm": 1.4527038796099208,
      "learning_rate": 4.104209430323921e-06,
      "loss": 0.7793,
      "step": 7777
    },
    {
      "epoch": 0.5706686721021305,
      "grad_norm": 1.55238719285691,
      "learning_rate": 4.103040483386998e-06,
      "loss": 0.8859,
      "step": 7778
    },
    {
      "epoch": 0.5707420416922696,
      "grad_norm": 1.5201177462116608,
      "learning_rate": 4.101871587103601e-06,
      "loss": 0.8263,
      "step": 7779
    },
    {
      "epoch": 0.5708154112824088,
      "grad_norm": 1.7508367396303195,
      "learning_rate": 4.100702741539739e-06,
      "loss": 0.7385,
      "step": 7780
    },
    {
      "epoch": 0.5708887808725478,
      "grad_norm": 0.642755737593707,
      "learning_rate": 4.09953394676142e-06,
      "loss": 0.511,
      "step": 7781
    },
    {
      "epoch": 0.570962150462687,
      "grad_norm": 1.4060926739091626,
      "learning_rate": 4.098365202834649e-06,
      "loss": 0.7155,
      "step": 7782
    },
    {
      "epoch": 0.5710355200528261,
      "grad_norm": 1.6097256289252562,
      "learning_rate": 4.097196509825429e-06,
      "loss": 0.7103,
      "step": 7783
    },
    {
      "epoch": 0.5711088896429652,
      "grad_norm": 0.5831309859613164,
      "learning_rate": 4.096027867799756e-06,
      "loss": 0.4913,
      "step": 7784
    },
    {
      "epoch": 0.5711088896429652,
      "lm_loss": 2.044,
      "step": 7784,
      "vm_loss": 0.0678
    },
    {
      "epoch": 0.5711088896429652,
      "lm_loss": 0.9989,
      "step": 7784,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.5711088896429652,
      "lm_loss": 0.0619,
      "step": 7784,
      "vm_loss": 0.1879
    },
    {
      "epoch": 0.5711088896429652,
      "lm_loss": 0.5007,
      "step": 7784,
      "vm_loss": 0.2262
    },
    {
      "epoch": 0.5711088896429652,
      "lm_loss": 0.5184,
      "step": 7784,
      "vm_loss": 0.0748
    },
    {
      "epoch": 0.5711088896429652,
      "lm_loss": 0.8102,
      "step": 7784,
      "vm_loss": 0.2009
    },
    {
      "epoch": 0.5711088896429652,
      "lm_loss": 0.88,
      "step": 7784,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.5711088896429652,
      "lm_loss": 0.6234,
      "step": 7784,
      "vm_loss": 0.162
    },
    {
      "epoch": 0.5711822592331044,
      "grad_norm": 1.4538475230723102,
      "learning_rate": 4.094859276823629e-06,
      "loss": 0.8468,
      "step": 7785
    },
    {
      "epoch": 0.5712556288232434,
      "grad_norm": 1.5182960145202893,
      "learning_rate": 4.093690736963041e-06,
      "loss": 0.7594,
      "step": 7786
    },
    {
      "epoch": 0.5713289984133826,
      "grad_norm": 1.3243920172249481,
      "learning_rate": 4.092522248283979e-06,
      "loss": 0.7124,
      "step": 7787
    },
    {
      "epoch": 0.5714023680035217,
      "grad_norm": 0.5782457924139225,
      "learning_rate": 4.091353810852436e-06,
      "loss": 0.4818,
      "step": 7788
    },
    {
      "epoch": 0.5714757375936609,
      "grad_norm": 0.5913096642572985,
      "learning_rate": 4.090185424734392e-06,
      "loss": 0.4997,
      "step": 7789
    },
    {
      "epoch": 0.5715491071838,
      "grad_norm": 0.5559820725115736,
      "learning_rate": 4.08901708999583e-06,
      "loss": 0.4946,
      "step": 7790
    },
    {
      "epoch": 0.5716224767739391,
      "grad_norm": 1.811781223518302,
      "learning_rate": 4.087848806702729e-06,
      "loss": 0.759,
      "step": 7791
    },
    {
      "epoch": 0.5716958463640782,
      "grad_norm": 1.4359756466756812,
      "learning_rate": 4.086680574921065e-06,
      "loss": 0.7389,
      "step": 7792
    },
    {
      "epoch": 0.5716958463640782,
      "lm_loss": 1.5247,
      "step": 7792,
      "vm_loss": 0.0563
    },
    {
      "epoch": 0.5716958463640782,
      "lm_loss": 0.3697,
      "step": 7792,
      "vm_loss": 0.0558
    },
    {
      "epoch": 0.5716958463640782,
      "lm_loss": 0.2521,
      "step": 7792,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.5716958463640782,
      "lm_loss": 0.7709,
      "step": 7792,
      "vm_loss": 0.1079
    },
    {
      "epoch": 0.5716958463640782,
      "lm_loss": 0.5026,
      "step": 7792,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.5716958463640782,
      "lm_loss": 0.6122,
      "step": 7792,
      "vm_loss": 0.1588
    },
    {
      "epoch": 0.5716958463640782,
      "lm_loss": 0.5908,
      "step": 7792,
      "vm_loss": 0.1327
    },
    {
      "epoch": 0.5716958463640782,
      "lm_loss": 0.7667,
      "step": 7792,
      "vm_loss": 0.1919
    },
    {
      "epoch": 0.5717692159542174,
      "grad_norm": 1.4186305096546339,
      "learning_rate": 4.085512394716811e-06,
      "loss": 0.9179,
      "step": 7793
    },
    {
      "epoch": 0.5718425855443565,
      "grad_norm": 1.4102846603342316,
      "learning_rate": 4.0843442661559365e-06,
      "loss": 0.7865,
      "step": 7794
    },
    {
      "epoch": 0.5719159551344957,
      "grad_norm": 1.4650369198729463,
      "learning_rate": 4.0831761893044085e-06,
      "loss": 0.7302,
      "step": 7795
    },
    {
      "epoch": 0.5719893247246347,
      "grad_norm": 1.426364008341397,
      "learning_rate": 4.082008164228192e-06,
      "loss": 0.7647,
      "step": 7796
    },
    {
      "epoch": 0.5720626943147739,
      "grad_norm": 1.373332757056959,
      "learning_rate": 4.080840190993248e-06,
      "loss": 0.7248,
      "step": 7797
    },
    {
      "epoch": 0.572136063904913,
      "grad_norm": 1.5804288906160358,
      "learning_rate": 4.079672269665534e-06,
      "loss": 0.798,
      "step": 7798
    },
    {
      "epoch": 0.5722094334950522,
      "grad_norm": 2.094317331803245,
      "learning_rate": 4.0785044003110075e-06,
      "loss": 0.8151,
      "step": 7799
    },
    {
      "epoch": 0.5722828030851913,
      "grad_norm": 1.47336951547927,
      "learning_rate": 4.077336582995618e-06,
      "loss": 0.8085,
      "step": 7800
    },
    {
      "epoch": 0.5722828030851913,
      "lm_loss": 1.4479,
      "step": 7800,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.5722828030851913,
      "lm_loss": 0.6126,
      "step": 7800,
      "vm_loss": 0.1301
    },
    {
      "epoch": 0.5722828030851913,
      "lm_loss": 0.873,
      "step": 7800,
      "vm_loss": 0.1901
    },
    {
      "epoch": 0.5722828030851913,
      "lm_loss": 0.4969,
      "step": 7800,
      "vm_loss": 0.1296
    },
    {
      "epoch": 0.5722828030851913,
      "lm_loss": 0.3913,
      "step": 7800,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.5722828030851913,
      "lm_loss": 0.8542,
      "step": 7800,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.5722828030851913,
      "lm_loss": 0.9642,
      "step": 7800,
      "vm_loss": 0.1295
    },
    {
      "epoch": 0.5722828030851913,
      "lm_loss": 0.6893,
      "step": 7800,
      "vm_loss": 0.1608
    },
    {
      "epoch": 0.5723561726753303,
      "grad_norm": 1.4673071960408948,
      "learning_rate": 4.076168817785317e-06,
      "loss": 0.8442,
      "step": 7801
    },
    {
      "epoch": 0.5724295422654695,
      "grad_norm": 0.6262205302025246,
      "learning_rate": 4.075001104746051e-06,
      "loss": 0.4875,
      "step": 7802
    },
    {
      "epoch": 0.5725029118556086,
      "grad_norm": 1.4346058129847872,
      "learning_rate": 4.073833443943763e-06,
      "loss": 0.743,
      "step": 7803
    },
    {
      "epoch": 0.5725762814457478,
      "grad_norm": 1.4634660062513372,
      "learning_rate": 4.072665835444395e-06,
      "loss": 0.736,
      "step": 7804
    },
    {
      "epoch": 0.5726496510358869,
      "grad_norm": 0.5733001753421708,
      "learning_rate": 4.071498279313883e-06,
      "loss": 0.4857,
      "step": 7805
    },
    {
      "epoch": 0.572723020626026,
      "grad_norm": 0.5565943494372547,
      "learning_rate": 4.070330775618162e-06,
      "loss": 0.496,
      "step": 7806
    },
    {
      "epoch": 0.5727963902161651,
      "grad_norm": 1.5006582672280424,
      "learning_rate": 4.069163324423165e-06,
      "loss": 0.7334,
      "step": 7807
    },
    {
      "epoch": 0.5728697598063043,
      "grad_norm": 1.4652020527264016,
      "learning_rate": 4.067995925794821e-06,
      "loss": 0.7605,
      "step": 7808
    },
    {
      "epoch": 0.5728697598063043,
      "lm_loss": 0.9152,
      "step": 7808,
      "vm_loss": 0.0755
    },
    {
      "epoch": 0.5728697598063043,
      "lm_loss": 1.0188,
      "step": 7808,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.5728697598063043,
      "lm_loss": 0.8149,
      "step": 7808,
      "vm_loss": 0.1568
    },
    {
      "epoch": 0.5728697598063043,
      "lm_loss": 0.6298,
      "step": 7808,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.5728697598063043,
      "lm_loss": 1.2274,
      "step": 7808,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.5728697598063043,
      "lm_loss": 0.9339,
      "step": 7808,
      "vm_loss": 0.1138
    },
    {
      "epoch": 0.5728697598063043,
      "lm_loss": 0.5591,
      "step": 7808,
      "vm_loss": 0.1209
    },
    {
      "epoch": 0.5728697598063043,
      "lm_loss": 0.7619,
      "step": 7808,
      "vm_loss": 0.0649
    },
    {
      "epoch": 0.5729431293964434,
      "grad_norm": 1.6559100151648294,
      "learning_rate": 4.066828579799054e-06,
      "loss": 0.7803,
      "step": 7809
    },
    {
      "epoch": 0.5730164989865826,
      "grad_norm": 1.6930773192228008,
      "learning_rate": 4.0656612865017885e-06,
      "loss": 0.7339,
      "step": 7810
    },
    {
      "epoch": 0.5730898685767216,
      "grad_norm": 0.5873218413653014,
      "learning_rate": 4.064494045968944e-06,
      "loss": 0.4861,
      "step": 7811
    },
    {
      "epoch": 0.5731632381668608,
      "grad_norm": 1.5909286118473018,
      "learning_rate": 4.063326858266438e-06,
      "loss": 0.8497,
      "step": 7812
    },
    {
      "epoch": 0.5732366077569999,
      "grad_norm": 1.450977962615102,
      "learning_rate": 4.062159723460183e-06,
      "loss": 0.7814,
      "step": 7813
    },
    {
      "epoch": 0.5733099773471391,
      "grad_norm": 0.5938277784039289,
      "learning_rate": 4.060992641616091e-06,
      "loss": 0.4872,
      "step": 7814
    },
    {
      "epoch": 0.5733833469372782,
      "grad_norm": 0.5758800392693966,
      "learning_rate": 4.05982561280007e-06,
      "loss": 0.5061,
      "step": 7815
    },
    {
      "epoch": 0.5734567165274173,
      "grad_norm": 1.7207489490052912,
      "learning_rate": 4.058658637078026e-06,
      "loss": 0.8727,
      "step": 7816
    },
    {
      "epoch": 0.5734567165274173,
      "lm_loss": 0.1542,
      "step": 7816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5734567165274173,
      "lm_loss": 0.4956,
      "step": 7816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5734567165274173,
      "lm_loss": 0.5019,
      "step": 7816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5734567165274173,
      "lm_loss": 0.3098,
      "step": 7816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5734567165274173,
      "lm_loss": 0.4727,
      "step": 7816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5734567165274173,
      "lm_loss": 0.4586,
      "step": 7816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5734567165274173,
      "lm_loss": 0.5738,
      "step": 7816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5734567165274173,
      "lm_loss": 0.4157,
      "step": 7816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5735300861175564,
      "grad_norm": 0.569785460752161,
      "learning_rate": 4.057491714515858e-06,
      "loss": 0.4717,
      "step": 7817
    },
    {
      "epoch": 0.5736034557076956,
      "grad_norm": 1.7047817470119995,
      "learning_rate": 4.056324845179469e-06,
      "loss": 0.7694,
      "step": 7818
    },
    {
      "epoch": 0.5736768252978347,
      "grad_norm": 1.6273625590637437,
      "learning_rate": 4.055158029134752e-06,
      "loss": 0.6832,
      "step": 7819
    },
    {
      "epoch": 0.5737501948879739,
      "grad_norm": 1.4208049319675562,
      "learning_rate": 4.053991266447601e-06,
      "loss": 0.6038,
      "step": 7820
    },
    {
      "epoch": 0.5738235644781129,
      "grad_norm": 1.4570180982813343,
      "learning_rate": 4.052824557183906e-06,
      "loss": 0.7141,
      "step": 7821
    },
    {
      "epoch": 0.573896934068252,
      "grad_norm": 0.598251154372337,
      "learning_rate": 4.051657901409553e-06,
      "loss": 0.4973,
      "step": 7822
    },
    {
      "epoch": 0.5739703036583912,
      "grad_norm": 0.5984092101793048,
      "learning_rate": 4.0504912991904286e-06,
      "loss": 0.523,
      "step": 7823
    },
    {
      "epoch": 0.5740436732485303,
      "grad_norm": 0.5829718300101787,
      "learning_rate": 4.049324750592411e-06,
      "loss": 0.4834,
      "step": 7824
    },
    {
      "epoch": 0.5740436732485303,
      "lm_loss": 0.1609,
      "step": 7824,
      "vm_loss": 0.0774
    },
    {
      "epoch": 0.5740436732485303,
      "lm_loss": 1.4277,
      "step": 7824,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.5740436732485303,
      "lm_loss": 0.8337,
      "step": 7824,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.5740436732485303,
      "lm_loss": 0.906,
      "step": 7824,
      "vm_loss": 0.1866
    },
    {
      "epoch": 0.5740436732485303,
      "lm_loss": 0.0692,
      "step": 7824,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.5740436732485303,
      "lm_loss": 0.5587,
      "step": 7824,
      "vm_loss": 0.1965
    },
    {
      "epoch": 0.5740436732485303,
      "lm_loss": 0.1142,
      "step": 7824,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.5740436732485303,
      "lm_loss": 0.8084,
      "step": 7824,
      "vm_loss": 0.1099
    },
    {
      "epoch": 0.5741170428386695,
      "grad_norm": 1.4735133870720454,
      "learning_rate": 4.048158255681379e-06,
      "loss": 0.7576,
      "step": 7825
    },
    {
      "epoch": 0.5741904124288085,
      "grad_norm": 1.4703062780430944,
      "learning_rate": 4.046991814523208e-06,
      "loss": 0.7534,
      "step": 7826
    },
    {
      "epoch": 0.5742637820189477,
      "grad_norm": 1.4875061961671505,
      "learning_rate": 4.0458254271837675e-06,
      "loss": 0.8003,
      "step": 7827
    },
    {
      "epoch": 0.5743371516090868,
      "grad_norm": 1.4259082574694182,
      "learning_rate": 4.04465909372893e-06,
      "loss": 0.7802,
      "step": 7828
    },
    {
      "epoch": 0.574410521199226,
      "grad_norm": 1.4918749036162782,
      "learning_rate": 4.043492814224559e-06,
      "loss": 0.6803,
      "step": 7829
    },
    {
      "epoch": 0.5744838907893651,
      "grad_norm": 1.560520401532413,
      "learning_rate": 4.042326588736517e-06,
      "loss": 0.6973,
      "step": 7830
    },
    {
      "epoch": 0.5745572603795042,
      "grad_norm": 1.4228473152373478,
      "learning_rate": 4.041160417330665e-06,
      "loss": 0.7578,
      "step": 7831
    },
    {
      "epoch": 0.5746306299696433,
      "grad_norm": 0.6056311983535071,
      "learning_rate": 4.039994300072859e-06,
      "loss": 0.5061,
      "step": 7832
    },
    {
      "epoch": 0.5746306299696433,
      "lm_loss": 1.3181,
      "step": 7832,
      "vm_loss": 0.1443
    },
    {
      "epoch": 0.5746306299696433,
      "lm_loss": 0.1517,
      "step": 7832,
      "vm_loss": 0.1609
    },
    {
      "epoch": 0.5746306299696433,
      "lm_loss": 0.2474,
      "step": 7832,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.5746306299696433,
      "lm_loss": 0.75,
      "step": 7832,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.5746306299696433,
      "lm_loss": 0.785,
      "step": 7832,
      "vm_loss": 0.2122
    },
    {
      "epoch": 0.5746306299696433,
      "lm_loss": 1.0909,
      "step": 7832,
      "vm_loss": 0.2102
    },
    {
      "epoch": 0.5746306299696433,
      "lm_loss": 0.0866,
      "step": 7832,
      "vm_loss": 0.0586
    },
    {
      "epoch": 0.5746306299696433,
      "lm_loss": 0.0052,
      "step": 7832,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.5747039995597825,
      "grad_norm": 1.4499846117167354,
      "learning_rate": 4.038828237028951e-06,
      "loss": 0.7796,
      "step": 7833
    },
    {
      "epoch": 0.5747773691499216,
      "grad_norm": 0.5670056945579842,
      "learning_rate": 4.037662228264793e-06,
      "loss": 0.4863,
      "step": 7834
    },
    {
      "epoch": 0.5748507387400608,
      "grad_norm": 1.6655467825224055,
      "learning_rate": 4.036496273846233e-06,
      "loss": 0.8111,
      "step": 7835
    },
    {
      "epoch": 0.5749241083301998,
      "grad_norm": 1.5417371252731444,
      "learning_rate": 4.035330373839113e-06,
      "loss": 0.8522,
      "step": 7836
    },
    {
      "epoch": 0.574997477920339,
      "grad_norm": 0.5601523842655772,
      "learning_rate": 4.0341645283092775e-06,
      "loss": 0.4713,
      "step": 7837
    },
    {
      "epoch": 0.5750708475104781,
      "grad_norm": 0.5482177345301695,
      "learning_rate": 4.032998737322561e-06,
      "loss": 0.4943,
      "step": 7838
    },
    {
      "epoch": 0.5751442171006172,
      "grad_norm": 0.5706362402489207,
      "learning_rate": 4.031833000944803e-06,
      "loss": 0.4892,
      "step": 7839
    },
    {
      "epoch": 0.5752175866907564,
      "grad_norm": 1.688767161734339,
      "learning_rate": 4.030667319241832e-06,
      "loss": 0.8035,
      "step": 7840
    },
    {
      "epoch": 0.5752175866907564,
      "lm_loss": 1.2178,
      "step": 7840,
      "vm_loss": 0.0244
    },
    {
      "epoch": 0.5752175866907564,
      "lm_loss": 0.1027,
      "step": 7840,
      "vm_loss": 0.0931
    },
    {
      "epoch": 0.5752175866907564,
      "lm_loss": 0.254,
      "step": 7840,
      "vm_loss": 0.2047
    },
    {
      "epoch": 0.5752175866907564,
      "lm_loss": 0.9938,
      "step": 7840,
      "vm_loss": 0.1507
    },
    {
      "epoch": 0.5752175866907564,
      "lm_loss": 0.7487,
      "step": 7840,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.5752175866907564,
      "lm_loss": 0.5706,
      "step": 7840,
      "vm_loss": 0.0651
    },
    {
      "epoch": 0.5752175866907564,
      "lm_loss": 0.5612,
      "step": 7840,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.5752175866907564,
      "lm_loss": 0.085,
      "step": 7840,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.5752909562808954,
      "grad_norm": 1.4328160268053993,
      "learning_rate": 4.029501692279478e-06,
      "loss": 0.6818,
      "step": 7841
    },
    {
      "epoch": 0.5753643258710346,
      "grad_norm": 1.4514738423205402,
      "learning_rate": 4.028336120123566e-06,
      "loss": 0.7757,
      "step": 7842
    },
    {
      "epoch": 0.5754376954611737,
      "grad_norm": 1.369775674919457,
      "learning_rate": 4.0271706028399205e-06,
      "loss": 0.6893,
      "step": 7843
    },
    {
      "epoch": 0.5755110650513129,
      "grad_norm": 1.482008822630063,
      "learning_rate": 4.0260051404943594e-06,
      "loss": 0.6783,
      "step": 7844
    },
    {
      "epoch": 0.575584434641452,
      "grad_norm": 1.7271630568756535,
      "learning_rate": 4.0248397331527014e-06,
      "loss": 0.7316,
      "step": 7845
    },
    {
      "epoch": 0.5756578042315911,
      "grad_norm": 1.5046337175395594,
      "learning_rate": 4.023674380880759e-06,
      "loss": 0.7808,
      "step": 7846
    },
    {
      "epoch": 0.5757311738217302,
      "grad_norm": 1.4524325507506182,
      "learning_rate": 4.022509083744341e-06,
      "loss": 0.7077,
      "step": 7847
    },
    {
      "epoch": 0.5758045434118694,
      "grad_norm": 1.5676467992253569,
      "learning_rate": 4.021343841809255e-06,
      "loss": 0.8012,
      "step": 7848
    },
    {
      "epoch": 0.5758045434118694,
      "lm_loss": 1.5854,
      "step": 7848,
      "vm_loss": 0.0424
    },
    {
      "epoch": 0.5758045434118694,
      "lm_loss": 0.0202,
      "step": 7848,
      "vm_loss": 0.1541
    },
    {
      "epoch": 0.5758045434118694,
      "lm_loss": 0.0579,
      "step": 7848,
      "vm_loss": 0.1175
    },
    {
      "epoch": 0.5758045434118694,
      "lm_loss": 0.4706,
      "step": 7848,
      "vm_loss": 0.1795
    },
    {
      "epoch": 0.5758045434118694,
      "lm_loss": 0.8884,
      "step": 7848,
      "vm_loss": 0.0949
    },
    {
      "epoch": 0.5758045434118694,
      "lm_loss": 0.6662,
      "step": 7848,
      "vm_loss": 0.1988
    },
    {
      "epoch": 0.5758045434118694,
      "lm_loss": 0.1175,
      "step": 7848,
      "vm_loss": 0.1839
    },
    {
      "epoch": 0.5758045434118694,
      "lm_loss": 1.1285,
      "step": 7848,
      "vm_loss": 0.1734
    },
    {
      "epoch": 0.5758779130020085,
      "grad_norm": 1.530286772624301,
      "learning_rate": 4.020178655141307e-06,
      "loss": 0.8132,
      "step": 7849
    },
    {
      "epoch": 0.5759512825921477,
      "grad_norm": 1.621243787979492,
      "learning_rate": 4.0190135238062974e-06,
      "loss": 0.7195,
      "step": 7850
    },
    {
      "epoch": 0.5760246521822867,
      "grad_norm": 1.5232548321309665,
      "learning_rate": 4.017848447870022e-06,
      "loss": 0.7067,
      "step": 7851
    },
    {
      "epoch": 0.5760980217724259,
      "grad_norm": 1.9062614267199465,
      "learning_rate": 4.0166834273982776e-06,
      "loss": 0.9091,
      "step": 7852
    },
    {
      "epoch": 0.576171391362565,
      "grad_norm": 0.6552357127334141,
      "learning_rate": 4.015518462456854e-06,
      "loss": 0.4895,
      "step": 7853
    },
    {
      "epoch": 0.5762447609527042,
      "grad_norm": 0.6146871303527552,
      "learning_rate": 4.014353553111543e-06,
      "loss": 0.5024,
      "step": 7854
    },
    {
      "epoch": 0.5763181305428433,
      "grad_norm": 1.4096064061775762,
      "learning_rate": 4.013188699428127e-06,
      "loss": 0.7837,
      "step": 7855
    },
    {
      "epoch": 0.5763915001329823,
      "grad_norm": 1.6909638101064433,
      "learning_rate": 4.012023901472388e-06,
      "loss": 0.861,
      "step": 7856
    },
    {
      "epoch": 0.5763915001329823,
      "lm_loss": 0.6584,
      "step": 7856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5763915001329823,
      "lm_loss": 0.6681,
      "step": 7856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5763915001329823,
      "lm_loss": 0.4751,
      "step": 7856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5763915001329823,
      "lm_loss": 0.3602,
      "step": 7856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5763915001329823,
      "lm_loss": 0.5198,
      "step": 7856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5763915001329823,
      "lm_loss": 0.4129,
      "step": 7856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5763915001329823,
      "lm_loss": 0.3822,
      "step": 7856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5763915001329823,
      "lm_loss": 0.5277,
      "step": 7856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5764648697231215,
      "grad_norm": 0.5837145911608245,
      "learning_rate": 4.010859159310106e-06,
      "loss": 0.501,
      "step": 7857
    },
    {
      "epoch": 0.5765382393132606,
      "grad_norm": 1.4362058179112633,
      "learning_rate": 4.009694473007059e-06,
      "loss": 0.7071,
      "step": 7858
    },
    {
      "epoch": 0.5766116089033998,
      "grad_norm": 0.5782954070599758,
      "learning_rate": 4.008529842629015e-06,
      "loss": 0.4955,
      "step": 7859
    },
    {
      "epoch": 0.5766849784935388,
      "grad_norm": 1.801059334827243,
      "learning_rate": 4.007365268241747e-06,
      "loss": 0.8308,
      "step": 7860
    },
    {
      "epoch": 0.576758348083678,
      "grad_norm": 1.5661631966565701,
      "learning_rate": 4.00620074991102e-06,
      "loss": 0.7755,
      "step": 7861
    },
    {
      "epoch": 0.5768317176738171,
      "grad_norm": 0.5654445158654549,
      "learning_rate": 4.005036287702599e-06,
      "loss": 0.526,
      "step": 7862
    },
    {
      "epoch": 0.5769050872639563,
      "grad_norm": 1.5602723081205818,
      "learning_rate": 4.003871881682243e-06,
      "loss": 0.7234,
      "step": 7863
    },
    {
      "epoch": 0.5769784568540954,
      "grad_norm": 1.6985807749308668,
      "learning_rate": 4.002707531915707e-06,
      "loss": 0.8046,
      "step": 7864
    },
    {
      "epoch": 0.5769784568540954,
      "lm_loss": 1.12,
      "step": 7864,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.5769784568540954,
      "lm_loss": 0.9364,
      "step": 7864,
      "vm_loss": 0.1321
    },
    {
      "epoch": 0.5769784568540954,
      "lm_loss": 0.3957,
      "step": 7864,
      "vm_loss": 0.1524
    },
    {
      "epoch": 0.5769784568540954,
      "lm_loss": 0.5082,
      "step": 7864,
      "vm_loss": 0.1017
    },
    {
      "epoch": 0.5769784568540954,
      "lm_loss": 0.8914,
      "step": 7864,
      "vm_loss": 0.1961
    },
    {
      "epoch": 0.5769784568540954,
      "lm_loss": 0.6172,
      "step": 7864,
      "vm_loss": 0.0641
    },
    {
      "epoch": 0.5769784568540954,
      "lm_loss": 0.7666,
      "step": 7864,
      "vm_loss": 0.1669
    },
    {
      "epoch": 0.5769784568540954,
      "lm_loss": 1.0591,
      "step": 7864,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.5770518264442345,
      "grad_norm": 2.224818164290153,
      "learning_rate": 4.001543238468748e-06,
      "loss": 0.7471,
      "step": 7865
    },
    {
      "epoch": 0.5771251960343736,
      "grad_norm": 0.5680603882902895,
      "learning_rate": 4.000379001407115e-06,
      "loss": 0.4916,
      "step": 7866
    },
    {
      "epoch": 0.5771985656245128,
      "grad_norm": 1.7897456947011368,
      "learning_rate": 3.999214820796554e-06,
      "loss": 0.7541,
      "step": 7867
    },
    {
      "epoch": 0.5772719352146519,
      "grad_norm": 0.5903056994493302,
      "learning_rate": 3.998050696702811e-06,
      "loss": 0.5033,
      "step": 7868
    },
    {
      "epoch": 0.5773453048047911,
      "grad_norm": 1.60232511824724,
      "learning_rate": 3.9968866291916254e-06,
      "loss": 0.7539,
      "step": 7869
    },
    {
      "epoch": 0.5774186743949301,
      "grad_norm": 0.54691061831592,
      "learning_rate": 3.995722618328739e-06,
      "loss": 0.492,
      "step": 7870
    },
    {
      "epoch": 0.5774920439850693,
      "grad_norm": 0.5518178147353303,
      "learning_rate": 3.994558664179881e-06,
      "loss": 0.4837,
      "step": 7871
    },
    {
      "epoch": 0.5775654135752084,
      "grad_norm": 1.646975720014975,
      "learning_rate": 3.9933947668107855e-06,
      "loss": 0.758,
      "step": 7872
    },
    {
      "epoch": 0.5775654135752084,
      "lm_loss": 1.202,
      "step": 7872,
      "vm_loss": 0.0824
    },
    {
      "epoch": 0.5775654135752084,
      "lm_loss": 0.3126,
      "step": 7872,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.5775654135752084,
      "lm_loss": 0.4606,
      "step": 7872,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.5775654135752084,
      "lm_loss": 1.7356,
      "step": 7872,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.5775654135752084,
      "lm_loss": 1.2225,
      "step": 7872,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.5775654135752084,
      "lm_loss": 0.5708,
      "step": 7872,
      "vm_loss": 0.2054
    },
    {
      "epoch": 0.5775654135752084,
      "lm_loss": 0.8826,
      "step": 7872,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.5775654135752084,
      "lm_loss": 1.0877,
      "step": 7872,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.5776387831653476,
      "grad_norm": 1.827438871908555,
      "learning_rate": 3.992230926287181e-06,
      "loss": 0.8332,
      "step": 7873
    },
    {
      "epoch": 0.5777121527554867,
      "grad_norm": 1.4337021320413432,
      "learning_rate": 3.9910671426747916e-06,
      "loss": 0.7686,
      "step": 7874
    },
    {
      "epoch": 0.5777855223456257,
      "grad_norm": 0.5589738399024372,
      "learning_rate": 3.989903416039338e-06,
      "loss": 0.4939,
      "step": 7875
    },
    {
      "epoch": 0.5778588919357649,
      "grad_norm": 1.4902051327869181,
      "learning_rate": 3.988739746446542e-06,
      "loss": 0.7371,
      "step": 7876
    },
    {
      "epoch": 0.577932261525904,
      "grad_norm": 1.6825625272229472,
      "learning_rate": 3.987576133962115e-06,
      "loss": 0.6626,
      "step": 7877
    },
    {
      "epoch": 0.5780056311160432,
      "grad_norm": 0.5628554626850664,
      "learning_rate": 3.986412578651772e-06,
      "loss": 0.4777,
      "step": 7878
    },
    {
      "epoch": 0.5780790007061823,
      "grad_norm": 1.7795099390715716,
      "learning_rate": 3.98524908058122e-06,
      "loss": 0.6918,
      "step": 7879
    },
    {
      "epoch": 0.5781523702963214,
      "grad_norm": 1.679392477939475,
      "learning_rate": 3.984085639816166e-06,
      "loss": 0.7581,
      "step": 7880
    },
    {
      "epoch": 0.5781523702963214,
      "lm_loss": 1.8045,
      "step": 7880,
      "vm_loss": 0.1487
    },
    {
      "epoch": 0.5781523702963214,
      "lm_loss": 1.3635,
      "step": 7880,
      "vm_loss": 0.1495
    },
    {
      "epoch": 0.5781523702963214,
      "lm_loss": 0.869,
      "step": 7880,
      "vm_loss": 0.0964
    },
    {
      "epoch": 0.5781523702963214,
      "lm_loss": 0.3832,
      "step": 7880,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.5781523702963214,
      "lm_loss": 0.677,
      "step": 7880,
      "vm_loss": 0.1121
    },
    {
      "epoch": 0.5781523702963214,
      "lm_loss": 0.9408,
      "step": 7880,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.5781523702963214,
      "lm_loss": 0.6343,
      "step": 7880,
      "vm_loss": 0.0571
    },
    {
      "epoch": 0.5781523702963214,
      "lm_loss": 0.1908,
      "step": 7880,
      "vm_loss": 0.1019
    },
    {
      "epoch": 0.5782257398864605,
      "grad_norm": 1.4737222165693804,
      "learning_rate": 3.982922256422313e-06,
      "loss": 0.8111,
      "step": 7881
    },
    {
      "epoch": 0.5782991094765997,
      "grad_norm": 0.5467296263816961,
      "learning_rate": 3.981758930465357e-06,
      "loss": 0.4824,
      "step": 7882
    },
    {
      "epoch": 0.5783724790667388,
      "grad_norm": 1.6850015133934448,
      "learning_rate": 3.9805956620109955e-06,
      "loss": 0.7383,
      "step": 7883
    },
    {
      "epoch": 0.578445848656878,
      "grad_norm": 0.5674347509760629,
      "learning_rate": 3.9794324511249234e-06,
      "loss": 0.4902,
      "step": 7884
    },
    {
      "epoch": 0.578519218247017,
      "grad_norm": 1.5995944942513929,
      "learning_rate": 3.978269297872828e-06,
      "loss": 0.7441,
      "step": 7885
    },
    {
      "epoch": 0.5785925878371562,
      "grad_norm": 1.5365315162764193,
      "learning_rate": 3.9771062023203934e-06,
      "loss": 0.8628,
      "step": 7886
    },
    {
      "epoch": 0.5786659574272953,
      "grad_norm": 0.5567057187524429,
      "learning_rate": 3.975943164533307e-06,
      "loss": 0.4775,
      "step": 7887
    },
    {
      "epoch": 0.5787393270174345,
      "grad_norm": 1.4880933392432316,
      "learning_rate": 3.974780184577245e-06,
      "loss": 0.8707,
      "step": 7888
    },
    {
      "epoch": 0.5787393270174345,
      "lm_loss": 1.3265,
      "step": 7888,
      "vm_loss": 0.1295
    },
    {
      "epoch": 0.5787393270174345,
      "lm_loss": 0.0916,
      "step": 7888,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.5787393270174345,
      "lm_loss": 0.9026,
      "step": 7888,
      "vm_loss": 0.1294
    },
    {
      "epoch": 0.5787393270174345,
      "lm_loss": 0.1897,
      "step": 7888,
      "vm_loss": 0.0856
    },
    {
      "epoch": 0.5787393270174345,
      "lm_loss": 0.1564,
      "step": 7888,
      "vm_loss": 0.1152
    },
    {
      "epoch": 0.5787393270174345,
      "lm_loss": 0.1285,
      "step": 7888,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.5787393270174345,
      "lm_loss": 0.5569,
      "step": 7888,
      "vm_loss": 0.1617
    },
    {
      "epoch": 0.5787393270174345,
      "lm_loss": 0.171,
      "step": 7888,
      "vm_loss": 0.1051
    },
    {
      "epoch": 0.5788126966075736,
      "grad_norm": 1.530134445409049,
      "learning_rate": 3.973617262517886e-06,
      "loss": 0.7906,
      "step": 7889
    },
    {
      "epoch": 0.5788860661977127,
      "grad_norm": 1.6943739832724085,
      "learning_rate": 3.9724543984209005e-06,
      "loss": 0.7465,
      "step": 7890
    },
    {
      "epoch": 0.5789594357878518,
      "grad_norm": 1.5016769377498176,
      "learning_rate": 3.97129159235196e-06,
      "loss": 0.8002,
      "step": 7891
    },
    {
      "epoch": 0.579032805377991,
      "grad_norm": 1.5301129798242348,
      "learning_rate": 3.970128844376732e-06,
      "loss": 0.657,
      "step": 7892
    },
    {
      "epoch": 0.5791061749681301,
      "grad_norm": 1.4043593702445307,
      "learning_rate": 3.968966154560878e-06,
      "loss": 0.79,
      "step": 7893
    },
    {
      "epoch": 0.5791795445582693,
      "grad_norm": 1.407233387086211,
      "learning_rate": 3.967803522970058e-06,
      "loss": 0.8262,
      "step": 7894
    },
    {
      "epoch": 0.5792529141484083,
      "grad_norm": 0.5667174370658674,
      "learning_rate": 3.96664094966993e-06,
      "loss": 0.4918,
      "step": 7895
    },
    {
      "epoch": 0.5793262837385474,
      "grad_norm": 1.4837869037224152,
      "learning_rate": 3.965478434726146e-06,
      "loss": 0.8653,
      "step": 7896
    },
    {
      "epoch": 0.5793262837385474,
      "lm_loss": 1.1985,
      "step": 7896,
      "vm_loss": 0.0657
    },
    {
      "epoch": 0.5793262837385474,
      "lm_loss": 0.0353,
      "step": 7896,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.5793262837385474,
      "lm_loss": 0.2412,
      "step": 7896,
      "vm_loss": 0.162
    },
    {
      "epoch": 0.5793262837385474,
      "lm_loss": 0.4891,
      "step": 7896,
      "vm_loss": 0.2207
    },
    {
      "epoch": 0.5793262837385474,
      "lm_loss": 0.1709,
      "step": 7896,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.5793262837385474,
      "lm_loss": 0.2192,
      "step": 7896,
      "vm_loss": 0.1182
    },
    {
      "epoch": 0.5793262837385474,
      "lm_loss": 0.783,
      "step": 7896,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.5793262837385474,
      "lm_loss": 0.7784,
      "step": 7896,
      "vm_loss": 0.1888
    },
    {
      "epoch": 0.5793996533286866,
      "grad_norm": 1.3824526202608154,
      "learning_rate": 3.964315978204357e-06,
      "loss": 0.7266,
      "step": 7897
    },
    {
      "epoch": 0.5794730229188257,
      "grad_norm": 1.4927165277079903,
      "learning_rate": 3.963153580170211e-06,
      "loss": 0.7315,
      "step": 7898
    },
    {
      "epoch": 0.5795463925089649,
      "grad_norm": 1.554732776971071,
      "learning_rate": 3.9619912406893475e-06,
      "loss": 0.7772,
      "step": 7899
    },
    {
      "epoch": 0.5796197620991039,
      "grad_norm": 1.658164726207744,
      "learning_rate": 3.9608289598274105e-06,
      "loss": 0.7865,
      "step": 7900
    },
    {
      "epoch": 0.5796931316892431,
      "grad_norm": 0.560419243985176,
      "learning_rate": 3.959666737650036e-06,
      "loss": 0.4979,
      "step": 7901
    },
    {
      "epoch": 0.5797665012793822,
      "grad_norm": 1.5101824231496404,
      "learning_rate": 3.958504574222856e-06,
      "loss": 0.7133,
      "step": 7902
    },
    {
      "epoch": 0.5798398708695214,
      "grad_norm": 1.4549483331994573,
      "learning_rate": 3.957342469611503e-06,
      "loss": 0.7921,
      "step": 7903
    },
    {
      "epoch": 0.5799132404596605,
      "grad_norm": 0.5659344862303188,
      "learning_rate": 3.956180423881602e-06,
      "loss": 0.5075,
      "step": 7904
    },
    {
      "epoch": 0.5799132404596605,
      "lm_loss": 1.5165,
      "step": 7904,
      "vm_loss": 0.0673
    },
    {
      "epoch": 0.5799132404596605,
      "lm_loss": 0.9027,
      "step": 7904,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.5799132404596605,
      "lm_loss": 0.7084,
      "step": 7904,
      "vm_loss": 0.1
    },
    {
      "epoch": 0.5799132404596605,
      "lm_loss": 0.2318,
      "step": 7904,
      "vm_loss": 0.0631
    },
    {
      "epoch": 0.5799132404596605,
      "lm_loss": 0.8991,
      "step": 7904,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.5799132404596605,
      "lm_loss": 0.2847,
      "step": 7904,
      "vm_loss": 0.1533
    },
    {
      "epoch": 0.5799132404596605,
      "lm_loss": 1.2477,
      "step": 7904,
      "vm_loss": 0.13
    },
    {
      "epoch": 0.5799132404596605,
      "lm_loss": 0.8061,
      "step": 7904,
      "vm_loss": 0.1971
    },
    {
      "epoch": 0.5799866100497996,
      "grad_norm": 1.6752122710596462,
      "learning_rate": 3.9550184370987785e-06,
      "loss": 0.8249,
      "step": 7905
    },
    {
      "epoch": 0.5800599796399387,
      "grad_norm": 1.6250043652442518,
      "learning_rate": 3.9538565093286505e-06,
      "loss": 0.8179,
      "step": 7906
    },
    {
      "epoch": 0.5801333492300779,
      "grad_norm": 1.5428525015753016,
      "learning_rate": 3.952694640636837e-06,
      "loss": 0.7456,
      "step": 7907
    },
    {
      "epoch": 0.580206718820217,
      "grad_norm": 1.3654139180820557,
      "learning_rate": 3.951532831088951e-06,
      "loss": 0.7355,
      "step": 7908
    },
    {
      "epoch": 0.5802800884103562,
      "grad_norm": 1.4817853646103296,
      "learning_rate": 3.950371080750602e-06,
      "loss": 0.7556,
      "step": 7909
    },
    {
      "epoch": 0.5803534580004952,
      "grad_norm": 1.526342570036064,
      "learning_rate": 3.949209389687396e-06,
      "loss": 0.739,
      "step": 7910
    },
    {
      "epoch": 0.5804268275906344,
      "grad_norm": 0.569910116103759,
      "learning_rate": 3.94804775796494e-06,
      "loss": 0.4951,
      "step": 7911
    },
    {
      "epoch": 0.5805001971807735,
      "grad_norm": 1.482920511353849,
      "learning_rate": 3.946886185648833e-06,
      "loss": 0.8337,
      "step": 7912
    },
    {
      "epoch": 0.5805001971807735,
      "lm_loss": 0.8804,
      "step": 7912,
      "vm_loss": 0.0491
    },
    {
      "epoch": 0.5805001971807735,
      "lm_loss": 0.1766,
      "step": 7912,
      "vm_loss": 0.0434
    },
    {
      "epoch": 0.5805001971807735,
      "lm_loss": 0.6524,
      "step": 7912,
      "vm_loss": 0.2059
    },
    {
      "epoch": 0.5805001971807735,
      "lm_loss": 0.3051,
      "step": 7912,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.5805001971807735,
      "lm_loss": 0.1783,
      "step": 7912,
      "vm_loss": 0.1653
    },
    {
      "epoch": 0.5805001971807735,
      "lm_loss": 0.4185,
      "step": 7912,
      "vm_loss": 0.0471
    },
    {
      "epoch": 0.5805001971807735,
      "lm_loss": 0.6704,
      "step": 7912,
      "vm_loss": 0.1529
    },
    {
      "epoch": 0.5805001971807735,
      "lm_loss": 1.1091,
      "step": 7912,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.5805735667709127,
      "grad_norm": 1.5908140184825976,
      "learning_rate": 3.9457246728046685e-06,
      "loss": 0.6821,
      "step": 7913
    },
    {
      "epoch": 0.5806469363610518,
      "grad_norm": 1.5694180874799029,
      "learning_rate": 3.944563219498044e-06,
      "loss": 0.8219,
      "step": 7914
    },
    {
      "epoch": 0.5807203059511908,
      "grad_norm": 1.5679014236982136,
      "learning_rate": 3.943401825794548e-06,
      "loss": 0.6834,
      "step": 7915
    },
    {
      "epoch": 0.58079367554133,
      "grad_norm": 1.5134644975944442,
      "learning_rate": 3.942240491759768e-06,
      "loss": 0.7909,
      "step": 7916
    },
    {
      "epoch": 0.5808670451314691,
      "grad_norm": 1.6474173196297968,
      "learning_rate": 3.941079217459286e-06,
      "loss": 0.8461,
      "step": 7917
    },
    {
      "epoch": 0.5809404147216083,
      "grad_norm": 1.444874760055077,
      "learning_rate": 3.939918002958684e-06,
      "loss": 0.8096,
      "step": 7918
    },
    {
      "epoch": 0.5810137843117474,
      "grad_norm": 0.5752414328086778,
      "learning_rate": 3.9387568483235366e-06,
      "loss": 0.4959,
      "step": 7919
    },
    {
      "epoch": 0.5810871539018865,
      "grad_norm": 1.501749628956168,
      "learning_rate": 3.9375957536194195e-06,
      "loss": 0.8043,
      "step": 7920
    },
    {
      "epoch": 0.5810871539018865,
      "lm_loss": 0.4785,
      "step": 7920,
      "vm_loss": 0.0561
    },
    {
      "epoch": 0.5810871539018865,
      "lm_loss": 0.4047,
      "step": 7920,
      "vm_loss": 0.1588
    },
    {
      "epoch": 0.5810871539018865,
      "lm_loss": 0.392,
      "step": 7920,
      "vm_loss": 0.1332
    },
    {
      "epoch": 0.5810871539018865,
      "lm_loss": 0.8515,
      "step": 7920,
      "vm_loss": 0.0837
    },
    {
      "epoch": 0.5810871539018865,
      "lm_loss": 0.5983,
      "step": 7920,
      "vm_loss": 0.0947
    },
    {
      "epoch": 0.5810871539018865,
      "lm_loss": 0.4396,
      "step": 7920,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.5810871539018865,
      "lm_loss": 0.1637,
      "step": 7920,
      "vm_loss": 0.0919
    },
    {
      "epoch": 0.5810871539018865,
      "lm_loss": 0.8749,
      "step": 7920,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.5811605234920256,
      "grad_norm": 1.6005177933298398,
      "learning_rate": 3.936434718911899e-06,
      "loss": 0.6885,
      "step": 7921
    },
    {
      "epoch": 0.5812338930821648,
      "grad_norm": 1.593608037212195,
      "learning_rate": 3.935273744266546e-06,
      "loss": 0.7696,
      "step": 7922
    },
    {
      "epoch": 0.5813072626723039,
      "grad_norm": 0.5636728414806453,
      "learning_rate": 3.9341128297489194e-06,
      "loss": 0.4893,
      "step": 7923
    },
    {
      "epoch": 0.5813806322624431,
      "grad_norm": 0.5530741179244671,
      "learning_rate": 3.932951975424582e-06,
      "loss": 0.5047,
      "step": 7924
    },
    {
      "epoch": 0.5814540018525821,
      "grad_norm": 1.5635763755866527,
      "learning_rate": 3.931791181359091e-06,
      "loss": 0.7778,
      "step": 7925
    },
    {
      "epoch": 0.5815273714427213,
      "grad_norm": 1.4419137717704624,
      "learning_rate": 3.930630447617994e-06,
      "loss": 0.7796,
      "step": 7926
    },
    {
      "epoch": 0.5816007410328604,
      "grad_norm": 0.5314352575880646,
      "learning_rate": 3.929469774266845e-06,
      "loss": 0.4802,
      "step": 7927
    },
    {
      "epoch": 0.5816741106229996,
      "grad_norm": 1.6042647353094885,
      "learning_rate": 3.928309161371191e-06,
      "loss": 0.7617,
      "step": 7928
    },
    {
      "epoch": 0.5816741106229996,
      "lm_loss": 0.6879,
      "step": 7928,
      "vm_loss": 0.0621
    },
    {
      "epoch": 0.5816741106229996,
      "lm_loss": 0.3568,
      "step": 7928,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.5816741106229996,
      "lm_loss": 0.8636,
      "step": 7928,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.5816741106229996,
      "lm_loss": 0.245,
      "step": 7928,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.5816741106229996,
      "lm_loss": 0.3188,
      "step": 7928,
      "vm_loss": 0.0917
    },
    {
      "epoch": 0.5816741106229996,
      "lm_loss": 0.1301,
      "step": 7928,
      "vm_loss": 0.0463
    },
    {
      "epoch": 0.5816741106229996,
      "lm_loss": 0.3942,
      "step": 7928,
      "vm_loss": 0.1399
    },
    {
      "epoch": 0.5816741106229996,
      "lm_loss": 0.7985,
      "step": 7928,
      "vm_loss": 0.2348
    },
    {
      "epoch": 0.5817474802131387,
      "grad_norm": 1.3894472944138667,
      "learning_rate": 3.927148608996569e-06,
      "loss": 0.6873,
      "step": 7929
    },
    {
      "epoch": 0.5818208498032778,
      "grad_norm": 1.4978793683589526,
      "learning_rate": 3.9259881172085244e-06,
      "loss": 0.6831,
      "step": 7930
    },
    {
      "epoch": 0.5818942193934169,
      "grad_norm": 1.3212466136906529,
      "learning_rate": 3.924827686072591e-06,
      "loss": 0.7184,
      "step": 7931
    },
    {
      "epoch": 0.581967588983556,
      "grad_norm": 0.5591917418766954,
      "learning_rate": 3.9236673156542994e-06,
      "loss": 0.4943,
      "step": 7932
    },
    {
      "epoch": 0.5820409585736952,
      "grad_norm": 1.4540175963983957,
      "learning_rate": 3.9225070060191806e-06,
      "loss": 0.6659,
      "step": 7933
    },
    {
      "epoch": 0.5821143281638343,
      "grad_norm": 1.429661940218592,
      "learning_rate": 3.921346757232759e-06,
      "loss": 0.69,
      "step": 7934
    },
    {
      "epoch": 0.5821876977539734,
      "grad_norm": 1.7427949621477645,
      "learning_rate": 3.920186569360557e-06,
      "loss": 0.7937,
      "step": 7935
    },
    {
      "epoch": 0.5822610673441125,
      "grad_norm": 0.5529791442549192,
      "learning_rate": 3.919026442468095e-06,
      "loss": 0.4725,
      "step": 7936
    },
    {
      "epoch": 0.5822610673441125,
      "lm_loss": 1.5358,
      "step": 7936,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.5822610673441125,
      "lm_loss": 0.336,
      "step": 7936,
      "vm_loss": 0.071
    },
    {
      "epoch": 0.5822610673441125,
      "lm_loss": 0.1518,
      "step": 7936,
      "vm_loss": 0.1109
    },
    {
      "epoch": 0.5822610673441125,
      "lm_loss": 0.0728,
      "step": 7936,
      "vm_loss": 0.1234
    },
    {
      "epoch": 0.5822610673441125,
      "lm_loss": 0.7926,
      "step": 7936,
      "vm_loss": 0.1814
    },
    {
      "epoch": 0.5822610673441125,
      "lm_loss": 0.7547,
      "step": 7936,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.5822610673441125,
      "lm_loss": 0.0765,
      "step": 7936,
      "vm_loss": 0.1521
    },
    {
      "epoch": 0.5822610673441125,
      "lm_loss": 0.2894,
      "step": 7936,
      "vm_loss": 0.0463
    },
    {
      "epoch": 0.5823344369342517,
      "grad_norm": 1.2947231509300496,
      "learning_rate": 3.9178663766208855e-06,
      "loss": 0.6847,
      "step": 7937
    },
    {
      "epoch": 0.5824078065243908,
      "grad_norm": 1.4317474954064882,
      "learning_rate": 3.916706371884443e-06,
      "loss": 0.737,
      "step": 7938
    },
    {
      "epoch": 0.5824811761145299,
      "grad_norm": 1.446699855514329,
      "learning_rate": 3.9155464283242725e-06,
      "loss": 0.7161,
      "step": 7939
    },
    {
      "epoch": 0.582554545704669,
      "grad_norm": 1.3525042518021488,
      "learning_rate": 3.914386546005881e-06,
      "loss": 0.7135,
      "step": 7940
    },
    {
      "epoch": 0.5826279152948082,
      "grad_norm": 1.5824861698850736,
      "learning_rate": 3.913226724994771e-06,
      "loss": 0.7924,
      "step": 7941
    },
    {
      "epoch": 0.5827012848849473,
      "grad_norm": 1.685159220488534,
      "learning_rate": 3.912066965356438e-06,
      "loss": 0.7417,
      "step": 7942
    },
    {
      "epoch": 0.5827746544750865,
      "grad_norm": 1.6096189545745492,
      "learning_rate": 3.910907267156379e-06,
      "loss": 0.8016,
      "step": 7943
    },
    {
      "epoch": 0.5828480240652255,
      "grad_norm": 0.557498543689259,
      "learning_rate": 3.909747630460083e-06,
      "loss": 0.4885,
      "step": 7944
    },
    {
      "epoch": 0.5828480240652255,
      "lm_loss": 0.3308,
      "step": 7944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5828480240652255,
      "lm_loss": 0.602,
      "step": 7944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5828480240652255,
      "lm_loss": 0.5189,
      "step": 7944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5828480240652255,
      "lm_loss": 0.5092,
      "step": 7944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5828480240652255,
      "lm_loss": 0.3079,
      "step": 7944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5828480240652255,
      "lm_loss": 0.3285,
      "step": 7944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5828480240652255,
      "lm_loss": 0.5053,
      "step": 7944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5828480240652255,
      "lm_loss": 0.5895,
      "step": 7944,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5829213936553647,
      "grad_norm": 0.5686775983425698,
      "learning_rate": 3.908588055333038e-06,
      "loss": 0.4913,
      "step": 7945
    },
    {
      "epoch": 0.5829947632455038,
      "grad_norm": 1.645732895730832,
      "learning_rate": 3.907428541840729e-06,
      "loss": 0.826,
      "step": 7946
    },
    {
      "epoch": 0.583068132835643,
      "grad_norm": 1.3712383372893773,
      "learning_rate": 3.9062690900486354e-06,
      "loss": 0.7921,
      "step": 7947
    },
    {
      "epoch": 0.5831415024257821,
      "grad_norm": 0.5586916563918713,
      "learning_rate": 3.905109700022235e-06,
      "loss": 0.4994,
      "step": 7948
    },
    {
      "epoch": 0.5832148720159211,
      "grad_norm": 0.5575889678028041,
      "learning_rate": 3.903950371827001e-06,
      "loss": 0.5005,
      "step": 7949
    },
    {
      "epoch": 0.5832882416060603,
      "grad_norm": 1.3873080145102326,
      "learning_rate": 3.902791105528404e-06,
      "loss": 0.7167,
      "step": 7950
    },
    {
      "epoch": 0.5833616111961994,
      "grad_norm": 1.457319745415717,
      "learning_rate": 3.901631901191911e-06,
      "loss": 0.6947,
      "step": 7951
    },
    {
      "epoch": 0.5834349807863386,
      "grad_norm": 1.6103273256424353,
      "learning_rate": 3.900472758882984e-06,
      "loss": 0.8752,
      "step": 7952
    },
    {
      "epoch": 0.5834349807863386,
      "lm_loss": 1.6267,
      "step": 7952,
      "vm_loss": 0.0621
    },
    {
      "epoch": 0.5834349807863386,
      "lm_loss": 1.595,
      "step": 7952,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.5834349807863386,
      "lm_loss": 0.0894,
      "step": 7952,
      "vm_loss": 0.1364
    },
    {
      "epoch": 0.5834349807863386,
      "lm_loss": 0.1998,
      "step": 7952,
      "vm_loss": 0.0895
    },
    {
      "epoch": 0.5834349807863386,
      "lm_loss": 0.2454,
      "step": 7952,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.5834349807863386,
      "lm_loss": 0.498,
      "step": 7952,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.5834349807863386,
      "lm_loss": 0.604,
      "step": 7952,
      "vm_loss": 0.0685
    },
    {
      "epoch": 0.5834349807863386,
      "lm_loss": 0.6382,
      "step": 7952,
      "vm_loss": 0.1094
    },
    {
      "epoch": 0.5835083503764777,
      "grad_norm": 1.62897694504031,
      "learning_rate": 3.899313678667083e-06,
      "loss": 0.7881,
      "step": 7953
    },
    {
      "epoch": 0.5835817199666168,
      "grad_norm": 1.339703335284471,
      "learning_rate": 3.898154660609665e-06,
      "loss": 0.768,
      "step": 7954
    },
    {
      "epoch": 0.5836550895567559,
      "grad_norm": 1.3796207857772944,
      "learning_rate": 3.896995704776183e-06,
      "loss": 0.6387,
      "step": 7955
    },
    {
      "epoch": 0.5837284591468951,
      "grad_norm": 1.6709647601453006,
      "learning_rate": 3.8958368112320835e-06,
      "loss": 0.7991,
      "step": 7956
    },
    {
      "epoch": 0.5838018287370342,
      "grad_norm": 1.5061977808298437,
      "learning_rate": 3.8946779800428145e-06,
      "loss": 0.7588,
      "step": 7957
    },
    {
      "epoch": 0.5838751983271734,
      "grad_norm": 0.6037373288854399,
      "learning_rate": 3.8935192112738165e-06,
      "loss": 0.4966,
      "step": 7958
    },
    {
      "epoch": 0.5839485679173124,
      "grad_norm": 1.5142440157822685,
      "learning_rate": 3.892360504990529e-06,
      "loss": 0.7579,
      "step": 7959
    },
    {
      "epoch": 0.5840219375074516,
      "grad_norm": 1.3512394528016867,
      "learning_rate": 3.8912018612583865e-06,
      "loss": 0.706,
      "step": 7960
    },
    {
      "epoch": 0.5840219375074516,
      "lm_loss": 1.3879,
      "step": 7960,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.5840219375074516,
      "lm_loss": 0.426,
      "step": 7960,
      "vm_loss": 0.0364
    },
    {
      "epoch": 0.5840219375074516,
      "lm_loss": 0.2326,
      "step": 7960,
      "vm_loss": 0.171
    },
    {
      "epoch": 0.5840219375074516,
      "lm_loss": 0.0703,
      "step": 7960,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.5840219375074516,
      "lm_loss": 1.004,
      "step": 7960,
      "vm_loss": 0.1648
    },
    {
      "epoch": 0.5840219375074516,
      "lm_loss": 0.1414,
      "step": 7960,
      "vm_loss": 0.0819
    },
    {
      "epoch": 0.5840219375074516,
      "lm_loss": 0.1765,
      "step": 7960,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.5840219375074516,
      "lm_loss": 0.6726,
      "step": 7960,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.5840953070975907,
      "grad_norm": 1.4978510646314476,
      "learning_rate": 3.890043280142821e-06,
      "loss": 0.6755,
      "step": 7961
    },
    {
      "epoch": 0.5841686766877299,
      "grad_norm": 1.4735940138999772,
      "learning_rate": 3.888884761709261e-06,
      "loss": 0.8033,
      "step": 7962
    },
    {
      "epoch": 0.584242046277869,
      "grad_norm": 1.445201885249143,
      "learning_rate": 3.88772630602313e-06,
      "loss": 0.7488,
      "step": 7963
    },
    {
      "epoch": 0.5843154158680081,
      "grad_norm": 0.5728653371959985,
      "learning_rate": 3.886567913149848e-06,
      "loss": 0.4913,
      "step": 7964
    },
    {
      "epoch": 0.5843887854581472,
      "grad_norm": 0.5647664316439112,
      "learning_rate": 3.885409583154834e-06,
      "loss": 0.4781,
      "step": 7965
    },
    {
      "epoch": 0.5844621550482864,
      "grad_norm": 0.5491764959176412,
      "learning_rate": 3.884251316103502e-06,
      "loss": 0.4922,
      "step": 7966
    },
    {
      "epoch": 0.5845355246384255,
      "grad_norm": 1.6951222512206447,
      "learning_rate": 3.883093112061259e-06,
      "loss": 0.72,
      "step": 7967
    },
    {
      "epoch": 0.5846088942285647,
      "grad_norm": 1.428141154819285,
      "learning_rate": 3.881934971093516e-06,
      "loss": 0.8406,
      "step": 7968
    },
    {
      "epoch": 0.5846088942285647,
      "lm_loss": 1.5286,
      "step": 7968,
      "vm_loss": 0.1383
    },
    {
      "epoch": 0.5846088942285647,
      "lm_loss": 0.8701,
      "step": 7968,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.5846088942285647,
      "lm_loss": 0.8072,
      "step": 7968,
      "vm_loss": 0.208
    },
    {
      "epoch": 0.5846088942285647,
      "lm_loss": 0.7506,
      "step": 7968,
      "vm_loss": 0.1728
    },
    {
      "epoch": 0.5846088942285647,
      "lm_loss": 0.0181,
      "step": 7968,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.5846088942285647,
      "lm_loss": 0.3768,
      "step": 7968,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.5846088942285647,
      "lm_loss": 1.0501,
      "step": 7968,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.5846088942285647,
      "lm_loss": 0.7719,
      "step": 7968,
      "vm_loss": 0.1328
    },
    {
      "epoch": 0.5846822638187037,
      "grad_norm": 1.5428446615823836,
      "learning_rate": 3.880776893265673e-06,
      "loss": 0.7712,
      "step": 7969
    },
    {
      "epoch": 0.5847556334088428,
      "grad_norm": 1.5416296196603345,
      "learning_rate": 3.87961887864313e-06,
      "loss": 0.6954,
      "step": 7970
    },
    {
      "epoch": 0.584829002998982,
      "grad_norm": 1.5479553909204355,
      "learning_rate": 3.878460927291286e-06,
      "loss": 0.8371,
      "step": 7971
    },
    {
      "epoch": 0.5849023725891211,
      "grad_norm": 1.5438799242950276,
      "learning_rate": 3.877303039275528e-06,
      "loss": 0.8424,
      "step": 7972
    },
    {
      "epoch": 0.5849757421792603,
      "grad_norm": 1.5844692496967332,
      "learning_rate": 3.876145214661249e-06,
      "loss": 0.7498,
      "step": 7973
    },
    {
      "epoch": 0.5850491117693993,
      "grad_norm": 0.5616924257914101,
      "learning_rate": 3.874987453513833e-06,
      "loss": 0.4984,
      "step": 7974
    },
    {
      "epoch": 0.5851224813595385,
      "grad_norm": 0.5719505273987289,
      "learning_rate": 3.87382975589866e-06,
      "loss": 0.5053,
      "step": 7975
    },
    {
      "epoch": 0.5851958509496776,
      "grad_norm": 1.4284885975248498,
      "learning_rate": 3.872672121881112e-06,
      "loss": 0.7533,
      "step": 7976
    },
    {
      "epoch": 0.5851958509496776,
      "lm_loss": 1.0606,
      "step": 7976,
      "vm_loss": 0.0704
    },
    {
      "epoch": 0.5851958509496776,
      "lm_loss": 0.8565,
      "step": 7976,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.5851958509496776,
      "lm_loss": 0.0852,
      "step": 7976,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.5851958509496776,
      "lm_loss": 0.7378,
      "step": 7976,
      "vm_loss": 0.142
    },
    {
      "epoch": 0.5851958509496776,
      "lm_loss": 1.1571,
      "step": 7976,
      "vm_loss": 0.0994
    },
    {
      "epoch": 0.5851958509496776,
      "lm_loss": 0.7543,
      "step": 7976,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.5851958509496776,
      "lm_loss": 0.0381,
      "step": 7976,
      "vm_loss": 0.088
    },
    {
      "epoch": 0.5851958509496776,
      "lm_loss": 0.0617,
      "step": 7976,
      "vm_loss": 0.0599
    },
    {
      "epoch": 0.5852692205398168,
      "grad_norm": 1.594191769734319,
      "learning_rate": 3.87151455152656e-06,
      "loss": 0.7861,
      "step": 7977
    },
    {
      "epoch": 0.5853425901299559,
      "grad_norm": 1.80240315899695,
      "learning_rate": 3.8703570449003755e-06,
      "loss": 0.7407,
      "step": 7978
    },
    {
      "epoch": 0.585415959720095,
      "grad_norm": 1.597447673585286,
      "learning_rate": 3.869199602067927e-06,
      "loss": 0.7392,
      "step": 7979
    },
    {
      "epoch": 0.5854893293102341,
      "grad_norm": 1.3236737240137082,
      "learning_rate": 3.868042223094576e-06,
      "loss": 0.6383,
      "step": 7980
    },
    {
      "epoch": 0.5855626989003733,
      "grad_norm": 1.47304345543237,
      "learning_rate": 3.866884908045686e-06,
      "loss": 0.8635,
      "step": 7981
    },
    {
      "epoch": 0.5856360684905124,
      "grad_norm": 1.6021968229323202,
      "learning_rate": 3.86572765698661e-06,
      "loss": 0.6839,
      "step": 7982
    },
    {
      "epoch": 0.5857094380806516,
      "grad_norm": 1.4744686889138667,
      "learning_rate": 3.864570469982702e-06,
      "loss": 0.7302,
      "step": 7983
    },
    {
      "epoch": 0.5857828076707906,
      "grad_norm": 1.6343215705364589,
      "learning_rate": 3.863413347099312e-06,
      "loss": 0.7475,
      "step": 7984
    },
    {
      "epoch": 0.5857828076707906,
      "lm_loss": 0.3063,
      "step": 7984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5857828076707906,
      "lm_loss": 0.5408,
      "step": 7984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5857828076707906,
      "lm_loss": 0.4034,
      "step": 7984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5857828076707906,
      "lm_loss": 0.5119,
      "step": 7984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5857828076707906,
      "lm_loss": 0.3711,
      "step": 7984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5857828076707906,
      "lm_loss": 0.4841,
      "step": 7984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5857828076707906,
      "lm_loss": 0.5431,
      "step": 7984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5857828076707906,
      "lm_loss": 0.5157,
      "step": 7984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5858561772609298,
      "grad_norm": 0.5588946781808531,
      "learning_rate": 3.862256288401785e-06,
      "loss": 0.4808,
      "step": 7985
    },
    {
      "epoch": 0.5859295468510689,
      "grad_norm": 1.454794218340278,
      "learning_rate": 3.861099293955462e-06,
      "loss": 0.742,
      "step": 7986
    },
    {
      "epoch": 0.586002916441208,
      "grad_norm": 1.449232324622789,
      "learning_rate": 3.859942363825684e-06,
      "loss": 0.8027,
      "step": 7987
    },
    {
      "epoch": 0.5860762860313472,
      "grad_norm": 0.5970242767536249,
      "learning_rate": 3.858785498077782e-06,
      "loss": 0.5043,
      "step": 7988
    },
    {
      "epoch": 0.5861496556214862,
      "grad_norm": 0.5684295948866263,
      "learning_rate": 3.85762869677709e-06,
      "loss": 0.4926,
      "step": 7989
    },
    {
      "epoch": 0.5862230252116254,
      "grad_norm": 0.5525396663815799,
      "learning_rate": 3.856471959988935e-06,
      "loss": 0.4874,
      "step": 7990
    },
    {
      "epoch": 0.5862963948017645,
      "grad_norm": 1.6740942708016,
      "learning_rate": 3.8553152877786395e-06,
      "loss": 0.7683,
      "step": 7991
    },
    {
      "epoch": 0.5863697643919037,
      "grad_norm": 1.5555477821563268,
      "learning_rate": 3.854158680211526e-06,
      "loss": 0.8413,
      "step": 7992
    },
    {
      "epoch": 0.5863697643919037,
      "lm_loss": 1.336,
      "step": 7992,
      "vm_loss": 0.0602
    },
    {
      "epoch": 0.5863697643919037,
      "lm_loss": 0.3535,
      "step": 7992,
      "vm_loss": 0.11
    },
    {
      "epoch": 0.5863697643919037,
      "lm_loss": 0.8716,
      "step": 7992,
      "vm_loss": 0.0743
    },
    {
      "epoch": 0.5863697643919037,
      "lm_loss": 0.6654,
      "step": 7992,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.5863697643919037,
      "lm_loss": 0.7224,
      "step": 7992,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.5863697643919037,
      "lm_loss": 0.362,
      "step": 7992,
      "vm_loss": 0.0689
    },
    {
      "epoch": 0.5863697643919037,
      "lm_loss": 0.4194,
      "step": 7992,
      "vm_loss": 0.0689
    },
    {
      "epoch": 0.5863697643919037,
      "lm_loss": 0.142,
      "step": 7992,
      "vm_loss": 0.1012
    },
    {
      "epoch": 0.5864431339820428,
      "grad_norm": 1.5912080138220068,
      "learning_rate": 3.8530021373529085e-06,
      "loss": 0.7473,
      "step": 7993
    },
    {
      "epoch": 0.5865165035721819,
      "grad_norm": 1.563601564310794,
      "learning_rate": 3.8518456592681e-06,
      "loss": 0.6936,
      "step": 7994
    },
    {
      "epoch": 0.586589873162321,
      "grad_norm": 1.6098800982102934,
      "learning_rate": 3.850689246022412e-06,
      "loss": 0.8187,
      "step": 7995
    },
    {
      "epoch": 0.5866632427524602,
      "grad_norm": 1.4892975968856694,
      "learning_rate": 3.8495328976811495e-06,
      "loss": 0.6718,
      "step": 7996
    },
    {
      "epoch": 0.5867366123425993,
      "grad_norm": 1.6927066896043437,
      "learning_rate": 3.848376614309611e-06,
      "loss": 0.7802,
      "step": 7997
    },
    {
      "epoch": 0.5868099819327385,
      "grad_norm": 1.5749542735308735,
      "learning_rate": 3.8472203959730975e-06,
      "loss": 0.6817,
      "step": 7998
    },
    {
      "epoch": 0.5868833515228775,
      "grad_norm": 1.4252663707540223,
      "learning_rate": 3.846064242736903e-06,
      "loss": 0.7065,
      "step": 7999
    },
    {
      "epoch": 0.5869567211130167,
      "grad_norm": 1.827368057544943,
      "learning_rate": 3.844908154666319e-06,
      "loss": 0.7928,
      "step": 8000
    },
    {
      "epoch": 0.5869567211130167,
      "lm_loss": 1.0691,
      "step": 8000,
      "vm_loss": 0.0881
    },
    {
      "epoch": 0.5869567211130167,
      "lm_loss": 0.2979,
      "step": 8000,
      "vm_loss": 0.1319
    },
    {
      "epoch": 0.5869567211130167,
      "lm_loss": 0.7714,
      "step": 8000,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.5869567211130167,
      "lm_loss": 0.2222,
      "step": 8000,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.5869567211130167,
      "lm_loss": 0.2506,
      "step": 8000,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.5869567211130167,
      "lm_loss": 0.2995,
      "step": 8000,
      "vm_loss": 0.0369
    },
    {
      "epoch": 0.5869567211130167,
      "lm_loss": 1.0033,
      "step": 8000,
      "vm_loss": 0.0676
    },
    {
      "epoch": 0.5869567211130167,
      "lm_loss": 1.3215,
      "step": 8000,
      "vm_loss": 0.149
    },
    {
      "epoch": 0.5870300907031558,
      "grad_norm": 1.5948555239495998,
      "learning_rate": 3.843752131826631e-06,
      "loss": 0.776,
      "step": 8001
    },
    {
      "epoch": 0.587103460293295,
      "grad_norm": 0.7009044731365994,
      "learning_rate": 3.842596174283124e-06,
      "loss": 0.5118,
      "step": 8002
    },
    {
      "epoch": 0.5871768298834341,
      "grad_norm": 1.7963039601672701,
      "learning_rate": 3.841440282101078e-06,
      "loss": 0.8432,
      "step": 8003
    },
    {
      "epoch": 0.5872501994735732,
      "grad_norm": 1.4714422901088953,
      "learning_rate": 3.840284455345768e-06,
      "loss": 0.7492,
      "step": 8004
    },
    {
      "epoch": 0.5873235690637123,
      "grad_norm": 1.5568166817192177,
      "learning_rate": 3.839128694082467e-06,
      "loss": 0.8148,
      "step": 8005
    },
    {
      "epoch": 0.5873969386538515,
      "grad_norm": 0.5797343774411277,
      "learning_rate": 3.837972998376442e-06,
      "loss": 0.481,
      "step": 8006
    },
    {
      "epoch": 0.5874703082439906,
      "grad_norm": 1.405802423074923,
      "learning_rate": 3.836817368292961e-06,
      "loss": 0.7516,
      "step": 8007
    },
    {
      "epoch": 0.5875436778341298,
      "grad_norm": 1.637048068079887,
      "learning_rate": 3.835661803897284e-06,
      "loss": 0.8579,
      "step": 8008
    },
    {
      "epoch": 0.5875436778341298,
      "lm_loss": 0.4248,
      "step": 8008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5875436778341298,
      "lm_loss": 0.5719,
      "step": 8008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5875436778341298,
      "lm_loss": 0.4204,
      "step": 8008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5875436778341298,
      "lm_loss": 0.5913,
      "step": 8008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5875436778341298,
      "lm_loss": 0.4282,
      "step": 8008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5875436778341298,
      "lm_loss": 0.384,
      "step": 8008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5875436778341298,
      "lm_loss": 0.521,
      "step": 8008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5875436778341298,
      "lm_loss": 0.578,
      "step": 8008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5876170474242688,
      "grad_norm": 0.5721214532523805,
      "learning_rate": 3.834506305254667e-06,
      "loss": 0.4782,
      "step": 8009
    },
    {
      "epoch": 0.5876904170144079,
      "grad_norm": 1.4870142728473104,
      "learning_rate": 3.833350872430366e-06,
      "loss": 0.7152,
      "step": 8010
    },
    {
      "epoch": 0.5877637866045471,
      "grad_norm": 1.499202660045448,
      "learning_rate": 3.8321955054896295e-06,
      "loss": 0.7288,
      "step": 8011
    },
    {
      "epoch": 0.5878371561946862,
      "grad_norm": 1.5290252312974064,
      "learning_rate": 3.831040204497706e-06,
      "loss": 0.6647,
      "step": 8012
    },
    {
      "epoch": 0.5879105257848254,
      "grad_norm": 0.5552448604574367,
      "learning_rate": 3.829884969519837e-06,
      "loss": 0.4919,
      "step": 8013
    },
    {
      "epoch": 0.5879838953749644,
      "grad_norm": 1.5521568356312934,
      "learning_rate": 3.8287298006212605e-06,
      "loss": 0.698,
      "step": 8014
    },
    {
      "epoch": 0.5880572649651036,
      "grad_norm": 1.4291987313097765,
      "learning_rate": 3.827574697867214e-06,
      "loss": 0.7447,
      "step": 8015
    },
    {
      "epoch": 0.5881306345552427,
      "grad_norm": 0.5598412126107517,
      "learning_rate": 3.8264196613229275e-06,
      "loss": 0.4824,
      "step": 8016
    },
    {
      "epoch": 0.5881306345552427,
      "lm_loss": 1.5447,
      "step": 8016,
      "vm_loss": 0.143
    },
    {
      "epoch": 0.5881306345552427,
      "lm_loss": 0.9193,
      "step": 8016,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.5881306345552427,
      "lm_loss": 0.1033,
      "step": 8016,
      "vm_loss": 0.223
    },
    {
      "epoch": 0.5881306345552427,
      "lm_loss": 0.3113,
      "step": 8016,
      "vm_loss": 0.1749
    },
    {
      "epoch": 0.5881306345552427,
      "lm_loss": 0.7099,
      "step": 8016,
      "vm_loss": 0.1144
    },
    {
      "epoch": 0.5881306345552427,
      "lm_loss": 0.5802,
      "step": 8016,
      "vm_loss": 0.0522
    },
    {
      "epoch": 0.5881306345552427,
      "lm_loss": 0.8055,
      "step": 8016,
      "vm_loss": 0.207
    },
    {
      "epoch": 0.5881306345552427,
      "lm_loss": 0.9516,
      "step": 8016,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.5882040041453819,
      "grad_norm": 1.5033056625813845,
      "learning_rate": 3.825264691053629e-06,
      "loss": 0.7052,
      "step": 8017
    },
    {
      "epoch": 0.588277373735521,
      "grad_norm": 1.5150507863109213,
      "learning_rate": 3.824109787124542e-06,
      "loss": 0.7741,
      "step": 8018
    },
    {
      "epoch": 0.5883507433256601,
      "grad_norm": 1.3543365163803989,
      "learning_rate": 3.822954949600887e-06,
      "loss": 0.7478,
      "step": 8019
    },
    {
      "epoch": 0.5884241129157992,
      "grad_norm": 0.5614488344020176,
      "learning_rate": 3.821800178547882e-06,
      "loss": 0.4939,
      "step": 8020
    },
    {
      "epoch": 0.5884974825059384,
      "grad_norm": 1.5924177720078532,
      "learning_rate": 3.8206454740307396e-06,
      "loss": 0.7424,
      "step": 8021
    },
    {
      "epoch": 0.5885708520960775,
      "grad_norm": 0.5468963213148053,
      "learning_rate": 3.819490836114665e-06,
      "loss": 0.4786,
      "step": 8022
    },
    {
      "epoch": 0.5886442216862166,
      "grad_norm": 1.7246104331094454,
      "learning_rate": 3.818336264864869e-06,
      "loss": 0.7952,
      "step": 8023
    },
    {
      "epoch": 0.5887175912763557,
      "grad_norm": 0.531205543872725,
      "learning_rate": 3.8171817603465495e-06,
      "loss": 0.4817,
      "step": 8024
    },
    {
      "epoch": 0.5887175912763557,
      "lm_loss": 1.2368,
      "step": 8024,
      "vm_loss": 0.0673
    },
    {
      "epoch": 0.5887175912763557,
      "lm_loss": 0.0936,
      "step": 8024,
      "vm_loss": 0.1286
    },
    {
      "epoch": 0.5887175912763557,
      "lm_loss": 0.7998,
      "step": 8024,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.5887175912763557,
      "lm_loss": 0.6364,
      "step": 8024,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.5887175912763557,
      "lm_loss": 0.7111,
      "step": 8024,
      "vm_loss": 0.1388
    },
    {
      "epoch": 0.5887175912763557,
      "lm_loss": 0.9355,
      "step": 8024,
      "vm_loss": 0.1823
    },
    {
      "epoch": 0.5887175912763557,
      "lm_loss": 0.6984,
      "step": 8024,
      "vm_loss": 0.0758
    },
    {
      "epoch": 0.5887175912763557,
      "lm_loss": 1.4989,
      "step": 8024,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.5887909608664949,
      "grad_norm": 1.5513297511018085,
      "learning_rate": 3.816027322624905e-06,
      "loss": 0.7115,
      "step": 8025
    },
    {
      "epoch": 0.588864330456634,
      "grad_norm": 1.3938411613973571,
      "learning_rate": 3.8148729517651294e-06,
      "loss": 0.7059,
      "step": 8026
    },
    {
      "epoch": 0.5889377000467731,
      "grad_norm": 1.556481373541172,
      "learning_rate": 3.813718647832414e-06,
      "loss": 0.7869,
      "step": 8027
    },
    {
      "epoch": 0.5890110696369122,
      "grad_norm": 1.471597331304362,
      "learning_rate": 3.8125644108919435e-06,
      "loss": 0.7828,
      "step": 8028
    },
    {
      "epoch": 0.5890844392270513,
      "grad_norm": 0.5462801036616126,
      "learning_rate": 3.811410241008902e-06,
      "loss": 0.5,
      "step": 8029
    },
    {
      "epoch": 0.5891578088171905,
      "grad_norm": 1.4723208387655158,
      "learning_rate": 3.810256138248467e-06,
      "loss": 0.7495,
      "step": 8030
    },
    {
      "epoch": 0.5892311784073296,
      "grad_norm": 1.4737432367036019,
      "learning_rate": 3.8091021026758144e-06,
      "loss": 0.6826,
      "step": 8031
    },
    {
      "epoch": 0.5893045479974688,
      "grad_norm": 1.5050184278498053,
      "learning_rate": 3.8079481343561163e-06,
      "loss": 0.7884,
      "step": 8032
    },
    {
      "epoch": 0.5893045479974688,
      "lm_loss": 0.981,
      "step": 8032,
      "vm_loss": 0.0404
    },
    {
      "epoch": 0.5893045479974688,
      "lm_loss": 1.457,
      "step": 8032,
      "vm_loss": 0.1818
    },
    {
      "epoch": 0.5893045479974688,
      "lm_loss": 0.501,
      "step": 8032,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.5893045479974688,
      "lm_loss": 0.5266,
      "step": 8032,
      "vm_loss": 0.1152
    },
    {
      "epoch": 0.5893045479974688,
      "lm_loss": 0.1439,
      "step": 8032,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.5893045479974688,
      "lm_loss": 0.1949,
      "step": 8032,
      "vm_loss": 0.1302
    },
    {
      "epoch": 0.5893045479974688,
      "lm_loss": 0.4276,
      "step": 8032,
      "vm_loss": 0.1068
    },
    {
      "epoch": 0.5893045479974688,
      "lm_loss": 0.6418,
      "step": 8032,
      "vm_loss": 0.0589
    },
    {
      "epoch": 0.5893779175876078,
      "grad_norm": 1.6304339709602884,
      "learning_rate": 3.8067942333545378e-06,
      "loss": 0.725,
      "step": 8033
    },
    {
      "epoch": 0.589451287177747,
      "grad_norm": 0.5513597514873492,
      "learning_rate": 3.8056403997362446e-06,
      "loss": 0.5101,
      "step": 8034
    },
    {
      "epoch": 0.5895246567678861,
      "grad_norm": 1.8916371351908414,
      "learning_rate": 3.804486633566396e-06,
      "loss": 0.7282,
      "step": 8035
    },
    {
      "epoch": 0.5895980263580253,
      "grad_norm": 0.5453338756446658,
      "learning_rate": 3.8033329349101465e-06,
      "loss": 0.5086,
      "step": 8036
    },
    {
      "epoch": 0.5896713959481644,
      "grad_norm": 0.5658564977114768,
      "learning_rate": 3.802179303832651e-06,
      "loss": 0.4968,
      "step": 8037
    },
    {
      "epoch": 0.5897447655383035,
      "grad_norm": 1.4031634869390672,
      "learning_rate": 3.8010257403990557e-06,
      "loss": 0.7269,
      "step": 8038
    },
    {
      "epoch": 0.5898181351284426,
      "grad_norm": 1.5673425381805852,
      "learning_rate": 3.7998722446745067e-06,
      "loss": 0.7612,
      "step": 8039
    },
    {
      "epoch": 0.5898915047185818,
      "grad_norm": 0.5713724667109171,
      "learning_rate": 3.7987188167241444e-06,
      "loss": 0.4915,
      "step": 8040
    },
    {
      "epoch": 0.5898915047185818,
      "lm_loss": 0.4674,
      "step": 8040,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5898915047185818,
      "lm_loss": 0.4794,
      "step": 8040,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5898915047185818,
      "lm_loss": 0.3912,
      "step": 8040,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5898915047185818,
      "lm_loss": 0.5092,
      "step": 8040,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5898915047185818,
      "lm_loss": 0.4852,
      "step": 8040,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5898915047185818,
      "lm_loss": 0.4595,
      "step": 8040,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5898915047185818,
      "lm_loss": 0.5422,
      "step": 8040,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5898915047185818,
      "lm_loss": 0.3175,
      "step": 8040,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5899648743087209,
      "grad_norm": 0.5735419383327683,
      "learning_rate": 3.7975654566131044e-06,
      "loss": 0.4922,
      "step": 8041
    },
    {
      "epoch": 0.5900382438988601,
      "grad_norm": 1.4815103698596053,
      "learning_rate": 3.7964121644065215e-06,
      "loss": 0.6837,
      "step": 8042
    },
    {
      "epoch": 0.5901116134889991,
      "grad_norm": 1.4473636840617143,
      "learning_rate": 3.795258940169525e-06,
      "loss": 0.7963,
      "step": 8043
    },
    {
      "epoch": 0.5901849830791382,
      "grad_norm": 1.6927888894024328,
      "learning_rate": 3.794105783967238e-06,
      "loss": 0.8454,
      "step": 8044
    },
    {
      "epoch": 0.5902583526692774,
      "grad_norm": 1.452661394252216,
      "learning_rate": 3.792952695864785e-06,
      "loss": 0.7243,
      "step": 8045
    },
    {
      "epoch": 0.5903317222594165,
      "grad_norm": 1.512816919808941,
      "learning_rate": 3.7917996759272813e-06,
      "loss": 0.748,
      "step": 8046
    },
    {
      "epoch": 0.5904050918495557,
      "grad_norm": 1.5255723490200832,
      "learning_rate": 3.790646724219843e-06,
      "loss": 0.7908,
      "step": 8047
    },
    {
      "epoch": 0.5904784614396947,
      "grad_norm": 1.395038894248472,
      "learning_rate": 3.7894938408075794e-06,
      "loss": 0.7756,
      "step": 8048
    },
    {
      "epoch": 0.5904784614396947,
      "lm_loss": 2.5127,
      "step": 8048,
      "vm_loss": 0.0649
    },
    {
      "epoch": 0.5904784614396947,
      "lm_loss": 1.6109,
      "step": 8048,
      "vm_loss": 0.1645
    },
    {
      "epoch": 0.5904784614396947,
      "lm_loss": 1.2686,
      "step": 8048,
      "vm_loss": 0.1571
    },
    {
      "epoch": 0.5904784614396947,
      "lm_loss": 0.0875,
      "step": 8048,
      "vm_loss": 0.2069
    },
    {
      "epoch": 0.5904784614396947,
      "lm_loss": 0.5045,
      "step": 8048,
      "vm_loss": 0.0898
    },
    {
      "epoch": 0.5904784614396947,
      "lm_loss": 0.5423,
      "step": 8048,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.5904784614396947,
      "lm_loss": 0.9624,
      "step": 8048,
      "vm_loss": 0.1623
    },
    {
      "epoch": 0.5904784614396947,
      "lm_loss": 0.7609,
      "step": 8048,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.5905518310298339,
      "grad_norm": 1.7407144729483293,
      "learning_rate": 3.788341025755595e-06,
      "loss": 0.8198,
      "step": 8049
    },
    {
      "epoch": 0.590625200619973,
      "grad_norm": 1.5736952328147704,
      "learning_rate": 3.787188279128995e-06,
      "loss": 0.7304,
      "step": 8050
    },
    {
      "epoch": 0.5906985702101122,
      "grad_norm": 1.6136324220647495,
      "learning_rate": 3.7860356009928757e-06,
      "loss": 0.813,
      "step": 8051
    },
    {
      "epoch": 0.5907719398002513,
      "grad_norm": 1.5065310633359443,
      "learning_rate": 3.7848829914123316e-06,
      "loss": 0.7351,
      "step": 8052
    },
    {
      "epoch": 0.5908453093903904,
      "grad_norm": 0.6031759273642497,
      "learning_rate": 3.7837304504524554e-06,
      "loss": 0.4936,
      "step": 8053
    },
    {
      "epoch": 0.5909186789805295,
      "grad_norm": 0.5973219968571143,
      "learning_rate": 3.782577978178332e-06,
      "loss": 0.484,
      "step": 8054
    },
    {
      "epoch": 0.5909920485706687,
      "grad_norm": 1.6810726598599584,
      "learning_rate": 3.781425574655045e-06,
      "loss": 0.7516,
      "step": 8055
    },
    {
      "epoch": 0.5910654181608078,
      "grad_norm": 1.3878324040167873,
      "learning_rate": 3.7802732399476736e-06,
      "loss": 0.635,
      "step": 8056
    },
    {
      "epoch": 0.5910654181608078,
      "lm_loss": 0.603,
      "step": 8056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5910654181608078,
      "lm_loss": 0.5396,
      "step": 8056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5910654181608078,
      "lm_loss": 0.507,
      "step": 8056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5910654181608078,
      "lm_loss": 0.4206,
      "step": 8056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5910654181608078,
      "lm_loss": 0.5487,
      "step": 8056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5910654181608078,
      "lm_loss": 0.6699,
      "step": 8056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5910654181608078,
      "lm_loss": 0.4351,
      "step": 8056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5910654181608078,
      "lm_loss": 0.4402,
      "step": 8056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.591138787750947,
      "grad_norm": 0.5600713021961302,
      "learning_rate": 3.779120974121293e-06,
      "loss": 0.496,
      "step": 8057
    },
    {
      "epoch": 0.591212157341086,
      "grad_norm": 0.5872928168611107,
      "learning_rate": 3.7779687772409747e-06,
      "loss": 0.5046,
      "step": 8058
    },
    {
      "epoch": 0.5912855269312252,
      "grad_norm": 1.5158110460954208,
      "learning_rate": 3.776816649371786e-06,
      "loss": 0.7427,
      "step": 8059
    },
    {
      "epoch": 0.5913588965213643,
      "grad_norm": 1.6337394756927015,
      "learning_rate": 3.7756645905787893e-06,
      "loss": 0.8609,
      "step": 8060
    },
    {
      "epoch": 0.5914322661115035,
      "grad_norm": 1.5242375022149286,
      "learning_rate": 3.7745126009270462e-06,
      "loss": 0.7033,
      "step": 8061
    },
    {
      "epoch": 0.5915056357016426,
      "grad_norm": 1.4554260372278731,
      "learning_rate": 3.7733606804816113e-06,
      "loss": 0.7576,
      "step": 8062
    },
    {
      "epoch": 0.5915790052917816,
      "grad_norm": 1.3062900280218712,
      "learning_rate": 3.772208829307536e-06,
      "loss": 0.8244,
      "step": 8063
    },
    {
      "epoch": 0.5916523748819208,
      "grad_norm": 1.4865511431136385,
      "learning_rate": 3.7710570474698694e-06,
      "loss": 0.7561,
      "step": 8064
    },
    {
      "epoch": 0.5916523748819208,
      "lm_loss": 0.4417,
      "step": 8064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5916523748819208,
      "lm_loss": 0.4438,
      "step": 8064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5916523748819208,
      "lm_loss": 0.4799,
      "step": 8064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5916523748819208,
      "lm_loss": 0.3957,
      "step": 8064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5916523748819208,
      "lm_loss": 0.423,
      "step": 8064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5916523748819208,
      "lm_loss": 0.3921,
      "step": 8064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5916523748819208,
      "lm_loss": 0.5651,
      "step": 8064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5916523748819208,
      "lm_loss": 0.498,
      "step": 8064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.59172574447206,
      "grad_norm": 0.5990997419926983,
      "learning_rate": 3.7699053350336536e-06,
      "loss": 0.4896,
      "step": 8065
    },
    {
      "epoch": 0.5917991140621991,
      "grad_norm": 1.397423574388287,
      "learning_rate": 3.7687536920639306e-06,
      "loss": 0.7215,
      "step": 8066
    },
    {
      "epoch": 0.5918724836523382,
      "grad_norm": 0.5879391069379925,
      "learning_rate": 3.7676021186257363e-06,
      "loss": 0.4943,
      "step": 8067
    },
    {
      "epoch": 0.5919458532424773,
      "grad_norm": 0.5826345395780489,
      "learning_rate": 3.7664506147841006e-06,
      "loss": 0.4882,
      "step": 8068
    },
    {
      "epoch": 0.5920192228326164,
      "grad_norm": 1.5874259235211934,
      "learning_rate": 3.765299180604055e-06,
      "loss": 0.8607,
      "step": 8069
    },
    {
      "epoch": 0.5920925924227556,
      "grad_norm": 1.6000427914091253,
      "learning_rate": 3.7641478161506227e-06,
      "loss": 0.7288,
      "step": 8070
    },
    {
      "epoch": 0.5921659620128947,
      "grad_norm": 0.5650251120269122,
      "learning_rate": 3.762996521488822e-06,
      "loss": 0.4892,
      "step": 8071
    },
    {
      "epoch": 0.5922393316030339,
      "grad_norm": 1.461487339454635,
      "learning_rate": 3.7618452966836726e-06,
      "loss": 0.6896,
      "step": 8072
    },
    {
      "epoch": 0.5922393316030339,
      "lm_loss": 0.5113,
      "step": 8072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5922393316030339,
      "lm_loss": 0.4872,
      "step": 8072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5922393316030339,
      "lm_loss": 0.5001,
      "step": 8072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5922393316030339,
      "lm_loss": 0.588,
      "step": 8072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5922393316030339,
      "lm_loss": 0.4903,
      "step": 8072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5922393316030339,
      "lm_loss": 0.6118,
      "step": 8072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5922393316030339,
      "lm_loss": 0.4936,
      "step": 8072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5922393316030339,
      "lm_loss": 0.6076,
      "step": 8072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5923127011931729,
      "grad_norm": 0.5820647704245551,
      "learning_rate": 3.7606941418001845e-06,
      "loss": 0.4729,
      "step": 8073
    },
    {
      "epoch": 0.5923860707833121,
      "grad_norm": 1.370234083300767,
      "learning_rate": 3.759543056903368e-06,
      "loss": 0.6919,
      "step": 8074
    },
    {
      "epoch": 0.5924594403734512,
      "grad_norm": 1.5426203181984897,
      "learning_rate": 3.7583920420582276e-06,
      "loss": 0.7469,
      "step": 8075
    },
    {
      "epoch": 0.5925328099635904,
      "grad_norm": 1.486615961361992,
      "learning_rate": 3.757241097329762e-06,
      "loss": 0.7271,
      "step": 8076
    },
    {
      "epoch": 0.5926061795537295,
      "grad_norm": 1.5455951744782301,
      "learning_rate": 3.75609022278297e-06,
      "loss": 0.6933,
      "step": 8077
    },
    {
      "epoch": 0.5926795491438686,
      "grad_norm": 1.5414950613001681,
      "learning_rate": 3.7549394184828442e-06,
      "loss": 0.8176,
      "step": 8078
    },
    {
      "epoch": 0.5927529187340077,
      "grad_norm": 0.5878723209103328,
      "learning_rate": 3.7537886844943716e-06,
      "loss": 0.4943,
      "step": 8079
    },
    {
      "epoch": 0.5928262883241469,
      "grad_norm": 0.5580245519408515,
      "learning_rate": 3.7526380208825386e-06,
      "loss": 0.4808,
      "step": 8080
    },
    {
      "epoch": 0.5928262883241469,
      "lm_loss": 1.2948,
      "step": 8080,
      "vm_loss": 0.0825
    },
    {
      "epoch": 0.5928262883241469,
      "lm_loss": 0.676,
      "step": 8080,
      "vm_loss": 0.1128
    },
    {
      "epoch": 0.5928262883241469,
      "lm_loss": 0.7643,
      "step": 8080,
      "vm_loss": 0.0784
    },
    {
      "epoch": 0.5928262883241469,
      "lm_loss": 0.6401,
      "step": 8080,
      "vm_loss": 0.0781
    },
    {
      "epoch": 0.5928262883241469,
      "lm_loss": 0.0715,
      "step": 8080,
      "vm_loss": 0.1323
    },
    {
      "epoch": 0.5928262883241469,
      "lm_loss": 0.6408,
      "step": 8080,
      "vm_loss": 0.0756
    },
    {
      "epoch": 0.5928262883241469,
      "lm_loss": 0.1297,
      "step": 8080,
      "vm_loss": 0.1249
    },
    {
      "epoch": 0.5928262883241469,
      "lm_loss": 0.675,
      "step": 8080,
      "vm_loss": 0.0713
    },
    {
      "epoch": 0.592899657914286,
      "grad_norm": 1.851904140564482,
      "learning_rate": 3.7514874277123266e-06,
      "loss": 0.7121,
      "step": 8081
    },
    {
      "epoch": 0.5929730275044252,
      "grad_norm": 1.8291949220636001,
      "learning_rate": 3.7503369050487093e-06,
      "loss": 0.7881,
      "step": 8082
    },
    {
      "epoch": 0.5930463970945642,
      "grad_norm": 1.4846622568296548,
      "learning_rate": 3.7491864529566636e-06,
      "loss": 0.8285,
      "step": 8083
    },
    {
      "epoch": 0.5931197666847033,
      "grad_norm": 1.7020064078493897,
      "learning_rate": 3.7480360715011555e-06,
      "loss": 0.7456,
      "step": 8084
    },
    {
      "epoch": 0.5931931362748425,
      "grad_norm": 1.5887816063315885,
      "learning_rate": 3.746885760747151e-06,
      "loss": 0.7619,
      "step": 8085
    },
    {
      "epoch": 0.5932665058649816,
      "grad_norm": 0.5613139263909721,
      "learning_rate": 3.745735520759612e-06,
      "loss": 0.4889,
      "step": 8086
    },
    {
      "epoch": 0.5933398754551208,
      "grad_norm": 1.567496807695652,
      "learning_rate": 3.7445853516034924e-06,
      "loss": 0.7917,
      "step": 8087
    },
    {
      "epoch": 0.5934132450452598,
      "grad_norm": 0.5762282453985961,
      "learning_rate": 3.7434352533437483e-06,
      "loss": 0.4923,
      "step": 8088
    },
    {
      "epoch": 0.5934132450452598,
      "lm_loss": 1.8359,
      "step": 8088,
      "vm_loss": 0.1426
    },
    {
      "epoch": 0.5934132450452598,
      "lm_loss": 0.4213,
      "step": 8088,
      "vm_loss": 0.0575
    },
    {
      "epoch": 0.5934132450452598,
      "lm_loss": 0.2151,
      "step": 8088,
      "vm_loss": 0.1999
    },
    {
      "epoch": 0.5934132450452598,
      "lm_loss": 0.7441,
      "step": 8088,
      "vm_loss": 0.2502
    },
    {
      "epoch": 0.5934132450452598,
      "lm_loss": 0.0672,
      "step": 8088,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.5934132450452598,
      "lm_loss": 0.778,
      "step": 8088,
      "vm_loss": 0.1633
    },
    {
      "epoch": 0.5934132450452598,
      "lm_loss": 0.3347,
      "step": 8088,
      "vm_loss": 0.0813
    },
    {
      "epoch": 0.5934132450452598,
      "lm_loss": 0.0276,
      "step": 8088,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.593486614635399,
      "grad_norm": 1.5420037211533328,
      "learning_rate": 3.7422852260453274e-06,
      "loss": 0.6639,
      "step": 8089
    },
    {
      "epoch": 0.5935599842255381,
      "grad_norm": 0.5728299500994409,
      "learning_rate": 3.7411352697731734e-06,
      "loss": 0.4908,
      "step": 8090
    },
    {
      "epoch": 0.5936333538156773,
      "grad_norm": 1.4469346137399233,
      "learning_rate": 3.739985384592229e-06,
      "loss": 0.7406,
      "step": 8091
    },
    {
      "epoch": 0.5937067234058164,
      "grad_norm": 1.5401233841747053,
      "learning_rate": 3.7388355705674297e-06,
      "loss": 0.8173,
      "step": 8092
    },
    {
      "epoch": 0.5937800929959555,
      "grad_norm": 1.482966861386797,
      "learning_rate": 3.7376858277637096e-06,
      "loss": 0.7204,
      "step": 8093
    },
    {
      "epoch": 0.5938534625860946,
      "grad_norm": 1.3999678066399788,
      "learning_rate": 3.7365361562459966e-06,
      "loss": 0.778,
      "step": 8094
    },
    {
      "epoch": 0.5939268321762338,
      "grad_norm": 1.5475803168688764,
      "learning_rate": 3.7353865560792156e-06,
      "loss": 0.7573,
      "step": 8095
    },
    {
      "epoch": 0.5940002017663729,
      "grad_norm": 1.4964430808797466,
      "learning_rate": 3.734237027328288e-06,
      "loss": 0.7249,
      "step": 8096
    },
    {
      "epoch": 0.5940002017663729,
      "lm_loss": 1.1368,
      "step": 8096,
      "vm_loss": 0.067
    },
    {
      "epoch": 0.5940002017663729,
      "lm_loss": 0.3209,
      "step": 8096,
      "vm_loss": 0.1151
    },
    {
      "epoch": 0.5940002017663729,
      "lm_loss": 0.5435,
      "step": 8096,
      "vm_loss": 0.0445
    },
    {
      "epoch": 0.5940002017663729,
      "lm_loss": 0.7422,
      "step": 8096,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.5940002017663729,
      "lm_loss": 0.7977,
      "step": 8096,
      "vm_loss": 0.2051
    },
    {
      "epoch": 0.5940002017663729,
      "lm_loss": 0.6338,
      "step": 8096,
      "vm_loss": 0.1052
    },
    {
      "epoch": 0.5940002017663729,
      "lm_loss": 0.1454,
      "step": 8096,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.5940002017663729,
      "lm_loss": 0.5866,
      "step": 8096,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.5940735713565121,
      "grad_norm": 1.4483194811006819,
      "learning_rate": 3.7330875700581305e-06,
      "loss": 0.7376,
      "step": 8097
    },
    {
      "epoch": 0.5941469409466511,
      "grad_norm": 1.5726406570944682,
      "learning_rate": 3.7319381843336546e-06,
      "loss": 0.8658,
      "step": 8098
    },
    {
      "epoch": 0.5942203105367903,
      "grad_norm": 1.580167457590635,
      "learning_rate": 3.73078887021977e-06,
      "loss": 0.8759,
      "step": 8099
    },
    {
      "epoch": 0.5942936801269294,
      "grad_norm": 0.620862540771081,
      "learning_rate": 3.7296396277813795e-06,
      "loss": 0.5043,
      "step": 8100
    },
    {
      "epoch": 0.5943670497170686,
      "grad_norm": 0.6126999552321292,
      "learning_rate": 3.7284904570833874e-06,
      "loss": 0.4847,
      "step": 8101
    },
    {
      "epoch": 0.5944404193072076,
      "grad_norm": 1.482683159818261,
      "learning_rate": 3.7273413581906876e-06,
      "loss": 0.7917,
      "step": 8102
    },
    {
      "epoch": 0.5945137888973467,
      "grad_norm": 1.574112528112573,
      "learning_rate": 3.726192331168172e-06,
      "loss": 0.6775,
      "step": 8103
    },
    {
      "epoch": 0.5945871584874859,
      "grad_norm": 1.438792535918883,
      "learning_rate": 3.7250433760807313e-06,
      "loss": 0.7462,
      "step": 8104
    },
    {
      "epoch": 0.5945871584874859,
      "lm_loss": 0.4316,
      "step": 8104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5945871584874859,
      "lm_loss": 0.5135,
      "step": 8104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5945871584874859,
      "lm_loss": 0.4763,
      "step": 8104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5945871584874859,
      "lm_loss": 0.5829,
      "step": 8104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5945871584874859,
      "lm_loss": 0.5378,
      "step": 8104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5945871584874859,
      "lm_loss": 0.47,
      "step": 8104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5945871584874859,
      "lm_loss": 0.5208,
      "step": 8104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.5945871584874859,
      "lm_loss": 0.549,
      "step": 8104,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.594660528077625,
      "grad_norm": 0.5792036265048978,
      "learning_rate": 3.723894492993248e-06,
      "loss": 0.5104,
      "step": 8105
    },
    {
      "epoch": 0.5947338976677642,
      "grad_norm": 1.5482128122855896,
      "learning_rate": 3.722745681970602e-06,
      "loss": 0.767,
      "step": 8106
    },
    {
      "epoch": 0.5948072672579032,
      "grad_norm": 1.5114323163774657,
      "learning_rate": 3.7215969430776718e-06,
      "loss": 0.7003,
      "step": 8107
    },
    {
      "epoch": 0.5948806368480424,
      "grad_norm": 1.421247918069999,
      "learning_rate": 3.7204482763793274e-06,
      "loss": 0.7397,
      "step": 8108
    },
    {
      "epoch": 0.5949540064381815,
      "grad_norm": 1.5245541643191571,
      "learning_rate": 3.719299681940437e-06,
      "loss": 0.8259,
      "step": 8109
    },
    {
      "epoch": 0.5950273760283207,
      "grad_norm": 1.8917215886309382,
      "learning_rate": 3.7181511598258667e-06,
      "loss": 0.9009,
      "step": 8110
    },
    {
      "epoch": 0.5951007456184598,
      "grad_norm": 1.3830135827105785,
      "learning_rate": 3.7170027101004734e-06,
      "loss": 0.7297,
      "step": 8111
    },
    {
      "epoch": 0.5951741152085989,
      "grad_norm": 0.631344834214322,
      "learning_rate": 3.715854332829115e-06,
      "loss": 0.513,
      "step": 8112
    },
    {
      "epoch": 0.5951741152085989,
      "lm_loss": 1.4637,
      "step": 8112,
      "vm_loss": 0.0543
    },
    {
      "epoch": 0.5951741152085989,
      "lm_loss": 1.0454,
      "step": 8112,
      "vm_loss": 0.1504
    },
    {
      "epoch": 0.5951741152085989,
      "lm_loss": 0.4356,
      "step": 8112,
      "vm_loss": 0.2282
    },
    {
      "epoch": 0.5951741152085989,
      "lm_loss": 0.0762,
      "step": 8112,
      "vm_loss": 0.2464
    },
    {
      "epoch": 0.5951741152085989,
      "lm_loss": 0.1427,
      "step": 8112,
      "vm_loss": 0.0438
    },
    {
      "epoch": 0.5951741152085989,
      "lm_loss": 0.1284,
      "step": 8112,
      "vm_loss": 0.1398
    },
    {
      "epoch": 0.5951741152085989,
      "lm_loss": 0.4314,
      "step": 8112,
      "vm_loss": 0.0479
    },
    {
      "epoch": 0.5951741152085989,
      "lm_loss": 0.2875,
      "step": 8112,
      "vm_loss": 0.073
    },
    {
      "epoch": 0.595247484798738,
      "grad_norm": 1.4620735189813687,
      "learning_rate": 3.7147060280766434e-06,
      "loss": 0.7959,
      "step": 8113
    },
    {
      "epoch": 0.5953208543888772,
      "grad_norm": 0.5930753538237453,
      "learning_rate": 3.7135577959079035e-06,
      "loss": 0.4942,
      "step": 8114
    },
    {
      "epoch": 0.5953942239790163,
      "grad_norm": 0.5956340205288773,
      "learning_rate": 3.7124096363877417e-06,
      "loss": 0.4985,
      "step": 8115
    },
    {
      "epoch": 0.5954675935691555,
      "grad_norm": 0.5529715415738183,
      "learning_rate": 3.7112615495809965e-06,
      "loss": 0.4863,
      "step": 8116
    },
    {
      "epoch": 0.5955409631592945,
      "grad_norm": 0.5643487878925251,
      "learning_rate": 3.710113535552503e-06,
      "loss": 0.5106,
      "step": 8117
    },
    {
      "epoch": 0.5956143327494337,
      "grad_norm": 1.5743423523285012,
      "learning_rate": 3.7089655943670925e-06,
      "loss": 0.7788,
      "step": 8118
    },
    {
      "epoch": 0.5956877023395728,
      "grad_norm": 1.533208536750342,
      "learning_rate": 3.7078177260895915e-06,
      "loss": 0.8117,
      "step": 8119
    },
    {
      "epoch": 0.595761071929712,
      "grad_norm": 1.465898126048735,
      "learning_rate": 3.7066699307848243e-06,
      "loss": 0.8207,
      "step": 8120
    },
    {
      "epoch": 0.595761071929712,
      "lm_loss": 1.4552,
      "step": 8120,
      "vm_loss": 0.052
    },
    {
      "epoch": 0.595761071929712,
      "lm_loss": 1.3328,
      "step": 8120,
      "vm_loss": 0.0308
    },
    {
      "epoch": 0.595761071929712,
      "lm_loss": 0.6869,
      "step": 8120,
      "vm_loss": 0.1681
    },
    {
      "epoch": 0.595761071929712,
      "lm_loss": 0.109,
      "step": 8120,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.595761071929712,
      "lm_loss": 0.7162,
      "step": 8120,
      "vm_loss": 0.1108
    },
    {
      "epoch": 0.595761071929712,
      "lm_loss": 0.0436,
      "step": 8120,
      "vm_loss": 0.0651
    },
    {
      "epoch": 0.595761071929712,
      "lm_loss": 1.0479,
      "step": 8120,
      "vm_loss": 0.1465
    },
    {
      "epoch": 0.595761071929712,
      "lm_loss": 0.8919,
      "step": 8120,
      "vm_loss": 0.1502
    },
    {
      "epoch": 0.5958344415198511,
      "grad_norm": 1.7383328601010999,
      "learning_rate": 3.705522208517609e-06,
      "loss": 0.7618,
      "step": 8121
    },
    {
      "epoch": 0.5959078111099901,
      "grad_norm": 0.6575894574882789,
      "learning_rate": 3.7043745593527598e-06,
      "loss": 0.4914,
      "step": 8122
    },
    {
      "epoch": 0.5959811807001293,
      "grad_norm": 1.619085126745735,
      "learning_rate": 3.703226983355088e-06,
      "loss": 0.791,
      "step": 8123
    },
    {
      "epoch": 0.5960545502902684,
      "grad_norm": 1.6258737950594222,
      "learning_rate": 3.7020794805894005e-06,
      "loss": 0.8034,
      "step": 8124
    },
    {
      "epoch": 0.5961279198804076,
      "grad_norm": 1.4457252628515502,
      "learning_rate": 3.7009320511204984e-06,
      "loss": 0.7894,
      "step": 8125
    },
    {
      "epoch": 0.5962012894705467,
      "grad_norm": 1.4676623935753683,
      "learning_rate": 3.69978469501318e-06,
      "loss": 0.6991,
      "step": 8126
    },
    {
      "epoch": 0.5962746590606858,
      "grad_norm": 1.366819187049393,
      "learning_rate": 3.69863741233224e-06,
      "loss": 0.8694,
      "step": 8127
    },
    {
      "epoch": 0.5963480286508249,
      "grad_norm": 1.357127055898182,
      "learning_rate": 3.6974902031424687e-06,
      "loss": 0.7761,
      "step": 8128
    },
    {
      "epoch": 0.5963480286508249,
      "lm_loss": 1.5395,
      "step": 8128,
      "vm_loss": 0.0719
    },
    {
      "epoch": 0.5963480286508249,
      "lm_loss": 1.5925,
      "step": 8128,
      "vm_loss": 0.0752
    },
    {
      "epoch": 0.5963480286508249,
      "lm_loss": 0.1791,
      "step": 8128,
      "vm_loss": 0.1218
    },
    {
      "epoch": 0.5963480286508249,
      "lm_loss": 0.1934,
      "step": 8128,
      "vm_loss": 0.103
    },
    {
      "epoch": 0.5963480286508249,
      "lm_loss": 0.0762,
      "step": 8128,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.5963480286508249,
      "lm_loss": 0.7131,
      "step": 8128,
      "vm_loss": 0.1187
    },
    {
      "epoch": 0.5963480286508249,
      "lm_loss": 0.6899,
      "step": 8128,
      "vm_loss": 0.1359
    },
    {
      "epoch": 0.5963480286508249,
      "lm_loss": 0.3739,
      "step": 8128,
      "vm_loss": 0.1465
    },
    {
      "epoch": 0.5964213982409641,
      "grad_norm": 1.4152150722266088,
      "learning_rate": 3.696343067508651e-06,
      "loss": 0.7118,
      "step": 8129
    },
    {
      "epoch": 0.5964947678311032,
      "grad_norm": 1.5537069925466518,
      "learning_rate": 3.695196005495568e-06,
      "loss": 0.7885,
      "step": 8130
    },
    {
      "epoch": 0.5965681374212424,
      "grad_norm": 1.4101748836436314,
      "learning_rate": 3.6940490171679988e-06,
      "loss": 0.7076,
      "step": 8131
    },
    {
      "epoch": 0.5966415070113814,
      "grad_norm": 1.4701214822454969,
      "learning_rate": 3.692902102590716e-06,
      "loss": 0.728,
      "step": 8132
    },
    {
      "epoch": 0.5967148766015206,
      "grad_norm": 1.4081415342624914,
      "learning_rate": 3.6917552618284874e-06,
      "loss": 0.8175,
      "step": 8133
    },
    {
      "epoch": 0.5967882461916597,
      "grad_norm": 0.5685742728975167,
      "learning_rate": 3.6906084949460794e-06,
      "loss": 0.4821,
      "step": 8134
    },
    {
      "epoch": 0.5968616157817989,
      "grad_norm": 0.5732121946036329,
      "learning_rate": 3.689461802008252e-06,
      "loss": 0.4799,
      "step": 8135
    },
    {
      "epoch": 0.596934985371938,
      "grad_norm": 1.3806389161365014,
      "learning_rate": 3.6883151830797626e-06,
      "loss": 0.7897,
      "step": 8136
    },
    {
      "epoch": 0.596934985371938,
      "lm_loss": 0.4953,
      "step": 8136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.596934985371938,
      "lm_loss": 0.361,
      "step": 8136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.596934985371938,
      "lm_loss": 0.4712,
      "step": 8136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.596934985371938,
      "lm_loss": 0.4821,
      "step": 8136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.596934985371938,
      "lm_loss": 0.442,
      "step": 8136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.596934985371938,
      "lm_loss": 0.5151,
      "step": 8136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.596934985371938,
      "lm_loss": 0.5168,
      "step": 8136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.596934985371938,
      "lm_loss": 0.4448,
      "step": 8136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.597008354962077,
      "grad_norm": 0.5405317430580376,
      "learning_rate": 3.6871686382253637e-06,
      "loss": 0.4602,
      "step": 8137
    },
    {
      "epoch": 0.5970817245522162,
      "grad_norm": 1.459666785560123,
      "learning_rate": 3.6860221675098016e-06,
      "loss": 0.8432,
      "step": 8138
    },
    {
      "epoch": 0.5971550941423553,
      "grad_norm": 1.574223041853965,
      "learning_rate": 3.6848757709978235e-06,
      "loss": 0.7493,
      "step": 8139
    },
    {
      "epoch": 0.5972284637324945,
      "grad_norm": 1.6426896592518638,
      "learning_rate": 3.6837294487541665e-06,
      "loss": 0.7774,
      "step": 8140
    },
    {
      "epoch": 0.5973018333226336,
      "grad_norm": 0.5499722993903622,
      "learning_rate": 3.682583200843566e-06,
      "loss": 0.4996,
      "step": 8141
    },
    {
      "epoch": 0.5973752029127727,
      "grad_norm": 1.5939148067622513,
      "learning_rate": 3.6814370273307564e-06,
      "loss": 0.7582,
      "step": 8142
    },
    {
      "epoch": 0.5974485725029118,
      "grad_norm": 0.5670299331181144,
      "learning_rate": 3.6802909282804635e-06,
      "loss": 0.4825,
      "step": 8143
    },
    {
      "epoch": 0.597521942093051,
      "grad_norm": 1.518354221995585,
      "learning_rate": 3.679144903757408e-06,
      "loss": 0.7558,
      "step": 8144
    },
    {
      "epoch": 0.597521942093051,
      "lm_loss": 0.8912,
      "step": 8144,
      "vm_loss": 0.0642
    },
    {
      "epoch": 0.597521942093051,
      "lm_loss": 1.2301,
      "step": 8144,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.597521942093051,
      "lm_loss": 1.2348,
      "step": 8144,
      "vm_loss": 0.1322
    },
    {
      "epoch": 0.597521942093051,
      "lm_loss": 0.2939,
      "step": 8144,
      "vm_loss": 0.0858
    },
    {
      "epoch": 0.597521942093051,
      "lm_loss": 0.093,
      "step": 8144,
      "vm_loss": 0.0881
    },
    {
      "epoch": 0.597521942093051,
      "lm_loss": 0.452,
      "step": 8144,
      "vm_loss": 0.082
    },
    {
      "epoch": 0.597521942093051,
      "lm_loss": 1.2088,
      "step": 8144,
      "vm_loss": 0.1636
    },
    {
      "epoch": 0.597521942093051,
      "lm_loss": 0.1173,
      "step": 8144,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.5975953116831901,
      "grad_norm": 1.6297015446159024,
      "learning_rate": 3.677998953826313e-06,
      "loss": 0.8293,
      "step": 8145
    },
    {
      "epoch": 0.5976686812733293,
      "grad_norm": 1.4249679806904012,
      "learning_rate": 3.676853078551891e-06,
      "loss": 0.7787,
      "step": 8146
    },
    {
      "epoch": 0.5977420508634683,
      "grad_norm": 1.4864062522576216,
      "learning_rate": 3.6757072779988516e-06,
      "loss": 0.8155,
      "step": 8147
    },
    {
      "epoch": 0.5978154204536075,
      "grad_norm": 1.5751189400286758,
      "learning_rate": 3.6745615522319013e-06,
      "loss": 0.8189,
      "step": 8148
    },
    {
      "epoch": 0.5978887900437466,
      "grad_norm": 1.3495849958483617,
      "learning_rate": 3.673415901315743e-06,
      "loss": 0.8432,
      "step": 8149
    },
    {
      "epoch": 0.5979621596338858,
      "grad_norm": 1.3318201049543636,
      "learning_rate": 3.6722703253150736e-06,
      "loss": 0.706,
      "step": 8150
    },
    {
      "epoch": 0.5980355292240249,
      "grad_norm": 1.532514106021706,
      "learning_rate": 3.6711248242945868e-06,
      "loss": 0.7726,
      "step": 8151
    },
    {
      "epoch": 0.598108898814164,
      "grad_norm": 1.512303697382611,
      "learning_rate": 3.669979398318973e-06,
      "loss": 0.8302,
      "step": 8152
    },
    {
      "epoch": 0.598108898814164,
      "lm_loss": 1.9428,
      "step": 8152,
      "vm_loss": 0.0752
    },
    {
      "epoch": 0.598108898814164,
      "lm_loss": 0.1311,
      "step": 8152,
      "vm_loss": 0.0669
    },
    {
      "epoch": 0.598108898814164,
      "lm_loss": 0.9131,
      "step": 8152,
      "vm_loss": 0.0846
    },
    {
      "epoch": 0.598108898814164,
      "lm_loss": 1.0583,
      "step": 8152,
      "vm_loss": 0.1316
    },
    {
      "epoch": 0.598108898814164,
      "lm_loss": 0.637,
      "step": 8152,
      "vm_loss": 0.1767
    },
    {
      "epoch": 0.598108898814164,
      "lm_loss": 0.2873,
      "step": 8152,
      "vm_loss": 0.1084
    },
    {
      "epoch": 0.598108898814164,
      "lm_loss": 0.065,
      "step": 8152,
      "vm_loss": 0.0573
    },
    {
      "epoch": 0.598108898814164,
      "lm_loss": 0.8522,
      "step": 8152,
      "vm_loss": 0.2488
    },
    {
      "epoch": 0.5981822684043031,
      "grad_norm": 1.5474070834402736,
      "learning_rate": 3.6688340474529147e-06,
      "loss": 0.7132,
      "step": 8153
    },
    {
      "epoch": 0.5982556379944423,
      "grad_norm": 1.5530303095728732,
      "learning_rate": 3.6676887717610965e-06,
      "loss": 0.738,
      "step": 8154
    },
    {
      "epoch": 0.5983290075845814,
      "grad_norm": 1.484930950863559,
      "learning_rate": 3.6665435713081922e-06,
      "loss": 0.7167,
      "step": 8155
    },
    {
      "epoch": 0.5984023771747206,
      "grad_norm": 1.4361543814819753,
      "learning_rate": 3.665398446158873e-06,
      "loss": 0.7845,
      "step": 8156
    },
    {
      "epoch": 0.5984757467648596,
      "grad_norm": 1.5140783325928713,
      "learning_rate": 3.6642533963778103e-06,
      "loss": 0.6481,
      "step": 8157
    },
    {
      "epoch": 0.5985491163549987,
      "grad_norm": 1.5193140432679197,
      "learning_rate": 3.663108422029666e-06,
      "loss": 0.8416,
      "step": 8158
    },
    {
      "epoch": 0.5986224859451379,
      "grad_norm": 1.5410071768109395,
      "learning_rate": 3.6619635231790985e-06,
      "loss": 0.7413,
      "step": 8159
    },
    {
      "epoch": 0.598695855535277,
      "grad_norm": 1.5754742056640625,
      "learning_rate": 3.6608186998907656e-06,
      "loss": 0.7195,
      "step": 8160
    },
    {
      "epoch": 0.598695855535277,
      "lm_loss": 1.2397,
      "step": 8160,
      "vm_loss": 0.0194
    },
    {
      "epoch": 0.598695855535277,
      "lm_loss": 0.0581,
      "step": 8160,
      "vm_loss": 0.1085
    },
    {
      "epoch": 0.598695855535277,
      "lm_loss": 0.2523,
      "step": 8160,
      "vm_loss": 0.1735
    },
    {
      "epoch": 0.598695855535277,
      "lm_loss": 0.7046,
      "step": 8160,
      "vm_loss": 0.2072
    },
    {
      "epoch": 0.598695855535277,
      "lm_loss": 0.598,
      "step": 8160,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.598695855535277,
      "lm_loss": 0.0808,
      "step": 8160,
      "vm_loss": 0.1454
    },
    {
      "epoch": 0.598695855535277,
      "lm_loss": 1.021,
      "step": 8160,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.598695855535277,
      "lm_loss": 0.5398,
      "step": 8160,
      "vm_loss": 0.1895
    },
    {
      "epoch": 0.5987692251254162,
      "grad_norm": 1.451603173889727,
      "learning_rate": 3.6596739522293166e-06,
      "loss": 0.7719,
      "step": 8161
    },
    {
      "epoch": 0.5988425947155552,
      "grad_norm": 1.4443086153854066,
      "learning_rate": 3.6585292802593997e-06,
      "loss": 0.7767,
      "step": 8162
    },
    {
      "epoch": 0.5989159643056944,
      "grad_norm": 1.5769588733748694,
      "learning_rate": 3.657384684045656e-06,
      "loss": 0.79,
      "step": 8163
    },
    {
      "epoch": 0.5989893338958335,
      "grad_norm": 1.473823682273607,
      "learning_rate": 3.6562401636527235e-06,
      "loss": 0.772,
      "step": 8164
    },
    {
      "epoch": 0.5990627034859727,
      "grad_norm": 0.587667591397217,
      "learning_rate": 3.655095719145238e-06,
      "loss": 0.508,
      "step": 8165
    },
    {
      "epoch": 0.5991360730761118,
      "grad_norm": 1.4177417531778695,
      "learning_rate": 3.6539513505878277e-06,
      "loss": 0.7906,
      "step": 8166
    },
    {
      "epoch": 0.5992094426662509,
      "grad_norm": 1.656496948904312,
      "learning_rate": 3.6528070580451168e-06,
      "loss": 0.6838,
      "step": 8167
    },
    {
      "epoch": 0.59928281225639,
      "grad_norm": 1.5224882976834828,
      "learning_rate": 3.6516628415817286e-06,
      "loss": 0.794,
      "step": 8168
    },
    {
      "epoch": 0.59928281225639,
      "lm_loss": 1.2365,
      "step": 8168,
      "vm_loss": 0.0945
    },
    {
      "epoch": 0.59928281225639,
      "lm_loss": 0.1705,
      "step": 8168,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.59928281225639,
      "lm_loss": 0.2068,
      "step": 8168,
      "vm_loss": 0.131
    },
    {
      "epoch": 0.59928281225639,
      "lm_loss": 0.4094,
      "step": 8168,
      "vm_loss": 0.1249
    },
    {
      "epoch": 0.59928281225639,
      "lm_loss": 0.1042,
      "step": 8168,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.59928281225639,
      "lm_loss": 0.7555,
      "step": 8168,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.59928281225639,
      "lm_loss": 0.8071,
      "step": 8168,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.59928281225639,
      "lm_loss": 1.4862,
      "step": 8168,
      "vm_loss": 0.1342
    },
    {
      "epoch": 0.5993561818465292,
      "grad_norm": 1.5548676787868425,
      "learning_rate": 3.650518701262278e-06,
      "loss": 0.8048,
      "step": 8169
    },
    {
      "epoch": 0.5994295514366683,
      "grad_norm": 1.5242632131648381,
      "learning_rate": 3.6493746371513787e-06,
      "loss": 0.7445,
      "step": 8170
    },
    {
      "epoch": 0.5995029210268075,
      "grad_norm": 0.552986216698013,
      "learning_rate": 3.6482306493136387e-06,
      "loss": 0.494,
      "step": 8171
    },
    {
      "epoch": 0.5995762906169465,
      "grad_norm": 1.5159090129649264,
      "learning_rate": 3.647086737813661e-06,
      "loss": 0.7148,
      "step": 8172
    },
    {
      "epoch": 0.5996496602070857,
      "grad_norm": 1.4006583937445636,
      "learning_rate": 3.6459429027160466e-06,
      "loss": 0.8008,
      "step": 8173
    },
    {
      "epoch": 0.5997230297972248,
      "grad_norm": 1.5422742899284796,
      "learning_rate": 3.644799144085389e-06,
      "loss": 0.8499,
      "step": 8174
    },
    {
      "epoch": 0.599796399387364,
      "grad_norm": 1.6785779738899378,
      "learning_rate": 3.6436554619862797e-06,
      "loss": 0.7816,
      "step": 8175
    },
    {
      "epoch": 0.5998697689775031,
      "grad_norm": 1.5826510613167941,
      "learning_rate": 3.642511856483306e-06,
      "loss": 0.7818,
      "step": 8176
    },
    {
      "epoch": 0.5998697689775031,
      "lm_loss": 1.0199,
      "step": 8176,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.5998697689775031,
      "lm_loss": 0.674,
      "step": 8176,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.5998697689775031,
      "lm_loss": 0.9204,
      "step": 8176,
      "vm_loss": 0.1769
    },
    {
      "epoch": 0.5998697689775031,
      "lm_loss": 0.6072,
      "step": 8176,
      "vm_loss": 0.2213
    },
    {
      "epoch": 0.5998697689775031,
      "lm_loss": 0.7631,
      "step": 8176,
      "vm_loss": 0.0851
    },
    {
      "epoch": 0.5998697689775031,
      "lm_loss": 0.82,
      "step": 8176,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.5998697689775031,
      "lm_loss": 0.0656,
      "step": 8176,
      "vm_loss": 0.0986
    },
    {
      "epoch": 0.5998697689775031,
      "lm_loss": 0.8733,
      "step": 8176,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.5999431385676421,
      "grad_norm": 1.590532739387375,
      "learning_rate": 3.6413683276410493e-06,
      "loss": 0.815,
      "step": 8177
    },
    {
      "epoch": 0.6000165081577813,
      "grad_norm": 1.4635653134815108,
      "learning_rate": 3.6402248755240877e-06,
      "loss": 0.7816,
      "step": 8178
    },
    {
      "epoch": 0.6000898777479204,
      "grad_norm": 1.559934714605077,
      "learning_rate": 3.6390815001969947e-06,
      "loss": 0.7245,
      "step": 8179
    },
    {
      "epoch": 0.6001632473380596,
      "grad_norm": 1.3761910050639947,
      "learning_rate": 3.6379382017243393e-06,
      "loss": 0.7135,
      "step": 8180
    },
    {
      "epoch": 0.6002366169281986,
      "grad_norm": 1.5021543427750745,
      "learning_rate": 3.636794980170687e-06,
      "loss": 0.7835,
      "step": 8181
    },
    {
      "epoch": 0.6003099865183378,
      "grad_norm": 0.5473092588347725,
      "learning_rate": 3.635651835600599e-06,
      "loss": 0.4659,
      "step": 8182
    },
    {
      "epoch": 0.6003833561084769,
      "grad_norm": 1.4388447637995414,
      "learning_rate": 3.634508768078629e-06,
      "loss": 0.6812,
      "step": 8183
    },
    {
      "epoch": 0.6004567256986161,
      "grad_norm": 1.4085827103822328,
      "learning_rate": 3.6333657776693314e-06,
      "loss": 0.7313,
      "step": 8184
    },
    {
      "epoch": 0.6004567256986161,
      "lm_loss": 1.3895,
      "step": 8184,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.6004567256986161,
      "lm_loss": 0.2944,
      "step": 8184,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.6004567256986161,
      "lm_loss": 0.7747,
      "step": 8184,
      "vm_loss": 0.1367
    },
    {
      "epoch": 0.6004567256986161,
      "lm_loss": 1.157,
      "step": 8184,
      "vm_loss": 0.1131
    },
    {
      "epoch": 0.6004567256986161,
      "lm_loss": 0.8475,
      "step": 8184,
      "vm_loss": 0.1851
    },
    {
      "epoch": 0.6004567256986161,
      "lm_loss": 0.0281,
      "step": 8184,
      "vm_loss": 0.1028
    },
    {
      "epoch": 0.6004567256986161,
      "lm_loss": 0.0992,
      "step": 8184,
      "vm_loss": 0.1005
    },
    {
      "epoch": 0.6004567256986161,
      "lm_loss": 0.838,
      "step": 8184,
      "vm_loss": 0.157
    },
    {
      "epoch": 0.6005300952887552,
      "grad_norm": 1.4836765331428043,
      "learning_rate": 3.6322228644372533e-06,
      "loss": 0.7774,
      "step": 8185
    },
    {
      "epoch": 0.6006034648788943,
      "grad_norm": 1.5835021405433094,
      "learning_rate": 3.631080028446936e-06,
      "loss": 0.693,
      "step": 8186
    },
    {
      "epoch": 0.6006768344690334,
      "grad_norm": 1.6578670230962957,
      "learning_rate": 3.62993726976292e-06,
      "loss": 0.69,
      "step": 8187
    },
    {
      "epoch": 0.6007502040591726,
      "grad_norm": 0.572489330467095,
      "learning_rate": 3.6287945884497387e-06,
      "loss": 0.4864,
      "step": 8188
    },
    {
      "epoch": 0.6008235736493117,
      "grad_norm": 1.4990280303218284,
      "learning_rate": 3.6276519845719237e-06,
      "loss": 0.7951,
      "step": 8189
    },
    {
      "epoch": 0.6008969432394509,
      "grad_norm": 1.4172918901986642,
      "learning_rate": 3.6265094581940003e-06,
      "loss": 0.6486,
      "step": 8190
    },
    {
      "epoch": 0.6009703128295899,
      "grad_norm": 1.636664251195115,
      "learning_rate": 3.6253670093804877e-06,
      "loss": 0.7764,
      "step": 8191
    },
    {
      "epoch": 0.601043682419729,
      "grad_norm": 0.5608355896964885,
      "learning_rate": 3.6242246381959058e-06,
      "loss": 0.4658,
      "step": 8192
    },
    {
      "epoch": 0.601043682419729,
      "lm_loss": 1.4202,
      "step": 8192,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.601043682419729,
      "lm_loss": 0.5089,
      "step": 8192,
      "vm_loss": 0.0552
    },
    {
      "epoch": 0.601043682419729,
      "lm_loss": 0.6497,
      "step": 8192,
      "vm_loss": 0.2666
    },
    {
      "epoch": 0.601043682419729,
      "lm_loss": 0.9994,
      "step": 8192,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.601043682419729,
      "lm_loss": 0.5024,
      "step": 8192,
      "vm_loss": 0.224
    },
    {
      "epoch": 0.601043682419729,
      "lm_loss": 0.6992,
      "step": 8192,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.601043682419729,
      "lm_loss": 0.2544,
      "step": 8192,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.601043682419729,
      "lm_loss": 0.0588,
      "step": 8192,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.6011170520098682,
      "grad_norm": 1.4541291450128015,
      "learning_rate": 3.623082344704765e-06,
      "loss": 0.7298,
      "step": 8193
    },
    {
      "epoch": 0.6011904216000074,
      "grad_norm": 0.5511271201486257,
      "learning_rate": 3.621940128971573e-06,
      "loss": 0.4963,
      "step": 8194
    },
    {
      "epoch": 0.6012637911901465,
      "grad_norm": 1.4089129389228283,
      "learning_rate": 3.620797991060836e-06,
      "loss": 0.7131,
      "step": 8195
    },
    {
      "epoch": 0.6013371607802855,
      "grad_norm": 0.5726139056945293,
      "learning_rate": 3.6196559310370507e-06,
      "loss": 0.4768,
      "step": 8196
    },
    {
      "epoch": 0.6014105303704247,
      "grad_norm": 1.5050390136257803,
      "learning_rate": 3.618513948964714e-06,
      "loss": 0.7305,
      "step": 8197
    },
    {
      "epoch": 0.6014838999605638,
      "grad_norm": 1.4784883022759845,
      "learning_rate": 3.617372044908316e-06,
      "loss": 0.8164,
      "step": 8198
    },
    {
      "epoch": 0.601557269550703,
      "grad_norm": 1.3154787716977054,
      "learning_rate": 3.6162302189323422e-06,
      "loss": 0.8204,
      "step": 8199
    },
    {
      "epoch": 0.6016306391408421,
      "grad_norm": 1.5042030447016914,
      "learning_rate": 3.6150884711012756e-06,
      "loss": 0.7765,
      "step": 8200
    },
    {
      "epoch": 0.6016306391408421,
      "lm_loss": 1.0695,
      "step": 8200,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.6016306391408421,
      "lm_loss": 0.0465,
      "step": 8200,
      "vm_loss": 0.1633
    },
    {
      "epoch": 0.6016306391408421,
      "lm_loss": 0.1506,
      "step": 8200,
      "vm_loss": 0.0543
    },
    {
      "epoch": 0.6016306391408421,
      "lm_loss": 0.636,
      "step": 8200,
      "vm_loss": 0.0581
    },
    {
      "epoch": 0.6016306391408421,
      "lm_loss": 1.0386,
      "step": 8200,
      "vm_loss": 0.2113
    },
    {
      "epoch": 0.6016306391408421,
      "lm_loss": 0.6768,
      "step": 8200,
      "vm_loss": 0.1795
    },
    {
      "epoch": 0.6016306391408421,
      "lm_loss": 0.8676,
      "step": 8200,
      "vm_loss": 0.1623
    },
    {
      "epoch": 0.6016306391408421,
      "lm_loss": 0.4064,
      "step": 8200,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.6017040087309812,
      "grad_norm": 1.4869372767017195,
      "learning_rate": 3.6139468014795927e-06,
      "loss": 0.7152,
      "step": 8201
    },
    {
      "epoch": 0.6017773783211203,
      "grad_norm": 1.6539020039059922,
      "learning_rate": 3.6128052101317653e-06,
      "loss": 0.711,
      "step": 8202
    },
    {
      "epoch": 0.6018507479112595,
      "grad_norm": 1.4961284794376857,
      "learning_rate": 3.611663697122264e-06,
      "loss": 0.7219,
      "step": 8203
    },
    {
      "epoch": 0.6019241175013986,
      "grad_norm": 1.4666844877978154,
      "learning_rate": 3.610522262515552e-06,
      "loss": 0.7206,
      "step": 8204
    },
    {
      "epoch": 0.6019974870915378,
      "grad_norm": 1.575471388685977,
      "learning_rate": 3.609380906376088e-06,
      "loss": 0.8766,
      "step": 8205
    },
    {
      "epoch": 0.6020708566816768,
      "grad_norm": 1.3794953827443193,
      "learning_rate": 3.6082396287683285e-06,
      "loss": 0.7392,
      "step": 8206
    },
    {
      "epoch": 0.602144226271816,
      "grad_norm": 1.4294582747410856,
      "learning_rate": 3.6070984297567226e-06,
      "loss": 0.821,
      "step": 8207
    },
    {
      "epoch": 0.6022175958619551,
      "grad_norm": 1.4688999001574083,
      "learning_rate": 3.60595730940572e-06,
      "loss": 0.6393,
      "step": 8208
    },
    {
      "epoch": 0.6022175958619551,
      "lm_loss": 0.354,
      "step": 8208,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.6022175958619551,
      "lm_loss": 1.5419,
      "step": 8208,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.6022175958619551,
      "lm_loss": 0.9079,
      "step": 8208,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.6022175958619551,
      "lm_loss": 0.7432,
      "step": 8208,
      "vm_loss": 0.1512
    },
    {
      "epoch": 0.6022175958619551,
      "lm_loss": 0.5307,
      "step": 8208,
      "vm_loss": 0.1468
    },
    {
      "epoch": 0.6022175958619551,
      "lm_loss": 0.6132,
      "step": 8208,
      "vm_loss": 0.2287
    },
    {
      "epoch": 0.6022175958619551,
      "lm_loss": 0.462,
      "step": 8208,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.6022175958619551,
      "lm_loss": 0.7966,
      "step": 8208,
      "vm_loss": 0.1301
    },
    {
      "epoch": 0.6022909654520943,
      "grad_norm": 1.3193185912886267,
      "learning_rate": 3.6048162677797595e-06,
      "loss": 0.8128,
      "step": 8209
    },
    {
      "epoch": 0.6023643350422334,
      "grad_norm": 1.404430387641183,
      "learning_rate": 3.603675304943278e-06,
      "loss": 0.8312,
      "step": 8210
    },
    {
      "epoch": 0.6024377046323725,
      "grad_norm": 1.4806937312731487,
      "learning_rate": 3.602534420960712e-06,
      "loss": 0.7596,
      "step": 8211
    },
    {
      "epoch": 0.6025110742225116,
      "grad_norm": 1.3649392831975402,
      "learning_rate": 3.601393615896488e-06,
      "loss": 0.8001,
      "step": 8212
    },
    {
      "epoch": 0.6025844438126507,
      "grad_norm": 1.5237979926057967,
      "learning_rate": 3.6002528898150284e-06,
      "loss": 0.6568,
      "step": 8213
    },
    {
      "epoch": 0.6026578134027899,
      "grad_norm": 1.3676191805383062,
      "learning_rate": 3.5991122427807557e-06,
      "loss": 0.7656,
      "step": 8214
    },
    {
      "epoch": 0.602731182992929,
      "grad_norm": 1.3280781239995452,
      "learning_rate": 3.5979716748580834e-06,
      "loss": 0.7621,
      "step": 8215
    },
    {
      "epoch": 0.6028045525830681,
      "grad_norm": 1.5065044827322356,
      "learning_rate": 3.5968311861114237e-06,
      "loss": 0.8052,
      "step": 8216
    },
    {
      "epoch": 0.6028045525830681,
      "lm_loss": 1.2069,
      "step": 8216,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.6028045525830681,
      "lm_loss": 1.5773,
      "step": 8216,
      "vm_loss": 0.1904
    },
    {
      "epoch": 0.6028045525830681,
      "lm_loss": 0.6678,
      "step": 8216,
      "vm_loss": 0.14
    },
    {
      "epoch": 0.6028045525830681,
      "lm_loss": 0.4525,
      "step": 8216,
      "vm_loss": 0.1878
    },
    {
      "epoch": 0.6028045525830681,
      "lm_loss": 0.327,
      "step": 8216,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.6028045525830681,
      "lm_loss": 0.3855,
      "step": 8216,
      "vm_loss": 0.1697
    },
    {
      "epoch": 0.6028045525830681,
      "lm_loss": 0.4775,
      "step": 8216,
      "vm_loss": 0.1689
    },
    {
      "epoch": 0.6028045525830681,
      "lm_loss": 0.1453,
      "step": 8216,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.6028779221732072,
      "grad_norm": 1.464783620246676,
      "learning_rate": 3.5956907766051806e-06,
      "loss": 0.7802,
      "step": 8217
    },
    {
      "epoch": 0.6029512917633464,
      "grad_norm": 1.5586674557788678,
      "learning_rate": 3.594550446403757e-06,
      "loss": 0.7269,
      "step": 8218
    },
    {
      "epoch": 0.6030246613534855,
      "grad_norm": 1.4967928697141306,
      "learning_rate": 3.5934101955715516e-06,
      "loss": 0.7843,
      "step": 8219
    },
    {
      "epoch": 0.6030980309436247,
      "grad_norm": 0.6272176841766084,
      "learning_rate": 3.5922700241729557e-06,
      "loss": 0.4888,
      "step": 8220
    },
    {
      "epoch": 0.6031714005337637,
      "grad_norm": 1.5289799711043726,
      "learning_rate": 3.5911299322723558e-06,
      "loss": 0.7575,
      "step": 8221
    },
    {
      "epoch": 0.6032447701239029,
      "grad_norm": 0.5856505265002832,
      "learning_rate": 3.589989919934139e-06,
      "loss": 0.4968,
      "step": 8222
    },
    {
      "epoch": 0.603318139714042,
      "grad_norm": 1.7823258142084875,
      "learning_rate": 3.588849987222683e-06,
      "loss": 0.792,
      "step": 8223
    },
    {
      "epoch": 0.6033915093041812,
      "grad_norm": 1.6804895110537694,
      "learning_rate": 3.5877101342023635e-06,
      "loss": 0.7895,
      "step": 8224
    },
    {
      "epoch": 0.6033915093041812,
      "lm_loss": 0.63,
      "step": 8224,
      "vm_loss": 0.068
    },
    {
      "epoch": 0.6033915093041812,
      "lm_loss": 0.3041,
      "step": 8224,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.6033915093041812,
      "lm_loss": 0.5118,
      "step": 8224,
      "vm_loss": 0.1354
    },
    {
      "epoch": 0.6033915093041812,
      "lm_loss": 0.47,
      "step": 8224,
      "vm_loss": 0.1693
    },
    {
      "epoch": 0.6033915093041812,
      "lm_loss": 0.0662,
      "step": 8224,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.6033915093041812,
      "lm_loss": 0.6895,
      "step": 8224,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.6033915093041812,
      "lm_loss": 0.6106,
      "step": 8224,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.6033915093041812,
      "lm_loss": 0.0703,
      "step": 8224,
      "vm_loss": 0.1416
    },
    {
      "epoch": 0.6034648788943203,
      "grad_norm": 1.50126834031341,
      "learning_rate": 3.58657036093755e-06,
      "loss": 0.7071,
      "step": 8225
    },
    {
      "epoch": 0.6035382484844594,
      "grad_norm": 1.4641470651211814,
      "learning_rate": 3.5854306674926077e-06,
      "loss": 0.7009,
      "step": 8226
    },
    {
      "epoch": 0.6036116180745985,
      "grad_norm": 1.47923416104518,
      "learning_rate": 3.5842910539319e-06,
      "loss": 0.6602,
      "step": 8227
    },
    {
      "epoch": 0.6036849876647377,
      "grad_norm": 1.3917566391204297,
      "learning_rate": 3.5831515203197814e-06,
      "loss": 0.7377,
      "step": 8228
    },
    {
      "epoch": 0.6037583572548768,
      "grad_norm": 0.6446427524928083,
      "learning_rate": 3.582012066720605e-06,
      "loss": 0.5283,
      "step": 8229
    },
    {
      "epoch": 0.603831726845016,
      "grad_norm": 1.4784337059964,
      "learning_rate": 3.5808726931987193e-06,
      "loss": 0.8377,
      "step": 8230
    },
    {
      "epoch": 0.603905096435155,
      "grad_norm": 1.5868195461146186,
      "learning_rate": 3.579733399818468e-06,
      "loss": 0.7285,
      "step": 8231
    },
    {
      "epoch": 0.6039784660252941,
      "grad_norm": 1.3918875176248702,
      "learning_rate": 3.5785941866441866e-06,
      "loss": 0.7573,
      "step": 8232
    },
    {
      "epoch": 0.6039784660252941,
      "lm_loss": 1.0359,
      "step": 8232,
      "vm_loss": 0.0764
    },
    {
      "epoch": 0.6039784660252941,
      "lm_loss": 0.4762,
      "step": 8232,
      "vm_loss": 0.1072
    },
    {
      "epoch": 0.6039784660252941,
      "lm_loss": 0.6604,
      "step": 8232,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.6039784660252941,
      "lm_loss": 0.3169,
      "step": 8232,
      "vm_loss": 0.203
    },
    {
      "epoch": 0.6039784660252941,
      "lm_loss": 0.0587,
      "step": 8232,
      "vm_loss": 0.1169
    },
    {
      "epoch": 0.6039784660252941,
      "lm_loss": 0.9648,
      "step": 8232,
      "vm_loss": 0.18
    },
    {
      "epoch": 0.6039784660252941,
      "lm_loss": 0.1254,
      "step": 8232,
      "vm_loss": 0.0683
    },
    {
      "epoch": 0.6039784660252941,
      "lm_loss": 0.4864,
      "step": 8232,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.6040518356154333,
      "grad_norm": 1.431644500364959,
      "learning_rate": 3.577455053740213e-06,
      "loss": 0.6488,
      "step": 8233
    },
    {
      "epoch": 0.6041252052055724,
      "grad_norm": 1.5898132926403683,
      "learning_rate": 3.576316001170875e-06,
      "loss": 0.7698,
      "step": 8234
    },
    {
      "epoch": 0.6041985747957116,
      "grad_norm": 1.6011542609046336,
      "learning_rate": 3.575177029000498e-06,
      "loss": 0.8697,
      "step": 8235
    },
    {
      "epoch": 0.6042719443858506,
      "grad_norm": 0.5723580660855818,
      "learning_rate": 3.574038137293404e-06,
      "loss": 0.4899,
      "step": 8236
    },
    {
      "epoch": 0.6043453139759898,
      "grad_norm": 1.4631793631342203,
      "learning_rate": 3.5728993261139056e-06,
      "loss": 0.7369,
      "step": 8237
    },
    {
      "epoch": 0.6044186835661289,
      "grad_norm": 0.5679598985511051,
      "learning_rate": 3.571760595526318e-06,
      "loss": 0.4959,
      "step": 8238
    },
    {
      "epoch": 0.6044920531562681,
      "grad_norm": 1.678092926582903,
      "learning_rate": 3.5706219455949462e-06,
      "loss": 0.7261,
      "step": 8239
    },
    {
      "epoch": 0.6045654227464072,
      "grad_norm": 1.8404967038953923,
      "learning_rate": 3.569483376384092e-06,
      "loss": 0.7599,
      "step": 8240
    },
    {
      "epoch": 0.6045654227464072,
      "lm_loss": 0.6708,
      "step": 8240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6045654227464072,
      "lm_loss": 0.5041,
      "step": 8240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6045654227464072,
      "lm_loss": 0.5266,
      "step": 8240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6045654227464072,
      "lm_loss": 0.3757,
      "step": 8240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6045654227464072,
      "lm_loss": 0.4288,
      "step": 8240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6045654227464072,
      "lm_loss": 0.5834,
      "step": 8240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6045654227464072,
      "lm_loss": 0.6135,
      "step": 8240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6045654227464072,
      "lm_loss": 0.5097,
      "step": 8240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6046387923365463,
      "grad_norm": 0.5633268003464481,
      "learning_rate": 3.5683448879580558e-06,
      "loss": 0.4883,
      "step": 8241
    },
    {
      "epoch": 0.6047121619266854,
      "grad_norm": 1.4403969434064532,
      "learning_rate": 3.5672064803811273e-06,
      "loss": 0.6972,
      "step": 8242
    },
    {
      "epoch": 0.6047855315168246,
      "grad_norm": 1.4408980283365465,
      "learning_rate": 3.566068153717599e-06,
      "loss": 0.7752,
      "step": 8243
    },
    {
      "epoch": 0.6048589011069637,
      "grad_norm": 0.5802506775975772,
      "learning_rate": 3.5649299080317523e-06,
      "loss": 0.4902,
      "step": 8244
    },
    {
      "epoch": 0.6049322706971029,
      "grad_norm": 1.476269873722977,
      "learning_rate": 3.5637917433878664e-06,
      "loss": 0.8438,
      "step": 8245
    },
    {
      "epoch": 0.6050056402872419,
      "grad_norm": 1.5345352344906404,
      "learning_rate": 3.5626536598502194e-06,
      "loss": 0.7052,
      "step": 8246
    },
    {
      "epoch": 0.6050790098773811,
      "grad_norm": 1.3913677018880144,
      "learning_rate": 3.5615156574830795e-06,
      "loss": 0.7485,
      "step": 8247
    },
    {
      "epoch": 0.6051523794675202,
      "grad_norm": 1.6303977768115663,
      "learning_rate": 3.5603777363507114e-06,
      "loss": 0.7729,
      "step": 8248
    },
    {
      "epoch": 0.6051523794675202,
      "lm_loss": 0.9635,
      "step": 8248,
      "vm_loss": 0.1146
    },
    {
      "epoch": 0.6051523794675202,
      "lm_loss": 0.6033,
      "step": 8248,
      "vm_loss": 0.0395
    },
    {
      "epoch": 0.6051523794675202,
      "lm_loss": 0.3867,
      "step": 8248,
      "vm_loss": 0.1594
    },
    {
      "epoch": 0.6051523794675202,
      "lm_loss": 0.5262,
      "step": 8248,
      "vm_loss": 0.2576
    },
    {
      "epoch": 0.6051523794675202,
      "lm_loss": 0.5324,
      "step": 8248,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.6051523794675202,
      "lm_loss": 0.1882,
      "step": 8248,
      "vm_loss": 0.1388
    },
    {
      "epoch": 0.6051523794675202,
      "lm_loss": 0.7376,
      "step": 8248,
      "vm_loss": 0.0936
    },
    {
      "epoch": 0.6051523794675202,
      "lm_loss": 0.0645,
      "step": 8248,
      "vm_loss": 0.165
    },
    {
      "epoch": 0.6052257490576594,
      "grad_norm": 1.354821029040143,
      "learning_rate": 3.559239896517379e-06,
      "loss": 0.7589,
      "step": 8249
    },
    {
      "epoch": 0.6052991186477985,
      "grad_norm": 1.4625504882478428,
      "learning_rate": 3.5581021380473362e-06,
      "loss": 0.7571,
      "step": 8250
    },
    {
      "epoch": 0.6053724882379375,
      "grad_norm": 0.5923385827413695,
      "learning_rate": 3.556964461004838e-06,
      "loss": 0.4995,
      "step": 8251
    },
    {
      "epoch": 0.6054458578280767,
      "grad_norm": 1.6287269471824992,
      "learning_rate": 3.55582686545413e-06,
      "loss": 0.8222,
      "step": 8252
    },
    {
      "epoch": 0.6055192274182158,
      "grad_norm": 1.549147084836811,
      "learning_rate": 3.5546893514594544e-06,
      "loss": 0.8093,
      "step": 8253
    },
    {
      "epoch": 0.605592597008355,
      "grad_norm": 0.5633616879823391,
      "learning_rate": 3.5535519190850513e-06,
      "loss": 0.482,
      "step": 8254
    },
    {
      "epoch": 0.6056659665984941,
      "grad_norm": 0.5621457472150523,
      "learning_rate": 3.552414568395154e-06,
      "loss": 0.4822,
      "step": 8255
    },
    {
      "epoch": 0.6057393361886332,
      "grad_norm": 1.4665418447384706,
      "learning_rate": 3.551277299453989e-06,
      "loss": 0.8663,
      "step": 8256
    },
    {
      "epoch": 0.6057393361886332,
      "lm_loss": 1.4988,
      "step": 8256,
      "vm_loss": 0.0591
    },
    {
      "epoch": 0.6057393361886332,
      "lm_loss": 0.0693,
      "step": 8256,
      "vm_loss": 0.1601
    },
    {
      "epoch": 0.6057393361886332,
      "lm_loss": 1.0465,
      "step": 8256,
      "vm_loss": 0.1176
    },
    {
      "epoch": 0.6057393361886332,
      "lm_loss": 0.5884,
      "step": 8256,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.6057393361886332,
      "lm_loss": 0.2781,
      "step": 8256,
      "vm_loss": 0.1806
    },
    {
      "epoch": 0.6057393361886332,
      "lm_loss": 0.8075,
      "step": 8256,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.6057393361886332,
      "lm_loss": 1.3743,
      "step": 8256,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.6057393361886332,
      "lm_loss": 0.6081,
      "step": 8256,
      "vm_loss": 0.0886
    },
    {
      "epoch": 0.6058127057787723,
      "grad_norm": 1.6396373794232582,
      "learning_rate": 3.550140112325784e-06,
      "loss": 0.9039,
      "step": 8257
    },
    {
      "epoch": 0.6058860753689115,
      "grad_norm": 1.4340332593260678,
      "learning_rate": 3.549003007074756e-06,
      "loss": 0.7313,
      "step": 8258
    },
    {
      "epoch": 0.6059594449590506,
      "grad_norm": 0.5665539875112123,
      "learning_rate": 3.547865983765123e-06,
      "loss": 0.5076,
      "step": 8259
    },
    {
      "epoch": 0.6060328145491898,
      "grad_norm": 1.5946374957202196,
      "learning_rate": 3.5467290424610946e-06,
      "loss": 0.6709,
      "step": 8260
    },
    {
      "epoch": 0.6061061841393288,
      "grad_norm": 1.4700104638357474,
      "learning_rate": 3.545592183226874e-06,
      "loss": 0.6922,
      "step": 8261
    },
    {
      "epoch": 0.606179553729468,
      "grad_norm": 1.722294062389912,
      "learning_rate": 3.544455406126666e-06,
      "loss": 0.7475,
      "step": 8262
    },
    {
      "epoch": 0.6062529233196071,
      "grad_norm": 1.5400317684660632,
      "learning_rate": 3.543318711224666e-06,
      "loss": 0.8098,
      "step": 8263
    },
    {
      "epoch": 0.6063262929097463,
      "grad_norm": 1.3871181028227113,
      "learning_rate": 3.5421820985850645e-06,
      "loss": 0.7482,
      "step": 8264
    },
    {
      "epoch": 0.6063262929097463,
      "lm_loss": 0.1187,
      "step": 8264,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.6063262929097463,
      "lm_loss": 0.1465,
      "step": 8264,
      "vm_loss": 0.0587
    },
    {
      "epoch": 0.6063262929097463,
      "lm_loss": 1.4484,
      "step": 8264,
      "vm_loss": 0.0471
    },
    {
      "epoch": 0.6063262929097463,
      "lm_loss": 0.9129,
      "step": 8264,
      "vm_loss": 0.1329
    },
    {
      "epoch": 0.6063262929097463,
      "lm_loss": 0.2506,
      "step": 8264,
      "vm_loss": 0.1895
    },
    {
      "epoch": 0.6063262929097463,
      "lm_loss": 0.1135,
      "step": 8264,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.6063262929097463,
      "lm_loss": 0.5448,
      "step": 8264,
      "vm_loss": 0.1147
    },
    {
      "epoch": 0.6063262929097463,
      "lm_loss": 0.66,
      "step": 8264,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.6063996624998853,
      "grad_norm": 1.4584386844987125,
      "learning_rate": 3.541045568272051e-06,
      "loss": 0.6664,
      "step": 8265
    },
    {
      "epoch": 0.6064730320900245,
      "grad_norm": 1.44994917475785,
      "learning_rate": 3.5399091203498083e-06,
      "loss": 0.7948,
      "step": 8266
    },
    {
      "epoch": 0.6065464016801636,
      "grad_norm": 1.5421040641017523,
      "learning_rate": 3.5387727548825113e-06,
      "loss": 0.7267,
      "step": 8267
    },
    {
      "epoch": 0.6066197712703028,
      "grad_norm": 1.3779169921311762,
      "learning_rate": 3.537636471934337e-06,
      "loss": 0.7984,
      "step": 8268
    },
    {
      "epoch": 0.6066931408604419,
      "grad_norm": 1.5820797405365112,
      "learning_rate": 3.536500271569452e-06,
      "loss": 0.7068,
      "step": 8269
    },
    {
      "epoch": 0.6067665104505809,
      "grad_norm": 1.551217255864601,
      "learning_rate": 3.535364153852021e-06,
      "loss": 0.7206,
      "step": 8270
    },
    {
      "epoch": 0.6068398800407201,
      "grad_norm": 0.5885570911262901,
      "learning_rate": 3.5342281188462042e-06,
      "loss": 0.5006,
      "step": 8271
    },
    {
      "epoch": 0.6069132496308592,
      "grad_norm": 1.6786482047554956,
      "learning_rate": 3.5330921666161536e-06,
      "loss": 0.8194,
      "step": 8272
    },
    {
      "epoch": 0.6069132496308592,
      "lm_loss": 0.6469,
      "step": 8272,
      "vm_loss": 0.0501
    },
    {
      "epoch": 0.6069132496308592,
      "lm_loss": 0.9857,
      "step": 8272,
      "vm_loss": 0.1882
    },
    {
      "epoch": 0.6069132496308592,
      "lm_loss": 0.4311,
      "step": 8272,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.6069132496308592,
      "lm_loss": 0.3224,
      "step": 8272,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.6069132496308592,
      "lm_loss": 0.4777,
      "step": 8272,
      "vm_loss": 0.153
    },
    {
      "epoch": 0.6069132496308592,
      "lm_loss": 1.1589,
      "step": 8272,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.6069132496308592,
      "lm_loss": 1.0891,
      "step": 8272,
      "vm_loss": 0.2312
    },
    {
      "epoch": 0.6069132496308592,
      "lm_loss": 0.8414,
      "step": 8272,
      "vm_loss": 0.1258
    },
    {
      "epoch": 0.6069866192209984,
      "grad_norm": 1.5065655672025402,
      "learning_rate": 3.531956297226023e-06,
      "loss": 0.7431,
      "step": 8273
    },
    {
      "epoch": 0.6070599888111375,
      "grad_norm": 1.5145847284108656,
      "learning_rate": 3.530820510739955e-06,
      "loss": 0.8199,
      "step": 8274
    },
    {
      "epoch": 0.6071333584012766,
      "grad_norm": 1.442809676766812,
      "learning_rate": 3.5296848072220907e-06,
      "loss": 0.7113,
      "step": 8275
    },
    {
      "epoch": 0.6072067279914157,
      "grad_norm": 1.540789763392788,
      "learning_rate": 3.5285491867365672e-06,
      "loss": 0.8142,
      "step": 8276
    },
    {
      "epoch": 0.6072800975815549,
      "grad_norm": 0.5780496705758957,
      "learning_rate": 3.527413649347514e-06,
      "loss": 0.4876,
      "step": 8277
    },
    {
      "epoch": 0.607353467171694,
      "grad_norm": 0.571716412774758,
      "learning_rate": 3.5262781951190604e-06,
      "loss": 0.495,
      "step": 8278
    },
    {
      "epoch": 0.6074268367618332,
      "grad_norm": 0.5644833012461378,
      "learning_rate": 3.5251428241153263e-06,
      "loss": 0.4892,
      "step": 8279
    },
    {
      "epoch": 0.6075002063519722,
      "grad_norm": 0.541251351595583,
      "learning_rate": 3.5240075364004288e-06,
      "loss": 0.4869,
      "step": 8280
    },
    {
      "epoch": 0.6075002063519722,
      "lm_loss": 0.7422,
      "step": 8280,
      "vm_loss": 0.0343
    },
    {
      "epoch": 0.6075002063519722,
      "lm_loss": 1.3404,
      "step": 8280,
      "vm_loss": 0.1489
    },
    {
      "epoch": 0.6075002063519722,
      "lm_loss": 1.105,
      "step": 8280,
      "vm_loss": 0.171
    },
    {
      "epoch": 0.6075002063519722,
      "lm_loss": 0.7921,
      "step": 8280,
      "vm_loss": 0.2142
    },
    {
      "epoch": 0.6075002063519722,
      "lm_loss": 0.6059,
      "step": 8280,
      "vm_loss": 0.0623
    },
    {
      "epoch": 0.6075002063519722,
      "lm_loss": 0.6882,
      "step": 8280,
      "vm_loss": 0.3281
    },
    {
      "epoch": 0.6075002063519722,
      "lm_loss": 0.8494,
      "step": 8280,
      "vm_loss": 0.076
    },
    {
      "epoch": 0.6075002063519722,
      "lm_loss": 0.2593,
      "step": 8280,
      "vm_loss": 0.0736
    },
    {
      "epoch": 0.6075735759421114,
      "grad_norm": 1.5325870187803003,
      "learning_rate": 3.5228723320384817e-06,
      "loss": 0.818,
      "step": 8281
    },
    {
      "epoch": 0.6076469455322505,
      "grad_norm": 1.4748234997708398,
      "learning_rate": 3.521737211093592e-06,
      "loss": 0.7696,
      "step": 8282
    },
    {
      "epoch": 0.6077203151223897,
      "grad_norm": 1.7048862070036808,
      "learning_rate": 3.5206021736298613e-06,
      "loss": 0.8275,
      "step": 8283
    },
    {
      "epoch": 0.6077936847125288,
      "grad_norm": 1.516275124114144,
      "learning_rate": 3.519467219711391e-06,
      "loss": 0.8043,
      "step": 8284
    },
    {
      "epoch": 0.6078670543026679,
      "grad_norm": 1.61318700308436,
      "learning_rate": 3.5183323494022727e-06,
      "loss": 0.7812,
      "step": 8285
    },
    {
      "epoch": 0.607940423892807,
      "grad_norm": 1.4905532945643183,
      "learning_rate": 3.517197562766597e-06,
      "loss": 0.6975,
      "step": 8286
    },
    {
      "epoch": 0.6080137934829462,
      "grad_norm": 1.5873344552901247,
      "learning_rate": 3.516062859868446e-06,
      "loss": 0.7972,
      "step": 8287
    },
    {
      "epoch": 0.6080871630730853,
      "grad_norm": 1.5134552460128812,
      "learning_rate": 3.5149282407719e-06,
      "loss": 0.7673,
      "step": 8288
    },
    {
      "epoch": 0.6080871630730853,
      "lm_loss": 1.3057,
      "step": 8288,
      "vm_loss": 0.044
    },
    {
      "epoch": 0.6080871630730853,
      "lm_loss": 0.7612,
      "step": 8288,
      "vm_loss": 0.1895
    },
    {
      "epoch": 0.6080871630730853,
      "lm_loss": 0.1867,
      "step": 8288,
      "vm_loss": 0.1752
    },
    {
      "epoch": 0.6080871630730853,
      "lm_loss": 0.7745,
      "step": 8288,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.6080871630730853,
      "lm_loss": 1.2484,
      "step": 8288,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.6080871630730853,
      "lm_loss": 0.3969,
      "step": 8288,
      "vm_loss": 0.081
    },
    {
      "epoch": 0.6080871630730853,
      "lm_loss": 1.0146,
      "step": 8288,
      "vm_loss": 0.0949
    },
    {
      "epoch": 0.6080871630730853,
      "lm_loss": 0.4639,
      "step": 8288,
      "vm_loss": 0.0585
    },
    {
      "epoch": 0.6081605326632245,
      "grad_norm": 1.4659041906729053,
      "learning_rate": 3.5137937055410343e-06,
      "loss": 0.7921,
      "step": 8289
    },
    {
      "epoch": 0.6082339022533635,
      "grad_norm": 1.3713000785707377,
      "learning_rate": 3.512659254239919e-06,
      "loss": 0.6807,
      "step": 8290
    },
    {
      "epoch": 0.6083072718435026,
      "grad_norm": 0.5833279337991415,
      "learning_rate": 3.511524886932618e-06,
      "loss": 0.513,
      "step": 8291
    },
    {
      "epoch": 0.6083806414336418,
      "grad_norm": 1.5859753502921288,
      "learning_rate": 3.5103906036831937e-06,
      "loss": 0.7767,
      "step": 8292
    },
    {
      "epoch": 0.6084540110237809,
      "grad_norm": 1.4604981398099042,
      "learning_rate": 3.509256404555701e-06,
      "loss": 0.7367,
      "step": 8293
    },
    {
      "epoch": 0.6085273806139201,
      "grad_norm": 1.58482594980553,
      "learning_rate": 3.5081222896141897e-06,
      "loss": 0.7822,
      "step": 8294
    },
    {
      "epoch": 0.6086007502040591,
      "grad_norm": 1.3467769179264046,
      "learning_rate": 3.506988258922709e-06,
      "loss": 0.7703,
      "step": 8295
    },
    {
      "epoch": 0.6086741197941983,
      "grad_norm": 0.5949140866953762,
      "learning_rate": 3.505854312545299e-06,
      "loss": 0.484,
      "step": 8296
    },
    {
      "epoch": 0.6086741197941983,
      "lm_loss": 1.634,
      "step": 8296,
      "vm_loss": 0.0483
    },
    {
      "epoch": 0.6086741197941983,
      "lm_loss": 0.4842,
      "step": 8296,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.6086741197941983,
      "lm_loss": 0.9019,
      "step": 8296,
      "vm_loss": 0.2223
    },
    {
      "epoch": 0.6086741197941983,
      "lm_loss": 0.458,
      "step": 8296,
      "vm_loss": 0.0949
    },
    {
      "epoch": 0.6086741197941983,
      "lm_loss": 0.9475,
      "step": 8296,
      "vm_loss": 0.1243
    },
    {
      "epoch": 0.6086741197941983,
      "lm_loss": 0.1433,
      "step": 8296,
      "vm_loss": 0.0808
    },
    {
      "epoch": 0.6086741197941983,
      "lm_loss": 0.5728,
      "step": 8296,
      "vm_loss": 0.1678
    },
    {
      "epoch": 0.6086741197941983,
      "lm_loss": 0.3664,
      "step": 8296,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.6087474893843374,
      "grad_norm": 1.3660264353936216,
      "learning_rate": 3.5047204505459953e-06,
      "loss": 0.7684,
      "step": 8297
    },
    {
      "epoch": 0.6088208589744766,
      "grad_norm": 0.5960464473786662,
      "learning_rate": 3.5035866729888302e-06,
      "loss": 0.487,
      "step": 8298
    },
    {
      "epoch": 0.6088942285646157,
      "grad_norm": 1.3541990832382114,
      "learning_rate": 3.5024529799378316e-06,
      "loss": 0.7135,
      "step": 8299
    },
    {
      "epoch": 0.6089675981547548,
      "grad_norm": 1.477491487334541,
      "learning_rate": 3.5013193714570225e-06,
      "loss": 0.7423,
      "step": 8300
    },
    {
      "epoch": 0.6090409677448939,
      "grad_norm": 1.5089965600334756,
      "learning_rate": 3.500185847610419e-06,
      "loss": 0.8157,
      "step": 8301
    },
    {
      "epoch": 0.6091143373350331,
      "grad_norm": 1.4071936461432362,
      "learning_rate": 3.499052408462035e-06,
      "loss": 0.74,
      "step": 8302
    },
    {
      "epoch": 0.6091877069251722,
      "grad_norm": 1.3756672329123942,
      "learning_rate": 3.4979190540758784e-06,
      "loss": 0.7762,
      "step": 8303
    },
    {
      "epoch": 0.6092610765153114,
      "grad_norm": 1.8211595636624265,
      "learning_rate": 3.496785784515954e-06,
      "loss": 0.7569,
      "step": 8304
    },
    {
      "epoch": 0.6092610765153114,
      "lm_loss": 0.4898,
      "step": 8304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6092610765153114,
      "lm_loss": 0.3927,
      "step": 8304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6092610765153114,
      "lm_loss": 0.4816,
      "step": 8304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6092610765153114,
      "lm_loss": 0.4974,
      "step": 8304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6092610765153114,
      "lm_loss": 0.4844,
      "step": 8304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6092610765153114,
      "lm_loss": 0.4274,
      "step": 8304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6092610765153114,
      "lm_loss": 0.5678,
      "step": 8304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6092610765153114,
      "lm_loss": 0.4513,
      "step": 8304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6093344461054504,
      "grad_norm": 0.5325239560630253,
      "learning_rate": 3.4956525998462574e-06,
      "loss": 0.4737,
      "step": 8305
    },
    {
      "epoch": 0.6094078156955895,
      "grad_norm": 1.5115845411797046,
      "learning_rate": 3.494519500130784e-06,
      "loss": 0.8215,
      "step": 8306
    },
    {
      "epoch": 0.6094811852857287,
      "grad_norm": 1.2571753263094092,
      "learning_rate": 3.493386485433524e-06,
      "loss": 0.7161,
      "step": 8307
    },
    {
      "epoch": 0.6095545548758678,
      "grad_norm": 1.4572674742698886,
      "learning_rate": 3.49225355581846e-06,
      "loss": 0.6891,
      "step": 8308
    },
    {
      "epoch": 0.609627924466007,
      "grad_norm": 1.533971253385178,
      "learning_rate": 3.4911207113495703e-06,
      "loss": 0.8352,
      "step": 8309
    },
    {
      "epoch": 0.609701294056146,
      "grad_norm": 1.6275295984144593,
      "learning_rate": 3.489987952090832e-06,
      "loss": 0.7509,
      "step": 8310
    },
    {
      "epoch": 0.6097746636462852,
      "grad_norm": 1.5255607317305921,
      "learning_rate": 3.4888552781062124e-06,
      "loss": 0.7478,
      "step": 8311
    },
    {
      "epoch": 0.6098480332364243,
      "grad_norm": 1.4715454511494273,
      "learning_rate": 3.487722689459678e-06,
      "loss": 0.6606,
      "step": 8312
    },
    {
      "epoch": 0.6098480332364243,
      "lm_loss": 0.9049,
      "step": 8312,
      "vm_loss": 0.0541
    },
    {
      "epoch": 0.6098480332364243,
      "lm_loss": 0.276,
      "step": 8312,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.6098480332364243,
      "lm_loss": 0.3464,
      "step": 8312,
      "vm_loss": 0.1423
    },
    {
      "epoch": 0.6098480332364243,
      "lm_loss": 0.9541,
      "step": 8312,
      "vm_loss": 0.082
    },
    {
      "epoch": 0.6098480332364243,
      "lm_loss": 0.9503,
      "step": 8312,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.6098480332364243,
      "lm_loss": 0.4658,
      "step": 8312,
      "vm_loss": 0.1814
    },
    {
      "epoch": 0.6098480332364243,
      "lm_loss": 1.3336,
      "step": 8312,
      "vm_loss": 0.1754
    },
    {
      "epoch": 0.6098480332364243,
      "lm_loss": 0.6325,
      "step": 8312,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.6099214028265635,
      "grad_norm": 1.7350963119916878,
      "learning_rate": 3.4865901862151895e-06,
      "loss": 0.752,
      "step": 8313
    },
    {
      "epoch": 0.6099947724167026,
      "grad_norm": 1.592638430653466,
      "learning_rate": 3.4854577684366997e-06,
      "loss": 0.8071,
      "step": 8314
    },
    {
      "epoch": 0.6100681420068417,
      "grad_norm": 1.5006549437159808,
      "learning_rate": 3.4843254361881605e-06,
      "loss": 0.9034,
      "step": 8315
    },
    {
      "epoch": 0.6101415115969808,
      "grad_norm": 1.517826831158295,
      "learning_rate": 3.4831931895335187e-06,
      "loss": 0.7078,
      "step": 8316
    },
    {
      "epoch": 0.61021488118712,
      "grad_norm": 1.595601628410689,
      "learning_rate": 3.4820610285367116e-06,
      "loss": 0.8244,
      "step": 8317
    },
    {
      "epoch": 0.6102882507772591,
      "grad_norm": 1.592345040267182,
      "learning_rate": 3.4809289532616797e-06,
      "loss": 0.6472,
      "step": 8318
    },
    {
      "epoch": 0.6103616203673983,
      "grad_norm": 1.517646452065176,
      "learning_rate": 3.4797969637723494e-06,
      "loss": 0.7516,
      "step": 8319
    },
    {
      "epoch": 0.6104349899575373,
      "grad_norm": 0.56725425205069,
      "learning_rate": 3.478665060132651e-06,
      "loss": 0.4836,
      "step": 8320
    },
    {
      "epoch": 0.6104349899575373,
      "lm_loss": 1.2654,
      "step": 8320,
      "vm_loss": 0.081
    },
    {
      "epoch": 0.6104349899575373,
      "lm_loss": 0.0123,
      "step": 8320,
      "vm_loss": 0.205
    },
    {
      "epoch": 0.6104349899575373,
      "lm_loss": 1.1534,
      "step": 8320,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.6104349899575373,
      "lm_loss": 1.421,
      "step": 8320,
      "vm_loss": 0.2151
    },
    {
      "epoch": 0.6104349899575373,
      "lm_loss": 0.2002,
      "step": 8320,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.6104349899575373,
      "lm_loss": 0.6709,
      "step": 8320,
      "vm_loss": 0.1146
    },
    {
      "epoch": 0.6104349899575373,
      "lm_loss": 0.0752,
      "step": 8320,
      "vm_loss": 0.0606
    },
    {
      "epoch": 0.6104349899575373,
      "lm_loss": 1.6023,
      "step": 8320,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.6105083595476765,
      "grad_norm": 1.615456110654775,
      "learning_rate": 3.4775332424065045e-06,
      "loss": 0.7532,
      "step": 8321
    },
    {
      "epoch": 0.6105817291378156,
      "grad_norm": 1.5512162841393138,
      "learning_rate": 3.476401510657824e-06,
      "loss": 0.7808,
      "step": 8322
    },
    {
      "epoch": 0.6106550987279548,
      "grad_norm": 1.330382572813642,
      "learning_rate": 3.4752698649505255e-06,
      "loss": 0.7243,
      "step": 8323
    },
    {
      "epoch": 0.6107284683180939,
      "grad_norm": 1.6088302624353803,
      "learning_rate": 3.4741383053485132e-06,
      "loss": 0.699,
      "step": 8324
    },
    {
      "epoch": 0.610801837908233,
      "grad_norm": 1.6202989514628485,
      "learning_rate": 3.4730068319156883e-06,
      "loss": 0.7301,
      "step": 8325
    },
    {
      "epoch": 0.6108752074983721,
      "grad_norm": 1.4142508873425057,
      "learning_rate": 3.471875444715951e-06,
      "loss": 0.7592,
      "step": 8326
    },
    {
      "epoch": 0.6109485770885112,
      "grad_norm": 0.5839555933543689,
      "learning_rate": 3.4707441438131904e-06,
      "loss": 0.486,
      "step": 8327
    },
    {
      "epoch": 0.6110219466786504,
      "grad_norm": 1.5302344127714342,
      "learning_rate": 3.469612929271295e-06,
      "loss": 0.7703,
      "step": 8328
    },
    {
      "epoch": 0.6110219466786504,
      "lm_loss": 1.0474,
      "step": 8328,
      "vm_loss": 0.0874
    },
    {
      "epoch": 0.6110219466786504,
      "lm_loss": 0.1592,
      "step": 8328,
      "vm_loss": 0.0861
    },
    {
      "epoch": 0.6110219466786504,
      "lm_loss": 0.3733,
      "step": 8328,
      "vm_loss": 0.0861
    },
    {
      "epoch": 0.6110219466786504,
      "lm_loss": 0.6184,
      "step": 8328,
      "vm_loss": 0.1317
    },
    {
      "epoch": 0.6110219466786504,
      "lm_loss": 0.9007,
      "step": 8328,
      "vm_loss": 0.1384
    },
    {
      "epoch": 0.6110219466786504,
      "lm_loss": 0.433,
      "step": 8328,
      "vm_loss": 0.0513
    },
    {
      "epoch": 0.6110219466786504,
      "lm_loss": 0.94,
      "step": 8328,
      "vm_loss": 0.1908
    },
    {
      "epoch": 0.6110219466786504,
      "lm_loss": 0.8766,
      "step": 8328,
      "vm_loss": 0.1826
    },
    {
      "epoch": 0.6110953162687895,
      "grad_norm": 1.4065357900225435,
      "learning_rate": 3.4684818011541484e-06,
      "loss": 0.8489,
      "step": 8329
    },
    {
      "epoch": 0.6111686858589286,
      "grad_norm": 1.6414839363235263,
      "learning_rate": 3.4673507595256257e-06,
      "loss": 0.8544,
      "step": 8330
    },
    {
      "epoch": 0.6112420554490677,
      "grad_norm": 1.5418831337605592,
      "learning_rate": 3.466219804449602e-06,
      "loss": 0.7933,
      "step": 8331
    },
    {
      "epoch": 0.6113154250392069,
      "grad_norm": 1.6493388410876413,
      "learning_rate": 3.4650889359899455e-06,
      "loss": 0.8328,
      "step": 8332
    },
    {
      "epoch": 0.611388794629346,
      "grad_norm": 1.5849907536322614,
      "learning_rate": 3.463958154210516e-06,
      "loss": 0.7477,
      "step": 8333
    },
    {
      "epoch": 0.6114621642194852,
      "grad_norm": 1.4118440299837887,
      "learning_rate": 3.4628274591751743e-06,
      "loss": 0.7149,
      "step": 8334
    },
    {
      "epoch": 0.6115355338096242,
      "grad_norm": 0.5450230842015873,
      "learning_rate": 3.461696850947773e-06,
      "loss": 0.4872,
      "step": 8335
    },
    {
      "epoch": 0.6116089033997634,
      "grad_norm": 1.7001453782136282,
      "learning_rate": 3.460566329592159e-06,
      "loss": 0.6312,
      "step": 8336
    },
    {
      "epoch": 0.6116089033997634,
      "lm_loss": 0.2334,
      "step": 8336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6116089033997634,
      "lm_loss": 0.4037,
      "step": 8336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6116089033997634,
      "lm_loss": 0.4111,
      "step": 8336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6116089033997634,
      "lm_loss": 0.3853,
      "step": 8336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6116089033997634,
      "lm_loss": 0.4966,
      "step": 8336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6116089033997634,
      "lm_loss": 0.3925,
      "step": 8336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6116089033997634,
      "lm_loss": 0.6238,
      "step": 8336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6116089033997634,
      "lm_loss": 0.3878,
      "step": 8336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6116822729899025,
      "grad_norm": 0.5603241706777142,
      "learning_rate": 3.4594358951721787e-06,
      "loss": 0.4864,
      "step": 8337
    },
    {
      "epoch": 0.6117556425800417,
      "grad_norm": 0.5473370457200719,
      "learning_rate": 3.458305547751666e-06,
      "loss": 0.4659,
      "step": 8338
    },
    {
      "epoch": 0.6118290121701808,
      "grad_norm": 1.6136797432963388,
      "learning_rate": 3.4571752873944585e-06,
      "loss": 0.7956,
      "step": 8339
    },
    {
      "epoch": 0.6119023817603199,
      "grad_norm": 1.6412003099139758,
      "learning_rate": 3.4560451141643837e-06,
      "loss": 0.7349,
      "step": 8340
    },
    {
      "epoch": 0.611975751350459,
      "grad_norm": 1.8408050364129815,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.745,
      "step": 8341
    },
    {
      "epoch": 0.6120491209405982,
      "grad_norm": 1.4475982359798913,
      "learning_rate": 3.453785029340919e-06,
      "loss": 0.7716,
      "step": 8342
    },
    {
      "epoch": 0.6121224905307373,
      "grad_norm": 1.4178139998282318,
      "learning_rate": 3.4526551178751635e-06,
      "loss": 0.74,
      "step": 8343
    },
    {
      "epoch": 0.6121958601208763,
      "grad_norm": 0.5749619838634243,
      "learning_rate": 3.4515252937918035e-06,
      "loss": 0.4739,
      "step": 8344
    },
    {
      "epoch": 0.6121958601208763,
      "lm_loss": 1.3722,
      "step": 8344,
      "vm_loss": 0.0386
    },
    {
      "epoch": 0.6121958601208763,
      "lm_loss": 0.0604,
      "step": 8344,
      "vm_loss": 0.1596
    },
    {
      "epoch": 0.6121958601208763,
      "lm_loss": 0.0187,
      "step": 8344,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.6121958601208763,
      "lm_loss": 0.5393,
      "step": 8344,
      "vm_loss": 0.178
    },
    {
      "epoch": 0.6121958601208763,
      "lm_loss": 0.4325,
      "step": 8344,
      "vm_loss": 0.0761
    },
    {
      "epoch": 0.6121958601208763,
      "lm_loss": 0.8856,
      "step": 8344,
      "vm_loss": 0.1434
    },
    {
      "epoch": 0.6121958601208763,
      "lm_loss": 0.2253,
      "step": 8344,
      "vm_loss": 0.1491
    },
    {
      "epoch": 0.6121958601208763,
      "lm_loss": 1.0495,
      "step": 8344,
      "vm_loss": 0.0897
    },
    {
      "epoch": 0.6122692297110155,
      "grad_norm": 1.5429111493080339,
      "learning_rate": 3.4503955571546465e-06,
      "loss": 0.7793,
      "step": 8345
    },
    {
      "epoch": 0.6123425993011546,
      "grad_norm": 1.5679667800927024,
      "learning_rate": 3.449265908027489e-06,
      "loss": 0.7028,
      "step": 8346
    },
    {
      "epoch": 0.6124159688912938,
      "grad_norm": 1.4581722187377282,
      "learning_rate": 3.4481363464741267e-06,
      "loss": 0.7528,
      "step": 8347
    },
    {
      "epoch": 0.6124893384814329,
      "grad_norm": 0.5646825519225992,
      "learning_rate": 3.447006872558348e-06,
      "loss": 0.4824,
      "step": 8348
    },
    {
      "epoch": 0.612562708071572,
      "grad_norm": 1.8094714365157167,
      "learning_rate": 3.4458774863439366e-06,
      "loss": 0.8857,
      "step": 8349
    },
    {
      "epoch": 0.6126360776617111,
      "grad_norm": 1.4263136340570124,
      "learning_rate": 3.4447481878946733e-06,
      "loss": 0.8281,
      "step": 8350
    },
    {
      "epoch": 0.6127094472518503,
      "grad_norm": 1.6281847980844635,
      "learning_rate": 3.4436189772743307e-06,
      "loss": 0.7165,
      "step": 8351
    },
    {
      "epoch": 0.6127828168419894,
      "grad_norm": 1.493918311919377,
      "learning_rate": 3.442489854546678e-06,
      "loss": 0.8316,
      "step": 8352
    },
    {
      "epoch": 0.6127828168419894,
      "lm_loss": 0.5123,
      "step": 8352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6127828168419894,
      "lm_loss": 0.591,
      "step": 8352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6127828168419894,
      "lm_loss": 0.5108,
      "step": 8352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6127828168419894,
      "lm_loss": 0.5449,
      "step": 8352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6127828168419894,
      "lm_loss": 0.5923,
      "step": 8352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6127828168419894,
      "lm_loss": 0.4783,
      "step": 8352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6127828168419894,
      "lm_loss": 0.5845,
      "step": 8352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6127828168419894,
      "lm_loss": 0.5146,
      "step": 8352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6128561864321286,
      "grad_norm": 0.5683278449108633,
      "learning_rate": 3.441360819775481e-06,
      "loss": 0.4927,
      "step": 8353
    },
    {
      "epoch": 0.6129295560222676,
      "grad_norm": 0.5880947457270892,
      "learning_rate": 3.4402318730244975e-06,
      "loss": 0.4918,
      "step": 8354
    },
    {
      "epoch": 0.6130029256124068,
      "grad_norm": 1.7171101035655538,
      "learning_rate": 3.4391030143574843e-06,
      "loss": 0.719,
      "step": 8355
    },
    {
      "epoch": 0.6130762952025459,
      "grad_norm": 1.758475821929073,
      "learning_rate": 3.4379742438381897e-06,
      "loss": 0.7396,
      "step": 8356
    },
    {
      "epoch": 0.6131496647926851,
      "grad_norm": 1.5066378083642904,
      "learning_rate": 3.4368455615303564e-06,
      "loss": 0.7212,
      "step": 8357
    },
    {
      "epoch": 0.6132230343828242,
      "grad_norm": 0.551938670180791,
      "learning_rate": 3.4357169674977265e-06,
      "loss": 0.4887,
      "step": 8358
    },
    {
      "epoch": 0.6132964039729633,
      "grad_norm": 1.738815620891874,
      "learning_rate": 3.4345884618040326e-06,
      "loss": 0.7657,
      "step": 8359
    },
    {
      "epoch": 0.6133697735631024,
      "grad_norm": 1.5739055401289561,
      "learning_rate": 3.4334600445130044e-06,
      "loss": 0.7439,
      "step": 8360
    },
    {
      "epoch": 0.6133697735631024,
      "lm_loss": 0.4497,
      "step": 8360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6133697735631024,
      "lm_loss": 0.3803,
      "step": 8360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6133697735631024,
      "lm_loss": 0.2979,
      "step": 8360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6133697735631024,
      "lm_loss": 0.5047,
      "step": 8360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6133697735631024,
      "lm_loss": 0.5699,
      "step": 8360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6133697735631024,
      "lm_loss": 0.5424,
      "step": 8360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6133697735631024,
      "lm_loss": 0.3888,
      "step": 8360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6133697735631024,
      "lm_loss": 0.5886,
      "step": 8360,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6134431431532416,
      "grad_norm": 0.554945198809737,
      "learning_rate": 3.4323317156883685e-06,
      "loss": 0.4909,
      "step": 8361
    },
    {
      "epoch": 0.6135165127433807,
      "grad_norm": 0.5629380586926991,
      "learning_rate": 3.431203475393842e-06,
      "loss": 0.472,
      "step": 8362
    },
    {
      "epoch": 0.6135898823335199,
      "grad_norm": 1.618176491466059,
      "learning_rate": 3.4300753236931393e-06,
      "loss": 0.7907,
      "step": 8363
    },
    {
      "epoch": 0.6136632519236589,
      "grad_norm": 0.5497470419830789,
      "learning_rate": 3.428947260649973e-06,
      "loss": 0.4671,
      "step": 8364
    },
    {
      "epoch": 0.613736621513798,
      "grad_norm": 1.4733870792419255,
      "learning_rate": 3.4278192863280435e-06,
      "loss": 0.7052,
      "step": 8365
    },
    {
      "epoch": 0.6138099911039372,
      "grad_norm": 0.5666163891417213,
      "learning_rate": 3.4266914007910533e-06,
      "loss": 0.485,
      "step": 8366
    },
    {
      "epoch": 0.6138833606940763,
      "grad_norm": 1.3563475099182059,
      "learning_rate": 3.425563604102697e-06,
      "loss": 0.7203,
      "step": 8367
    },
    {
      "epoch": 0.6139567302842155,
      "grad_norm": 1.5309342369976693,
      "learning_rate": 3.424435896326661e-06,
      "loss": 0.6562,
      "step": 8368
    },
    {
      "epoch": 0.6139567302842155,
      "lm_loss": 0.6064,
      "step": 8368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6139567302842155,
      "lm_loss": 0.6567,
      "step": 8368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6139567302842155,
      "lm_loss": 0.3429,
      "step": 8368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6139567302842155,
      "lm_loss": 0.4909,
      "step": 8368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6139567302842155,
      "lm_loss": 0.3854,
      "step": 8368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6139567302842155,
      "lm_loss": 0.4979,
      "step": 8368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6139567302842155,
      "lm_loss": 0.5399,
      "step": 8368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6139567302842155,
      "lm_loss": 0.5918,
      "step": 8368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6140300998743545,
      "grad_norm": 0.5490363917169787,
      "learning_rate": 3.423308277526633e-06,
      "loss": 0.4743,
      "step": 8369
    },
    {
      "epoch": 0.6141034694644937,
      "grad_norm": 1.6602142901409038,
      "learning_rate": 3.4221807477662914e-06,
      "loss": 0.8159,
      "step": 8370
    },
    {
      "epoch": 0.6141768390546328,
      "grad_norm": 1.5945127180334102,
      "learning_rate": 3.4210533071093086e-06,
      "loss": 0.7304,
      "step": 8371
    },
    {
      "epoch": 0.614250208644772,
      "grad_norm": 1.499664359413628,
      "learning_rate": 3.4199259556193576e-06,
      "loss": 0.6419,
      "step": 8372
    },
    {
      "epoch": 0.6143235782349111,
      "grad_norm": 1.525552210450253,
      "learning_rate": 3.4187986933601004e-06,
      "loss": 0.7328,
      "step": 8373
    },
    {
      "epoch": 0.6143969478250502,
      "grad_norm": 1.5592101650261017,
      "learning_rate": 3.4176715203951967e-06,
      "loss": 0.677,
      "step": 8374
    },
    {
      "epoch": 0.6144703174151893,
      "grad_norm": 1.6060812857054465,
      "learning_rate": 3.416544436788301e-06,
      "loss": 0.7993,
      "step": 8375
    },
    {
      "epoch": 0.6145436870053285,
      "grad_norm": 1.4506441310530875,
      "learning_rate": 3.415417442603062e-06,
      "loss": 0.7308,
      "step": 8376
    },
    {
      "epoch": 0.6145436870053285,
      "lm_loss": 1.034,
      "step": 8376,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.6145436870053285,
      "lm_loss": 1.2954,
      "step": 8376,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.6145436870053285,
      "lm_loss": 0.4115,
      "step": 8376,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.6145436870053285,
      "lm_loss": 0.7041,
      "step": 8376,
      "vm_loss": 0.1108
    },
    {
      "epoch": 0.6145436870053285,
      "lm_loss": 0.0563,
      "step": 8376,
      "vm_loss": 0.1761
    },
    {
      "epoch": 0.6145436870053285,
      "lm_loss": 0.3598,
      "step": 8376,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.6145436870053285,
      "lm_loss": 0.364,
      "step": 8376,
      "vm_loss": 0.178
    },
    {
      "epoch": 0.6145436870053285,
      "lm_loss": 1.0738,
      "step": 8376,
      "vm_loss": 0.1595
    },
    {
      "epoch": 0.6146170565954676,
      "grad_norm": 1.6761956822155495,
      "learning_rate": 3.4142905379031243e-06,
      "loss": 0.8055,
      "step": 8377
    },
    {
      "epoch": 0.6146904261856068,
      "grad_norm": 0.5961556435003157,
      "learning_rate": 3.413163722752127e-06,
      "loss": 0.4949,
      "step": 8378
    },
    {
      "epoch": 0.6147637957757458,
      "grad_norm": 1.5192323930551275,
      "learning_rate": 3.412036997213703e-06,
      "loss": 0.7532,
      "step": 8379
    },
    {
      "epoch": 0.614837165365885,
      "grad_norm": 1.5823795746767466,
      "learning_rate": 3.410910361351484e-06,
      "loss": 0.9433,
      "step": 8380
    },
    {
      "epoch": 0.6149105349560241,
      "grad_norm": 1.5901770585560033,
      "learning_rate": 3.40978381522909e-06,
      "loss": 0.6763,
      "step": 8381
    },
    {
      "epoch": 0.6149839045461633,
      "grad_norm": 1.467801650800877,
      "learning_rate": 3.408657358910144e-06,
      "loss": 0.8014,
      "step": 8382
    },
    {
      "epoch": 0.6150572741363024,
      "grad_norm": 1.3332619910587693,
      "learning_rate": 3.4075309924582577e-06,
      "loss": 0.763,
      "step": 8383
    },
    {
      "epoch": 0.6151306437264414,
      "grad_norm": 0.5471455241843816,
      "learning_rate": 3.406404715937038e-06,
      "loss": 0.4857,
      "step": 8384
    },
    {
      "epoch": 0.6151306437264414,
      "lm_loss": 1.7739,
      "step": 8384,
      "vm_loss": 0.0756
    },
    {
      "epoch": 0.6151306437264414,
      "lm_loss": 1.2154,
      "step": 8384,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.6151306437264414,
      "lm_loss": 1.3188,
      "step": 8384,
      "vm_loss": 0.1515
    },
    {
      "epoch": 0.6151306437264414,
      "lm_loss": 0.9959,
      "step": 8384,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.6151306437264414,
      "lm_loss": 0.0717,
      "step": 8384,
      "vm_loss": 0.1112
    },
    {
      "epoch": 0.6151306437264414,
      "lm_loss": 0.8167,
      "step": 8384,
      "vm_loss": 0.0626
    },
    {
      "epoch": 0.6151306437264414,
      "lm_loss": 1.1643,
      "step": 8384,
      "vm_loss": 0.2771
    },
    {
      "epoch": 0.6151306437264414,
      "lm_loss": 0.974,
      "step": 8384,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.6152040133165806,
      "grad_norm": 1.625014424041434,
      "learning_rate": 3.4052785294100933e-06,
      "loss": 0.7801,
      "step": 8385
    },
    {
      "epoch": 0.6152773829067197,
      "grad_norm": 1.6441986118447731,
      "learning_rate": 3.4041524329410182e-06,
      "loss": 0.7799,
      "step": 8386
    },
    {
      "epoch": 0.6153507524968589,
      "grad_norm": 1.4882049888724278,
      "learning_rate": 3.403026426593406e-06,
      "loss": 0.7227,
      "step": 8387
    },
    {
      "epoch": 0.615424122086998,
      "grad_norm": 1.4203937963634965,
      "learning_rate": 3.401900510430848e-06,
      "loss": 0.7716,
      "step": 8388
    },
    {
      "epoch": 0.6154974916771371,
      "grad_norm": 1.5431739927868626,
      "learning_rate": 3.4007746845169253e-06,
      "loss": 0.817,
      "step": 8389
    },
    {
      "epoch": 0.6155708612672762,
      "grad_norm": 1.594408957956618,
      "learning_rate": 3.399648948915216e-06,
      "loss": 0.7395,
      "step": 8390
    },
    {
      "epoch": 0.6156442308574154,
      "grad_norm": 1.400969292042043,
      "learning_rate": 3.398523303689294e-06,
      "loss": 0.8358,
      "step": 8391
    },
    {
      "epoch": 0.6157176004475545,
      "grad_norm": 0.5522651908922558,
      "learning_rate": 3.3973977489027264e-06,
      "loss": 0.4951,
      "step": 8392
    },
    {
      "epoch": 0.6157176004475545,
      "lm_loss": 0.237,
      "step": 8392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6157176004475545,
      "lm_loss": 0.413,
      "step": 8392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6157176004475545,
      "lm_loss": 0.4523,
      "step": 8392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6157176004475545,
      "lm_loss": 0.5505,
      "step": 8392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6157176004475545,
      "lm_loss": 0.5787,
      "step": 8392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6157176004475545,
      "lm_loss": 0.5201,
      "step": 8392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6157176004475545,
      "lm_loss": 0.5139,
      "step": 8392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6157176004475545,
      "lm_loss": 0.453,
      "step": 8392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6157909700376937,
      "grad_norm": 0.5598574217369925,
      "learning_rate": 3.3962722846190784e-06,
      "loss": 0.4913,
      "step": 8393
    },
    {
      "epoch": 0.6158643396278327,
      "grad_norm": 1.9873974054427614,
      "learning_rate": 3.3951469109019055e-06,
      "loss": 0.684,
      "step": 8394
    },
    {
      "epoch": 0.6159377092179719,
      "grad_norm": 1.4444045024841696,
      "learning_rate": 3.3940216278147595e-06,
      "loss": 0.7156,
      "step": 8395
    },
    {
      "epoch": 0.616011078808111,
      "grad_norm": 1.4428606169071907,
      "learning_rate": 3.3928964354211906e-06,
      "loss": 0.8076,
      "step": 8396
    },
    {
      "epoch": 0.6160844483982502,
      "grad_norm": 1.5893027034443081,
      "learning_rate": 3.3917713337847404e-06,
      "loss": 0.6938,
      "step": 8397
    },
    {
      "epoch": 0.6161578179883893,
      "grad_norm": 0.5633241554265024,
      "learning_rate": 3.390646322968944e-06,
      "loss": 0.4944,
      "step": 8398
    },
    {
      "epoch": 0.6162311875785283,
      "grad_norm": 1.5289797343716771,
      "learning_rate": 3.389521403037337e-06,
      "loss": 0.802,
      "step": 8399
    },
    {
      "epoch": 0.6163045571686675,
      "grad_norm": 1.5076096357546607,
      "learning_rate": 3.3883965740534432e-06,
      "loss": 0.6677,
      "step": 8400
    },
    {
      "epoch": 0.6163045571686675,
      "lm_loss": 2.306,
      "step": 8400,
      "vm_loss": 0.0644
    },
    {
      "epoch": 0.6163045571686675,
      "lm_loss": 0.7358,
      "step": 8400,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.6163045571686675,
      "lm_loss": 0.7971,
      "step": 8400,
      "vm_loss": 0.2145
    },
    {
      "epoch": 0.6163045571686675,
      "lm_loss": 0.3317,
      "step": 8400,
      "vm_loss": 0.048
    },
    {
      "epoch": 0.6163045571686675,
      "lm_loss": 0.9569,
      "step": 8400,
      "vm_loss": 0.1342
    },
    {
      "epoch": 0.6163045571686675,
      "lm_loss": 0.058,
      "step": 8400,
      "vm_loss": 0.1515
    },
    {
      "epoch": 0.6163045571686675,
      "lm_loss": 1.3879,
      "step": 8400,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.6163045571686675,
      "lm_loss": 0.869,
      "step": 8400,
      "vm_loss": 0.1547
    },
    {
      "epoch": 0.6163779267588066,
      "grad_norm": 1.5273938851892954,
      "learning_rate": 3.3872718360807867e-06,
      "loss": 0.7808,
      "step": 8401
    },
    {
      "epoch": 0.6164512963489458,
      "grad_norm": 1.4408082317029802,
      "learning_rate": 3.3861471891828834e-06,
      "loss": 0.7632,
      "step": 8402
    },
    {
      "epoch": 0.616524665939085,
      "grad_norm": 0.574355887952962,
      "learning_rate": 3.3850226334232446e-06,
      "loss": 0.4938,
      "step": 8403
    },
    {
      "epoch": 0.616598035529224,
      "grad_norm": 1.4872840718176716,
      "learning_rate": 3.3838981688653776e-06,
      "loss": 0.8406,
      "step": 8404
    },
    {
      "epoch": 0.6166714051193631,
      "grad_norm": 0.5501950757851752,
      "learning_rate": 3.3827737955727835e-06,
      "loss": 0.4666,
      "step": 8405
    },
    {
      "epoch": 0.6167447747095023,
      "grad_norm": 1.4748713938229943,
      "learning_rate": 3.381649513608957e-06,
      "loss": 0.7739,
      "step": 8406
    },
    {
      "epoch": 0.6168181442996414,
      "grad_norm": 1.4840540758025484,
      "learning_rate": 3.380525323037391e-06,
      "loss": 0.6346,
      "step": 8407
    },
    {
      "epoch": 0.6168915138897806,
      "grad_norm": 1.5443522594882915,
      "learning_rate": 3.3794012239215702e-06,
      "loss": 0.8367,
      "step": 8408
    },
    {
      "epoch": 0.6168915138897806,
      "lm_loss": 1.366,
      "step": 8408,
      "vm_loss": 0.0517
    },
    {
      "epoch": 0.6168915138897806,
      "lm_loss": 0.0946,
      "step": 8408,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.6168915138897806,
      "lm_loss": 0.2204,
      "step": 8408,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.6168915138897806,
      "lm_loss": 0.205,
      "step": 8408,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.6168915138897806,
      "lm_loss": 1.0276,
      "step": 8408,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.6168915138897806,
      "lm_loss": 0.3185,
      "step": 8408,
      "vm_loss": 0.1189
    },
    {
      "epoch": 0.6168915138897806,
      "lm_loss": 0.055,
      "step": 8408,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.6168915138897806,
      "lm_loss": 0.4828,
      "step": 8408,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.6169648834799196,
      "grad_norm": 1.4289533147703157,
      "learning_rate": 3.3782772163249767e-06,
      "loss": 0.7076,
      "step": 8409
    },
    {
      "epoch": 0.6170382530700588,
      "grad_norm": 1.4481179930150467,
      "learning_rate": 3.3771533003110848e-06,
      "loss": 0.8252,
      "step": 8410
    },
    {
      "epoch": 0.6171116226601979,
      "grad_norm": 1.479945829763666,
      "learning_rate": 3.3760294759433633e-06,
      "loss": 0.7658,
      "step": 8411
    },
    {
      "epoch": 0.6171849922503371,
      "grad_norm": 0.5732590374238326,
      "learning_rate": 3.37490574328528e-06,
      "loss": 0.4881,
      "step": 8412
    },
    {
      "epoch": 0.6172583618404762,
      "grad_norm": 1.4323924239318153,
      "learning_rate": 3.373782102400294e-06,
      "loss": 0.7479,
      "step": 8413
    },
    {
      "epoch": 0.6173317314306153,
      "grad_norm": 1.4498948643773673,
      "learning_rate": 3.37265855335186e-06,
      "loss": 0.7483,
      "step": 8414
    },
    {
      "epoch": 0.6174051010207544,
      "grad_norm": 0.5737454212641125,
      "learning_rate": 3.3715350962034273e-06,
      "loss": 0.5048,
      "step": 8415
    },
    {
      "epoch": 0.6174784706108936,
      "grad_norm": 1.7848916567810433,
      "learning_rate": 3.370411731018441e-06,
      "loss": 0.821,
      "step": 8416
    },
    {
      "epoch": 0.6174784706108936,
      "lm_loss": 1.5481,
      "step": 8416,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.6174784706108936,
      "lm_loss": 0.7142,
      "step": 8416,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.6174784706108936,
      "lm_loss": 0.8343,
      "step": 8416,
      "vm_loss": 0.0497
    },
    {
      "epoch": 0.6174784706108936,
      "lm_loss": 0.4571,
      "step": 8416,
      "vm_loss": 0.0776
    },
    {
      "epoch": 0.6174784706108936,
      "lm_loss": 0.6046,
      "step": 8416,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.6174784706108936,
      "lm_loss": 0.7757,
      "step": 8416,
      "vm_loss": 0.1683
    },
    {
      "epoch": 0.6174784706108936,
      "lm_loss": 0.5586,
      "step": 8416,
      "vm_loss": 0.068
    },
    {
      "epoch": 0.6174784706108936,
      "lm_loss": 0.0715,
      "step": 8416,
      "vm_loss": 0.198
    },
    {
      "epoch": 0.6175518402010327,
      "grad_norm": 1.8546008832203582,
      "learning_rate": 3.3692884578603386e-06,
      "loss": 0.8679,
      "step": 8417
    },
    {
      "epoch": 0.6176252097911719,
      "grad_norm": 1.629678513907402,
      "learning_rate": 3.3681652767925565e-06,
      "loss": 0.7751,
      "step": 8418
    },
    {
      "epoch": 0.6176985793813109,
      "grad_norm": 1.4686454606354706,
      "learning_rate": 3.3670421878785213e-06,
      "loss": 0.8852,
      "step": 8419
    },
    {
      "epoch": 0.61777194897145,
      "grad_norm": 1.5046766171874872,
      "learning_rate": 3.365919191181659e-06,
      "loss": 0.7457,
      "step": 8420
    },
    {
      "epoch": 0.6178453185615892,
      "grad_norm": 0.5706816010228204,
      "learning_rate": 3.3647962867653864e-06,
      "loss": 0.4754,
      "step": 8421
    },
    {
      "epoch": 0.6179186881517283,
      "grad_norm": 1.4232818126662914,
      "learning_rate": 3.3636734746931167e-06,
      "loss": 0.754,
      "step": 8422
    },
    {
      "epoch": 0.6179920577418674,
      "grad_norm": 1.7648988819895708,
      "learning_rate": 3.3625507550282583e-06,
      "loss": 0.7299,
      "step": 8423
    },
    {
      "epoch": 0.6180654273320065,
      "grad_norm": 1.5423578622607288,
      "learning_rate": 3.361428127834214e-06,
      "loss": 0.841,
      "step": 8424
    },
    {
      "epoch": 0.6180654273320065,
      "lm_loss": 0.5691,
      "step": 8424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6180654273320065,
      "lm_loss": 0.4761,
      "step": 8424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6180654273320065,
      "lm_loss": 0.5578,
      "step": 8424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6180654273320065,
      "lm_loss": 0.5475,
      "step": 8424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6180654273320065,
      "lm_loss": 0.4326,
      "step": 8424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6180654273320065,
      "lm_loss": 0.497,
      "step": 8424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6180654273320065,
      "lm_loss": 0.4393,
      "step": 8424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6180654273320065,
      "lm_loss": 0.5142,
      "step": 8424,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6181387969221457,
      "grad_norm": 0.5745826703258982,
      "learning_rate": 3.3603055931743805e-06,
      "loss": 0.4935,
      "step": 8425
    },
    {
      "epoch": 0.6182121665122848,
      "grad_norm": 1.4729787161716932,
      "learning_rate": 3.359183151112152e-06,
      "loss": 0.7767,
      "step": 8426
    },
    {
      "epoch": 0.618285536102424,
      "grad_norm": 1.632200516241648,
      "learning_rate": 3.3580608017109127e-06,
      "loss": 0.7395,
      "step": 8427
    },
    {
      "epoch": 0.618358905692563,
      "grad_norm": 0.5871263114448128,
      "learning_rate": 3.3569385450340473e-06,
      "loss": 0.518,
      "step": 8428
    },
    {
      "epoch": 0.6184322752827022,
      "grad_norm": 0.551167255987209,
      "learning_rate": 3.3558163811449317e-06,
      "loss": 0.4891,
      "step": 8429
    },
    {
      "epoch": 0.6185056448728413,
      "grad_norm": 0.5488151237873121,
      "learning_rate": 3.354694310106935e-06,
      "loss": 0.474,
      "step": 8430
    },
    {
      "epoch": 0.6185790144629805,
      "grad_norm": 1.694282236274622,
      "learning_rate": 3.353572331983427e-06,
      "loss": 0.8127,
      "step": 8431
    },
    {
      "epoch": 0.6186523840531196,
      "grad_norm": 1.546911069770297,
      "learning_rate": 3.352450446837766e-06,
      "loss": 0.7291,
      "step": 8432
    },
    {
      "epoch": 0.6186523840531196,
      "lm_loss": 1.0314,
      "step": 8432,
      "vm_loss": 0.1197
    },
    {
      "epoch": 0.6186523840531196,
      "lm_loss": 0.2475,
      "step": 8432,
      "vm_loss": 0.0457
    },
    {
      "epoch": 0.6186523840531196,
      "lm_loss": 0.8111,
      "step": 8432,
      "vm_loss": 0.1646
    },
    {
      "epoch": 0.6186523840531196,
      "lm_loss": 0.2955,
      "step": 8432,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.6186523840531196,
      "lm_loss": 0.6305,
      "step": 8432,
      "vm_loss": 0.3378
    },
    {
      "epoch": 0.6186523840531196,
      "lm_loss": 0.3089,
      "step": 8432,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.6186523840531196,
      "lm_loss": 0.736,
      "step": 8432,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.6186523840531196,
      "lm_loss": 0.9805,
      "step": 8432,
      "vm_loss": 0.1556
    },
    {
      "epoch": 0.6187257536432587,
      "grad_norm": 1.6710026914628557,
      "learning_rate": 3.3513286547333077e-06,
      "loss": 0.8334,
      "step": 8433
    },
    {
      "epoch": 0.6187991232333978,
      "grad_norm": 1.4104276347103275,
      "learning_rate": 3.350206955733404e-06,
      "loss": 0.7762,
      "step": 8434
    },
    {
      "epoch": 0.618872492823537,
      "grad_norm": 1.5598380388402322,
      "learning_rate": 3.349085349901399e-06,
      "loss": 0.809,
      "step": 8435
    },
    {
      "epoch": 0.6189458624136761,
      "grad_norm": 1.4232047554693998,
      "learning_rate": 3.3479638373006323e-06,
      "loss": 0.719,
      "step": 8436
    },
    {
      "epoch": 0.6190192320038153,
      "grad_norm": 0.6233239571050607,
      "learning_rate": 3.34684241799444e-06,
      "loss": 0.4828,
      "step": 8437
    },
    {
      "epoch": 0.6190926015939543,
      "grad_norm": 1.6470989303112356,
      "learning_rate": 3.3457210920461488e-06,
      "loss": 0.7352,
      "step": 8438
    },
    {
      "epoch": 0.6191659711840934,
      "grad_norm": 0.6454934459027737,
      "learning_rate": 3.3445998595190852e-06,
      "loss": 0.4852,
      "step": 8439
    },
    {
      "epoch": 0.6192393407742326,
      "grad_norm": 1.485744765452482,
      "learning_rate": 3.343478720476567e-06,
      "loss": 0.737,
      "step": 8440
    },
    {
      "epoch": 0.6192393407742326,
      "lm_loss": 1.7271,
      "step": 8440,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.6192393407742326,
      "lm_loss": 0.4111,
      "step": 8440,
      "vm_loss": 0.1685
    },
    {
      "epoch": 0.6192393407742326,
      "lm_loss": 0.3596,
      "step": 8440,
      "vm_loss": 0.1176
    },
    {
      "epoch": 0.6192393407742326,
      "lm_loss": 0.9527,
      "step": 8440,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.6192393407742326,
      "lm_loss": 0.9223,
      "step": 8440,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.6192393407742326,
      "lm_loss": 1.9525,
      "step": 8440,
      "vm_loss": 0.1881
    },
    {
      "epoch": 0.6192393407742326,
      "lm_loss": 0.2464,
      "step": 8440,
      "vm_loss": 0.1888
    },
    {
      "epoch": 0.6192393407742326,
      "lm_loss": 0.1818,
      "step": 8440,
      "vm_loss": 0.0947
    },
    {
      "epoch": 0.6193127103643717,
      "grad_norm": 1.5395050586998442,
      "learning_rate": 3.3423576749819077e-06,
      "loss": 0.6982,
      "step": 8441
    },
    {
      "epoch": 0.6193860799545109,
      "grad_norm": 1.4487929021659836,
      "learning_rate": 3.3412367230984157e-06,
      "loss": 0.7464,
      "step": 8442
    },
    {
      "epoch": 0.6194594495446499,
      "grad_norm": 0.5556196304326237,
      "learning_rate": 3.340115864889394e-06,
      "loss": 0.4655,
      "step": 8443
    },
    {
      "epoch": 0.6195328191347891,
      "grad_norm": 0.5623143558647508,
      "learning_rate": 3.338995100418139e-06,
      "loss": 0.4968,
      "step": 8444
    },
    {
      "epoch": 0.6196061887249282,
      "grad_norm": 0.5513200731977439,
      "learning_rate": 3.3378744297479446e-06,
      "loss": 0.467,
      "step": 8445
    },
    {
      "epoch": 0.6196795583150674,
      "grad_norm": 1.5964052268140625,
      "learning_rate": 3.3367538529420974e-06,
      "loss": 0.7165,
      "step": 8446
    },
    {
      "epoch": 0.6197529279052065,
      "grad_norm": 1.5690007545821891,
      "learning_rate": 3.3356333700638815e-06,
      "loss": 0.7695,
      "step": 8447
    },
    {
      "epoch": 0.6198262974953456,
      "grad_norm": 0.5969086482879211,
      "learning_rate": 3.3345129811765682e-06,
      "loss": 0.5082,
      "step": 8448
    },
    {
      "epoch": 0.6198262974953456,
      "lm_loss": 0.8543,
      "step": 8448,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.6198262974953456,
      "lm_loss": 1.1479,
      "step": 8448,
      "vm_loss": 0.1161
    },
    {
      "epoch": 0.6198262974953456,
      "lm_loss": 0.2662,
      "step": 8448,
      "vm_loss": 0.1741
    },
    {
      "epoch": 0.6198262974953456,
      "lm_loss": 0.9678,
      "step": 8448,
      "vm_loss": 0.1068
    },
    {
      "epoch": 0.6198262974953456,
      "lm_loss": 0.7562,
      "step": 8448,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.6198262974953456,
      "lm_loss": 0.7044,
      "step": 8448,
      "vm_loss": 0.0723
    },
    {
      "epoch": 0.6198262974953456,
      "lm_loss": 0.4175,
      "step": 8448,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.6198262974953456,
      "lm_loss": 0.2274,
      "step": 8448,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.6198996670854847,
      "grad_norm": 1.477915825956189,
      "learning_rate": 3.3333926863434317e-06,
      "loss": 0.8345,
      "step": 8449
    },
    {
      "epoch": 0.6199730366756239,
      "grad_norm": 0.5741960371798124,
      "learning_rate": 3.3322724856277373e-06,
      "loss": 0.5126,
      "step": 8450
    },
    {
      "epoch": 0.620046406265763,
      "grad_norm": 1.4469820025527012,
      "learning_rate": 3.3311523790927464e-06,
      "loss": 0.7361,
      "step": 8451
    },
    {
      "epoch": 0.6201197758559022,
      "grad_norm": 1.3133265328181387,
      "learning_rate": 3.330032366801713e-06,
      "loss": 0.7084,
      "step": 8452
    },
    {
      "epoch": 0.6201931454460412,
      "grad_norm": 0.5735613293596978,
      "learning_rate": 3.328912448817887e-06,
      "loss": 0.5127,
      "step": 8453
    },
    {
      "epoch": 0.6202665150361804,
      "grad_norm": 1.5420306966236714,
      "learning_rate": 3.3277926252045144e-06,
      "loss": 0.7195,
      "step": 8454
    },
    {
      "epoch": 0.6203398846263195,
      "grad_norm": 0.575277583025478,
      "learning_rate": 3.3266728960248335e-06,
      "loss": 0.5154,
      "step": 8455
    },
    {
      "epoch": 0.6204132542164587,
      "grad_norm": 0.555723129821364,
      "learning_rate": 3.325553261342077e-06,
      "loss": 0.4981,
      "step": 8456
    },
    {
      "epoch": 0.6204132542164587,
      "lm_loss": 1.0167,
      "step": 8456,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.6204132542164587,
      "lm_loss": 0.7277,
      "step": 8456,
      "vm_loss": 0.0249
    },
    {
      "epoch": 0.6204132542164587,
      "lm_loss": 0.0943,
      "step": 8456,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.6204132542164587,
      "lm_loss": 0.8189,
      "step": 8456,
      "vm_loss": 0.1248
    },
    {
      "epoch": 0.6204132542164587,
      "lm_loss": 0.7166,
      "step": 8456,
      "vm_loss": 0.1549
    },
    {
      "epoch": 0.6204132542164587,
      "lm_loss": 0.7908,
      "step": 8456,
      "vm_loss": 0.1456
    },
    {
      "epoch": 0.6204132542164587,
      "lm_loss": 0.4705,
      "step": 8456,
      "vm_loss": 0.1674
    },
    {
      "epoch": 0.6204132542164587,
      "lm_loss": 0.129,
      "step": 8456,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.6204866238065978,
      "grad_norm": 1.5124182550999936,
      "learning_rate": 3.3244337212194757e-06,
      "loss": 0.7346,
      "step": 8457
    },
    {
      "epoch": 0.6205599933967368,
      "grad_norm": 0.5527782585187611,
      "learning_rate": 3.323314275720251e-06,
      "loss": 0.4835,
      "step": 8458
    },
    {
      "epoch": 0.620633362986876,
      "grad_norm": 0.5697195286188377,
      "learning_rate": 3.3221949249076203e-06,
      "loss": 0.4948,
      "step": 8459
    },
    {
      "epoch": 0.6207067325770151,
      "grad_norm": 1.4752796467300024,
      "learning_rate": 3.3210756688447986e-06,
      "loss": 0.7467,
      "step": 8460
    },
    {
      "epoch": 0.6207801021671543,
      "grad_norm": 1.393844402274014,
      "learning_rate": 3.319956507594991e-06,
      "loss": 0.6709,
      "step": 8461
    },
    {
      "epoch": 0.6208534717572934,
      "grad_norm": 1.4940108893002193,
      "learning_rate": 3.318837441221401e-06,
      "loss": 0.7764,
      "step": 8462
    },
    {
      "epoch": 0.6209268413474325,
      "grad_norm": 0.5655747521524852,
      "learning_rate": 3.3177184697872235e-06,
      "loss": 0.4897,
      "step": 8463
    },
    {
      "epoch": 0.6210002109375716,
      "grad_norm": 1.5183885319701471,
      "learning_rate": 3.3165995933556486e-06,
      "loss": 0.8144,
      "step": 8464
    },
    {
      "epoch": 0.6210002109375716,
      "lm_loss": 0.4594,
      "step": 8464,
      "vm_loss": 0.0396
    },
    {
      "epoch": 0.6210002109375716,
      "lm_loss": 0.6592,
      "step": 8464,
      "vm_loss": 0.0484
    },
    {
      "epoch": 0.6210002109375716,
      "lm_loss": 0.8972,
      "step": 8464,
      "vm_loss": 0.143
    },
    {
      "epoch": 0.6210002109375716,
      "lm_loss": 0.3369,
      "step": 8464,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.6210002109375716,
      "lm_loss": 0.7378,
      "step": 8464,
      "vm_loss": 0.1389
    },
    {
      "epoch": 0.6210002109375716,
      "lm_loss": 1.3679,
      "step": 8464,
      "vm_loss": 0.2301
    },
    {
      "epoch": 0.6210002109375716,
      "lm_loss": 0.0343,
      "step": 8464,
      "vm_loss": 0.0898
    },
    {
      "epoch": 0.6210002109375716,
      "lm_loss": 0.4456,
      "step": 8464,
      "vm_loss": 0.1686
    },
    {
      "epoch": 0.6210735805277108,
      "grad_norm": 1.3733695737375176,
      "learning_rate": 3.315480811989866e-06,
      "loss": 0.7505,
      "step": 8465
    },
    {
      "epoch": 0.6211469501178499,
      "grad_norm": 1.4725083750440657,
      "learning_rate": 3.3143621257530522e-06,
      "loss": 0.7643,
      "step": 8466
    },
    {
      "epoch": 0.6212203197079891,
      "grad_norm": 1.4953850641328692,
      "learning_rate": 3.3132435347083834e-06,
      "loss": 0.7666,
      "step": 8467
    },
    {
      "epoch": 0.6212936892981281,
      "grad_norm": 1.5610751649260033,
      "learning_rate": 3.3121250389190303e-06,
      "loss": 0.8456,
      "step": 8468
    },
    {
      "epoch": 0.6213670588882673,
      "grad_norm": 1.367050244674106,
      "learning_rate": 3.311006638448155e-06,
      "loss": 0.7704,
      "step": 8469
    },
    {
      "epoch": 0.6214404284784064,
      "grad_norm": 0.5563977814188534,
      "learning_rate": 3.309888333358918e-06,
      "loss": 0.4811,
      "step": 8470
    },
    {
      "epoch": 0.6215137980685456,
      "grad_norm": 0.5648972509513879,
      "learning_rate": 3.3087701237144733e-06,
      "loss": 0.5067,
      "step": 8471
    },
    {
      "epoch": 0.6215871676586847,
      "grad_norm": 1.6051855240642983,
      "learning_rate": 3.307652009577966e-06,
      "loss": 0.6776,
      "step": 8472
    },
    {
      "epoch": 0.6215871676586847,
      "lm_loss": 0.0258,
      "step": 8472,
      "vm_loss": 0.0613
    },
    {
      "epoch": 0.6215871676586847,
      "lm_loss": 0.254,
      "step": 8472,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0.6215871676586847,
      "lm_loss": 0.8241,
      "step": 8472,
      "vm_loss": 0.2523
    },
    {
      "epoch": 0.6215871676586847,
      "lm_loss": 0.613,
      "step": 8472,
      "vm_loss": 0.191
    },
    {
      "epoch": 0.6215871676586847,
      "lm_loss": 0.7495,
      "step": 8472,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.6215871676586847,
      "lm_loss": 1.0761,
      "step": 8472,
      "vm_loss": 0.1686
    },
    {
      "epoch": 0.6215871676586847,
      "lm_loss": 1.0339,
      "step": 8472,
      "vm_loss": 0.1677
    },
    {
      "epoch": 0.6215871676586847,
      "lm_loss": 0.6811,
      "step": 8472,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.6216605372488238,
      "grad_norm": 1.6196118902882037,
      "learning_rate": 3.3065339910125426e-06,
      "loss": 0.848,
      "step": 8473
    },
    {
      "epoch": 0.6217339068389629,
      "grad_norm": 1.5883281070819302,
      "learning_rate": 3.3054160680813384e-06,
      "loss": 0.8674,
      "step": 8474
    },
    {
      "epoch": 0.621807276429102,
      "grad_norm": 1.4490186172734862,
      "learning_rate": 3.3042982408474842e-06,
      "loss": 0.8509,
      "step": 8475
    },
    {
      "epoch": 0.6218806460192412,
      "grad_norm": 0.5417363908331475,
      "learning_rate": 3.3031805093741085e-06,
      "loss": 0.4867,
      "step": 8476
    },
    {
      "epoch": 0.6219540156093804,
      "grad_norm": 1.6541912992745236,
      "learning_rate": 3.302062873724331e-06,
      "loss": 0.6495,
      "step": 8477
    },
    {
      "epoch": 0.6220273851995194,
      "grad_norm": 1.4737154939072843,
      "learning_rate": 3.300945333961269e-06,
      "loss": 0.8296,
      "step": 8478
    },
    {
      "epoch": 0.6221007547896585,
      "grad_norm": 1.5057611664064456,
      "learning_rate": 3.299827890148031e-06,
      "loss": 0.7998,
      "step": 8479
    },
    {
      "epoch": 0.6221741243797977,
      "grad_norm": 1.3874202601704968,
      "learning_rate": 3.2987105423477216e-06,
      "loss": 0.742,
      "step": 8480
    },
    {
      "epoch": 0.6221741243797977,
      "lm_loss": 1.6789,
      "step": 8480,
      "vm_loss": 0.0697
    },
    {
      "epoch": 0.6221741243797977,
      "lm_loss": 0.0835,
      "step": 8480,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.6221741243797977,
      "lm_loss": 0.3969,
      "step": 8480,
      "vm_loss": 0.1673
    },
    {
      "epoch": 0.6221741243797977,
      "lm_loss": 0.1304,
      "step": 8480,
      "vm_loss": 0.0962
    },
    {
      "epoch": 0.6221741243797977,
      "lm_loss": 0.41,
      "step": 8480,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.6221741243797977,
      "lm_loss": 0.7924,
      "step": 8480,
      "vm_loss": 0.1622
    },
    {
      "epoch": 0.6221741243797977,
      "lm_loss": 0.7359,
      "step": 8480,
      "vm_loss": 0.165
    },
    {
      "epoch": 0.6221741243797977,
      "lm_loss": 0.5076,
      "step": 8480,
      "vm_loss": 0.1099
    },
    {
      "epoch": 0.6222474939699368,
      "grad_norm": 1.4776367837033388,
      "learning_rate": 3.2975932906234425e-06,
      "loss": 0.702,
      "step": 8481
    },
    {
      "epoch": 0.622320863560076,
      "grad_norm": 1.5082234679432152,
      "learning_rate": 3.2964761350382864e-06,
      "loss": 0.815,
      "step": 8482
    },
    {
      "epoch": 0.622394233150215,
      "grad_norm": 0.5573330235609074,
      "learning_rate": 3.2953590756553398e-06,
      "loss": 0.4957,
      "step": 8483
    },
    {
      "epoch": 0.6224676027403542,
      "grad_norm": 0.5479020762621903,
      "learning_rate": 3.294242112537689e-06,
      "loss": 0.4849,
      "step": 8484
    },
    {
      "epoch": 0.6225409723304933,
      "grad_norm": 0.5378203377756998,
      "learning_rate": 3.293125245748411e-06,
      "loss": 0.4873,
      "step": 8485
    },
    {
      "epoch": 0.6226143419206325,
      "grad_norm": 1.409403798122743,
      "learning_rate": 3.2920084753505754e-06,
      "loss": 0.6881,
      "step": 8486
    },
    {
      "epoch": 0.6226877115107716,
      "grad_norm": 1.6567179474173561,
      "learning_rate": 3.290891801407252e-06,
      "loss": 0.7821,
      "step": 8487
    },
    {
      "epoch": 0.6227610811009107,
      "grad_norm": 0.5672310466910934,
      "learning_rate": 3.2897752239815005e-06,
      "loss": 0.4859,
      "step": 8488
    },
    {
      "epoch": 0.6227610811009107,
      "lm_loss": 1.641,
      "step": 8488,
      "vm_loss": 0.1321
    },
    {
      "epoch": 0.6227610811009107,
      "lm_loss": 0.5139,
      "step": 8488,
      "vm_loss": 0.1467
    },
    {
      "epoch": 0.6227610811009107,
      "lm_loss": 0.1648,
      "step": 8488,
      "vm_loss": 0.1224
    },
    {
      "epoch": 0.6227610811009107,
      "lm_loss": 0.9282,
      "step": 8488,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.6227610811009107,
      "lm_loss": 1.0078,
      "step": 8488,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.6227610811009107,
      "lm_loss": 0.9017,
      "step": 8488,
      "vm_loss": 0.1605
    },
    {
      "epoch": 0.6227610811009107,
      "lm_loss": 1.1137,
      "step": 8488,
      "vm_loss": 0.1709
    },
    {
      "epoch": 0.6227610811009107,
      "lm_loss": 0.6855,
      "step": 8488,
      "vm_loss": 0.0939
    },
    {
      "epoch": 0.6228344506910498,
      "grad_norm": 1.4748109523910944,
      "learning_rate": 3.288658743136378e-06,
      "loss": 0.8203,
      "step": 8489
    },
    {
      "epoch": 0.622907820281189,
      "grad_norm": 1.379927485809558,
      "learning_rate": 3.2875423589349355e-06,
      "loss": 0.7463,
      "step": 8490
    },
    {
      "epoch": 0.6229811898713281,
      "grad_norm": 1.6421888844980659,
      "learning_rate": 3.2864260714402147e-06,
      "loss": 0.6948,
      "step": 8491
    },
    {
      "epoch": 0.6230545594614673,
      "grad_norm": 1.3405111978474318,
      "learning_rate": 3.2853098807152583e-06,
      "loss": 0.6566,
      "step": 8492
    },
    {
      "epoch": 0.6231279290516063,
      "grad_norm": 1.619295010621982,
      "learning_rate": 3.2841937868231e-06,
      "loss": 0.8208,
      "step": 8493
    },
    {
      "epoch": 0.6232012986417454,
      "grad_norm": 1.5275805193573602,
      "learning_rate": 3.283077789826766e-06,
      "loss": 0.7356,
      "step": 8494
    },
    {
      "epoch": 0.6232746682318846,
      "grad_norm": 0.5746568633116693,
      "learning_rate": 3.2819618897892825e-06,
      "loss": 0.482,
      "step": 8495
    },
    {
      "epoch": 0.6233480378220237,
      "grad_norm": 1.6449740944300124,
      "learning_rate": 3.2808460867736647e-06,
      "loss": 0.7988,
      "step": 8496
    },
    {
      "epoch": 0.6233480378220237,
      "lm_loss": 2.1997,
      "step": 8496,
      "vm_loss": 0.0583
    },
    {
      "epoch": 0.6233480378220237,
      "lm_loss": 0.0937,
      "step": 8496,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.6233480378220237,
      "lm_loss": 0.7087,
      "step": 8496,
      "vm_loss": 0.2115
    },
    {
      "epoch": 0.6233480378220237,
      "lm_loss": 0.7881,
      "step": 8496,
      "vm_loss": 0.0706
    },
    {
      "epoch": 0.6233480378220237,
      "lm_loss": 0.4751,
      "step": 8496,
      "vm_loss": 0.0962
    },
    {
      "epoch": 0.6233480378220237,
      "lm_loss": 0.3972,
      "step": 8496,
      "vm_loss": 0.1481
    },
    {
      "epoch": 0.6233480378220237,
      "lm_loss": 0.9257,
      "step": 8496,
      "vm_loss": 0.1745
    },
    {
      "epoch": 0.6233480378220237,
      "lm_loss": 0.0315,
      "step": 8496,
      "vm_loss": 0.1316
    },
    {
      "epoch": 0.6234214074121629,
      "grad_norm": 1.4768396939733872,
      "learning_rate": 3.279730380842926e-06,
      "loss": 0.8203,
      "step": 8497
    },
    {
      "epoch": 0.6234947770023019,
      "grad_norm": 1.4515898951605073,
      "learning_rate": 3.2786147720600737e-06,
      "loss": 0.7994,
      "step": 8498
    },
    {
      "epoch": 0.6235681465924411,
      "grad_norm": 0.5810941394949458,
      "learning_rate": 3.277499260488107e-06,
      "loss": 0.5027,
      "step": 8499
    },
    {
      "epoch": 0.6236415161825802,
      "grad_norm": 1.389010683297104,
      "learning_rate": 3.276383846190024e-06,
      "loss": 0.7036,
      "step": 8500
    },
    {
      "epoch": 0.6237148857727194,
      "grad_norm": 1.545411486343649,
      "learning_rate": 3.275268529228813e-06,
      "loss": 0.7415,
      "step": 8501
    },
    {
      "epoch": 0.6237882553628585,
      "grad_norm": 1.5516997366659162,
      "learning_rate": 3.274153309667459e-06,
      "loss": 0.7213,
      "step": 8502
    },
    {
      "epoch": 0.6238616249529976,
      "grad_norm": 1.5172447243740668,
      "learning_rate": 3.2730381875689423e-06,
      "loss": 0.6966,
      "step": 8503
    },
    {
      "epoch": 0.6239349945431367,
      "grad_norm": 1.4733133697747796,
      "learning_rate": 3.271923162996235e-06,
      "loss": 0.7996,
      "step": 8504
    },
    {
      "epoch": 0.6239349945431367,
      "lm_loss": 1.2699,
      "step": 8504,
      "vm_loss": 0.0337
    },
    {
      "epoch": 0.6239349945431367,
      "lm_loss": 0.1125,
      "step": 8504,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.6239349945431367,
      "lm_loss": 0.8866,
      "step": 8504,
      "vm_loss": 0.1117
    },
    {
      "epoch": 0.6239349945431367,
      "lm_loss": 0.1833,
      "step": 8504,
      "vm_loss": 0.0883
    },
    {
      "epoch": 0.6239349945431367,
      "lm_loss": 0.0291,
      "step": 8504,
      "vm_loss": 0.0996
    },
    {
      "epoch": 0.6239349945431367,
      "lm_loss": 0.3807,
      "step": 8504,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.6239349945431367,
      "lm_loss": 0.1001,
      "step": 8504,
      "vm_loss": 0.0919
    },
    {
      "epoch": 0.6239349945431367,
      "lm_loss": 0.5375,
      "step": 8504,
      "vm_loss": 0.1041
    },
    {
      "epoch": 0.6240083641332759,
      "grad_norm": 1.3857847349482653,
      "learning_rate": 3.2708082360123073e-06,
      "loss": 0.6547,
      "step": 8505
    },
    {
      "epoch": 0.624081733723415,
      "grad_norm": 1.5601697993269517,
      "learning_rate": 3.26969340668012e-06,
      "loss": 0.7232,
      "step": 8506
    },
    {
      "epoch": 0.6241551033135541,
      "grad_norm": 1.590563291940997,
      "learning_rate": 3.2685786750626315e-06,
      "loss": 0.7182,
      "step": 8507
    },
    {
      "epoch": 0.6242284729036932,
      "grad_norm": 0.5904873282716914,
      "learning_rate": 3.2674640412227933e-06,
      "loss": 0.5084,
      "step": 8508
    },
    {
      "epoch": 0.6243018424938324,
      "grad_norm": 0.5790785256297821,
      "learning_rate": 3.2663495052235505e-06,
      "loss": 0.4901,
      "step": 8509
    },
    {
      "epoch": 0.6243752120839715,
      "grad_norm": 1.5225716416256387,
      "learning_rate": 3.2652350671278443e-06,
      "loss": 0.6419,
      "step": 8510
    },
    {
      "epoch": 0.6244485816741107,
      "grad_norm": 1.5489825779375774,
      "learning_rate": 3.264120726998611e-06,
      "loss": 0.8045,
      "step": 8511
    },
    {
      "epoch": 0.6245219512642497,
      "grad_norm": 1.4571628034616135,
      "learning_rate": 3.2630064848987788e-06,
      "loss": 0.6602,
      "step": 8512
    },
    {
      "epoch": 0.6245219512642497,
      "lm_loss": 0.839,
      "step": 8512,
      "vm_loss": 0.0556
    },
    {
      "epoch": 0.6245219512642497,
      "lm_loss": 1.297,
      "step": 8512,
      "vm_loss": 0.2166
    },
    {
      "epoch": 0.6245219512642497,
      "lm_loss": 0.3492,
      "step": 8512,
      "vm_loss": 0.0577
    },
    {
      "epoch": 0.6245219512642497,
      "lm_loss": 0.7194,
      "step": 8512,
      "vm_loss": 0.1437
    },
    {
      "epoch": 0.6245219512642497,
      "lm_loss": 0.2231,
      "step": 8512,
      "vm_loss": 0.0698
    },
    {
      "epoch": 0.6245219512642497,
      "lm_loss": 0.5111,
      "step": 8512,
      "vm_loss": 0.0937
    },
    {
      "epoch": 0.6245219512642497,
      "lm_loss": 1.1977,
      "step": 8512,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.6245219512642497,
      "lm_loss": 0.1233,
      "step": 8512,
      "vm_loss": 0.1241
    },
    {
      "epoch": 0.6245953208543888,
      "grad_norm": 1.596121541508274,
      "learning_rate": 3.2618923408912707e-06,
      "loss": 0.778,
      "step": 8513
    },
    {
      "epoch": 0.624668690444528,
      "grad_norm": 0.5370935410904691,
      "learning_rate": 3.2607782950390075e-06,
      "loss": 0.4952,
      "step": 8514
    },
    {
      "epoch": 0.6247420600346671,
      "grad_norm": 1.478665301766354,
      "learning_rate": 3.2596643474049006e-06,
      "loss": 0.6964,
      "step": 8515
    },
    {
      "epoch": 0.6248154296248063,
      "grad_norm": 1.5899571190512618,
      "learning_rate": 3.258550498051859e-06,
      "loss": 0.7695,
      "step": 8516
    },
    {
      "epoch": 0.6248887992149453,
      "grad_norm": 1.5629557215624017,
      "learning_rate": 3.2574367470427827e-06,
      "loss": 0.7741,
      "step": 8517
    },
    {
      "epoch": 0.6249621688050845,
      "grad_norm": 0.5753204378050412,
      "learning_rate": 3.256323094440568e-06,
      "loss": 0.5162,
      "step": 8518
    },
    {
      "epoch": 0.6250355383952236,
      "grad_norm": 0.555879092107671,
      "learning_rate": 3.2552095403081075e-06,
      "loss": 0.4946,
      "step": 8519
    },
    {
      "epoch": 0.6251089079853628,
      "grad_norm": 1.772879157700095,
      "learning_rate": 3.254096084708285e-06,
      "loss": 0.8134,
      "step": 8520
    },
    {
      "epoch": 0.6251089079853628,
      "lm_loss": 1.4562,
      "step": 8520,
      "vm_loss": 0.0742
    },
    {
      "epoch": 0.6251089079853628,
      "lm_loss": 1.2109,
      "step": 8520,
      "vm_loss": 0.1391
    },
    {
      "epoch": 0.6251089079853628,
      "lm_loss": 0.3464,
      "step": 8520,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.6251089079853628,
      "lm_loss": 1.1657,
      "step": 8520,
      "vm_loss": 0.2183
    },
    {
      "epoch": 0.6251089079853628,
      "lm_loss": 0.743,
      "step": 8520,
      "vm_loss": 0.1478
    },
    {
      "epoch": 0.6251089079853628,
      "lm_loss": 0.2796,
      "step": 8520,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.6251089079853628,
      "lm_loss": 1.0042,
      "step": 8520,
      "vm_loss": 0.1271
    },
    {
      "epoch": 0.6251089079853628,
      "lm_loss": 0.7803,
      "step": 8520,
      "vm_loss": 0.1593
    },
    {
      "epoch": 0.6251822775755019,
      "grad_norm": 1.717269328964594,
      "learning_rate": 3.2529827277039794e-06,
      "loss": 0.7328,
      "step": 8521
    },
    {
      "epoch": 0.625255647165641,
      "grad_norm": 1.716819368148663,
      "learning_rate": 3.251869469358066e-06,
      "loss": 0.7488,
      "step": 8522
    },
    {
      "epoch": 0.6253290167557801,
      "grad_norm": 0.57636062168998,
      "learning_rate": 3.250756309733413e-06,
      "loss": 0.4964,
      "step": 8523
    },
    {
      "epoch": 0.6254023863459193,
      "grad_norm": 1.845071618393683,
      "learning_rate": 3.2496432488928832e-06,
      "loss": 0.7409,
      "step": 8524
    },
    {
      "epoch": 0.6254757559360584,
      "grad_norm": 1.4495215442553688,
      "learning_rate": 3.2485302868993333e-06,
      "loss": 0.7477,
      "step": 8525
    },
    {
      "epoch": 0.6255491255261976,
      "grad_norm": 0.5364745987954681,
      "learning_rate": 3.2474174238156155e-06,
      "loss": 0.4704,
      "step": 8526
    },
    {
      "epoch": 0.6256224951163366,
      "grad_norm": 1.4960823323474377,
      "learning_rate": 3.2463046597045766e-06,
      "loss": 0.7633,
      "step": 8527
    },
    {
      "epoch": 0.6256958647064758,
      "grad_norm": 1.5940008312359697,
      "learning_rate": 3.2451919946290565e-06,
      "loss": 0.673,
      "step": 8528
    },
    {
      "epoch": 0.6256958647064758,
      "lm_loss": 0.6485,
      "step": 8528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6256958647064758,
      "lm_loss": 0.5467,
      "step": 8528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6256958647064758,
      "lm_loss": 0.4931,
      "step": 8528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6256958647064758,
      "lm_loss": 0.3721,
      "step": 8528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6256958647064758,
      "lm_loss": 0.4714,
      "step": 8528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6256958647064758,
      "lm_loss": 0.5444,
      "step": 8528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6256958647064758,
      "lm_loss": 0.3942,
      "step": 8528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6256958647064758,
      "lm_loss": 0.5264,
      "step": 8528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6257692342966149,
      "grad_norm": 0.5393336263605666,
      "learning_rate": 3.2440794286518896e-06,
      "loss": 0.4708,
      "step": 8529
    },
    {
      "epoch": 0.6258426038867541,
      "grad_norm": 1.3940247756955684,
      "learning_rate": 3.242966961835906e-06,
      "loss": 0.7245,
      "step": 8530
    },
    {
      "epoch": 0.6259159734768932,
      "grad_norm": 1.4787410535417616,
      "learning_rate": 3.2418545942439295e-06,
      "loss": 0.7409,
      "step": 8531
    },
    {
      "epoch": 0.6259893430670322,
      "grad_norm": 0.5545252423849067,
      "learning_rate": 3.2407423259387783e-06,
      "loss": 0.4722,
      "step": 8532
    },
    {
      "epoch": 0.6260627126571714,
      "grad_norm": 0.564321192998847,
      "learning_rate": 3.2396301569832646e-06,
      "loss": 0.4953,
      "step": 8533
    },
    {
      "epoch": 0.6261360822473105,
      "grad_norm": 1.4539708377576872,
      "learning_rate": 3.2385180874401946e-06,
      "loss": 0.736,
      "step": 8534
    },
    {
      "epoch": 0.6262094518374497,
      "grad_norm": 1.6270442294672978,
      "learning_rate": 3.2374061173723715e-06,
      "loss": 0.8325,
      "step": 8535
    },
    {
      "epoch": 0.6262828214275888,
      "grad_norm": 1.493521917749391,
      "learning_rate": 3.2362942468425893e-06,
      "loss": 0.7473,
      "step": 8536
    },
    {
      "epoch": 0.6262828214275888,
      "lm_loss": 1.3436,
      "step": 8536,
      "vm_loss": 0.0806
    },
    {
      "epoch": 0.6262828214275888,
      "lm_loss": 1.0234,
      "step": 8536,
      "vm_loss": 0.0825
    },
    {
      "epoch": 0.6262828214275888,
      "lm_loss": 0.1133,
      "step": 8536,
      "vm_loss": 0.1597
    },
    {
      "epoch": 0.6262828214275888,
      "lm_loss": 0.1332,
      "step": 8536,
      "vm_loss": 0.1396
    },
    {
      "epoch": 0.6262828214275888,
      "lm_loss": 0.9696,
      "step": 8536,
      "vm_loss": 0.1449
    },
    {
      "epoch": 0.6262828214275888,
      "lm_loss": 0.1694,
      "step": 8536,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.6262828214275888,
      "lm_loss": 0.5542,
      "step": 8536,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.6262828214275888,
      "lm_loss": 0.9355,
      "step": 8536,
      "vm_loss": 0.1549
    },
    {
      "epoch": 0.6263561910177279,
      "grad_norm": 1.4448258838590449,
      "learning_rate": 3.2351824759136386e-06,
      "loss": 0.6856,
      "step": 8537
    },
    {
      "epoch": 0.626429560607867,
      "grad_norm": 1.63890113682678,
      "learning_rate": 3.2340708046483043e-06,
      "loss": 0.8505,
      "step": 8538
    },
    {
      "epoch": 0.6265029301980062,
      "grad_norm": 1.5317201816354105,
      "learning_rate": 3.232959233109365e-06,
      "loss": 0.6353,
      "step": 8539
    },
    {
      "epoch": 0.6265762997881453,
      "grad_norm": 1.5661782761977592,
      "learning_rate": 3.2318477613595935e-06,
      "loss": 0.7801,
      "step": 8540
    },
    {
      "epoch": 0.6266496693782845,
      "grad_norm": 1.7407901050926198,
      "learning_rate": 3.2307363894617583e-06,
      "loss": 0.81,
      "step": 8541
    },
    {
      "epoch": 0.6267230389684235,
      "grad_norm": 1.3989389006876662,
      "learning_rate": 3.2296251174786197e-06,
      "loss": 0.7263,
      "step": 8542
    },
    {
      "epoch": 0.6267964085585627,
      "grad_norm": 1.4972698767564292,
      "learning_rate": 3.228513945472935e-06,
      "loss": 0.7583,
      "step": 8543
    },
    {
      "epoch": 0.6268697781487018,
      "grad_norm": 0.5671538250474141,
      "learning_rate": 3.2274028735074565e-06,
      "loss": 0.4813,
      "step": 8544
    },
    {
      "epoch": 0.6268697781487018,
      "lm_loss": 1.3182,
      "step": 8544,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.6268697781487018,
      "lm_loss": 0.9722,
      "step": 8544,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.6268697781487018,
      "lm_loss": 0.6101,
      "step": 8544,
      "vm_loss": 0.0987
    },
    {
      "epoch": 0.6268697781487018,
      "lm_loss": 0.2571,
      "step": 8544,
      "vm_loss": 0.1197
    },
    {
      "epoch": 0.6268697781487018,
      "lm_loss": 0.531,
      "step": 8544,
      "vm_loss": 0.2164
    },
    {
      "epoch": 0.6268697781487018,
      "lm_loss": 0.7388,
      "step": 8544,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.6268697781487018,
      "lm_loss": 0.6829,
      "step": 8544,
      "vm_loss": 0.212
    },
    {
      "epoch": 0.6268697781487018,
      "lm_loss": 0.2342,
      "step": 8544,
      "vm_loss": 0.0936
    },
    {
      "epoch": 0.626943147738841,
      "grad_norm": 1.808552878232875,
      "learning_rate": 3.2262919016449255e-06,
      "loss": 0.7889,
      "step": 8545
    },
    {
      "epoch": 0.6270165173289801,
      "grad_norm": 0.5740175779652664,
      "learning_rate": 3.2251810299480846e-06,
      "loss": 0.4721,
      "step": 8546
    },
    {
      "epoch": 0.6270898869191192,
      "grad_norm": 1.202842720566795,
      "learning_rate": 3.2240702584796655e-06,
      "loss": 0.63,
      "step": 8547
    },
    {
      "epoch": 0.6271632565092583,
      "grad_norm": 0.5812834105986813,
      "learning_rate": 3.222959587302397e-06,
      "loss": 0.4905,
      "step": 8548
    },
    {
      "epoch": 0.6272366260993975,
      "grad_norm": 1.5591062738015722,
      "learning_rate": 3.2218490164790015e-06,
      "loss": 0.783,
      "step": 8549
    },
    {
      "epoch": 0.6273099956895366,
      "grad_norm": 0.5487227226174524,
      "learning_rate": 3.2207385460721954e-06,
      "loss": 0.4807,
      "step": 8550
    },
    {
      "epoch": 0.6273833652796758,
      "grad_norm": 0.5415779565050703,
      "learning_rate": 3.21962817614469e-06,
      "loss": 0.4639,
      "step": 8551
    },
    {
      "epoch": 0.6274567348698148,
      "grad_norm": 0.5488717864235572,
      "learning_rate": 3.2185179067591915e-06,
      "loss": 0.5056,
      "step": 8552
    },
    {
      "epoch": 0.6274567348698148,
      "lm_loss": 0.4744,
      "step": 8552,
      "vm_loss": 0.0673
    },
    {
      "epoch": 0.6274567348698148,
      "lm_loss": 1.0687,
      "step": 8552,
      "vm_loss": 0.1633
    },
    {
      "epoch": 0.6274567348698148,
      "lm_loss": 0.3916,
      "step": 8552,
      "vm_loss": 0.0624
    },
    {
      "epoch": 0.6274567348698148,
      "lm_loss": 0.0615,
      "step": 8552,
      "vm_loss": 0.1101
    },
    {
      "epoch": 0.6274567348698148,
      "lm_loss": 0.8847,
      "step": 8552,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.6274567348698148,
      "lm_loss": 0.6949,
      "step": 8552,
      "vm_loss": 0.1204
    },
    {
      "epoch": 0.6274567348698148,
      "lm_loss": 0.5573,
      "step": 8552,
      "vm_loss": 0.0967
    },
    {
      "epoch": 0.6274567348698148,
      "lm_loss": 0.5764,
      "step": 8552,
      "vm_loss": 0.1206
    },
    {
      "epoch": 0.6275301044599539,
      "grad_norm": 1.4784446977954557,
      "learning_rate": 3.217407737978397e-06,
      "loss": 0.783,
      "step": 8553
    },
    {
      "epoch": 0.6276034740500931,
      "grad_norm": 0.561239650820141,
      "learning_rate": 3.2162976698650027e-06,
      "loss": 0.5091,
      "step": 8554
    },
    {
      "epoch": 0.6276768436402322,
      "grad_norm": 1.5337646667817344,
      "learning_rate": 3.215187702481697e-06,
      "loss": 0.8184,
      "step": 8555
    },
    {
      "epoch": 0.6277502132303714,
      "grad_norm": 0.5586851002511642,
      "learning_rate": 3.2140778358911604e-06,
      "loss": 0.502,
      "step": 8556
    },
    {
      "epoch": 0.6278235828205104,
      "grad_norm": 0.5626439035876718,
      "learning_rate": 3.212968070156072e-06,
      "loss": 0.4785,
      "step": 8557
    },
    {
      "epoch": 0.6278969524106496,
      "grad_norm": 1.5673867608869516,
      "learning_rate": 3.211858405339101e-06,
      "loss": 0.8574,
      "step": 8558
    },
    {
      "epoch": 0.6279703220007887,
      "grad_norm": 1.6554141956311352,
      "learning_rate": 3.2107488415029157e-06,
      "loss": 0.7447,
      "step": 8559
    },
    {
      "epoch": 0.6280436915909279,
      "grad_norm": 0.5447133358965661,
      "learning_rate": 3.209639378710174e-06,
      "loss": 0.4942,
      "step": 8560
    },
    {
      "epoch": 0.6280436915909279,
      "lm_loss": 1.0896,
      "step": 8560,
      "vm_loss": 0.0706
    },
    {
      "epoch": 0.6280436915909279,
      "lm_loss": 1.2637,
      "step": 8560,
      "vm_loss": 0.1217
    },
    {
      "epoch": 0.6280436915909279,
      "lm_loss": 0.0526,
      "step": 8560,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.6280436915909279,
      "lm_loss": 0.6174,
      "step": 8560,
      "vm_loss": 0.2092
    },
    {
      "epoch": 0.6280436915909279,
      "lm_loss": 0.2162,
      "step": 8560,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.6280436915909279,
      "lm_loss": 0.1887,
      "step": 8560,
      "vm_loss": 0.2035
    },
    {
      "epoch": 0.6280436915909279,
      "lm_loss": 0.1396,
      "step": 8560,
      "vm_loss": 0.043
    },
    {
      "epoch": 0.6280436915909279,
      "lm_loss": 0.0769,
      "step": 8560,
      "vm_loss": 0.1355
    },
    {
      "epoch": 0.628117061181067,
      "grad_norm": 1.339453642678871,
      "learning_rate": 3.20853001702353e-06,
      "loss": 0.7654,
      "step": 8561
    },
    {
      "epoch": 0.6281904307712061,
      "grad_norm": 1.7157937049093366,
      "learning_rate": 3.2074207565056324e-06,
      "loss": 0.7576,
      "step": 8562
    },
    {
      "epoch": 0.6282638003613452,
      "grad_norm": 1.5644995596621698,
      "learning_rate": 3.2063115972191246e-06,
      "loss": 0.7424,
      "step": 8563
    },
    {
      "epoch": 0.6283371699514844,
      "grad_norm": 1.3042045816968122,
      "learning_rate": 3.2052025392266417e-06,
      "loss": 0.6937,
      "step": 8564
    },
    {
      "epoch": 0.6284105395416235,
      "grad_norm": 1.4733488631213503,
      "learning_rate": 3.204093582590817e-06,
      "loss": 0.8048,
      "step": 8565
    },
    {
      "epoch": 0.6284839091317627,
      "grad_norm": 1.5454980512699348,
      "learning_rate": 3.2029847273742753e-06,
      "loss": 0.7512,
      "step": 8566
    },
    {
      "epoch": 0.6285572787219017,
      "grad_norm": 1.5256709665698316,
      "learning_rate": 3.2018759736396354e-06,
      "loss": 0.6586,
      "step": 8567
    },
    {
      "epoch": 0.6286306483120409,
      "grad_norm": 0.5481289830355889,
      "learning_rate": 3.2007673214495126e-06,
      "loss": 0.4851,
      "step": 8568
    },
    {
      "epoch": 0.6286306483120409,
      "lm_loss": 1.3865,
      "step": 8568,
      "vm_loss": 0.0364
    },
    {
      "epoch": 0.6286306483120409,
      "lm_loss": 1.4378,
      "step": 8568,
      "vm_loss": 0.0569
    },
    {
      "epoch": 0.6286306483120409,
      "lm_loss": 0.8824,
      "step": 8568,
      "vm_loss": 0.1694
    },
    {
      "epoch": 0.6286306483120409,
      "lm_loss": 1.28,
      "step": 8568,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.6286306483120409,
      "lm_loss": 1.1044,
      "step": 8568,
      "vm_loss": 0.0734
    },
    {
      "epoch": 0.6286306483120409,
      "lm_loss": 0.8303,
      "step": 8568,
      "vm_loss": 0.1498
    },
    {
      "epoch": 0.6286306483120409,
      "lm_loss": 0.3364,
      "step": 8568,
      "vm_loss": 0.1251
    },
    {
      "epoch": 0.6286306483120409,
      "lm_loss": 0.9649,
      "step": 8568,
      "vm_loss": 0.1769
    },
    {
      "epoch": 0.62870401790218,
      "grad_norm": 1.3985622923061662,
      "learning_rate": 3.199658770866515e-06,
      "loss": 0.8247,
      "step": 8569
    },
    {
      "epoch": 0.6287773874923192,
      "grad_norm": 0.556849584170358,
      "learning_rate": 3.1985503219532456e-06,
      "loss": 0.4918,
      "step": 8570
    },
    {
      "epoch": 0.6288507570824583,
      "grad_norm": 1.6160005281070056,
      "learning_rate": 3.1974419747723006e-06,
      "loss": 0.755,
      "step": 8571
    },
    {
      "epoch": 0.6289241266725973,
      "grad_norm": 1.9151294205543679,
      "learning_rate": 3.19633372938627e-06,
      "loss": 0.6793,
      "step": 8572
    },
    {
      "epoch": 0.6289974962627365,
      "grad_norm": 1.5938107877000918,
      "learning_rate": 3.195225585857742e-06,
      "loss": 0.7364,
      "step": 8573
    },
    {
      "epoch": 0.6290708658528756,
      "grad_norm": 1.3066896541348292,
      "learning_rate": 3.194117544249294e-06,
      "loss": 0.6828,
      "step": 8574
    },
    {
      "epoch": 0.6291442354430148,
      "grad_norm": 1.499386195347291,
      "learning_rate": 3.1930096046235e-06,
      "loss": 0.7753,
      "step": 8575
    },
    {
      "epoch": 0.6292176050331539,
      "grad_norm": 1.5430093718542273,
      "learning_rate": 3.1919017670429285e-06,
      "loss": 0.7378,
      "step": 8576
    },
    {
      "epoch": 0.6292176050331539,
      "lm_loss": 0.5076,
      "step": 8576,
      "vm_loss": 0.0501
    },
    {
      "epoch": 0.6292176050331539,
      "lm_loss": 0.0172,
      "step": 8576,
      "vm_loss": 0.0854
    },
    {
      "epoch": 0.6292176050331539,
      "lm_loss": 0.6398,
      "step": 8576,
      "vm_loss": 0.1496
    },
    {
      "epoch": 0.6292176050331539,
      "lm_loss": 0.8952,
      "step": 8576,
      "vm_loss": 0.1498
    },
    {
      "epoch": 0.6292176050331539,
      "lm_loss": 0.3121,
      "step": 8576,
      "vm_loss": 0.1533
    },
    {
      "epoch": 0.6292176050331539,
      "lm_loss": 0.2406,
      "step": 8576,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.6292176050331539,
      "lm_loss": 0.3585,
      "step": 8576,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.6292176050331539,
      "lm_loss": 0.8063,
      "step": 8576,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.629290974623293,
      "grad_norm": 1.5603583689160077,
      "learning_rate": 3.1907940315701426e-06,
      "loss": 0.7968,
      "step": 8577
    },
    {
      "epoch": 0.6293643442134321,
      "grad_norm": 0.5451342338996806,
      "learning_rate": 3.1896863982676977e-06,
      "loss": 0.4726,
      "step": 8578
    },
    {
      "epoch": 0.6294377138035713,
      "grad_norm": 1.6383400375838242,
      "learning_rate": 3.1885788671981455e-06,
      "loss": 0.8303,
      "step": 8579
    },
    {
      "epoch": 0.6295110833937104,
      "grad_norm": 1.6384113369900535,
      "learning_rate": 3.18747143842403e-06,
      "loss": 0.7989,
      "step": 8580
    },
    {
      "epoch": 0.6295844529838496,
      "grad_norm": 0.5572085116180802,
      "learning_rate": 3.1863641120078913e-06,
      "loss": 0.4952,
      "step": 8581
    },
    {
      "epoch": 0.6296578225739886,
      "grad_norm": 0.5514762820786169,
      "learning_rate": 3.1852568880122635e-06,
      "loss": 0.474,
      "step": 8582
    },
    {
      "epoch": 0.6297311921641278,
      "grad_norm": 1.4853220801725968,
      "learning_rate": 3.184149766499672e-06,
      "loss": 0.7527,
      "step": 8583
    },
    {
      "epoch": 0.6298045617542669,
      "grad_norm": 0.5460815522476546,
      "learning_rate": 3.183042747532641e-06,
      "loss": 0.4729,
      "step": 8584
    },
    {
      "epoch": 0.6298045617542669,
      "lm_loss": 1.2358,
      "step": 8584,
      "vm_loss": 0.0747
    },
    {
      "epoch": 0.6298045617542669,
      "lm_loss": 0.5812,
      "step": 8584,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.6298045617542669,
      "lm_loss": 0.6401,
      "step": 8584,
      "vm_loss": 0.2261
    },
    {
      "epoch": 0.6298045617542669,
      "lm_loss": 0.0941,
      "step": 8584,
      "vm_loss": 0.2099
    },
    {
      "epoch": 0.6298045617542669,
      "lm_loss": 0.1143,
      "step": 8584,
      "vm_loss": 0.1505
    },
    {
      "epoch": 0.6298045617542669,
      "lm_loss": 0.1683,
      "step": 8584,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.6298045617542669,
      "lm_loss": 0.801,
      "step": 8584,
      "vm_loss": 0.1589
    },
    {
      "epoch": 0.6298045617542669,
      "lm_loss": 0.9097,
      "step": 8584,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.6298779313444061,
      "grad_norm": 1.5432110084925488,
      "learning_rate": 3.1819358311736846e-06,
      "loss": 0.688,
      "step": 8585
    },
    {
      "epoch": 0.6299513009345451,
      "grad_norm": 1.5668713344766698,
      "learning_rate": 3.180829017485315e-06,
      "loss": 0.8146,
      "step": 8586
    },
    {
      "epoch": 0.6300246705246842,
      "grad_norm": 1.4879782269655357,
      "learning_rate": 3.1797223065300364e-06,
      "loss": 0.8903,
      "step": 8587
    },
    {
      "epoch": 0.6300980401148234,
      "grad_norm": 0.591699896959889,
      "learning_rate": 3.178615698370345e-06,
      "loss": 0.5056,
      "step": 8588
    },
    {
      "epoch": 0.6301714097049625,
      "grad_norm": 0.5647285940516691,
      "learning_rate": 3.1775091930687374e-06,
      "loss": 0.4921,
      "step": 8589
    },
    {
      "epoch": 0.6302447792951017,
      "grad_norm": 1.5140836993255016,
      "learning_rate": 3.1764027906876992e-06,
      "loss": 0.7755,
      "step": 8590
    },
    {
      "epoch": 0.6303181488852407,
      "grad_norm": 0.5633872525881977,
      "learning_rate": 3.17529649128971e-06,
      "loss": 0.4884,
      "step": 8591
    },
    {
      "epoch": 0.6303915184753799,
      "grad_norm": 0.5496445183802413,
      "learning_rate": 3.174190294937248e-06,
      "loss": 0.4758,
      "step": 8592
    },
    {
      "epoch": 0.6303915184753799,
      "lm_loss": 1.3102,
      "step": 8592,
      "vm_loss": 0.0978
    },
    {
      "epoch": 0.6303915184753799,
      "lm_loss": 0.0925,
      "step": 8592,
      "vm_loss": 0.0635
    },
    {
      "epoch": 0.6303915184753799,
      "lm_loss": 0.9474,
      "step": 8592,
      "vm_loss": 0.1665
    },
    {
      "epoch": 0.6303915184753799,
      "lm_loss": 0.3365,
      "step": 8592,
      "vm_loss": 0.0798
    },
    {
      "epoch": 0.6303915184753799,
      "lm_loss": 0.454,
      "step": 8592,
      "vm_loss": 0.2115
    },
    {
      "epoch": 0.6303915184753799,
      "lm_loss": 1.0772,
      "step": 8592,
      "vm_loss": 0.0983
    },
    {
      "epoch": 0.6303915184753799,
      "lm_loss": 0.5365,
      "step": 8592,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.6303915184753799,
      "lm_loss": 0.324,
      "step": 8592,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.630464888065519,
      "grad_norm": 1.5364382231560016,
      "learning_rate": 3.1730842016927822e-06,
      "loss": 0.7403,
      "step": 8593
    },
    {
      "epoch": 0.6305382576556582,
      "grad_norm": 1.4361911350709915,
      "learning_rate": 3.171978211618774e-06,
      "loss": 0.6561,
      "step": 8594
    },
    {
      "epoch": 0.6306116272457973,
      "grad_norm": 1.613545853494467,
      "learning_rate": 3.1708723247776842e-06,
      "loss": 0.6762,
      "step": 8595
    },
    {
      "epoch": 0.6306849968359364,
      "grad_norm": 1.560202410203422,
      "learning_rate": 3.1697665412319636e-06,
      "loss": 0.766,
      "step": 8596
    },
    {
      "epoch": 0.6307583664260755,
      "grad_norm": 1.4956717425828292,
      "learning_rate": 3.168660861044061e-06,
      "loss": 0.7785,
      "step": 8597
    },
    {
      "epoch": 0.6308317360162147,
      "grad_norm": 1.5605835439885951,
      "learning_rate": 3.1675552842764122e-06,
      "loss": 0.77,
      "step": 8598
    },
    {
      "epoch": 0.6309051056063538,
      "grad_norm": 1.473439269448491,
      "learning_rate": 3.1664498109914554e-06,
      "loss": 0.7272,
      "step": 8599
    },
    {
      "epoch": 0.630978475196493,
      "grad_norm": 1.3730626616428663,
      "learning_rate": 3.1653444412516177e-06,
      "loss": 0.8487,
      "step": 8600
    },
    {
      "epoch": 0.630978475196493,
      "lm_loss": 1.379,
      "step": 8600,
      "vm_loss": 0.0519
    },
    {
      "epoch": 0.630978475196493,
      "lm_loss": 0.911,
      "step": 8600,
      "vm_loss": 0.199
    },
    {
      "epoch": 0.630978475196493,
      "lm_loss": 0.8025,
      "step": 8600,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.630978475196493,
      "lm_loss": 0.6535,
      "step": 8600,
      "vm_loss": 0.1273
    },
    {
      "epoch": 0.630978475196493,
      "lm_loss": 0.0849,
      "step": 8600,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.630978475196493,
      "lm_loss": 0.528,
      "step": 8600,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.630978475196493,
      "lm_loss": 0.926,
      "step": 8600,
      "vm_loss": 0.1238
    },
    {
      "epoch": 0.630978475196493,
      "lm_loss": 0.2321,
      "step": 8600,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.631051844786632,
      "grad_norm": 1.3449018847653222,
      "learning_rate": 3.164239175119324e-06,
      "loss": 0.6862,
      "step": 8601
    },
    {
      "epoch": 0.6311252143767712,
      "grad_norm": 0.636652345538311,
      "learning_rate": 3.1631340126569896e-06,
      "loss": 0.4991,
      "step": 8602
    },
    {
      "epoch": 0.6311985839669103,
      "grad_norm": 1.646746625688827,
      "learning_rate": 3.1620289539270266e-06,
      "loss": 0.8477,
      "step": 8603
    },
    {
      "epoch": 0.6312719535570495,
      "grad_norm": 1.4529848941142842,
      "learning_rate": 3.1609239989918405e-06,
      "loss": 0.7613,
      "step": 8604
    },
    {
      "epoch": 0.6313453231471886,
      "grad_norm": 1.3840268183101492,
      "learning_rate": 3.1598191479138307e-06,
      "loss": 0.745,
      "step": 8605
    },
    {
      "epoch": 0.6314186927373276,
      "grad_norm": 0.571994305992122,
      "learning_rate": 3.1587144007553896e-06,
      "loss": 0.4882,
      "step": 8606
    },
    {
      "epoch": 0.6314920623274668,
      "grad_norm": 1.6281712005338962,
      "learning_rate": 3.1576097575789066e-06,
      "loss": 0.7903,
      "step": 8607
    },
    {
      "epoch": 0.631565431917606,
      "grad_norm": 1.3240457607061946,
      "learning_rate": 3.1565052184467636e-06,
      "loss": 0.7968,
      "step": 8608
    },
    {
      "epoch": 0.631565431917606,
      "lm_loss": 1.0095,
      "step": 8608,
      "vm_loss": 0.0556
    },
    {
      "epoch": 0.631565431917606,
      "lm_loss": 0.5843,
      "step": 8608,
      "vm_loss": 0.157
    },
    {
      "epoch": 0.631565431917606,
      "lm_loss": 1.1045,
      "step": 8608,
      "vm_loss": 0.1672
    },
    {
      "epoch": 0.631565431917606,
      "lm_loss": 0.3449,
      "step": 8608,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.631565431917606,
      "lm_loss": 0.3551,
      "step": 8608,
      "vm_loss": 0.0567
    },
    {
      "epoch": 0.631565431917606,
      "lm_loss": 0.2313,
      "step": 8608,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.631565431917606,
      "lm_loss": 0.0496,
      "step": 8608,
      "vm_loss": 0.0728
    },
    {
      "epoch": 0.631565431917606,
      "lm_loss": 0.1579,
      "step": 8608,
      "vm_loss": 0.1769
    },
    {
      "epoch": 0.6316388015077451,
      "grad_norm": 1.7026118492847995,
      "learning_rate": 3.1554007834213357e-06,
      "loss": 0.7809,
      "step": 8609
    },
    {
      "epoch": 0.6317121710978842,
      "grad_norm": 1.5932901678830877,
      "learning_rate": 3.1542964525649945e-06,
      "loss": 0.7733,
      "step": 8610
    },
    {
      "epoch": 0.6317855406880233,
      "grad_norm": 1.516613857454559,
      "learning_rate": 3.153192225940102e-06,
      "loss": 0.808,
      "step": 8611
    },
    {
      "epoch": 0.6318589102781624,
      "grad_norm": 0.5512193270765047,
      "learning_rate": 3.1520881036090197e-06,
      "loss": 0.4837,
      "step": 8612
    },
    {
      "epoch": 0.6319322798683016,
      "grad_norm": 0.5593181586074732,
      "learning_rate": 3.150984085634098e-06,
      "loss": 0.4914,
      "step": 8613
    },
    {
      "epoch": 0.6320056494584407,
      "grad_norm": 1.6808519452737976,
      "learning_rate": 3.149880172077683e-06,
      "loss": 0.7638,
      "step": 8614
    },
    {
      "epoch": 0.6320790190485799,
      "grad_norm": 1.5616950946943657,
      "learning_rate": 3.1487763630021173e-06,
      "loss": 0.7185,
      "step": 8615
    },
    {
      "epoch": 0.6321523886387189,
      "grad_norm": 1.623946847220705,
      "learning_rate": 3.1476726584697353e-06,
      "loss": 0.7856,
      "step": 8616
    },
    {
      "epoch": 0.6321523886387189,
      "lm_loss": 1.9826,
      "step": 8616,
      "vm_loss": 0.0439
    },
    {
      "epoch": 0.6321523886387189,
      "lm_loss": 0.589,
      "step": 8616,
      "vm_loss": 0.142
    },
    {
      "epoch": 0.6321523886387189,
      "lm_loss": 0.3472,
      "step": 8616,
      "vm_loss": 0.1527
    },
    {
      "epoch": 0.6321523886387189,
      "lm_loss": 1.0569,
      "step": 8616,
      "vm_loss": 0.1648
    },
    {
      "epoch": 0.6321523886387189,
      "lm_loss": 0.3499,
      "step": 8616,
      "vm_loss": 0.0758
    },
    {
      "epoch": 0.6321523886387189,
      "lm_loss": 0.6167,
      "step": 8616,
      "vm_loss": 0.1878
    },
    {
      "epoch": 0.6321523886387189,
      "lm_loss": 0.5386,
      "step": 8616,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.6321523886387189,
      "lm_loss": 0.4677,
      "step": 8616,
      "vm_loss": 0.2381
    },
    {
      "epoch": 0.6322257582288581,
      "grad_norm": 1.4706643908974877,
      "learning_rate": 3.146569058542864e-06,
      "loss": 0.7253,
      "step": 8617
    },
    {
      "epoch": 0.6322991278189972,
      "grad_norm": 0.5700237369413484,
      "learning_rate": 3.1454655632838292e-06,
      "loss": 0.516,
      "step": 8618
    },
    {
      "epoch": 0.6323724974091364,
      "grad_norm": 0.5766004798284456,
      "learning_rate": 3.1443621727549456e-06,
      "loss": 0.4886,
      "step": 8619
    },
    {
      "epoch": 0.6324458669992755,
      "grad_norm": 1.4563392817817513,
      "learning_rate": 3.1432588870185272e-06,
      "loss": 0.6872,
      "step": 8620
    },
    {
      "epoch": 0.6325192365894146,
      "grad_norm": 1.5912849525441495,
      "learning_rate": 3.142155706136877e-06,
      "loss": 0.7312,
      "step": 8621
    },
    {
      "epoch": 0.6325926061795537,
      "grad_norm": 1.5843068125757378,
      "learning_rate": 3.141052630172294e-06,
      "loss": 0.7151,
      "step": 8622
    },
    {
      "epoch": 0.6326659757696929,
      "grad_norm": 1.3563905535342662,
      "learning_rate": 3.139949659187073e-06,
      "loss": 0.7,
      "step": 8623
    },
    {
      "epoch": 0.632739345359832,
      "grad_norm": 0.5786178474785943,
      "learning_rate": 3.138846793243502e-06,
      "loss": 0.482,
      "step": 8624
    },
    {
      "epoch": 0.632739345359832,
      "lm_loss": 0.461,
      "step": 8624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.632739345359832,
      "lm_loss": 0.3419,
      "step": 8624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.632739345359832,
      "lm_loss": 0.6007,
      "step": 8624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.632739345359832,
      "lm_loss": 0.5044,
      "step": 8624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.632739345359832,
      "lm_loss": 0.4809,
      "step": 8624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.632739345359832,
      "lm_loss": 0.432,
      "step": 8624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.632739345359832,
      "lm_loss": 0.4816,
      "step": 8624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.632739345359832,
      "lm_loss": 0.5422,
      "step": 8624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6328127149499712,
      "grad_norm": 0.5571792245868082,
      "learning_rate": 3.13774403240386e-06,
      "loss": 0.4934,
      "step": 8625
    },
    {
      "epoch": 0.6328860845401102,
      "grad_norm": 1.5257962862655814,
      "learning_rate": 3.136641376730426e-06,
      "loss": 0.7728,
      "step": 8626
    },
    {
      "epoch": 0.6329594541302493,
      "grad_norm": 1.5436914352394915,
      "learning_rate": 3.1355388262854657e-06,
      "loss": 0.7929,
      "step": 8627
    },
    {
      "epoch": 0.6330328237203885,
      "grad_norm": 0.5739665268058889,
      "learning_rate": 3.134436381131247e-06,
      "loss": 0.5137,
      "step": 8628
    },
    {
      "epoch": 0.6331061933105276,
      "grad_norm": 1.5833710420515694,
      "learning_rate": 3.1333340413300263e-06,
      "loss": 0.7051,
      "step": 8629
    },
    {
      "epoch": 0.6331795629006668,
      "grad_norm": 1.3988811380008532,
      "learning_rate": 3.132231806944054e-06,
      "loss": 0.6844,
      "step": 8630
    },
    {
      "epoch": 0.6332529324908058,
      "grad_norm": 1.64506243621862,
      "learning_rate": 3.131129678035577e-06,
      "loss": 0.7915,
      "step": 8631
    },
    {
      "epoch": 0.633326302080945,
      "grad_norm": 1.7611841593664928,
      "learning_rate": 3.1300276546668363e-06,
      "loss": 0.8321,
      "step": 8632
    },
    {
      "epoch": 0.633326302080945,
      "lm_loss": 1.7415,
      "step": 8632,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.633326302080945,
      "lm_loss": 0.4642,
      "step": 8632,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.633326302080945,
      "lm_loss": 0.9516,
      "step": 8632,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.633326302080945,
      "lm_loss": 0.3813,
      "step": 8632,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.633326302080945,
      "lm_loss": 0.3563,
      "step": 8632,
      "vm_loss": 0.1663
    },
    {
      "epoch": 0.633326302080945,
      "lm_loss": 0.914,
      "step": 8632,
      "vm_loss": 0.1301
    },
    {
      "epoch": 0.633326302080945,
      "lm_loss": 0.6259,
      "step": 8632,
      "vm_loss": 0.1261
    },
    {
      "epoch": 0.633326302080945,
      "lm_loss": 0.4198,
      "step": 8632,
      "vm_loss": 0.1115
    },
    {
      "epoch": 0.6333996716710841,
      "grad_norm": 1.5471949610428688,
      "learning_rate": 3.128925736900063e-06,
      "loss": 0.724,
      "step": 8633
    },
    {
      "epoch": 0.6334730412612233,
      "grad_norm": 1.5949550155976908,
      "learning_rate": 3.1278239247974886e-06,
      "loss": 0.8187,
      "step": 8634
    },
    {
      "epoch": 0.6335464108513624,
      "grad_norm": 0.5771355626820066,
      "learning_rate": 3.1267222184213332e-06,
      "loss": 0.5026,
      "step": 8635
    },
    {
      "epoch": 0.6336197804415015,
      "grad_norm": 1.4664661470627485,
      "learning_rate": 3.1256206178338124e-06,
      "loss": 0.7351,
      "step": 8636
    },
    {
      "epoch": 0.6336931500316406,
      "grad_norm": 1.6850115647103954,
      "learning_rate": 3.1245191230971372e-06,
      "loss": 0.7878,
      "step": 8637
    },
    {
      "epoch": 0.6337665196217798,
      "grad_norm": 1.6085988465531893,
      "learning_rate": 3.123417734273512e-06,
      "loss": 0.8157,
      "step": 8638
    },
    {
      "epoch": 0.6338398892119189,
      "grad_norm": 1.7058357531324193,
      "learning_rate": 3.1223164514251345e-06,
      "loss": 0.7561,
      "step": 8639
    },
    {
      "epoch": 0.6339132588020581,
      "grad_norm": 1.6139094179950293,
      "learning_rate": 3.1212152746141976e-06,
      "loss": 0.8153,
      "step": 8640
    },
    {
      "epoch": 0.6339132588020581,
      "lm_loss": 0.6012,
      "step": 8640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6339132588020581,
      "lm_loss": 0.346,
      "step": 8640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6339132588020581,
      "lm_loss": 0.3026,
      "step": 8640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6339132588020581,
      "lm_loss": 0.3485,
      "step": 8640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6339132588020581,
      "lm_loss": 0.5368,
      "step": 8640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6339132588020581,
      "lm_loss": 0.4976,
      "step": 8640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6339132588020581,
      "lm_loss": 0.4262,
      "step": 8640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6339132588020581,
      "lm_loss": 0.49,
      "step": 8640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6339866283921971,
      "grad_norm": 0.5786594554785291,
      "learning_rate": 3.1201142039028863e-06,
      "loss": 0.477,
      "step": 8641
    },
    {
      "epoch": 0.6340599979823363,
      "grad_norm": 1.5949370610607294,
      "learning_rate": 3.119013239353382e-06,
      "loss": 0.8234,
      "step": 8642
    },
    {
      "epoch": 0.6341333675724754,
      "grad_norm": 1.847824295872118,
      "learning_rate": 3.1179123810278582e-06,
      "loss": 0.7678,
      "step": 8643
    },
    {
      "epoch": 0.6342067371626146,
      "grad_norm": 1.4107043484055204,
      "learning_rate": 3.116811628988482e-06,
      "loss": 0.6736,
      "step": 8644
    },
    {
      "epoch": 0.6342801067527537,
      "grad_norm": 0.5587410445208759,
      "learning_rate": 3.115710983297418e-06,
      "loss": 0.4954,
      "step": 8645
    },
    {
      "epoch": 0.6343534763428927,
      "grad_norm": 0.5824436873794316,
      "learning_rate": 3.11461044401682e-06,
      "loss": 0.4812,
      "step": 8646
    },
    {
      "epoch": 0.6344268459330319,
      "grad_norm": 1.5208861810538572,
      "learning_rate": 3.11351001120884e-06,
      "loss": 0.7415,
      "step": 8647
    },
    {
      "epoch": 0.634500215523171,
      "grad_norm": 1.5229668344248704,
      "learning_rate": 3.112409684935622e-06,
      "loss": 0.7727,
      "step": 8648
    },
    {
      "epoch": 0.634500215523171,
      "lm_loss": 1.5395,
      "step": 8648,
      "vm_loss": 0.115
    },
    {
      "epoch": 0.634500215523171,
      "lm_loss": 1.3952,
      "step": 8648,
      "vm_loss": 0.1216
    },
    {
      "epoch": 0.634500215523171,
      "lm_loss": 0.9864,
      "step": 8648,
      "vm_loss": 0.1959
    },
    {
      "epoch": 0.634500215523171,
      "lm_loss": 1.0624,
      "step": 8648,
      "vm_loss": 0.2182
    },
    {
      "epoch": 0.634500215523171,
      "lm_loss": 0.9847,
      "step": 8648,
      "vm_loss": 0.219
    },
    {
      "epoch": 0.634500215523171,
      "lm_loss": 0.3179,
      "step": 8648,
      "vm_loss": 0.0883
    },
    {
      "epoch": 0.634500215523171,
      "lm_loss": 0.1429,
      "step": 8648,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.634500215523171,
      "lm_loss": 0.7885,
      "step": 8648,
      "vm_loss": 0.076
    },
    {
      "epoch": 0.6345735851133102,
      "grad_norm": 1.4716564876280165,
      "learning_rate": 3.1113094652593023e-06,
      "loss": 0.8232,
      "step": 8649
    },
    {
      "epoch": 0.6346469547034493,
      "grad_norm": 1.4341977678810591,
      "learning_rate": 3.1102093522420163e-06,
      "loss": 0.7709,
      "step": 8650
    },
    {
      "epoch": 0.6347203242935884,
      "grad_norm": 1.6925177650302794,
      "learning_rate": 3.109109345945887e-06,
      "loss": 0.742,
      "step": 8651
    },
    {
      "epoch": 0.6347936938837275,
      "grad_norm": 0.5676512989816914,
      "learning_rate": 3.1080094464330353e-06,
      "loss": 0.4929,
      "step": 8652
    },
    {
      "epoch": 0.6348670634738667,
      "grad_norm": 1.45064387450857,
      "learning_rate": 3.1069096537655764e-06,
      "loss": 0.8324,
      "step": 8653
    },
    {
      "epoch": 0.6349404330640058,
      "grad_norm": 1.481700307055817,
      "learning_rate": 3.1058099680056164e-06,
      "loss": 0.7198,
      "step": 8654
    },
    {
      "epoch": 0.635013802654145,
      "grad_norm": 1.5297347288800678,
      "learning_rate": 3.104710389215259e-06,
      "loss": 0.7811,
      "step": 8655
    },
    {
      "epoch": 0.635087172244284,
      "grad_norm": 0.5703655124347021,
      "learning_rate": 3.1036109174566e-06,
      "loss": 0.4991,
      "step": 8656
    },
    {
      "epoch": 0.635087172244284,
      "lm_loss": 0.5111,
      "step": 8656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.635087172244284,
      "lm_loss": 0.5497,
      "step": 8656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.635087172244284,
      "lm_loss": 0.4857,
      "step": 8656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.635087172244284,
      "lm_loss": 0.4687,
      "step": 8656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.635087172244284,
      "lm_loss": 0.4439,
      "step": 8656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.635087172244284,
      "lm_loss": 0.4955,
      "step": 8656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.635087172244284,
      "lm_loss": 0.5237,
      "step": 8656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.635087172244284,
      "lm_loss": 0.4233,
      "step": 8656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6351605418344232,
      "grad_norm": 0.5421713374357022,
      "learning_rate": 3.1025115527917284e-06,
      "loss": 0.4983,
      "step": 8657
    },
    {
      "epoch": 0.6352339114245623,
      "grad_norm": 1.4296772820368104,
      "learning_rate": 3.1014122952827285e-06,
      "loss": 0.6964,
      "step": 8658
    },
    {
      "epoch": 0.6353072810147015,
      "grad_norm": 1.5337388553241078,
      "learning_rate": 3.100313144991678e-06,
      "loss": 0.6153,
      "step": 8659
    },
    {
      "epoch": 0.6353806506048406,
      "grad_norm": 1.4250884389074654,
      "learning_rate": 3.0992141019806483e-06,
      "loss": 0.735,
      "step": 8660
    },
    {
      "epoch": 0.6354540201949797,
      "grad_norm": 0.5610828590604463,
      "learning_rate": 3.0981151663117058e-06,
      "loss": 0.4826,
      "step": 8661
    },
    {
      "epoch": 0.6355273897851188,
      "grad_norm": 0.5686916229480389,
      "learning_rate": 3.0970163380469107e-06,
      "loss": 0.498,
      "step": 8662
    },
    {
      "epoch": 0.635600759375258,
      "grad_norm": 1.5881026559093196,
      "learning_rate": 3.095917617248314e-06,
      "loss": 0.6165,
      "step": 8663
    },
    {
      "epoch": 0.6356741289653971,
      "grad_norm": 1.5256569564695954,
      "learning_rate": 3.0948190039779665e-06,
      "loss": 0.8417,
      "step": 8664
    },
    {
      "epoch": 0.6356741289653971,
      "lm_loss": 1.2755,
      "step": 8664,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.6356741289653971,
      "lm_loss": 1.6916,
      "step": 8664,
      "vm_loss": 0.1547
    },
    {
      "epoch": 0.6356741289653971,
      "lm_loss": 0.806,
      "step": 8664,
      "vm_loss": 0.1271
    },
    {
      "epoch": 0.6356741289653971,
      "lm_loss": 0.8669,
      "step": 8664,
      "vm_loss": 0.2005
    },
    {
      "epoch": 0.6356741289653971,
      "lm_loss": 0.0897,
      "step": 8664,
      "vm_loss": 0.1448
    },
    {
      "epoch": 0.6356741289653971,
      "lm_loss": 0.5084,
      "step": 8664,
      "vm_loss": 0.1369
    },
    {
      "epoch": 0.6356741289653971,
      "lm_loss": 0.9215,
      "step": 8664,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.6356741289653971,
      "lm_loss": 0.1054,
      "step": 8664,
      "vm_loss": 0.0742
    },
    {
      "epoch": 0.6357474985555361,
      "grad_norm": 1.4268016721576964,
      "learning_rate": 3.0937204982979063e-06,
      "loss": 0.8035,
      "step": 8665
    },
    {
      "epoch": 0.6358208681456753,
      "grad_norm": 1.423723143936862,
      "learning_rate": 3.0926221002701717e-06,
      "loss": 0.7301,
      "step": 8666
    },
    {
      "epoch": 0.6358942377358144,
      "grad_norm": 0.550669924856615,
      "learning_rate": 3.09152380995679e-06,
      "loss": 0.4782,
      "step": 8667
    },
    {
      "epoch": 0.6359676073259536,
      "grad_norm": 1.5185542377892984,
      "learning_rate": 3.0904256274197853e-06,
      "loss": 0.7169,
      "step": 8668
    },
    {
      "epoch": 0.6360409769160927,
      "grad_norm": 0.5426150431554956,
      "learning_rate": 3.0893275527211742e-06,
      "loss": 0.4794,
      "step": 8669
    },
    {
      "epoch": 0.6361143465062318,
      "grad_norm": 1.3286297509963998,
      "learning_rate": 3.088229585922968e-06,
      "loss": 0.6484,
      "step": 8670
    },
    {
      "epoch": 0.6361877160963709,
      "grad_norm": 1.4899125219578948,
      "learning_rate": 3.0871317270871716e-06,
      "loss": 0.828,
      "step": 8671
    },
    {
      "epoch": 0.6362610856865101,
      "grad_norm": 0.5692180950968588,
      "learning_rate": 3.086033976275784e-06,
      "loss": 0.4962,
      "step": 8672
    },
    {
      "epoch": 0.6362610856865101,
      "lm_loss": 1.2476,
      "step": 8672,
      "vm_loss": 0.0299
    },
    {
      "epoch": 0.6362610856865101,
      "lm_loss": 1.1533,
      "step": 8672,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.6362610856865101,
      "lm_loss": 0.9639,
      "step": 8672,
      "vm_loss": 0.1513
    },
    {
      "epoch": 0.6362610856865101,
      "lm_loss": 0.4844,
      "step": 8672,
      "vm_loss": 0.0785
    },
    {
      "epoch": 0.6362610856865101,
      "lm_loss": 0.9919,
      "step": 8672,
      "vm_loss": 0.182
    },
    {
      "epoch": 0.6362610856865101,
      "lm_loss": 1.0244,
      "step": 8672,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.6362610856865101,
      "lm_loss": 0.2139,
      "step": 8672,
      "vm_loss": 0.1252
    },
    {
      "epoch": 0.6362610856865101,
      "lm_loss": 0.7035,
      "step": 8672,
      "vm_loss": 0.2017
    },
    {
      "epoch": 0.6363344552766492,
      "grad_norm": 1.5273507217242641,
      "learning_rate": 3.0849363335507966e-06,
      "loss": 0.9068,
      "step": 8673
    },
    {
      "epoch": 0.6364078248667884,
      "grad_norm": 1.607571655789728,
      "learning_rate": 3.083838798974198e-06,
      "loss": 0.7266,
      "step": 8674
    },
    {
      "epoch": 0.6364811944569274,
      "grad_norm": 1.4285954092657585,
      "learning_rate": 3.0827413726079675e-06,
      "loss": 0.7644,
      "step": 8675
    },
    {
      "epoch": 0.6365545640470666,
      "grad_norm": 0.5487754895432454,
      "learning_rate": 3.081644054514079e-06,
      "loss": 0.4975,
      "step": 8676
    },
    {
      "epoch": 0.6366279336372057,
      "grad_norm": 1.4284724282507395,
      "learning_rate": 3.080546844754502e-06,
      "loss": 0.7568,
      "step": 8677
    },
    {
      "epoch": 0.6367013032273449,
      "grad_norm": 1.5750235233794652,
      "learning_rate": 3.0794497433911986e-06,
      "loss": 0.7992,
      "step": 8678
    },
    {
      "epoch": 0.636774672817484,
      "grad_norm": 1.4563501640470482,
      "learning_rate": 3.078352750486123e-06,
      "loss": 0.6622,
      "step": 8679
    },
    {
      "epoch": 0.636848042407623,
      "grad_norm": 1.5594224180100782,
      "learning_rate": 3.077255866101228e-06,
      "loss": 0.7199,
      "step": 8680
    },
    {
      "epoch": 0.636848042407623,
      "lm_loss": 2.076,
      "step": 8680,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.636848042407623,
      "lm_loss": 0.9447,
      "step": 8680,
      "vm_loss": 0.1316
    },
    {
      "epoch": 0.636848042407623,
      "lm_loss": 0.0373,
      "step": 8680,
      "vm_loss": 0.1149
    },
    {
      "epoch": 0.636848042407623,
      "lm_loss": 0.8149,
      "step": 8680,
      "vm_loss": 0.1233
    },
    {
      "epoch": 0.636848042407623,
      "lm_loss": 0.4792,
      "step": 8680,
      "vm_loss": 0.1672
    },
    {
      "epoch": 0.636848042407623,
      "lm_loss": 1.7613,
      "step": 8680,
      "vm_loss": 0.1927
    },
    {
      "epoch": 0.636848042407623,
      "lm_loss": 0.4189,
      "step": 8680,
      "vm_loss": 0.1456
    },
    {
      "epoch": 0.636848042407623,
      "lm_loss": 0.2255,
      "step": 8680,
      "vm_loss": 0.1858
    },
    {
      "epoch": 0.6369214119977622,
      "grad_norm": 1.7729345826197689,
      "learning_rate": 3.0761590902984545e-06,
      "loss": 0.7693,
      "step": 8681
    },
    {
      "epoch": 0.6369947815879013,
      "grad_norm": 1.5233884288999884,
      "learning_rate": 3.0750624231397424e-06,
      "loss": 0.8106,
      "step": 8682
    },
    {
      "epoch": 0.6370681511780405,
      "grad_norm": 1.5293318032517378,
      "learning_rate": 3.0739658646870225e-06,
      "loss": 0.8689,
      "step": 8683
    },
    {
      "epoch": 0.6371415207681796,
      "grad_norm": 1.547274488032128,
      "learning_rate": 3.0728694150022188e-06,
      "loss": 0.8612,
      "step": 8684
    },
    {
      "epoch": 0.6372148903583187,
      "grad_norm": 1.3665778027571422,
      "learning_rate": 3.0717730741472528e-06,
      "loss": 0.7622,
      "step": 8685
    },
    {
      "epoch": 0.6372882599484578,
      "grad_norm": 1.4166994463576228,
      "learning_rate": 3.0706768421840367e-06,
      "loss": 0.7374,
      "step": 8686
    },
    {
      "epoch": 0.637361629538597,
      "grad_norm": 0.5919156793622621,
      "learning_rate": 3.069580719174476e-06,
      "loss": 0.4946,
      "step": 8687
    },
    {
      "epoch": 0.6374349991287361,
      "grad_norm": 0.5665962445309735,
      "learning_rate": 3.0684847051804733e-06,
      "loss": 0.489,
      "step": 8688
    },
    {
      "epoch": 0.6374349991287361,
      "lm_loss": 0.8412,
      "step": 8688,
      "vm_loss": 0.0552
    },
    {
      "epoch": 0.6374349991287361,
      "lm_loss": 0.1142,
      "step": 8688,
      "vm_loss": 0.198
    },
    {
      "epoch": 0.6374349991287361,
      "lm_loss": 0.1351,
      "step": 8688,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.6374349991287361,
      "lm_loss": 0.7533,
      "step": 8688,
      "vm_loss": 0.126
    },
    {
      "epoch": 0.6374349991287361,
      "lm_loss": 0.8718,
      "step": 8688,
      "vm_loss": 0.2312
    },
    {
      "epoch": 0.6374349991287361,
      "lm_loss": 0.0485,
      "step": 8688,
      "vm_loss": 0.094
    },
    {
      "epoch": 0.6374349991287361,
      "lm_loss": 0.3752,
      "step": 8688,
      "vm_loss": 0.2199
    },
    {
      "epoch": 0.6374349991287361,
      "lm_loss": 0.1538,
      "step": 8688,
      "vm_loss": 0.0391
    },
    {
      "epoch": 0.6375083687188753,
      "grad_norm": 1.3770821064752057,
      "learning_rate": 3.067388800263923e-06,
      "loss": 0.7079,
      "step": 8689
    },
    {
      "epoch": 0.6375817383090143,
      "grad_norm": 1.4168710896093952,
      "learning_rate": 3.066293004486712e-06,
      "loss": 0.6783,
      "step": 8690
    },
    {
      "epoch": 0.6376551078991535,
      "grad_norm": 1.480989275175506,
      "learning_rate": 3.0651973179107254e-06,
      "loss": 0.6783,
      "step": 8691
    },
    {
      "epoch": 0.6377284774892926,
      "grad_norm": 1.5767473254331512,
      "learning_rate": 3.064101740597836e-06,
      "loss": 0.719,
      "step": 8692
    },
    {
      "epoch": 0.6378018470794318,
      "grad_norm": 0.5483683737331947,
      "learning_rate": 3.0630062726099163e-06,
      "loss": 0.4836,
      "step": 8693
    },
    {
      "epoch": 0.6378752166695709,
      "grad_norm": 0.5624997396043014,
      "learning_rate": 3.0619109140088297e-06,
      "loss": 0.4835,
      "step": 8694
    },
    {
      "epoch": 0.63794858625971,
      "grad_norm": 1.4299148705177689,
      "learning_rate": 3.0608156648564325e-06,
      "loss": 0.7371,
      "step": 8695
    },
    {
      "epoch": 0.6380219558498491,
      "grad_norm": 0.5395302805968517,
      "learning_rate": 3.059720525214578e-06,
      "loss": 0.4722,
      "step": 8696
    },
    {
      "epoch": 0.6380219558498491,
      "lm_loss": 0.2026,
      "step": 8696,
      "vm_loss": 0.0353
    },
    {
      "epoch": 0.6380219558498491,
      "lm_loss": 1.1534,
      "step": 8696,
      "vm_loss": 0.1468
    },
    {
      "epoch": 0.6380219558498491,
      "lm_loss": 1.0808,
      "step": 8696,
      "vm_loss": 0.1844
    },
    {
      "epoch": 0.6380219558498491,
      "lm_loss": 0.6805,
      "step": 8696,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.6380219558498491,
      "lm_loss": 0.2308,
      "step": 8696,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.6380219558498491,
      "lm_loss": 0.0154,
      "step": 8696,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.6380219558498491,
      "lm_loss": 0.3693,
      "step": 8696,
      "vm_loss": 0.1312
    },
    {
      "epoch": 0.6380219558498491,
      "lm_loss": 1.1651,
      "step": 8696,
      "vm_loss": 0.1841
    },
    {
      "epoch": 0.6380953254399883,
      "grad_norm": 1.4115893421783199,
      "learning_rate": 3.0586254951451104e-06,
      "loss": 0.7809,
      "step": 8697
    },
    {
      "epoch": 0.6381686950301274,
      "grad_norm": 1.4253573773980799,
      "learning_rate": 3.0575305747098683e-06,
      "loss": 0.7053,
      "step": 8698
    },
    {
      "epoch": 0.6382420646202666,
      "grad_norm": 0.5455493966084078,
      "learning_rate": 3.0564357639706853e-06,
      "loss": 0.476,
      "step": 8699
    },
    {
      "epoch": 0.6383154342104056,
      "grad_norm": 1.4093874519814034,
      "learning_rate": 3.055341062989387e-06,
      "loss": 0.7453,
      "step": 8700
    },
    {
      "epoch": 0.6383888038005447,
      "grad_norm": 1.6631100440086812,
      "learning_rate": 3.054246471827796e-06,
      "loss": 0.8511,
      "step": 8701
    },
    {
      "epoch": 0.6384621733906839,
      "grad_norm": 0.5797392041403877,
      "learning_rate": 3.053151990547725e-06,
      "loss": 0.4857,
      "step": 8702
    },
    {
      "epoch": 0.638535542980823,
      "grad_norm": 1.4263777527468509,
      "learning_rate": 3.052057619210982e-06,
      "loss": 0.7321,
      "step": 8703
    },
    {
      "epoch": 0.6386089125709622,
      "grad_norm": 1.6856515819934992,
      "learning_rate": 3.0509633578793698e-06,
      "loss": 0.8029,
      "step": 8704
    },
    {
      "epoch": 0.6386089125709622,
      "lm_loss": 1.2179,
      "step": 8704,
      "vm_loss": 0.1219
    },
    {
      "epoch": 0.6386089125709622,
      "lm_loss": 0.6665,
      "step": 8704,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.6386089125709622,
      "lm_loss": 1.1691,
      "step": 8704,
      "vm_loss": 0.1751
    },
    {
      "epoch": 0.6386089125709622,
      "lm_loss": 0.9407,
      "step": 8704,
      "vm_loss": 0.191
    },
    {
      "epoch": 0.6386089125709622,
      "lm_loss": 0.3764,
      "step": 8704,
      "vm_loss": 0.1224
    },
    {
      "epoch": 0.6386089125709622,
      "lm_loss": 0.8619,
      "step": 8704,
      "vm_loss": 0.0779
    },
    {
      "epoch": 0.6386089125709622,
      "lm_loss": 0.0947,
      "step": 8704,
      "vm_loss": 0.1143
    },
    {
      "epoch": 0.6386089125709622,
      "lm_loss": 0.0789,
      "step": 8704,
      "vm_loss": 0.1381
    },
    {
      "epoch": 0.6386822821611012,
      "grad_norm": 1.4128093219399784,
      "learning_rate": 3.049869206614683e-06,
      "loss": 0.7422,
      "step": 8705
    },
    {
      "epoch": 0.6387556517512404,
      "grad_norm": 1.62771999278561,
      "learning_rate": 3.048775165478711e-06,
      "loss": 0.7793,
      "step": 8706
    },
    {
      "epoch": 0.6388290213413795,
      "grad_norm": 1.4980625204492606,
      "learning_rate": 3.0476812345332385e-06,
      "loss": 0.7502,
      "step": 8707
    },
    {
      "epoch": 0.6389023909315187,
      "grad_norm": 1.4011765460544308,
      "learning_rate": 3.046587413840041e-06,
      "loss": 0.7776,
      "step": 8708
    },
    {
      "epoch": 0.6389757605216578,
      "grad_norm": 1.5873908170626982,
      "learning_rate": 3.04549370346089e-06,
      "loss": 0.8296,
      "step": 8709
    },
    {
      "epoch": 0.6390491301117969,
      "grad_norm": 1.4087166507218691,
      "learning_rate": 3.04440010345755e-06,
      "loss": 0.6989,
      "step": 8710
    },
    {
      "epoch": 0.639122499701936,
      "grad_norm": 1.5452447276565644,
      "learning_rate": 3.0433066138917786e-06,
      "loss": 0.691,
      "step": 8711
    },
    {
      "epoch": 0.6391958692920752,
      "grad_norm": 1.6227823822081413,
      "learning_rate": 3.0422132348253292e-06,
      "loss": 0.7435,
      "step": 8712
    },
    {
      "epoch": 0.6391958692920752,
      "lm_loss": 1.6518,
      "step": 8712,
      "vm_loss": 0.0894
    },
    {
      "epoch": 0.6391958692920752,
      "lm_loss": 1.5499,
      "step": 8712,
      "vm_loss": 0.0807
    },
    {
      "epoch": 0.6391958692920752,
      "lm_loss": 0.4537,
      "step": 8712,
      "vm_loss": 0.1669
    },
    {
      "epoch": 0.6391958692920752,
      "lm_loss": 0.4471,
      "step": 8712,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.6391958692920752,
      "lm_loss": 0.1068,
      "step": 8712,
      "vm_loss": 0.0801
    },
    {
      "epoch": 0.6391958692920752,
      "lm_loss": 0.4372,
      "step": 8712,
      "vm_loss": 0.1149
    },
    {
      "epoch": 0.6391958692920752,
      "lm_loss": 0.8081,
      "step": 8712,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.6391958692920752,
      "lm_loss": 0.1622,
      "step": 8712,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.6392692388822143,
      "grad_norm": 1.4195393410791728,
      "learning_rate": 3.0411199663199466e-06,
      "loss": 0.7292,
      "step": 8713
    },
    {
      "epoch": 0.6393426084723535,
      "grad_norm": 1.7057733361156417,
      "learning_rate": 3.0400268084373697e-06,
      "loss": 0.7937,
      "step": 8714
    },
    {
      "epoch": 0.6394159780624925,
      "grad_norm": 1.519290576985616,
      "learning_rate": 3.038933761239333e-06,
      "loss": 0.7288,
      "step": 8715
    },
    {
      "epoch": 0.6394893476526317,
      "grad_norm": 1.4101300848147786,
      "learning_rate": 3.0378408247875647e-06,
      "loss": 0.7942,
      "step": 8716
    },
    {
      "epoch": 0.6395627172427708,
      "grad_norm": 1.7024084849033718,
      "learning_rate": 3.036747999143782e-06,
      "loss": 0.6973,
      "step": 8717
    },
    {
      "epoch": 0.63963608683291,
      "grad_norm": 1.433125116387071,
      "learning_rate": 3.0356552843697036e-06,
      "loss": 0.7759,
      "step": 8718
    },
    {
      "epoch": 0.6397094564230491,
      "grad_norm": 0.5731875244944025,
      "learning_rate": 3.0345626805270343e-06,
      "loss": 0.4911,
      "step": 8719
    },
    {
      "epoch": 0.6397828260131881,
      "grad_norm": 1.4496489368096053,
      "learning_rate": 3.033470187677479e-06,
      "loss": 0.7941,
      "step": 8720
    },
    {
      "epoch": 0.6397828260131881,
      "lm_loss": 1.1619,
      "step": 8720,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.6397828260131881,
      "lm_loss": 0.9211,
      "step": 8720,
      "vm_loss": 0.1576
    },
    {
      "epoch": 0.6397828260131881,
      "lm_loss": 0.9046,
      "step": 8720,
      "vm_loss": 0.1564
    },
    {
      "epoch": 0.6397828260131881,
      "lm_loss": 0.5865,
      "step": 8720,
      "vm_loss": 0.1461
    },
    {
      "epoch": 0.6397828260131881,
      "lm_loss": 0.8242,
      "step": 8720,
      "vm_loss": 0.1276
    },
    {
      "epoch": 0.6397828260131881,
      "lm_loss": 1.4664,
      "step": 8720,
      "vm_loss": 0.0872
    },
    {
      "epoch": 0.6397828260131881,
      "lm_loss": 0.4139,
      "step": 8720,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.6397828260131881,
      "lm_loss": 0.507,
      "step": 8720,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.6398561956033273,
      "grad_norm": 1.3620240975139684,
      "learning_rate": 3.0323778058827322e-06,
      "loss": 0.7913,
      "step": 8721
    },
    {
      "epoch": 0.6399295651934664,
      "grad_norm": 1.5469129006241205,
      "learning_rate": 3.0312855352044825e-06,
      "loss": 0.703,
      "step": 8722
    },
    {
      "epoch": 0.6400029347836056,
      "grad_norm": 0.5813042264756945,
      "learning_rate": 3.0301933757044155e-06,
      "loss": 0.4892,
      "step": 8723
    },
    {
      "epoch": 0.6400763043737447,
      "grad_norm": 0.5500252403339105,
      "learning_rate": 3.029101327444206e-06,
      "loss": 0.5011,
      "step": 8724
    },
    {
      "epoch": 0.6401496739638838,
      "grad_norm": 1.513668281702591,
      "learning_rate": 3.0280093904855244e-06,
      "loss": 0.8343,
      "step": 8725
    },
    {
      "epoch": 0.6402230435540229,
      "grad_norm": 1.5397269803950737,
      "learning_rate": 3.026917564890037e-06,
      "loss": 0.7652,
      "step": 8726
    },
    {
      "epoch": 0.6402964131441621,
      "grad_norm": 1.5898522244442346,
      "learning_rate": 3.0258258507194006e-06,
      "loss": 0.8016,
      "step": 8727
    },
    {
      "epoch": 0.6403697827343012,
      "grad_norm": 1.3079660258528492,
      "learning_rate": 3.024734248035268e-06,
      "loss": 0.6272,
      "step": 8728
    },
    {
      "epoch": 0.6403697827343012,
      "lm_loss": 1.7204,
      "step": 8728,
      "vm_loss": 0.0762
    },
    {
      "epoch": 0.6403697827343012,
      "lm_loss": 0.9689,
      "step": 8728,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.6403697827343012,
      "lm_loss": 0.23,
      "step": 8728,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.6403697827343012,
      "lm_loss": 0.5532,
      "step": 8728,
      "vm_loss": 0.1094
    },
    {
      "epoch": 0.6403697827343012,
      "lm_loss": 0.2129,
      "step": 8728,
      "vm_loss": 0.105
    },
    {
      "epoch": 0.6403697827343012,
      "lm_loss": 0.7032,
      "step": 8728,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.6403697827343012,
      "lm_loss": 0.1969,
      "step": 8728,
      "vm_loss": 0.1981
    },
    {
      "epoch": 0.6403697827343012,
      "lm_loss": 0.2888,
      "step": 8728,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.6404431523244404,
      "grad_norm": 1.5743036850331704,
      "learning_rate": 3.0236427568992845e-06,
      "loss": 0.7189,
      "step": 8729
    },
    {
      "epoch": 0.6405165219145794,
      "grad_norm": 1.516331373646308,
      "learning_rate": 3.0225513773730873e-06,
      "loss": 0.845,
      "step": 8730
    },
    {
      "epoch": 0.6405898915047186,
      "grad_norm": 1.5130058977908545,
      "learning_rate": 3.0214601095183116e-06,
      "loss": 0.7391,
      "step": 8731
    },
    {
      "epoch": 0.6406632610948577,
      "grad_norm": 1.6054706381015655,
      "learning_rate": 3.020368953396584e-06,
      "loss": 0.6642,
      "step": 8732
    },
    {
      "epoch": 0.6407366306849969,
      "grad_norm": 0.5909158896189881,
      "learning_rate": 3.019277909069523e-06,
      "loss": 0.4892,
      "step": 8733
    },
    {
      "epoch": 0.640810000275136,
      "grad_norm": 0.5749028186478893,
      "learning_rate": 3.0181869765987447e-06,
      "loss": 0.469,
      "step": 8734
    },
    {
      "epoch": 0.640883369865275,
      "grad_norm": 1.3989690252809488,
      "learning_rate": 3.0170961560458545e-06,
      "loss": 0.8318,
      "step": 8735
    },
    {
      "epoch": 0.6409567394554142,
      "grad_norm": 1.5099334969383624,
      "learning_rate": 3.0160054474724555e-06,
      "loss": 0.7546,
      "step": 8736
    },
    {
      "epoch": 0.6409567394554142,
      "lm_loss": 1.1993,
      "step": 8736,
      "vm_loss": 0.0566
    },
    {
      "epoch": 0.6409567394554142,
      "lm_loss": 1.18,
      "step": 8736,
      "vm_loss": 0.1682
    },
    {
      "epoch": 0.6409567394554142,
      "lm_loss": 0.5473,
      "step": 8736,
      "vm_loss": 0.1651
    },
    {
      "epoch": 0.6409567394554142,
      "lm_loss": 0.6518,
      "step": 8736,
      "vm_loss": 0.1872
    },
    {
      "epoch": 0.6409567394554142,
      "lm_loss": 0.5046,
      "step": 8736,
      "vm_loss": 0.1775
    },
    {
      "epoch": 0.6409567394554142,
      "lm_loss": 0.3062,
      "step": 8736,
      "vm_loss": 0.0309
    },
    {
      "epoch": 0.6409567394554142,
      "lm_loss": 0.7966,
      "step": 8736,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.6409567394554142,
      "lm_loss": 0.6498,
      "step": 8736,
      "vm_loss": 0.168
    },
    {
      "epoch": 0.6410301090455534,
      "grad_norm": 1.5356914090179348,
      "learning_rate": 3.0149148509401426e-06,
      "loss": 0.6919,
      "step": 8737
    },
    {
      "epoch": 0.6411034786356925,
      "grad_norm": 1.3780544688474545,
      "learning_rate": 3.0138243665105033e-06,
      "loss": 0.7436,
      "step": 8738
    },
    {
      "epoch": 0.6411768482258317,
      "grad_norm": 1.4000178994968744,
      "learning_rate": 3.012733994245122e-06,
      "loss": 0.7913,
      "step": 8739
    },
    {
      "epoch": 0.6412502178159707,
      "grad_norm": 1.4419787983416477,
      "learning_rate": 3.0116437342055725e-06,
      "loss": 0.736,
      "step": 8740
    },
    {
      "epoch": 0.6413235874061098,
      "grad_norm": 1.4147527272029243,
      "learning_rate": 3.010553586453425e-06,
      "loss": 0.7021,
      "step": 8741
    },
    {
      "epoch": 0.641396956996249,
      "grad_norm": 1.5372412792274002,
      "learning_rate": 3.0094635510502436e-06,
      "loss": 0.8269,
      "step": 8742
    },
    {
      "epoch": 0.6414703265863881,
      "grad_norm": 1.3511714753553912,
      "learning_rate": 3.0083736280575854e-06,
      "loss": 0.6223,
      "step": 8743
    },
    {
      "epoch": 0.6415436961765272,
      "grad_norm": 0.5656846820396765,
      "learning_rate": 3.0072838175369995e-06,
      "loss": 0.4924,
      "step": 8744
    },
    {
      "epoch": 0.6415436961765272,
      "lm_loss": 1.2068,
      "step": 8744,
      "vm_loss": 0.0785
    },
    {
      "epoch": 0.6415436961765272,
      "lm_loss": 0.9232,
      "step": 8744,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.6415436961765272,
      "lm_loss": 0.9397,
      "step": 8744,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.6415436961765272,
      "lm_loss": 0.9144,
      "step": 8744,
      "vm_loss": 0.1353
    },
    {
      "epoch": 0.6415436961765272,
      "lm_loss": 0.1005,
      "step": 8744,
      "vm_loss": 0.1033
    },
    {
      "epoch": 0.6415436961765272,
      "lm_loss": 0.7408,
      "step": 8744,
      "vm_loss": 0.1079
    },
    {
      "epoch": 0.6415436961765272,
      "lm_loss": 0.3915,
      "step": 8744,
      "vm_loss": 0.1481
    },
    {
      "epoch": 0.6415436961765272,
      "lm_loss": 0.4338,
      "step": 8744,
      "vm_loss": 0.0903
    },
    {
      "epoch": 0.6416170657666663,
      "grad_norm": 1.469714200620805,
      "learning_rate": 3.0061941195500323e-06,
      "loss": 0.721,
      "step": 8745
    },
    {
      "epoch": 0.6416904353568055,
      "grad_norm": 1.5705898802172618,
      "learning_rate": 3.00510453415822e-06,
      "loss": 0.704,
      "step": 8746
    },
    {
      "epoch": 0.6417638049469446,
      "grad_norm": 1.3790585374663473,
      "learning_rate": 3.004015061423098e-06,
      "loss": 0.6848,
      "step": 8747
    },
    {
      "epoch": 0.6418371745370838,
      "grad_norm": 1.3886413203142551,
      "learning_rate": 3.0029257014061854e-06,
      "loss": 0.6918,
      "step": 8748
    },
    {
      "epoch": 0.6419105441272228,
      "grad_norm": 1.5582271352840764,
      "learning_rate": 3.0018364541690048e-06,
      "loss": 0.8083,
      "step": 8749
    },
    {
      "epoch": 0.641983913717362,
      "grad_norm": 1.4623102069546792,
      "learning_rate": 3.000747319773067e-06,
      "loss": 0.7512,
      "step": 8750
    },
    {
      "epoch": 0.6420572833075011,
      "grad_norm": 0.5596184656171346,
      "learning_rate": 2.9996582982798804e-06,
      "loss": 0.4736,
      "step": 8751
    },
    {
      "epoch": 0.6421306528976403,
      "grad_norm": 1.5008241644518403,
      "learning_rate": 2.998569389750943e-06,
      "loss": 0.6974,
      "step": 8752
    },
    {
      "epoch": 0.6421306528976403,
      "lm_loss": 1.4958,
      "step": 8752,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.6421306528976403,
      "lm_loss": 0.1206,
      "step": 8752,
      "vm_loss": 0.082
    },
    {
      "epoch": 0.6421306528976403,
      "lm_loss": 0.1001,
      "step": 8752,
      "vm_loss": 0.1116
    },
    {
      "epoch": 0.6421306528976403,
      "lm_loss": 0.1023,
      "step": 8752,
      "vm_loss": 0.1131
    },
    {
      "epoch": 0.6421306528976403,
      "lm_loss": 0.4138,
      "step": 8752,
      "vm_loss": 0.241
    },
    {
      "epoch": 0.6421306528976403,
      "lm_loss": 0.4536,
      "step": 8752,
      "vm_loss": 0.1494
    },
    {
      "epoch": 0.6421306528976403,
      "lm_loss": 0.0428,
      "step": 8752,
      "vm_loss": 0.1596
    },
    {
      "epoch": 0.6421306528976403,
      "lm_loss": 0.0513,
      "step": 8752,
      "vm_loss": 0.0431
    },
    {
      "epoch": 0.6422040224877794,
      "grad_norm": 1.451900106196321,
      "learning_rate": 2.997480594247748e-06,
      "loss": 0.6587,
      "step": 8753
    },
    {
      "epoch": 0.6422773920779185,
      "grad_norm": 1.433698806606715,
      "learning_rate": 2.996391911831784e-06,
      "loss": 0.7845,
      "step": 8754
    },
    {
      "epoch": 0.6423507616680576,
      "grad_norm": 1.4474656556588048,
      "learning_rate": 2.9953033425645305e-06,
      "loss": 0.723,
      "step": 8755
    },
    {
      "epoch": 0.6424241312581968,
      "grad_norm": 1.536379240218032,
      "learning_rate": 2.994214886507461e-06,
      "loss": 0.8031,
      "step": 8756
    },
    {
      "epoch": 0.6424975008483359,
      "grad_norm": 0.5609753427850447,
      "learning_rate": 2.9931265437220448e-06,
      "loss": 0.4903,
      "step": 8757
    },
    {
      "epoch": 0.642570870438475,
      "grad_norm": 0.5418800889633448,
      "learning_rate": 2.992038314269743e-06,
      "loss": 0.4665,
      "step": 8758
    },
    {
      "epoch": 0.6426442400286141,
      "grad_norm": 1.4715633889319277,
      "learning_rate": 2.9909501982120088e-06,
      "loss": 0.6688,
      "step": 8759
    },
    {
      "epoch": 0.6427176096187532,
      "grad_norm": 1.6414265685090086,
      "learning_rate": 2.9898621956102937e-06,
      "loss": 0.8498,
      "step": 8760
    },
    {
      "epoch": 0.6427176096187532,
      "lm_loss": 1.3082,
      "step": 8760,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.6427176096187532,
      "lm_loss": 1.2736,
      "step": 8760,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.6427176096187532,
      "lm_loss": 1.1595,
      "step": 8760,
      "vm_loss": 0.1233
    },
    {
      "epoch": 0.6427176096187532,
      "lm_loss": 0.0309,
      "step": 8760,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.6427176096187532,
      "lm_loss": 0.1449,
      "step": 8760,
      "vm_loss": 0.0945
    },
    {
      "epoch": 0.6427176096187532,
      "lm_loss": 1.1165,
      "step": 8760,
      "vm_loss": 0.1528
    },
    {
      "epoch": 0.6427176096187532,
      "lm_loss": 0.1676,
      "step": 8760,
      "vm_loss": 0.0722
    },
    {
      "epoch": 0.6427176096187532,
      "lm_loss": 0.2604,
      "step": 8760,
      "vm_loss": 0.1986
    },
    {
      "epoch": 0.6427909792088924,
      "grad_norm": 1.5448485416423459,
      "learning_rate": 2.9887743065260365e-06,
      "loss": 0.7825,
      "step": 8761
    },
    {
      "epoch": 0.6428643487990315,
      "grad_norm": 1.5184733911201445,
      "learning_rate": 2.987686531020677e-06,
      "loss": 0.7744,
      "step": 8762
    },
    {
      "epoch": 0.6429377183891707,
      "grad_norm": 1.521586149365441,
      "learning_rate": 2.986598869155642e-06,
      "loss": 0.7808,
      "step": 8763
    },
    {
      "epoch": 0.6430110879793097,
      "grad_norm": 1.7553453166805102,
      "learning_rate": 2.985511320992354e-06,
      "loss": 0.8474,
      "step": 8764
    },
    {
      "epoch": 0.6430844575694489,
      "grad_norm": 0.5398375197893774,
      "learning_rate": 2.9844238865922317e-06,
      "loss": 0.4909,
      "step": 8765
    },
    {
      "epoch": 0.643157827159588,
      "grad_norm": 0.537028698186642,
      "learning_rate": 2.9833365660166833e-06,
      "loss": 0.4742,
      "step": 8766
    },
    {
      "epoch": 0.6432311967497272,
      "grad_norm": 0.587976924328851,
      "learning_rate": 2.9822493593271127e-06,
      "loss": 0.505,
      "step": 8767
    },
    {
      "epoch": 0.6433045663398663,
      "grad_norm": 0.5480713468210195,
      "learning_rate": 2.9811622665849195e-06,
      "loss": 0.4891,
      "step": 8768
    },
    {
      "epoch": 0.6433045663398663,
      "lm_loss": 1.1243,
      "step": 8768,
      "vm_loss": 0.034
    },
    {
      "epoch": 0.6433045663398663,
      "lm_loss": 0.2235,
      "step": 8768,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.6433045663398663,
      "lm_loss": 1.0173,
      "step": 8768,
      "vm_loss": 0.1917
    },
    {
      "epoch": 0.6433045663398663,
      "lm_loss": 0.951,
      "step": 8768,
      "vm_loss": 0.0833
    },
    {
      "epoch": 0.6433045663398663,
      "lm_loss": 1.186,
      "step": 8768,
      "vm_loss": 0.2358
    },
    {
      "epoch": 0.6433045663398663,
      "lm_loss": 0.1095,
      "step": 8768,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.6433045663398663,
      "lm_loss": 0.562,
      "step": 8768,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.6433045663398663,
      "lm_loss": 0.6137,
      "step": 8768,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.6433779359300054,
      "grad_norm": 1.5859674675855588,
      "learning_rate": 2.9800752878514903e-06,
      "loss": 0.7057,
      "step": 8769
    },
    {
      "epoch": 0.6434513055201445,
      "grad_norm": 1.6095267142306076,
      "learning_rate": 2.978988423188214e-06,
      "loss": 0.7923,
      "step": 8770
    },
    {
      "epoch": 0.6435246751102837,
      "grad_norm": 0.5516333627259451,
      "learning_rate": 2.977901672656466e-06,
      "loss": 0.4869,
      "step": 8771
    },
    {
      "epoch": 0.6435980447004228,
      "grad_norm": 0.5425331349288781,
      "learning_rate": 2.976815036317616e-06,
      "loss": 0.4952,
      "step": 8772
    },
    {
      "epoch": 0.643671414290562,
      "grad_norm": 0.5839339422026623,
      "learning_rate": 2.975728514233034e-06,
      "loss": 0.5026,
      "step": 8773
    },
    {
      "epoch": 0.643744783880701,
      "grad_norm": 1.5061478934415364,
      "learning_rate": 2.9746421064640753e-06,
      "loss": 0.798,
      "step": 8774
    },
    {
      "epoch": 0.6438181534708401,
      "grad_norm": 1.573555995518081,
      "learning_rate": 2.973555813072091e-06,
      "loss": 0.691,
      "step": 8775
    },
    {
      "epoch": 0.6438915230609793,
      "grad_norm": 0.5527884408739877,
      "learning_rate": 2.9724696341184293e-06,
      "loss": 0.4882,
      "step": 8776
    },
    {
      "epoch": 0.6438915230609793,
      "lm_loss": 1.3497,
      "step": 8776,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.6438915230609793,
      "lm_loss": 0.1496,
      "step": 8776,
      "vm_loss": 0.0568
    },
    {
      "epoch": 0.6438915230609793,
      "lm_loss": 0.2257,
      "step": 8776,
      "vm_loss": 0.158
    },
    {
      "epoch": 0.6438915230609793,
      "lm_loss": 0.1263,
      "step": 8776,
      "vm_loss": 0.1137
    },
    {
      "epoch": 0.6438915230609793,
      "lm_loss": 0.7951,
      "step": 8776,
      "vm_loss": 0.1526
    },
    {
      "epoch": 0.6438915230609793,
      "lm_loss": 0.5623,
      "step": 8776,
      "vm_loss": 0.1195
    },
    {
      "epoch": 0.6438915230609793,
      "lm_loss": 0.7617,
      "step": 8776,
      "vm_loss": 0.2303
    },
    {
      "epoch": 0.6438915230609793,
      "lm_loss": 1.1437,
      "step": 8776,
      "vm_loss": 0.1739
    },
    {
      "epoch": 0.6439648926511184,
      "grad_norm": 1.8357422793841869,
      "learning_rate": 2.971383569664428e-06,
      "loss": 0.703,
      "step": 8777
    },
    {
      "epoch": 0.6440382622412576,
      "grad_norm": 1.4596949483157233,
      "learning_rate": 2.970297619771421e-06,
      "loss": 0.6996,
      "step": 8778
    },
    {
      "epoch": 0.6441116318313966,
      "grad_norm": 1.5835692007120927,
      "learning_rate": 2.969211784500734e-06,
      "loss": 0.7386,
      "step": 8779
    },
    {
      "epoch": 0.6441850014215358,
      "grad_norm": 1.647487589003145,
      "learning_rate": 2.968126063913686e-06,
      "loss": 0.6394,
      "step": 8780
    },
    {
      "epoch": 0.6442583710116749,
      "grad_norm": 1.5472017312627946,
      "learning_rate": 2.967040458071592e-06,
      "loss": 0.7211,
      "step": 8781
    },
    {
      "epoch": 0.6443317406018141,
      "grad_norm": 1.5230135827988034,
      "learning_rate": 2.9659549670357574e-06,
      "loss": 0.7389,
      "step": 8782
    },
    {
      "epoch": 0.6444051101919532,
      "grad_norm": 1.605654677229932,
      "learning_rate": 2.964869590867482e-06,
      "loss": 0.7589,
      "step": 8783
    },
    {
      "epoch": 0.6444784797820923,
      "grad_norm": 1.786179894764187,
      "learning_rate": 2.9637843296280623e-06,
      "loss": 0.7293,
      "step": 8784
    },
    {
      "epoch": 0.6444784797820923,
      "lm_loss": 1.6562,
      "step": 8784,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.6444784797820923,
      "lm_loss": 1.071,
      "step": 8784,
      "vm_loss": 0.1093
    },
    {
      "epoch": 0.6444784797820923,
      "lm_loss": 0.6506,
      "step": 8784,
      "vm_loss": 0.0506
    },
    {
      "epoch": 0.6444784797820923,
      "lm_loss": 0.4077,
      "step": 8784,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.6444784797820923,
      "lm_loss": 0.2798,
      "step": 8784,
      "vm_loss": 0.1065
    },
    {
      "epoch": 0.6444784797820923,
      "lm_loss": 0.875,
      "step": 8784,
      "vm_loss": 0.1492
    },
    {
      "epoch": 0.6444784797820923,
      "lm_loss": 0.5215,
      "step": 8784,
      "vm_loss": 0.0948
    },
    {
      "epoch": 0.6444784797820923,
      "lm_loss": 0.6692,
      "step": 8784,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.6445518493722314,
      "grad_norm": 1.3379055958584438,
      "learning_rate": 2.962699183378783e-06,
      "loss": 0.6994,
      "step": 8785
    },
    {
      "epoch": 0.6446252189623706,
      "grad_norm": 0.5526149537417613,
      "learning_rate": 2.961614152180926e-06,
      "loss": 0.4928,
      "step": 8786
    },
    {
      "epoch": 0.6446985885525097,
      "grad_norm": 1.657334916575957,
      "learning_rate": 2.9605292360957673e-06,
      "loss": 0.7255,
      "step": 8787
    },
    {
      "epoch": 0.6447719581426489,
      "grad_norm": 1.7267490859418206,
      "learning_rate": 2.959444435184572e-06,
      "loss": 0.7414,
      "step": 8788
    },
    {
      "epoch": 0.6448453277327879,
      "grad_norm": 0.5810360068889293,
      "learning_rate": 2.958359749508603e-06,
      "loss": 0.4981,
      "step": 8789
    },
    {
      "epoch": 0.6449186973229271,
      "grad_norm": 1.514160912512999,
      "learning_rate": 2.9572751791291152e-06,
      "loss": 0.7713,
      "step": 8790
    },
    {
      "epoch": 0.6449920669130662,
      "grad_norm": 1.4924577451730165,
      "learning_rate": 2.9561907241073563e-06,
      "loss": 0.7726,
      "step": 8791
    },
    {
      "epoch": 0.6450654365032054,
      "grad_norm": 0.5350091475350711,
      "learning_rate": 2.9551063845045692e-06,
      "loss": 0.4708,
      "step": 8792
    },
    {
      "epoch": 0.6450654365032054,
      "lm_loss": 1.2109,
      "step": 8792,
      "vm_loss": 0.1216
    },
    {
      "epoch": 0.6450654365032054,
      "lm_loss": 0.8979,
      "step": 8792,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.6450654365032054,
      "lm_loss": 0.6742,
      "step": 8792,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.6450654365032054,
      "lm_loss": 0.2892,
      "step": 8792,
      "vm_loss": 0.1499
    },
    {
      "epoch": 0.6450654365032054,
      "lm_loss": 0.9158,
      "step": 8792,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.6450654365032054,
      "lm_loss": 0.1329,
      "step": 8792,
      "vm_loss": 0.1706
    },
    {
      "epoch": 0.6450654365032054,
      "lm_loss": 1.0616,
      "step": 8792,
      "vm_loss": 0.1786
    },
    {
      "epoch": 0.6450654365032054,
      "lm_loss": 0.2191,
      "step": 8792,
      "vm_loss": 0.054
    },
    {
      "epoch": 0.6451388060933445,
      "grad_norm": 1.4831492235212198,
      "learning_rate": 2.9540221603819885e-06,
      "loss": 0.6127,
      "step": 8793
    },
    {
      "epoch": 0.6452121756834835,
      "grad_norm": 1.5237400922828515,
      "learning_rate": 2.952938051800843e-06,
      "loss": 0.6585,
      "step": 8794
    },
    {
      "epoch": 0.6452855452736227,
      "grad_norm": 1.585566481379604,
      "learning_rate": 2.9518540588223556e-06,
      "loss": 0.7364,
      "step": 8795
    },
    {
      "epoch": 0.6453589148637618,
      "grad_norm": 1.5337930084938265,
      "learning_rate": 2.9507701815077407e-06,
      "loss": 0.8348,
      "step": 8796
    },
    {
      "epoch": 0.645432284453901,
      "grad_norm": 0.5433921201887775,
      "learning_rate": 2.9496864199182094e-06,
      "loss": 0.4823,
      "step": 8797
    },
    {
      "epoch": 0.6455056540440401,
      "grad_norm": 1.6366244324717436,
      "learning_rate": 2.9486027741149637e-06,
      "loss": 0.6659,
      "step": 8798
    },
    {
      "epoch": 0.6455790236341792,
      "grad_norm": 1.6116780371742472,
      "learning_rate": 2.9475192441591982e-06,
      "loss": 0.785,
      "step": 8799
    },
    {
      "epoch": 0.6456523932243183,
      "grad_norm": 0.5704765180469974,
      "learning_rate": 2.9464358301121053e-06,
      "loss": 0.4802,
      "step": 8800
    },
    {
      "epoch": 0.6456523932243183,
      "lm_loss": 0.8666,
      "step": 8800,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.6456523932243183,
      "lm_loss": 1.3933,
      "step": 8800,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.6456523932243183,
      "lm_loss": 0.8132,
      "step": 8800,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.6456523932243183,
      "lm_loss": 0.5469,
      "step": 8800,
      "vm_loss": 0.1244
    },
    {
      "epoch": 0.6456523932243183,
      "lm_loss": 0.3137,
      "step": 8800,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.6456523932243183,
      "lm_loss": 0.2391,
      "step": 8800,
      "vm_loss": 0.1274
    },
    {
      "epoch": 0.6456523932243183,
      "lm_loss": 0.0601,
      "step": 8800,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.6456523932243183,
      "lm_loss": 0.9383,
      "step": 8800,
      "vm_loss": 0.0983
    },
    {
      "epoch": 0.6457257628144575,
      "grad_norm": 1.4731830166712323,
      "learning_rate": 2.9453525320348664e-06,
      "loss": 0.727,
      "step": 8801
    },
    {
      "epoch": 0.6457991324045966,
      "grad_norm": 1.623726598103132,
      "learning_rate": 2.9442693499886573e-06,
      "loss": 0.7682,
      "step": 8802
    },
    {
      "epoch": 0.6458725019947358,
      "grad_norm": 1.492388597823805,
      "learning_rate": 2.94318628403465e-06,
      "loss": 0.7174,
      "step": 8803
    },
    {
      "epoch": 0.6459458715848748,
      "grad_norm": 0.5646850714379221,
      "learning_rate": 2.9421033342340053e-06,
      "loss": 0.4766,
      "step": 8804
    },
    {
      "epoch": 0.646019241175014,
      "grad_norm": 1.6290251909841778,
      "learning_rate": 2.9410205006478833e-06,
      "loss": 0.7582,
      "step": 8805
    },
    {
      "epoch": 0.6460926107651531,
      "grad_norm": 1.6992275196400417,
      "learning_rate": 2.939937783337432e-06,
      "loss": 0.7386,
      "step": 8806
    },
    {
      "epoch": 0.6461659803552923,
      "grad_norm": 1.49591274879167,
      "learning_rate": 2.9388551823637958e-06,
      "loss": 0.8209,
      "step": 8807
    },
    {
      "epoch": 0.6462393499454314,
      "grad_norm": 1.4752449841388018,
      "learning_rate": 2.9377726977881117e-06,
      "loss": 0.6996,
      "step": 8808
    },
    {
      "epoch": 0.6462393499454314,
      "lm_loss": 1.2581,
      "step": 8808,
      "vm_loss": 0.1017
    },
    {
      "epoch": 0.6462393499454314,
      "lm_loss": 0.0845,
      "step": 8808,
      "vm_loss": 0.0543
    },
    {
      "epoch": 0.6462393499454314,
      "lm_loss": 0.2214,
      "step": 8808,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.6462393499454314,
      "lm_loss": 0.6675,
      "step": 8808,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.6462393499454314,
      "lm_loss": 1.0086,
      "step": 8808,
      "vm_loss": 0.1991
    },
    {
      "epoch": 0.6462393499454314,
      "lm_loss": 0.1484,
      "step": 8808,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.6462393499454314,
      "lm_loss": 0.756,
      "step": 8808,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.6462393499454314,
      "lm_loss": 0.2359,
      "step": 8808,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.6463127195355705,
      "grad_norm": 1.5261384849788353,
      "learning_rate": 2.936690329671511e-06,
      "loss": 0.7563,
      "step": 8809
    },
    {
      "epoch": 0.6463860891257096,
      "grad_norm": 1.3648835411692468,
      "learning_rate": 2.9356080780751163e-06,
      "loss": 0.5902,
      "step": 8810
    },
    {
      "epoch": 0.6464594587158488,
      "grad_norm": 1.553870108694523,
      "learning_rate": 2.934525943060046e-06,
      "loss": 0.8238,
      "step": 8811
    },
    {
      "epoch": 0.6465328283059879,
      "grad_norm": 1.6286723478188292,
      "learning_rate": 2.9334439246874115e-06,
      "loss": 0.7445,
      "step": 8812
    },
    {
      "epoch": 0.646606197896127,
      "grad_norm": 1.406339024899846,
      "learning_rate": 2.9323620230183146e-06,
      "loss": 0.7838,
      "step": 8813
    },
    {
      "epoch": 0.6466795674862661,
      "grad_norm": 1.3939722227675604,
      "learning_rate": 2.931280238113856e-06,
      "loss": 0.639,
      "step": 8814
    },
    {
      "epoch": 0.6467529370764052,
      "grad_norm": 1.4765765077973232,
      "learning_rate": 2.9301985700351244e-06,
      "loss": 0.6239,
      "step": 8815
    },
    {
      "epoch": 0.6468263066665444,
      "grad_norm": 0.5897048561456901,
      "learning_rate": 2.929117018843206e-06,
      "loss": 0.491,
      "step": 8816
    },
    {
      "epoch": 0.6468263066665444,
      "lm_loss": 0.358,
      "step": 8816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6468263066665444,
      "lm_loss": 0.4267,
      "step": 8816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6468263066665444,
      "lm_loss": 0.4675,
      "step": 8816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6468263066665444,
      "lm_loss": 0.5282,
      "step": 8816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6468263066665444,
      "lm_loss": 0.4035,
      "step": 8816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6468263066665444,
      "lm_loss": 0.5449,
      "step": 8816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6468263066665444,
      "lm_loss": 0.5415,
      "step": 8816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6468263066665444,
      "lm_loss": 0.5038,
      "step": 8816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6468996762566835,
      "grad_norm": 0.5961328810474539,
      "learning_rate": 2.9280355845991785e-06,
      "loss": 0.498,
      "step": 8817
    },
    {
      "epoch": 0.6469730458468227,
      "grad_norm": 1.442318416373778,
      "learning_rate": 2.9269542673641114e-06,
      "loss": 0.6701,
      "step": 8818
    },
    {
      "epoch": 0.6470464154369617,
      "grad_norm": 1.4529103491383002,
      "learning_rate": 2.925873067199072e-06,
      "loss": 0.7056,
      "step": 8819
    },
    {
      "epoch": 0.6471197850271009,
      "grad_norm": 0.5601031754218625,
      "learning_rate": 2.9247919841651173e-06,
      "loss": 0.479,
      "step": 8820
    },
    {
      "epoch": 0.64719315461724,
      "grad_norm": 1.4904229817299968,
      "learning_rate": 2.9237110183232964e-06,
      "loss": 0.7619,
      "step": 8821
    },
    {
      "epoch": 0.6472665242073792,
      "grad_norm": 1.5184411711419703,
      "learning_rate": 2.922630169734658e-06,
      "loss": 0.7906,
      "step": 8822
    },
    {
      "epoch": 0.6473398937975183,
      "grad_norm": 0.5747162320641698,
      "learning_rate": 2.9215494384602373e-06,
      "loss": 0.5073,
      "step": 8823
    },
    {
      "epoch": 0.6474132633876574,
      "grad_norm": 0.5619586728929246,
      "learning_rate": 2.920468824561068e-06,
      "loss": 0.4867,
      "step": 8824
    },
    {
      "epoch": 0.6474132633876574,
      "lm_loss": 1.0307,
      "step": 8824,
      "vm_loss": 0.0665
    },
    {
      "epoch": 0.6474132633876574,
      "lm_loss": 0.9028,
      "step": 8824,
      "vm_loss": 0.1617
    },
    {
      "epoch": 0.6474132633876574,
      "lm_loss": 0.02,
      "step": 8824,
      "vm_loss": 0.0433
    },
    {
      "epoch": 0.6474132633876574,
      "lm_loss": 0.4682,
      "step": 8824,
      "vm_loss": 0.1796
    },
    {
      "epoch": 0.6474132633876574,
      "lm_loss": 1.2076,
      "step": 8824,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.6474132633876574,
      "lm_loss": 0.2168,
      "step": 8824,
      "vm_loss": 0.0991
    },
    {
      "epoch": 0.6474132633876574,
      "lm_loss": 0.0537,
      "step": 8824,
      "vm_loss": 0.1338
    },
    {
      "epoch": 0.6474132633876574,
      "lm_loss": 0.6896,
      "step": 8824,
      "vm_loss": 0.1009
    },
    {
      "epoch": 0.6474866329777965,
      "grad_norm": 1.6968841495222176,
      "learning_rate": 2.919388328098174e-06,
      "loss": 0.8036,
      "step": 8825
    },
    {
      "epoch": 0.6475600025679357,
      "grad_norm": 1.6077362595109872,
      "learning_rate": 2.918307949132573e-06,
      "loss": 0.7083,
      "step": 8826
    },
    {
      "epoch": 0.6476333721580748,
      "grad_norm": 1.5937078237321318,
      "learning_rate": 2.9172276877252757e-06,
      "loss": 0.8104,
      "step": 8827
    },
    {
      "epoch": 0.6477067417482139,
      "grad_norm": 1.566554270855392,
      "learning_rate": 2.916147543937291e-06,
      "loss": 0.7632,
      "step": 8828
    },
    {
      "epoch": 0.647780111338353,
      "grad_norm": 1.5000105738085048,
      "learning_rate": 2.915067517829615e-06,
      "loss": 0.7234,
      "step": 8829
    },
    {
      "epoch": 0.6478534809284922,
      "grad_norm": 1.4279340308689943,
      "learning_rate": 2.913987609463239e-06,
      "loss": 0.6872,
      "step": 8830
    },
    {
      "epoch": 0.6479268505186313,
      "grad_norm": 0.5802464998479968,
      "learning_rate": 2.91290781889915e-06,
      "loss": 0.4776,
      "step": 8831
    },
    {
      "epoch": 0.6480002201087705,
      "grad_norm": 1.5575006521134782,
      "learning_rate": 2.911828146198324e-06,
      "loss": 0.7183,
      "step": 8832
    },
    {
      "epoch": 0.6480002201087705,
      "lm_loss": 0.6837,
      "step": 8832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6480002201087705,
      "lm_loss": 0.468,
      "step": 8832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6480002201087705,
      "lm_loss": 0.4296,
      "step": 8832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6480002201087705,
      "lm_loss": 0.4618,
      "step": 8832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6480002201087705,
      "lm_loss": 0.4374,
      "step": 8832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6480002201087705,
      "lm_loss": 0.4522,
      "step": 8832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6480002201087705,
      "lm_loss": 0.5138,
      "step": 8832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6480002201087705,
      "lm_loss": 0.5951,
      "step": 8832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6480735896989095,
      "grad_norm": 0.5682267929278895,
      "learning_rate": 2.9107485914217315e-06,
      "loss": 0.4832,
      "step": 8833
    },
    {
      "epoch": 0.6481469592890486,
      "grad_norm": 1.4147631838368597,
      "learning_rate": 2.909669154630343e-06,
      "loss": 0.7145,
      "step": 8834
    },
    {
      "epoch": 0.6482203288791878,
      "grad_norm": 1.4970974698252206,
      "learning_rate": 2.9085898358851135e-06,
      "loss": 0.8782,
      "step": 8835
    },
    {
      "epoch": 0.6482936984693269,
      "grad_norm": 1.4404501194091306,
      "learning_rate": 2.907510635246995e-06,
      "loss": 0.8736,
      "step": 8836
    },
    {
      "epoch": 0.6483670680594661,
      "grad_norm": 1.5224074005671349,
      "learning_rate": 2.9064315527769333e-06,
      "loss": 0.733,
      "step": 8837
    },
    {
      "epoch": 0.6484404376496051,
      "grad_norm": 0.5437551792092258,
      "learning_rate": 2.905352588535864e-06,
      "loss": 0.4933,
      "step": 8838
    },
    {
      "epoch": 0.6485138072397443,
      "grad_norm": 1.6179086159698168,
      "learning_rate": 2.9042737425847235e-06,
      "loss": 0.8459,
      "step": 8839
    },
    {
      "epoch": 0.6485871768298834,
      "grad_norm": 0.5460733552609085,
      "learning_rate": 2.903195014984435e-06,
      "loss": 0.4872,
      "step": 8840
    },
    {
      "epoch": 0.6485871768298834,
      "lm_loss": 1.0601,
      "step": 8840,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.6485871768298834,
      "lm_loss": 0.0412,
      "step": 8840,
      "vm_loss": 0.0891
    },
    {
      "epoch": 0.6485871768298834,
      "lm_loss": 0.2605,
      "step": 8840,
      "vm_loss": 0.0657
    },
    {
      "epoch": 0.6485871768298834,
      "lm_loss": 0.7877,
      "step": 8840,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.6485871768298834,
      "lm_loss": 1.6644,
      "step": 8840,
      "vm_loss": 0.1296
    },
    {
      "epoch": 0.6485871768298834,
      "lm_loss": 0.3307,
      "step": 8840,
      "vm_loss": 0.2574
    },
    {
      "epoch": 0.6485871768298834,
      "lm_loss": 0.8183,
      "step": 8840,
      "vm_loss": 0.1544
    },
    {
      "epoch": 0.6485871768298834,
      "lm_loss": 0.4696,
      "step": 8840,
      "vm_loss": 0.1369
    },
    {
      "epoch": 0.6486605464200226,
      "grad_norm": 1.5257543617160039,
      "learning_rate": 2.902116405795917e-06,
      "loss": 0.8014,
      "step": 8841
    },
    {
      "epoch": 0.6487339160101617,
      "grad_norm": 0.5621280631144719,
      "learning_rate": 2.90103791508008e-06,
      "loss": 0.4877,
      "step": 8842
    },
    {
      "epoch": 0.6488072856003008,
      "grad_norm": 1.5650061255658334,
      "learning_rate": 2.89995954289783e-06,
      "loss": 0.828,
      "step": 8843
    },
    {
      "epoch": 0.6488806551904399,
      "grad_norm": 1.4846456189256312,
      "learning_rate": 2.8988812893100637e-06,
      "loss": 0.6816,
      "step": 8844
    },
    {
      "epoch": 0.6489540247805791,
      "grad_norm": 1.6153989288376236,
      "learning_rate": 2.897803154377676e-06,
      "loss": 0.7417,
      "step": 8845
    },
    {
      "epoch": 0.6490273943707182,
      "grad_norm": 1.6282602284375187,
      "learning_rate": 2.896725138161551e-06,
      "loss": 0.8009,
      "step": 8846
    },
    {
      "epoch": 0.6491007639608574,
      "grad_norm": 1.4655576906983458,
      "learning_rate": 2.8956472407225654e-06,
      "loss": 0.7273,
      "step": 8847
    },
    {
      "epoch": 0.6491741335509964,
      "grad_norm": 1.6116995452904204,
      "learning_rate": 2.8945694621215913e-06,
      "loss": 0.8342,
      "step": 8848
    },
    {
      "epoch": 0.6491741335509964,
      "lm_loss": 1.3643,
      "step": 8848,
      "vm_loss": 0.0491
    },
    {
      "epoch": 0.6491741335509964,
      "lm_loss": 0.2291,
      "step": 8848,
      "vm_loss": 0.2442
    },
    {
      "epoch": 0.6491741335509964,
      "lm_loss": 1.0538,
      "step": 8848,
      "vm_loss": 0.1997
    },
    {
      "epoch": 0.6491741335509964,
      "lm_loss": 0.7685,
      "step": 8848,
      "vm_loss": 0.1721
    },
    {
      "epoch": 0.6491741335509964,
      "lm_loss": 1.1148,
      "step": 8848,
      "vm_loss": 0.087
    },
    {
      "epoch": 0.6491741335509964,
      "lm_loss": 0.0899,
      "step": 8848,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.6491741335509964,
      "lm_loss": 1.0832,
      "step": 8848,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.6491741335509964,
      "lm_loss": 0.8864,
      "step": 8848,
      "vm_loss": 0.0897
    },
    {
      "epoch": 0.6492475031411356,
      "grad_norm": 1.6731310665145187,
      "learning_rate": 2.893491802419492e-06,
      "loss": 0.7578,
      "step": 8849
    },
    {
      "epoch": 0.6493208727312747,
      "grad_norm": 1.4344568289169064,
      "learning_rate": 2.8924142616771287e-06,
      "loss": 0.7482,
      "step": 8850
    },
    {
      "epoch": 0.6493942423214139,
      "grad_norm": 1.472011171888956,
      "learning_rate": 2.8913368399553524e-06,
      "loss": 0.6753,
      "step": 8851
    },
    {
      "epoch": 0.649467611911553,
      "grad_norm": 1.792703981995075,
      "learning_rate": 2.8902595373150066e-06,
      "loss": 0.7667,
      "step": 8852
    },
    {
      "epoch": 0.649540981501692,
      "grad_norm": 1.5608176007049355,
      "learning_rate": 2.889182353816929e-06,
      "loss": 0.7715,
      "step": 8853
    },
    {
      "epoch": 0.6496143510918312,
      "grad_norm": 0.5860536638922199,
      "learning_rate": 2.8881052895219493e-06,
      "loss": 0.4891,
      "step": 8854
    },
    {
      "epoch": 0.6496877206819703,
      "grad_norm": 1.6334015774342499,
      "learning_rate": 2.8870283444908958e-06,
      "loss": 0.7554,
      "step": 8855
    },
    {
      "epoch": 0.6497610902721095,
      "grad_norm": 0.5737327679071921,
      "learning_rate": 2.8859515187845843e-06,
      "loss": 0.4772,
      "step": 8856
    },
    {
      "epoch": 0.6497610902721095,
      "lm_loss": 1.1364,
      "step": 8856,
      "vm_loss": 0.053
    },
    {
      "epoch": 0.6497610902721095,
      "lm_loss": 0.0442,
      "step": 8856,
      "vm_loss": 0.0773
    },
    {
      "epoch": 0.6497610902721095,
      "lm_loss": 1.3623,
      "step": 8856,
      "vm_loss": 0.1675
    },
    {
      "epoch": 0.6497610902721095,
      "lm_loss": 0.4326,
      "step": 8856,
      "vm_loss": 0.29
    },
    {
      "epoch": 0.6497610902721095,
      "lm_loss": 0.7581,
      "step": 8856,
      "vm_loss": 0.1296
    },
    {
      "epoch": 0.6497610902721095,
      "lm_loss": 0.7174,
      "step": 8856,
      "vm_loss": 0.0425
    },
    {
      "epoch": 0.6497610902721095,
      "lm_loss": 0.2383,
      "step": 8856,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.6497610902721095,
      "lm_loss": 1.1741,
      "step": 8856,
      "vm_loss": 0.1827
    },
    {
      "epoch": 0.6498344598622486,
      "grad_norm": 1.4656949886529027,
      "learning_rate": 2.8848748124638266e-06,
      "loss": 0.7107,
      "step": 8857
    },
    {
      "epoch": 0.6499078294523877,
      "grad_norm": 0.563853275617256,
      "learning_rate": 2.8837982255894257e-06,
      "loss": 0.4801,
      "step": 8858
    },
    {
      "epoch": 0.6499811990425268,
      "grad_norm": 1.527811818501745,
      "learning_rate": 2.88272175822218e-06,
      "loss": 0.7558,
      "step": 8859
    },
    {
      "epoch": 0.650054568632666,
      "grad_norm": 0.5745869401226279,
      "learning_rate": 2.8816454104228777e-06,
      "loss": 0.4909,
      "step": 8860
    },
    {
      "epoch": 0.6501279382228051,
      "grad_norm": 1.5130589980075484,
      "learning_rate": 2.8805691822523068e-06,
      "loss": 0.7596,
      "step": 8861
    },
    {
      "epoch": 0.6502013078129443,
      "grad_norm": 1.3541164331940474,
      "learning_rate": 2.8794930737712433e-06,
      "loss": 0.6712,
      "step": 8862
    },
    {
      "epoch": 0.6502746774030833,
      "grad_norm": 1.6243591429078288,
      "learning_rate": 2.8784170850404567e-06,
      "loss": 0.6952,
      "step": 8863
    },
    {
      "epoch": 0.6503480469932225,
      "grad_norm": 0.5630868431418244,
      "learning_rate": 2.877341216120712e-06,
      "loss": 0.4833,
      "step": 8864
    },
    {
      "epoch": 0.6503480469932225,
      "lm_loss": 0.3176,
      "step": 8864,
      "vm_loss": 0.0888
    },
    {
      "epoch": 0.6503480469932225,
      "lm_loss": 1.4377,
      "step": 8864,
      "vm_loss": 0.079
    },
    {
      "epoch": 0.6503480469932225,
      "lm_loss": 0.0561,
      "step": 8864,
      "vm_loss": 0.1192
    },
    {
      "epoch": 0.6503480469932225,
      "lm_loss": 1.4006,
      "step": 8864,
      "vm_loss": 0.1934
    },
    {
      "epoch": 0.6503480469932225,
      "lm_loss": 1.0447,
      "step": 8864,
      "vm_loss": 0.1528
    },
    {
      "epoch": 0.6503480469932225,
      "lm_loss": 0.9993,
      "step": 8864,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.6503480469932225,
      "lm_loss": 0.2069,
      "step": 8864,
      "vm_loss": 0.1371
    },
    {
      "epoch": 0.6503480469932225,
      "lm_loss": 0.6185,
      "step": 8864,
      "vm_loss": 0.0639
    },
    {
      "epoch": 0.6504214165833616,
      "grad_norm": 1.4957818399265128,
      "learning_rate": 2.8762654670727636e-06,
      "loss": 0.7776,
      "step": 8865
    },
    {
      "epoch": 0.6504947861735008,
      "grad_norm": 0.5636252788544057,
      "learning_rate": 2.8751898379573652e-06,
      "loss": 0.4809,
      "step": 8866
    },
    {
      "epoch": 0.6505681557636399,
      "grad_norm": 0.5608984570030705,
      "learning_rate": 2.8741143288352584e-06,
      "loss": 0.4711,
      "step": 8867
    },
    {
      "epoch": 0.650641525353779,
      "grad_norm": 1.6242641502781354,
      "learning_rate": 2.8730389397671808e-06,
      "loss": 0.8344,
      "step": 8868
    },
    {
      "epoch": 0.6507148949439181,
      "grad_norm": 1.5411594459695654,
      "learning_rate": 2.871963670813861e-06,
      "loss": 0.7232,
      "step": 8869
    },
    {
      "epoch": 0.6507882645340572,
      "grad_norm": 1.589764659014919,
      "learning_rate": 2.8708885220360226e-06,
      "loss": 0.7964,
      "step": 8870
    },
    {
      "epoch": 0.6508616341241964,
      "grad_norm": 1.5270448978010491,
      "learning_rate": 2.869813493494381e-06,
      "loss": 0.757,
      "step": 8871
    },
    {
      "epoch": 0.6509350037143355,
      "grad_norm": 0.5698879536109515,
      "learning_rate": 2.868738585249648e-06,
      "loss": 0.4822,
      "step": 8872
    },
    {
      "epoch": 0.6509350037143355,
      "lm_loss": 0.2069,
      "step": 8872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6509350037143355,
      "lm_loss": 0.4366,
      "step": 8872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6509350037143355,
      "lm_loss": 0.6916,
      "step": 8872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6509350037143355,
      "lm_loss": 0.3899,
      "step": 8872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6509350037143355,
      "lm_loss": 0.4576,
      "step": 8872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6509350037143355,
      "lm_loss": 0.4869,
      "step": 8872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6509350037143355,
      "lm_loss": 0.4316,
      "step": 8872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6509350037143355,
      "lm_loss": 0.5234,
      "step": 8872,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6510083733044746,
      "grad_norm": 0.5650958477675926,
      "learning_rate": 2.8676637973625254e-06,
      "loss": 0.4809,
      "step": 8873
    },
    {
      "epoch": 0.6510817428946137,
      "grad_norm": 1.5632943765085097,
      "learning_rate": 2.8665891298937088e-06,
      "loss": 0.7075,
      "step": 8874
    },
    {
      "epoch": 0.6511551124847529,
      "grad_norm": 1.524091453649059,
      "learning_rate": 2.8655145829038867e-06,
      "loss": 0.6806,
      "step": 8875
    },
    {
      "epoch": 0.651228482074892,
      "grad_norm": 1.4971245606572778,
      "learning_rate": 2.8644401564537403e-06,
      "loss": 0.7014,
      "step": 8876
    },
    {
      "epoch": 0.6513018516650312,
      "grad_norm": 1.5140135166329958,
      "learning_rate": 2.8633658506039497e-06,
      "loss": 0.7922,
      "step": 8877
    },
    {
      "epoch": 0.6513752212551702,
      "grad_norm": 1.4921367936767125,
      "learning_rate": 2.86229166541518e-06,
      "loss": 0.731,
      "step": 8878
    },
    {
      "epoch": 0.6514485908453094,
      "grad_norm": 0.5682934629645984,
      "learning_rate": 2.8612176009480943e-06,
      "loss": 0.5048,
      "step": 8879
    },
    {
      "epoch": 0.6515219604354485,
      "grad_norm": 1.6180447941373892,
      "learning_rate": 2.8601436572633466e-06,
      "loss": 0.7648,
      "step": 8880
    },
    {
      "epoch": 0.6515219604354485,
      "lm_loss": 1.6352,
      "step": 8880,
      "vm_loss": 0.0324
    },
    {
      "epoch": 0.6515219604354485,
      "lm_loss": 1.4214,
      "step": 8880,
      "vm_loss": 0.1198
    },
    {
      "epoch": 0.6515219604354485,
      "lm_loss": 0.4586,
      "step": 8880,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.6515219604354485,
      "lm_loss": 0.4797,
      "step": 8880,
      "vm_loss": 0.1732
    },
    {
      "epoch": 0.6515219604354485,
      "lm_loss": 0.5477,
      "step": 8880,
      "vm_loss": 0.101
    },
    {
      "epoch": 0.6515219604354485,
      "lm_loss": 0.2491,
      "step": 8880,
      "vm_loss": 0.1196
    },
    {
      "epoch": 0.6515219604354485,
      "lm_loss": 0.186,
      "step": 8880,
      "vm_loss": 0.11
    },
    {
      "epoch": 0.6515219604354485,
      "lm_loss": 0.4392,
      "step": 8880,
      "vm_loss": 0.0659
    },
    {
      "epoch": 0.6515953300255877,
      "grad_norm": 1.4424872334670331,
      "learning_rate": 2.8590698344215843e-06,
      "loss": 0.801,
      "step": 8881
    },
    {
      "epoch": 0.6516686996157268,
      "grad_norm": 1.5425062763780824,
      "learning_rate": 2.8579961324834525e-06,
      "loss": 0.6951,
      "step": 8882
    },
    {
      "epoch": 0.6517420692058659,
      "grad_norm": 0.5559008758925329,
      "learning_rate": 2.8569225515095834e-06,
      "loss": 0.4691,
      "step": 8883
    },
    {
      "epoch": 0.651815438796005,
      "grad_norm": 1.4143556586913644,
      "learning_rate": 2.855849091560605e-06,
      "loss": 0.6456,
      "step": 8884
    },
    {
      "epoch": 0.6518888083861442,
      "grad_norm": 1.3531654032285245,
      "learning_rate": 2.8547757526971385e-06,
      "loss": 0.7181,
      "step": 8885
    },
    {
      "epoch": 0.6519621779762833,
      "grad_norm": 1.599661807088699,
      "learning_rate": 2.853702534979798e-06,
      "loss": 0.8501,
      "step": 8886
    },
    {
      "epoch": 0.6520355475664225,
      "grad_norm": 1.5138479517680663,
      "learning_rate": 2.8526294384691888e-06,
      "loss": 0.7574,
      "step": 8887
    },
    {
      "epoch": 0.6521089171565615,
      "grad_norm": 1.4763017978313264,
      "learning_rate": 2.851556463225914e-06,
      "loss": 0.7805,
      "step": 8888
    },
    {
      "epoch": 0.6521089171565615,
      "lm_loss": 1.263,
      "step": 8888,
      "vm_loss": 0.0469
    },
    {
      "epoch": 0.6521089171565615,
      "lm_loss": 0.9475,
      "step": 8888,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.6521089171565615,
      "lm_loss": 1.0543,
      "step": 8888,
      "vm_loss": 0.1254
    },
    {
      "epoch": 0.6521089171565615,
      "lm_loss": 0.6291,
      "step": 8888,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.6521089171565615,
      "lm_loss": 0.5844,
      "step": 8888,
      "vm_loss": 0.163
    },
    {
      "epoch": 0.6521089171565615,
      "lm_loss": 0.4496,
      "step": 8888,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.6521089171565615,
      "lm_loss": 0.1504,
      "step": 8888,
      "vm_loss": 0.055
    },
    {
      "epoch": 0.6521089171565615,
      "lm_loss": 0.2773,
      "step": 8888,
      "vm_loss": 0.1476
    },
    {
      "epoch": 0.6521822867467006,
      "grad_norm": 1.47780025618228,
      "learning_rate": 2.850483609310567e-06,
      "loss": 0.708,
      "step": 8889
    },
    {
      "epoch": 0.6522556563368398,
      "grad_norm": 1.4625464133226296,
      "learning_rate": 2.849410876783734e-06,
      "loss": 0.7341,
      "step": 8890
    },
    {
      "epoch": 0.652329025926979,
      "grad_norm": 1.5105710624238369,
      "learning_rate": 2.8483382657059945e-06,
      "loss": 0.7527,
      "step": 8891
    },
    {
      "epoch": 0.6524023955171181,
      "grad_norm": 1.4487935655425574,
      "learning_rate": 2.8472657761379196e-06,
      "loss": 0.7429,
      "step": 8892
    },
    {
      "epoch": 0.6524757651072571,
      "grad_norm": 1.5033442375737793,
      "learning_rate": 2.8461934081400787e-06,
      "loss": 0.7868,
      "step": 8893
    },
    {
      "epoch": 0.6525491346973963,
      "grad_norm": 1.506162232809494,
      "learning_rate": 2.845121161773031e-06,
      "loss": 0.7344,
      "step": 8894
    },
    {
      "epoch": 0.6526225042875354,
      "grad_norm": 0.557985862699962,
      "learning_rate": 2.844049037097327e-06,
      "loss": 0.4906,
      "step": 8895
    },
    {
      "epoch": 0.6526958738776746,
      "grad_norm": 1.8136872876843275,
      "learning_rate": 2.8429770341735135e-06,
      "loss": 0.6968,
      "step": 8896
    },
    {
      "epoch": 0.6526958738776746,
      "lm_loss": 0.447,
      "step": 8896,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.6526958738776746,
      "lm_loss": 0.3894,
      "step": 8896,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.6526958738776746,
      "lm_loss": 0.8293,
      "step": 8896,
      "vm_loss": 0.1284
    },
    {
      "epoch": 0.6526958738776746,
      "lm_loss": 0.6416,
      "step": 8896,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.6526958738776746,
      "lm_loss": 0.8723,
      "step": 8896,
      "vm_loss": 0.1772
    },
    {
      "epoch": 0.6526958738776746,
      "lm_loss": 0.2558,
      "step": 8896,
      "vm_loss": 0.1149
    },
    {
      "epoch": 0.6526958738776746,
      "lm_loss": 1.0021,
      "step": 8896,
      "vm_loss": 0.1642
    },
    {
      "epoch": 0.6526958738776746,
      "lm_loss": 0.8602,
      "step": 8896,
      "vm_loss": 0.1618
    },
    {
      "epoch": 0.6527692434678137,
      "grad_norm": 1.561122010556319,
      "learning_rate": 2.8419051530621284e-06,
      "loss": 0.7469,
      "step": 8897
    },
    {
      "epoch": 0.6528426130579528,
      "grad_norm": 1.6276084592330855,
      "learning_rate": 2.840833393823704e-06,
      "loss": 0.8022,
      "step": 8898
    },
    {
      "epoch": 0.6529159826480919,
      "grad_norm": 0.5458288017645251,
      "learning_rate": 2.839761756518764e-06,
      "loss": 0.4973,
      "step": 8899
    },
    {
      "epoch": 0.6529893522382311,
      "grad_norm": 0.5557103036170621,
      "learning_rate": 2.838690241207829e-06,
      "loss": 0.4823,
      "step": 8900
    },
    {
      "epoch": 0.6530627218283702,
      "grad_norm": 1.568478045884549,
      "learning_rate": 2.8376188479514065e-06,
      "loss": 0.7863,
      "step": 8901
    },
    {
      "epoch": 0.6531360914185094,
      "grad_norm": 0.5676612562945589,
      "learning_rate": 2.836547576810002e-06,
      "loss": 0.4842,
      "step": 8902
    },
    {
      "epoch": 0.6532094610086484,
      "grad_norm": 0.5582344511446371,
      "learning_rate": 2.8354764278441153e-06,
      "loss": 0.4753,
      "step": 8903
    },
    {
      "epoch": 0.6532828305987876,
      "grad_norm": 1.6584667746398885,
      "learning_rate": 2.8344054011142354e-06,
      "loss": 0.7375,
      "step": 8904
    },
    {
      "epoch": 0.6532828305987876,
      "lm_loss": 1.4232,
      "step": 8904,
      "vm_loss": 0.0392
    },
    {
      "epoch": 0.6532828305987876,
      "lm_loss": 0.994,
      "step": 8904,
      "vm_loss": 0.2484
    },
    {
      "epoch": 0.6532828305987876,
      "lm_loss": 0.5417,
      "step": 8904,
      "vm_loss": 0.1252
    },
    {
      "epoch": 0.6532828305987876,
      "lm_loss": 0.0618,
      "step": 8904,
      "vm_loss": 0.0837
    },
    {
      "epoch": 0.6532828305987876,
      "lm_loss": 0.1792,
      "step": 8904,
      "vm_loss": 0.1323
    },
    {
      "epoch": 0.6532828305987876,
      "lm_loss": 0.2627,
      "step": 8904,
      "vm_loss": 0.1016
    },
    {
      "epoch": 0.6532828305987876,
      "lm_loss": 0.5717,
      "step": 8904,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.6532828305987876,
      "lm_loss": 0.0524,
      "step": 8904,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.6533562001889267,
      "grad_norm": 1.383581695602857,
      "learning_rate": 2.8333344966808453e-06,
      "loss": 0.6544,
      "step": 8905
    },
    {
      "epoch": 0.6534295697790659,
      "grad_norm": 0.5452785699621037,
      "learning_rate": 2.8322637146044214e-06,
      "loss": 0.4632,
      "step": 8906
    },
    {
      "epoch": 0.6535029393692049,
      "grad_norm": 1.416067296387947,
      "learning_rate": 2.831193054945432e-06,
      "loss": 0.8431,
      "step": 8907
    },
    {
      "epoch": 0.653576308959344,
      "grad_norm": 1.372748541640117,
      "learning_rate": 2.830122517764344e-06,
      "loss": 0.7634,
      "step": 8908
    },
    {
      "epoch": 0.6536496785494832,
      "grad_norm": 1.5865641766794218,
      "learning_rate": 2.829052103121611e-06,
      "loss": 0.7319,
      "step": 8909
    },
    {
      "epoch": 0.6537230481396223,
      "grad_norm": 1.4571159071537223,
      "learning_rate": 2.8279818110776814e-06,
      "loss": 0.8255,
      "step": 8910
    },
    {
      "epoch": 0.6537964177297615,
      "grad_norm": 1.5048100725087221,
      "learning_rate": 2.8269116416929985e-06,
      "loss": 0.7916,
      "step": 8911
    },
    {
      "epoch": 0.6538697873199005,
      "grad_norm": 0.547682727824264,
      "learning_rate": 2.8258415950279965e-06,
      "loss": 0.4609,
      "step": 8912
    },
    {
      "epoch": 0.6538697873199005,
      "lm_loss": 1.5438,
      "step": 8912,
      "vm_loss": 0.0731
    },
    {
      "epoch": 0.6538697873199005,
      "lm_loss": 1.0553,
      "step": 8912,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.6538697873199005,
      "lm_loss": 0.2314,
      "step": 8912,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.6538697873199005,
      "lm_loss": 0.8686,
      "step": 8912,
      "vm_loss": 0.1754
    },
    {
      "epoch": 0.6538697873199005,
      "lm_loss": 0.1737,
      "step": 8912,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.6538697873199005,
      "lm_loss": 0.0369,
      "step": 8912,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.6538697873199005,
      "lm_loss": 0.4083,
      "step": 8912,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.6538697873199005,
      "lm_loss": 0.213,
      "step": 8912,
      "vm_loss": 0.2429
    },
    {
      "epoch": 0.6539431569100397,
      "grad_norm": 1.561742695483222,
      "learning_rate": 2.824771671143102e-06,
      "loss": 0.7292,
      "step": 8913
    },
    {
      "epoch": 0.6540165265001788,
      "grad_norm": 1.5648766568840244,
      "learning_rate": 2.8237018700987403e-06,
      "loss": 0.8116,
      "step": 8914
    },
    {
      "epoch": 0.654089896090318,
      "grad_norm": 1.372730653319662,
      "learning_rate": 2.822632191955323e-06,
      "loss": 0.6317,
      "step": 8915
    },
    {
      "epoch": 0.6541632656804571,
      "grad_norm": 1.5975725827165617,
      "learning_rate": 2.821562636773258e-06,
      "loss": 0.8625,
      "step": 8916
    },
    {
      "epoch": 0.6542366352705962,
      "grad_norm": 1.4478104972629253,
      "learning_rate": 2.8204932046129463e-06,
      "loss": 0.8884,
      "step": 8917
    },
    {
      "epoch": 0.6543100048607353,
      "grad_norm": 1.580196442283409,
      "learning_rate": 2.819423895534778e-06,
      "loss": 0.7766,
      "step": 8918
    },
    {
      "epoch": 0.6543833744508745,
      "grad_norm": 1.7229413023106477,
      "learning_rate": 2.818354709599145e-06,
      "loss": 0.8793,
      "step": 8919
    },
    {
      "epoch": 0.6544567440410136,
      "grad_norm": 1.376561030789579,
      "learning_rate": 2.817285646866424e-06,
      "loss": 0.7587,
      "step": 8920
    },
    {
      "epoch": 0.6544567440410136,
      "lm_loss": 0.8536,
      "step": 8920,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.6544567440410136,
      "lm_loss": 0.0278,
      "step": 8920,
      "vm_loss": 0.137
    },
    {
      "epoch": 0.6544567440410136,
      "lm_loss": 0.8651,
      "step": 8920,
      "vm_loss": 0.1133
    },
    {
      "epoch": 0.6544567440410136,
      "lm_loss": 0.8626,
      "step": 8920,
      "vm_loss": 0.1215
    },
    {
      "epoch": 0.6544567440410136,
      "lm_loss": 0.3262,
      "step": 8920,
      "vm_loss": 0.1933
    },
    {
      "epoch": 0.6544567440410136,
      "lm_loss": 1.0462,
      "step": 8920,
      "vm_loss": 0.122
    },
    {
      "epoch": 0.6544567440410136,
      "lm_loss": 0.3474,
      "step": 8920,
      "vm_loss": 0.0839
    },
    {
      "epoch": 0.6544567440410136,
      "lm_loss": 1.1238,
      "step": 8920,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.6545301136311528,
      "grad_norm": 1.5532859743507006,
      "learning_rate": 2.8162167073969882e-06,
      "loss": 0.7783,
      "step": 8921
    },
    {
      "epoch": 0.6546034832212918,
      "grad_norm": 1.4382672566086883,
      "learning_rate": 2.8151478912512026e-06,
      "loss": 0.6518,
      "step": 8922
    },
    {
      "epoch": 0.654676852811431,
      "grad_norm": 1.361602168237538,
      "learning_rate": 2.8140791984894245e-06,
      "loss": 0.769,
      "step": 8923
    },
    {
      "epoch": 0.6547502224015701,
      "grad_norm": 0.5578183842505796,
      "learning_rate": 2.8130106291720094e-06,
      "loss": 0.4843,
      "step": 8924
    },
    {
      "epoch": 0.6548235919917093,
      "grad_norm": 1.5553533439833234,
      "learning_rate": 2.8119421833593003e-06,
      "loss": 0.761,
      "step": 8925
    },
    {
      "epoch": 0.6548969615818484,
      "grad_norm": 0.5530835448396884,
      "learning_rate": 2.8108738611116343e-06,
      "loss": 0.4777,
      "step": 8926
    },
    {
      "epoch": 0.6549703311719874,
      "grad_norm": 1.4554543579888497,
      "learning_rate": 2.8098056624893434e-06,
      "loss": 0.7311,
      "step": 8927
    },
    {
      "epoch": 0.6550437007621266,
      "grad_norm": 1.725817398314223,
      "learning_rate": 2.80873758755275e-06,
      "loss": 0.8734,
      "step": 8928
    },
    {
      "epoch": 0.6550437007621266,
      "lm_loss": 1.0209,
      "step": 8928,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.6550437007621266,
      "lm_loss": 0.8533,
      "step": 8928,
      "vm_loss": 0.1712
    },
    {
      "epoch": 0.6550437007621266,
      "lm_loss": 0.2651,
      "step": 8928,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.6550437007621266,
      "lm_loss": 0.5806,
      "step": 8928,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.6550437007621266,
      "lm_loss": 0.0558,
      "step": 8928,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.6550437007621266,
      "lm_loss": 0.6228,
      "step": 8928,
      "vm_loss": 0.1377
    },
    {
      "epoch": 0.6550437007621266,
      "lm_loss": 1.1632,
      "step": 8928,
      "vm_loss": 0.2578
    },
    {
      "epoch": 0.6550437007621266,
      "lm_loss": 0.5746,
      "step": 8928,
      "vm_loss": 0.1272
    },
    {
      "epoch": 0.6551170703522657,
      "grad_norm": 1.3827198312107252,
      "learning_rate": 2.807669636362169e-06,
      "loss": 0.7663,
      "step": 8929
    },
    {
      "epoch": 0.6551904399424049,
      "grad_norm": 1.507958168861972,
      "learning_rate": 2.8066018089779158e-06,
      "loss": 0.7252,
      "step": 8930
    },
    {
      "epoch": 0.655263809532544,
      "grad_norm": 1.5948384500479185,
      "learning_rate": 2.8055341054602904e-06,
      "loss": 0.731,
      "step": 8931
    },
    {
      "epoch": 0.6553371791226831,
      "grad_norm": 0.5611520217033679,
      "learning_rate": 2.804466525869588e-06,
      "loss": 0.4953,
      "step": 8932
    },
    {
      "epoch": 0.6554105487128222,
      "grad_norm": 0.5710483472323374,
      "learning_rate": 2.8033990702660983e-06,
      "loss": 0.494,
      "step": 8933
    },
    {
      "epoch": 0.6554839183029614,
      "grad_norm": 0.5585013457076561,
      "learning_rate": 2.802331738710101e-06,
      "loss": 0.4829,
      "step": 8934
    },
    {
      "epoch": 0.6555572878931005,
      "grad_norm": 0.5496720037362832,
      "learning_rate": 2.8012645312618746e-06,
      "loss": 0.488,
      "step": 8935
    },
    {
      "epoch": 0.6556306574832397,
      "grad_norm": 1.5576758274900255,
      "learning_rate": 2.800197447981685e-06,
      "loss": 0.7072,
      "step": 8936
    },
    {
      "epoch": 0.6556306574832397,
      "lm_loss": 0.6195,
      "step": 8936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6556306574832397,
      "lm_loss": 0.4202,
      "step": 8936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6556306574832397,
      "lm_loss": 0.5177,
      "step": 8936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6556306574832397,
      "lm_loss": 0.5343,
      "step": 8936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6556306574832397,
      "lm_loss": 0.3422,
      "step": 8936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6556306574832397,
      "lm_loss": 0.4995,
      "step": 8936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6556306574832397,
      "lm_loss": 0.6105,
      "step": 8936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6556306574832397,
      "lm_loss": 0.6479,
      "step": 8936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6557040270733787,
      "grad_norm": 0.5310417355041519,
      "learning_rate": 2.7991304889297936e-06,
      "loss": 0.4858,
      "step": 8937
    },
    {
      "epoch": 0.6557773966635179,
      "grad_norm": 1.467471537058233,
      "learning_rate": 2.7980636541664543e-06,
      "loss": 0.7564,
      "step": 8938
    },
    {
      "epoch": 0.655850766253657,
      "grad_norm": 1.5017858128189463,
      "learning_rate": 2.796996943751913e-06,
      "loss": 0.6837,
      "step": 8939
    },
    {
      "epoch": 0.6559241358437962,
      "grad_norm": 1.4612180863256952,
      "learning_rate": 2.7959303577464077e-06,
      "loss": 0.6843,
      "step": 8940
    },
    {
      "epoch": 0.6559975054339353,
      "grad_norm": 1.527135668776397,
      "learning_rate": 2.7948638962101756e-06,
      "loss": 0.8391,
      "step": 8941
    },
    {
      "epoch": 0.6560708750240744,
      "grad_norm": 1.423820977750725,
      "learning_rate": 2.79379755920344e-06,
      "loss": 0.7677,
      "step": 8942
    },
    {
      "epoch": 0.6561442446142135,
      "grad_norm": 1.3399337418423425,
      "learning_rate": 2.7927313467864205e-06,
      "loss": 0.7481,
      "step": 8943
    },
    {
      "epoch": 0.6562176142043527,
      "grad_norm": 1.5835506751980366,
      "learning_rate": 2.7916652590193274e-06,
      "loss": 0.7369,
      "step": 8944
    },
    {
      "epoch": 0.6562176142043527,
      "lm_loss": 1.3047,
      "step": 8944,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.6562176142043527,
      "lm_loss": 0.5593,
      "step": 8944,
      "vm_loss": 0.0457
    },
    {
      "epoch": 0.6562176142043527,
      "lm_loss": 0.3528,
      "step": 8944,
      "vm_loss": 0.058
    },
    {
      "epoch": 0.6562176142043527,
      "lm_loss": 0.1412,
      "step": 8944,
      "vm_loss": 0.102
    },
    {
      "epoch": 0.6562176142043527,
      "lm_loss": 0.8593,
      "step": 8944,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.6562176142043527,
      "lm_loss": 0.1119,
      "step": 8944,
      "vm_loss": 0.0569
    },
    {
      "epoch": 0.6562176142043527,
      "lm_loss": 0.2426,
      "step": 8944,
      "vm_loss": 0.1162
    },
    {
      "epoch": 0.6562176142043527,
      "lm_loss": 0.2053,
      "step": 8944,
      "vm_loss": 0.0839
    },
    {
      "epoch": 0.6562909837944918,
      "grad_norm": 1.3486642917930596,
      "learning_rate": 2.790599295962364e-06,
      "loss": 0.6239,
      "step": 8945
    },
    {
      "epoch": 0.656364353384631,
      "grad_norm": 1.544196633985761,
      "learning_rate": 2.7895334576757314e-06,
      "loss": 0.7782,
      "step": 8946
    },
    {
      "epoch": 0.65643772297477,
      "grad_norm": 0.5860242546674765,
      "learning_rate": 2.7884677442196195e-06,
      "loss": 0.4989,
      "step": 8947
    },
    {
      "epoch": 0.6565110925649091,
      "grad_norm": 1.3920088138814692,
      "learning_rate": 2.78740215565421e-06,
      "loss": 0.7977,
      "step": 8948
    },
    {
      "epoch": 0.6565844621550483,
      "grad_norm": 1.4377480597169778,
      "learning_rate": 2.7863366920396805e-06,
      "loss": 0.6984,
      "step": 8949
    },
    {
      "epoch": 0.6566578317451874,
      "grad_norm": 1.6835765837839665,
      "learning_rate": 2.7852713534361987e-06,
      "loss": 0.7638,
      "step": 8950
    },
    {
      "epoch": 0.6567312013353266,
      "grad_norm": 1.5363890387827943,
      "learning_rate": 2.78420613990393e-06,
      "loss": 0.7176,
      "step": 8951
    },
    {
      "epoch": 0.6568045709254656,
      "grad_norm": 1.496084251415976,
      "learning_rate": 2.783141051503027e-06,
      "loss": 0.7876,
      "step": 8952
    },
    {
      "epoch": 0.6568045709254656,
      "lm_loss": 0.0437,
      "step": 8952,
      "vm_loss": 0.0607
    },
    {
      "epoch": 0.6568045709254656,
      "lm_loss": 0.2013,
      "step": 8952,
      "vm_loss": 0.048
    },
    {
      "epoch": 0.6568045709254656,
      "lm_loss": 0.2188,
      "step": 8952,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.6568045709254656,
      "lm_loss": 0.9469,
      "step": 8952,
      "vm_loss": 0.1492
    },
    {
      "epoch": 0.6568045709254656,
      "lm_loss": 0.3923,
      "step": 8952,
      "vm_loss": 0.0603
    },
    {
      "epoch": 0.6568045709254656,
      "lm_loss": 0.3598,
      "step": 8952,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.6568045709254656,
      "lm_loss": 0.4969,
      "step": 8952,
      "vm_loss": 0.1078
    },
    {
      "epoch": 0.6568045709254656,
      "lm_loss": 0.8284,
      "step": 8952,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.6568779405156048,
      "grad_norm": 1.5651698002390895,
      "learning_rate": 2.7820760882936397e-06,
      "loss": 0.7186,
      "step": 8953
    },
    {
      "epoch": 0.6569513101057439,
      "grad_norm": 1.5286780497364114,
      "learning_rate": 2.7810112503359075e-06,
      "loss": 0.7611,
      "step": 8954
    },
    {
      "epoch": 0.6570246796958831,
      "grad_norm": 0.5777591796445879,
      "learning_rate": 2.779946537689966e-06,
      "loss": 0.4922,
      "step": 8955
    },
    {
      "epoch": 0.6570980492860222,
      "grad_norm": 0.5822100514735792,
      "learning_rate": 2.778881950415939e-06,
      "loss": 0.4991,
      "step": 8956
    },
    {
      "epoch": 0.6571714188761613,
      "grad_norm": 0.568437008921842,
      "learning_rate": 2.777817488573951e-06,
      "loss": 0.4943,
      "step": 8957
    },
    {
      "epoch": 0.6572447884663004,
      "grad_norm": 1.4019073171260352,
      "learning_rate": 2.776753152224113e-06,
      "loss": 0.6685,
      "step": 8958
    },
    {
      "epoch": 0.6573181580564396,
      "grad_norm": 0.5433604101339597,
      "learning_rate": 2.7756889414265298e-06,
      "loss": 0.5005,
      "step": 8959
    },
    {
      "epoch": 0.6573915276465787,
      "grad_norm": 1.5550857033989622,
      "learning_rate": 2.7746248562413002e-06,
      "loss": 0.8474,
      "step": 8960
    },
    {
      "epoch": 0.6573915276465787,
      "lm_loss": 1.1517,
      "step": 8960,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.6573915276465787,
      "lm_loss": 0.2475,
      "step": 8960,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.6573915276465787,
      "lm_loss": 0.2609,
      "step": 8960,
      "vm_loss": 0.0485
    },
    {
      "epoch": 0.6573915276465787,
      "lm_loss": 0.6584,
      "step": 8960,
      "vm_loss": 0.1473
    },
    {
      "epoch": 0.6573915276465787,
      "lm_loss": 0.762,
      "step": 8960,
      "vm_loss": 0.1008
    },
    {
      "epoch": 0.6573915276465787,
      "lm_loss": 0.5229,
      "step": 8960,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.6573915276465787,
      "lm_loss": 0.6078,
      "step": 8960,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.6573915276465787,
      "lm_loss": 0.0808,
      "step": 8960,
      "vm_loss": 0.063
    },
    {
      "epoch": 0.6574648972367179,
      "grad_norm": 1.525567137729184,
      "learning_rate": 2.7735608967285144e-06,
      "loss": 0.6797,
      "step": 8961
    },
    {
      "epoch": 0.6575382668268569,
      "grad_norm": 1.3580559242074721,
      "learning_rate": 2.77249706294826e-06,
      "loss": 0.7507,
      "step": 8962
    },
    {
      "epoch": 0.657611636416996,
      "grad_norm": 0.5336966768173192,
      "learning_rate": 2.771433354960613e-06,
      "loss": 0.475,
      "step": 8963
    },
    {
      "epoch": 0.6576850060071352,
      "grad_norm": 1.8477629786523873,
      "learning_rate": 2.7703697728256435e-06,
      "loss": 0.7865,
      "step": 8964
    },
    {
      "epoch": 0.6577583755972743,
      "grad_norm": 0.5558284485689149,
      "learning_rate": 2.7693063166034144e-06,
      "loss": 0.4778,
      "step": 8965
    },
    {
      "epoch": 0.6578317451874135,
      "grad_norm": 1.6788680076425417,
      "learning_rate": 2.7682429863539817e-06,
      "loss": 0.7054,
      "step": 8966
    },
    {
      "epoch": 0.6579051147775525,
      "grad_norm": 0.5448262061884248,
      "learning_rate": 2.7671797821373926e-06,
      "loss": 0.485,
      "step": 8967
    },
    {
      "epoch": 0.6579784843676917,
      "grad_norm": 0.5531645320702349,
      "learning_rate": 2.766116704013693e-06,
      "loss": 0.4922,
      "step": 8968
    },
    {
      "epoch": 0.6579784843676917,
      "lm_loss": 1.5448,
      "step": 8968,
      "vm_loss": 0.0242
    },
    {
      "epoch": 0.6579784843676917,
      "lm_loss": 0.8578,
      "step": 8968,
      "vm_loss": 0.1843
    },
    {
      "epoch": 0.6579784843676917,
      "lm_loss": 0.6528,
      "step": 8968,
      "vm_loss": 0.2113
    },
    {
      "epoch": 0.6579784843676917,
      "lm_loss": 0.4998,
      "step": 8968,
      "vm_loss": 0.2034
    },
    {
      "epoch": 0.6579784843676917,
      "lm_loss": 0.1243,
      "step": 8968,
      "vm_loss": 0.1538
    },
    {
      "epoch": 0.6579784843676917,
      "lm_loss": 1.2791,
      "step": 8968,
      "vm_loss": 0.1087
    },
    {
      "epoch": 0.6579784843676917,
      "lm_loss": 0.3934,
      "step": 8968,
      "vm_loss": 0.1397
    },
    {
      "epoch": 0.6579784843676917,
      "lm_loss": 0.0604,
      "step": 8968,
      "vm_loss": 0.089
    },
    {
      "epoch": 0.6580518539578308,
      "grad_norm": 1.4691646070734716,
      "learning_rate": 2.765053752042915e-06,
      "loss": 0.7484,
      "step": 8969
    },
    {
      "epoch": 0.65812522354797,
      "grad_norm": 0.5541036920219073,
      "learning_rate": 2.7639909262850863e-06,
      "loss": 0.4982,
      "step": 8970
    },
    {
      "epoch": 0.6581985931381091,
      "grad_norm": 1.6298464066596887,
      "learning_rate": 2.762928226800227e-06,
      "loss": 0.692,
      "step": 8971
    },
    {
      "epoch": 0.6582719627282482,
      "grad_norm": 0.5601797063804417,
      "learning_rate": 2.761865653648349e-06,
      "loss": 0.4791,
      "step": 8972
    },
    {
      "epoch": 0.6583453323183873,
      "grad_norm": 1.8530348688112253,
      "learning_rate": 2.760803206889463e-06,
      "loss": 0.7801,
      "step": 8973
    },
    {
      "epoch": 0.6584187019085265,
      "grad_norm": 1.545641239382511,
      "learning_rate": 2.759740886583564e-06,
      "loss": 0.8014,
      "step": 8974
    },
    {
      "epoch": 0.6584920714986656,
      "grad_norm": 0.5781949120046247,
      "learning_rate": 2.7586786927906457e-06,
      "loss": 0.4927,
      "step": 8975
    },
    {
      "epoch": 0.6585654410888048,
      "grad_norm": 1.5156939986733358,
      "learning_rate": 2.757616625570692e-06,
      "loss": 0.6273,
      "step": 8976
    },
    {
      "epoch": 0.6585654410888048,
      "lm_loss": 0.4761,
      "step": 8976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6585654410888048,
      "lm_loss": 0.4788,
      "step": 8976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6585654410888048,
      "lm_loss": 0.4281,
      "step": 8976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6585654410888048,
      "lm_loss": 0.5714,
      "step": 8976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6585654410888048,
      "lm_loss": 0.4345,
      "step": 8976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6585654410888048,
      "lm_loss": 0.5541,
      "step": 8976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6585654410888048,
      "lm_loss": 0.4175,
      "step": 8976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6585654410888048,
      "lm_loss": 0.4749,
      "step": 8976,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6586388106789438,
      "grad_norm": 0.5579206272546241,
      "learning_rate": 2.7565546849836788e-06,
      "loss": 0.4954,
      "step": 8977
    },
    {
      "epoch": 0.658712180269083,
      "grad_norm": 0.5433270441833737,
      "learning_rate": 2.75549287108958e-06,
      "loss": 0.4688,
      "step": 8978
    },
    {
      "epoch": 0.6587855498592221,
      "grad_norm": 1.537730447524622,
      "learning_rate": 2.754431183948357e-06,
      "loss": 0.7222,
      "step": 8979
    },
    {
      "epoch": 0.6588589194493613,
      "grad_norm": 1.410002800805454,
      "learning_rate": 2.753369623619967e-06,
      "loss": 0.7421,
      "step": 8980
    },
    {
      "epoch": 0.6589322890395004,
      "grad_norm": 0.5531854611038083,
      "learning_rate": 2.752308190164357e-06,
      "loss": 0.4821,
      "step": 8981
    },
    {
      "epoch": 0.6590056586296394,
      "grad_norm": 1.5568757187582551,
      "learning_rate": 2.7512468836414707e-06,
      "loss": 0.7389,
      "step": 8982
    },
    {
      "epoch": 0.6590790282197786,
      "grad_norm": 1.7005208179150713,
      "learning_rate": 2.750185704111239e-06,
      "loss": 0.8872,
      "step": 8983
    },
    {
      "epoch": 0.6591523978099177,
      "grad_norm": 0.5672468527460482,
      "learning_rate": 2.7491246516335944e-06,
      "loss": 0.4872,
      "step": 8984
    },
    {
      "epoch": 0.6591523978099177,
      "lm_loss": 0.9278,
      "step": 8984,
      "vm_loss": 0.0563
    },
    {
      "epoch": 0.6591523978099177,
      "lm_loss": 0.1545,
      "step": 8984,
      "vm_loss": 0.1631
    },
    {
      "epoch": 0.6591523978099177,
      "lm_loss": 0.3774,
      "step": 8984,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.6591523978099177,
      "lm_loss": 0.6405,
      "step": 8984,
      "vm_loss": 0.1696
    },
    {
      "epoch": 0.6591523978099177,
      "lm_loss": 0.0588,
      "step": 8984,
      "vm_loss": 0.128
    },
    {
      "epoch": 0.6591523978099177,
      "lm_loss": 0.1452,
      "step": 8984,
      "vm_loss": 0.0727
    },
    {
      "epoch": 0.6591523978099177,
      "lm_loss": 0.8658,
      "step": 8984,
      "vm_loss": 0.0682
    },
    {
      "epoch": 0.6591523978099177,
      "lm_loss": 0.7045,
      "step": 8984,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.6592257674000569,
      "grad_norm": 1.5434504443318844,
      "learning_rate": 2.7480637262684556e-06,
      "loss": 0.7756,
      "step": 8985
    },
    {
      "epoch": 0.6592991369901959,
      "grad_norm": 1.3714448298997592,
      "learning_rate": 2.747002928075735e-06,
      "loss": 0.745,
      "step": 8986
    },
    {
      "epoch": 0.6593725065803351,
      "grad_norm": 1.6484035459728361,
      "learning_rate": 2.7459422571153372e-06,
      "loss": 0.703,
      "step": 8987
    },
    {
      "epoch": 0.6594458761704742,
      "grad_norm": 1.555101501891748,
      "learning_rate": 2.7448817134471616e-06,
      "loss": 0.7538,
      "step": 8988
    },
    {
      "epoch": 0.6595192457606134,
      "grad_norm": 1.5731946478623229,
      "learning_rate": 2.7438212971311016e-06,
      "loss": 0.8272,
      "step": 8989
    },
    {
      "epoch": 0.6595926153507525,
      "grad_norm": 1.4442592126691354,
      "learning_rate": 2.7427610082270407e-06,
      "loss": 0.81,
      "step": 8990
    },
    {
      "epoch": 0.6596659849408916,
      "grad_norm": 1.5120691725191588,
      "learning_rate": 2.741700846794855e-06,
      "loss": 0.8048,
      "step": 8991
    },
    {
      "epoch": 0.6597393545310307,
      "grad_norm": 1.5915104344347963,
      "learning_rate": 2.7406408128944154e-06,
      "loss": 0.7604,
      "step": 8992
    },
    {
      "epoch": 0.6597393545310307,
      "lm_loss": 1.2434,
      "step": 8992,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.6597393545310307,
      "lm_loss": 0.11,
      "step": 8992,
      "vm_loss": 0.1874
    },
    {
      "epoch": 0.6597393545310307,
      "lm_loss": 1.0046,
      "step": 8992,
      "vm_loss": 0.0853
    },
    {
      "epoch": 0.6597393545310307,
      "lm_loss": 0.9461,
      "step": 8992,
      "vm_loss": 0.1002
    },
    {
      "epoch": 0.6597393545310307,
      "lm_loss": 0.5364,
      "step": 8992,
      "vm_loss": 0.1948
    },
    {
      "epoch": 0.6597393545310307,
      "lm_loss": 0.6289,
      "step": 8992,
      "vm_loss": 0.1654
    },
    {
      "epoch": 0.6597393545310307,
      "lm_loss": 0.4749,
      "step": 8992,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.6597393545310307,
      "lm_loss": 1.0545,
      "step": 8992,
      "vm_loss": 0.126
    },
    {
      "epoch": 0.6598127241211699,
      "grad_norm": 1.526332721539851,
      "learning_rate": 2.7395809065855848e-06,
      "loss": 0.7423,
      "step": 8993
    },
    {
      "epoch": 0.659886093711309,
      "grad_norm": 0.5583196342473141,
      "learning_rate": 2.7385211279282153e-06,
      "loss": 0.4718,
      "step": 8994
    },
    {
      "epoch": 0.6599594633014482,
      "grad_norm": 1.4540471652485363,
      "learning_rate": 2.7374614769821606e-06,
      "loss": 0.6713,
      "step": 8995
    },
    {
      "epoch": 0.6600328328915872,
      "grad_norm": 1.700371749740009,
      "learning_rate": 2.7364019538072596e-06,
      "loss": 0.8589,
      "step": 8996
    },
    {
      "epoch": 0.6601062024817264,
      "grad_norm": 1.5391155153196205,
      "learning_rate": 2.7353425584633464e-06,
      "loss": 0.7761,
      "step": 8997
    },
    {
      "epoch": 0.6601795720718655,
      "grad_norm": 1.4283034442035862,
      "learning_rate": 2.7342832910102464e-06,
      "loss": 0.77,
      "step": 8998
    },
    {
      "epoch": 0.6602529416620047,
      "grad_norm": 0.5676879716223945,
      "learning_rate": 2.7332241515077784e-06,
      "loss": 0.4864,
      "step": 8999
    },
    {
      "epoch": 0.6603263112521438,
      "grad_norm": 1.5108924469065053,
      "learning_rate": 2.732165140015758e-06,
      "loss": 0.8174,
      "step": 9000
    },
    {
      "epoch": 0.6603263112521438,
      "lm_loss": 1.2772,
      "step": 9000,
      "vm_loss": 0.0365
    },
    {
      "epoch": 0.6603263112521438,
      "lm_loss": 0.7896,
      "step": 9000,
      "vm_loss": 0.1206
    },
    {
      "epoch": 0.6603263112521438,
      "lm_loss": 0.3403,
      "step": 9000,
      "vm_loss": 0.0492
    },
    {
      "epoch": 0.6603263112521438,
      "lm_loss": 0.4561,
      "step": 9000,
      "vm_loss": 0.1347
    },
    {
      "epoch": 0.6603263112521438,
      "lm_loss": 1.0286,
      "step": 9000,
      "vm_loss": 0.1055
    },
    {
      "epoch": 0.6603263112521438,
      "lm_loss": 0.9223,
      "step": 9000,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.6603263112521438,
      "lm_loss": 0.5507,
      "step": 9000,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.6603263112521438,
      "lm_loss": 0.8364,
      "step": 9000,
      "vm_loss": 0.1236
    },
    {
      "epoch": 0.6603996808422828,
      "grad_norm": 1.62367522895999,
      "learning_rate": 2.7311062565939876e-06,
      "loss": 0.7383,
      "step": 9001
    },
    {
      "epoch": 0.660473050432422,
      "grad_norm": 1.51770929316972,
      "learning_rate": 2.7300475013022666e-06,
      "loss": 0.7856,
      "step": 9002
    },
    {
      "epoch": 0.6605464200225611,
      "grad_norm": 0.562236564580216,
      "learning_rate": 2.7289888742003844e-06,
      "loss": 0.489,
      "step": 9003
    },
    {
      "epoch": 0.6606197896127003,
      "grad_norm": 0.5659675987490357,
      "learning_rate": 2.7279303753481224e-06,
      "loss": 0.4896,
      "step": 9004
    },
    {
      "epoch": 0.6606931592028394,
      "grad_norm": 0.5601049453693193,
      "learning_rate": 2.726872004805261e-06,
      "loss": 0.4783,
      "step": 9005
    },
    {
      "epoch": 0.6607665287929785,
      "grad_norm": 1.5936045860162136,
      "learning_rate": 2.725813762631567e-06,
      "loss": 0.7196,
      "step": 9006
    },
    {
      "epoch": 0.6608398983831176,
      "grad_norm": 0.5654860197958739,
      "learning_rate": 2.724755648886801e-06,
      "loss": 0.4807,
      "step": 9007
    },
    {
      "epoch": 0.6609132679732568,
      "grad_norm": 1.4511822870028237,
      "learning_rate": 2.723697663630719e-06,
      "loss": 0.6533,
      "step": 9008
    },
    {
      "epoch": 0.6609132679732568,
      "lm_loss": 1.0278,
      "step": 9008,
      "vm_loss": 0.0451
    },
    {
      "epoch": 0.6609132679732568,
      "lm_loss": 0.0017,
      "step": 9008,
      "vm_loss": 0.0515
    },
    {
      "epoch": 0.6609132679732568,
      "lm_loss": 0.1749,
      "step": 9008,
      "vm_loss": 0.1601
    },
    {
      "epoch": 0.6609132679732568,
      "lm_loss": 0.4723,
      "step": 9008,
      "vm_loss": 0.1925
    },
    {
      "epoch": 0.6609132679732568,
      "lm_loss": 0.1628,
      "step": 9008,
      "vm_loss": 0.0564
    },
    {
      "epoch": 0.6609132679732568,
      "lm_loss": 0.1551,
      "step": 9008,
      "vm_loss": 0.1897
    },
    {
      "epoch": 0.6609132679732568,
      "lm_loss": 0.1287,
      "step": 9008,
      "vm_loss": 0.1294
    },
    {
      "epoch": 0.6609132679732568,
      "lm_loss": 0.1567,
      "step": 9008,
      "vm_loss": 0.1631
    },
    {
      "epoch": 0.6609866375633959,
      "grad_norm": 1.5499376330643826,
      "learning_rate": 2.722639806923066e-06,
      "loss": 0.6772,
      "step": 9009
    },
    {
      "epoch": 0.6610600071535351,
      "grad_norm": 1.5460084151606175,
      "learning_rate": 2.7215820788235823e-06,
      "loss": 0.7065,
      "step": 9010
    },
    {
      "epoch": 0.6611333767436741,
      "grad_norm": 1.454112671828495,
      "learning_rate": 2.720524479392002e-06,
      "loss": 0.6611,
      "step": 9011
    },
    {
      "epoch": 0.6612067463338133,
      "grad_norm": 1.3880595177938493,
      "learning_rate": 2.7194670086880497e-06,
      "loss": 0.7875,
      "step": 9012
    },
    {
      "epoch": 0.6612801159239524,
      "grad_norm": 1.4780868579161544,
      "learning_rate": 2.718409666771443e-06,
      "loss": 0.7091,
      "step": 9013
    },
    {
      "epoch": 0.6613534855140916,
      "grad_norm": 1.6916795931561068,
      "learning_rate": 2.7173524537018924e-06,
      "loss": 0.7717,
      "step": 9014
    },
    {
      "epoch": 0.6614268551042307,
      "grad_norm": 1.4919166532739103,
      "learning_rate": 2.7162953695391002e-06,
      "loss": 0.7119,
      "step": 9015
    },
    {
      "epoch": 0.6615002246943698,
      "grad_norm": 0.5591803876157311,
      "learning_rate": 2.7152384143427655e-06,
      "loss": 0.4796,
      "step": 9016
    },
    {
      "epoch": 0.6615002246943698,
      "lm_loss": 0.4154,
      "step": 9016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6615002246943698,
      "lm_loss": 0.4904,
      "step": 9016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6615002246943698,
      "lm_loss": 0.5275,
      "step": 9016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6615002246943698,
      "lm_loss": 0.5708,
      "step": 9016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6615002246943698,
      "lm_loss": 0.5576,
      "step": 9016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6615002246943698,
      "lm_loss": 0.4267,
      "step": 9016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6615002246943698,
      "lm_loss": 0.547,
      "step": 9016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6615002246943698,
      "lm_loss": 0.4616,
      "step": 9016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6615735942845089,
      "grad_norm": 0.5823098391039012,
      "learning_rate": 2.7141815881725765e-06,
      "loss": 0.4959,
      "step": 9017
    },
    {
      "epoch": 0.661646963874648,
      "grad_norm": 1.5657512407564957,
      "learning_rate": 2.7131248910882122e-06,
      "loss": 0.766,
      "step": 9018
    },
    {
      "epoch": 0.6617203334647872,
      "grad_norm": 0.5788258006113101,
      "learning_rate": 2.7120683231493503e-06,
      "loss": 0.4842,
      "step": 9019
    },
    {
      "epoch": 0.6617937030549264,
      "grad_norm": 1.5238404532189995,
      "learning_rate": 2.711011884415655e-06,
      "loss": 0.7602,
      "step": 9020
    },
    {
      "epoch": 0.6618670726450654,
      "grad_norm": 1.6493035249969166,
      "learning_rate": 2.7099555749467866e-06,
      "loss": 0.6908,
      "step": 9021
    },
    {
      "epoch": 0.6619404422352045,
      "grad_norm": 1.6367606653564593,
      "learning_rate": 2.708899394802399e-06,
      "loss": 0.7676,
      "step": 9022
    },
    {
      "epoch": 0.6620138118253437,
      "grad_norm": 1.4999872603827047,
      "learning_rate": 2.707843344042137e-06,
      "loss": 0.7376,
      "step": 9023
    },
    {
      "epoch": 0.6620871814154828,
      "grad_norm": 0.5531923791204593,
      "learning_rate": 2.7067874227256374e-06,
      "loss": 0.4793,
      "step": 9024
    },
    {
      "epoch": 0.6620871814154828,
      "lm_loss": 0.5125,
      "step": 9024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6620871814154828,
      "lm_loss": 0.3046,
      "step": 9024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6620871814154828,
      "lm_loss": 0.4394,
      "step": 9024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6620871814154828,
      "lm_loss": 0.328,
      "step": 9024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6620871814154828,
      "lm_loss": 0.434,
      "step": 9024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6620871814154828,
      "lm_loss": 0.3361,
      "step": 9024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6620871814154828,
      "lm_loss": 0.5771,
      "step": 9024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6620871814154828,
      "lm_loss": 0.5252,
      "step": 9024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.662160551005622,
      "grad_norm": 0.5638586037238387,
      "learning_rate": 2.705731630912531e-06,
      "loss": 0.4776,
      "step": 9025
    },
    {
      "epoch": 0.662233920595761,
      "grad_norm": 1.3848761545493633,
      "learning_rate": 2.7046759686624393e-06,
      "loss": 0.7612,
      "step": 9026
    },
    {
      "epoch": 0.6623072901859002,
      "grad_norm": 1.3754870302021598,
      "learning_rate": 2.7036204360349815e-06,
      "loss": 0.7879,
      "step": 9027
    },
    {
      "epoch": 0.6623806597760393,
      "grad_norm": 0.5589622101877182,
      "learning_rate": 2.7025650330897646e-06,
      "loss": 0.4857,
      "step": 9028
    },
    {
      "epoch": 0.6624540293661785,
      "grad_norm": 1.5189099732071534,
      "learning_rate": 2.7015097598863906e-06,
      "loss": 0.7636,
      "step": 9029
    },
    {
      "epoch": 0.6625273989563176,
      "grad_norm": 0.5682982623152036,
      "learning_rate": 2.7004546164844515e-06,
      "loss": 0.4827,
      "step": 9030
    },
    {
      "epoch": 0.6626007685464567,
      "grad_norm": 0.5607257923255172,
      "learning_rate": 2.6993996029435337e-06,
      "loss": 0.4919,
      "step": 9031
    },
    {
      "epoch": 0.6626741381365958,
      "grad_norm": 1.4894334943095888,
      "learning_rate": 2.6983447193232195e-06,
      "loss": 0.7884,
      "step": 9032
    },
    {
      "epoch": 0.6626741381365958,
      "lm_loss": 0.3036,
      "step": 9032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6626741381365958,
      "lm_loss": 0.3999,
      "step": 9032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6626741381365958,
      "lm_loss": 0.4983,
      "step": 9032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6626741381365958,
      "lm_loss": 0.5214,
      "step": 9032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6626741381365958,
      "lm_loss": 0.5098,
      "step": 9032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6626741381365958,
      "lm_loss": 0.4923,
      "step": 9032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6626741381365958,
      "lm_loss": 0.4419,
      "step": 9032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6626741381365958,
      "lm_loss": 0.4345,
      "step": 9032,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.662747507726735,
      "grad_norm": 0.5532889908705995,
      "learning_rate": 2.6972899656830786e-06,
      "loss": 0.4979,
      "step": 9033
    },
    {
      "epoch": 0.6628208773168741,
      "grad_norm": 1.3716519107655836,
      "learning_rate": 2.6962353420826766e-06,
      "loss": 0.7177,
      "step": 9034
    },
    {
      "epoch": 0.6628942469070133,
      "grad_norm": 1.6249449004891474,
      "learning_rate": 2.6951808485815695e-06,
      "loss": 0.7229,
      "step": 9035
    },
    {
      "epoch": 0.6629676164971523,
      "grad_norm": 1.505028659458241,
      "learning_rate": 2.694126485239308e-06,
      "loss": 0.727,
      "step": 9036
    },
    {
      "epoch": 0.6630409860872915,
      "grad_norm": 0.5861932781501663,
      "learning_rate": 2.6930722521154318e-06,
      "loss": 0.4935,
      "step": 9037
    },
    {
      "epoch": 0.6631143556774306,
      "grad_norm": 1.480134786542435,
      "learning_rate": 2.692018149269481e-06,
      "loss": 0.8538,
      "step": 9038
    },
    {
      "epoch": 0.6631877252675697,
      "grad_norm": 1.5548802897670668,
      "learning_rate": 2.6909641767609806e-06,
      "loss": 0.6971,
      "step": 9039
    },
    {
      "epoch": 0.6632610948577089,
      "grad_norm": 1.6095617545976288,
      "learning_rate": 2.6899103346494514e-06,
      "loss": 0.7174,
      "step": 9040
    },
    {
      "epoch": 0.6632610948577089,
      "lm_loss": 0.3908,
      "step": 9040,
      "vm_loss": 0.0443
    },
    {
      "epoch": 0.6632610948577089,
      "lm_loss": 1.0317,
      "step": 9040,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.6632610948577089,
      "lm_loss": 0.2717,
      "step": 9040,
      "vm_loss": 0.0494
    },
    {
      "epoch": 0.6632610948577089,
      "lm_loss": 0.5384,
      "step": 9040,
      "vm_loss": 0.0775
    },
    {
      "epoch": 0.6632610948577089,
      "lm_loss": 0.9829,
      "step": 9040,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.6632610948577089,
      "lm_loss": 0.0139,
      "step": 9040,
      "vm_loss": 0.173
    },
    {
      "epoch": 0.6632610948577089,
      "lm_loss": 0.3219,
      "step": 9040,
      "vm_loss": 0.1389
    },
    {
      "epoch": 0.6632610948577089,
      "lm_loss": 0.2601,
      "step": 9040,
      "vm_loss": 0.085
    },
    {
      "epoch": 0.6633344644478479,
      "grad_norm": 1.6029552988967375,
      "learning_rate": 2.6888566229944063e-06,
      "loss": 0.8365,
      "step": 9041
    },
    {
      "epoch": 0.6634078340379871,
      "grad_norm": 0.5776146765403523,
      "learning_rate": 2.687803041855349e-06,
      "loss": 0.4743,
      "step": 9042
    },
    {
      "epoch": 0.6634812036281262,
      "grad_norm": 1.4618155364384608,
      "learning_rate": 2.686749591291783e-06,
      "loss": 0.7356,
      "step": 9043
    },
    {
      "epoch": 0.6635545732182654,
      "grad_norm": 1.3375617414633054,
      "learning_rate": 2.685696271363195e-06,
      "loss": 0.6789,
      "step": 9044
    },
    {
      "epoch": 0.6636279428084045,
      "grad_norm": 0.5594203878112592,
      "learning_rate": 2.6846430821290708e-06,
      "loss": 0.4926,
      "step": 9045
    },
    {
      "epoch": 0.6637013123985436,
      "grad_norm": 1.4533589638197852,
      "learning_rate": 2.6835900236488856e-06,
      "loss": 0.7645,
      "step": 9046
    },
    {
      "epoch": 0.6637746819886827,
      "grad_norm": 1.4532584858898545,
      "learning_rate": 2.6825370959821073e-06,
      "loss": 0.8188,
      "step": 9047
    },
    {
      "epoch": 0.6638480515788219,
      "grad_norm": 1.571999951642575,
      "learning_rate": 2.6814842991881997e-06,
      "loss": 0.7721,
      "step": 9048
    },
    {
      "epoch": 0.6638480515788219,
      "lm_loss": 1.396,
      "step": 9048,
      "vm_loss": 0.0659
    },
    {
      "epoch": 0.6638480515788219,
      "lm_loss": 1.3341,
      "step": 9048,
      "vm_loss": 0.1855
    },
    {
      "epoch": 0.6638480515788219,
      "lm_loss": 0.0127,
      "step": 9048,
      "vm_loss": 0.1152
    },
    {
      "epoch": 0.6638480515788219,
      "lm_loss": 1.0602,
      "step": 9048,
      "vm_loss": 0.1304
    },
    {
      "epoch": 0.6638480515788219,
      "lm_loss": 0.7125,
      "step": 9048,
      "vm_loss": 0.1529
    },
    {
      "epoch": 0.6638480515788219,
      "lm_loss": 0.0183,
      "step": 9048,
      "vm_loss": 0.1596
    },
    {
      "epoch": 0.6638480515788219,
      "lm_loss": 1.0564,
      "step": 9048,
      "vm_loss": 0.1196
    },
    {
      "epoch": 0.6638480515788219,
      "lm_loss": 0.9742,
      "step": 9048,
      "vm_loss": 0.1352
    },
    {
      "epoch": 0.663921421168961,
      "grad_norm": 1.5509694093566742,
      "learning_rate": 2.680431633326614e-06,
      "loss": 0.7831,
      "step": 9049
    },
    {
      "epoch": 0.6639947907591002,
      "grad_norm": 0.5647288439503718,
      "learning_rate": 2.6793790984567993e-06,
      "loss": 0.4832,
      "step": 9050
    },
    {
      "epoch": 0.6640681603492392,
      "grad_norm": 0.5632137433525058,
      "learning_rate": 2.678326694638193e-06,
      "loss": 0.4879,
      "step": 9051
    },
    {
      "epoch": 0.6641415299393784,
      "grad_norm": 1.4580765151441426,
      "learning_rate": 2.677274421930226e-06,
      "loss": 0.7725,
      "step": 9052
    },
    {
      "epoch": 0.6642148995295175,
      "grad_norm": 1.3853447544700683,
      "learning_rate": 2.6762222803923273e-06,
      "loss": 0.7214,
      "step": 9053
    },
    {
      "epoch": 0.6642882691196567,
      "grad_norm": 0.5504015206849537,
      "learning_rate": 2.6751702700839106e-06,
      "loss": 0.481,
      "step": 9054
    },
    {
      "epoch": 0.6643616387097958,
      "grad_norm": 1.6278474909187675,
      "learning_rate": 2.6741183910643858e-06,
      "loss": 0.7776,
      "step": 9055
    },
    {
      "epoch": 0.6644350082999348,
      "grad_norm": 1.5508999997930033,
      "learning_rate": 2.6730666433931564e-06,
      "loss": 0.6755,
      "step": 9056
    },
    {
      "epoch": 0.6644350082999348,
      "lm_loss": 0.6125,
      "step": 9056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6644350082999348,
      "lm_loss": 0.4428,
      "step": 9056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6644350082999348,
      "lm_loss": 0.5078,
      "step": 9056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6644350082999348,
      "lm_loss": 0.5199,
      "step": 9056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6644350082999348,
      "lm_loss": 0.563,
      "step": 9056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6644350082999348,
      "lm_loss": 0.4298,
      "step": 9056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6644350082999348,
      "lm_loss": 0.4278,
      "step": 9056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6644350082999348,
      "lm_loss": 0.5058,
      "step": 9056,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.664508377890074,
      "grad_norm": 0.5345378477109627,
      "learning_rate": 2.672015027129613e-06,
      "loss": 0.4828,
      "step": 9057
    },
    {
      "epoch": 0.6645817474802131,
      "grad_norm": 1.5153916286730582,
      "learning_rate": 2.6709635423331487e-06,
      "loss": 0.7528,
      "step": 9058
    },
    {
      "epoch": 0.6646551170703523,
      "grad_norm": 1.4952695218636143,
      "learning_rate": 2.669912189063141e-06,
      "loss": 0.7216,
      "step": 9059
    },
    {
      "epoch": 0.6647284866604914,
      "grad_norm": 0.5483675532513511,
      "learning_rate": 2.6688609673789625e-06,
      "loss": 0.4902,
      "step": 9060
    },
    {
      "epoch": 0.6648018562506305,
      "grad_norm": 1.515224916680528,
      "learning_rate": 2.6678098773399773e-06,
      "loss": 0.7388,
      "step": 9061
    },
    {
      "epoch": 0.6648752258407696,
      "grad_norm": 0.569330749128332,
      "learning_rate": 2.666758919005545e-06,
      "loss": 0.5006,
      "step": 9062
    },
    {
      "epoch": 0.6649485954309088,
      "grad_norm": 1.75656745889406,
      "learning_rate": 2.665708092435012e-06,
      "loss": 0.6845,
      "step": 9063
    },
    {
      "epoch": 0.6650219650210479,
      "grad_norm": 0.5674338962157656,
      "learning_rate": 2.664657397687726e-06,
      "loss": 0.4849,
      "step": 9064
    },
    {
      "epoch": 0.6650219650210479,
      "lm_loss": 1.4625,
      "step": 9064,
      "vm_loss": 0.0798
    },
    {
      "epoch": 0.6650219650210479,
      "lm_loss": 0.5175,
      "step": 9064,
      "vm_loss": 0.1141
    },
    {
      "epoch": 0.6650219650210479,
      "lm_loss": 0.3801,
      "step": 9064,
      "vm_loss": 0.146
    },
    {
      "epoch": 0.6650219650210479,
      "lm_loss": 0.9198,
      "step": 9064,
      "vm_loss": 0.151
    },
    {
      "epoch": 0.6650219650210479,
      "lm_loss": 0.0927,
      "step": 9064,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.6650219650210479,
      "lm_loss": 0.0075,
      "step": 9064,
      "vm_loss": 0.0763
    },
    {
      "epoch": 0.6650219650210479,
      "lm_loss": 0.1058,
      "step": 9064,
      "vm_loss": 0.214
    },
    {
      "epoch": 0.6650219650210479,
      "lm_loss": 1.4798,
      "step": 9064,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.6650953346111871,
      "grad_norm": 1.46949547663491,
      "learning_rate": 2.6636068348230202e-06,
      "loss": 0.7419,
      "step": 9065
    },
    {
      "epoch": 0.6651687042013261,
      "grad_norm": 1.551510862084649,
      "learning_rate": 2.662556403900223e-06,
      "loss": 0.7141,
      "step": 9066
    },
    {
      "epoch": 0.6652420737914653,
      "grad_norm": 1.6755270600941852,
      "learning_rate": 2.661506104978653e-06,
      "loss": 0.8022,
      "step": 9067
    },
    {
      "epoch": 0.6653154433816044,
      "grad_norm": 1.4130074838457998,
      "learning_rate": 2.660455938117623e-06,
      "loss": 0.7597,
      "step": 9068
    },
    {
      "epoch": 0.6653888129717436,
      "grad_norm": 1.3534860478752884,
      "learning_rate": 2.659405903376442e-06,
      "loss": 0.6374,
      "step": 9069
    },
    {
      "epoch": 0.6654621825618826,
      "grad_norm": 1.5208072857111445,
      "learning_rate": 2.658356000814406e-06,
      "loss": 0.7713,
      "step": 9070
    },
    {
      "epoch": 0.6655355521520218,
      "grad_norm": 0.5720150592363681,
      "learning_rate": 2.657306230490806e-06,
      "loss": 0.4955,
      "step": 9071
    },
    {
      "epoch": 0.6656089217421609,
      "grad_norm": 1.3487935137236786,
      "learning_rate": 2.656256592464924e-06,
      "loss": 0.6315,
      "step": 9072
    },
    {
      "epoch": 0.6656089217421609,
      "lm_loss": 0.4939,
      "step": 9072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6656089217421609,
      "lm_loss": 0.4905,
      "step": 9072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6656089217421609,
      "lm_loss": 0.6109,
      "step": 9072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6656089217421609,
      "lm_loss": 0.6542,
      "step": 9072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6656089217421609,
      "lm_loss": 0.383,
      "step": 9072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6656089217421609,
      "lm_loss": 0.5809,
      "step": 9072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6656089217421609,
      "lm_loss": 0.4084,
      "step": 9072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6656089217421609,
      "lm_loss": 0.6009,
      "step": 9072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6656822913323001,
      "grad_norm": 0.5709580611201911,
      "learning_rate": 2.655207086796035e-06,
      "loss": 0.5082,
      "step": 9073
    },
    {
      "epoch": 0.6657556609224392,
      "grad_norm": 0.5545022501094007,
      "learning_rate": 2.6541577135434103e-06,
      "loss": 0.483,
      "step": 9074
    },
    {
      "epoch": 0.6658290305125782,
      "grad_norm": 0.560616607663133,
      "learning_rate": 2.6531084727663093e-06,
      "loss": 0.4846,
      "step": 9075
    },
    {
      "epoch": 0.6659024001027174,
      "grad_norm": 1.4519254500146797,
      "learning_rate": 2.6520593645239843e-06,
      "loss": 0.7385,
      "step": 9076
    },
    {
      "epoch": 0.6659757696928565,
      "grad_norm": 1.4484639139411775,
      "learning_rate": 2.6510103888756822e-06,
      "loss": 0.7696,
      "step": 9077
    },
    {
      "epoch": 0.6660491392829957,
      "grad_norm": 1.5653107604338639,
      "learning_rate": 2.64996154588064e-06,
      "loss": 0.7019,
      "step": 9078
    },
    {
      "epoch": 0.6661225088731348,
      "grad_norm": 0.5456914053351232,
      "learning_rate": 2.6489128355980877e-06,
      "loss": 0.4964,
      "step": 9079
    },
    {
      "epoch": 0.6661958784632739,
      "grad_norm": 0.5530259893839988,
      "learning_rate": 2.647864258087251e-06,
      "loss": 0.4849,
      "step": 9080
    },
    {
      "epoch": 0.6661958784632739,
      "lm_loss": 1.3877,
      "step": 9080,
      "vm_loss": 0.0687
    },
    {
      "epoch": 0.6661958784632739,
      "lm_loss": 0.4151,
      "step": 9080,
      "vm_loss": 0.1144
    },
    {
      "epoch": 0.6661958784632739,
      "lm_loss": 0.4358,
      "step": 9080,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.6661958784632739,
      "lm_loss": 1.542,
      "step": 9080,
      "vm_loss": 0.1015
    },
    {
      "epoch": 0.6661958784632739,
      "lm_loss": 0.2172,
      "step": 9080,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.6661958784632739,
      "lm_loss": 1.0319,
      "step": 9080,
      "vm_loss": 0.1606
    },
    {
      "epoch": 0.6661958784632739,
      "lm_loss": 0.031,
      "step": 9080,
      "vm_loss": 0.0728
    },
    {
      "epoch": 0.6661958784632739,
      "lm_loss": 0.1084,
      "step": 9080,
      "vm_loss": 0.0943
    },
    {
      "epoch": 0.666269248053413,
      "grad_norm": 1.8238716002493698,
      "learning_rate": 2.6468158134073452e-06,
      "loss": 0.7772,
      "step": 9081
    },
    {
      "epoch": 0.6663426176435522,
      "grad_norm": 0.547083799957933,
      "learning_rate": 2.6457675016175775e-06,
      "loss": 0.485,
      "step": 9082
    },
    {
      "epoch": 0.6664159872336913,
      "grad_norm": 1.46381950251115,
      "learning_rate": 2.6447193227771496e-06,
      "loss": 0.7032,
      "step": 9083
    },
    {
      "epoch": 0.6664893568238305,
      "grad_norm": 1.5876266785374142,
      "learning_rate": 2.643671276945252e-06,
      "loss": 0.7679,
      "step": 9084
    },
    {
      "epoch": 0.6665627264139695,
      "grad_norm": 0.5805578147309179,
      "learning_rate": 2.6426233641810757e-06,
      "loss": 0.502,
      "step": 9085
    },
    {
      "epoch": 0.6666360960041087,
      "grad_norm": 1.4684755196981742,
      "learning_rate": 2.6415755845437954e-06,
      "loss": 0.6567,
      "step": 9086
    },
    {
      "epoch": 0.6667094655942478,
      "grad_norm": 1.4309486548588406,
      "learning_rate": 2.640527938092582e-06,
      "loss": 0.722,
      "step": 9087
    },
    {
      "epoch": 0.666782835184387,
      "grad_norm": 1.4658434554258086,
      "learning_rate": 2.639480424886599e-06,
      "loss": 0.6935,
      "step": 9088
    },
    {
      "epoch": 0.666782835184387,
      "lm_loss": 1.0865,
      "step": 9088,
      "vm_loss": 0.0683
    },
    {
      "epoch": 0.666782835184387,
      "lm_loss": 0.7552,
      "step": 9088,
      "vm_loss": 0.1871
    },
    {
      "epoch": 0.666782835184387,
      "lm_loss": 0.7151,
      "step": 9088,
      "vm_loss": 0.106
    },
    {
      "epoch": 0.666782835184387,
      "lm_loss": 0.1071,
      "step": 9088,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.666782835184387,
      "lm_loss": 0.8798,
      "step": 9088,
      "vm_loss": 0.1252
    },
    {
      "epoch": 0.666782835184387,
      "lm_loss": 0.0604,
      "step": 9088,
      "vm_loss": 0.1789
    },
    {
      "epoch": 0.666782835184387,
      "lm_loss": 1.1309,
      "step": 9088,
      "vm_loss": 0.1206
    },
    {
      "epoch": 0.666782835184387,
      "lm_loss": 0.0165,
      "step": 9088,
      "vm_loss": 0.0426
    },
    {
      "epoch": 0.6668562047745261,
      "grad_norm": 1.4046200786544196,
      "learning_rate": 2.6384330449850028e-06,
      "loss": 0.786,
      "step": 9089
    },
    {
      "epoch": 0.6669295743646652,
      "grad_norm": 1.616904525600034,
      "learning_rate": 2.6373857984469385e-06,
      "loss": 0.7812,
      "step": 9090
    },
    {
      "epoch": 0.6670029439548043,
      "grad_norm": 0.5621070557102501,
      "learning_rate": 2.636338685331551e-06,
      "loss": 0.4738,
      "step": 9091
    },
    {
      "epoch": 0.6670763135449435,
      "grad_norm": 0.573313335190657,
      "learning_rate": 2.6352917056979715e-06,
      "loss": 0.4807,
      "step": 9092
    },
    {
      "epoch": 0.6671496831350826,
      "grad_norm": 1.559710633785105,
      "learning_rate": 2.6342448596053256e-06,
      "loss": 0.8082,
      "step": 9093
    },
    {
      "epoch": 0.6672230527252218,
      "grad_norm": 1.4452968669536852,
      "learning_rate": 2.633198147112731e-06,
      "loss": 0.7965,
      "step": 9094
    },
    {
      "epoch": 0.6672964223153608,
      "grad_norm": 1.4915195084498505,
      "learning_rate": 2.6321515682792965e-06,
      "loss": 0.7786,
      "step": 9095
    },
    {
      "epoch": 0.6673697919054999,
      "grad_norm": 1.4129497000504694,
      "learning_rate": 2.631105123164129e-06,
      "loss": 0.7044,
      "step": 9096
    },
    {
      "epoch": 0.6673697919054999,
      "lm_loss": 0.0443,
      "step": 9096,
      "vm_loss": 0.0253
    },
    {
      "epoch": 0.6673697919054999,
      "lm_loss": 0.2659,
      "step": 9096,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.6673697919054999,
      "lm_loss": 0.6257,
      "step": 9096,
      "vm_loss": 0.1446
    },
    {
      "epoch": 0.6673697919054999,
      "lm_loss": 0.5948,
      "step": 9096,
      "vm_loss": 0.2361
    },
    {
      "epoch": 0.6673697919054999,
      "lm_loss": 0.7949,
      "step": 9096,
      "vm_loss": 0.1985
    },
    {
      "epoch": 0.6673697919054999,
      "lm_loss": 0.5379,
      "step": 9096,
      "vm_loss": 0.0279
    },
    {
      "epoch": 0.6673697919054999,
      "lm_loss": 0.7277,
      "step": 9096,
      "vm_loss": 0.1544
    },
    {
      "epoch": 0.6673697919054999,
      "lm_loss": 0.4871,
      "step": 9096,
      "vm_loss": 0.149
    },
    {
      "epoch": 0.6674431614956391,
      "grad_norm": 1.660617370330867,
      "learning_rate": 2.630058811826321e-06,
      "loss": 0.6946,
      "step": 9097
    },
    {
      "epoch": 0.6675165310857782,
      "grad_norm": 0.5668299104601251,
      "learning_rate": 2.6290126343249613e-06,
      "loss": 0.4812,
      "step": 9098
    },
    {
      "epoch": 0.6675899006759174,
      "grad_norm": 0.5350435397344306,
      "learning_rate": 2.62796659071913e-06,
      "loss": 0.485,
      "step": 9099
    },
    {
      "epoch": 0.6676632702660564,
      "grad_norm": 1.8329475265966404,
      "learning_rate": 2.626920681067897e-06,
      "loss": 0.7853,
      "step": 9100
    },
    {
      "epoch": 0.6677366398561956,
      "grad_norm": 1.5877669907277203,
      "learning_rate": 2.625874905430333e-06,
      "loss": 0.7708,
      "step": 9101
    },
    {
      "epoch": 0.6678100094463347,
      "grad_norm": 1.5125152304843712,
      "learning_rate": 2.6248292638654917e-06,
      "loss": 0.729,
      "step": 9102
    },
    {
      "epoch": 0.6678833790364739,
      "grad_norm": 1.5001915824011356,
      "learning_rate": 2.6237837564324244e-06,
      "loss": 0.6818,
      "step": 9103
    },
    {
      "epoch": 0.667956748626613,
      "grad_norm": 0.54761153956589,
      "learning_rate": 2.622738383190172e-06,
      "loss": 0.4752,
      "step": 9104
    },
    {
      "epoch": 0.667956748626613,
      "lm_loss": 1.7664,
      "step": 9104,
      "vm_loss": 0.0805
    },
    {
      "epoch": 0.667956748626613,
      "lm_loss": 1.6845,
      "step": 9104,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.667956748626613,
      "lm_loss": 0.7453,
      "step": 9104,
      "vm_loss": 0.0791
    },
    {
      "epoch": 0.667956748626613,
      "lm_loss": 0.2187,
      "step": 9104,
      "vm_loss": 0.1393
    },
    {
      "epoch": 0.667956748626613,
      "lm_loss": 0.7919,
      "step": 9104,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.667956748626613,
      "lm_loss": 0.2639,
      "step": 9104,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.667956748626613,
      "lm_loss": 0.5065,
      "step": 9104,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.667956748626613,
      "lm_loss": 0.2676,
      "step": 9104,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.6680301182167521,
      "grad_norm": 1.7125277358581767,
      "learning_rate": 2.621693144197771e-06,
      "loss": 0.7813,
      "step": 9105
    },
    {
      "epoch": 0.6681034878068912,
      "grad_norm": 1.7015012426065375,
      "learning_rate": 2.6206480395142462e-06,
      "loss": 0.7172,
      "step": 9106
    },
    {
      "epoch": 0.6681768573970304,
      "grad_norm": 0.5637536928341116,
      "learning_rate": 2.6196030691986206e-06,
      "loss": 0.496,
      "step": 9107
    },
    {
      "epoch": 0.6682502269871695,
      "grad_norm": 1.9151271569155157,
      "learning_rate": 2.6185582333099046e-06,
      "loss": 0.7821,
      "step": 9108
    },
    {
      "epoch": 0.6683235965773087,
      "grad_norm": 0.5340893317390574,
      "learning_rate": 2.617513531907103e-06,
      "loss": 0.4749,
      "step": 9109
    },
    {
      "epoch": 0.6683969661674477,
      "grad_norm": 1.6656904471393528,
      "learning_rate": 2.6164689650492116e-06,
      "loss": 0.7523,
      "step": 9110
    },
    {
      "epoch": 0.6684703357575869,
      "grad_norm": 1.891927296291667,
      "learning_rate": 2.6154245327952196e-06,
      "loss": 0.7633,
      "step": 9111
    },
    {
      "epoch": 0.668543705347726,
      "grad_norm": 1.6771905224692347,
      "learning_rate": 2.6143802352041103e-06,
      "loss": 0.6812,
      "step": 9112
    },
    {
      "epoch": 0.668543705347726,
      "lm_loss": 2.1126,
      "step": 9112,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.668543705347726,
      "lm_loss": 0.7893,
      "step": 9112,
      "vm_loss": 0.0357
    },
    {
      "epoch": 0.668543705347726,
      "lm_loss": 1.0535,
      "step": 9112,
      "vm_loss": 0.1689
    },
    {
      "epoch": 0.668543705347726,
      "lm_loss": 0.9759,
      "step": 9112,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.668543705347726,
      "lm_loss": 0.576,
      "step": 9112,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.668543705347726,
      "lm_loss": 0.6101,
      "step": 9112,
      "vm_loss": 0.1192
    },
    {
      "epoch": 0.668543705347726,
      "lm_loss": 1.1467,
      "step": 9112,
      "vm_loss": 0.0733
    },
    {
      "epoch": 0.668543705347726,
      "lm_loss": 0.1831,
      "step": 9112,
      "vm_loss": 0.153
    },
    {
      "epoch": 0.6686170749378652,
      "grad_norm": 1.4981475359974352,
      "learning_rate": 2.6133360723348576e-06,
      "loss": 0.7638,
      "step": 9113
    },
    {
      "epoch": 0.6686904445280043,
      "grad_norm": 1.6189348704178887,
      "learning_rate": 2.6122920442464276e-06,
      "loss": 0.7505,
      "step": 9114
    },
    {
      "epoch": 0.6687638141181433,
      "grad_norm": 1.8974243019654922,
      "learning_rate": 2.611248150997779e-06,
      "loss": 0.7051,
      "step": 9115
    },
    {
      "epoch": 0.6688371837082825,
      "grad_norm": 1.4590903942326525,
      "learning_rate": 2.6102043926478627e-06,
      "loss": 0.6885,
      "step": 9116
    },
    {
      "epoch": 0.6689105532984216,
      "grad_norm": 0.5556090746041847,
      "learning_rate": 2.609160769255621e-06,
      "loss": 0.4927,
      "step": 9117
    },
    {
      "epoch": 0.6689839228885608,
      "grad_norm": 0.5416141896002183,
      "learning_rate": 2.6081172808799928e-06,
      "loss": 0.4833,
      "step": 9118
    },
    {
      "epoch": 0.6690572924786999,
      "grad_norm": 1.421959129034809,
      "learning_rate": 2.607073927579905e-06,
      "loss": 0.6412,
      "step": 9119
    },
    {
      "epoch": 0.669130662068839,
      "grad_norm": 1.6932978070364493,
      "learning_rate": 2.6060307094142788e-06,
      "loss": 0.7119,
      "step": 9120
    },
    {
      "epoch": 0.669130662068839,
      "lm_loss": 0.2983,
      "step": 9120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.669130662068839,
      "lm_loss": 0.6063,
      "step": 9120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.669130662068839,
      "lm_loss": 0.6425,
      "step": 9120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.669130662068839,
      "lm_loss": 0.458,
      "step": 9120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.669130662068839,
      "lm_loss": 0.5201,
      "step": 9120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.669130662068839,
      "lm_loss": 0.6954,
      "step": 9120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.669130662068839,
      "lm_loss": 0.4114,
      "step": 9120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.669130662068839,
      "lm_loss": 0.628,
      "step": 9120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6692040316589781,
      "grad_norm": 0.5595081805549323,
      "learning_rate": 2.604987626442027e-06,
      "loss": 0.4697,
      "step": 9121
    },
    {
      "epoch": 0.6692774012491173,
      "grad_norm": 1.4061400527888932,
      "learning_rate": 2.6039446787220535e-06,
      "loss": 0.7705,
      "step": 9122
    },
    {
      "epoch": 0.6693507708392564,
      "grad_norm": 1.621754771279566,
      "learning_rate": 2.6029018663132593e-06,
      "loss": 0.7597,
      "step": 9123
    },
    {
      "epoch": 0.6694241404293956,
      "grad_norm": 0.5517672633408194,
      "learning_rate": 2.6018591892745332e-06,
      "loss": 0.4812,
      "step": 9124
    },
    {
      "epoch": 0.6694975100195346,
      "grad_norm": 0.5550098373110229,
      "learning_rate": 2.600816647664758e-06,
      "loss": 0.5025,
      "step": 9125
    },
    {
      "epoch": 0.6695708796096738,
      "grad_norm": 1.4802914995454686,
      "learning_rate": 2.5997742415428072e-06,
      "loss": 0.9147,
      "step": 9126
    },
    {
      "epoch": 0.6696442491998129,
      "grad_norm": 1.6355791910456354,
      "learning_rate": 2.5987319709675474e-06,
      "loss": 0.6809,
      "step": 9127
    },
    {
      "epoch": 0.6697176187899521,
      "grad_norm": 1.4252040751066524,
      "learning_rate": 2.597689835997842e-06,
      "loss": 0.6839,
      "step": 9128
    },
    {
      "epoch": 0.6697176187899521,
      "lm_loss": 0.4422,
      "step": 9128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6697176187899521,
      "lm_loss": 0.2572,
      "step": 9128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6697176187899521,
      "lm_loss": 0.5637,
      "step": 9128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6697176187899521,
      "lm_loss": 0.448,
      "step": 9128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6697176187899521,
      "lm_loss": 0.5533,
      "step": 9128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6697176187899521,
      "lm_loss": 0.4112,
      "step": 9128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6697176187899521,
      "lm_loss": 0.6465,
      "step": 9128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6697176187899521,
      "lm_loss": 0.689,
      "step": 9128,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6697909883800912,
      "grad_norm": 0.5495516751972687,
      "learning_rate": 2.5966478366925406e-06,
      "loss": 0.4959,
      "step": 9129
    },
    {
      "epoch": 0.6698643579702303,
      "grad_norm": 1.71983408141525,
      "learning_rate": 2.5956059731104876e-06,
      "loss": 0.7118,
      "step": 9130
    },
    {
      "epoch": 0.6699377275603694,
      "grad_norm": 1.361851165233703,
      "learning_rate": 2.59456424531052e-06,
      "loss": 0.672,
      "step": 9131
    },
    {
      "epoch": 0.6700110971505085,
      "grad_norm": 1.514320306511683,
      "learning_rate": 2.5935226533514667e-06,
      "loss": 0.6809,
      "step": 9132
    },
    {
      "epoch": 0.6700844667406477,
      "grad_norm": 1.3349733521280185,
      "learning_rate": 2.592481197292146e-06,
      "loss": 0.6458,
      "step": 9133
    },
    {
      "epoch": 0.6701578363307868,
      "grad_norm": 1.5185340285524924,
      "learning_rate": 2.591439877191377e-06,
      "loss": 0.7753,
      "step": 9134
    },
    {
      "epoch": 0.6702312059209259,
      "grad_norm": 0.5418688479836139,
      "learning_rate": 2.590398693107962e-06,
      "loss": 0.4785,
      "step": 9135
    },
    {
      "epoch": 0.670304575511065,
      "grad_norm": 1.3524351053381805,
      "learning_rate": 2.589357645100701e-06,
      "loss": 0.6514,
      "step": 9136
    },
    {
      "epoch": 0.670304575511065,
      "lm_loss": 1.3912,
      "step": 9136,
      "vm_loss": 0.0489
    },
    {
      "epoch": 0.670304575511065,
      "lm_loss": 0.9046,
      "step": 9136,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.670304575511065,
      "lm_loss": 0.887,
      "step": 9136,
      "vm_loss": 0.0872
    },
    {
      "epoch": 0.670304575511065,
      "lm_loss": 1.1319,
      "step": 9136,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.670304575511065,
      "lm_loss": 0.7305,
      "step": 9136,
      "vm_loss": 0.1405
    },
    {
      "epoch": 0.670304575511065,
      "lm_loss": 0.9749,
      "step": 9136,
      "vm_loss": 0.193
    },
    {
      "epoch": 0.670304575511065,
      "lm_loss": 0.8414,
      "step": 9136,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.670304575511065,
      "lm_loss": 0.3636,
      "step": 9136,
      "vm_loss": 0.0848
    },
    {
      "epoch": 0.6703779451012042,
      "grad_norm": 1.4406454869734167,
      "learning_rate": 2.5883167332283824e-06,
      "loss": 0.8532,
      "step": 9137
    },
    {
      "epoch": 0.6704513146913433,
      "grad_norm": 1.4078034820941099,
      "learning_rate": 2.5872759575497893e-06,
      "loss": 0.7524,
      "step": 9138
    },
    {
      "epoch": 0.6705246842814825,
      "grad_norm": 1.7763325109625445,
      "learning_rate": 2.5862353181236998e-06,
      "loss": 0.7525,
      "step": 9139
    },
    {
      "epoch": 0.6705980538716215,
      "grad_norm": 0.5439108484271293,
      "learning_rate": 2.585194815008879e-06,
      "loss": 0.4798,
      "step": 9140
    },
    {
      "epoch": 0.6706714234617607,
      "grad_norm": 0.5661285271283613,
      "learning_rate": 2.584154448264087e-06,
      "loss": 0.4653,
      "step": 9141
    },
    {
      "epoch": 0.6707447930518998,
      "grad_norm": 0.5549798700732385,
      "learning_rate": 2.583114217948077e-06,
      "loss": 0.4705,
      "step": 9142
    },
    {
      "epoch": 0.670818162642039,
      "grad_norm": 1.533041769628648,
      "learning_rate": 2.5820741241195913e-06,
      "loss": 0.7455,
      "step": 9143
    },
    {
      "epoch": 0.6708915322321781,
      "grad_norm": 0.556299237974991,
      "learning_rate": 2.5810341668373666e-06,
      "loss": 0.4904,
      "step": 9144
    },
    {
      "epoch": 0.6708915322321781,
      "lm_loss": 0.3292,
      "step": 9144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6708915322321781,
      "lm_loss": 0.4163,
      "step": 9144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6708915322321781,
      "lm_loss": 0.438,
      "step": 9144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6708915322321781,
      "lm_loss": 0.5906,
      "step": 9144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6708915322321781,
      "lm_loss": 0.5089,
      "step": 9144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6708915322321781,
      "lm_loss": 0.3155,
      "step": 9144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6708915322321781,
      "lm_loss": 0.4756,
      "step": 9144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6708915322321781,
      "lm_loss": 0.5174,
      "step": 9144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6709649018223172,
      "grad_norm": 0.5557880948628791,
      "learning_rate": 2.5799943461601353e-06,
      "loss": 0.4834,
      "step": 9145
    },
    {
      "epoch": 0.6710382714124563,
      "grad_norm": 1.4170910849715987,
      "learning_rate": 2.578954662146615e-06,
      "loss": 0.7194,
      "step": 9146
    },
    {
      "epoch": 0.6711116410025955,
      "grad_norm": 0.5498789278408772,
      "learning_rate": 2.5779151148555214e-06,
      "loss": 0.4792,
      "step": 9147
    },
    {
      "epoch": 0.6711850105927346,
      "grad_norm": 0.5768360102609362,
      "learning_rate": 2.5768757043455596e-06,
      "loss": 0.494,
      "step": 9148
    },
    {
      "epoch": 0.6712583801828736,
      "grad_norm": 1.4512152250897215,
      "learning_rate": 2.5758364306754247e-06,
      "loss": 0.7055,
      "step": 9149
    },
    {
      "epoch": 0.6713317497730128,
      "grad_norm": 1.4953908675473597,
      "learning_rate": 2.5747972939038122e-06,
      "loss": 0.7839,
      "step": 9150
    },
    {
      "epoch": 0.671405119363152,
      "grad_norm": 1.4872501928374429,
      "learning_rate": 2.573758294089403e-06,
      "loss": 0.6465,
      "step": 9151
    },
    {
      "epoch": 0.6714784889532911,
      "grad_norm": 1.5275343696360164,
      "learning_rate": 2.572719431290871e-06,
      "loss": 0.8206,
      "step": 9152
    },
    {
      "epoch": 0.6714784889532911,
      "lm_loss": 0.2054,
      "step": 9152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6714784889532911,
      "lm_loss": 0.4895,
      "step": 9152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6714784889532911,
      "lm_loss": 0.4839,
      "step": 9152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6714784889532911,
      "lm_loss": 0.456,
      "step": 9152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6714784889532911,
      "lm_loss": 0.512,
      "step": 9152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6714784889532911,
      "lm_loss": 0.581,
      "step": 9152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6714784889532911,
      "lm_loss": 0.4913,
      "step": 9152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6714784889532911,
      "lm_loss": 0.3541,
      "step": 9152,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6715518585434302,
      "grad_norm": 0.5511835288825703,
      "learning_rate": 2.5716807055668837e-06,
      "loss": 0.4776,
      "step": 9153
    },
    {
      "epoch": 0.6716252281335693,
      "grad_norm": 1.570720097304356,
      "learning_rate": 2.5706421169760986e-06,
      "loss": 0.6533,
      "step": 9154
    },
    {
      "epoch": 0.6716985977237084,
      "grad_norm": 1.5941546089604777,
      "learning_rate": 2.5696036655771716e-06,
      "loss": 0.721,
      "step": 9155
    },
    {
      "epoch": 0.6717719673138476,
      "grad_norm": 1.4473629265985182,
      "learning_rate": 2.5685653514287444e-06,
      "loss": 0.7914,
      "step": 9156
    },
    {
      "epoch": 0.6718453369039867,
      "grad_norm": 1.442262133880337,
      "learning_rate": 2.5675271745894524e-06,
      "loss": 0.7063,
      "step": 9157
    },
    {
      "epoch": 0.6719187064941259,
      "grad_norm": 1.5688648393147355,
      "learning_rate": 2.5664891351179256e-06,
      "loss": 0.7391,
      "step": 9158
    },
    {
      "epoch": 0.6719920760842649,
      "grad_norm": 1.5224922486217276,
      "learning_rate": 2.5654512330727837e-06,
      "loss": 0.6858,
      "step": 9159
    },
    {
      "epoch": 0.6720654456744041,
      "grad_norm": 1.4377918689705307,
      "learning_rate": 2.5644134685126376e-06,
      "loss": 0.8186,
      "step": 9160
    },
    {
      "epoch": 0.6720654456744041,
      "lm_loss": 0.9458,
      "step": 9160,
      "vm_loss": 0.0653
    },
    {
      "epoch": 0.6720654456744041,
      "lm_loss": 0.0086,
      "step": 9160,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.6720654456744041,
      "lm_loss": 0.4974,
      "step": 9160,
      "vm_loss": 0.1496
    },
    {
      "epoch": 0.6720654456744041,
      "lm_loss": 0.0524,
      "step": 9160,
      "vm_loss": 0.1434
    },
    {
      "epoch": 0.6720654456744041,
      "lm_loss": 0.6308,
      "step": 9160,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.6720654456744041,
      "lm_loss": 1.6579,
      "step": 9160,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.6720654456744041,
      "lm_loss": 0.3911,
      "step": 9160,
      "vm_loss": 0.1671
    },
    {
      "epoch": 0.6720654456744041,
      "lm_loss": 1.0846,
      "step": 9160,
      "vm_loss": 0.1149
    },
    {
      "epoch": 0.6721388152645432,
      "grad_norm": 1.5043296562868533,
      "learning_rate": 2.563375841496097e-06,
      "loss": 0.7147,
      "step": 9161
    },
    {
      "epoch": 0.6722121848546824,
      "grad_norm": 1.3937942795028553,
      "learning_rate": 2.5623383520817563e-06,
      "loss": 0.7131,
      "step": 9162
    },
    {
      "epoch": 0.6722855544448215,
      "grad_norm": 0.5788816488537857,
      "learning_rate": 2.5613010003282055e-06,
      "loss": 0.4971,
      "step": 9163
    },
    {
      "epoch": 0.6723589240349606,
      "grad_norm": 1.493514336021518,
      "learning_rate": 2.560263786294027e-06,
      "loss": 0.6526,
      "step": 9164
    },
    {
      "epoch": 0.6724322936250997,
      "grad_norm": 1.316594022104498,
      "learning_rate": 2.559226710037792e-06,
      "loss": 0.7713,
      "step": 9165
    },
    {
      "epoch": 0.6725056632152389,
      "grad_norm": 1.5926025164821693,
      "learning_rate": 2.5581897716180717e-06,
      "loss": 0.6965,
      "step": 9166
    },
    {
      "epoch": 0.672579032805378,
      "grad_norm": 1.474023826452332,
      "learning_rate": 2.557152971093422e-06,
      "loss": 0.7083,
      "step": 9167
    },
    {
      "epoch": 0.6726524023955172,
      "grad_norm": 0.5465478134029931,
      "learning_rate": 2.5561163085223926e-06,
      "loss": 0.4668,
      "step": 9168
    },
    {
      "epoch": 0.6726524023955172,
      "lm_loss": 1.4687,
      "step": 9168,
      "vm_loss": 0.0996
    },
    {
      "epoch": 0.6726524023955172,
      "lm_loss": 0.0917,
      "step": 9168,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.6726524023955172,
      "lm_loss": 1.0811,
      "step": 9168,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.6726524023955172,
      "lm_loss": 0.4175,
      "step": 9168,
      "vm_loss": 0.272
    },
    {
      "epoch": 0.6726524023955172,
      "lm_loss": 0.5504,
      "step": 9168,
      "vm_loss": 0.1093
    },
    {
      "epoch": 0.6726524023955172,
      "lm_loss": 0.796,
      "step": 9168,
      "vm_loss": 0.0721
    },
    {
      "epoch": 0.6726524023955172,
      "lm_loss": 0.3425,
      "step": 9168,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.6726524023955172,
      "lm_loss": 0.9316,
      "step": 9168,
      "vm_loss": 0.1121
    },
    {
      "epoch": 0.6727257719856562,
      "grad_norm": 1.624450567673824,
      "learning_rate": 2.5550797839635283e-06,
      "loss": 0.8249,
      "step": 9169
    },
    {
      "epoch": 0.6727991415757953,
      "grad_norm": 0.5433048266595922,
      "learning_rate": 2.554043397475363e-06,
      "loss": 0.4822,
      "step": 9170
    },
    {
      "epoch": 0.6728725111659345,
      "grad_norm": 0.5328311842697052,
      "learning_rate": 2.5530071491164225e-06,
      "loss": 0.4836,
      "step": 9171
    },
    {
      "epoch": 0.6729458807560736,
      "grad_norm": 0.5586015477018899,
      "learning_rate": 2.55197103894523e-06,
      "loss": 0.5038,
      "step": 9172
    },
    {
      "epoch": 0.6730192503462128,
      "grad_norm": 1.3439681387767246,
      "learning_rate": 2.550935067020296e-06,
      "loss": 0.7302,
      "step": 9173
    },
    {
      "epoch": 0.6730926199363518,
      "grad_norm": 1.4313238618207544,
      "learning_rate": 2.549899233400123e-06,
      "loss": 0.7175,
      "step": 9174
    },
    {
      "epoch": 0.673165989526491,
      "grad_norm": 0.5469733776599354,
      "learning_rate": 2.5488635381432086e-06,
      "loss": 0.4671,
      "step": 9175
    },
    {
      "epoch": 0.6732393591166301,
      "grad_norm": 1.5162114959466095,
      "learning_rate": 2.547827981308038e-06,
      "loss": 0.7937,
      "step": 9176
    },
    {
      "epoch": 0.6732393591166301,
      "lm_loss": 1.7898,
      "step": 9176,
      "vm_loss": 0.0953
    },
    {
      "epoch": 0.6732393591166301,
      "lm_loss": 0.7105,
      "step": 9176,
      "vm_loss": 0.1251
    },
    {
      "epoch": 0.6732393591166301,
      "lm_loss": 1.1851,
      "step": 9176,
      "vm_loss": 0.2236
    },
    {
      "epoch": 0.6732393591166301,
      "lm_loss": 0.84,
      "step": 9176,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.6732393591166301,
      "lm_loss": 0.8901,
      "step": 9176,
      "vm_loss": 0.1038
    },
    {
      "epoch": 0.6732393591166301,
      "lm_loss": 0.5076,
      "step": 9176,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.6732393591166301,
      "lm_loss": 0.2038,
      "step": 9176,
      "vm_loss": 0.0781
    },
    {
      "epoch": 0.6732393591166301,
      "lm_loss": 0.5876,
      "step": 9176,
      "vm_loss": 0.2352
    },
    {
      "epoch": 0.6733127287067693,
      "grad_norm": 1.5970326997460311,
      "learning_rate": 2.546792562953096e-06,
      "loss": 0.7237,
      "step": 9177
    },
    {
      "epoch": 0.6733860982969084,
      "grad_norm": 1.363375174520849,
      "learning_rate": 2.5457572831368526e-06,
      "loss": 0.7698,
      "step": 9178
    },
    {
      "epoch": 0.6734594678870475,
      "grad_norm": 1.427625364317131,
      "learning_rate": 2.5447221419177743e-06,
      "loss": 0.7467,
      "step": 9179
    },
    {
      "epoch": 0.6735328374771866,
      "grad_norm": 1.5955115972726455,
      "learning_rate": 2.5436871393543168e-06,
      "loss": 0.8048,
      "step": 9180
    },
    {
      "epoch": 0.6736062070673258,
      "grad_norm": 1.5859089261608095,
      "learning_rate": 2.542652275504927e-06,
      "loss": 0.7348,
      "step": 9181
    },
    {
      "epoch": 0.6736795766574649,
      "grad_norm": 1.5576565127553934,
      "learning_rate": 2.541617550428051e-06,
      "loss": 0.7595,
      "step": 9182
    },
    {
      "epoch": 0.6737529462476041,
      "grad_norm": 1.4739587646708696,
      "learning_rate": 2.54058296418212e-06,
      "loss": 0.7651,
      "step": 9183
    },
    {
      "epoch": 0.6738263158377431,
      "grad_norm": 1.49057395823749,
      "learning_rate": 2.5395485168255596e-06,
      "loss": 0.831,
      "step": 9184
    },
    {
      "epoch": 0.6738263158377431,
      "lm_loss": 1.1949,
      "step": 9184,
      "vm_loss": 0.069
    },
    {
      "epoch": 0.6738263158377431,
      "lm_loss": 0.1195,
      "step": 9184,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.6738263158377431,
      "lm_loss": 0.9278,
      "step": 9184,
      "vm_loss": 0.1863
    },
    {
      "epoch": 0.6738263158377431,
      "lm_loss": 0.0595,
      "step": 9184,
      "vm_loss": 0.1381
    },
    {
      "epoch": 0.6738263158377431,
      "lm_loss": 0.2225,
      "step": 9184,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.6738263158377431,
      "lm_loss": 0.4867,
      "step": 9184,
      "vm_loss": 0.1002
    },
    {
      "epoch": 0.6738263158377431,
      "lm_loss": 0.6716,
      "step": 9184,
      "vm_loss": 0.0949
    },
    {
      "epoch": 0.6738263158377431,
      "lm_loss": 0.6137,
      "step": 9184,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.6738996854278823,
      "grad_norm": 1.5734922697904812,
      "learning_rate": 2.538514208416788e-06,
      "loss": 0.7601,
      "step": 9185
    },
    {
      "epoch": 0.6739730550180214,
      "grad_norm": 1.3661043776016124,
      "learning_rate": 2.537480039014214e-06,
      "loss": 0.7646,
      "step": 9186
    },
    {
      "epoch": 0.6740464246081606,
      "grad_norm": 1.330664432829632,
      "learning_rate": 2.5364460086762393e-06,
      "loss": 0.7586,
      "step": 9187
    },
    {
      "epoch": 0.6741197941982997,
      "grad_norm": 1.708162546883758,
      "learning_rate": 2.5354121174612612e-06,
      "loss": 0.8571,
      "step": 9188
    },
    {
      "epoch": 0.6741931637884387,
      "grad_norm": 1.4068223851387882,
      "learning_rate": 2.5343783654276644e-06,
      "loss": 0.7648,
      "step": 9189
    },
    {
      "epoch": 0.6742665333785779,
      "grad_norm": 0.5764369038670779,
      "learning_rate": 2.5333447526338272e-06,
      "loss": 0.4637,
      "step": 9190
    },
    {
      "epoch": 0.674339902968717,
      "grad_norm": 1.4000123065961643,
      "learning_rate": 2.53231127913812e-06,
      "loss": 0.673,
      "step": 9191
    },
    {
      "epoch": 0.6744132725588562,
      "grad_norm": 1.3909466629126253,
      "learning_rate": 2.5312779449989045e-06,
      "loss": 0.743,
      "step": 9192
    },
    {
      "epoch": 0.6744132725588562,
      "lm_loss": 1.0232,
      "step": 9192,
      "vm_loss": 0.1036
    },
    {
      "epoch": 0.6744132725588562,
      "lm_loss": 0.603,
      "step": 9192,
      "vm_loss": 0.175
    },
    {
      "epoch": 0.6744132725588562,
      "lm_loss": 0.0443,
      "step": 9192,
      "vm_loss": 0.2005
    },
    {
      "epoch": 0.6744132725588562,
      "lm_loss": 0.9439,
      "step": 9192,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.6744132725588562,
      "lm_loss": 0.4246,
      "step": 9192,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.6744132725588562,
      "lm_loss": 0.7567,
      "step": 9192,
      "vm_loss": 0.1959
    },
    {
      "epoch": 0.6744132725588562,
      "lm_loss": 0.4926,
      "step": 9192,
      "vm_loss": 0.1543
    },
    {
      "epoch": 0.6744132725588562,
      "lm_loss": 0.2816,
      "step": 9192,
      "vm_loss": 0.1604
    },
    {
      "epoch": 0.6744866421489953,
      "grad_norm": 1.2965030491457097,
      "learning_rate": 2.5302447502745385e-06,
      "loss": 0.7017,
      "step": 9193
    },
    {
      "epoch": 0.6745600117391344,
      "grad_norm": 1.5077340998519668,
      "learning_rate": 2.5292116950233684e-06,
      "loss": 0.7079,
      "step": 9194
    },
    {
      "epoch": 0.6746333813292735,
      "grad_norm": 0.5538956130383365,
      "learning_rate": 2.528178779303733e-06,
      "loss": 0.4696,
      "step": 9195
    },
    {
      "epoch": 0.6747067509194127,
      "grad_norm": 1.6906565148520873,
      "learning_rate": 2.5271460031739624e-06,
      "loss": 0.8158,
      "step": 9196
    },
    {
      "epoch": 0.6747801205095518,
      "grad_norm": 0.5557203238602005,
      "learning_rate": 2.526113366692381e-06,
      "loss": 0.4884,
      "step": 9197
    },
    {
      "epoch": 0.674853490099691,
      "grad_norm": 1.4930751707769465,
      "learning_rate": 2.5250808699173046e-06,
      "loss": 0.7952,
      "step": 9198
    },
    {
      "epoch": 0.67492685968983,
      "grad_norm": 1.4289709554699426,
      "learning_rate": 2.5240485129070403e-06,
      "loss": 0.7211,
      "step": 9199
    },
    {
      "epoch": 0.6750002292799692,
      "grad_norm": 1.6330497818260934,
      "learning_rate": 2.523016295719888e-06,
      "loss": 0.7646,
      "step": 9200
    },
    {
      "epoch": 0.6750002292799692,
      "lm_loss": 0.6858,
      "step": 9200,
      "vm_loss": 0.0314
    },
    {
      "epoch": 0.6750002292799692,
      "lm_loss": 1.471,
      "step": 9200,
      "vm_loss": 0.0719
    },
    {
      "epoch": 0.6750002292799692,
      "lm_loss": 0.5719,
      "step": 9200,
      "vm_loss": 0.1525
    },
    {
      "epoch": 0.6750002292799692,
      "lm_loss": 0.2306,
      "step": 9200,
      "vm_loss": 0.1796
    },
    {
      "epoch": 0.6750002292799692,
      "lm_loss": 0.5376,
      "step": 9200,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.6750002292799692,
      "lm_loss": 0.6739,
      "step": 9200,
      "vm_loss": 0.1965
    },
    {
      "epoch": 0.6750002292799692,
      "lm_loss": 0.5854,
      "step": 9200,
      "vm_loss": 0.172
    },
    {
      "epoch": 0.6750002292799692,
      "lm_loss": 0.8674,
      "step": 9200,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.6750735988701083,
      "grad_norm": 1.473579755865854,
      "learning_rate": 2.5219842184141397e-06,
      "loss": 0.761,
      "step": 9201
    },
    {
      "epoch": 0.6751469684602475,
      "grad_norm": 1.62840159731373,
      "learning_rate": 2.520952281048078e-06,
      "loss": 0.879,
      "step": 9202
    },
    {
      "epoch": 0.6752203380503866,
      "grad_norm": 1.4984263059607161,
      "learning_rate": 2.519920483679981e-06,
      "loss": 0.8122,
      "step": 9203
    },
    {
      "epoch": 0.6752937076405257,
      "grad_norm": 0.5479175922607493,
      "learning_rate": 2.518888826368117e-06,
      "loss": 0.4787,
      "step": 9204
    },
    {
      "epoch": 0.6753670772306648,
      "grad_norm": 0.5406956333343944,
      "learning_rate": 2.5178573091707446e-06,
      "loss": 0.479,
      "step": 9205
    },
    {
      "epoch": 0.675440446820804,
      "grad_norm": 0.547461493148154,
      "learning_rate": 2.5168259321461177e-06,
      "loss": 0.5024,
      "step": 9206
    },
    {
      "epoch": 0.6755138164109431,
      "grad_norm": 0.5496897986990887,
      "learning_rate": 2.515794695352478e-06,
      "loss": 0.4945,
      "step": 9207
    },
    {
      "epoch": 0.6755871860010823,
      "grad_norm": 1.5259148233856907,
      "learning_rate": 2.514763598848065e-06,
      "loss": 0.7793,
      "step": 9208
    },
    {
      "epoch": 0.6755871860010823,
      "lm_loss": 2.0385,
      "step": 9208,
      "vm_loss": 0.0516
    },
    {
      "epoch": 0.6755871860010823,
      "lm_loss": 1.5978,
      "step": 9208,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.6755871860010823,
      "lm_loss": 0.0469,
      "step": 9208,
      "vm_loss": 0.1535
    },
    {
      "epoch": 0.6755871860010823,
      "lm_loss": 0.2272,
      "step": 9208,
      "vm_loss": 0.1844
    },
    {
      "epoch": 0.6755871860010823,
      "lm_loss": 0.156,
      "step": 9208,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.6755871860010823,
      "lm_loss": 0.2107,
      "step": 9208,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.6755871860010823,
      "lm_loss": 0.0568,
      "step": 9208,
      "vm_loss": 0.1342
    },
    {
      "epoch": 0.6755871860010823,
      "lm_loss": 0.1226,
      "step": 9208,
      "vm_loss": 0.1021
    },
    {
      "epoch": 0.6756605555912213,
      "grad_norm": 1.3741855094669955,
      "learning_rate": 2.5137326426911067e-06,
      "loss": 0.7624,
      "step": 9209
    },
    {
      "epoch": 0.6757339251813604,
      "grad_norm": 1.6323746794505387,
      "learning_rate": 2.512701826939823e-06,
      "loss": 0.7709,
      "step": 9210
    },
    {
      "epoch": 0.6758072947714996,
      "grad_norm": 1.4049779820218287,
      "learning_rate": 2.5116711516524272e-06,
      "loss": 0.841,
      "step": 9211
    },
    {
      "epoch": 0.6758806643616387,
      "grad_norm": 1.7786459650086843,
      "learning_rate": 2.5106406168871235e-06,
      "loss": 0.7057,
      "step": 9212
    },
    {
      "epoch": 0.6759540339517779,
      "grad_norm": 1.4231638649525247,
      "learning_rate": 2.509610222702107e-06,
      "loss": 0.6644,
      "step": 9213
    },
    {
      "epoch": 0.6760274035419169,
      "grad_norm": 1.5066731356262806,
      "learning_rate": 2.50857996915557e-06,
      "loss": 0.6883,
      "step": 9214
    },
    {
      "epoch": 0.6761007731320561,
      "grad_norm": 0.5895710289713629,
      "learning_rate": 2.5075498563056926e-06,
      "loss": 0.4838,
      "step": 9215
    },
    {
      "epoch": 0.6761741427221952,
      "grad_norm": 1.5460070497194096,
      "learning_rate": 2.5065198842106475e-06,
      "loss": 0.8009,
      "step": 9216
    },
    {
      "epoch": 0.6761741427221952,
      "lm_loss": 0.9769,
      "step": 9216,
      "vm_loss": 0.034
    },
    {
      "epoch": 0.6761741427221952,
      "lm_loss": 0.9651,
      "step": 9216,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.6761741427221952,
      "lm_loss": 0.5557,
      "step": 9216,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.6761741427221952,
      "lm_loss": 0.7286,
      "step": 9216,
      "vm_loss": 0.1283
    },
    {
      "epoch": 0.6761741427221952,
      "lm_loss": 1.0752,
      "step": 9216,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.6761741427221952,
      "lm_loss": 0.6729,
      "step": 9216,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.6761741427221952,
      "lm_loss": 0.4729,
      "step": 9216,
      "vm_loss": 0.1004
    },
    {
      "epoch": 0.6761741427221952,
      "lm_loss": 0.1083,
      "step": 9216,
      "vm_loss": 0.0733
    },
    {
      "epoch": 0.6762475123123344,
      "grad_norm": 1.527243273926884,
      "learning_rate": 2.5054900529285987e-06,
      "loss": 0.7144,
      "step": 9217
    },
    {
      "epoch": 0.6763208819024735,
      "grad_norm": 1.5557047786149931,
      "learning_rate": 2.5044603625177026e-06,
      "loss": 0.7426,
      "step": 9218
    },
    {
      "epoch": 0.6763942514926126,
      "grad_norm": 1.5490660459133014,
      "learning_rate": 2.503430813036112e-06,
      "loss": 0.7693,
      "step": 9219
    },
    {
      "epoch": 0.6764676210827517,
      "grad_norm": 1.592010412585841,
      "learning_rate": 2.5024014045419664e-06,
      "loss": 0.7232,
      "step": 9220
    },
    {
      "epoch": 0.6765409906728909,
      "grad_norm": 1.3586492822354657,
      "learning_rate": 2.501372137093398e-06,
      "loss": 0.7507,
      "step": 9221
    },
    {
      "epoch": 0.67661436026303,
      "grad_norm": 0.5691486672724742,
      "learning_rate": 2.500343010748533e-06,
      "loss": 0.4839,
      "step": 9222
    },
    {
      "epoch": 0.6766877298531692,
      "grad_norm": 1.4414177763247122,
      "learning_rate": 2.4993140255654867e-06,
      "loss": 0.7848,
      "step": 9223
    },
    {
      "epoch": 0.6767610994433082,
      "grad_norm": 1.5121145687836455,
      "learning_rate": 2.4982851816023724e-06,
      "loss": 0.7771,
      "step": 9224
    },
    {
      "epoch": 0.6767610994433082,
      "lm_loss": 0.1337,
      "step": 9224,
      "vm_loss": 0.0382
    },
    {
      "epoch": 0.6767610994433082,
      "lm_loss": 1.0705,
      "step": 9224,
      "vm_loss": 0.1842
    },
    {
      "epoch": 0.6767610994433082,
      "lm_loss": 0.4756,
      "step": 9224,
      "vm_loss": 0.0967
    },
    {
      "epoch": 0.6767610994433082,
      "lm_loss": 0.4205,
      "step": 9224,
      "vm_loss": 0.057
    },
    {
      "epoch": 0.6767610994433082,
      "lm_loss": 0.3305,
      "step": 9224,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.6767610994433082,
      "lm_loss": 0.9331,
      "step": 9224,
      "vm_loss": 0.186
    },
    {
      "epoch": 0.6767610994433082,
      "lm_loss": 1.0397,
      "step": 9224,
      "vm_loss": 0.2162
    },
    {
      "epoch": 0.6767610994433082,
      "lm_loss": 0.1087,
      "step": 9224,
      "vm_loss": 0.0743
    },
    {
      "epoch": 0.6768344690334474,
      "grad_norm": 1.5035909394417581,
      "learning_rate": 2.497256478917289e-06,
      "loss": 0.7249,
      "step": 9225
    },
    {
      "epoch": 0.6769078386235865,
      "grad_norm": 2.559632076303319,
      "learning_rate": 2.4962279175683297e-06,
      "loss": 0.7657,
      "step": 9226
    },
    {
      "epoch": 0.6769812082137256,
      "grad_norm": 1.6599929974284162,
      "learning_rate": 2.4951994976135805e-06,
      "loss": 0.757,
      "step": 9227
    },
    {
      "epoch": 0.6770545778038647,
      "grad_norm": 0.545824397758957,
      "learning_rate": 2.4941712191111194e-06,
      "loss": 0.4853,
      "step": 9228
    },
    {
      "epoch": 0.6771279473940038,
      "grad_norm": 0.539563326409682,
      "learning_rate": 2.493143082119013e-06,
      "loss": 0.502,
      "step": 9229
    },
    {
      "epoch": 0.677201316984143,
      "grad_norm": 1.584365321006314,
      "learning_rate": 2.4921150866953265e-06,
      "loss": 0.7558,
      "step": 9230
    },
    {
      "epoch": 0.6772746865742821,
      "grad_norm": 1.4858310189617243,
      "learning_rate": 2.4910872328981116e-06,
      "loss": 0.6893,
      "step": 9231
    },
    {
      "epoch": 0.6773480561644213,
      "grad_norm": 1.558340808667666,
      "learning_rate": 2.4900595207854145e-06,
      "loss": 0.7511,
      "step": 9232
    },
    {
      "epoch": 0.6773480561644213,
      "lm_loss": 0.5853,
      "step": 9232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6773480561644213,
      "lm_loss": 0.4186,
      "step": 9232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6773480561644213,
      "lm_loss": 0.5738,
      "step": 9232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6773480561644213,
      "lm_loss": 0.4031,
      "step": 9232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6773480561644213,
      "lm_loss": 0.5686,
      "step": 9232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6773480561644213,
      "lm_loss": 0.4049,
      "step": 9232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6773480561644213,
      "lm_loss": 0.4496,
      "step": 9232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6773480561644213,
      "lm_loss": 0.5888,
      "step": 9232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6774214257545603,
      "grad_norm": 0.5480905079442877,
      "learning_rate": 2.4890319504152717e-06,
      "loss": 0.4955,
      "step": 9233
    },
    {
      "epoch": 0.6774947953446995,
      "grad_norm": 1.4105356714364583,
      "learning_rate": 2.4880045218457107e-06,
      "loss": 0.7065,
      "step": 9234
    },
    {
      "epoch": 0.6775681649348386,
      "grad_norm": 1.3516401116255854,
      "learning_rate": 2.4869772351347576e-06,
      "loss": 0.5961,
      "step": 9235
    },
    {
      "epoch": 0.6776415345249778,
      "grad_norm": 1.519920511469836,
      "learning_rate": 2.4859500903404234e-06,
      "loss": 0.7909,
      "step": 9236
    },
    {
      "epoch": 0.6777149041151169,
      "grad_norm": 1.4812899192704017,
      "learning_rate": 2.484923087520713e-06,
      "loss": 0.7349,
      "step": 9237
    },
    {
      "epoch": 0.677788273705256,
      "grad_norm": 1.4002780878089878,
      "learning_rate": 2.4838962267336247e-06,
      "loss": 0.7549,
      "step": 9238
    },
    {
      "epoch": 0.6778616432953951,
      "grad_norm": 1.4620838441367494,
      "learning_rate": 2.4828695080371474e-06,
      "loss": 0.7711,
      "step": 9239
    },
    {
      "epoch": 0.6779350128855343,
      "grad_norm": 1.4594690955718583,
      "learning_rate": 2.481842931489261e-06,
      "loss": 0.7319,
      "step": 9240
    },
    {
      "epoch": 0.6779350128855343,
      "lm_loss": 0.3223,
      "step": 9240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6779350128855343,
      "lm_loss": 0.4722,
      "step": 9240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6779350128855343,
      "lm_loss": 0.3625,
      "step": 9240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6779350128855343,
      "lm_loss": 0.3743,
      "step": 9240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6779350128855343,
      "lm_loss": 0.479,
      "step": 9240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6779350128855343,
      "lm_loss": 0.5483,
      "step": 9240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6779350128855343,
      "lm_loss": 0.5525,
      "step": 9240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6779350128855343,
      "lm_loss": 0.4952,
      "step": 9240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6780083824756734,
      "grad_norm": 0.5444986592680182,
      "learning_rate": 2.480816497147942e-06,
      "loss": 0.4736,
      "step": 9241
    },
    {
      "epoch": 0.6780817520658126,
      "grad_norm": 0.5521701182207843,
      "learning_rate": 2.479790205071154e-06,
      "loss": 0.475,
      "step": 9242
    },
    {
      "epoch": 0.6781551216559516,
      "grad_norm": 1.5584180969595296,
      "learning_rate": 2.4787640553168547e-06,
      "loss": 0.7895,
      "step": 9243
    },
    {
      "epoch": 0.6782284912460907,
      "grad_norm": 1.466775009065041,
      "learning_rate": 2.4777380479429925e-06,
      "loss": 0.8067,
      "step": 9244
    },
    {
      "epoch": 0.6783018608362299,
      "grad_norm": 1.3788489871103888,
      "learning_rate": 2.476712183007507e-06,
      "loss": 0.6886,
      "step": 9245
    },
    {
      "epoch": 0.678375230426369,
      "grad_norm": 0.5434080095391174,
      "learning_rate": 2.4756864605683357e-06,
      "loss": 0.4709,
      "step": 9246
    },
    {
      "epoch": 0.6784486000165082,
      "grad_norm": 0.5722085453191779,
      "learning_rate": 2.474660880683401e-06,
      "loss": 0.4853,
      "step": 9247
    },
    {
      "epoch": 0.6785219696066472,
      "grad_norm": 0.5711256707440199,
      "learning_rate": 2.4736354434106202e-06,
      "loss": 0.4724,
      "step": 9248
    },
    {
      "epoch": 0.6785219696066472,
      "lm_loss": 0.7258,
      "step": 9248,
      "vm_loss": 0.0284
    },
    {
      "epoch": 0.6785219696066472,
      "lm_loss": 1.1104,
      "step": 9248,
      "vm_loss": 0.0984
    },
    {
      "epoch": 0.6785219696066472,
      "lm_loss": 0.349,
      "step": 9248,
      "vm_loss": 0.1534
    },
    {
      "epoch": 0.6785219696066472,
      "lm_loss": 0.996,
      "step": 9248,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.6785219696066472,
      "lm_loss": 0.2962,
      "step": 9248,
      "vm_loss": 0.1387
    },
    {
      "epoch": 0.6785219696066472,
      "lm_loss": 0.9242,
      "step": 9248,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.6785219696066472,
      "lm_loss": 0.8552,
      "step": 9248,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.6785219696066472,
      "lm_loss": 0.4871,
      "step": 9248,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.6785953391967864,
      "grad_norm": 1.4611525995130272,
      "learning_rate": 2.472610148807903e-06,
      "loss": 0.6992,
      "step": 9249
    },
    {
      "epoch": 0.6786687087869255,
      "grad_norm": 1.5277754931977094,
      "learning_rate": 2.471584996933147e-06,
      "loss": 0.6906,
      "step": 9250
    },
    {
      "epoch": 0.6787420783770647,
      "grad_norm": 1.448714561425361,
      "learning_rate": 2.47055998784425e-06,
      "loss": 0.7627,
      "step": 9251
    },
    {
      "epoch": 0.6788154479672038,
      "grad_norm": 0.5498175272587813,
      "learning_rate": 2.469535121599094e-06,
      "loss": 0.4999,
      "step": 9252
    },
    {
      "epoch": 0.6788888175573429,
      "grad_norm": 1.552859372364063,
      "learning_rate": 2.468510398255557e-06,
      "loss": 0.7457,
      "step": 9253
    },
    {
      "epoch": 0.678962187147482,
      "grad_norm": 0.5509918619217565,
      "learning_rate": 2.4674858178715065e-06,
      "loss": 0.487,
      "step": 9254
    },
    {
      "epoch": 0.6790355567376212,
      "grad_norm": 1.3901039762282528,
      "learning_rate": 2.4664613805048042e-06,
      "loss": 0.7046,
      "step": 9255
    },
    {
      "epoch": 0.6791089263277603,
      "grad_norm": 1.3278515727257738,
      "learning_rate": 2.465437086213299e-06,
      "loss": 0.6931,
      "step": 9256
    },
    {
      "epoch": 0.6791089263277603,
      "lm_loss": 0.0241,
      "step": 9256,
      "vm_loss": 0.0516
    },
    {
      "epoch": 0.6791089263277603,
      "lm_loss": 0.9654,
      "step": 9256,
      "vm_loss": 0.1536
    },
    {
      "epoch": 0.6791089263277603,
      "lm_loss": 0.3384,
      "step": 9256,
      "vm_loss": 0.1776
    },
    {
      "epoch": 0.6791089263277603,
      "lm_loss": 1.2126,
      "step": 9256,
      "vm_loss": 0.1038
    },
    {
      "epoch": 0.6791089263277603,
      "lm_loss": 0.7931,
      "step": 9256,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.6791089263277603,
      "lm_loss": 0.943,
      "step": 9256,
      "vm_loss": 0.1671
    },
    {
      "epoch": 0.6791089263277603,
      "lm_loss": 0.1102,
      "step": 9256,
      "vm_loss": 0.0976
    },
    {
      "epoch": 0.6791089263277603,
      "lm_loss": 0.9924,
      "step": 9256,
      "vm_loss": 0.1238
    },
    {
      "epoch": 0.6791822959178995,
      "grad_norm": 1.5859261825445075,
      "learning_rate": 2.4644129350548412e-06,
      "loss": 0.7789,
      "step": 9257
    },
    {
      "epoch": 0.6792556655080385,
      "grad_norm": 1.5257636024293932,
      "learning_rate": 2.4633889270872636e-06,
      "loss": 0.7211,
      "step": 9258
    },
    {
      "epoch": 0.6793290350981777,
      "grad_norm": 0.5569062276810798,
      "learning_rate": 2.4623650623683954e-06,
      "loss": 0.4938,
      "step": 9259
    },
    {
      "epoch": 0.6794024046883168,
      "grad_norm": 1.5705969212232074,
      "learning_rate": 2.4613413409560568e-06,
      "loss": 0.7645,
      "step": 9260
    },
    {
      "epoch": 0.679475774278456,
      "grad_norm": 1.3910259044381434,
      "learning_rate": 2.4603177629080564e-06,
      "loss": 0.7369,
      "step": 9261
    },
    {
      "epoch": 0.6795491438685951,
      "grad_norm": 1.4101916193345962,
      "learning_rate": 2.4592943282822048e-06,
      "loss": 0.7963,
      "step": 9262
    },
    {
      "epoch": 0.6796225134587341,
      "grad_norm": 1.663852952807792,
      "learning_rate": 2.4582710371362932e-06,
      "loss": 0.6994,
      "step": 9263
    },
    {
      "epoch": 0.6796958830488733,
      "grad_norm": 1.480516138431028,
      "learning_rate": 2.457247889528111e-06,
      "loss": 0.7153,
      "step": 9264
    },
    {
      "epoch": 0.6796958830488733,
      "lm_loss": 0.5266,
      "step": 9264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6796958830488733,
      "lm_loss": 0.3149,
      "step": 9264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6796958830488733,
      "lm_loss": 0.4027,
      "step": 9264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6796958830488733,
      "lm_loss": 0.3885,
      "step": 9264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6796958830488733,
      "lm_loss": 0.4166,
      "step": 9264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6796958830488733,
      "lm_loss": 0.5387,
      "step": 9264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6796958830488733,
      "lm_loss": 0.4023,
      "step": 9264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6796958830488733,
      "lm_loss": 0.521,
      "step": 9264,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6797692526390124,
      "grad_norm": 0.5425862559784201,
      "learning_rate": 2.456224885515438e-06,
      "loss": 0.4869,
      "step": 9265
    },
    {
      "epoch": 0.6798426222291516,
      "grad_norm": 2.2572854259790383,
      "learning_rate": 2.455202025156045e-06,
      "loss": 0.9024,
      "step": 9266
    },
    {
      "epoch": 0.6799159918192907,
      "grad_norm": 0.5677709264607561,
      "learning_rate": 2.454179308507693e-06,
      "loss": 0.487,
      "step": 9267
    },
    {
      "epoch": 0.6799893614094298,
      "grad_norm": 1.5603786218738915,
      "learning_rate": 2.4531567356281435e-06,
      "loss": 0.7305,
      "step": 9268
    },
    {
      "epoch": 0.6800627309995689,
      "grad_norm": 1.4495771624749125,
      "learning_rate": 2.452134306575139e-06,
      "loss": 0.6771,
      "step": 9269
    },
    {
      "epoch": 0.6801361005897081,
      "grad_norm": 0.5481233714063353,
      "learning_rate": 2.4511120214064204e-06,
      "loss": 0.4689,
      "step": 9270
    },
    {
      "epoch": 0.6802094701798472,
      "grad_norm": 0.5497528155816468,
      "learning_rate": 2.450089880179718e-06,
      "loss": 0.4685,
      "step": 9271
    },
    {
      "epoch": 0.6802828397699864,
      "grad_norm": 0.5603663318821894,
      "learning_rate": 2.449067882952753e-06,
      "loss": 0.481,
      "step": 9272
    },
    {
      "epoch": 0.6802828397699864,
      "lm_loss": 1.3932,
      "step": 9272,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.6802828397699864,
      "lm_loss": 0.8485,
      "step": 9272,
      "vm_loss": 0.0611
    },
    {
      "epoch": 0.6802828397699864,
      "lm_loss": 0.1967,
      "step": 9272,
      "vm_loss": 0.1928
    },
    {
      "epoch": 0.6802828397699864,
      "lm_loss": 0.4524,
      "step": 9272,
      "vm_loss": 0.0503
    },
    {
      "epoch": 0.6802828397699864,
      "lm_loss": 0.5301,
      "step": 9272,
      "vm_loss": 0.0755
    },
    {
      "epoch": 0.6802828397699864,
      "lm_loss": 1.0787,
      "step": 9272,
      "vm_loss": 0.0692
    },
    {
      "epoch": 0.6802828397699864,
      "lm_loss": 0.7773,
      "step": 9272,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.6802828397699864,
      "lm_loss": 0.9357,
      "step": 9272,
      "vm_loss": 0.1513
    },
    {
      "epoch": 0.6803562093601254,
      "grad_norm": 1.3919078040976611,
      "learning_rate": 2.448046029783244e-06,
      "loss": 0.6676,
      "step": 9273
    },
    {
      "epoch": 0.6804295789502646,
      "grad_norm": 1.4707540248109852,
      "learning_rate": 2.4470243207288953e-06,
      "loss": 0.7115,
      "step": 9274
    },
    {
      "epoch": 0.6805029485404037,
      "grad_norm": 1.4882857950599127,
      "learning_rate": 2.446002755847405e-06,
      "loss": 0.7223,
      "step": 9275
    },
    {
      "epoch": 0.6805763181305429,
      "grad_norm": 1.5408237236486848,
      "learning_rate": 2.444981335196465e-06,
      "loss": 0.7556,
      "step": 9276
    },
    {
      "epoch": 0.680649687720682,
      "grad_norm": 0.5476403143038323,
      "learning_rate": 2.4439600588337544e-06,
      "loss": 0.4688,
      "step": 9277
    },
    {
      "epoch": 0.6807230573108211,
      "grad_norm": 1.5740146120940657,
      "learning_rate": 2.4429389268169505e-06,
      "loss": 0.7518,
      "step": 9278
    },
    {
      "epoch": 0.6807964269009602,
      "grad_norm": 1.7341443737676627,
      "learning_rate": 2.441917939203718e-06,
      "loss": 0.7391,
      "step": 9279
    },
    {
      "epoch": 0.6808697964910994,
      "grad_norm": 1.4842837594876606,
      "learning_rate": 2.4408970960517146e-06,
      "loss": 0.7566,
      "step": 9280
    },
    {
      "epoch": 0.6808697964910994,
      "lm_loss": 1.3444,
      "step": 9280,
      "vm_loss": 0.107
    },
    {
      "epoch": 0.6808697964910994,
      "lm_loss": 0.5293,
      "step": 9280,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.6808697964910994,
      "lm_loss": 1.0665,
      "step": 9280,
      "vm_loss": 0.0894
    },
    {
      "epoch": 0.6808697964910994,
      "lm_loss": 0.0365,
      "step": 9280,
      "vm_loss": 0.0851
    },
    {
      "epoch": 0.6808697964910994,
      "lm_loss": 0.369,
      "step": 9280,
      "vm_loss": 0.063
    },
    {
      "epoch": 0.6808697964910994,
      "lm_loss": 0.899,
      "step": 9280,
      "vm_loss": 0.1165
    },
    {
      "epoch": 0.6808697964910994,
      "lm_loss": 0.5961,
      "step": 9280,
      "vm_loss": 0.1719
    },
    {
      "epoch": 0.6808697964910994,
      "lm_loss": 0.047,
      "step": 9280,
      "vm_loss": 0.0343
    },
    {
      "epoch": 0.6809431660812385,
      "grad_norm": 1.5458439976341545,
      "learning_rate": 2.4398763974185897e-06,
      "loss": 0.7869,
      "step": 9281
    },
    {
      "epoch": 0.6810165356713777,
      "grad_norm": 1.53104973245996,
      "learning_rate": 2.4388558433619847e-06,
      "loss": 0.741,
      "step": 9282
    },
    {
      "epoch": 0.6810899052615167,
      "grad_norm": 1.5516780259026148,
      "learning_rate": 2.437835433939531e-06,
      "loss": 0.8413,
      "step": 9283
    },
    {
      "epoch": 0.6811632748516558,
      "grad_norm": 0.5700554969728594,
      "learning_rate": 2.436815169208857e-06,
      "loss": 0.5059,
      "step": 9284
    },
    {
      "epoch": 0.681236644441795,
      "grad_norm": 1.4242021972906398,
      "learning_rate": 2.4357950492275785e-06,
      "loss": 0.6731,
      "step": 9285
    },
    {
      "epoch": 0.6813100140319341,
      "grad_norm": 1.511580508208926,
      "learning_rate": 2.4347750740533033e-06,
      "loss": 0.7976,
      "step": 9286
    },
    {
      "epoch": 0.6813833836220733,
      "grad_norm": 1.4323111535444848,
      "learning_rate": 2.4337552437436325e-06,
      "loss": 0.757,
      "step": 9287
    },
    {
      "epoch": 0.6814567532122123,
      "grad_norm": 1.354173202774537,
      "learning_rate": 2.4327355583561562e-06,
      "loss": 0.7023,
      "step": 9288
    },
    {
      "epoch": 0.6814567532122123,
      "lm_loss": 1.213,
      "step": 9288,
      "vm_loss": 0.0376
    },
    {
      "epoch": 0.6814567532122123,
      "lm_loss": 0.8051,
      "step": 9288,
      "vm_loss": 0.1156
    },
    {
      "epoch": 0.6814567532122123,
      "lm_loss": 1.1092,
      "step": 9288,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.6814567532122123,
      "lm_loss": 0.5297,
      "step": 9288,
      "vm_loss": 0.1898
    },
    {
      "epoch": 0.6814567532122123,
      "lm_loss": 0.8557,
      "step": 9288,
      "vm_loss": 0.1475
    },
    {
      "epoch": 0.6814567532122123,
      "lm_loss": 0.9051,
      "step": 9288,
      "vm_loss": 0.1818
    },
    {
      "epoch": 0.6814567532122123,
      "lm_loss": 0.493,
      "step": 9288,
      "vm_loss": 0.1384
    },
    {
      "epoch": 0.6814567532122123,
      "lm_loss": 0.4305,
      "step": 9288,
      "vm_loss": 0.1543
    },
    {
      "epoch": 0.6815301228023515,
      "grad_norm": 1.7520181155087382,
      "learning_rate": 2.431716017948462e-06,
      "loss": 0.7731,
      "step": 9289
    },
    {
      "epoch": 0.6816034923924906,
      "grad_norm": 1.607905224123054,
      "learning_rate": 2.430696622578126e-06,
      "loss": 0.7521,
      "step": 9290
    },
    {
      "epoch": 0.6816768619826298,
      "grad_norm": 0.5533909787498267,
      "learning_rate": 2.4296773723027134e-06,
      "loss": 0.4789,
      "step": 9291
    },
    {
      "epoch": 0.6817502315727689,
      "grad_norm": 1.5000883434635015,
      "learning_rate": 2.428658267179785e-06,
      "loss": 0.7559,
      "step": 9292
    },
    {
      "epoch": 0.681823601162908,
      "grad_norm": 0.5456229799524341,
      "learning_rate": 2.4276393072668923e-06,
      "loss": 0.4825,
      "step": 9293
    },
    {
      "epoch": 0.6818969707530471,
      "grad_norm": 0.5521508979799934,
      "learning_rate": 2.4266204926215765e-06,
      "loss": 0.4812,
      "step": 9294
    },
    {
      "epoch": 0.6819703403431863,
      "grad_norm": 0.541748481480492,
      "learning_rate": 2.4256018233013757e-06,
      "loss": 0.4933,
      "step": 9295
    },
    {
      "epoch": 0.6820437099333254,
      "grad_norm": 0.5391922208289952,
      "learning_rate": 2.4245832993638156e-06,
      "loss": 0.4638,
      "step": 9296
    },
    {
      "epoch": 0.6820437099333254,
      "lm_loss": 1.0914,
      "step": 9296,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.6820437099333254,
      "lm_loss": 0.0872,
      "step": 9296,
      "vm_loss": 0.1863
    },
    {
      "epoch": 0.6820437099333254,
      "lm_loss": 0.7396,
      "step": 9296,
      "vm_loss": 0.0978
    },
    {
      "epoch": 0.6820437099333254,
      "lm_loss": 0.8093,
      "step": 9296,
      "vm_loss": 0.15
    },
    {
      "epoch": 0.6820437099333254,
      "lm_loss": 0.997,
      "step": 9296,
      "vm_loss": 0.2075
    },
    {
      "epoch": 0.6820437099333254,
      "lm_loss": 0.7213,
      "step": 9296,
      "vm_loss": 0.0544
    },
    {
      "epoch": 0.6820437099333254,
      "lm_loss": 0.2152,
      "step": 9296,
      "vm_loss": 0.0363
    },
    {
      "epoch": 0.6820437099333254,
      "lm_loss": 1.1358,
      "step": 9296,
      "vm_loss": 0.1371
    },
    {
      "epoch": 0.6821170795234646,
      "grad_norm": 1.5250363936273394,
      "learning_rate": 2.4235649208664145e-06,
      "loss": 0.7648,
      "step": 9297
    },
    {
      "epoch": 0.6821904491136036,
      "grad_norm": 1.550554753415571,
      "learning_rate": 2.422546687866683e-06,
      "loss": 0.6783,
      "step": 9298
    },
    {
      "epoch": 0.6822638187037428,
      "grad_norm": 0.5563745999764662,
      "learning_rate": 2.421528600422121e-06,
      "loss": 0.4709,
      "step": 9299
    },
    {
      "epoch": 0.6823371882938819,
      "grad_norm": 0.5518481268761946,
      "learning_rate": 2.420510658590226e-06,
      "loss": 0.5039,
      "step": 9300
    },
    {
      "epoch": 0.682410557884021,
      "grad_norm": 1.4488435626465892,
      "learning_rate": 2.4194928624284827e-06,
      "loss": 0.755,
      "step": 9301
    },
    {
      "epoch": 0.6824839274741602,
      "grad_norm": 1.5975518288576642,
      "learning_rate": 2.4184752119943673e-06,
      "loss": 0.7586,
      "step": 9302
    },
    {
      "epoch": 0.6825572970642992,
      "grad_norm": 0.5491089492220124,
      "learning_rate": 2.4174577073453503e-06,
      "loss": 0.4875,
      "step": 9303
    },
    {
      "epoch": 0.6826306666544384,
      "grad_norm": 1.4898429456096014,
      "learning_rate": 2.4164403485388897e-06,
      "loss": 0.655,
      "step": 9304
    },
    {
      "epoch": 0.6826306666544384,
      "lm_loss": 0.222,
      "step": 9304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6826306666544384,
      "lm_loss": 0.3891,
      "step": 9304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6826306666544384,
      "lm_loss": 0.4892,
      "step": 9304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6826306666544384,
      "lm_loss": 0.4254,
      "step": 9304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6826306666544384,
      "lm_loss": 0.5479,
      "step": 9304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6826306666544384,
      "lm_loss": 0.5316,
      "step": 9304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6826306666544384,
      "lm_loss": 0.5278,
      "step": 9304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6826306666544384,
      "lm_loss": 0.555,
      "step": 9304,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6827040362445775,
      "grad_norm": 0.545519816492103,
      "learning_rate": 2.415423135632443e-06,
      "loss": 0.4839,
      "step": 9305
    },
    {
      "epoch": 0.6827774058347167,
      "grad_norm": 0.5263092568090258,
      "learning_rate": 2.4144060686834524e-06,
      "loss": 0.4859,
      "step": 9306
    },
    {
      "epoch": 0.6828507754248558,
      "grad_norm": 1.5865934333484546,
      "learning_rate": 2.4133891477493537e-06,
      "loss": 0.7331,
      "step": 9307
    },
    {
      "epoch": 0.6829241450149949,
      "grad_norm": 1.5291076424773273,
      "learning_rate": 2.412372372887576e-06,
      "loss": 0.7609,
      "step": 9308
    },
    {
      "epoch": 0.682997514605134,
      "grad_norm": 0.5453318345675823,
      "learning_rate": 2.4113557441555384e-06,
      "loss": 0.4708,
      "step": 9309
    },
    {
      "epoch": 0.6830708841952732,
      "grad_norm": 1.5077210900502127,
      "learning_rate": 2.4103392616106507e-06,
      "loss": 0.6525,
      "step": 9310
    },
    {
      "epoch": 0.6831442537854123,
      "grad_norm": 1.4573039961934169,
      "learning_rate": 2.4093229253103203e-06,
      "loss": 0.6916,
      "step": 9311
    },
    {
      "epoch": 0.6832176233755514,
      "grad_norm": 1.4758443596577013,
      "learning_rate": 2.4083067353119392e-06,
      "loss": 0.7871,
      "step": 9312
    },
    {
      "epoch": 0.6832176233755514,
      "lm_loss": 0.8055,
      "step": 9312,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.6832176233755514,
      "lm_loss": 1.018,
      "step": 9312,
      "vm_loss": 0.1143
    },
    {
      "epoch": 0.6832176233755514,
      "lm_loss": 0.7134,
      "step": 9312,
      "vm_loss": 0.0886
    },
    {
      "epoch": 0.6832176233755514,
      "lm_loss": 1.0516,
      "step": 9312,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.6832176233755514,
      "lm_loss": 0.5861,
      "step": 9312,
      "vm_loss": 0.241
    },
    {
      "epoch": 0.6832176233755514,
      "lm_loss": 0.513,
      "step": 9312,
      "vm_loss": 0.1584
    },
    {
      "epoch": 0.6832176233755514,
      "lm_loss": 0.9625,
      "step": 9312,
      "vm_loss": 0.0678
    },
    {
      "epoch": 0.6832176233755514,
      "lm_loss": 1.1499,
      "step": 9312,
      "vm_loss": 0.168
    },
    {
      "epoch": 0.6832909929656905,
      "grad_norm": 1.6682070623646315,
      "learning_rate": 2.4072906916728957e-06,
      "loss": 0.8628,
      "step": 9313
    },
    {
      "epoch": 0.6833643625558297,
      "grad_norm": 0.5418244619762628,
      "learning_rate": 2.4062747944505673e-06,
      "loss": 0.4582,
      "step": 9314
    },
    {
      "epoch": 0.6834377321459688,
      "grad_norm": 0.581669154915933,
      "learning_rate": 2.4052590437023227e-06,
      "loss": 0.4954,
      "step": 9315
    },
    {
      "epoch": 0.683511101736108,
      "grad_norm": 1.5866314570492726,
      "learning_rate": 2.4042434394855272e-06,
      "loss": 0.8563,
      "step": 9316
    },
    {
      "epoch": 0.683584471326247,
      "grad_norm": 1.4759400686267568,
      "learning_rate": 2.403227981857534e-06,
      "loss": 0.6236,
      "step": 9317
    },
    {
      "epoch": 0.6836578409163862,
      "grad_norm": 1.5221397822686773,
      "learning_rate": 2.4022126708756864e-06,
      "loss": 0.7598,
      "step": 9318
    },
    {
      "epoch": 0.6837312105065253,
      "grad_norm": 1.4876508187882065,
      "learning_rate": 2.401197506597323e-06,
      "loss": 0.7,
      "step": 9319
    },
    {
      "epoch": 0.6838045800966644,
      "grad_norm": 1.4723161312456885,
      "learning_rate": 2.400182489079772e-06,
      "loss": 0.7205,
      "step": 9320
    },
    {
      "epoch": 0.6838045800966644,
      "lm_loss": 1.1304,
      "step": 9320,
      "vm_loss": 0.0864
    },
    {
      "epoch": 0.6838045800966644,
      "lm_loss": 0.0001,
      "step": 9320,
      "vm_loss": 0.0735
    },
    {
      "epoch": 0.6838045800966644,
      "lm_loss": 1.4171,
      "step": 9320,
      "vm_loss": 0.1278
    },
    {
      "epoch": 0.6838045800966644,
      "lm_loss": 0.9948,
      "step": 9320,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.6838045800966644,
      "lm_loss": 0.6245,
      "step": 9320,
      "vm_loss": 0.0331
    },
    {
      "epoch": 0.6838045800966644,
      "lm_loss": 0.4779,
      "step": 9320,
      "vm_loss": 0.2627
    },
    {
      "epoch": 0.6838045800966644,
      "lm_loss": 0.2941,
      "step": 9320,
      "vm_loss": 0.1337
    },
    {
      "epoch": 0.6838045800966644,
      "lm_loss": 0.0633,
      "step": 9320,
      "vm_loss": 0.0651
    },
    {
      "epoch": 0.6838779496868036,
      "grad_norm": 1.4506183300730136,
      "learning_rate": 2.3991676183803524e-06,
      "loss": 0.6927,
      "step": 9321
    },
    {
      "epoch": 0.6839513192769426,
      "grad_norm": 0.5440739727213818,
      "learning_rate": 2.39815289455638e-06,
      "loss": 0.4654,
      "step": 9322
    },
    {
      "epoch": 0.6840246888670818,
      "grad_norm": 1.5417683964898454,
      "learning_rate": 2.3971383176651575e-06,
      "loss": 0.7022,
      "step": 9323
    },
    {
      "epoch": 0.6840980584572209,
      "grad_norm": 1.3444712566216332,
      "learning_rate": 2.39612388776398e-06,
      "loss": 0.7464,
      "step": 9324
    },
    {
      "epoch": 0.6841714280473601,
      "grad_norm": 0.5609640652291443,
      "learning_rate": 2.3951096049101346e-06,
      "loss": 0.4787,
      "step": 9325
    },
    {
      "epoch": 0.6842447976374992,
      "grad_norm": 1.3873762664084517,
      "learning_rate": 2.3940954691608986e-06,
      "loss": 0.7068,
      "step": 9326
    },
    {
      "epoch": 0.6843181672276383,
      "grad_norm": 0.5669880351503277,
      "learning_rate": 2.3930814805735474e-06,
      "loss": 0.4909,
      "step": 9327
    },
    {
      "epoch": 0.6843915368177774,
      "grad_norm": 1.5190070782336662,
      "learning_rate": 2.3920676392053407e-06,
      "loss": 0.7294,
      "step": 9328
    },
    {
      "epoch": 0.6843915368177774,
      "lm_loss": 1.3779,
      "step": 9328,
      "vm_loss": 0.058
    },
    {
      "epoch": 0.6843915368177774,
      "lm_loss": 0.1385,
      "step": 9328,
      "vm_loss": 0.1118
    },
    {
      "epoch": 0.6843915368177774,
      "lm_loss": 0.8439,
      "step": 9328,
      "vm_loss": 0.2007
    },
    {
      "epoch": 0.6843915368177774,
      "lm_loss": 0.3211,
      "step": 9328,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.6843915368177774,
      "lm_loss": 0.9063,
      "step": 9328,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.6843915368177774,
      "lm_loss": 0.5202,
      "step": 9328,
      "vm_loss": 0.18
    },
    {
      "epoch": 0.6843915368177774,
      "lm_loss": 0.8958,
      "step": 9328,
      "vm_loss": 0.1679
    },
    {
      "epoch": 0.6843915368177774,
      "lm_loss": 1.0471,
      "step": 9328,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.6844649064079166,
      "grad_norm": 1.5391477849273456,
      "learning_rate": 2.391053945113533e-06,
      "loss": 0.7722,
      "step": 9329
    },
    {
      "epoch": 0.6845382759980557,
      "grad_norm": 1.6723078654979322,
      "learning_rate": 2.3900403983553695e-06,
      "loss": 0.8187,
      "step": 9330
    },
    {
      "epoch": 0.6846116455881949,
      "grad_norm": 1.458005352844151,
      "learning_rate": 2.3890269989880866e-06,
      "loss": 0.7534,
      "step": 9331
    },
    {
      "epoch": 0.6846850151783339,
      "grad_norm": 0.5465286086911664,
      "learning_rate": 2.388013747068916e-06,
      "loss": 0.4869,
      "step": 9332
    },
    {
      "epoch": 0.6847583847684731,
      "grad_norm": 0.5651829610472658,
      "learning_rate": 2.387000642655079e-06,
      "loss": 0.4805,
      "step": 9333
    },
    {
      "epoch": 0.6848317543586122,
      "grad_norm": 1.7106906196103069,
      "learning_rate": 2.385987685803786e-06,
      "loss": 0.6652,
      "step": 9334
    },
    {
      "epoch": 0.6849051239487514,
      "grad_norm": 1.5110453681127656,
      "learning_rate": 2.384974876572242e-06,
      "loss": 0.8014,
      "step": 9335
    },
    {
      "epoch": 0.6849784935388905,
      "grad_norm": 1.3891784422589792,
      "learning_rate": 2.3839622150176426e-06,
      "loss": 0.7491,
      "step": 9336
    },
    {
      "epoch": 0.6849784935388905,
      "lm_loss": 0.4856,
      "step": 9336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6849784935388905,
      "lm_loss": 0.5439,
      "step": 9336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6849784935388905,
      "lm_loss": 0.4644,
      "step": 9336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6849784935388905,
      "lm_loss": 0.4048,
      "step": 9336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6849784935388905,
      "lm_loss": 0.6318,
      "step": 9336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6849784935388905,
      "lm_loss": 0.5119,
      "step": 9336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6849784935388905,
      "lm_loss": 0.5875,
      "step": 9336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6849784935388905,
      "lm_loss": 0.5196,
      "step": 9336,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6850518631290295,
      "grad_norm": 0.546079804769215,
      "learning_rate": 2.3829497011971735e-06,
      "loss": 0.4955,
      "step": 9337
    },
    {
      "epoch": 0.6851252327191687,
      "grad_norm": 1.625463473198287,
      "learning_rate": 2.381937335168018e-06,
      "loss": 0.6931,
      "step": 9338
    },
    {
      "epoch": 0.6851986023093078,
      "grad_norm": 0.5308797361210092,
      "learning_rate": 2.380925116987345e-06,
      "loss": 0.4899,
      "step": 9339
    },
    {
      "epoch": 0.685271971899447,
      "grad_norm": 1.6980677111630158,
      "learning_rate": 2.379913046712316e-06,
      "loss": 0.778,
      "step": 9340
    },
    {
      "epoch": 0.6853453414895861,
      "grad_norm": 0.5495191039704314,
      "learning_rate": 2.378901124400086e-06,
      "loss": 0.4805,
      "step": 9341
    },
    {
      "epoch": 0.6854187110797252,
      "grad_norm": 1.5383590132091174,
      "learning_rate": 2.3778893501077983e-06,
      "loss": 0.6721,
      "step": 9342
    },
    {
      "epoch": 0.6854920806698643,
      "grad_norm": 1.7120924825276989,
      "learning_rate": 2.376877723892595e-06,
      "loss": 0.7161,
      "step": 9343
    },
    {
      "epoch": 0.6855654502600035,
      "grad_norm": 1.4287425087530647,
      "learning_rate": 2.3758662458116022e-06,
      "loss": 0.6637,
      "step": 9344
    },
    {
      "epoch": 0.6855654502600035,
      "lm_loss": 0.964,
      "step": 9344,
      "vm_loss": 0.0474
    },
    {
      "epoch": 0.6855654502600035,
      "lm_loss": 0.3973,
      "step": 9344,
      "vm_loss": 0.0907
    },
    {
      "epoch": 0.6855654502600035,
      "lm_loss": 1.0215,
      "step": 9344,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.6855654502600035,
      "lm_loss": 0.6916,
      "step": 9344,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.6855654502600035,
      "lm_loss": 0.8464,
      "step": 9344,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.6855654502600035,
      "lm_loss": 0.0199,
      "step": 9344,
      "vm_loss": 0.0653
    },
    {
      "epoch": 0.6855654502600035,
      "lm_loss": 0.2268,
      "step": 9344,
      "vm_loss": 0.2135
    },
    {
      "epoch": 0.6855654502600035,
      "lm_loss": 0.8815,
      "step": 9344,
      "vm_loss": 0.1375
    },
    {
      "epoch": 0.6856388198501426,
      "grad_norm": 1.5535559777967525,
      "learning_rate": 2.3748549159219415e-06,
      "loss": 0.7828,
      "step": 9345
    },
    {
      "epoch": 0.6857121894402818,
      "grad_norm": 1.6395183530999247,
      "learning_rate": 2.3738437342807246e-06,
      "loss": 0.7847,
      "step": 9346
    },
    {
      "epoch": 0.6857855590304208,
      "grad_norm": 1.557690322369639,
      "learning_rate": 2.3728327009450537e-06,
      "loss": 0.7915,
      "step": 9347
    },
    {
      "epoch": 0.68585892862056,
      "grad_norm": 0.5494972012662473,
      "learning_rate": 2.3718218159720295e-06,
      "loss": 0.4733,
      "step": 9348
    },
    {
      "epoch": 0.6859322982106991,
      "grad_norm": 1.506818618053865,
      "learning_rate": 2.370811079418735e-06,
      "loss": 0.7165,
      "step": 9349
    },
    {
      "epoch": 0.6860056678008383,
      "grad_norm": 0.5495311992055743,
      "learning_rate": 2.3698004913422495e-06,
      "loss": 0.4818,
      "step": 9350
    },
    {
      "epoch": 0.6860790373909774,
      "grad_norm": 1.352021730675892,
      "learning_rate": 2.3687900517996437e-06,
      "loss": 0.6835,
      "step": 9351
    },
    {
      "epoch": 0.6861524069811165,
      "grad_norm": 0.5625364153875522,
      "learning_rate": 2.367779760847978e-06,
      "loss": 0.494,
      "step": 9352
    },
    {
      "epoch": 0.6861524069811165,
      "lm_loss": 1.5315,
      "step": 9352,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.6861524069811165,
      "lm_loss": 1.08,
      "step": 9352,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.6861524069811165,
      "lm_loss": 0.4682,
      "step": 9352,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.6861524069811165,
      "lm_loss": 0.0161,
      "step": 9352,
      "vm_loss": 0.0802
    },
    {
      "epoch": 0.6861524069811165,
      "lm_loss": 0.841,
      "step": 9352,
      "vm_loss": 0.2147
    },
    {
      "epoch": 0.6861524069811165,
      "lm_loss": 0.2506,
      "step": 9352,
      "vm_loss": 0.2329
    },
    {
      "epoch": 0.6861524069811165,
      "lm_loss": 0.1181,
      "step": 9352,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.6861524069811165,
      "lm_loss": 0.2207,
      "step": 9352,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.6862257765712556,
      "grad_norm": 1.4894865424027743,
      "learning_rate": 2.3667696185443094e-06,
      "loss": 0.7607,
      "step": 9353
    },
    {
      "epoch": 0.6862991461613948,
      "grad_norm": 1.339657610221749,
      "learning_rate": 2.365759624945682e-06,
      "loss": 0.7188,
      "step": 9354
    },
    {
      "epoch": 0.6863725157515339,
      "grad_norm": 1.490692933682775,
      "learning_rate": 2.364749780109132e-06,
      "loss": 0.7666,
      "step": 9355
    },
    {
      "epoch": 0.6864458853416731,
      "grad_norm": 1.5519674123587122,
      "learning_rate": 2.363740084091688e-06,
      "loss": 0.694,
      "step": 9356
    },
    {
      "epoch": 0.6865192549318121,
      "grad_norm": 1.4311162823647747,
      "learning_rate": 2.3627305369503677e-06,
      "loss": 0.7534,
      "step": 9357
    },
    {
      "epoch": 0.6865926245219512,
      "grad_norm": 1.5522440175331025,
      "learning_rate": 2.361721138742187e-06,
      "loss": 0.7309,
      "step": 9358
    },
    {
      "epoch": 0.6866659941120904,
      "grad_norm": 0.5656444719780891,
      "learning_rate": 2.360711889524148e-06,
      "loss": 0.4879,
      "step": 9359
    },
    {
      "epoch": 0.6867393637022295,
      "grad_norm": 0.5650175456350948,
      "learning_rate": 2.3597027893532444e-06,
      "loss": 0.475,
      "step": 9360
    },
    {
      "epoch": 0.6867393637022295,
      "lm_loss": 1.2203,
      "step": 9360,
      "vm_loss": 0.0574
    },
    {
      "epoch": 0.6867393637022295,
      "lm_loss": 0.1119,
      "step": 9360,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.6867393637022295,
      "lm_loss": 0.197,
      "step": 9360,
      "vm_loss": 0.1671
    },
    {
      "epoch": 0.6867393637022295,
      "lm_loss": 0.498,
      "step": 9360,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.6867393637022295,
      "lm_loss": 0.4853,
      "step": 9360,
      "vm_loss": 0.0695
    },
    {
      "epoch": 0.6867393637022295,
      "lm_loss": 0.9204,
      "step": 9360,
      "vm_loss": 0.2188
    },
    {
      "epoch": 0.6867393637022295,
      "lm_loss": 0.8378,
      "step": 9360,
      "vm_loss": 0.1021
    },
    {
      "epoch": 0.6867393637022295,
      "lm_loss": 0.1315,
      "step": 9360,
      "vm_loss": 0.2433
    },
    {
      "epoch": 0.6868127332923687,
      "grad_norm": 1.4259965751757422,
      "learning_rate": 2.3586938382864626e-06,
      "loss": 0.6865,
      "step": 9361
    },
    {
      "epoch": 0.6868861028825077,
      "grad_norm": 0.5404529911669772,
      "learning_rate": 2.3576850363807805e-06,
      "loss": 0.4916,
      "step": 9362
    },
    {
      "epoch": 0.6869594724726469,
      "grad_norm": 1.6746171922542508,
      "learning_rate": 2.3566763836931667e-06,
      "loss": 0.7739,
      "step": 9363
    },
    {
      "epoch": 0.687032842062786,
      "grad_norm": 0.5593263334763796,
      "learning_rate": 2.3556678802805853e-06,
      "loss": 0.5003,
      "step": 9364
    },
    {
      "epoch": 0.6871062116529252,
      "grad_norm": 0.5379363446961648,
      "learning_rate": 2.354659526199987e-06,
      "loss": 0.4635,
      "step": 9365
    },
    {
      "epoch": 0.6871795812430643,
      "grad_norm": 1.427658032410899,
      "learning_rate": 2.3536513215083167e-06,
      "loss": 0.7442,
      "step": 9366
    },
    {
      "epoch": 0.6872529508332034,
      "grad_norm": 1.5528634578937657,
      "learning_rate": 2.3526432662625097e-06,
      "loss": 0.739,
      "step": 9367
    },
    {
      "epoch": 0.6873263204233425,
      "grad_norm": 1.4487474004990348,
      "learning_rate": 2.351635360519492e-06,
      "loss": 0.6982,
      "step": 9368
    },
    {
      "epoch": 0.6873263204233425,
      "lm_loss": 1.3855,
      "step": 9368,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.6873263204233425,
      "lm_loss": 0.9447,
      "step": 9368,
      "vm_loss": 0.0888
    },
    {
      "epoch": 0.6873263204233425,
      "lm_loss": 0.9659,
      "step": 9368,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.6873263204233425,
      "lm_loss": 1.0351,
      "step": 9368,
      "vm_loss": 0.1648
    },
    {
      "epoch": 0.6873263204233425,
      "lm_loss": 0.5327,
      "step": 9368,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.6873263204233425,
      "lm_loss": 0.7137,
      "step": 9368,
      "vm_loss": 0.207
    },
    {
      "epoch": 0.6873263204233425,
      "lm_loss": 0.4889,
      "step": 9368,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.6873263204233425,
      "lm_loss": 0.7121,
      "step": 9368,
      "vm_loss": 0.1341
    },
    {
      "epoch": 0.6873996900134817,
      "grad_norm": 1.5785833518186647,
      "learning_rate": 2.350627604336186e-06,
      "loss": 0.757,
      "step": 9369
    },
    {
      "epoch": 0.6874730596036208,
      "grad_norm": 1.5390780811583893,
      "learning_rate": 2.3496199977694993e-06,
      "loss": 0.7715,
      "step": 9370
    },
    {
      "epoch": 0.68754642919376,
      "grad_norm": 0.5350831659455522,
      "learning_rate": 2.3486125408763356e-06,
      "loss": 0.4805,
      "step": 9371
    },
    {
      "epoch": 0.687619798783899,
      "grad_norm": 1.8365308516032726,
      "learning_rate": 2.3476052337135876e-06,
      "loss": 0.9174,
      "step": 9372
    },
    {
      "epoch": 0.6876931683740382,
      "grad_norm": 1.5970697570223724,
      "learning_rate": 2.3465980763381384e-06,
      "loss": 0.8472,
      "step": 9373
    },
    {
      "epoch": 0.6877665379641773,
      "grad_norm": 1.606787645123884,
      "learning_rate": 2.3455910688068694e-06,
      "loss": 0.8385,
      "step": 9374
    },
    {
      "epoch": 0.6878399075543165,
      "grad_norm": 1.5458030833522207,
      "learning_rate": 2.3445842111766457e-06,
      "loss": 0.7101,
      "step": 9375
    },
    {
      "epoch": 0.6879132771444556,
      "grad_norm": 1.5188451106512912,
      "learning_rate": 2.3435775035043278e-06,
      "loss": 0.719,
      "step": 9376
    },
    {
      "epoch": 0.6879132771444556,
      "lm_loss": 1.2991,
      "step": 9376,
      "vm_loss": 0.0819
    },
    {
      "epoch": 0.6879132771444556,
      "lm_loss": 1.198,
      "step": 9376,
      "vm_loss": 0.2177
    },
    {
      "epoch": 0.6879132771444556,
      "lm_loss": 0.3697,
      "step": 9376,
      "vm_loss": 0.0802
    },
    {
      "epoch": 0.6879132771444556,
      "lm_loss": 0.3209,
      "step": 9376,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.6879132771444556,
      "lm_loss": 0.4895,
      "step": 9376,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.6879132771444556,
      "lm_loss": 0.2691,
      "step": 9376,
      "vm_loss": 0.0779
    },
    {
      "epoch": 0.6879132771444556,
      "lm_loss": 0.1529,
      "step": 9376,
      "vm_loss": 0.0955
    },
    {
      "epoch": 0.6879132771444556,
      "lm_loss": 0.7739,
      "step": 9376,
      "vm_loss": 0.1288
    },
    {
      "epoch": 0.6879866467345946,
      "grad_norm": 1.4287155626180006,
      "learning_rate": 2.342570945846767e-06,
      "loss": 0.7207,
      "step": 9377
    },
    {
      "epoch": 0.6880600163247338,
      "grad_norm": 1.558281905820211,
      "learning_rate": 2.341564538260806e-06,
      "loss": 0.7825,
      "step": 9378
    },
    {
      "epoch": 0.6881333859148729,
      "grad_norm": 1.4860510659843829,
      "learning_rate": 2.340558280803277e-06,
      "loss": 0.7568,
      "step": 9379
    },
    {
      "epoch": 0.6882067555050121,
      "grad_norm": 1.5461285063625427,
      "learning_rate": 2.3395521735310097e-06,
      "loss": 0.7284,
      "step": 9380
    },
    {
      "epoch": 0.6882801250951512,
      "grad_norm": 0.5645008295144791,
      "learning_rate": 2.3385462165008193e-06,
      "loss": 0.4709,
      "step": 9381
    },
    {
      "epoch": 0.6883534946852903,
      "grad_norm": 1.5353592243366596,
      "learning_rate": 2.3375404097695155e-06,
      "loss": 0.8242,
      "step": 9382
    },
    {
      "epoch": 0.6884268642754294,
      "grad_norm": 1.3896782167322062,
      "learning_rate": 2.336534753393898e-06,
      "loss": 0.712,
      "step": 9383
    },
    {
      "epoch": 0.6885002338655686,
      "grad_norm": 1.554739566582903,
      "learning_rate": 2.335529247430757e-06,
      "loss": 0.7117,
      "step": 9384
    },
    {
      "epoch": 0.6885002338655686,
      "lm_loss": 1.2881,
      "step": 9384,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.6885002338655686,
      "lm_loss": 0.7363,
      "step": 9384,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.6885002338655686,
      "lm_loss": 0.3353,
      "step": 9384,
      "vm_loss": 0.0896
    },
    {
      "epoch": 0.6885002338655686,
      "lm_loss": 1.2745,
      "step": 9384,
      "vm_loss": 0.1086
    },
    {
      "epoch": 0.6885002338655686,
      "lm_loss": 0.445,
      "step": 9384,
      "vm_loss": 0.188
    },
    {
      "epoch": 0.6885002338655686,
      "lm_loss": 0.4275,
      "step": 9384,
      "vm_loss": 0.1491
    },
    {
      "epoch": 0.6885002338655686,
      "lm_loss": 0.1031,
      "step": 9384,
      "vm_loss": 0.0481
    },
    {
      "epoch": 0.6885002338655686,
      "lm_loss": 0.5009,
      "step": 9384,
      "vm_loss": 0.0559
    },
    {
      "epoch": 0.6885736034557077,
      "grad_norm": 1.6810184528105718,
      "learning_rate": 2.33452389193688e-06,
      "loss": 0.734,
      "step": 9385
    },
    {
      "epoch": 0.6886469730458469,
      "grad_norm": 0.5507185141785902,
      "learning_rate": 2.3335186869690396e-06,
      "loss": 0.4898,
      "step": 9386
    },
    {
      "epoch": 0.6887203426359859,
      "grad_norm": 1.5074623063873862,
      "learning_rate": 2.3325136325840025e-06,
      "loss": 0.8084,
      "step": 9387
    },
    {
      "epoch": 0.6887937122261251,
      "grad_norm": 1.5909619113452742,
      "learning_rate": 2.331508728838526e-06,
      "loss": 0.7277,
      "step": 9388
    },
    {
      "epoch": 0.6888670818162642,
      "grad_norm": 1.5841103032448371,
      "learning_rate": 2.330503975789361e-06,
      "loss": 0.6826,
      "step": 9389
    },
    {
      "epoch": 0.6889404514064034,
      "grad_norm": 1.4830347783954287,
      "learning_rate": 2.3294993734932447e-06,
      "loss": 0.8488,
      "step": 9390
    },
    {
      "epoch": 0.6890138209965424,
      "grad_norm": 1.5645505479766324,
      "learning_rate": 2.328494922006914e-06,
      "loss": 0.711,
      "step": 9391
    },
    {
      "epoch": 0.6890871905866816,
      "grad_norm": 1.4509794827937907,
      "learning_rate": 2.3274906213870914e-06,
      "loss": 0.8264,
      "step": 9392
    },
    {
      "epoch": 0.6890871905866816,
      "lm_loss": 1.3609,
      "step": 9392,
      "vm_loss": 0.061
    },
    {
      "epoch": 0.6890871905866816,
      "lm_loss": 0.7923,
      "step": 9392,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.6890871905866816,
      "lm_loss": 1.1082,
      "step": 9392,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.6890871905866816,
      "lm_loss": 0.0119,
      "step": 9392,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.6890871905866816,
      "lm_loss": 0.3698,
      "step": 9392,
      "vm_loss": 0.0755
    },
    {
      "epoch": 0.6890871905866816,
      "lm_loss": 0.0183,
      "step": 9392,
      "vm_loss": 0.1387
    },
    {
      "epoch": 0.6890871905866816,
      "lm_loss": 0.093,
      "step": 9392,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.6890871905866816,
      "lm_loss": 1.082,
      "step": 9392,
      "vm_loss": 0.1458
    },
    {
      "epoch": 0.6891605601768207,
      "grad_norm": 1.372192239724408,
      "learning_rate": 2.3264864716904918e-06,
      "loss": 0.6437,
      "step": 9393
    },
    {
      "epoch": 0.6892339297669599,
      "grad_norm": 0.5484436986948353,
      "learning_rate": 2.325482472973821e-06,
      "loss": 0.476,
      "step": 9394
    },
    {
      "epoch": 0.689307299357099,
      "grad_norm": 1.5434338520156168,
      "learning_rate": 2.324478625293778e-06,
      "loss": 0.7988,
      "step": 9395
    },
    {
      "epoch": 0.689380668947238,
      "grad_norm": 1.4617258986369182,
      "learning_rate": 2.3234749287070535e-06,
      "loss": 0.68,
      "step": 9396
    },
    {
      "epoch": 0.6894540385373772,
      "grad_norm": 1.5634614706299734,
      "learning_rate": 2.3224713832703283e-06,
      "loss": 0.7813,
      "step": 9397
    },
    {
      "epoch": 0.6895274081275163,
      "grad_norm": 1.6855159147683934,
      "learning_rate": 2.321467989040275e-06,
      "loss": 0.7815,
      "step": 9398
    },
    {
      "epoch": 0.6896007777176555,
      "grad_norm": 1.507855950864232,
      "learning_rate": 2.320464746073558e-06,
      "loss": 0.7355,
      "step": 9399
    },
    {
      "epoch": 0.6896741473077946,
      "grad_norm": 1.3179478641462874,
      "learning_rate": 2.31946165442683e-06,
      "loss": 0.6713,
      "step": 9400
    },
    {
      "epoch": 0.6896741473077946,
      "lm_loss": 0.97,
      "step": 9400,
      "vm_loss": 0.0473
    },
    {
      "epoch": 0.6896741473077946,
      "lm_loss": 0.6274,
      "step": 9400,
      "vm_loss": 0.0954
    },
    {
      "epoch": 0.6896741473077946,
      "lm_loss": 0.0022,
      "step": 9400,
      "vm_loss": 0.0802
    },
    {
      "epoch": 0.6896741473077946,
      "lm_loss": 0.3663,
      "step": 9400,
      "vm_loss": 0.1324
    },
    {
      "epoch": 0.6896741473077946,
      "lm_loss": 0.1036,
      "step": 9400,
      "vm_loss": 0.0587
    },
    {
      "epoch": 0.6896741473077946,
      "lm_loss": 0.2231,
      "step": 9400,
      "vm_loss": 0.1568
    },
    {
      "epoch": 0.6896741473077946,
      "lm_loss": 1.7002,
      "step": 9400,
      "vm_loss": 0.1441
    },
    {
      "epoch": 0.6896741473077946,
      "lm_loss": 0.6832,
      "step": 9400,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.6897475168979337,
      "grad_norm": 1.618356165803855,
      "learning_rate": 2.318458714156743e-06,
      "loss": 0.824,
      "step": 9401
    },
    {
      "epoch": 0.6898208864880728,
      "grad_norm": 0.5778166181350125,
      "learning_rate": 2.3174559253199323e-06,
      "loss": 0.48,
      "step": 9402
    },
    {
      "epoch": 0.689894256078212,
      "grad_norm": 1.4729527847602004,
      "learning_rate": 2.3164532879730293e-06,
      "loss": 0.7875,
      "step": 9403
    },
    {
      "epoch": 0.6899676256683511,
      "grad_norm": 0.5591235125831179,
      "learning_rate": 2.315450802172655e-06,
      "loss": 0.481,
      "step": 9404
    },
    {
      "epoch": 0.6900409952584903,
      "grad_norm": 1.56222472396852,
      "learning_rate": 2.314448467975422e-06,
      "loss": 0.6587,
      "step": 9405
    },
    {
      "epoch": 0.6901143648486293,
      "grad_norm": 1.6120786082923122,
      "learning_rate": 2.3134462854379332e-06,
      "loss": 0.7751,
      "step": 9406
    },
    {
      "epoch": 0.6901877344387685,
      "grad_norm": 1.4529887302670863,
      "learning_rate": 2.312444254616787e-06,
      "loss": 0.7822,
      "step": 9407
    },
    {
      "epoch": 0.6902611040289076,
      "grad_norm": 1.3598405587853506,
      "learning_rate": 2.3114423755685693e-06,
      "loss": 0.6078,
      "step": 9408
    },
    {
      "epoch": 0.6902611040289076,
      "lm_loss": 1.1928,
      "step": 9408,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.6902611040289076,
      "lm_loss": 0.4413,
      "step": 9408,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.6902611040289076,
      "lm_loss": 0.6792,
      "step": 9408,
      "vm_loss": 0.1065
    },
    {
      "epoch": 0.6902611040289076,
      "lm_loss": 0.3788,
      "step": 9408,
      "vm_loss": 0.0801
    },
    {
      "epoch": 0.6902611040289076,
      "lm_loss": 0.4812,
      "step": 9408,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.6902611040289076,
      "lm_loss": 0.9842,
      "step": 9408,
      "vm_loss": 0.1052
    },
    {
      "epoch": 0.6902611040289076,
      "lm_loss": 0.5445,
      "step": 9408,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.6902611040289076,
      "lm_loss": 0.1768,
      "step": 9408,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.6903344736190468,
      "grad_norm": 1.6551637489416504,
      "learning_rate": 2.3104406483498593e-06,
      "loss": 0.793,
      "step": 9409
    },
    {
      "epoch": 0.6904078432091859,
      "grad_norm": 1.6410117926844612,
      "learning_rate": 2.309439073017226e-06,
      "loss": 0.945,
      "step": 9410
    },
    {
      "epoch": 0.690481212799325,
      "grad_norm": 0.5664777764215697,
      "learning_rate": 2.30843764962723e-06,
      "loss": 0.499,
      "step": 9411
    },
    {
      "epoch": 0.6905545823894641,
      "grad_norm": 0.5518418875640454,
      "learning_rate": 2.3074363782364265e-06,
      "loss": 0.4928,
      "step": 9412
    },
    {
      "epoch": 0.6906279519796032,
      "grad_norm": 1.515653418325139,
      "learning_rate": 2.3064352589013594e-06,
      "loss": 0.6816,
      "step": 9413
    },
    {
      "epoch": 0.6907013215697424,
      "grad_norm": 1.4429224834634342,
      "learning_rate": 2.305434291678563e-06,
      "loss": 0.7152,
      "step": 9414
    },
    {
      "epoch": 0.6907746911598815,
      "grad_norm": 0.544625972523802,
      "learning_rate": 2.3044334766245663e-06,
      "loss": 0.48,
      "step": 9415
    },
    {
      "epoch": 0.6908480607500206,
      "grad_norm": 0.5462664678742479,
      "learning_rate": 2.3034328137958855e-06,
      "loss": 0.4853,
      "step": 9416
    },
    {
      "epoch": 0.6908480607500206,
      "lm_loss": 1.3551,
      "step": 9416,
      "vm_loss": 0.0598
    },
    {
      "epoch": 0.6908480607500206,
      "lm_loss": 0.7482,
      "step": 9416,
      "vm_loss": 0.1538
    },
    {
      "epoch": 0.6908480607500206,
      "lm_loss": 1.0714,
      "step": 9416,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.6908480607500206,
      "lm_loss": 0.0065,
      "step": 9416,
      "vm_loss": 0.0727
    },
    {
      "epoch": 0.6908480607500206,
      "lm_loss": 0.8969,
      "step": 9416,
      "vm_loss": 0.1523
    },
    {
      "epoch": 0.6908480607500206,
      "lm_loss": 0.8747,
      "step": 9416,
      "vm_loss": 0.154
    },
    {
      "epoch": 0.6908480607500206,
      "lm_loss": 0.9443,
      "step": 9416,
      "vm_loss": 0.0695
    },
    {
      "epoch": 0.6908480607500206,
      "lm_loss": 0.752,
      "step": 9416,
      "vm_loss": 0.0721
    },
    {
      "epoch": 0.6909214303401597,
      "grad_norm": 1.5299148803976155,
      "learning_rate": 2.3024323032490304e-06,
      "loss": 0.7393,
      "step": 9417
    },
    {
      "epoch": 0.6909947999302989,
      "grad_norm": 0.5828715498437036,
      "learning_rate": 2.3014319450405054e-06,
      "loss": 0.4878,
      "step": 9418
    },
    {
      "epoch": 0.691068169520438,
      "grad_norm": 1.576932518795896,
      "learning_rate": 2.300431739226801e-06,
      "loss": 0.797,
      "step": 9419
    },
    {
      "epoch": 0.6911415391105772,
      "grad_norm": 1.5656146213837945,
      "learning_rate": 2.2994316858644022e-06,
      "loss": 0.7419,
      "step": 9420
    },
    {
      "epoch": 0.6912149087007162,
      "grad_norm": 1.6547524090384167,
      "learning_rate": 2.2984317850097835e-06,
      "loss": 0.8099,
      "step": 9421
    },
    {
      "epoch": 0.6912882782908554,
      "grad_norm": 0.5760435772217413,
      "learning_rate": 2.297432036719411e-06,
      "loss": 0.5155,
      "step": 9422
    },
    {
      "epoch": 0.6913616478809945,
      "grad_norm": 1.3609024642384553,
      "learning_rate": 2.296432441049745e-06,
      "loss": 0.677,
      "step": 9423
    },
    {
      "epoch": 0.6914350174711337,
      "grad_norm": 1.440236301658163,
      "learning_rate": 2.295432998057235e-06,
      "loss": 0.8179,
      "step": 9424
    },
    {
      "epoch": 0.6914350174711337,
      "lm_loss": 1.3666,
      "step": 9424,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.6914350174711337,
      "lm_loss": 0.7512,
      "step": 9424,
      "vm_loss": 0.1326
    },
    {
      "epoch": 0.6914350174711337,
      "lm_loss": 0.8742,
      "step": 9424,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.6914350174711337,
      "lm_loss": 0.4594,
      "step": 9424,
      "vm_loss": 0.1093
    },
    {
      "epoch": 0.6914350174711337,
      "lm_loss": 1.2662,
      "step": 9424,
      "vm_loss": 0.1319
    },
    {
      "epoch": 0.6914350174711337,
      "lm_loss": 0.6511,
      "step": 9424,
      "vm_loss": 0.1878
    },
    {
      "epoch": 0.6914350174711337,
      "lm_loss": 0.2048,
      "step": 9424,
      "vm_loss": 0.0705
    },
    {
      "epoch": 0.6914350174711337,
      "lm_loss": 0.1276,
      "step": 9424,
      "vm_loss": 0.1549
    },
    {
      "epoch": 0.6915083870612728,
      "grad_norm": 1.5004428327251456,
      "learning_rate": 2.2944337077983213e-06,
      "loss": 0.6764,
      "step": 9425
    },
    {
      "epoch": 0.6915817566514119,
      "grad_norm": 1.426268157632546,
      "learning_rate": 2.2934345703294363e-06,
      "loss": 0.7373,
      "step": 9426
    },
    {
      "epoch": 0.691655126241551,
      "grad_norm": 1.5120987214446107,
      "learning_rate": 2.292435585707002e-06,
      "loss": 0.7299,
      "step": 9427
    },
    {
      "epoch": 0.6917284958316902,
      "grad_norm": 0.5454671916290298,
      "learning_rate": 2.2914367539874367e-06,
      "loss": 0.4744,
      "step": 9428
    },
    {
      "epoch": 0.6918018654218293,
      "grad_norm": 0.5516330628312766,
      "learning_rate": 2.290438075227146e-06,
      "loss": 0.4895,
      "step": 9429
    },
    {
      "epoch": 0.6918752350119685,
      "grad_norm": 0.5461600262080395,
      "learning_rate": 2.2894395494825274e-06,
      "loss": 0.4835,
      "step": 9430
    },
    {
      "epoch": 0.6919486046021075,
      "grad_norm": 0.5603399296045094,
      "learning_rate": 2.2884411768099696e-06,
      "loss": 0.4816,
      "step": 9431
    },
    {
      "epoch": 0.6920219741922466,
      "grad_norm": 1.4433947644407057,
      "learning_rate": 2.287442957265854e-06,
      "loss": 0.7221,
      "step": 9432
    },
    {
      "epoch": 0.6920219741922466,
      "lm_loss": 0.517,
      "step": 9432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6920219741922466,
      "lm_loss": 0.3784,
      "step": 9432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6920219741922466,
      "lm_loss": 0.3406,
      "step": 9432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6920219741922466,
      "lm_loss": 0.4735,
      "step": 9432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6920219741922466,
      "lm_loss": 0.7768,
      "step": 9432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6920219741922466,
      "lm_loss": 0.5721,
      "step": 9432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6920219741922466,
      "lm_loss": 0.5441,
      "step": 9432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6920219741922466,
      "lm_loss": 0.4782,
      "step": 9432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6920953437823858,
      "grad_norm": 0.5703905482393449,
      "learning_rate": 2.28644489090655e-06,
      "loss": 0.4931,
      "step": 9433
    },
    {
      "epoch": 0.692168713372525,
      "grad_norm": 1.3796595631354833,
      "learning_rate": 2.2854469777884253e-06,
      "loss": 0.5486,
      "step": 9434
    },
    {
      "epoch": 0.6922420829626641,
      "grad_norm": 0.5338541332109091,
      "learning_rate": 2.284449217967832e-06,
      "loss": 0.4584,
      "step": 9435
    },
    {
      "epoch": 0.6923154525528031,
      "grad_norm": 0.5469913878608537,
      "learning_rate": 2.2834516115011163e-06,
      "loss": 0.4823,
      "step": 9436
    },
    {
      "epoch": 0.6923888221429423,
      "grad_norm": 0.5505863411185388,
      "learning_rate": 2.2824541584446153e-06,
      "loss": 0.4796,
      "step": 9437
    },
    {
      "epoch": 0.6924621917330814,
      "grad_norm": 1.6605035522068967,
      "learning_rate": 2.2814568588546563e-06,
      "loss": 0.713,
      "step": 9438
    },
    {
      "epoch": 0.6925355613232206,
      "grad_norm": 1.3946968677534988,
      "learning_rate": 2.2804597127875628e-06,
      "loss": 0.7815,
      "step": 9439
    },
    {
      "epoch": 0.6926089309133597,
      "grad_norm": 1.4911746419387364,
      "learning_rate": 2.279462720299644e-06,
      "loss": 0.7606,
      "step": 9440
    },
    {
      "epoch": 0.6926089309133597,
      "lm_loss": 0.556,
      "step": 9440,
      "vm_loss": 0.0698
    },
    {
      "epoch": 0.6926089309133597,
      "lm_loss": 0.8226,
      "step": 9440,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.6926089309133597,
      "lm_loss": 0.341,
      "step": 9440,
      "vm_loss": 0.1228
    },
    {
      "epoch": 0.6926089309133597,
      "lm_loss": 0.7105,
      "step": 9440,
      "vm_loss": 0.1344
    },
    {
      "epoch": 0.6926089309133597,
      "lm_loss": 0.3466,
      "step": 9440,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.6926089309133597,
      "lm_loss": 1.069,
      "step": 9440,
      "vm_loss": 0.1364
    },
    {
      "epoch": 0.6926089309133597,
      "lm_loss": 0.3012,
      "step": 9440,
      "vm_loss": 0.197
    },
    {
      "epoch": 0.6926089309133597,
      "lm_loss": 0.688,
      "step": 9440,
      "vm_loss": 0.1444
    },
    {
      "epoch": 0.6926823005034988,
      "grad_norm": 1.5180912253631547,
      "learning_rate": 2.2784658814472026e-06,
      "loss": 0.7652,
      "step": 9441
    },
    {
      "epoch": 0.6927556700936379,
      "grad_norm": 0.564047655783184,
      "learning_rate": 2.2774691962865325e-06,
      "loss": 0.4877,
      "step": 9442
    },
    {
      "epoch": 0.6928290396837771,
      "grad_norm": 0.5591190486303903,
      "learning_rate": 2.2764726648739196e-06,
      "loss": 0.4659,
      "step": 9443
    },
    {
      "epoch": 0.6929024092739162,
      "grad_norm": 1.4999992477191415,
      "learning_rate": 2.2754762872656377e-06,
      "loss": 0.7212,
      "step": 9444
    },
    {
      "epoch": 0.6929757788640554,
      "grad_norm": 1.6823415100228807,
      "learning_rate": 2.274480063517959e-06,
      "loss": 0.7602,
      "step": 9445
    },
    {
      "epoch": 0.6930491484541944,
      "grad_norm": 1.486780598295607,
      "learning_rate": 2.2734839936871413e-06,
      "loss": 0.6775,
      "step": 9446
    },
    {
      "epoch": 0.6931225180443336,
      "grad_norm": 0.5673192744601568,
      "learning_rate": 2.272488077829434e-06,
      "loss": 0.4866,
      "step": 9447
    },
    {
      "epoch": 0.6931958876344727,
      "grad_norm": 0.5771180388818397,
      "learning_rate": 2.2714923160010794e-06,
      "loss": 0.4782,
      "step": 9448
    },
    {
      "epoch": 0.6931958876344727,
      "lm_loss": 0.4878,
      "step": 9448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6931958876344727,
      "lm_loss": 0.5094,
      "step": 9448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6931958876344727,
      "lm_loss": 0.4711,
      "step": 9448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6931958876344727,
      "lm_loss": 0.549,
      "step": 9448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6931958876344727,
      "lm_loss": 0.4792,
      "step": 9448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6931958876344727,
      "lm_loss": 0.4987,
      "step": 9448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6931958876344727,
      "lm_loss": 0.5704,
      "step": 9448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6931958876344727,
      "lm_loss": 0.4135,
      "step": 9448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6932692572246119,
      "grad_norm": 0.5408793591655157,
      "learning_rate": 2.270496708258309e-06,
      "loss": 0.4667,
      "step": 9449
    },
    {
      "epoch": 0.693342626814751,
      "grad_norm": 0.5651664734490566,
      "learning_rate": 2.2695012546573506e-06,
      "loss": 0.4758,
      "step": 9450
    },
    {
      "epoch": 0.69341599640489,
      "grad_norm": 0.5430354832277426,
      "learning_rate": 2.2685059552544187e-06,
      "loss": 0.489,
      "step": 9451
    },
    {
      "epoch": 0.6934893659950292,
      "grad_norm": 1.6426140331281038,
      "learning_rate": 2.2675108101057202e-06,
      "loss": 0.7881,
      "step": 9452
    },
    {
      "epoch": 0.6935627355851683,
      "grad_norm": 1.3517279216553157,
      "learning_rate": 2.2665158192674525e-06,
      "loss": 0.7357,
      "step": 9453
    },
    {
      "epoch": 0.6936361051753075,
      "grad_norm": 0.5712623019210097,
      "learning_rate": 2.2655209827958046e-06,
      "loss": 0.5076,
      "step": 9454
    },
    {
      "epoch": 0.6937094747654466,
      "grad_norm": 1.4654630142392302,
      "learning_rate": 2.2645263007469604e-06,
      "loss": 0.8079,
      "step": 9455
    },
    {
      "epoch": 0.6937828443555857,
      "grad_norm": 1.4286319553414781,
      "learning_rate": 2.26353177317709e-06,
      "loss": 0.727,
      "step": 9456
    },
    {
      "epoch": 0.6937828443555857,
      "lm_loss": 1.5707,
      "step": 9456,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.6937828443555857,
      "lm_loss": 1.347,
      "step": 9456,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.6937828443555857,
      "lm_loss": 0.0451,
      "step": 9456,
      "vm_loss": 0.0853
    },
    {
      "epoch": 0.6937828443555857,
      "lm_loss": 0.7755,
      "step": 9456,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.6937828443555857,
      "lm_loss": 0.398,
      "step": 9456,
      "vm_loss": 0.0967
    },
    {
      "epoch": 0.6937828443555857,
      "lm_loss": 0.7277,
      "step": 9456,
      "vm_loss": 0.187
    },
    {
      "epoch": 0.6937828443555857,
      "lm_loss": 0.3198,
      "step": 9456,
      "vm_loss": 0.1189
    },
    {
      "epoch": 0.6937828443555857,
      "lm_loss": 0.1452,
      "step": 9456,
      "vm_loss": 0.1365
    },
    {
      "epoch": 0.6938562139457248,
      "grad_norm": 1.475132143192354,
      "learning_rate": 2.262537400142357e-06,
      "loss": 0.8234,
      "step": 9457
    },
    {
      "epoch": 0.693929583535864,
      "grad_norm": 1.3199965470604917,
      "learning_rate": 2.2615431816989162e-06,
      "loss": 0.7619,
      "step": 9458
    },
    {
      "epoch": 0.6940029531260031,
      "grad_norm": 0.5436479392236779,
      "learning_rate": 2.2605491179029142e-06,
      "loss": 0.4748,
      "step": 9459
    },
    {
      "epoch": 0.6940763227161423,
      "grad_norm": 0.5686894367980665,
      "learning_rate": 2.2595552088104857e-06,
      "loss": 0.469,
      "step": 9460
    },
    {
      "epoch": 0.6941496923062813,
      "grad_norm": 0.5542403639973227,
      "learning_rate": 2.258561454477762e-06,
      "loss": 0.4822,
      "step": 9461
    },
    {
      "epoch": 0.6942230618964205,
      "grad_norm": 1.4135084491453307,
      "learning_rate": 2.257567854960863e-06,
      "loss": 0.7704,
      "step": 9462
    },
    {
      "epoch": 0.6942964314865596,
      "grad_norm": 1.485046425245632,
      "learning_rate": 2.2565744103158987e-06,
      "loss": 0.7344,
      "step": 9463
    },
    {
      "epoch": 0.6943698010766988,
      "grad_norm": 1.4327280910594864,
      "learning_rate": 2.2555811205989715e-06,
      "loss": 0.6702,
      "step": 9464
    },
    {
      "epoch": 0.6943698010766988,
      "lm_loss": 1.3534,
      "step": 9464,
      "vm_loss": 0.0593
    },
    {
      "epoch": 0.6943698010766988,
      "lm_loss": 0.1154,
      "step": 9464,
      "vm_loss": 0.105
    },
    {
      "epoch": 0.6943698010766988,
      "lm_loss": 0.7707,
      "step": 9464,
      "vm_loss": 0.1723
    },
    {
      "epoch": 0.6943698010766988,
      "lm_loss": 1.7287,
      "step": 9464,
      "vm_loss": 0.2138
    },
    {
      "epoch": 0.6943698010766988,
      "lm_loss": 0.1084,
      "step": 9464,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.6943698010766988,
      "lm_loss": 1.1265,
      "step": 9464,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.6943698010766988,
      "lm_loss": 0.7302,
      "step": 9464,
      "vm_loss": 0.0607
    },
    {
      "epoch": 0.6943698010766988,
      "lm_loss": 1.3497,
      "step": 9464,
      "vm_loss": 0.155
    },
    {
      "epoch": 0.6944431706668379,
      "grad_norm": 1.6599852478522155,
      "learning_rate": 2.2545879858661724e-06,
      "loss": 0.759,
      "step": 9465
    },
    {
      "epoch": 0.694516540256977,
      "grad_norm": 1.4203019858001888,
      "learning_rate": 2.2535950061735907e-06,
      "loss": 0.7607,
      "step": 9466
    },
    {
      "epoch": 0.6945899098471161,
      "grad_norm": 1.4783128295239947,
      "learning_rate": 2.252602181577301e-06,
      "loss": 0.7662,
      "step": 9467
    },
    {
      "epoch": 0.6946632794372553,
      "grad_norm": 1.5739287147152914,
      "learning_rate": 2.251609512133369e-06,
      "loss": 0.6794,
      "step": 9468
    },
    {
      "epoch": 0.6947366490273944,
      "grad_norm": 1.7006055828164994,
      "learning_rate": 2.2506169978978543e-06,
      "loss": 0.7744,
      "step": 9469
    },
    {
      "epoch": 0.6948100186175334,
      "grad_norm": 1.4270053369137121,
      "learning_rate": 2.249624638926805e-06,
      "loss": 0.7052,
      "step": 9470
    },
    {
      "epoch": 0.6948833882076726,
      "grad_norm": 0.5450830272511328,
      "learning_rate": 2.2486324352762653e-06,
      "loss": 0.4959,
      "step": 9471
    },
    {
      "epoch": 0.6949567577978117,
      "grad_norm": 0.5469908849570904,
      "learning_rate": 2.2476403870022658e-06,
      "loss": 0.4771,
      "step": 9472
    },
    {
      "epoch": 0.6949567577978117,
      "lm_loss": 1.4991,
      "step": 9472,
      "vm_loss": 0.045
    },
    {
      "epoch": 0.6949567577978117,
      "lm_loss": 0.7384,
      "step": 9472,
      "vm_loss": 0.1517
    },
    {
      "epoch": 0.6949567577978117,
      "lm_loss": 0.4741,
      "step": 9472,
      "vm_loss": 0.0938
    },
    {
      "epoch": 0.6949567577978117,
      "lm_loss": 0.875,
      "step": 9472,
      "vm_loss": 0.1641
    },
    {
      "epoch": 0.6949567577978117,
      "lm_loss": 0.7443,
      "step": 9472,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.6949567577978117,
      "lm_loss": 0.1445,
      "step": 9472,
      "vm_loss": 0.09
    },
    {
      "epoch": 0.6949567577978117,
      "lm_loss": 0.2777,
      "step": 9472,
      "vm_loss": 0.1692
    },
    {
      "epoch": 0.6949567577978117,
      "lm_loss": 0.8415,
      "step": 9472,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.6950301273879509,
      "grad_norm": 1.5126263140688667,
      "learning_rate": 2.2466484941608293e-06,
      "loss": 0.6344,
      "step": 9473
    },
    {
      "epoch": 0.69510349697809,
      "grad_norm": 1.58107163891612,
      "learning_rate": 2.2456567568079714e-06,
      "loss": 0.7216,
      "step": 9474
    },
    {
      "epoch": 0.6951768665682291,
      "grad_norm": 0.5381655085787529,
      "learning_rate": 2.2446651749996966e-06,
      "loss": 0.4873,
      "step": 9475
    },
    {
      "epoch": 0.6952502361583682,
      "grad_norm": 1.517991281720871,
      "learning_rate": 2.243673748792002e-06,
      "loss": 0.738,
      "step": 9476
    },
    {
      "epoch": 0.6953236057485074,
      "grad_norm": 0.5584618048847524,
      "learning_rate": 2.242682478240878e-06,
      "loss": 0.4831,
      "step": 9477
    },
    {
      "epoch": 0.6953969753386465,
      "grad_norm": 1.6650034338556001,
      "learning_rate": 2.241691363402303e-06,
      "loss": 0.7416,
      "step": 9478
    },
    {
      "epoch": 0.6954703449287857,
      "grad_norm": 0.5486168286110437,
      "learning_rate": 2.240700404332247e-06,
      "loss": 0.4821,
      "step": 9479
    },
    {
      "epoch": 0.6955437145189247,
      "grad_norm": 1.607329334338195,
      "learning_rate": 2.239709601086673e-06,
      "loss": 0.7241,
      "step": 9480
    },
    {
      "epoch": 0.6955437145189247,
      "lm_loss": 1.1924,
      "step": 9480,
      "vm_loss": 0.1551
    },
    {
      "epoch": 0.6955437145189247,
      "lm_loss": 1.76,
      "step": 9480,
      "vm_loss": 0.1109
    },
    {
      "epoch": 0.6955437145189247,
      "lm_loss": 0.8725,
      "step": 9480,
      "vm_loss": 0.2256
    },
    {
      "epoch": 0.6955437145189247,
      "lm_loss": 1.0679,
      "step": 9480,
      "vm_loss": 0.2582
    },
    {
      "epoch": 0.6955437145189247,
      "lm_loss": 0.7054,
      "step": 9480,
      "vm_loss": 0.1083
    },
    {
      "epoch": 0.6955437145189247,
      "lm_loss": 0.5145,
      "step": 9480,
      "vm_loss": 0.0919
    },
    {
      "epoch": 0.6955437145189247,
      "lm_loss": 0.521,
      "step": 9480,
      "vm_loss": 0.0626
    },
    {
      "epoch": 0.6955437145189247,
      "lm_loss": 0.5784,
      "step": 9480,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.6956170841090639,
      "grad_norm": 1.4223569959170226,
      "learning_rate": 2.238718953721532e-06,
      "loss": 0.754,
      "step": 9481
    },
    {
      "epoch": 0.695690453699203,
      "grad_norm": 0.5489637222457671,
      "learning_rate": 2.2377284622927713e-06,
      "loss": 0.4726,
      "step": 9482
    },
    {
      "epoch": 0.6957638232893422,
      "grad_norm": 1.5968250609973043,
      "learning_rate": 2.2367381268563255e-06,
      "loss": 0.6906,
      "step": 9483
    },
    {
      "epoch": 0.6958371928794813,
      "grad_norm": 1.419941528704186,
      "learning_rate": 2.23574794746812e-06,
      "loss": 0.8317,
      "step": 9484
    },
    {
      "epoch": 0.6959105624696204,
      "grad_norm": 0.5581562786538595,
      "learning_rate": 2.234757924184074e-06,
      "loss": 0.4688,
      "step": 9485
    },
    {
      "epoch": 0.6959839320597595,
      "grad_norm": 1.5211601559713381,
      "learning_rate": 2.2337680570600955e-06,
      "loss": 0.7739,
      "step": 9486
    },
    {
      "epoch": 0.6960573016498987,
      "grad_norm": 0.560966084519363,
      "learning_rate": 2.232778346152084e-06,
      "loss": 0.4722,
      "step": 9487
    },
    {
      "epoch": 0.6961306712400378,
      "grad_norm": 1.5887081103060383,
      "learning_rate": 2.2317887915159338e-06,
      "loss": 0.7491,
      "step": 9488
    },
    {
      "epoch": 0.6961306712400378,
      "lm_loss": 0.4458,
      "step": 9488,
      "vm_loss": 0.0358
    },
    {
      "epoch": 0.6961306712400378,
      "lm_loss": 0.4484,
      "step": 9488,
      "vm_loss": 0.0503
    },
    {
      "epoch": 0.6961306712400378,
      "lm_loss": 1.0529,
      "step": 9488,
      "vm_loss": 0.1243
    },
    {
      "epoch": 0.6961306712400378,
      "lm_loss": 0.7094,
      "step": 9488,
      "vm_loss": 0.1709
    },
    {
      "epoch": 0.6961306712400378,
      "lm_loss": 0.5965,
      "step": 9488,
      "vm_loss": 0.1245
    },
    {
      "epoch": 0.6961306712400378,
      "lm_loss": 0.1663,
      "step": 9488,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.6961306712400378,
      "lm_loss": 0.8398,
      "step": 9488,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.6961306712400378,
      "lm_loss": 0.7816,
      "step": 9488,
      "vm_loss": 0.1452
    },
    {
      "epoch": 0.696204040830177,
      "grad_norm": 1.3905565147464811,
      "learning_rate": 2.230799393207526e-06,
      "loss": 0.7275,
      "step": 9489
    },
    {
      "epoch": 0.696277410420316,
      "grad_norm": 1.4274809525651395,
      "learning_rate": 2.229810151282734e-06,
      "loss": 0.6056,
      "step": 9490
    },
    {
      "epoch": 0.6963507800104551,
      "grad_norm": 1.5487205695392856,
      "learning_rate": 2.2288210657974237e-06,
      "loss": 0.7192,
      "step": 9491
    },
    {
      "epoch": 0.6964241496005943,
      "grad_norm": 0.5530978344879303,
      "learning_rate": 2.227832136807448e-06,
      "loss": 0.4718,
      "step": 9492
    },
    {
      "epoch": 0.6964975191907334,
      "grad_norm": 0.5444945246357153,
      "learning_rate": 2.2268433643686586e-06,
      "loss": 0.4841,
      "step": 9493
    },
    {
      "epoch": 0.6965708887808726,
      "grad_norm": 1.318452608496139,
      "learning_rate": 2.2258547485368926e-06,
      "loss": 0.7005,
      "step": 9494
    },
    {
      "epoch": 0.6966442583710116,
      "grad_norm": 0.5712549352663742,
      "learning_rate": 2.2248662893679783e-06,
      "loss": 0.493,
      "step": 9495
    },
    {
      "epoch": 0.6967176279611508,
      "grad_norm": 1.7236437956519353,
      "learning_rate": 2.2238779869177373e-06,
      "loss": 0.7512,
      "step": 9496
    },
    {
      "epoch": 0.6967176279611508,
      "lm_loss": 1.2704,
      "step": 9496,
      "vm_loss": 0.0387
    },
    {
      "epoch": 0.6967176279611508,
      "lm_loss": 0.0456,
      "step": 9496,
      "vm_loss": 0.1836
    },
    {
      "epoch": 0.6967176279611508,
      "lm_loss": 0.9742,
      "step": 9496,
      "vm_loss": 0.1391
    },
    {
      "epoch": 0.6967176279611508,
      "lm_loss": 0.1769,
      "step": 9496,
      "vm_loss": 0.1115
    },
    {
      "epoch": 0.6967176279611508,
      "lm_loss": 1.1333,
      "step": 9496,
      "vm_loss": 0.1853
    },
    {
      "epoch": 0.6967176279611508,
      "lm_loss": 0.3156,
      "step": 9496,
      "vm_loss": 0.0576
    },
    {
      "epoch": 0.6967176279611508,
      "lm_loss": 1.443,
      "step": 9496,
      "vm_loss": 0.0986
    },
    {
      "epoch": 0.6967176279611508,
      "lm_loss": 0.1866,
      "step": 9496,
      "vm_loss": 0.0798
    },
    {
      "epoch": 0.6967909975512899,
      "grad_norm": 1.53376724649885,
      "learning_rate": 2.2228898412419798e-06,
      "loss": 0.7805,
      "step": 9497
    },
    {
      "epoch": 0.6968643671414291,
      "grad_norm": 0.5231362865702409,
      "learning_rate": 2.2219018523965137e-06,
      "loss": 0.4856,
      "step": 9498
    },
    {
      "epoch": 0.6969377367315682,
      "grad_norm": 0.5489577931043123,
      "learning_rate": 2.220914020437128e-06,
      "loss": 0.4633,
      "step": 9499
    },
    {
      "epoch": 0.6970111063217073,
      "grad_norm": 0.5538689890470255,
      "learning_rate": 2.2199263454196103e-06,
      "loss": 0.4884,
      "step": 9500
    },
    {
      "epoch": 0.6970844759118464,
      "grad_norm": 1.4437854442484517,
      "learning_rate": 2.2189388273997364e-06,
      "loss": 0.7287,
      "step": 9501
    },
    {
      "epoch": 0.6971578455019856,
      "grad_norm": 1.3349756238337973,
      "learning_rate": 2.217951466433272e-06,
      "loss": 0.7332,
      "step": 9502
    },
    {
      "epoch": 0.6972312150921247,
      "grad_norm": 0.5494313311276644,
      "learning_rate": 2.21696426257598e-06,
      "loss": 0.4764,
      "step": 9503
    },
    {
      "epoch": 0.6973045846822639,
      "grad_norm": 1.717774096517284,
      "learning_rate": 2.215977215883608e-06,
      "loss": 0.857,
      "step": 9504
    },
    {
      "epoch": 0.6973045846822639,
      "lm_loss": 0.5126,
      "step": 9504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6973045846822639,
      "lm_loss": 0.4955,
      "step": 9504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6973045846822639,
      "lm_loss": 0.4095,
      "step": 9504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6973045846822639,
      "lm_loss": 0.5353,
      "step": 9504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6973045846822639,
      "lm_loss": 0.4474,
      "step": 9504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6973045846822639,
      "lm_loss": 0.5389,
      "step": 9504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6973045846822639,
      "lm_loss": 0.4551,
      "step": 9504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6973045846822639,
      "lm_loss": 0.4713,
      "step": 9504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6973779542724029,
      "grad_norm": 0.5599942578760617,
      "learning_rate": 2.214990326411897e-06,
      "loss": 0.488,
      "step": 9505
    },
    {
      "epoch": 0.697451323862542,
      "grad_norm": 1.389202568853785,
      "learning_rate": 2.21400359421658e-06,
      "loss": 0.727,
      "step": 9506
    },
    {
      "epoch": 0.6975246934526812,
      "grad_norm": 0.5381426788446921,
      "learning_rate": 2.213017019353377e-06,
      "loss": 0.4895,
      "step": 9507
    },
    {
      "epoch": 0.6975980630428203,
      "grad_norm": 0.5619618772049749,
      "learning_rate": 2.2120306018780065e-06,
      "loss": 0.5002,
      "step": 9508
    },
    {
      "epoch": 0.6976714326329595,
      "grad_norm": 1.3586899901496339,
      "learning_rate": 2.2110443418461723e-06,
      "loss": 0.7075,
      "step": 9509
    },
    {
      "epoch": 0.6977448022230985,
      "grad_norm": 1.5261108611274476,
      "learning_rate": 2.2100582393135712e-06,
      "loss": 0.9215,
      "step": 9510
    },
    {
      "epoch": 0.6978181718132377,
      "grad_norm": 1.4115623148329546,
      "learning_rate": 2.20907229433589e-06,
      "loss": 0.729,
      "step": 9511
    },
    {
      "epoch": 0.6978915414033768,
      "grad_norm": 1.5827741484759257,
      "learning_rate": 2.2080865069688084e-06,
      "loss": 0.7761,
      "step": 9512
    },
    {
      "epoch": 0.6978915414033768,
      "lm_loss": 1.0901,
      "step": 9512,
      "vm_loss": 0.0599
    },
    {
      "epoch": 0.6978915414033768,
      "lm_loss": 0.9911,
      "step": 9512,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.6978915414033768,
      "lm_loss": 0.7773,
      "step": 9512,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.6978915414033768,
      "lm_loss": 0.0694,
      "step": 9512,
      "vm_loss": 0.155
    },
    {
      "epoch": 0.6978915414033768,
      "lm_loss": 0.9722,
      "step": 9512,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.6978915414033768,
      "lm_loss": 0.3282,
      "step": 9512,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.6978915414033768,
      "lm_loss": 0.5699,
      "step": 9512,
      "vm_loss": 0.1426
    },
    {
      "epoch": 0.6978915414033768,
      "lm_loss": 0.3747,
      "step": 9512,
      "vm_loss": 0.1518
    },
    {
      "epoch": 0.697964910993516,
      "grad_norm": 1.4513151083775626,
      "learning_rate": 2.2071008772679937e-06,
      "loss": 0.7175,
      "step": 9513
    },
    {
      "epoch": 0.6980382805836551,
      "grad_norm": 1.4343067530623923,
      "learning_rate": 2.2061154052891104e-06,
      "loss": 0.6094,
      "step": 9514
    },
    {
      "epoch": 0.6981116501737942,
      "grad_norm": 1.508746227825454,
      "learning_rate": 2.2051300910878096e-06,
      "loss": 0.7998,
      "step": 9515
    },
    {
      "epoch": 0.6981850197639333,
      "grad_norm": 1.5281960338665894,
      "learning_rate": 2.204144934719734e-06,
      "loss": 0.7169,
      "step": 9516
    },
    {
      "epoch": 0.6982583893540725,
      "grad_norm": 1.5300459576258854,
      "learning_rate": 2.2031599362405174e-06,
      "loss": 0.8193,
      "step": 9517
    },
    {
      "epoch": 0.6983317589442116,
      "grad_norm": 1.4838136473048253,
      "learning_rate": 2.202175095705784e-06,
      "loss": 0.8517,
      "step": 9518
    },
    {
      "epoch": 0.6984051285343508,
      "grad_norm": 1.4705560406465163,
      "learning_rate": 2.2011904131711536e-06,
      "loss": 0.6145,
      "step": 9519
    },
    {
      "epoch": 0.6984784981244898,
      "grad_norm": 1.63167954161804,
      "learning_rate": 2.200205888692231e-06,
      "loss": 0.8122,
      "step": 9520
    },
    {
      "epoch": 0.6984784981244898,
      "lm_loss": 0.9674,
      "step": 9520,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.6984784981244898,
      "lm_loss": 2.4124,
      "step": 9520,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.6984784981244898,
      "lm_loss": 1.1484,
      "step": 9520,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.6984784981244898,
      "lm_loss": 0.8638,
      "step": 9520,
      "vm_loss": 0.0745
    },
    {
      "epoch": 0.6984784981244898,
      "lm_loss": 0.5153,
      "step": 9520,
      "vm_loss": 0.1033
    },
    {
      "epoch": 0.6984784981244898,
      "lm_loss": 0.4305,
      "step": 9520,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.6984784981244898,
      "lm_loss": 0.6483,
      "step": 9520,
      "vm_loss": 0.1409
    },
    {
      "epoch": 0.6984784981244898,
      "lm_loss": 0.044,
      "step": 9520,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.698551867714629,
      "grad_norm": 1.4049008544548083,
      "learning_rate": 2.199221522324616e-06,
      "loss": 0.8069,
      "step": 9521
    },
    {
      "epoch": 0.6986252373047681,
      "grad_norm": 1.5225493219159911,
      "learning_rate": 2.1982373141238976e-06,
      "loss": 0.7096,
      "step": 9522
    },
    {
      "epoch": 0.6986986068949073,
      "grad_norm": 0.5474977808440136,
      "learning_rate": 2.1972532641456545e-06,
      "loss": 0.4785,
      "step": 9523
    },
    {
      "epoch": 0.6987719764850464,
      "grad_norm": 0.5360293861640706,
      "learning_rate": 2.1962693724454626e-06,
      "loss": 0.465,
      "step": 9524
    },
    {
      "epoch": 0.6988453460751854,
      "grad_norm": 1.5866806934834632,
      "learning_rate": 2.1952856390788826e-06,
      "loss": 0.8289,
      "step": 9525
    },
    {
      "epoch": 0.6989187156653246,
      "grad_norm": 1.5078974619219383,
      "learning_rate": 2.194302064101468e-06,
      "loss": 0.7253,
      "step": 9526
    },
    {
      "epoch": 0.6989920852554637,
      "grad_norm": 1.3486386518858884,
      "learning_rate": 2.193318647568764e-06,
      "loss": 0.7216,
      "step": 9527
    },
    {
      "epoch": 0.6990654548456029,
      "grad_norm": 1.548160495241336,
      "learning_rate": 2.1923353895363062e-06,
      "loss": 0.7131,
      "step": 9528
    },
    {
      "epoch": 0.6990654548456029,
      "lm_loss": 0.5183,
      "step": 9528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6990654548456029,
      "lm_loss": 0.398,
      "step": 9528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6990654548456029,
      "lm_loss": 0.4324,
      "step": 9528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6990654548456029,
      "lm_loss": 0.3793,
      "step": 9528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6990654548456029,
      "lm_loss": 0.5712,
      "step": 9528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6990654548456029,
      "lm_loss": 0.4569,
      "step": 9528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6990654548456029,
      "lm_loss": 0.5186,
      "step": 9528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.6990654548456029,
      "lm_loss": 0.5284,
      "step": 9528,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.699138824435742,
      "grad_norm": 0.5580476120230318,
      "learning_rate": 2.191352290059621e-06,
      "loss": 0.4788,
      "step": 9529
    },
    {
      "epoch": 0.6992121940258811,
      "grad_norm": 1.5090879641845514,
      "learning_rate": 2.1903693491942292e-06,
      "loss": 0.7359,
      "step": 9530
    },
    {
      "epoch": 0.6992855636160202,
      "grad_norm": 0.5658125252203493,
      "learning_rate": 2.1893865669956376e-06,
      "loss": 0.4731,
      "step": 9531
    },
    {
      "epoch": 0.6993589332061594,
      "grad_norm": 1.4849570691302934,
      "learning_rate": 2.1884039435193478e-06,
      "loss": 0.6577,
      "step": 9532
    },
    {
      "epoch": 0.6994323027962985,
      "grad_norm": 0.5563315839279367,
      "learning_rate": 2.1874214788208492e-06,
      "loss": 0.4604,
      "step": 9533
    },
    {
      "epoch": 0.6995056723864377,
      "grad_norm": 0.5621783025207032,
      "learning_rate": 2.186439172955624e-06,
      "loss": 0.504,
      "step": 9534
    },
    {
      "epoch": 0.6995790419765767,
      "grad_norm": 0.5828560852964745,
      "learning_rate": 2.1854570259791476e-06,
      "loss": 0.4854,
      "step": 9535
    },
    {
      "epoch": 0.6996524115667159,
      "grad_norm": 1.859553254188654,
      "learning_rate": 2.1844750379468833e-06,
      "loss": 0.7461,
      "step": 9536
    },
    {
      "epoch": 0.6996524115667159,
      "lm_loss": 0.9988,
      "step": 9536,
      "vm_loss": 0.1038
    },
    {
      "epoch": 0.6996524115667159,
      "lm_loss": 1.3457,
      "step": 9536,
      "vm_loss": 0.0948
    },
    {
      "epoch": 0.6996524115667159,
      "lm_loss": 0.1299,
      "step": 9536,
      "vm_loss": 0.1629
    },
    {
      "epoch": 0.6996524115667159,
      "lm_loss": 0.9599,
      "step": 9536,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.6996524115667159,
      "lm_loss": 0.3799,
      "step": 9536,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.6996524115667159,
      "lm_loss": 0.3337,
      "step": 9536,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.6996524115667159,
      "lm_loss": 1.3221,
      "step": 9536,
      "vm_loss": 0.2763
    },
    {
      "epoch": 0.6996524115667159,
      "lm_loss": 0.1509,
      "step": 9536,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.699725781156855,
      "grad_norm": 1.4000090745267897,
      "learning_rate": 2.1834932089142855e-06,
      "loss": 0.7213,
      "step": 9537
    },
    {
      "epoch": 0.6997991507469942,
      "grad_norm": 0.5495840104998827,
      "learning_rate": 2.182511538936802e-06,
      "loss": 0.4863,
      "step": 9538
    },
    {
      "epoch": 0.6998725203371333,
      "grad_norm": 1.3397747973931518,
      "learning_rate": 2.1815300280698686e-06,
      "loss": 0.6821,
      "step": 9539
    },
    {
      "epoch": 0.6999458899272724,
      "grad_norm": 1.5113642961619915,
      "learning_rate": 2.180548676368913e-06,
      "loss": 0.7937,
      "step": 9540
    },
    {
      "epoch": 0.7000192595174115,
      "grad_norm": 1.5744441494728896,
      "learning_rate": 2.1795674838893577e-06,
      "loss": 0.76,
      "step": 9541
    },
    {
      "epoch": 0.7000926291075507,
      "grad_norm": 0.5340516806019233,
      "learning_rate": 2.1785864506866112e-06,
      "loss": 0.4706,
      "step": 9542
    },
    {
      "epoch": 0.7001659986976898,
      "grad_norm": 0.5575804001570648,
      "learning_rate": 2.1776055768160748e-06,
      "loss": 0.4856,
      "step": 9543
    },
    {
      "epoch": 0.700239368287829,
      "grad_norm": 0.5402089658500577,
      "learning_rate": 2.176624862333141e-06,
      "loss": 0.4779,
      "step": 9544
    },
    {
      "epoch": 0.700239368287829,
      "lm_loss": 1.3604,
      "step": 9544,
      "vm_loss": 0.0189
    },
    {
      "epoch": 0.700239368287829,
      "lm_loss": 0.8103,
      "step": 9544,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.700239368287829,
      "lm_loss": 0.7132,
      "step": 9544,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.700239368287829,
      "lm_loss": 0.9731,
      "step": 9544,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.700239368287829,
      "lm_loss": 0.0649,
      "step": 9544,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.700239368287829,
      "lm_loss": 0.1175,
      "step": 9544,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.700239368287829,
      "lm_loss": 0.3421,
      "step": 9544,
      "vm_loss": 0.0936
    },
    {
      "epoch": 0.700239368287829,
      "lm_loss": 0.5539,
      "step": 9544,
      "vm_loss": 0.1288
    },
    {
      "epoch": 0.700312737877968,
      "grad_norm": 1.5520065072562932,
      "learning_rate": 2.1756443072931914e-06,
      "loss": 0.733,
      "step": 9545
    },
    {
      "epoch": 0.7003861074681071,
      "grad_norm": 1.3509610549142492,
      "learning_rate": 2.1746639117516043e-06,
      "loss": 0.6458,
      "step": 9546
    },
    {
      "epoch": 0.7004594770582463,
      "grad_norm": 1.4878653828692117,
      "learning_rate": 2.1736836757637423e-06,
      "loss": 0.6978,
      "step": 9547
    },
    {
      "epoch": 0.7005328466483854,
      "grad_norm": 1.6473189949300535,
      "learning_rate": 2.1727035993849633e-06,
      "loss": 0.7682,
      "step": 9548
    },
    {
      "epoch": 0.7006062162385246,
      "grad_norm": 1.4138495534581086,
      "learning_rate": 2.171723682670613e-06,
      "loss": 0.7386,
      "step": 9549
    },
    {
      "epoch": 0.7006795858286636,
      "grad_norm": 0.5646904103405255,
      "learning_rate": 2.1707439256760287e-06,
      "loss": 0.4741,
      "step": 9550
    },
    {
      "epoch": 0.7007529554188028,
      "grad_norm": 1.629299265945226,
      "learning_rate": 2.1697643284565433e-06,
      "loss": 0.7367,
      "step": 9551
    },
    {
      "epoch": 0.7008263250089419,
      "grad_norm": 1.5003855163293243,
      "learning_rate": 2.1687848910674753e-06,
      "loss": 0.7009,
      "step": 9552
    },
    {
      "epoch": 0.7008263250089419,
      "lm_loss": 1.1855,
      "step": 9552,
      "vm_loss": 0.081
    },
    {
      "epoch": 0.7008263250089419,
      "lm_loss": 0.3546,
      "step": 9552,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.7008263250089419,
      "lm_loss": 0.2114,
      "step": 9552,
      "vm_loss": 0.166
    },
    {
      "epoch": 0.7008263250089419,
      "lm_loss": 0.0102,
      "step": 9552,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.7008263250089419,
      "lm_loss": 0.221,
      "step": 9552,
      "vm_loss": 0.1251
    },
    {
      "epoch": 0.7008263250089419,
      "lm_loss": 1.6584,
      "step": 9552,
      "vm_loss": 0.137
    },
    {
      "epoch": 0.7008263250089419,
      "lm_loss": 0.527,
      "step": 9552,
      "vm_loss": 0.0554
    },
    {
      "epoch": 0.7008263250089419,
      "lm_loss": 0.5462,
      "step": 9552,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.7008996945990811,
      "grad_norm": 1.6374432450027194,
      "learning_rate": 2.1678056135641364e-06,
      "loss": 0.7127,
      "step": 9553
    },
    {
      "epoch": 0.7009730641892201,
      "grad_norm": 1.3047643822012487,
      "learning_rate": 2.1668264960018276e-06,
      "loss": 0.6847,
      "step": 9554
    },
    {
      "epoch": 0.7010464337793593,
      "grad_norm": 0.5695743290676534,
      "learning_rate": 2.165847538435843e-06,
      "loss": 0.4932,
      "step": 9555
    },
    {
      "epoch": 0.7011198033694984,
      "grad_norm": 0.5577794003764136,
      "learning_rate": 2.1648687409214646e-06,
      "loss": 0.4749,
      "step": 9556
    },
    {
      "epoch": 0.7011931729596376,
      "grad_norm": 1.5622700089439647,
      "learning_rate": 2.1638901035139715e-06,
      "loss": 0.6856,
      "step": 9557
    },
    {
      "epoch": 0.7012665425497767,
      "grad_norm": 1.552865331469774,
      "learning_rate": 2.1629116262686277e-06,
      "loss": 0.6977,
      "step": 9558
    },
    {
      "epoch": 0.7013399121399158,
      "grad_norm": 1.4756452958967847,
      "learning_rate": 2.1619333092406908e-06,
      "loss": 0.7336,
      "step": 9559
    },
    {
      "epoch": 0.7014132817300549,
      "grad_norm": 0.5469140159100121,
      "learning_rate": 2.1609551524854076e-06,
      "loss": 0.4745,
      "step": 9560
    },
    {
      "epoch": 0.7014132817300549,
      "lm_loss": 0.5348,
      "step": 9560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7014132817300549,
      "lm_loss": 0.4873,
      "step": 9560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7014132817300549,
      "lm_loss": 0.4763,
      "step": 9560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7014132817300549,
      "lm_loss": 0.3564,
      "step": 9560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7014132817300549,
      "lm_loss": 0.4492,
      "step": 9560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7014132817300549,
      "lm_loss": 0.4822,
      "step": 9560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7014132817300549,
      "lm_loss": 0.5846,
      "step": 9560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7014132817300549,
      "lm_loss": 0.7151,
      "step": 9560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.701486651320194,
      "grad_norm": 0.5489737075294912,
      "learning_rate": 2.1599771560580165e-06,
      "loss": 0.4763,
      "step": 9561
    },
    {
      "epoch": 0.7015600209103332,
      "grad_norm": 1.4575891871547548,
      "learning_rate": 2.1589993200137495e-06,
      "loss": 0.6978,
      "step": 9562
    },
    {
      "epoch": 0.7016333905004724,
      "grad_norm": 1.489497892665564,
      "learning_rate": 2.1580216444078274e-06,
      "loss": 0.6794,
      "step": 9563
    },
    {
      "epoch": 0.7017067600906114,
      "grad_norm": 1.4844797291202356,
      "learning_rate": 2.15704412929546e-06,
      "loss": 0.8942,
      "step": 9564
    },
    {
      "epoch": 0.7017801296807505,
      "grad_norm": 1.538912837591466,
      "learning_rate": 2.1560667747318525e-06,
      "loss": 0.7181,
      "step": 9565
    },
    {
      "epoch": 0.7018534992708897,
      "grad_norm": 1.5498144209865647,
      "learning_rate": 2.1550895807721944e-06,
      "loss": 0.7419,
      "step": 9566
    },
    {
      "epoch": 0.7019268688610288,
      "grad_norm": 0.541829607814262,
      "learning_rate": 2.154112547471675e-06,
      "loss": 0.4984,
      "step": 9567
    },
    {
      "epoch": 0.702000238451168,
      "grad_norm": 0.5540507344946761,
      "learning_rate": 2.1531356748854677e-06,
      "loss": 0.459,
      "step": 9568
    },
    {
      "epoch": 0.702000238451168,
      "lm_loss": 0.5545,
      "step": 9568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702000238451168,
      "lm_loss": 0.4502,
      "step": 9568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702000238451168,
      "lm_loss": 0.6228,
      "step": 9568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702000238451168,
      "lm_loss": 0.5537,
      "step": 9568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702000238451168,
      "lm_loss": 0.4243,
      "step": 9568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702000238451168,
      "lm_loss": 0.4348,
      "step": 9568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702000238451168,
      "lm_loss": 0.3862,
      "step": 9568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702000238451168,
      "lm_loss": 0.6071,
      "step": 9568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702073608041307,
      "grad_norm": 0.5439511430408932,
      "learning_rate": 2.152158963068739e-06,
      "loss": 0.4777,
      "step": 9569
    },
    {
      "epoch": 0.7021469776314462,
      "grad_norm": 1.3287777010908706,
      "learning_rate": 2.1511824120766467e-06,
      "loss": 0.747,
      "step": 9570
    },
    {
      "epoch": 0.7022203472215853,
      "grad_norm": 1.5970946473037073,
      "learning_rate": 2.150206021964338e-06,
      "loss": 0.7215,
      "step": 9571
    },
    {
      "epoch": 0.7022937168117245,
      "grad_norm": 1.4431969653154844,
      "learning_rate": 2.149229792786951e-06,
      "loss": 0.6719,
      "step": 9572
    },
    {
      "epoch": 0.7023670864018636,
      "grad_norm": 1.4702223820737563,
      "learning_rate": 2.148253724599619e-06,
      "loss": 0.6352,
      "step": 9573
    },
    {
      "epoch": 0.7024404559920027,
      "grad_norm": 1.4769384223464528,
      "learning_rate": 2.1472778174574614e-06,
      "loss": 0.6928,
      "step": 9574
    },
    {
      "epoch": 0.7025138255821418,
      "grad_norm": 1.3913635614824464,
      "learning_rate": 2.14630207141559e-06,
      "loss": 0.7117,
      "step": 9575
    },
    {
      "epoch": 0.702587195172281,
      "grad_norm": 1.4658039178101872,
      "learning_rate": 2.1453264865291086e-06,
      "loss": 0.6996,
      "step": 9576
    },
    {
      "epoch": 0.702587195172281,
      "lm_loss": 0.5781,
      "step": 9576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702587195172281,
      "lm_loss": 0.3395,
      "step": 9576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702587195172281,
      "lm_loss": 0.4405,
      "step": 9576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702587195172281,
      "lm_loss": 0.3894,
      "step": 9576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702587195172281,
      "lm_loss": 0.6575,
      "step": 9576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702587195172281,
      "lm_loss": 0.4832,
      "step": 9576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702587195172281,
      "lm_loss": 0.4416,
      "step": 9576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.702587195172281,
      "lm_loss": 0.3291,
      "step": 9576,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7026605647624201,
      "grad_norm": 0.5529948432376216,
      "learning_rate": 2.1443510628531073e-06,
      "loss": 0.4872,
      "step": 9577
    },
    {
      "epoch": 0.7027339343525593,
      "grad_norm": 1.543702943609996,
      "learning_rate": 2.143375800442675e-06,
      "loss": 0.7542,
      "step": 9578
    },
    {
      "epoch": 0.7028073039426983,
      "grad_norm": 1.4771506058181507,
      "learning_rate": 2.1424006993528868e-06,
      "loss": 0.7834,
      "step": 9579
    },
    {
      "epoch": 0.7028806735328375,
      "grad_norm": 1.5233095030502788,
      "learning_rate": 2.1414257596388067e-06,
      "loss": 0.8304,
      "step": 9580
    },
    {
      "epoch": 0.7029540431229766,
      "grad_norm": 1.4844798638716776,
      "learning_rate": 2.1404509813554934e-06,
      "loss": 0.7057,
      "step": 9581
    },
    {
      "epoch": 0.7030274127131158,
      "grad_norm": 0.5279894338952116,
      "learning_rate": 2.1394763645579946e-06,
      "loss": 0.484,
      "step": 9582
    },
    {
      "epoch": 0.7031007823032549,
      "grad_norm": 0.5612747870998179,
      "learning_rate": 2.138501909301348e-06,
      "loss": 0.4698,
      "step": 9583
    },
    {
      "epoch": 0.7031741518933939,
      "grad_norm": 1.5029649207961342,
      "learning_rate": 2.1375276156405866e-06,
      "loss": 0.6139,
      "step": 9584
    },
    {
      "epoch": 0.7031741518933939,
      "lm_loss": 1.0595,
      "step": 9584,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.7031741518933939,
      "lm_loss": 0.2811,
      "step": 9584,
      "vm_loss": 0.022
    },
    {
      "epoch": 0.7031741518933939,
      "lm_loss": 0.3598,
      "step": 9584,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.7031741518933939,
      "lm_loss": 0.7802,
      "step": 9584,
      "vm_loss": 0.2385
    },
    {
      "epoch": 0.7031741518933939,
      "lm_loss": 0.817,
      "step": 9584,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.7031741518933939,
      "lm_loss": 0.3756,
      "step": 9584,
      "vm_loss": 0.0589
    },
    {
      "epoch": 0.7031741518933939,
      "lm_loss": 0.3087,
      "step": 9584,
      "vm_loss": 0.2541
    },
    {
      "epoch": 0.7031741518933939,
      "lm_loss": 0.8219,
      "step": 9584,
      "vm_loss": 0.0553
    },
    {
      "epoch": 0.7032475214835331,
      "grad_norm": 1.6369888036792162,
      "learning_rate": 2.1365534836307295e-06,
      "loss": 0.7256,
      "step": 9585
    },
    {
      "epoch": 0.7033208910736722,
      "grad_norm": 0.5415306605211117,
      "learning_rate": 2.1355795133267883e-06,
      "loss": 0.4783,
      "step": 9586
    },
    {
      "epoch": 0.7033942606638114,
      "grad_norm": 1.497872571370818,
      "learning_rate": 2.1346057047837657e-06,
      "loss": 0.7646,
      "step": 9587
    },
    {
      "epoch": 0.7034676302539505,
      "grad_norm": 1.3267460530830875,
      "learning_rate": 2.133632058056653e-06,
      "loss": 0.8231,
      "step": 9588
    },
    {
      "epoch": 0.7035409998440896,
      "grad_norm": 1.6913938923348595,
      "learning_rate": 2.1326585732004384e-06,
      "loss": 0.7945,
      "step": 9589
    },
    {
      "epoch": 0.7036143694342287,
      "grad_norm": 0.5454680981217427,
      "learning_rate": 2.131685250270095e-06,
      "loss": 0.4822,
      "step": 9590
    },
    {
      "epoch": 0.7036877390243679,
      "grad_norm": 1.6056306103033535,
      "learning_rate": 2.130712089320589e-06,
      "loss": 0.7936,
      "step": 9591
    },
    {
      "epoch": 0.703761108614507,
      "grad_norm": 1.571726531379578,
      "learning_rate": 2.129739090406877e-06,
      "loss": 0.7994,
      "step": 9592
    },
    {
      "epoch": 0.703761108614507,
      "lm_loss": 1.3752,
      "step": 9592,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.703761108614507,
      "lm_loss": 0.2632,
      "step": 9592,
      "vm_loss": 0.0625
    },
    {
      "epoch": 0.703761108614507,
      "lm_loss": 0.8502,
      "step": 9592,
      "vm_loss": 0.0583
    },
    {
      "epoch": 0.703761108614507,
      "lm_loss": 0.1203,
      "step": 9592,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.703761108614507,
      "lm_loss": 1.1647,
      "step": 9592,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.703761108614507,
      "lm_loss": 1.1788,
      "step": 9592,
      "vm_loss": 0.1188
    },
    {
      "epoch": 0.703761108614507,
      "lm_loss": 0.2681,
      "step": 9592,
      "vm_loss": 0.0602
    },
    {
      "epoch": 0.703761108614507,
      "lm_loss": 1.0781,
      "step": 9592,
      "vm_loss": 0.1603
    },
    {
      "epoch": 0.7038344782046462,
      "grad_norm": 1.5935067256401203,
      "learning_rate": 2.1287662535839044e-06,
      "loss": 0.7299,
      "step": 9593
    },
    {
      "epoch": 0.7039078477947852,
      "grad_norm": 1.7310890543139554,
      "learning_rate": 2.127793578906614e-06,
      "loss": 0.7723,
      "step": 9594
    },
    {
      "epoch": 0.7039812173849244,
      "grad_norm": 1.5484945812724351,
      "learning_rate": 2.1268210664299333e-06,
      "loss": 0.793,
      "step": 9595
    },
    {
      "epoch": 0.7040545869750635,
      "grad_norm": 1.5409497789464566,
      "learning_rate": 2.125848716208782e-06,
      "loss": 0.7815,
      "step": 9596
    },
    {
      "epoch": 0.7041279565652027,
      "grad_norm": 0.565315313756044,
      "learning_rate": 2.1248765282980715e-06,
      "loss": 0.4907,
      "step": 9597
    },
    {
      "epoch": 0.7042013261553418,
      "grad_norm": 1.6658466394481062,
      "learning_rate": 2.1239045027527033e-06,
      "loss": 0.7644,
      "step": 9598
    },
    {
      "epoch": 0.7042746957454808,
      "grad_norm": 1.6107725604817305,
      "learning_rate": 2.122932639627569e-06,
      "loss": 0.6659,
      "step": 9599
    },
    {
      "epoch": 0.70434806533562,
      "grad_norm": 1.477910788597474,
      "learning_rate": 2.1219609389775554e-06,
      "loss": 0.7373,
      "step": 9600
    },
    {
      "epoch": 0.70434806533562,
      "lm_loss": 1.6207,
      "step": 9600,
      "vm_loss": 0.1933
    },
    {
      "epoch": 0.70434806533562,
      "lm_loss": 1.4047,
      "step": 9600,
      "vm_loss": 0.0756
    },
    {
      "epoch": 0.70434806533562,
      "lm_loss": 0.0222,
      "step": 9600,
      "vm_loss": 0.1856
    },
    {
      "epoch": 0.70434806533562,
      "lm_loss": 0.755,
      "step": 9600,
      "vm_loss": 0.1817
    },
    {
      "epoch": 0.70434806533562,
      "lm_loss": 0.6177,
      "step": 9600,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.70434806533562,
      "lm_loss": 0.8501,
      "step": 9600,
      "vm_loss": 0.1801
    },
    {
      "epoch": 0.70434806533562,
      "lm_loss": 0.1387,
      "step": 9600,
      "vm_loss": 0.1565
    },
    {
      "epoch": 0.70434806533562,
      "lm_loss": 0.5245,
      "step": 9600,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.7044214349257591,
      "grad_norm": 1.4805078663242657,
      "learning_rate": 2.1209894008575342e-06,
      "loss": 0.7945,
      "step": 9601
    },
    {
      "epoch": 0.7044948045158983,
      "grad_norm": 1.6854848602164778,
      "learning_rate": 2.120018025322372e-06,
      "loss": 0.8107,
      "step": 9602
    },
    {
      "epoch": 0.7045681741060374,
      "grad_norm": 1.4303186243290877,
      "learning_rate": 2.1190468124269236e-06,
      "loss": 0.663,
      "step": 9603
    },
    {
      "epoch": 0.7046415436961765,
      "grad_norm": 1.490482681307804,
      "learning_rate": 2.1180757622260346e-06,
      "loss": 0.7177,
      "step": 9604
    },
    {
      "epoch": 0.7047149132863156,
      "grad_norm": 1.6525101532398774,
      "learning_rate": 2.117104874774546e-06,
      "loss": 0.7753,
      "step": 9605
    },
    {
      "epoch": 0.7047882828764548,
      "grad_norm": 1.4258638350032615,
      "learning_rate": 2.116134150127285e-06,
      "loss": 0.6259,
      "step": 9606
    },
    {
      "epoch": 0.7048616524665939,
      "grad_norm": 1.609051198429087,
      "learning_rate": 2.1151635883390696e-06,
      "loss": 0.6805,
      "step": 9607
    },
    {
      "epoch": 0.7049350220567331,
      "grad_norm": 0.5658990357061419,
      "learning_rate": 2.1141931894647104e-06,
      "loss": 0.4866,
      "step": 9608
    },
    {
      "epoch": 0.7049350220567331,
      "lm_loss": 1.2022,
      "step": 9608,
      "vm_loss": 0.0534
    },
    {
      "epoch": 0.7049350220567331,
      "lm_loss": 0.7671,
      "step": 9608,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.7049350220567331,
      "lm_loss": 0.1953,
      "step": 9608,
      "vm_loss": 0.1936
    },
    {
      "epoch": 0.7049350220567331,
      "lm_loss": 0.2496,
      "step": 9608,
      "vm_loss": 0.1375
    },
    {
      "epoch": 0.7049350220567331,
      "lm_loss": 0.8753,
      "step": 9608,
      "vm_loss": 0.1088
    },
    {
      "epoch": 0.7049350220567331,
      "lm_loss": 0.9054,
      "step": 9608,
      "vm_loss": 0.2199
    },
    {
      "epoch": 0.7049350220567331,
      "lm_loss": 0.5597,
      "step": 9608,
      "vm_loss": 0.1368
    },
    {
      "epoch": 0.7049350220567331,
      "lm_loss": 0.7534,
      "step": 9608,
      "vm_loss": 0.0676
    },
    {
      "epoch": 0.7050083916468721,
      "grad_norm": 1.6597595562850227,
      "learning_rate": 2.1132229535590092e-06,
      "loss": 0.6864,
      "step": 9609
    },
    {
      "epoch": 0.7050817612370113,
      "grad_norm": 0.5290554218842016,
      "learning_rate": 2.1122528806767546e-06,
      "loss": 0.4621,
      "step": 9610
    },
    {
      "epoch": 0.7051551308271504,
      "grad_norm": 1.5726956163171537,
      "learning_rate": 2.1112829708727326e-06,
      "loss": 0.6775,
      "step": 9611
    },
    {
      "epoch": 0.7052285004172896,
      "grad_norm": 0.5509964697740557,
      "learning_rate": 2.110313224201716e-06,
      "loss": 0.4746,
      "step": 9612
    },
    {
      "epoch": 0.7053018700074287,
      "grad_norm": 0.552184667812731,
      "learning_rate": 2.109343640718467e-06,
      "loss": 0.4941,
      "step": 9613
    },
    {
      "epoch": 0.7053752395975678,
      "grad_norm": 1.5723748016734942,
      "learning_rate": 2.1083742204777415e-06,
      "loss": 0.698,
      "step": 9614
    },
    {
      "epoch": 0.7054486091877069,
      "grad_norm": 1.468746175874083,
      "learning_rate": 2.1074049635342832e-06,
      "loss": 0.7343,
      "step": 9615
    },
    {
      "epoch": 0.7055219787778461,
      "grad_norm": 1.600155368684121,
      "learning_rate": 2.106435869942831e-06,
      "loss": 0.8909,
      "step": 9616
    },
    {
      "epoch": 0.7055219787778461,
      "lm_loss": 1.2326,
      "step": 9616,
      "vm_loss": 0.0553
    },
    {
      "epoch": 0.7055219787778461,
      "lm_loss": 1.0126,
      "step": 9616,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.7055219787778461,
      "lm_loss": 0.1018,
      "step": 9616,
      "vm_loss": 0.1066
    },
    {
      "epoch": 0.7055219787778461,
      "lm_loss": 0.2111,
      "step": 9616,
      "vm_loss": 0.1644
    },
    {
      "epoch": 0.7055219787778461,
      "lm_loss": 1.1601,
      "step": 9616,
      "vm_loss": 0.1317
    },
    {
      "epoch": 0.7055219787778461,
      "lm_loss": 0.3745,
      "step": 9616,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.7055219787778461,
      "lm_loss": 0.6836,
      "step": 9616,
      "vm_loss": 0.1379
    },
    {
      "epoch": 0.7055219787778461,
      "lm_loss": 0.3528,
      "step": 9616,
      "vm_loss": 0.1162
    },
    {
      "epoch": 0.7055953483679852,
      "grad_norm": 1.5391289748672645,
      "learning_rate": 2.105466939758112e-06,
      "loss": 0.7423,
      "step": 9617
    },
    {
      "epoch": 0.7056687179581244,
      "grad_norm": 1.443750772726773,
      "learning_rate": 2.1044981730348422e-06,
      "loss": 0.7856,
      "step": 9618
    },
    {
      "epoch": 0.7057420875482634,
      "grad_norm": 1.525909572298653,
      "learning_rate": 2.103529569827732e-06,
      "loss": 0.7532,
      "step": 9619
    },
    {
      "epoch": 0.7058154571384025,
      "grad_norm": 1.6415580223720476,
      "learning_rate": 2.102561130191478e-06,
      "loss": 0.7354,
      "step": 9620
    },
    {
      "epoch": 0.7058888267285417,
      "grad_norm": 0.5557941494261943,
      "learning_rate": 2.1015928541807734e-06,
      "loss": 0.4725,
      "step": 9621
    },
    {
      "epoch": 0.7059621963186808,
      "grad_norm": 1.54818243240124,
      "learning_rate": 2.100624741850299e-06,
      "loss": 0.6339,
      "step": 9622
    },
    {
      "epoch": 0.70603556590882,
      "grad_norm": 1.404204749401092,
      "learning_rate": 2.099656793254725e-06,
      "loss": 0.7997,
      "step": 9623
    },
    {
      "epoch": 0.706108935498959,
      "grad_norm": 1.4948597193302389,
      "learning_rate": 2.098689008448715e-06,
      "loss": 0.7747,
      "step": 9624
    },
    {
      "epoch": 0.706108935498959,
      "lm_loss": 0.3351,
      "step": 9624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.706108935498959,
      "lm_loss": 0.4912,
      "step": 9624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.706108935498959,
      "lm_loss": 0.5413,
      "step": 9624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.706108935498959,
      "lm_loss": 0.4852,
      "step": 9624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.706108935498959,
      "lm_loss": 0.4917,
      "step": 9624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.706108935498959,
      "lm_loss": 0.4451,
      "step": 9624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.706108935498959,
      "lm_loss": 0.5545,
      "step": 9624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.706108935498959,
      "lm_loss": 0.4693,
      "step": 9624,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7061823050890982,
      "grad_norm": 0.566178848445674,
      "learning_rate": 2.097721387486921e-06,
      "loss": 0.4945,
      "step": 9625
    },
    {
      "epoch": 0.7062556746792373,
      "grad_norm": 0.564776726182577,
      "learning_rate": 2.0967539304239865e-06,
      "loss": 0.4574,
      "step": 9626
    },
    {
      "epoch": 0.7063290442693765,
      "grad_norm": 0.5462398185383183,
      "learning_rate": 2.0957866373145487e-06,
      "loss": 0.4874,
      "step": 9627
    },
    {
      "epoch": 0.7064024138595156,
      "grad_norm": 1.5979360004348742,
      "learning_rate": 2.0948195082132315e-06,
      "loss": 0.7973,
      "step": 9628
    },
    {
      "epoch": 0.7064757834496547,
      "grad_norm": 1.681574752368375,
      "learning_rate": 2.093852543174652e-06,
      "loss": 0.7222,
      "step": 9629
    },
    {
      "epoch": 0.7065491530397938,
      "grad_norm": 1.6562524997784407,
      "learning_rate": 2.0928857422534154e-06,
      "loss": 0.8271,
      "step": 9630
    },
    {
      "epoch": 0.706622522629933,
      "grad_norm": 1.4502200518012698,
      "learning_rate": 2.0919191055041193e-06,
      "loss": 0.7054,
      "step": 9631
    },
    {
      "epoch": 0.7066958922200721,
      "grad_norm": 1.6098328457488407,
      "learning_rate": 2.0909526329813547e-06,
      "loss": 0.7112,
      "step": 9632
    },
    {
      "epoch": 0.7066958922200721,
      "lm_loss": 1.0126,
      "step": 9632,
      "vm_loss": 0.0893
    },
    {
      "epoch": 0.7066958922200721,
      "lm_loss": 0.0588,
      "step": 9632,
      "vm_loss": 0.0469
    },
    {
      "epoch": 0.7066958922200721,
      "lm_loss": 0.2161,
      "step": 9632,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.7066958922200721,
      "lm_loss": 0.9859,
      "step": 9632,
      "vm_loss": 0.1956
    },
    {
      "epoch": 0.7066958922200721,
      "lm_loss": 0.31,
      "step": 9632,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.7066958922200721,
      "lm_loss": 0.7332,
      "step": 9632,
      "vm_loss": 0.0664
    },
    {
      "epoch": 0.7066958922200721,
      "lm_loss": 1.0035,
      "step": 9632,
      "vm_loss": 0.088
    },
    {
      "epoch": 0.7066958922200721,
      "lm_loss": 0.5618,
      "step": 9632,
      "vm_loss": 0.1392
    },
    {
      "epoch": 0.7067692618102112,
      "grad_norm": 1.6591314967305488,
      "learning_rate": 2.0899863247396985e-06,
      "loss": 0.6987,
      "step": 9633
    },
    {
      "epoch": 0.7068426314003503,
      "grad_norm": 0.5690903414358042,
      "learning_rate": 2.089020180833721e-06,
      "loss": 0.4966,
      "step": 9634
    },
    {
      "epoch": 0.7069160009904895,
      "grad_norm": 1.6395752116645215,
      "learning_rate": 2.088054201317984e-06,
      "loss": 0.7447,
      "step": 9635
    },
    {
      "epoch": 0.7069893705806286,
      "grad_norm": 0.5789617016060303,
      "learning_rate": 2.0870883862470366e-06,
      "loss": 0.491,
      "step": 9636
    },
    {
      "epoch": 0.7070627401707678,
      "grad_norm": 1.4661963751510225,
      "learning_rate": 2.08612273567542e-06,
      "loss": 0.6524,
      "step": 9637
    },
    {
      "epoch": 0.7071361097609068,
      "grad_norm": 1.8988357626758803,
      "learning_rate": 2.0851572496576705e-06,
      "loss": 0.7907,
      "step": 9638
    },
    {
      "epoch": 0.7072094793510459,
      "grad_norm": 1.8016056275701444,
      "learning_rate": 2.0841919282483097e-06,
      "loss": 0.7501,
      "step": 9639
    },
    {
      "epoch": 0.7072828489411851,
      "grad_norm": 1.4730291366385317,
      "learning_rate": 2.083226771501851e-06,
      "loss": 0.6609,
      "step": 9640
    },
    {
      "epoch": 0.7072828489411851,
      "lm_loss": 1.0185,
      "step": 9640,
      "vm_loss": 0.0356
    },
    {
      "epoch": 0.7072828489411851,
      "lm_loss": 0.6753,
      "step": 9640,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.7072828489411851,
      "lm_loss": 0.8891,
      "step": 9640,
      "vm_loss": 0.1394
    },
    {
      "epoch": 0.7072828489411851,
      "lm_loss": 0.7192,
      "step": 9640,
      "vm_loss": 0.1625
    },
    {
      "epoch": 0.7072828489411851,
      "lm_loss": 0.1158,
      "step": 9640,
      "vm_loss": 0.1035
    },
    {
      "epoch": 0.7072828489411851,
      "lm_loss": 1.0796,
      "step": 9640,
      "vm_loss": 0.1706
    },
    {
      "epoch": 0.7072828489411851,
      "lm_loss": 0.1501,
      "step": 9640,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.7072828489411851,
      "lm_loss": 0.0712,
      "step": 9640,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.7073562185313242,
      "grad_norm": 1.4358091265154571,
      "learning_rate": 2.0822617794728006e-06,
      "loss": 0.6452,
      "step": 9641
    },
    {
      "epoch": 0.7074295881214634,
      "grad_norm": 1.3893928258699146,
      "learning_rate": 2.081296952215651e-06,
      "loss": 0.6486,
      "step": 9642
    },
    {
      "epoch": 0.7075029577116024,
      "grad_norm": 1.708550529509632,
      "learning_rate": 2.080332289784893e-06,
      "loss": 0.7369,
      "step": 9643
    },
    {
      "epoch": 0.7075763273017416,
      "grad_norm": 0.5348055146506956,
      "learning_rate": 2.0793677922349996e-06,
      "loss": 0.4822,
      "step": 9644
    },
    {
      "epoch": 0.7076496968918807,
      "grad_norm": 0.5581606560410106,
      "learning_rate": 2.0784034596204413e-06,
      "loss": 0.4712,
      "step": 9645
    },
    {
      "epoch": 0.7077230664820199,
      "grad_norm": 0.5432433878822039,
      "learning_rate": 2.077439291995674e-06,
      "loss": 0.4707,
      "step": 9646
    },
    {
      "epoch": 0.707796436072159,
      "grad_norm": 0.537953412249347,
      "learning_rate": 2.0764752894151457e-06,
      "loss": 0.4831,
      "step": 9647
    },
    {
      "epoch": 0.7078698056622981,
      "grad_norm": 0.5549861440132069,
      "learning_rate": 2.0755114519333016e-06,
      "loss": 0.484,
      "step": 9648
    },
    {
      "epoch": 0.7078698056622981,
      "lm_loss": 0.9819,
      "step": 9648,
      "vm_loss": 0.0593
    },
    {
      "epoch": 0.7078698056622981,
      "lm_loss": 0.9922,
      "step": 9648,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.7078698056622981,
      "lm_loss": 0.8874,
      "step": 9648,
      "vm_loss": 0.1297
    },
    {
      "epoch": 0.7078698056622981,
      "lm_loss": 0.4502,
      "step": 9648,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.7078698056622981,
      "lm_loss": 0.4272,
      "step": 9648,
      "vm_loss": 0.0949
    },
    {
      "epoch": 0.7078698056622981,
      "lm_loss": 0.8053,
      "step": 9648,
      "vm_loss": 0.0897
    },
    {
      "epoch": 0.7078698056622981,
      "lm_loss": 0.2295,
      "step": 9648,
      "vm_loss": 0.1128
    },
    {
      "epoch": 0.7078698056622981,
      "lm_loss": 0.804,
      "step": 9648,
      "vm_loss": 0.1347
    },
    {
      "epoch": 0.7079431752524372,
      "grad_norm": 1.5186251567084643,
      "learning_rate": 2.0745477796045664e-06,
      "loss": 0.7147,
      "step": 9649
    },
    {
      "epoch": 0.7080165448425764,
      "grad_norm": 0.5861057848691918,
      "learning_rate": 2.073584272483363e-06,
      "loss": 0.4954,
      "step": 9650
    },
    {
      "epoch": 0.7080899144327155,
      "grad_norm": 0.5709620437206079,
      "learning_rate": 2.0726209306241025e-06,
      "loss": 0.4917,
      "step": 9651
    },
    {
      "epoch": 0.7081632840228547,
      "grad_norm": 1.7902501123409515,
      "learning_rate": 2.0716577540811863e-06,
      "loss": 0.745,
      "step": 9652
    },
    {
      "epoch": 0.7082366536129937,
      "grad_norm": 0.5551483874405392,
      "learning_rate": 2.0706947429090096e-06,
      "loss": 0.4869,
      "step": 9653
    },
    {
      "epoch": 0.7083100232031329,
      "grad_norm": 1.6507990925675444,
      "learning_rate": 2.0697318971619555e-06,
      "loss": 0.7904,
      "step": 9654
    },
    {
      "epoch": 0.708383392793272,
      "grad_norm": 0.540046581574552,
      "learning_rate": 2.068769216894398e-06,
      "loss": 0.4853,
      "step": 9655
    },
    {
      "epoch": 0.7084567623834112,
      "grad_norm": 1.6757504669454435,
      "learning_rate": 2.067806702160702e-06,
      "loss": 0.7393,
      "step": 9656
    },
    {
      "epoch": 0.7084567623834112,
      "lm_loss": 1.278,
      "step": 9656,
      "vm_loss": 0.0816
    },
    {
      "epoch": 0.7084567623834112,
      "lm_loss": 0.0663,
      "step": 9656,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.7084567623834112,
      "lm_loss": 0.8767,
      "step": 9656,
      "vm_loss": 0.1498
    },
    {
      "epoch": 0.7084567623834112,
      "lm_loss": 0.362,
      "step": 9656,
      "vm_loss": 0.0991
    },
    {
      "epoch": 0.7084567623834112,
      "lm_loss": 0.7754,
      "step": 9656,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.7084567623834112,
      "lm_loss": 0.1971,
      "step": 9656,
      "vm_loss": 0.0876
    },
    {
      "epoch": 0.7084567623834112,
      "lm_loss": 0.7969,
      "step": 9656,
      "vm_loss": 0.2725
    },
    {
      "epoch": 0.7084567623834112,
      "lm_loss": 0.679,
      "step": 9656,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.7085301319735503,
      "grad_norm": 1.5582439539034867,
      "learning_rate": 2.066844353015221e-06,
      "loss": 0.7226,
      "step": 9657
    },
    {
      "epoch": 0.7086035015636893,
      "grad_norm": 0.5567457507794924,
      "learning_rate": 2.065882169512305e-06,
      "loss": 0.4805,
      "step": 9658
    },
    {
      "epoch": 0.7086768711538285,
      "grad_norm": 1.4924785882472504,
      "learning_rate": 2.064920151706289e-06,
      "loss": 0.7479,
      "step": 9659
    },
    {
      "epoch": 0.7087502407439676,
      "grad_norm": 0.5599600556875194,
      "learning_rate": 2.0639582996515006e-06,
      "loss": 0.481,
      "step": 9660
    },
    {
      "epoch": 0.7088236103341068,
      "grad_norm": 1.47395174264322,
      "learning_rate": 2.0629966134022584e-06,
      "loss": 0.7565,
      "step": 9661
    },
    {
      "epoch": 0.7088969799242459,
      "grad_norm": 1.4292190726733178,
      "learning_rate": 2.0620350930128686e-06,
      "loss": 0.6689,
      "step": 9662
    },
    {
      "epoch": 0.708970349514385,
      "grad_norm": 1.6829208404520073,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.7367,
      "step": 9663
    },
    {
      "epoch": 0.7090437191045241,
      "grad_norm": 1.4347378076818524,
      "learning_rate": 2.060112550030845e-06,
      "loss": 0.8701,
      "step": 9664
    },
    {
      "epoch": 0.7090437191045241,
      "lm_loss": 1.3634,
      "step": 9664,
      "vm_loss": 0.0589
    },
    {
      "epoch": 0.7090437191045241,
      "lm_loss": 1.1051,
      "step": 9664,
      "vm_loss": 0.2058
    },
    {
      "epoch": 0.7090437191045241,
      "lm_loss": 0.8295,
      "step": 9664,
      "vm_loss": 0.1579
    },
    {
      "epoch": 0.7090437191045241,
      "lm_loss": 0.1049,
      "step": 9664,
      "vm_loss": 0.1339
    },
    {
      "epoch": 0.7090437191045241,
      "lm_loss": 0.9629,
      "step": 9664,
      "vm_loss": 0.1867
    },
    {
      "epoch": 0.7090437191045241,
      "lm_loss": 0.5086,
      "step": 9664,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.7090437191045241,
      "lm_loss": 0.9177,
      "step": 9664,
      "vm_loss": 0.0537
    },
    {
      "epoch": 0.7090437191045241,
      "lm_loss": 0.5365,
      "step": 9664,
      "vm_loss": 0.1116
    },
    {
      "epoch": 0.7091170886946633,
      "grad_norm": 1.5385160723115916,
      "learning_rate": 2.0591515275467807e-06,
      "loss": 0.755,
      "step": 9665
    },
    {
      "epoch": 0.7091904582848024,
      "grad_norm": 0.5673431703828647,
      "learning_rate": 2.0581906711397115e-06,
      "loss": 0.4534,
      "step": 9666
    },
    {
      "epoch": 0.7092638278749416,
      "grad_norm": 1.4328883898330766,
      "learning_rate": 2.0572299808639006e-06,
      "loss": 0.6764,
      "step": 9667
    },
    {
      "epoch": 0.7093371974650806,
      "grad_norm": 1.4393030253748942,
      "learning_rate": 2.0562694567735993e-06,
      "loss": 0.7851,
      "step": 9668
    },
    {
      "epoch": 0.7094105670552198,
      "grad_norm": 1.2833064892857386,
      "learning_rate": 2.0553090989230527e-06,
      "loss": 0.688,
      "step": 9669
    },
    {
      "epoch": 0.7094839366453589,
      "grad_norm": 1.411209130087144,
      "learning_rate": 2.0543489073664942e-06,
      "loss": 0.7122,
      "step": 9670
    },
    {
      "epoch": 0.7095573062354981,
      "grad_norm": 1.507878642725676,
      "learning_rate": 2.0533888821581476e-06,
      "loss": 0.6802,
      "step": 9671
    },
    {
      "epoch": 0.7096306758256372,
      "grad_norm": 1.460060017052969,
      "learning_rate": 2.0524290233522276e-06,
      "loss": 0.717,
      "step": 9672
    },
    {
      "epoch": 0.7096306758256372,
      "lm_loss": 0.3735,
      "step": 9672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7096306758256372,
      "lm_loss": 0.5277,
      "step": 9672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7096306758256372,
      "lm_loss": 0.411,
      "step": 9672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7096306758256372,
      "lm_loss": 0.5026,
      "step": 9672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7096306758256372,
      "lm_loss": 0.4594,
      "step": 9672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7096306758256372,
      "lm_loss": 0.3759,
      "step": 9672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7096306758256372,
      "lm_loss": 0.4819,
      "step": 9672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7096306758256372,
      "lm_loss": 0.5089,
      "step": 9672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7097040454157763,
      "grad_norm": 0.5666923379559148,
      "learning_rate": 2.0514693310029388e-06,
      "loss": 0.4765,
      "step": 9673
    },
    {
      "epoch": 0.7097774150059154,
      "grad_norm": 0.5544836578508501,
      "learning_rate": 2.0505098051644807e-06,
      "loss": 0.4992,
      "step": 9674
    },
    {
      "epoch": 0.7098507845960546,
      "grad_norm": 1.403900489559663,
      "learning_rate": 2.049550445891038e-06,
      "loss": 0.7219,
      "step": 9675
    },
    {
      "epoch": 0.7099241541861937,
      "grad_norm": 1.4562596351261112,
      "learning_rate": 2.048591253236788e-06,
      "loss": 0.7898,
      "step": 9676
    },
    {
      "epoch": 0.7099975237763329,
      "grad_norm": 0.5656718864495011,
      "learning_rate": 2.0476322272558995e-06,
      "loss": 0.4858,
      "step": 9677
    },
    {
      "epoch": 0.7100708933664719,
      "grad_norm": 1.4296384428273197,
      "learning_rate": 2.0466733680025303e-06,
      "loss": 0.6893,
      "step": 9678
    },
    {
      "epoch": 0.710144262956611,
      "grad_norm": 1.4651728386015306,
      "learning_rate": 2.0457146755308276e-06,
      "loss": 0.86,
      "step": 9679
    },
    {
      "epoch": 0.7102176325467502,
      "grad_norm": 0.5296757217312098,
      "learning_rate": 2.044756149894936e-06,
      "loss": 0.4768,
      "step": 9680
    },
    {
      "epoch": 0.7102176325467502,
      "lm_loss": 0.3963,
      "step": 9680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7102176325467502,
      "lm_loss": 0.495,
      "step": 9680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7102176325467502,
      "lm_loss": 0.5027,
      "step": 9680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7102176325467502,
      "lm_loss": 0.4716,
      "step": 9680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7102176325467502,
      "lm_loss": 0.6631,
      "step": 9680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7102176325467502,
      "lm_loss": 0.5256,
      "step": 9680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7102176325467502,
      "lm_loss": 0.402,
      "step": 9680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7102176325467502,
      "lm_loss": 0.4201,
      "step": 9680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7102910021368893,
      "grad_norm": 0.557865024200173,
      "learning_rate": 2.0437977911489825e-06,
      "loss": 0.4841,
      "step": 9681
    },
    {
      "epoch": 0.7103643717270285,
      "grad_norm": 0.5767914204600525,
      "learning_rate": 2.042839599347089e-06,
      "loss": 0.4881,
      "step": 9682
    },
    {
      "epoch": 0.7104377413171675,
      "grad_norm": 1.4449478673617995,
      "learning_rate": 2.0418815745433663e-06,
      "loss": 0.7967,
      "step": 9683
    },
    {
      "epoch": 0.7105111109073067,
      "grad_norm": 1.4506444176247788,
      "learning_rate": 2.040923716791915e-06,
      "loss": 0.6927,
      "step": 9684
    },
    {
      "epoch": 0.7105844804974458,
      "grad_norm": 1.5176192811608573,
      "learning_rate": 2.039966026146831e-06,
      "loss": 0.6567,
      "step": 9685
    },
    {
      "epoch": 0.710657850087585,
      "grad_norm": 0.5496382817516777,
      "learning_rate": 2.0390085026621964e-06,
      "loss": 0.5003,
      "step": 9686
    },
    {
      "epoch": 0.7107312196777241,
      "grad_norm": 1.5790240502040718,
      "learning_rate": 2.0380511463920844e-06,
      "loss": 0.7073,
      "step": 9687
    },
    {
      "epoch": 0.7108045892678632,
      "grad_norm": 1.5482192157666255,
      "learning_rate": 2.037093957390559e-06,
      "loss": 0.745,
      "step": 9688
    },
    {
      "epoch": 0.7108045892678632,
      "lm_loss": 0.1884,
      "step": 9688,
      "vm_loss": 0.0884
    },
    {
      "epoch": 0.7108045892678632,
      "lm_loss": 1.1446,
      "step": 9688,
      "vm_loss": 0.0758
    },
    {
      "epoch": 0.7108045892678632,
      "lm_loss": 0.4377,
      "step": 9688,
      "vm_loss": 0.1451
    },
    {
      "epoch": 0.7108045892678632,
      "lm_loss": 0.3079,
      "step": 9688,
      "vm_loss": 0.2828
    },
    {
      "epoch": 0.7108045892678632,
      "lm_loss": 0.6284,
      "step": 9688,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.7108045892678632,
      "lm_loss": 0.1924,
      "step": 9688,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.7108045892678632,
      "lm_loss": 0.0546,
      "step": 9688,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.7108045892678632,
      "lm_loss": 0.1799,
      "step": 9688,
      "vm_loss": 0.094
    },
    {
      "epoch": 0.7108779588580023,
      "grad_norm": 1.4563194941948099,
      "learning_rate": 2.036136935711674e-06,
      "loss": 0.7742,
      "step": 9689
    },
    {
      "epoch": 0.7109513284481415,
      "grad_norm": 1.505493308982518,
      "learning_rate": 2.0351800814094774e-06,
      "loss": 0.669,
      "step": 9690
    },
    {
      "epoch": 0.7110246980382806,
      "grad_norm": 1.3807889730040814,
      "learning_rate": 2.0342233945380043e-06,
      "loss": 0.7983,
      "step": 9691
    },
    {
      "epoch": 0.7110980676284198,
      "grad_norm": 1.433521362439275,
      "learning_rate": 2.0332668751512804e-06,
      "loss": 0.7084,
      "step": 9692
    },
    {
      "epoch": 0.7111714372185588,
      "grad_norm": 1.4061171051304777,
      "learning_rate": 2.032310523303323e-06,
      "loss": 0.7486,
      "step": 9693
    },
    {
      "epoch": 0.711244806808698,
      "grad_norm": 0.5590937777472761,
      "learning_rate": 2.03135433904814e-06,
      "loss": 0.4806,
      "step": 9694
    },
    {
      "epoch": 0.7113181763988371,
      "grad_norm": 0.540501167120514,
      "learning_rate": 2.030398322439727e-06,
      "loss": 0.4703,
      "step": 9695
    },
    {
      "epoch": 0.7113915459889762,
      "grad_norm": 0.5323801531467311,
      "learning_rate": 2.029442473532077e-06,
      "loss": 0.4572,
      "step": 9696
    },
    {
      "epoch": 0.7113915459889762,
      "lm_loss": 1.6062,
      "step": 9696,
      "vm_loss": 0.058
    },
    {
      "epoch": 0.7113915459889762,
      "lm_loss": 0.0204,
      "step": 9696,
      "vm_loss": 0.1084
    },
    {
      "epoch": 0.7113915459889762,
      "lm_loss": 0.7837,
      "step": 9696,
      "vm_loss": 0.1444
    },
    {
      "epoch": 0.7113915459889762,
      "lm_loss": 0.0289,
      "step": 9696,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.7113915459889762,
      "lm_loss": 0.0757,
      "step": 9696,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.7113915459889762,
      "lm_loss": 0.2938,
      "step": 9696,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.7113915459889762,
      "lm_loss": 0.3455,
      "step": 9696,
      "vm_loss": 0.0597
    },
    {
      "epoch": 0.7113915459889762,
      "lm_loss": 0.4972,
      "step": 9696,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.7114649155791154,
      "grad_norm": 1.5407113929387222,
      "learning_rate": 2.0284867923791668e-06,
      "loss": 0.7429,
      "step": 9697
    },
    {
      "epoch": 0.7115382851692544,
      "grad_norm": 1.5053179128520042,
      "learning_rate": 2.0275312790349665e-06,
      "loss": 0.7187,
      "step": 9698
    },
    {
      "epoch": 0.7116116547593936,
      "grad_norm": 1.5104579004852559,
      "learning_rate": 2.0265759335534357e-06,
      "loss": 0.7412,
      "step": 9699
    },
    {
      "epoch": 0.7116850243495327,
      "grad_norm": 1.4587041497220639,
      "learning_rate": 2.025620755988524e-06,
      "loss": 0.7189,
      "step": 9700
    },
    {
      "epoch": 0.7117583939396719,
      "grad_norm": 1.3799922325935874,
      "learning_rate": 2.024665746394176e-06,
      "loss": 0.6757,
      "step": 9701
    },
    {
      "epoch": 0.711831763529811,
      "grad_norm": 1.4025495358524351,
      "learning_rate": 2.0237109048243214e-06,
      "loss": 0.7329,
      "step": 9702
    },
    {
      "epoch": 0.7119051331199501,
      "grad_norm": 1.5445186423821744,
      "learning_rate": 2.0227562313328823e-06,
      "loss": 0.7935,
      "step": 9703
    },
    {
      "epoch": 0.7119785027100892,
      "grad_norm": 1.5104138615975575,
      "learning_rate": 2.021801725973773e-06,
      "loss": 0.779,
      "step": 9704
    },
    {
      "epoch": 0.7119785027100892,
      "lm_loss": 0.627,
      "step": 9704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7119785027100892,
      "lm_loss": 0.4773,
      "step": 9704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7119785027100892,
      "lm_loss": 0.5356,
      "step": 9704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7119785027100892,
      "lm_loss": 0.3771,
      "step": 9704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7119785027100892,
      "lm_loss": 0.3903,
      "step": 9704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7119785027100892,
      "lm_loss": 0.5558,
      "step": 9704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7119785027100892,
      "lm_loss": 0.5126,
      "step": 9704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7119785027100892,
      "lm_loss": 0.4958,
      "step": 9704,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7120518723002284,
      "grad_norm": 0.548440958296708,
      "learning_rate": 2.0208473888008944e-06,
      "loss": 0.4806,
      "step": 9705
    },
    {
      "epoch": 0.7121252418903675,
      "grad_norm": 1.3433711917291762,
      "learning_rate": 2.0198932198681397e-06,
      "loss": 0.8032,
      "step": 9706
    },
    {
      "epoch": 0.7121986114805067,
      "grad_norm": 0.5608270590044979,
      "learning_rate": 2.0189392192293976e-06,
      "loss": 0.4929,
      "step": 9707
    },
    {
      "epoch": 0.7122719810706457,
      "grad_norm": 1.4988808349136928,
      "learning_rate": 2.0179853869385397e-06,
      "loss": 0.6907,
      "step": 9708
    },
    {
      "epoch": 0.7123453506607849,
      "grad_norm": 1.4820222314337772,
      "learning_rate": 2.017031723049432e-06,
      "loss": 0.8499,
      "step": 9709
    },
    {
      "epoch": 0.712418720250924,
      "grad_norm": 0.5493484480747957,
      "learning_rate": 2.01607822761593e-06,
      "loss": 0.4828,
      "step": 9710
    },
    {
      "epoch": 0.7124920898410632,
      "grad_norm": 1.5219993384980353,
      "learning_rate": 2.0151249006918786e-06,
      "loss": 0.6671,
      "step": 9711
    },
    {
      "epoch": 0.7125654594312022,
      "grad_norm": 1.5638211306223344,
      "learning_rate": 2.0141717423311174e-06,
      "loss": 0.7231,
      "step": 9712
    },
    {
      "epoch": 0.7125654594312022,
      "lm_loss": 0.5181,
      "step": 9712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7125654594312022,
      "lm_loss": 0.4799,
      "step": 9712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7125654594312022,
      "lm_loss": 0.673,
      "step": 9712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7125654594312022,
      "lm_loss": 0.3972,
      "step": 9712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7125654594312022,
      "lm_loss": 0.331,
      "step": 9712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7125654594312022,
      "lm_loss": 0.3406,
      "step": 9712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7125654594312022,
      "lm_loss": 0.3606,
      "step": 9712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7125654594312022,
      "lm_loss": 0.3422,
      "step": 9712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7126388290213413,
      "grad_norm": 0.5483853304984345,
      "learning_rate": 2.0132187525874723e-06,
      "loss": 0.47,
      "step": 9713
    },
    {
      "epoch": 0.7127121986114805,
      "grad_norm": 0.5656900807763523,
      "learning_rate": 2.012265931514761e-06,
      "loss": 0.4822,
      "step": 9714
    },
    {
      "epoch": 0.7127855682016196,
      "grad_norm": 1.5237453414163389,
      "learning_rate": 2.0113132791667915e-06,
      "loss": 0.7446,
      "step": 9715
    },
    {
      "epoch": 0.7128589377917588,
      "grad_norm": 1.529194636748353,
      "learning_rate": 2.0103607955973607e-06,
      "loss": 0.7647,
      "step": 9716
    },
    {
      "epoch": 0.7129323073818978,
      "grad_norm": 0.5274625881137082,
      "learning_rate": 2.009408480860261e-06,
      "loss": 0.4649,
      "step": 9717
    },
    {
      "epoch": 0.713005676972037,
      "grad_norm": 1.446830490274539,
      "learning_rate": 2.008456335009271e-06,
      "loss": 0.8261,
      "step": 9718
    },
    {
      "epoch": 0.7130790465621761,
      "grad_norm": 1.4702870738239444,
      "learning_rate": 2.0075043580981594e-06,
      "loss": 0.6951,
      "step": 9719
    },
    {
      "epoch": 0.7131524161523153,
      "grad_norm": 0.5632556772738163,
      "learning_rate": 2.0065525501806882e-06,
      "loss": 0.4947,
      "step": 9720
    },
    {
      "epoch": 0.7131524161523153,
      "lm_loss": 0.2385,
      "step": 9720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7131524161523153,
      "lm_loss": 0.4767,
      "step": 9720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7131524161523153,
      "lm_loss": 0.422,
      "step": 9720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7131524161523153,
      "lm_loss": 0.5205,
      "step": 9720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7131524161523153,
      "lm_loss": 0.5679,
      "step": 9720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7131524161523153,
      "lm_loss": 0.5758,
      "step": 9720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7131524161523153,
      "lm_loss": 0.4814,
      "step": 9720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7131524161523153,
      "lm_loss": 0.3849,
      "step": 9720,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7132257857424544,
      "grad_norm": 0.5582405592654373,
      "learning_rate": 2.0056009113106073e-06,
      "loss": 0.4813,
      "step": 9721
    },
    {
      "epoch": 0.7132991553325935,
      "grad_norm": 1.6640977303155475,
      "learning_rate": 2.0046494415416567e-06,
      "loss": 0.8052,
      "step": 9722
    },
    {
      "epoch": 0.7133725249227326,
      "grad_norm": 0.5644815815279773,
      "learning_rate": 2.003698140927572e-06,
      "loss": 0.4954,
      "step": 9723
    },
    {
      "epoch": 0.7134458945128718,
      "grad_norm": 1.565171296658589,
      "learning_rate": 2.0027470095220735e-06,
      "loss": 0.7364,
      "step": 9724
    },
    {
      "epoch": 0.7135192641030109,
      "grad_norm": 1.6141938293032294,
      "learning_rate": 2.0017960473788734e-06,
      "loss": 0.7977,
      "step": 9725
    },
    {
      "epoch": 0.7135926336931501,
      "grad_norm": 0.5622188728750905,
      "learning_rate": 2.000845254551676e-06,
      "loss": 0.4843,
      "step": 9726
    },
    {
      "epoch": 0.7136660032832891,
      "grad_norm": 1.7088655740193779,
      "learning_rate": 1.9998946310941725e-06,
      "loss": 0.7935,
      "step": 9727
    },
    {
      "epoch": 0.7137393728734283,
      "grad_norm": 0.5470485597977672,
      "learning_rate": 1.9989441770600505e-06,
      "loss": 0.4913,
      "step": 9728
    },
    {
      "epoch": 0.7137393728734283,
      "lm_loss": 1.5743,
      "step": 9728,
      "vm_loss": 0.1845
    },
    {
      "epoch": 0.7137393728734283,
      "lm_loss": 0.5155,
      "step": 9728,
      "vm_loss": 0.0928
    },
    {
      "epoch": 0.7137393728734283,
      "lm_loss": 0.17,
      "step": 9728,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.7137393728734283,
      "lm_loss": 0.3974,
      "step": 9728,
      "vm_loss": 0.2067
    },
    {
      "epoch": 0.7137393728734283,
      "lm_loss": 0.9052,
      "step": 9728,
      "vm_loss": 0.2179
    },
    {
      "epoch": 0.7137393728734283,
      "lm_loss": 0.1263,
      "step": 9728,
      "vm_loss": 0.0781
    },
    {
      "epoch": 0.7137393728734283,
      "lm_loss": 0.5944,
      "step": 9728,
      "vm_loss": 0.1482
    },
    {
      "epoch": 0.7137393728734283,
      "lm_loss": 0.1035,
      "step": 9728,
      "vm_loss": 0.0724
    },
    {
      "epoch": 0.7138127424635674,
      "grad_norm": 1.3924724523837022,
      "learning_rate": 1.997993892502983e-06,
      "loss": 0.7089,
      "step": 9729
    },
    {
      "epoch": 0.7138861120537066,
      "grad_norm": 0.5474925452564012,
      "learning_rate": 1.9970437774766343e-06,
      "loss": 0.4726,
      "step": 9730
    },
    {
      "epoch": 0.7139594816438457,
      "grad_norm": 1.4611150338561596,
      "learning_rate": 1.99609383203466e-06,
      "loss": 0.7683,
      "step": 9731
    },
    {
      "epoch": 0.7140328512339847,
      "grad_norm": 1.5207779235997154,
      "learning_rate": 1.9951440562307067e-06,
      "loss": 0.7266,
      "step": 9732
    },
    {
      "epoch": 0.7141062208241239,
      "grad_norm": 1.5183781526863982,
      "learning_rate": 1.994194450118408e-06,
      "loss": 0.7348,
      "step": 9733
    },
    {
      "epoch": 0.714179590414263,
      "grad_norm": 1.4544584013409674,
      "learning_rate": 1.9932450137513944e-06,
      "loss": 0.6616,
      "step": 9734
    },
    {
      "epoch": 0.7142529600044022,
      "grad_norm": 1.4217455011502125,
      "learning_rate": 1.9922957471832805e-06,
      "loss": 0.7136,
      "step": 9735
    },
    {
      "epoch": 0.7143263295945413,
      "grad_norm": 1.5568304613890422,
      "learning_rate": 1.9913466504676743e-06,
      "loss": 0.8293,
      "step": 9736
    },
    {
      "epoch": 0.7143263295945413,
      "lm_loss": 0.7433,
      "step": 9736,
      "vm_loss": 0.0891
    },
    {
      "epoch": 0.7143263295945413,
      "lm_loss": 1.2334,
      "step": 9736,
      "vm_loss": 0.1526
    },
    {
      "epoch": 0.7143263295945413,
      "lm_loss": 0.2721,
      "step": 9736,
      "vm_loss": 0.1551
    },
    {
      "epoch": 0.7143263295945413,
      "lm_loss": 0.3473,
      "step": 9736,
      "vm_loss": 0.1326
    },
    {
      "epoch": 0.7143263295945413,
      "lm_loss": 0.2923,
      "step": 9736,
      "vm_loss": 0.1422
    },
    {
      "epoch": 0.7143263295945413,
      "lm_loss": 0.7752,
      "step": 9736,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.7143263295945413,
      "lm_loss": 0.8102,
      "step": 9736,
      "vm_loss": 0.1518
    },
    {
      "epoch": 0.7143263295945413,
      "lm_loss": 0.964,
      "step": 9736,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.7143996991846804,
      "grad_norm": 1.4789395186125127,
      "learning_rate": 1.990397723658174e-06,
      "loss": 0.8328,
      "step": 9737
    },
    {
      "epoch": 0.7144730687748195,
      "grad_norm": 1.5227145561907616,
      "learning_rate": 1.989448966808365e-06,
      "loss": 0.7706,
      "step": 9738
    },
    {
      "epoch": 0.7145464383649587,
      "grad_norm": 0.5699868429442796,
      "learning_rate": 1.9885003799718304e-06,
      "loss": 0.4923,
      "step": 9739
    },
    {
      "epoch": 0.7146198079550978,
      "grad_norm": 1.6557536023674713,
      "learning_rate": 1.9875519632021374e-06,
      "loss": 0.7995,
      "step": 9740
    },
    {
      "epoch": 0.714693177545237,
      "grad_norm": 1.4226371981731645,
      "learning_rate": 1.986603716552845e-06,
      "loss": 0.6692,
      "step": 9741
    },
    {
      "epoch": 0.714766547135376,
      "grad_norm": 1.5808427568555559,
      "learning_rate": 1.9856556400775034e-06,
      "loss": 0.7889,
      "step": 9742
    },
    {
      "epoch": 0.7148399167255152,
      "grad_norm": 1.4873036113217635,
      "learning_rate": 1.9847077338296512e-06,
      "loss": 0.7703,
      "step": 9743
    },
    {
      "epoch": 0.7149132863156543,
      "grad_norm": 0.5526434357467195,
      "learning_rate": 1.983759997862822e-06,
      "loss": 0.4778,
      "step": 9744
    },
    {
      "epoch": 0.7149132863156543,
      "lm_loss": 0.4207,
      "step": 9744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7149132863156543,
      "lm_loss": 0.4282,
      "step": 9744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7149132863156543,
      "lm_loss": 0.4202,
      "step": 9744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7149132863156543,
      "lm_loss": 0.3715,
      "step": 9744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7149132863156543,
      "lm_loss": 0.5006,
      "step": 9744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7149132863156543,
      "lm_loss": 0.5611,
      "step": 9744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7149132863156543,
      "lm_loss": 0.423,
      "step": 9744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7149132863156543,
      "lm_loss": 0.4517,
      "step": 9744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7149866559057935,
      "grad_norm": 0.5634898803158122,
      "learning_rate": 1.9828124322305354e-06,
      "loss": 0.4732,
      "step": 9745
    },
    {
      "epoch": 0.7150600254959326,
      "grad_norm": 0.5587842589076044,
      "learning_rate": 1.9818650369863024e-06,
      "loss": 0.4933,
      "step": 9746
    },
    {
      "epoch": 0.7151333950860717,
      "grad_norm": 1.5434458296175377,
      "learning_rate": 1.9809178121836255e-06,
      "loss": 0.7126,
      "step": 9747
    },
    {
      "epoch": 0.7152067646762108,
      "grad_norm": 1.3778639919509934,
      "learning_rate": 1.9799707578759958e-06,
      "loss": 0.6767,
      "step": 9748
    },
    {
      "epoch": 0.71528013426635,
      "grad_norm": 1.5419298263387031,
      "learning_rate": 1.979023874116895e-06,
      "loss": 0.8167,
      "step": 9749
    },
    {
      "epoch": 0.7153535038564891,
      "grad_norm": 1.4299663814738448,
      "learning_rate": 1.9780771609597988e-06,
      "loss": 0.7114,
      "step": 9750
    },
    {
      "epoch": 0.7154268734466283,
      "grad_norm": 1.6379837127949444,
      "learning_rate": 1.977130618458169e-06,
      "loss": 0.7983,
      "step": 9751
    },
    {
      "epoch": 0.7155002430367673,
      "grad_norm": 0.5968391402519421,
      "learning_rate": 1.9761842466654596e-06,
      "loss": 0.498,
      "step": 9752
    },
    {
      "epoch": 0.7155002430367673,
      "lm_loss": 0.8821,
      "step": 9752,
      "vm_loss": 0.0538
    },
    {
      "epoch": 0.7155002430367673,
      "lm_loss": 1.0011,
      "step": 9752,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.7155002430367673,
      "lm_loss": 0.7976,
      "step": 9752,
      "vm_loss": 0.168
    },
    {
      "epoch": 0.7155002430367673,
      "lm_loss": 0.7767,
      "step": 9752,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.7155002430367673,
      "lm_loss": 0.6964,
      "step": 9752,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.7155002430367673,
      "lm_loss": 0.78,
      "step": 9752,
      "vm_loss": 0.203
    },
    {
      "epoch": 0.7155002430367673,
      "lm_loss": 0.8617,
      "step": 9752,
      "vm_loss": 0.092
    },
    {
      "epoch": 0.7155002430367673,
      "lm_loss": 0.0731,
      "step": 9752,
      "vm_loss": 0.0714
    },
    {
      "epoch": 0.7155736126269064,
      "grad_norm": 1.6121887880130223,
      "learning_rate": 1.9752380456351138e-06,
      "loss": 0.7066,
      "step": 9753
    },
    {
      "epoch": 0.7156469822170456,
      "grad_norm": 1.4375601241297098,
      "learning_rate": 1.974292015420565e-06,
      "loss": 0.7102,
      "step": 9754
    },
    {
      "epoch": 0.7157203518071847,
      "grad_norm": 0.5819784465257754,
      "learning_rate": 1.97334615607524e-06,
      "loss": 0.4723,
      "step": 9755
    },
    {
      "epoch": 0.7157937213973239,
      "grad_norm": 1.5347601955537693,
      "learning_rate": 1.972400467652554e-06,
      "loss": 0.6813,
      "step": 9756
    },
    {
      "epoch": 0.7158670909874629,
      "grad_norm": 0.5393898027147317,
      "learning_rate": 1.9714549502059107e-06,
      "loss": 0.4744,
      "step": 9757
    },
    {
      "epoch": 0.7159404605776021,
      "grad_norm": 1.5166148053094073,
      "learning_rate": 1.9705096037887065e-06,
      "loss": 0.7322,
      "step": 9758
    },
    {
      "epoch": 0.7160138301677412,
      "grad_norm": 1.5812312086848201,
      "learning_rate": 1.9695644284543276e-06,
      "loss": 0.7219,
      "step": 9759
    },
    {
      "epoch": 0.7160871997578804,
      "grad_norm": 0.5565086877398078,
      "learning_rate": 1.968619424256148e-06,
      "loss": 0.4965,
      "step": 9760
    },
    {
      "epoch": 0.7160871997578804,
      "lm_loss": 1.2033,
      "step": 9760,
      "vm_loss": 0.0228
    },
    {
      "epoch": 0.7160871997578804,
      "lm_loss": 0.3678,
      "step": 9760,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.7160871997578804,
      "lm_loss": 0.0734,
      "step": 9760,
      "vm_loss": 0.0617
    },
    {
      "epoch": 0.7160871997578804,
      "lm_loss": 0.7644,
      "step": 9760,
      "vm_loss": 0.2234
    },
    {
      "epoch": 0.7160871997578804,
      "lm_loss": 0.6374,
      "step": 9760,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.7160871997578804,
      "lm_loss": 0.8939,
      "step": 9760,
      "vm_loss": 0.094
    },
    {
      "epoch": 0.7160871997578804,
      "lm_loss": 0.7037,
      "step": 9760,
      "vm_loss": 0.266
    },
    {
      "epoch": 0.7160871997578804,
      "lm_loss": 0.5683,
      "step": 9760,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.7161605693480195,
      "grad_norm": 1.6970964220799927,
      "learning_rate": 1.9676745912475387e-06,
      "loss": 0.7846,
      "step": 9761
    },
    {
      "epoch": 0.7162339389381586,
      "grad_norm": 1.6522295336399146,
      "learning_rate": 1.9667299294818544e-06,
      "loss": 0.7225,
      "step": 9762
    },
    {
      "epoch": 0.7163073085282977,
      "grad_norm": 0.5437379044310822,
      "learning_rate": 1.965785439012443e-06,
      "loss": 0.4763,
      "step": 9763
    },
    {
      "epoch": 0.7163806781184369,
      "grad_norm": 1.4249179683002584,
      "learning_rate": 1.964841119892642e-06,
      "loss": 0.6563,
      "step": 9764
    },
    {
      "epoch": 0.716454047708576,
      "grad_norm": 1.551504036065491,
      "learning_rate": 1.963896972175777e-06,
      "loss": 0.7872,
      "step": 9765
    },
    {
      "epoch": 0.7165274172987152,
      "grad_norm": 1.4845129961234393,
      "learning_rate": 1.962952995915171e-06,
      "loss": 0.6993,
      "step": 9766
    },
    {
      "epoch": 0.7166007868888542,
      "grad_norm": 1.6112747634034688,
      "learning_rate": 1.962009191164131e-06,
      "loss": 0.7144,
      "step": 9767
    },
    {
      "epoch": 0.7166741564789934,
      "grad_norm": 1.52593910376039,
      "learning_rate": 1.961065557975955e-06,
      "loss": 0.7933,
      "step": 9768
    },
    {
      "epoch": 0.7166741564789934,
      "lm_loss": 0.5795,
      "step": 9768,
      "vm_loss": 0.0631
    },
    {
      "epoch": 0.7166741564789934,
      "lm_loss": 1.0102,
      "step": 9768,
      "vm_loss": 0.2167
    },
    {
      "epoch": 0.7166741564789934,
      "lm_loss": 0.6333,
      "step": 9768,
      "vm_loss": 0.1027
    },
    {
      "epoch": 0.7166741564789934,
      "lm_loss": 0.3649,
      "step": 9768,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.7166741564789934,
      "lm_loss": 0.0542,
      "step": 9768,
      "vm_loss": 0.0546
    },
    {
      "epoch": 0.7166741564789934,
      "lm_loss": 0.2455,
      "step": 9768,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.7166741564789934,
      "lm_loss": 1.0423,
      "step": 9768,
      "vm_loss": 0.1691
    },
    {
      "epoch": 0.7166741564789934,
      "lm_loss": 0.661,
      "step": 9768,
      "vm_loss": 0.0851
    },
    {
      "epoch": 0.7167475260691325,
      "grad_norm": 1.3956614332173283,
      "learning_rate": 1.9601220964039324e-06,
      "loss": 0.7477,
      "step": 9769
    },
    {
      "epoch": 0.7168208956592717,
      "grad_norm": 1.5325803472319302,
      "learning_rate": 1.9591788065013417e-06,
      "loss": 0.7622,
      "step": 9770
    },
    {
      "epoch": 0.7168942652494108,
      "grad_norm": 0.5638800370352918,
      "learning_rate": 1.958235688321456e-06,
      "loss": 0.4796,
      "step": 9771
    },
    {
      "epoch": 0.7169676348395498,
      "grad_norm": 1.5658203998900242,
      "learning_rate": 1.9572927419175337e-06,
      "loss": 0.7822,
      "step": 9772
    },
    {
      "epoch": 0.717041004429689,
      "grad_norm": 0.5944759992140887,
      "learning_rate": 1.9563499673428255e-06,
      "loss": 0.4941,
      "step": 9773
    },
    {
      "epoch": 0.7171143740198281,
      "grad_norm": 1.491480593068952,
      "learning_rate": 1.9554073646505717e-06,
      "loss": 0.7975,
      "step": 9774
    },
    {
      "epoch": 0.7171877436099673,
      "grad_norm": 1.5325674439573962,
      "learning_rate": 1.954464933894004e-06,
      "loss": 0.7839,
      "step": 9775
    },
    {
      "epoch": 0.7172611132001064,
      "grad_norm": 1.3936761330475826,
      "learning_rate": 1.953522675126342e-06,
      "loss": 0.7398,
      "step": 9776
    },
    {
      "epoch": 0.7172611132001064,
      "lm_loss": 0.7036,
      "step": 9776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7172611132001064,
      "lm_loss": 0.4907,
      "step": 9776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7172611132001064,
      "lm_loss": 0.634,
      "step": 9776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7172611132001064,
      "lm_loss": 0.5112,
      "step": 9776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7172611132001064,
      "lm_loss": 0.403,
      "step": 9776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7172611132001064,
      "lm_loss": 0.3218,
      "step": 9776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7172611132001064,
      "lm_loss": 0.5294,
      "step": 9776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7172611132001064,
      "lm_loss": 0.6304,
      "step": 9776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7173344827902455,
      "grad_norm": 0.5484590327941054,
      "learning_rate": 1.952580588400801e-06,
      "loss": 0.4715,
      "step": 9777
    },
    {
      "epoch": 0.7174078523803846,
      "grad_norm": 0.552041157428447,
      "learning_rate": 1.9516386737705807e-06,
      "loss": 0.4884,
      "step": 9778
    },
    {
      "epoch": 0.7174812219705238,
      "grad_norm": 1.52587764633335,
      "learning_rate": 1.950696931288874e-06,
      "loss": 0.7192,
      "step": 9779
    },
    {
      "epoch": 0.7175545915606629,
      "grad_norm": 0.5484663065907767,
      "learning_rate": 1.9497553610088626e-06,
      "loss": 0.4779,
      "step": 9780
    },
    {
      "epoch": 0.7176279611508021,
      "grad_norm": 1.546668032490406,
      "learning_rate": 1.948813962983718e-06,
      "loss": 0.7475,
      "step": 9781
    },
    {
      "epoch": 0.7177013307409411,
      "grad_norm": 1.3595641946585137,
      "learning_rate": 1.947872737266607e-06,
      "loss": 0.814,
      "step": 9782
    },
    {
      "epoch": 0.7177747003310803,
      "grad_norm": 0.5493740470991794,
      "learning_rate": 1.946931683910681e-06,
      "loss": 0.4753,
      "step": 9783
    },
    {
      "epoch": 0.7178480699212194,
      "grad_norm": 1.6674502651563476,
      "learning_rate": 1.945990802969084e-06,
      "loss": 0.7857,
      "step": 9784
    },
    {
      "epoch": 0.7178480699212194,
      "lm_loss": 0.2365,
      "step": 9784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7178480699212194,
      "lm_loss": 0.6122,
      "step": 9784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7178480699212194,
      "lm_loss": 0.4522,
      "step": 9784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7178480699212194,
      "lm_loss": 0.4873,
      "step": 9784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7178480699212194,
      "lm_loss": 0.569,
      "step": 9784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7178480699212194,
      "lm_loss": 0.5383,
      "step": 9784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7178480699212194,
      "lm_loss": 0.7075,
      "step": 9784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7178480699212194,
      "lm_loss": 0.4811,
      "step": 9784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7179214395113586,
      "grad_norm": 0.561706674907988,
      "learning_rate": 1.9450500944949487e-06,
      "loss": 0.4859,
      "step": 9785
    },
    {
      "epoch": 0.7179948091014977,
      "grad_norm": 0.5512794352184555,
      "learning_rate": 1.9441095585414005e-06,
      "loss": 0.4742,
      "step": 9786
    },
    {
      "epoch": 0.7180681786916367,
      "grad_norm": 1.5733474807337284,
      "learning_rate": 1.9431691951615513e-06,
      "loss": 0.7636,
      "step": 9787
    },
    {
      "epoch": 0.7181415482817759,
      "grad_norm": 1.5953503122322077,
      "learning_rate": 1.9422290044085096e-06,
      "loss": 0.6762,
      "step": 9788
    },
    {
      "epoch": 0.718214917871915,
      "grad_norm": 1.4284652552809327,
      "learning_rate": 1.9412889863353683e-06,
      "loss": 0.6684,
      "step": 9789
    },
    {
      "epoch": 0.7182882874620542,
      "grad_norm": 0.5347126538174548,
      "learning_rate": 1.9403491409952136e-06,
      "loss": 0.4839,
      "step": 9790
    },
    {
      "epoch": 0.7183616570521933,
      "grad_norm": 1.6560259376575932,
      "learning_rate": 1.939409468441119e-06,
      "loss": 0.7193,
      "step": 9791
    },
    {
      "epoch": 0.7184350266423324,
      "grad_norm": 1.5580847864162055,
      "learning_rate": 1.93846996872615e-06,
      "loss": 0.671,
      "step": 9792
    },
    {
      "epoch": 0.7184350266423324,
      "lm_loss": 1.2949,
      "step": 9792,
      "vm_loss": 0.0377
    },
    {
      "epoch": 0.7184350266423324,
      "lm_loss": 0.6261,
      "step": 9792,
      "vm_loss": 0.0477
    },
    {
      "epoch": 0.7184350266423324,
      "lm_loss": 0.5324,
      "step": 9792,
      "vm_loss": 0.2026
    },
    {
      "epoch": 0.7184350266423324,
      "lm_loss": 0.4066,
      "step": 9792,
      "vm_loss": 0.085
    },
    {
      "epoch": 0.7184350266423324,
      "lm_loss": 0.6046,
      "step": 9792,
      "vm_loss": 0.0603
    },
    {
      "epoch": 0.7184350266423324,
      "lm_loss": 2.0101,
      "step": 9792,
      "vm_loss": 0.1586
    },
    {
      "epoch": 0.7184350266423324,
      "lm_loss": 0.1069,
      "step": 9792,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.7184350266423324,
      "lm_loss": 0.288,
      "step": 9792,
      "vm_loss": 0.1333
    },
    {
      "epoch": 0.7185083962324715,
      "grad_norm": 1.6358541503958173,
      "learning_rate": 1.9375306419033646e-06,
      "loss": 0.7727,
      "step": 9793
    },
    {
      "epoch": 0.7185817658226107,
      "grad_norm": 1.7361476254014978,
      "learning_rate": 1.9365914880258085e-06,
      "loss": 0.6852,
      "step": 9794
    },
    {
      "epoch": 0.7186551354127498,
      "grad_norm": 1.5369555985696381,
      "learning_rate": 1.9356525071465178e-06,
      "loss": 0.8076,
      "step": 9795
    },
    {
      "epoch": 0.7187285050028889,
      "grad_norm": 1.3747211235164118,
      "learning_rate": 1.9347136993185184e-06,
      "loss": 0.7161,
      "step": 9796
    },
    {
      "epoch": 0.718801874593028,
      "grad_norm": 1.3407213967336735,
      "learning_rate": 1.9337750645948257e-06,
      "loss": 0.6129,
      "step": 9797
    },
    {
      "epoch": 0.7188752441831672,
      "grad_norm": 1.7749086838695145,
      "learning_rate": 1.93283660302845e-06,
      "loss": 0.7799,
      "step": 9798
    },
    {
      "epoch": 0.7189486137733063,
      "grad_norm": 0.5852635667312635,
      "learning_rate": 1.9318983146723893e-06,
      "loss": 0.4914,
      "step": 9799
    },
    {
      "epoch": 0.7190219833634455,
      "grad_norm": 1.4966546340269755,
      "learning_rate": 1.9309601995796263e-06,
      "loss": 0.7514,
      "step": 9800
    },
    {
      "epoch": 0.7190219833634455,
      "lm_loss": 1.2101,
      "step": 9800,
      "vm_loss": 0.0523
    },
    {
      "epoch": 0.7190219833634455,
      "lm_loss": 0.0521,
      "step": 9800,
      "vm_loss": 0.059
    },
    {
      "epoch": 0.7190219833634455,
      "lm_loss": 0.7599,
      "step": 9800,
      "vm_loss": 0.1431
    },
    {
      "epoch": 0.7190219833634455,
      "lm_loss": 0.2708,
      "step": 9800,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.7190219833634455,
      "lm_loss": 1.8019,
      "step": 9800,
      "vm_loss": 0.1244
    },
    {
      "epoch": 0.7190219833634455,
      "lm_loss": 0.3969,
      "step": 9800,
      "vm_loss": 0.1415
    },
    {
      "epoch": 0.7190219833634455,
      "lm_loss": 0.962,
      "step": 9800,
      "vm_loss": 0.1238
    },
    {
      "epoch": 0.7190219833634455,
      "lm_loss": 0.4751,
      "step": 9800,
      "vm_loss": 0.1774
    },
    {
      "epoch": 0.7190953529535845,
      "grad_norm": 1.5112308322841925,
      "learning_rate": 1.9300222578031413e-06,
      "loss": 0.6813,
      "step": 9801
    },
    {
      "epoch": 0.7191687225437237,
      "grad_norm": 0.5617141239008723,
      "learning_rate": 1.929084489395901e-06,
      "loss": 0.4902,
      "step": 9802
    },
    {
      "epoch": 0.7192420921338628,
      "grad_norm": 1.5390235172101507,
      "learning_rate": 1.9281468944108655e-06,
      "loss": 0.7384,
      "step": 9803
    },
    {
      "epoch": 0.719315461724002,
      "grad_norm": 1.553668864464114,
      "learning_rate": 1.9272094729009826e-06,
      "loss": 0.7049,
      "step": 9804
    },
    {
      "epoch": 0.7193888313141411,
      "grad_norm": 0.5398985204187939,
      "learning_rate": 1.9262722249191905e-06,
      "loss": 0.4859,
      "step": 9805
    },
    {
      "epoch": 0.7194622009042801,
      "grad_norm": 0.5458866101816136,
      "learning_rate": 1.9253351505184183e-06,
      "loss": 0.4731,
      "step": 9806
    },
    {
      "epoch": 0.7195355704944193,
      "grad_norm": 1.4659329841371271,
      "learning_rate": 1.9243982497515817e-06,
      "loss": 0.744,
      "step": 9807
    },
    {
      "epoch": 0.7196089400845584,
      "grad_norm": 0.5614568922047207,
      "learning_rate": 1.9234615226715945e-06,
      "loss": 0.4958,
      "step": 9808
    },
    {
      "epoch": 0.7196089400845584,
      "lm_loss": 0.6441,
      "step": 9808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7196089400845584,
      "lm_loss": 0.404,
      "step": 9808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7196089400845584,
      "lm_loss": 0.4888,
      "step": 9808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7196089400845584,
      "lm_loss": 0.4301,
      "step": 9808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7196089400845584,
      "lm_loss": 0.4791,
      "step": 9808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7196089400845584,
      "lm_loss": 0.3942,
      "step": 9808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7196089400845584,
      "lm_loss": 0.2225,
      "step": 9808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7196089400845584,
      "lm_loss": 0.5032,
      "step": 9808,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7196823096746976,
      "grad_norm": 0.5464882085505416,
      "learning_rate": 1.9225249693313547e-06,
      "loss": 0.4676,
      "step": 9809
    },
    {
      "epoch": 0.7197556792648367,
      "grad_norm": 0.5469065384645841,
      "learning_rate": 1.9215885897837506e-06,
      "loss": 0.469,
      "step": 9810
    },
    {
      "epoch": 0.7198290488549758,
      "grad_norm": 1.5391207517289307,
      "learning_rate": 1.9206523840816626e-06,
      "loss": 0.8345,
      "step": 9811
    },
    {
      "epoch": 0.7199024184451149,
      "grad_norm": 1.5031522411086444,
      "learning_rate": 1.919716352277958e-06,
      "loss": 0.7684,
      "step": 9812
    },
    {
      "epoch": 0.7199757880352541,
      "grad_norm": 0.5607311195748919,
      "learning_rate": 1.918780494425502e-06,
      "loss": 0.4831,
      "step": 9813
    },
    {
      "epoch": 0.7200491576253932,
      "grad_norm": 1.5337397151067511,
      "learning_rate": 1.917844810577141e-06,
      "loss": 0.767,
      "step": 9814
    },
    {
      "epoch": 0.7201225272155324,
      "grad_norm": 1.511294495219173,
      "learning_rate": 1.916909300785717e-06,
      "loss": 0.7532,
      "step": 9815
    },
    {
      "epoch": 0.7201958968056714,
      "grad_norm": 1.4876507163000499,
      "learning_rate": 1.915973965104059e-06,
      "loss": 0.8153,
      "step": 9816
    },
    {
      "epoch": 0.7201958968056714,
      "lm_loss": 0.4897,
      "step": 9816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7201958968056714,
      "lm_loss": 0.4931,
      "step": 9816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7201958968056714,
      "lm_loss": 0.3771,
      "step": 9816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7201958968056714,
      "lm_loss": 0.4867,
      "step": 9816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7201958968056714,
      "lm_loss": 0.4665,
      "step": 9816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7201958968056714,
      "lm_loss": 0.6021,
      "step": 9816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7201958968056714,
      "lm_loss": 0.4716,
      "step": 9816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7201958968056714,
      "lm_loss": 0.7109,
      "step": 9816,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7202692663958106,
      "grad_norm": 0.5476515550352345,
      "learning_rate": 1.91503880358499e-06,
      "loss": 0.4776,
      "step": 9817
    },
    {
      "epoch": 0.7203426359859497,
      "grad_norm": 1.4847567227393261,
      "learning_rate": 1.914103816281317e-06,
      "loss": 0.6798,
      "step": 9818
    },
    {
      "epoch": 0.7204160055760889,
      "grad_norm": 1.5414876631014238,
      "learning_rate": 1.9131690032458454e-06,
      "loss": 0.7186,
      "step": 9819
    },
    {
      "epoch": 0.720489375166228,
      "grad_norm": 1.5572682093289003,
      "learning_rate": 1.9122343645313648e-06,
      "loss": 0.7197,
      "step": 9820
    },
    {
      "epoch": 0.7205627447563671,
      "grad_norm": 0.5613674467815896,
      "learning_rate": 1.9112999001906564e-06,
      "loss": 0.4813,
      "step": 9821
    },
    {
      "epoch": 0.7206361143465062,
      "grad_norm": 1.4721490679865514,
      "learning_rate": 1.910365610276491e-06,
      "loss": 0.7426,
      "step": 9822
    },
    {
      "epoch": 0.7207094839366454,
      "grad_norm": 1.3188747258570983,
      "learning_rate": 1.9094314948416297e-06,
      "loss": 0.6412,
      "step": 9823
    },
    {
      "epoch": 0.7207828535267845,
      "grad_norm": 0.541529054627036,
      "learning_rate": 1.908497553938827e-06,
      "loss": 0.488,
      "step": 9824
    },
    {
      "epoch": 0.7207828535267845,
      "lm_loss": 1.1461,
      "step": 9824,
      "vm_loss": 0.0861
    },
    {
      "epoch": 0.7207828535267845,
      "lm_loss": 1.3556,
      "step": 9824,
      "vm_loss": 0.0498
    },
    {
      "epoch": 0.7207828535267845,
      "lm_loss": 0.8802,
      "step": 9824,
      "vm_loss": 0.1689
    },
    {
      "epoch": 0.7207828535267845,
      "lm_loss": 0.8373,
      "step": 9824,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.7207828535267845,
      "lm_loss": 0.1255,
      "step": 9824,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.7207828535267845,
      "lm_loss": 0.0454,
      "step": 9824,
      "vm_loss": 0.0966
    },
    {
      "epoch": 0.7207828535267845,
      "lm_loss": 0.9569,
      "step": 9824,
      "vm_loss": 0.1069
    },
    {
      "epoch": 0.7207828535267845,
      "lm_loss": 0.6949,
      "step": 9824,
      "vm_loss": 0.267
    },
    {
      "epoch": 0.7208562231169237,
      "grad_norm": 1.4796257288543397,
      "learning_rate": 1.907563787620824e-06,
      "loss": 0.8762,
      "step": 9825
    },
    {
      "epoch": 0.7209295927070627,
      "grad_norm": 1.420490891194973,
      "learning_rate": 1.906630195940352e-06,
      "loss": 0.7879,
      "step": 9826
    },
    {
      "epoch": 0.7210029622972018,
      "grad_norm": 1.5327327773439472,
      "learning_rate": 1.9056967789501336e-06,
      "loss": 0.7628,
      "step": 9827
    },
    {
      "epoch": 0.721076331887341,
      "grad_norm": 1.4958818471480908,
      "learning_rate": 1.9047635367028805e-06,
      "loss": 0.6842,
      "step": 9828
    },
    {
      "epoch": 0.7211497014774801,
      "grad_norm": 0.5455780086038389,
      "learning_rate": 1.9038304692512943e-06,
      "loss": 0.4677,
      "step": 9829
    },
    {
      "epoch": 0.7212230710676193,
      "grad_norm": 1.483201539652911,
      "learning_rate": 1.9028975766480705e-06,
      "loss": 0.7426,
      "step": 9830
    },
    {
      "epoch": 0.7212964406577583,
      "grad_norm": 0.5597240417644669,
      "learning_rate": 1.9019648589458905e-06,
      "loss": 0.4949,
      "step": 9831
    },
    {
      "epoch": 0.7213698102478975,
      "grad_norm": 1.4740895819672433,
      "learning_rate": 1.901032316197427e-06,
      "loss": 0.7879,
      "step": 9832
    },
    {
      "epoch": 0.7213698102478975,
      "lm_loss": 0.2818,
      "step": 9832,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.7213698102478975,
      "lm_loss": 0.0836,
      "step": 9832,
      "vm_loss": 0.0634
    },
    {
      "epoch": 0.7213698102478975,
      "lm_loss": 0.0766,
      "step": 9832,
      "vm_loss": 0.1157
    },
    {
      "epoch": 0.7213698102478975,
      "lm_loss": 0.8071,
      "step": 9832,
      "vm_loss": 0.2138
    },
    {
      "epoch": 0.7213698102478975,
      "lm_loss": 0.9527,
      "step": 9832,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.7213698102478975,
      "lm_loss": 1.0451,
      "step": 9832,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.7213698102478975,
      "lm_loss": 0.4448,
      "step": 9832,
      "vm_loss": 0.2048
    },
    {
      "epoch": 0.7213698102478975,
      "lm_loss": 0.6779,
      "step": 9832,
      "vm_loss": 0.1349
    },
    {
      "epoch": 0.7214431798380366,
      "grad_norm": 1.3545997375452812,
      "learning_rate": 1.9000999484553434e-06,
      "loss": 0.6498,
      "step": 9833
    },
    {
      "epoch": 0.7215165494281758,
      "grad_norm": 1.4778634356707365,
      "learning_rate": 1.8991677557722903e-06,
      "loss": 0.7205,
      "step": 9834
    },
    {
      "epoch": 0.7215899190183149,
      "grad_norm": 1.4537955355615544,
      "learning_rate": 1.8982357382009153e-06,
      "loss": 0.7591,
      "step": 9835
    },
    {
      "epoch": 0.721663288608454,
      "grad_norm": 1.4443745434919926,
      "learning_rate": 1.8973038957938495e-06,
      "loss": 0.7869,
      "step": 9836
    },
    {
      "epoch": 0.7217366581985931,
      "grad_norm": 1.5386757216068616,
      "learning_rate": 1.8963722286037162e-06,
      "loss": 0.723,
      "step": 9837
    },
    {
      "epoch": 0.7218100277887323,
      "grad_norm": 0.5372025909783403,
      "learning_rate": 1.895440736683129e-06,
      "loss": 0.4643,
      "step": 9838
    },
    {
      "epoch": 0.7218833973788714,
      "grad_norm": 1.4690723073411516,
      "learning_rate": 1.8945094200846903e-06,
      "loss": 0.7805,
      "step": 9839
    },
    {
      "epoch": 0.7219567669690106,
      "grad_norm": 1.5617256604557193,
      "learning_rate": 1.8935782788609969e-06,
      "loss": 0.8788,
      "step": 9840
    },
    {
      "epoch": 0.7219567669690106,
      "lm_loss": 1.3667,
      "step": 9840,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.7219567669690106,
      "lm_loss": 1.1859,
      "step": 9840,
      "vm_loss": 0.1745
    },
    {
      "epoch": 0.7219567669690106,
      "lm_loss": 0.002,
      "step": 9840,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.7219567669690106,
      "lm_loss": 0.4712,
      "step": 9840,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.7219567669690106,
      "lm_loss": 1.1765,
      "step": 9840,
      "vm_loss": 0.0796
    },
    {
      "epoch": 0.7219567669690106,
      "lm_loss": 0.8859,
      "step": 9840,
      "vm_loss": 0.0775
    },
    {
      "epoch": 0.7219567669690106,
      "lm_loss": 0.8954,
      "step": 9840,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.7219567669690106,
      "lm_loss": 0.1556,
      "step": 9840,
      "vm_loss": 0.1861
    },
    {
      "epoch": 0.7220301365591496,
      "grad_norm": 1.5144715983173838,
      "learning_rate": 1.892647313064631e-06,
      "loss": 0.7576,
      "step": 9841
    },
    {
      "epoch": 0.7221035061492888,
      "grad_norm": 1.5163210633990345,
      "learning_rate": 1.8917165227481665e-06,
      "loss": 0.7044,
      "step": 9842
    },
    {
      "epoch": 0.7221768757394279,
      "grad_norm": 1.5407478256126907,
      "learning_rate": 1.8907859079641676e-06,
      "loss": 0.8131,
      "step": 9843
    },
    {
      "epoch": 0.722250245329567,
      "grad_norm": 1.4685614777154556,
      "learning_rate": 1.8898554687651883e-06,
      "loss": 0.7279,
      "step": 9844
    },
    {
      "epoch": 0.7223236149197062,
      "grad_norm": 1.2460882220184193,
      "learning_rate": 1.888925205203771e-06,
      "loss": 0.5454,
      "step": 9845
    },
    {
      "epoch": 0.7223969845098452,
      "grad_norm": 1.5486260863860382,
      "learning_rate": 1.8879951173324534e-06,
      "loss": 0.6758,
      "step": 9846
    },
    {
      "epoch": 0.7224703540999844,
      "grad_norm": 1.5313842153659758,
      "learning_rate": 1.8870652052037586e-06,
      "loss": 0.724,
      "step": 9847
    },
    {
      "epoch": 0.7225437236901235,
      "grad_norm": 1.4157713133713272,
      "learning_rate": 1.8861354688702005e-06,
      "loss": 0.7005,
      "step": 9848
    },
    {
      "epoch": 0.7225437236901235,
      "lm_loss": 0.157,
      "step": 9848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7225437236901235,
      "lm_loss": 0.3569,
      "step": 9848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7225437236901235,
      "lm_loss": 0.4613,
      "step": 9848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7225437236901235,
      "lm_loss": 0.5056,
      "step": 9848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7225437236901235,
      "lm_loss": 0.3487,
      "step": 9848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7225437236901235,
      "lm_loss": 0.4576,
      "step": 9848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7225437236901235,
      "lm_loss": 0.6002,
      "step": 9848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7225437236901235,
      "lm_loss": 0.6094,
      "step": 9848,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7226170932802627,
      "grad_norm": 0.5378632671209056,
      "learning_rate": 1.8852059083842838e-06,
      "loss": 0.4755,
      "step": 9849
    },
    {
      "epoch": 0.7226904628704018,
      "grad_norm": 0.5473589505417098,
      "learning_rate": 1.8842765237985017e-06,
      "loss": 0.489,
      "step": 9850
    },
    {
      "epoch": 0.7227638324605409,
      "grad_norm": 1.49908876046192,
      "learning_rate": 1.883347315165342e-06,
      "loss": 0.832,
      "step": 9851
    },
    {
      "epoch": 0.72283720205068,
      "grad_norm": 0.5476670105930213,
      "learning_rate": 1.8824182825372772e-06,
      "loss": 0.4731,
      "step": 9852
    },
    {
      "epoch": 0.7229105716408192,
      "grad_norm": 1.5320942439731522,
      "learning_rate": 1.8814894259667727e-06,
      "loss": 0.6794,
      "step": 9853
    },
    {
      "epoch": 0.7229839412309583,
      "grad_norm": 1.7272747839454419,
      "learning_rate": 1.8805607455062835e-06,
      "loss": 0.7512,
      "step": 9854
    },
    {
      "epoch": 0.7230573108210975,
      "grad_norm": 1.5693422476588212,
      "learning_rate": 1.879632241208254e-06,
      "loss": 0.7738,
      "step": 9855
    },
    {
      "epoch": 0.7231306804112365,
      "grad_norm": 0.5485206917031124,
      "learning_rate": 1.8787039131251172e-06,
      "loss": 0.4781,
      "step": 9856
    },
    {
      "epoch": 0.7231306804112365,
      "lm_loss": 1.081,
      "step": 9856,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.7231306804112365,
      "lm_loss": 1.2836,
      "step": 9856,
      "vm_loss": 0.1249
    },
    {
      "epoch": 0.7231306804112365,
      "lm_loss": 0.8275,
      "step": 9856,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.7231306804112365,
      "lm_loss": 0.8478,
      "step": 9856,
      "vm_loss": 0.1763
    },
    {
      "epoch": 0.7231306804112365,
      "lm_loss": 2.0088,
      "step": 9856,
      "vm_loss": 0.2311
    },
    {
      "epoch": 0.7231306804112365,
      "lm_loss": 0.146,
      "step": 9856,
      "vm_loss": 0.1655
    },
    {
      "epoch": 0.7231306804112365,
      "lm_loss": 0.7735,
      "step": 9856,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.7231306804112365,
      "lm_loss": 0.1816,
      "step": 9856,
      "vm_loss": 0.0695
    },
    {
      "epoch": 0.7232040500013757,
      "grad_norm": 1.4025355715761976,
      "learning_rate": 1.8777757613093022e-06,
      "loss": 0.7943,
      "step": 9857
    },
    {
      "epoch": 0.7232774195915148,
      "grad_norm": 0.5692786733033529,
      "learning_rate": 1.8768477858132223e-06,
      "loss": 0.5049,
      "step": 9858
    },
    {
      "epoch": 0.723350789181654,
      "grad_norm": 1.5146466885615746,
      "learning_rate": 1.875919986689282e-06,
      "loss": 0.7427,
      "step": 9859
    },
    {
      "epoch": 0.7234241587717931,
      "grad_norm": 0.54367267524396,
      "learning_rate": 1.8749923639898766e-06,
      "loss": 0.4951,
      "step": 9860
    },
    {
      "epoch": 0.7234975283619322,
      "grad_norm": 1.501359003027669,
      "learning_rate": 1.8740649177673897e-06,
      "loss": 0.6848,
      "step": 9861
    },
    {
      "epoch": 0.7235708979520713,
      "grad_norm": 0.544699083949236,
      "learning_rate": 1.8731376480742002e-06,
      "loss": 0.4689,
      "step": 9862
    },
    {
      "epoch": 0.7236442675422105,
      "grad_norm": 0.5583556286913164,
      "learning_rate": 1.8722105549626707e-06,
      "loss": 0.4833,
      "step": 9863
    },
    {
      "epoch": 0.7237176371323496,
      "grad_norm": 0.5729644179145702,
      "learning_rate": 1.8712836384851573e-06,
      "loss": 0.5106,
      "step": 9864
    },
    {
      "epoch": 0.7237176371323496,
      "lm_loss": 0.4677,
      "step": 9864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7237176371323496,
      "lm_loss": 0.3728,
      "step": 9864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7237176371323496,
      "lm_loss": 0.4629,
      "step": 9864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7237176371323496,
      "lm_loss": 0.397,
      "step": 9864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7237176371323496,
      "lm_loss": 0.4835,
      "step": 9864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7237176371323496,
      "lm_loss": 0.5775,
      "step": 9864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7237176371323496,
      "lm_loss": 0.6364,
      "step": 9864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7237176371323496,
      "lm_loss": 0.4876,
      "step": 9864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7237910067224888,
      "grad_norm": 0.5553261310135708,
      "learning_rate": 1.870356898694005e-06,
      "loss": 0.482,
      "step": 9865
    },
    {
      "epoch": 0.7238643763126278,
      "grad_norm": 1.4118924026926003,
      "learning_rate": 1.8694303356415473e-06,
      "loss": 0.6805,
      "step": 9866
    },
    {
      "epoch": 0.7239377459027669,
      "grad_norm": 1.5866978162596217,
      "learning_rate": 1.8685039493801132e-06,
      "loss": 0.7176,
      "step": 9867
    },
    {
      "epoch": 0.7240111154929061,
      "grad_norm": 1.4389440273908833,
      "learning_rate": 1.867577739962016e-06,
      "loss": 0.6656,
      "step": 9868
    },
    {
      "epoch": 0.7240844850830452,
      "grad_norm": 1.4522298426749904,
      "learning_rate": 1.8666517074395607e-06,
      "loss": 0.7571,
      "step": 9869
    },
    {
      "epoch": 0.7241578546731844,
      "grad_norm": 1.3730994482591943,
      "learning_rate": 1.8657258518650435e-06,
      "loss": 0.7024,
      "step": 9870
    },
    {
      "epoch": 0.7242312242633234,
      "grad_norm": 1.4706005834646265,
      "learning_rate": 1.8648001732907494e-06,
      "loss": 0.839,
      "step": 9871
    },
    {
      "epoch": 0.7243045938534626,
      "grad_norm": 1.5811344807233512,
      "learning_rate": 1.8638746717689515e-06,
      "loss": 0.7938,
      "step": 9872
    },
    {
      "epoch": 0.7243045938534626,
      "lm_loss": 1.4516,
      "step": 9872,
      "vm_loss": 0.0189
    },
    {
      "epoch": 0.7243045938534626,
      "lm_loss": 0.649,
      "step": 9872,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.7243045938534626,
      "lm_loss": 0.4572,
      "step": 9872,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.7243045938534626,
      "lm_loss": 0.757,
      "step": 9872,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.7243045938534626,
      "lm_loss": 0.3824,
      "step": 9872,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.7243045938534626,
      "lm_loss": 0.5222,
      "step": 9872,
      "vm_loss": 0.138
    },
    {
      "epoch": 0.7243045938534626,
      "lm_loss": 0.3819,
      "step": 9872,
      "vm_loss": 0.1672
    },
    {
      "epoch": 0.7243045938534626,
      "lm_loss": 0.0715,
      "step": 9872,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.7243779634436017,
      "grad_norm": 1.635521584102396,
      "learning_rate": 1.8629493473519201e-06,
      "loss": 0.7785,
      "step": 9873
    },
    {
      "epoch": 0.7244513330337409,
      "grad_norm": 0.5709683905165047,
      "learning_rate": 1.862024200091907e-06,
      "loss": 0.4822,
      "step": 9874
    },
    {
      "epoch": 0.7245247026238799,
      "grad_norm": 0.550299912884194,
      "learning_rate": 1.8610992300411584e-06,
      "loss": 0.4463,
      "step": 9875
    },
    {
      "epoch": 0.7245980722140191,
      "grad_norm": 0.56041348457022,
      "learning_rate": 1.8601744372519098e-06,
      "loss": 0.5032,
      "step": 9876
    },
    {
      "epoch": 0.7246714418041582,
      "grad_norm": 1.7476109635692787,
      "learning_rate": 1.8592498217763844e-06,
      "loss": 0.76,
      "step": 9877
    },
    {
      "epoch": 0.7247448113942974,
      "grad_norm": 1.6581597869099216,
      "learning_rate": 1.8583253836668018e-06,
      "loss": 0.6579,
      "step": 9878
    },
    {
      "epoch": 0.7248181809844365,
      "grad_norm": 0.5363710827645476,
      "learning_rate": 1.8574011229753646e-06,
      "loss": 0.456,
      "step": 9879
    },
    {
      "epoch": 0.7248915505745755,
      "grad_norm": 0.5502583214580227,
      "learning_rate": 1.8564770397542686e-06,
      "loss": 0.487,
      "step": 9880
    },
    {
      "epoch": 0.7248915505745755,
      "lm_loss": 1.374,
      "step": 9880,
      "vm_loss": 0.0602
    },
    {
      "epoch": 0.7248915505745755,
      "lm_loss": 1.4843,
      "step": 9880,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.7248915505745755,
      "lm_loss": 0.6919,
      "step": 9880,
      "vm_loss": 0.0477
    },
    {
      "epoch": 0.7248915505745755,
      "lm_loss": 0.0747,
      "step": 9880,
      "vm_loss": 0.0988
    },
    {
      "epoch": 0.7248915505745755,
      "lm_loss": 0.4001,
      "step": 9880,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.7248915505745755,
      "lm_loss": 0.1432,
      "step": 9880,
      "vm_loss": 0.1648
    },
    {
      "epoch": 0.7248915505745755,
      "lm_loss": 0.9824,
      "step": 9880,
      "vm_loss": 0.185
    },
    {
      "epoch": 0.7248915505745755,
      "lm_loss": 0.0589,
      "step": 9880,
      "vm_loss": 0.1899
    },
    {
      "epoch": 0.7249649201647147,
      "grad_norm": 1.4599935586630477,
      "learning_rate": 1.8555531340556987e-06,
      "loss": 0.7364,
      "step": 9881
    },
    {
      "epoch": 0.7250382897548538,
      "grad_norm": 0.5510520337829726,
      "learning_rate": 1.854629405931831e-06,
      "loss": 0.477,
      "step": 9882
    },
    {
      "epoch": 0.725111659344993,
      "grad_norm": 1.6330850702768651,
      "learning_rate": 1.853705855434828e-06,
      "loss": 0.7772,
      "step": 9883
    },
    {
      "epoch": 0.7251850289351321,
      "grad_norm": 0.5566111607745451,
      "learning_rate": 1.8527824826168488e-06,
      "loss": 0.4647,
      "step": 9884
    },
    {
      "epoch": 0.7252583985252712,
      "grad_norm": 0.5505695284178027,
      "learning_rate": 1.8518592875300368e-06,
      "loss": 0.4814,
      "step": 9885
    },
    {
      "epoch": 0.7253317681154103,
      "grad_norm": 1.5529261054935295,
      "learning_rate": 1.850936270226527e-06,
      "loss": 0.8063,
      "step": 9886
    },
    {
      "epoch": 0.7254051377055495,
      "grad_norm": 1.7513366857318287,
      "learning_rate": 1.8500134307584444e-06,
      "loss": 0.7958,
      "step": 9887
    },
    {
      "epoch": 0.7254785072956886,
      "grad_norm": 1.5034025776296123,
      "learning_rate": 1.8490907691779026e-06,
      "loss": 0.7279,
      "step": 9888
    },
    {
      "epoch": 0.7254785072956886,
      "lm_loss": 0.8858,
      "step": 9888,
      "vm_loss": 0.0506
    },
    {
      "epoch": 0.7254785072956886,
      "lm_loss": 0.2628,
      "step": 9888,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.7254785072956886,
      "lm_loss": 1.1394,
      "step": 9888,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.7254785072956886,
      "lm_loss": 0.9697,
      "step": 9888,
      "vm_loss": 0.1259
    },
    {
      "epoch": 0.7254785072956886,
      "lm_loss": 0.3995,
      "step": 9888,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.7254785072956886,
      "lm_loss": 0.9319,
      "step": 9888,
      "vm_loss": 0.2024
    },
    {
      "epoch": 0.7254785072956886,
      "lm_loss": 0.1929,
      "step": 9888,
      "vm_loss": 0.1443
    },
    {
      "epoch": 0.7254785072956886,
      "lm_loss": 0.5332,
      "step": 9888,
      "vm_loss": 0.1687
    },
    {
      "epoch": 0.7255518768858278,
      "grad_norm": 1.5281272128390169,
      "learning_rate": 1.8481682855370098e-06,
      "loss": 0.7317,
      "step": 9889
    },
    {
      "epoch": 0.7256252464759668,
      "grad_norm": 1.5461650534970195,
      "learning_rate": 1.8472459798878594e-06,
      "loss": 0.6957,
      "step": 9890
    },
    {
      "epoch": 0.725698616066106,
      "grad_norm": 1.7083636247379554,
      "learning_rate": 1.846323852282536e-06,
      "loss": 0.8463,
      "step": 9891
    },
    {
      "epoch": 0.7257719856562451,
      "grad_norm": 0.5567849600094578,
      "learning_rate": 1.8454019027731152e-06,
      "loss": 0.4751,
      "step": 9892
    },
    {
      "epoch": 0.7258453552463843,
      "grad_norm": 1.5329777121068455,
      "learning_rate": 1.844480131411659e-06,
      "loss": 0.7547,
      "step": 9893
    },
    {
      "epoch": 0.7259187248365234,
      "grad_norm": 0.5577010384919606,
      "learning_rate": 1.8435585382502258e-06,
      "loss": 0.4885,
      "step": 9894
    },
    {
      "epoch": 0.7259920944266625,
      "grad_norm": 1.4504291986737792,
      "learning_rate": 1.842637123340859e-06,
      "loss": 0.7211,
      "step": 9895
    },
    {
      "epoch": 0.7260654640168016,
      "grad_norm": 1.7135857313005893,
      "learning_rate": 1.8417158867355934e-06,
      "loss": 0.8105,
      "step": 9896
    },
    {
      "epoch": 0.7260654640168016,
      "lm_loss": 0.6944,
      "step": 9896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7260654640168016,
      "lm_loss": 0.4246,
      "step": 9896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7260654640168016,
      "lm_loss": 0.4728,
      "step": 9896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7260654640168016,
      "lm_loss": 0.4296,
      "step": 9896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7260654640168016,
      "lm_loss": 0.6471,
      "step": 9896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7260654640168016,
      "lm_loss": 0.5374,
      "step": 9896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7260654640168016,
      "lm_loss": 0.5787,
      "step": 9896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7260654640168016,
      "lm_loss": 0.5015,
      "step": 9896,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7261388336069408,
      "grad_norm": 0.5495279066516421,
      "learning_rate": 1.8407948284864524e-06,
      "loss": 0.4814,
      "step": 9897
    },
    {
      "epoch": 0.7262122031970799,
      "grad_norm": 1.5881565011020438,
      "learning_rate": 1.8398739486454509e-06,
      "loss": 0.6541,
      "step": 9898
    },
    {
      "epoch": 0.7262855727872191,
      "grad_norm": 0.54368107524251,
      "learning_rate": 1.8389532472645921e-06,
      "loss": 0.477,
      "step": 9899
    },
    {
      "epoch": 0.7263589423773581,
      "grad_norm": 0.5578175147396558,
      "learning_rate": 1.8380327243958735e-06,
      "loss": 0.4876,
      "step": 9900
    },
    {
      "epoch": 0.7264323119674972,
      "grad_norm": 1.5207946374281116,
      "learning_rate": 1.837112380091277e-06,
      "loss": 0.7179,
      "step": 9901
    },
    {
      "epoch": 0.7265056815576364,
      "grad_norm": 0.5516192748223031,
      "learning_rate": 1.8361922144027777e-06,
      "loss": 0.4813,
      "step": 9902
    },
    {
      "epoch": 0.7265790511477755,
      "grad_norm": 0.5367626541772875,
      "learning_rate": 1.8352722273823386e-06,
      "loss": 0.4744,
      "step": 9903
    },
    {
      "epoch": 0.7266524207379147,
      "grad_norm": 1.4559587832938659,
      "learning_rate": 1.8343524190819128e-06,
      "loss": 0.7956,
      "step": 9904
    },
    {
      "epoch": 0.7266524207379147,
      "lm_loss": 0.2023,
      "step": 9904,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7266524207379147,
      "lm_loss": 0.3598,
      "step": 9904,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7266524207379147,
      "lm_loss": 0.3945,
      "step": 9904,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7266524207379147,
      "lm_loss": 0.3833,
      "step": 9904,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7266524207379147,
      "lm_loss": 0.4189,
      "step": 9904,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7266524207379147,
      "lm_loss": 0.7049,
      "step": 9904,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7266524207379147,
      "lm_loss": 0.4812,
      "step": 9904,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7266524207379147,
      "lm_loss": 0.4756,
      "step": 9904,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7267257903280537,
      "grad_norm": 0.5283249526109073,
      "learning_rate": 1.833432789553447e-06,
      "loss": 0.4693,
      "step": 9905
    },
    {
      "epoch": 0.7267991599181929,
      "grad_norm": 1.5393999206094706,
      "learning_rate": 1.8325133388488737e-06,
      "loss": 0.6405,
      "step": 9906
    },
    {
      "epoch": 0.726872529508332,
      "grad_norm": 0.5608596597840424,
      "learning_rate": 1.8315940670201155e-06,
      "loss": 0.4756,
      "step": 9907
    },
    {
      "epoch": 0.7269458990984712,
      "grad_norm": 1.4771607068875865,
      "learning_rate": 1.8306749741190877e-06,
      "loss": 0.7671,
      "step": 9908
    },
    {
      "epoch": 0.7270192686886103,
      "grad_norm": 1.6375328060897762,
      "learning_rate": 1.829756060197692e-06,
      "loss": 0.8274,
      "step": 9909
    },
    {
      "epoch": 0.7270926382787494,
      "grad_norm": 1.34426506288744,
      "learning_rate": 1.8288373253078217e-06,
      "loss": 0.7353,
      "step": 9910
    },
    {
      "epoch": 0.7271660078688885,
      "grad_norm": 1.4194513327701166,
      "learning_rate": 1.8279187695013617e-06,
      "loss": 0.6701,
      "step": 9911
    },
    {
      "epoch": 0.7272393774590277,
      "grad_norm": 0.5261321661130357,
      "learning_rate": 1.827000392830185e-06,
      "loss": 0.4821,
      "step": 9912
    },
    {
      "epoch": 0.7272393774590277,
      "lm_loss": 0.7392,
      "step": 9912,
      "vm_loss": 0.0672
    },
    {
      "epoch": 0.7272393774590277,
      "lm_loss": 0.1716,
      "step": 9912,
      "vm_loss": 0.0631
    },
    {
      "epoch": 0.7272393774590277,
      "lm_loss": 0.3594,
      "step": 9912,
      "vm_loss": 0.1035
    },
    {
      "epoch": 0.7272393774590277,
      "lm_loss": 0.5925,
      "step": 9912,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.7272393774590277,
      "lm_loss": 0.6295,
      "step": 9912,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.7272393774590277,
      "lm_loss": 0.9811,
      "step": 9912,
      "vm_loss": 0.111
    },
    {
      "epoch": 0.7272393774590277,
      "lm_loss": 0.9495,
      "step": 9912,
      "vm_loss": 0.1218
    },
    {
      "epoch": 0.7272393774590277,
      "lm_loss": 0.5131,
      "step": 9912,
      "vm_loss": 0.0823
    },
    {
      "epoch": 0.7273127470491668,
      "grad_norm": 1.5945251094097794,
      "learning_rate": 1.8260821953461539e-06,
      "loss": 0.7003,
      "step": 9913
    },
    {
      "epoch": 0.727386116639306,
      "grad_norm": 0.564019408066403,
      "learning_rate": 1.8251641771011208e-06,
      "loss": 0.4765,
      "step": 9914
    },
    {
      "epoch": 0.727459486229445,
      "grad_norm": 1.4901935955988768,
      "learning_rate": 1.8242463381469278e-06,
      "loss": 0.7403,
      "step": 9915
    },
    {
      "epoch": 0.7275328558195842,
      "grad_norm": 1.3652314287745517,
      "learning_rate": 1.8233286785354098e-06,
      "loss": 0.6548,
      "step": 9916
    },
    {
      "epoch": 0.7276062254097233,
      "grad_norm": 1.3923267034227165,
      "learning_rate": 1.822411198318389e-06,
      "loss": 0.7892,
      "step": 9917
    },
    {
      "epoch": 0.7276795949998625,
      "grad_norm": 1.4574947574287107,
      "learning_rate": 1.8214938975476765e-06,
      "loss": 0.7515,
      "step": 9918
    },
    {
      "epoch": 0.7277529645900016,
      "grad_norm": 0.537166545845892,
      "learning_rate": 1.8205767762750752e-06,
      "loss": 0.4762,
      "step": 9919
    },
    {
      "epoch": 0.7278263341801406,
      "grad_norm": 0.5452874145264683,
      "learning_rate": 1.8196598345523753e-06,
      "loss": 0.4685,
      "step": 9920
    },
    {
      "epoch": 0.7278263341801406,
      "lm_loss": 0.2626,
      "step": 9920,
      "vm_loss": 0.0506
    },
    {
      "epoch": 0.7278263341801406,
      "lm_loss": 1.0964,
      "step": 9920,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.7278263341801406,
      "lm_loss": 0.5888,
      "step": 9920,
      "vm_loss": 0.1419
    },
    {
      "epoch": 0.7278263341801406,
      "lm_loss": 0.8202,
      "step": 9920,
      "vm_loss": 0.201
    },
    {
      "epoch": 0.7278263341801406,
      "lm_loss": 0.9342,
      "step": 9920,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.7278263341801406,
      "lm_loss": 0.7231,
      "step": 9920,
      "vm_loss": 0.1784
    },
    {
      "epoch": 0.7278263341801406,
      "lm_loss": 0.7896,
      "step": 9920,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.7278263341801406,
      "lm_loss": 0.233,
      "step": 9920,
      "vm_loss": 0.1263
    },
    {
      "epoch": 0.7278997037702798,
      "grad_norm": 1.4859887773688454,
      "learning_rate": 1.8187430724313621e-06,
      "loss": 0.6519,
      "step": 9921
    },
    {
      "epoch": 0.7279730733604189,
      "grad_norm": 1.44283658628558,
      "learning_rate": 1.817826489963806e-06,
      "loss": 0.6587,
      "step": 9922
    },
    {
      "epoch": 0.7280464429505581,
      "grad_norm": 0.5364173856034312,
      "learning_rate": 1.8169100872014682e-06,
      "loss": 0.4782,
      "step": 9923
    },
    {
      "epoch": 0.7281198125406972,
      "grad_norm": 1.5805237436469217,
      "learning_rate": 1.815993864196101e-06,
      "loss": 0.7001,
      "step": 9924
    },
    {
      "epoch": 0.7281931821308363,
      "grad_norm": 1.5744904030646578,
      "learning_rate": 1.815077820999445e-06,
      "loss": 0.7152,
      "step": 9925
    },
    {
      "epoch": 0.7282665517209754,
      "grad_norm": 1.5892405596263917,
      "learning_rate": 1.8141619576632307e-06,
      "loss": 0.7396,
      "step": 9926
    },
    {
      "epoch": 0.7283399213111146,
      "grad_norm": 1.4265254793084718,
      "learning_rate": 1.8132462742391816e-06,
      "loss": 0.7699,
      "step": 9927
    },
    {
      "epoch": 0.7284132909012537,
      "grad_norm": 1.5135961354704885,
      "learning_rate": 1.8123307707790071e-06,
      "loss": 0.7027,
      "step": 9928
    },
    {
      "epoch": 0.7284132909012537,
      "lm_loss": 1.2583,
      "step": 9928,
      "vm_loss": 0.0605
    },
    {
      "epoch": 0.7284132909012537,
      "lm_loss": 0.7548,
      "step": 9928,
      "vm_loss": 0.0705
    },
    {
      "epoch": 0.7284132909012537,
      "lm_loss": 0.4201,
      "step": 9928,
      "vm_loss": 0.1161
    },
    {
      "epoch": 0.7284132909012537,
      "lm_loss": 0.6182,
      "step": 9928,
      "vm_loss": 0.1379
    },
    {
      "epoch": 0.7284132909012537,
      "lm_loss": 0.2078,
      "step": 9928,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.7284132909012537,
      "lm_loss": 0.6858,
      "step": 9928,
      "vm_loss": 0.0696
    },
    {
      "epoch": 0.7284132909012537,
      "lm_loss": 0.2094,
      "step": 9928,
      "vm_loss": 0.1711
    },
    {
      "epoch": 0.7284132909012537,
      "lm_loss": 0.9198,
      "step": 9928,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.7284866604913929,
      "grad_norm": 1.534809354858958,
      "learning_rate": 1.8114154473344081e-06,
      "loss": 0.6798,
      "step": 9929
    },
    {
      "epoch": 0.7285600300815319,
      "grad_norm": 0.5454985049613591,
      "learning_rate": 1.810500303957075e-06,
      "loss": 0.4675,
      "step": 9930
    },
    {
      "epoch": 0.7286333996716711,
      "grad_norm": 0.5774515476437674,
      "learning_rate": 1.8095853406986874e-06,
      "loss": 0.4974,
      "step": 9931
    },
    {
      "epoch": 0.7287067692618102,
      "grad_norm": 0.5517330031565341,
      "learning_rate": 1.8086705576109176e-06,
      "loss": 0.4734,
      "step": 9932
    },
    {
      "epoch": 0.7287801388519494,
      "grad_norm": 1.664572093537438,
      "learning_rate": 1.807755954745425e-06,
      "loss": 0.7525,
      "step": 9933
    },
    {
      "epoch": 0.7288535084420885,
      "grad_norm": 1.5582190543952152,
      "learning_rate": 1.806841532153859e-06,
      "loss": 0.6985,
      "step": 9934
    },
    {
      "epoch": 0.7289268780322276,
      "grad_norm": 1.5008995986232754,
      "learning_rate": 1.8059272898878594e-06,
      "loss": 0.6466,
      "step": 9935
    },
    {
      "epoch": 0.7290002476223667,
      "grad_norm": 1.4303181133218976,
      "learning_rate": 1.8050132279990551e-06,
      "loss": 0.8114,
      "step": 9936
    },
    {
      "epoch": 0.7290002476223667,
      "lm_loss": 0.9079,
      "step": 9936,
      "vm_loss": 0.0433
    },
    {
      "epoch": 0.7290002476223667,
      "lm_loss": 0.1694,
      "step": 9936,
      "vm_loss": 0.2754
    },
    {
      "epoch": 0.7290002476223667,
      "lm_loss": 0.1259,
      "step": 9936,
      "vm_loss": 0.1152
    },
    {
      "epoch": 0.7290002476223667,
      "lm_loss": 0.8538,
      "step": 9936,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.7290002476223667,
      "lm_loss": 1.0736,
      "step": 9936,
      "vm_loss": 0.0967
    },
    {
      "epoch": 0.7290002476223667,
      "lm_loss": 0.1752,
      "step": 9936,
      "vm_loss": 0.175
    },
    {
      "epoch": 0.7290002476223667,
      "lm_loss": 0.0652,
      "step": 9936,
      "vm_loss": 0.0575
    },
    {
      "epoch": 0.7290002476223667,
      "lm_loss": 0.0709,
      "step": 9936,
      "vm_loss": 0.2011
    },
    {
      "epoch": 0.7290736172125059,
      "grad_norm": 1.453054650890821,
      "learning_rate": 1.8040993465390649e-06,
      "loss": 0.711,
      "step": 9937
    },
    {
      "epoch": 0.729146986802645,
      "grad_norm": 1.4726814569269528,
      "learning_rate": 1.803185645559501e-06,
      "loss": 0.7137,
      "step": 9938
    },
    {
      "epoch": 0.7292203563927842,
      "grad_norm": 1.4708020463072946,
      "learning_rate": 1.80227212511196e-06,
      "loss": 0.7825,
      "step": 9939
    },
    {
      "epoch": 0.7292937259829232,
      "grad_norm": 0.537033299215979,
      "learning_rate": 1.801358785248032e-06,
      "loss": 0.4719,
      "step": 9940
    },
    {
      "epoch": 0.7293670955730623,
      "grad_norm": 1.599637335886885,
      "learning_rate": 1.8004456260192938e-06,
      "loss": 0.7635,
      "step": 9941
    },
    {
      "epoch": 0.7294404651632015,
      "grad_norm": 1.7626169183595373,
      "learning_rate": 1.7995326474773135e-06,
      "loss": 0.7634,
      "step": 9942
    },
    {
      "epoch": 0.7295138347533406,
      "grad_norm": 1.551906994814536,
      "learning_rate": 1.7986198496736518e-06,
      "loss": 0.7274,
      "step": 9943
    },
    {
      "epoch": 0.7295872043434798,
      "grad_norm": 1.5862400845163418,
      "learning_rate": 1.7977072326598555e-06,
      "loss": 0.7939,
      "step": 9944
    },
    {
      "epoch": 0.7295872043434798,
      "lm_loss": 1.5346,
      "step": 9944,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.7295872043434798,
      "lm_loss": 0.5189,
      "step": 9944,
      "vm_loss": 0.1692
    },
    {
      "epoch": 0.7295872043434798,
      "lm_loss": 0.5014,
      "step": 9944,
      "vm_loss": 0.1557
    },
    {
      "epoch": 0.7295872043434798,
      "lm_loss": 0.1655,
      "step": 9944,
      "vm_loss": 0.0977
    },
    {
      "epoch": 0.7295872043434798,
      "lm_loss": 0.1182,
      "step": 9944,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.7295872043434798,
      "lm_loss": 1.0291,
      "step": 9944,
      "vm_loss": 0.1391
    },
    {
      "epoch": 0.7295872043434798,
      "lm_loss": 0.3329,
      "step": 9944,
      "vm_loss": 0.0869
    },
    {
      "epoch": 0.7295872043434798,
      "lm_loss": 0.1094,
      "step": 9944,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.7296605739336188,
      "grad_norm": 1.56696646669973,
      "learning_rate": 1.7967947964874622e-06,
      "loss": 0.8195,
      "step": 9945
    },
    {
      "epoch": 0.729733943523758,
      "grad_norm": 1.4107416255706544,
      "learning_rate": 1.7958825412079988e-06,
      "loss": 0.6983,
      "step": 9946
    },
    {
      "epoch": 0.7298073131138971,
      "grad_norm": 0.526524165894021,
      "learning_rate": 1.7949704668729818e-06,
      "loss": 0.4657,
      "step": 9947
    },
    {
      "epoch": 0.7298806827040363,
      "grad_norm": 1.423374639827219,
      "learning_rate": 1.7940585735339216e-06,
      "loss": 0.785,
      "step": 9948
    },
    {
      "epoch": 0.7299540522941754,
      "grad_norm": 1.4273502502367934,
      "learning_rate": 1.7931468612423142e-06,
      "loss": 0.7231,
      "step": 9949
    },
    {
      "epoch": 0.7300274218843145,
      "grad_norm": 0.5423464258565731,
      "learning_rate": 1.7922353300496442e-06,
      "loss": 0.4835,
      "step": 9950
    },
    {
      "epoch": 0.7301007914744536,
      "grad_norm": 1.6934813328738403,
      "learning_rate": 1.7913239800073883e-06,
      "loss": 0.7079,
      "step": 9951
    },
    {
      "epoch": 0.7301741610645928,
      "grad_norm": 0.5397621402799759,
      "learning_rate": 1.790412811167012e-06,
      "loss": 0.4781,
      "step": 9952
    },
    {
      "epoch": 0.7301741610645928,
      "lm_loss": 1.2644,
      "step": 9952,
      "vm_loss": 0.0658
    },
    {
      "epoch": 0.7301741610645928,
      "lm_loss": 0.6244,
      "step": 9952,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.7301741610645928,
      "lm_loss": 0.6844,
      "step": 9952,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.7301741610645928,
      "lm_loss": 0.3268,
      "step": 9952,
      "vm_loss": 0.232
    },
    {
      "epoch": 0.7301741610645928,
      "lm_loss": 1.281,
      "step": 9952,
      "vm_loss": 0.1317
    },
    {
      "epoch": 0.7301741610645928,
      "lm_loss": 0.4973,
      "step": 9952,
      "vm_loss": 0.0448
    },
    {
      "epoch": 0.7301741610645928,
      "lm_loss": 0.1925,
      "step": 9952,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.7301741610645928,
      "lm_loss": 0.8224,
      "step": 9952,
      "vm_loss": 0.1782
    },
    {
      "epoch": 0.7302475306547319,
      "grad_norm": 1.5045218723966782,
      "learning_rate": 1.789501823579975e-06,
      "loss": 0.7387,
      "step": 9953
    },
    {
      "epoch": 0.730320900244871,
      "grad_norm": 1.4984252493435244,
      "learning_rate": 1.78859101729772e-06,
      "loss": 0.6242,
      "step": 9954
    },
    {
      "epoch": 0.7303942698350101,
      "grad_norm": 1.6337684252536644,
      "learning_rate": 1.7876803923716834e-06,
      "loss": 0.7911,
      "step": 9955
    },
    {
      "epoch": 0.7304676394251493,
      "grad_norm": 0.5591598657791461,
      "learning_rate": 1.7867699488532898e-06,
      "loss": 0.4674,
      "step": 9956
    },
    {
      "epoch": 0.7305410090152884,
      "grad_norm": 1.548487698296362,
      "learning_rate": 1.7858596867939532e-06,
      "loss": 0.7995,
      "step": 9957
    },
    {
      "epoch": 0.7306143786054276,
      "grad_norm": 1.6450963300540038,
      "learning_rate": 1.7849496062450816e-06,
      "loss": 0.7463,
      "step": 9958
    },
    {
      "epoch": 0.7306877481955666,
      "grad_norm": 1.5424574087611491,
      "learning_rate": 1.7840397072580678e-06,
      "loss": 0.8312,
      "step": 9959
    },
    {
      "epoch": 0.7307611177857057,
      "grad_norm": 1.486580526697543,
      "learning_rate": 1.7831299898842958e-06,
      "loss": 0.6991,
      "step": 9960
    },
    {
      "epoch": 0.7307611177857057,
      "lm_loss": 1.3508,
      "step": 9960,
      "vm_loss": 0.0577
    },
    {
      "epoch": 0.7307611177857057,
      "lm_loss": 0.1505,
      "step": 9960,
      "vm_loss": 0.1155
    },
    {
      "epoch": 0.7307611177857057,
      "lm_loss": 0.0714,
      "step": 9960,
      "vm_loss": 0.143
    },
    {
      "epoch": 0.7307611177857057,
      "lm_loss": 0.5534,
      "step": 9960,
      "vm_loss": 0.1909
    },
    {
      "epoch": 0.7307611177857057,
      "lm_loss": 0.8729,
      "step": 9960,
      "vm_loss": 0.0772
    },
    {
      "epoch": 0.7307611177857057,
      "lm_loss": 0.7984,
      "step": 9960,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.7307611177857057,
      "lm_loss": 0.9782,
      "step": 9960,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.7307611177857057,
      "lm_loss": 0.0834,
      "step": 9960,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.7308344873758449,
      "grad_norm": 1.7541619495065868,
      "learning_rate": 1.7822204541751392e-06,
      "loss": 0.7348,
      "step": 9961
    },
    {
      "epoch": 0.730907856965984,
      "grad_norm": 1.4938212378442313,
      "learning_rate": 1.7813111001819611e-06,
      "loss": 0.6905,
      "step": 9962
    },
    {
      "epoch": 0.7309812265561232,
      "grad_norm": 1.6504891936194948,
      "learning_rate": 1.780401927956118e-06,
      "loss": 0.6539,
      "step": 9963
    },
    {
      "epoch": 0.7310545961462622,
      "grad_norm": 1.5500672526748474,
      "learning_rate": 1.7794929375489506e-06,
      "loss": 0.6555,
      "step": 9964
    },
    {
      "epoch": 0.7311279657364014,
      "grad_norm": 1.527413878741288,
      "learning_rate": 1.7785841290117927e-06,
      "loss": 0.7318,
      "step": 9965
    },
    {
      "epoch": 0.7312013353265405,
      "grad_norm": 1.6001224178958489,
      "learning_rate": 1.777675502395967e-06,
      "loss": 0.7302,
      "step": 9966
    },
    {
      "epoch": 0.7312747049166797,
      "grad_norm": 1.5700015189928918,
      "learning_rate": 1.7767670577527856e-06,
      "loss": 0.7587,
      "step": 9967
    },
    {
      "epoch": 0.7313480745068188,
      "grad_norm": 1.4728551514601311,
      "learning_rate": 1.7758587951335487e-06,
      "loss": 0.7048,
      "step": 9968
    },
    {
      "epoch": 0.7313480745068188,
      "lm_loss": 0.7156,
      "step": 9968,
      "vm_loss": 0.0534
    },
    {
      "epoch": 0.7313480745068188,
      "lm_loss": 1.1922,
      "step": 9968,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.7313480745068188,
      "lm_loss": 0.8211,
      "step": 9968,
      "vm_loss": 0.1081
    },
    {
      "epoch": 0.7313480745068188,
      "lm_loss": 0.6669,
      "step": 9968,
      "vm_loss": 0.0929
    },
    {
      "epoch": 0.7313480745068188,
      "lm_loss": 0.8034,
      "step": 9968,
      "vm_loss": 0.1226
    },
    {
      "epoch": 0.7313480745068188,
      "lm_loss": 0.0027,
      "step": 9968,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.7313480745068188,
      "lm_loss": 0.0709,
      "step": 9968,
      "vm_loss": 0.0807
    },
    {
      "epoch": 0.7313480745068188,
      "lm_loss": 0.1812,
      "step": 9968,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.7314214440969579,
      "grad_norm": 1.671614266561536,
      "learning_rate": 1.7749507145895518e-06,
      "loss": 0.8286,
      "step": 9969
    },
    {
      "epoch": 0.731494813687097,
      "grad_norm": 1.7438179911063305,
      "learning_rate": 1.7740428161720751e-06,
      "loss": 0.7872,
      "step": 9970
    },
    {
      "epoch": 0.7315681832772362,
      "grad_norm": 0.5661447096265155,
      "learning_rate": 1.7731350999323893e-06,
      "loss": 0.4861,
      "step": 9971
    },
    {
      "epoch": 0.7316415528673753,
      "grad_norm": 1.5572740267732097,
      "learning_rate": 1.7722275659217553e-06,
      "loss": 0.7268,
      "step": 9972
    },
    {
      "epoch": 0.7317149224575145,
      "grad_norm": 0.5706531325037523,
      "learning_rate": 1.7713202141914226e-06,
      "loss": 0.4754,
      "step": 9973
    },
    {
      "epoch": 0.7317882920476535,
      "grad_norm": 0.5634899455103533,
      "learning_rate": 1.770413044792635e-06,
      "loss": 0.4836,
      "step": 9974
    },
    {
      "epoch": 0.7318616616377926,
      "grad_norm": 1.5782790915613338,
      "learning_rate": 1.7695060577766204e-06,
      "loss": 0.7813,
      "step": 9975
    },
    {
      "epoch": 0.7319350312279318,
      "grad_norm": 1.6191471972979457,
      "learning_rate": 1.7685992531945995e-06,
      "loss": 0.6683,
      "step": 9976
    },
    {
      "epoch": 0.7319350312279318,
      "lm_loss": 1.2409,
      "step": 9976,
      "vm_loss": 0.0607
    },
    {
      "epoch": 0.7319350312279318,
      "lm_loss": 1.1628,
      "step": 9976,
      "vm_loss": 0.1229
    },
    {
      "epoch": 0.7319350312279318,
      "lm_loss": 0.8994,
      "step": 9976,
      "vm_loss": 0.1777
    },
    {
      "epoch": 0.7319350312279318,
      "lm_loss": 0.2505,
      "step": 9976,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.7319350312279318,
      "lm_loss": 0.1921,
      "step": 9976,
      "vm_loss": 0.175
    },
    {
      "epoch": 0.7319350312279318,
      "lm_loss": 0.7918,
      "step": 9976,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.7319350312279318,
      "lm_loss": 1.076,
      "step": 9976,
      "vm_loss": 0.0898
    },
    {
      "epoch": 0.7319350312279318,
      "lm_loss": 0.6013,
      "step": 9976,
      "vm_loss": 0.1331
    },
    {
      "epoch": 0.732008400818071,
      "grad_norm": 1.4373073338011126,
      "learning_rate": 1.7676926310977805e-06,
      "loss": 0.7166,
      "step": 9977
    },
    {
      "epoch": 0.7320817704082101,
      "grad_norm": 1.5179297063054005,
      "learning_rate": 1.7667861915373636e-06,
      "loss": 0.8055,
      "step": 9978
    },
    {
      "epoch": 0.7321551399983491,
      "grad_norm": 1.4940596596861937,
      "learning_rate": 1.7658799345645356e-06,
      "loss": 0.756,
      "step": 9979
    },
    {
      "epoch": 0.7322285095884883,
      "grad_norm": 1.5069147780127905,
      "learning_rate": 1.7649738602304788e-06,
      "loss": 0.7521,
      "step": 9980
    },
    {
      "epoch": 0.7323018791786274,
      "grad_norm": 1.3918435230623665,
      "learning_rate": 1.7640679685863592e-06,
      "loss": 0.7523,
      "step": 9981
    },
    {
      "epoch": 0.7323752487687666,
      "grad_norm": 0.5517337250704428,
      "learning_rate": 1.7631622596833353e-06,
      "loss": 0.4779,
      "step": 9982
    },
    {
      "epoch": 0.7324486183589057,
      "grad_norm": 1.4166377226268263,
      "learning_rate": 1.7622567335725544e-06,
      "loss": 0.6596,
      "step": 9983
    },
    {
      "epoch": 0.7325219879490448,
      "grad_norm": 0.5603148029225378,
      "learning_rate": 1.7613513903051521e-06,
      "loss": 0.4786,
      "step": 9984
    },
    {
      "epoch": 0.7325219879490448,
      "lm_loss": 0.4945,
      "step": 9984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7325219879490448,
      "lm_loss": 0.3255,
      "step": 9984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7325219879490448,
      "lm_loss": 0.5564,
      "step": 9984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7325219879490448,
      "lm_loss": 0.4763,
      "step": 9984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7325219879490448,
      "lm_loss": 0.4776,
      "step": 9984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7325219879490448,
      "lm_loss": 0.5474,
      "step": 9984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7325219879490448,
      "lm_loss": 0.3197,
      "step": 9984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7325219879490448,
      "lm_loss": 0.5161,
      "step": 9984,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7325953575391839,
      "grad_norm": 0.5579297329635171,
      "learning_rate": 1.7604462299322589e-06,
      "loss": 0.4853,
      "step": 9985
    },
    {
      "epoch": 0.7326687271293231,
      "grad_norm": 1.5040607215048312,
      "learning_rate": 1.7595412525049898e-06,
      "loss": 0.6105,
      "step": 9986
    },
    {
      "epoch": 0.7327420967194622,
      "grad_norm": 1.5987909275636023,
      "learning_rate": 1.758636458074451e-06,
      "loss": 0.7747,
      "step": 9987
    },
    {
      "epoch": 0.7328154663096014,
      "grad_norm": 1.5713732023091822,
      "learning_rate": 1.7577318466917382e-06,
      "loss": 0.8274,
      "step": 9988
    },
    {
      "epoch": 0.7328888358997404,
      "grad_norm": 1.6052177541824952,
      "learning_rate": 1.756827418407936e-06,
      "loss": 0.6993,
      "step": 9989
    },
    {
      "epoch": 0.7329622054898796,
      "grad_norm": 1.5167199131147153,
      "learning_rate": 1.7559231732741228e-06,
      "loss": 0.8427,
      "step": 9990
    },
    {
      "epoch": 0.7330355750800187,
      "grad_norm": 0.5323011808104257,
      "learning_rate": 1.7550191113413618e-06,
      "loss": 0.4633,
      "step": 9991
    },
    {
      "epoch": 0.7331089446701579,
      "grad_norm": 0.546462087917002,
      "learning_rate": 1.7541152326607076e-06,
      "loss": 0.4673,
      "step": 9992
    },
    {
      "epoch": 0.7331089446701579,
      "lm_loss": 0.1193,
      "step": 9992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7331089446701579,
      "lm_loss": 0.3772,
      "step": 9992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7331089446701579,
      "lm_loss": 0.3543,
      "step": 9992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7331089446701579,
      "lm_loss": 0.4947,
      "step": 9992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7331089446701579,
      "lm_loss": 0.3965,
      "step": 9992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7331089446701579,
      "lm_loss": 0.4098,
      "step": 9992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7331089446701579,
      "lm_loss": 0.545,
      "step": 9992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7331089446701579,
      "lm_loss": 0.5612,
      "step": 9992,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.733182314260297,
      "grad_norm": 0.5635319974080673,
      "learning_rate": 1.7532115372832042e-06,
      "loss": 0.4766,
      "step": 9993
    },
    {
      "epoch": 0.733255683850436,
      "grad_norm": 1.4530372965778289,
      "learning_rate": 1.7523080252598862e-06,
      "loss": 0.7176,
      "step": 9994
    },
    {
      "epoch": 0.7333290534405752,
      "grad_norm": 1.3658863715504177,
      "learning_rate": 1.7514046966417752e-06,
      "loss": 0.7377,
      "step": 9995
    },
    {
      "epoch": 0.7334024230307143,
      "grad_norm": 1.4564855770832432,
      "learning_rate": 1.7505015514798874e-06,
      "loss": 0.8504,
      "step": 9996
    },
    {
      "epoch": 0.7334757926208535,
      "grad_norm": 1.3596894589964432,
      "learning_rate": 1.7495985898252244e-06,
      "loss": 0.69,
      "step": 9997
    },
    {
      "epoch": 0.7335491622109926,
      "grad_norm": 1.6058377436885418,
      "learning_rate": 1.7486958117287783e-06,
      "loss": 0.7496,
      "step": 9998
    },
    {
      "epoch": 0.7336225318011317,
      "grad_norm": 1.7520262510715239,
      "learning_rate": 1.7477932172415317e-06,
      "loss": 0.7192,
      "step": 9999
    },
    {
      "epoch": 0.7336959013912708,
      "grad_norm": 1.2992120000475775,
      "learning_rate": 1.7468908064144541e-06,
      "loss": 0.6159,
      "step": 10000
    },
    {
      "epoch": 0.7336959013912708,
      "lm_loss": 1.3597,
      "step": 10000,
      "vm_loss": 0.0539
    },
    {
      "epoch": 0.7336959013912708,
      "lm_loss": 1.5051,
      "step": 10000,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.7336959013912708,
      "lm_loss": 0.175,
      "step": 10000,
      "vm_loss": 0.1589
    },
    {
      "epoch": 0.7336959013912708,
      "lm_loss": 0.2719,
      "step": 10000,
      "vm_loss": 0.1881
    },
    {
      "epoch": 0.7336959013912708,
      "lm_loss": 1.0362,
      "step": 10000,
      "vm_loss": 0.181
    },
    {
      "epoch": 0.7336959013912708,
      "lm_loss": 0.5494,
      "step": 10000,
      "vm_loss": 0.1401
    },
    {
      "epoch": 0.7336959013912708,
      "lm_loss": 1.4657,
      "step": 10000,
      "vm_loss": 0.1398
    },
    {
      "epoch": 0.7336959013912708,
      "lm_loss": 0.6017,
      "step": 10000,
      "vm_loss": 0.073
    },
    {
      "epoch": 0.73376927098141,
      "grad_norm": 1.7743087088393483,
      "learning_rate": 1.745988579298511e-06,
      "loss": 0.9159,
      "step": 10001
    },
    {
      "epoch": 0.7338426405715491,
      "grad_norm": 0.5675316370940457,
      "learning_rate": 1.7450865359446505e-06,
      "loss": 0.4801,
      "step": 10002
    },
    {
      "epoch": 0.7339160101616883,
      "grad_norm": 1.4015995945014388,
      "learning_rate": 1.7441846764038146e-06,
      "loss": 0.7846,
      "step": 10003
    },
    {
      "epoch": 0.7339893797518273,
      "grad_norm": 0.5564622775360136,
      "learning_rate": 1.743283000726933e-06,
      "loss": 0.4784,
      "step": 10004
    },
    {
      "epoch": 0.7340627493419665,
      "grad_norm": 1.3497951178409422,
      "learning_rate": 1.7423815089649248e-06,
      "loss": 0.8141,
      "step": 10005
    },
    {
      "epoch": 0.7341361189321056,
      "grad_norm": 1.4247702241701703,
      "learning_rate": 1.7414802011686994e-06,
      "loss": 0.6077,
      "step": 10006
    },
    {
      "epoch": 0.7342094885222448,
      "grad_norm": 1.556701618604699,
      "learning_rate": 1.7405790773891573e-06,
      "loss": 0.7697,
      "step": 10007
    },
    {
      "epoch": 0.7342828581123839,
      "grad_norm": 1.4920386999144941,
      "learning_rate": 1.739678137677187e-06,
      "loss": 0.6349,
      "step": 10008
    },
    {
      "epoch": 0.7342828581123839,
      "lm_loss": 1.404,
      "step": 10008,
      "vm_loss": 0.0456
    },
    {
      "epoch": 0.7342828581123839,
      "lm_loss": 0.5756,
      "step": 10008,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.7342828581123839,
      "lm_loss": 0.7898,
      "step": 10008,
      "vm_loss": 0.0965
    },
    {
      "epoch": 0.7342828581123839,
      "lm_loss": 0.3938,
      "step": 10008,
      "vm_loss": 0.1397
    },
    {
      "epoch": 0.7342828581123839,
      "lm_loss": 1.0168,
      "step": 10008,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.7342828581123839,
      "lm_loss": 0.6632,
      "step": 10008,
      "vm_loss": 0.0862
    },
    {
      "epoch": 0.7342828581123839,
      "lm_loss": 0.0442,
      "step": 10008,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.7342828581123839,
      "lm_loss": 1.9014,
      "step": 10008,
      "vm_loss": 0.174
    },
    {
      "epoch": 0.734356227702523,
      "grad_norm": 1.5554511379486997,
      "learning_rate": 1.7387773820836668e-06,
      "loss": 0.7376,
      "step": 10009
    },
    {
      "epoch": 0.7344295972926621,
      "grad_norm": 1.513681315918054,
      "learning_rate": 1.7378768106594635e-06,
      "loss": 0.739,
      "step": 10010
    },
    {
      "epoch": 0.7345029668828013,
      "grad_norm": 0.5876844498310136,
      "learning_rate": 1.7369764234554337e-06,
      "loss": 0.4976,
      "step": 10011
    },
    {
      "epoch": 0.7345763364729404,
      "grad_norm": 0.5523708607504302,
      "learning_rate": 1.7360762205224274e-06,
      "loss": 0.4773,
      "step": 10012
    },
    {
      "epoch": 0.7346497060630796,
      "grad_norm": 1.4884445421741623,
      "learning_rate": 1.7351762019112805e-06,
      "loss": 0.6969,
      "step": 10013
    },
    {
      "epoch": 0.7347230756532186,
      "grad_norm": 1.5417134720729413,
      "learning_rate": 1.734276367672818e-06,
      "loss": 0.8952,
      "step": 10014
    },
    {
      "epoch": 0.7347964452433577,
      "grad_norm": 1.4256734348816495,
      "learning_rate": 1.7333767178578569e-06,
      "loss": 0.6903,
      "step": 10015
    },
    {
      "epoch": 0.7348698148334969,
      "grad_norm": 1.6386907151189196,
      "learning_rate": 1.7324772525172006e-06,
      "loss": 0.7683,
      "step": 10016
    },
    {
      "epoch": 0.7348698148334969,
      "lm_loss": 0.471,
      "step": 10016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7348698148334969,
      "lm_loss": 0.509,
      "step": 10016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7348698148334969,
      "lm_loss": 0.5103,
      "step": 10016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7348698148334969,
      "lm_loss": 0.4187,
      "step": 10016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7348698148334969,
      "lm_loss": 0.483,
      "step": 10016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7348698148334969,
      "lm_loss": 0.4277,
      "step": 10016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7348698148334969,
      "lm_loss": 0.4356,
      "step": 10016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7348698148334969,
      "lm_loss": 0.5119,
      "step": 10016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.734943184423636,
      "grad_norm": 0.5489174197186372,
      "learning_rate": 1.7315779717016473e-06,
      "loss": 0.4833,
      "step": 10017
    },
    {
      "epoch": 0.7350165540137752,
      "grad_norm": 0.5401973636792239,
      "learning_rate": 1.7306788754619797e-06,
      "loss": 0.4696,
      "step": 10018
    },
    {
      "epoch": 0.7350899236039142,
      "grad_norm": 0.5401032654187364,
      "learning_rate": 1.7297799638489726e-06,
      "loss": 0.4722,
      "step": 10019
    },
    {
      "epoch": 0.7351632931940534,
      "grad_norm": 1.4518857831248868,
      "learning_rate": 1.7288812369133895e-06,
      "loss": 0.7601,
      "step": 10020
    },
    {
      "epoch": 0.7352366627841925,
      "grad_norm": 1.5857794425582705,
      "learning_rate": 1.7279826947059837e-06,
      "loss": 0.8342,
      "step": 10021
    },
    {
      "epoch": 0.7353100323743317,
      "grad_norm": 1.5858081171783582,
      "learning_rate": 1.7270843372774965e-06,
      "loss": 0.6457,
      "step": 10022
    },
    {
      "epoch": 0.7353834019644708,
      "grad_norm": 0.5256187888435726,
      "learning_rate": 1.7261861646786636e-06,
      "loss": 0.4654,
      "step": 10023
    },
    {
      "epoch": 0.7354567715546099,
      "grad_norm": 1.5103610724838235,
      "learning_rate": 1.725288176960206e-06,
      "loss": 0.731,
      "step": 10024
    },
    {
      "epoch": 0.7354567715546099,
      "lm_loss": 1.3749,
      "step": 10024,
      "vm_loss": 0.0755
    },
    {
      "epoch": 0.7354567715546099,
      "lm_loss": 1.2382,
      "step": 10024,
      "vm_loss": 0.1349
    },
    {
      "epoch": 0.7354567715546099,
      "lm_loss": 1.0626,
      "step": 10024,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.7354567715546099,
      "lm_loss": 1.1356,
      "step": 10024,
      "vm_loss": 0.137
    },
    {
      "epoch": 0.7354567715546099,
      "lm_loss": 0.2238,
      "step": 10024,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.7354567715546099,
      "lm_loss": 0.6245,
      "step": 10024,
      "vm_loss": 0.0789
    },
    {
      "epoch": 0.7354567715546099,
      "lm_loss": 0.278,
      "step": 10024,
      "vm_loss": 0.1287
    },
    {
      "epoch": 0.7354567715546099,
      "lm_loss": 0.7952,
      "step": 10024,
      "vm_loss": 0.1581
    },
    {
      "epoch": 0.735530141144749,
      "grad_norm": 1.4799315240808641,
      "learning_rate": 1.724390374172834e-06,
      "loss": 0.6971,
      "step": 10025
    },
    {
      "epoch": 0.7356035107348882,
      "grad_norm": 1.5694529338295797,
      "learning_rate": 1.7234927563672498e-06,
      "loss": 0.833,
      "step": 10026
    },
    {
      "epoch": 0.7356768803250273,
      "grad_norm": 0.5611945246100918,
      "learning_rate": 1.7225953235941423e-06,
      "loss": 0.4751,
      "step": 10027
    },
    {
      "epoch": 0.7357502499151665,
      "grad_norm": 1.6110233965210023,
      "learning_rate": 1.7216980759041946e-06,
      "loss": 0.7084,
      "step": 10028
    },
    {
      "epoch": 0.7358236195053055,
      "grad_norm": 1.5079503591617707,
      "learning_rate": 1.7208010133480751e-06,
      "loss": 0.7224,
      "step": 10029
    },
    {
      "epoch": 0.7358969890954447,
      "grad_norm": 1.604536045397315,
      "learning_rate": 1.7199041359764434e-06,
      "loss": 0.7484,
      "step": 10030
    },
    {
      "epoch": 0.7359703586855838,
      "grad_norm": 0.567090488026084,
      "learning_rate": 1.719007443839948e-06,
      "loss": 0.4983,
      "step": 10031
    },
    {
      "epoch": 0.736043728275723,
      "grad_norm": 0.5396556324412256,
      "learning_rate": 1.7181109369892273e-06,
      "loss": 0.4931,
      "step": 10032
    },
    {
      "epoch": 0.736043728275723,
      "lm_loss": 1.2969,
      "step": 10032,
      "vm_loss": 0.1042
    },
    {
      "epoch": 0.736043728275723,
      "lm_loss": 0.0016,
      "step": 10032,
      "vm_loss": 0.1655
    },
    {
      "epoch": 0.736043728275723,
      "lm_loss": 1.0452,
      "step": 10032,
      "vm_loss": 0.1646
    },
    {
      "epoch": 0.736043728275723,
      "lm_loss": 1.002,
      "step": 10032,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.736043728275723,
      "lm_loss": 0.3657,
      "step": 10032,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.736043728275723,
      "lm_loss": 0.2685,
      "step": 10032,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.736043728275723,
      "lm_loss": 0.15,
      "step": 10032,
      "vm_loss": 0.0762
    },
    {
      "epoch": 0.736043728275723,
      "lm_loss": 0.0629,
      "step": 10032,
      "vm_loss": 0.1072
    },
    {
      "epoch": 0.7361170978658621,
      "grad_norm": 1.6669533381003627,
      "learning_rate": 1.717214615474908e-06,
      "loss": 0.7904,
      "step": 10033
    },
    {
      "epoch": 0.7361904674560011,
      "grad_norm": 0.5330527327450757,
      "learning_rate": 1.7163184793476107e-06,
      "loss": 0.4685,
      "step": 10034
    },
    {
      "epoch": 0.7362638370461403,
      "grad_norm": 1.4426018048795612,
      "learning_rate": 1.7154225286579406e-06,
      "loss": 0.7135,
      "step": 10035
    },
    {
      "epoch": 0.7363372066362794,
      "grad_norm": 1.81467957624823,
      "learning_rate": 1.714526763456495e-06,
      "loss": 0.7635,
      "step": 10036
    },
    {
      "epoch": 0.7364105762264186,
      "grad_norm": 0.5439426353521015,
      "learning_rate": 1.7136311837938591e-06,
      "loss": 0.4595,
      "step": 10037
    },
    {
      "epoch": 0.7364839458165576,
      "grad_norm": 1.560163993267872,
      "learning_rate": 1.712735789720607e-06,
      "loss": 0.733,
      "step": 10038
    },
    {
      "epoch": 0.7365573154066968,
      "grad_norm": 1.6086054839034987,
      "learning_rate": 1.7118405812873074e-06,
      "loss": 0.7555,
      "step": 10039
    },
    {
      "epoch": 0.7366306849968359,
      "grad_norm": 1.464274828751438,
      "learning_rate": 1.7109455585445129e-06,
      "loss": 0.7009,
      "step": 10040
    },
    {
      "epoch": 0.7366306849968359,
      "lm_loss": 0.83,
      "step": 10040,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.7366306849968359,
      "lm_loss": 0.138,
      "step": 10040,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.7366306849968359,
      "lm_loss": 0.1569,
      "step": 10040,
      "vm_loss": 0.069
    },
    {
      "epoch": 0.7366306849968359,
      "lm_loss": 0.6219,
      "step": 10040,
      "vm_loss": 0.1292
    },
    {
      "epoch": 0.7366306849968359,
      "lm_loss": 1.2764,
      "step": 10040,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.7366306849968359,
      "lm_loss": 0.1602,
      "step": 10040,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.7366306849968359,
      "lm_loss": 0.6542,
      "step": 10040,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.7366306849968359,
      "lm_loss": 0.5364,
      "step": 10040,
      "vm_loss": 0.1903
    },
    {
      "epoch": 0.7367040545869751,
      "grad_norm": 1.643862626736659,
      "learning_rate": 1.7100507215427681e-06,
      "loss": 0.7217,
      "step": 10041
    },
    {
      "epoch": 0.7367774241771142,
      "grad_norm": 0.5627521077671601,
      "learning_rate": 1.7091560703326065e-06,
      "loss": 0.4728,
      "step": 10042
    },
    {
      "epoch": 0.7368507937672533,
      "grad_norm": 1.395231027894143,
      "learning_rate": 1.7082616049645495e-06,
      "loss": 0.6863,
      "step": 10043
    },
    {
      "epoch": 0.7369241633573924,
      "grad_norm": 1.5022608363585024,
      "learning_rate": 1.7073673254891138e-06,
      "loss": 0.727,
      "step": 10044
    },
    {
      "epoch": 0.7369975329475316,
      "grad_norm": 1.4659313761771589,
      "learning_rate": 1.7064732319567983e-06,
      "loss": 0.7153,
      "step": 10045
    },
    {
      "epoch": 0.7370709025376707,
      "grad_norm": 1.4976362051482826,
      "learning_rate": 1.7055793244180963e-06,
      "loss": 0.8121,
      "step": 10046
    },
    {
      "epoch": 0.7371442721278099,
      "grad_norm": 1.474577338622998,
      "learning_rate": 1.704685602923488e-06,
      "loss": 0.7259,
      "step": 10047
    },
    {
      "epoch": 0.7372176417179489,
      "grad_norm": 1.382196063880797,
      "learning_rate": 1.7037920675234448e-06,
      "loss": 0.7261,
      "step": 10048
    },
    {
      "epoch": 0.7372176417179489,
      "lm_loss": 1.3216,
      "step": 10048,
      "vm_loss": 0.078
    },
    {
      "epoch": 0.7372176417179489,
      "lm_loss": 0.1549,
      "step": 10048,
      "vm_loss": 0.0796
    },
    {
      "epoch": 0.7372176417179489,
      "lm_loss": 0.7577,
      "step": 10048,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.7372176417179489,
      "lm_loss": 0.5627,
      "step": 10048,
      "vm_loss": 0.1515
    },
    {
      "epoch": 0.7372176417179489,
      "lm_loss": 0.52,
      "step": 10048,
      "vm_loss": 0.0405
    },
    {
      "epoch": 0.7372176417179489,
      "lm_loss": 0.569,
      "step": 10048,
      "vm_loss": 0.0801
    },
    {
      "epoch": 0.7372176417179489,
      "lm_loss": 0.3493,
      "step": 10048,
      "vm_loss": 0.1427
    },
    {
      "epoch": 0.7372176417179489,
      "lm_loss": 0.1662,
      "step": 10048,
      "vm_loss": 0.0754
    },
    {
      "epoch": 0.737291011308088,
      "grad_norm": 1.6698767850584613,
      "learning_rate": 1.7028987182684248e-06,
      "loss": 0.7312,
      "step": 10049
    },
    {
      "epoch": 0.7373643808982272,
      "grad_norm": 1.3460915321266762,
      "learning_rate": 1.7020055552088805e-06,
      "loss": 0.7383,
      "step": 10050
    },
    {
      "epoch": 0.7374377504883664,
      "grad_norm": 1.5184384988071176,
      "learning_rate": 1.7011125783952503e-06,
      "loss": 0.7329,
      "step": 10051
    },
    {
      "epoch": 0.7375111200785055,
      "grad_norm": 1.571492963290935,
      "learning_rate": 1.7002197878779625e-06,
      "loss": 0.7139,
      "step": 10052
    },
    {
      "epoch": 0.7375844896686445,
      "grad_norm": 1.3243403631029866,
      "learning_rate": 1.6993271837074343e-06,
      "loss": 0.6142,
      "step": 10053
    },
    {
      "epoch": 0.7376578592587837,
      "grad_norm": 1.4131294902104532,
      "learning_rate": 1.6984347659340727e-06,
      "loss": 0.736,
      "step": 10054
    },
    {
      "epoch": 0.7377312288489228,
      "grad_norm": 0.5599929575041822,
      "learning_rate": 1.697542534608278e-06,
      "loss": 0.4759,
      "step": 10055
    },
    {
      "epoch": 0.737804598439062,
      "grad_norm": 1.498918780851656,
      "learning_rate": 1.6966504897804348e-06,
      "loss": 0.764,
      "step": 10056
    },
    {
      "epoch": 0.737804598439062,
      "lm_loss": 1.3351,
      "step": 10056,
      "vm_loss": 0.079
    },
    {
      "epoch": 0.737804598439062,
      "lm_loss": 0.3819,
      "step": 10056,
      "vm_loss": 0.1231
    },
    {
      "epoch": 0.737804598439062,
      "lm_loss": 0.555,
      "step": 10056,
      "vm_loss": 0.1394
    },
    {
      "epoch": 0.737804598439062,
      "lm_loss": 0.368,
      "step": 10056,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.737804598439062,
      "lm_loss": 0.2105,
      "step": 10056,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.737804598439062,
      "lm_loss": 1.1119,
      "step": 10056,
      "vm_loss": 0.1517
    },
    {
      "epoch": 0.737804598439062,
      "lm_loss": 0.2941,
      "step": 10056,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.737804598439062,
      "lm_loss": 0.7234,
      "step": 10056,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.7378779680292011,
      "grad_norm": 1.6247336012028597,
      "learning_rate": 1.6957586315009189e-06,
      "loss": 0.7601,
      "step": 10057
    },
    {
      "epoch": 0.7379513376193402,
      "grad_norm": 0.5736288289693899,
      "learning_rate": 1.6948669598200956e-06,
      "loss": 0.491,
      "step": 10058
    },
    {
      "epoch": 0.7380247072094793,
      "grad_norm": 1.4476666847508717,
      "learning_rate": 1.6939754747883202e-06,
      "loss": 0.7028,
      "step": 10059
    },
    {
      "epoch": 0.7380980767996185,
      "grad_norm": 0.5419935284744066,
      "learning_rate": 1.6930841764559353e-06,
      "loss": 0.4738,
      "step": 10060
    },
    {
      "epoch": 0.7381714463897576,
      "grad_norm": 1.6462242315434827,
      "learning_rate": 1.6921930648732782e-06,
      "loss": 0.7808,
      "step": 10061
    },
    {
      "epoch": 0.7382448159798968,
      "grad_norm": 1.5572032943661058,
      "learning_rate": 1.6913021400906704e-06,
      "loss": 0.7124,
      "step": 10062
    },
    {
      "epoch": 0.7383181855700358,
      "grad_norm": 1.572481535497898,
      "learning_rate": 1.6904114021584244e-06,
      "loss": 0.7837,
      "step": 10063
    },
    {
      "epoch": 0.738391555160175,
      "grad_norm": 1.4740846689908427,
      "learning_rate": 1.689520851126843e-06,
      "loss": 0.7846,
      "step": 10064
    },
    {
      "epoch": 0.738391555160175,
      "lm_loss": 0.7888,
      "step": 10064,
      "vm_loss": 0.1039
    },
    {
      "epoch": 0.738391555160175,
      "lm_loss": 1.6809,
      "step": 10064,
      "vm_loss": 0.0367
    },
    {
      "epoch": 0.738391555160175,
      "lm_loss": 1.5033,
      "step": 10064,
      "vm_loss": 0.1569
    },
    {
      "epoch": 0.738391555160175,
      "lm_loss": 0.0709,
      "step": 10064,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.738391555160175,
      "lm_loss": 0.3555,
      "step": 10064,
      "vm_loss": 0.1576
    },
    {
      "epoch": 0.738391555160175,
      "lm_loss": 0.4362,
      "step": 10064,
      "vm_loss": 0.2361
    },
    {
      "epoch": 0.738391555160175,
      "lm_loss": 0.3124,
      "step": 10064,
      "vm_loss": 0.1244
    },
    {
      "epoch": 0.738391555160175,
      "lm_loss": 1.2166,
      "step": 10064,
      "vm_loss": 0.153
    },
    {
      "epoch": 0.7384649247503141,
      "grad_norm": 1.393110734849835,
      "learning_rate": 1.6886304870462156e-06,
      "loss": 0.8717,
      "step": 10065
    },
    {
      "epoch": 0.7385382943404533,
      "grad_norm": 1.564515138346865,
      "learning_rate": 1.6877403099668267e-06,
      "loss": 0.7942,
      "step": 10066
    },
    {
      "epoch": 0.7386116639305924,
      "grad_norm": 0.55447007059176,
      "learning_rate": 1.6868503199389447e-06,
      "loss": 0.4772,
      "step": 10067
    },
    {
      "epoch": 0.7386850335207314,
      "grad_norm": 1.54724042897263,
      "learning_rate": 1.6859605170128307e-06,
      "loss": 0.7551,
      "step": 10068
    },
    {
      "epoch": 0.7387584031108706,
      "grad_norm": 1.5117080154414293,
      "learning_rate": 1.6850709012387328e-06,
      "loss": 0.7491,
      "step": 10069
    },
    {
      "epoch": 0.7388317727010097,
      "grad_norm": 1.451666187240752,
      "learning_rate": 1.6841814726668892e-06,
      "loss": 0.676,
      "step": 10070
    },
    {
      "epoch": 0.7389051422911489,
      "grad_norm": 1.4592606405996453,
      "learning_rate": 1.6832922313475308e-06,
      "loss": 0.8255,
      "step": 10071
    },
    {
      "epoch": 0.738978511881288,
      "grad_norm": 0.5614963860350618,
      "learning_rate": 1.6824031773308735e-06,
      "loss": 0.4682,
      "step": 10072
    },
    {
      "epoch": 0.738978511881288,
      "lm_loss": 1.7773,
      "step": 10072,
      "vm_loss": 0.0481
    },
    {
      "epoch": 0.738978511881288,
      "lm_loss": 0.4149,
      "step": 10072,
      "vm_loss": 0.0613
    },
    {
      "epoch": 0.738978511881288,
      "lm_loss": 0.8124,
      "step": 10072,
      "vm_loss": 0.0982
    },
    {
      "epoch": 0.738978511881288,
      "lm_loss": 0.6583,
      "step": 10072,
      "vm_loss": 0.184
    },
    {
      "epoch": 0.738978511881288,
      "lm_loss": 0.9106,
      "step": 10072,
      "vm_loss": 0.101
    },
    {
      "epoch": 0.738978511881288,
      "lm_loss": 1.1039,
      "step": 10072,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.738978511881288,
      "lm_loss": 0.065,
      "step": 10072,
      "vm_loss": 0.137
    },
    {
      "epoch": 0.738978511881288,
      "lm_loss": 0.4211,
      "step": 10072,
      "vm_loss": 0.1107
    },
    {
      "epoch": 0.7390518814714271,
      "grad_norm": 1.6028408134131953,
      "learning_rate": 1.681514310667125e-06,
      "loss": 0.8598,
      "step": 10073
    },
    {
      "epoch": 0.7391252510615662,
      "grad_norm": 1.5191726090932602,
      "learning_rate": 1.6806256314064818e-06,
      "loss": 0.7066,
      "step": 10074
    },
    {
      "epoch": 0.7391986206517054,
      "grad_norm": 0.5577003653907799,
      "learning_rate": 1.6797371395991286e-06,
      "loss": 0.502,
      "step": 10075
    },
    {
      "epoch": 0.7392719902418445,
      "grad_norm": 0.5656192400970866,
      "learning_rate": 1.6788488352952403e-06,
      "loss": 0.4784,
      "step": 10076
    },
    {
      "epoch": 0.7393453598319837,
      "grad_norm": 0.5616113965526449,
      "learning_rate": 1.6779607185449853e-06,
      "loss": 0.4774,
      "step": 10077
    },
    {
      "epoch": 0.7394187294221227,
      "grad_norm": 0.5575341971798579,
      "learning_rate": 1.6770727893985145e-06,
      "loss": 0.4724,
      "step": 10078
    },
    {
      "epoch": 0.7394920990122619,
      "grad_norm": 1.5745780018274735,
      "learning_rate": 1.6761850479059727e-06,
      "loss": 0.8496,
      "step": 10079
    },
    {
      "epoch": 0.739565468602401,
      "grad_norm": 1.4957778999839828,
      "learning_rate": 1.6752974941174927e-06,
      "loss": 0.7288,
      "step": 10080
    },
    {
      "epoch": 0.739565468602401,
      "lm_loss": 0.3878,
      "step": 10080,
      "vm_loss": 0.0853
    },
    {
      "epoch": 0.739565468602401,
      "lm_loss": 1.4008,
      "step": 10080,
      "vm_loss": 0.1456
    },
    {
      "epoch": 0.739565468602401,
      "lm_loss": 1.3241,
      "step": 10080,
      "vm_loss": 0.0927
    },
    {
      "epoch": 0.739565468602401,
      "lm_loss": 0.5069,
      "step": 10080,
      "vm_loss": 0.1461
    },
    {
      "epoch": 0.739565468602401,
      "lm_loss": 0.064,
      "step": 10080,
      "vm_loss": 0.185
    },
    {
      "epoch": 0.739565468602401,
      "lm_loss": 0.147,
      "step": 10080,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.739565468602401,
      "lm_loss": 0.594,
      "step": 10080,
      "vm_loss": 0.1435
    },
    {
      "epoch": 0.739565468602401,
      "lm_loss": 0.0593,
      "step": 10080,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.7396388381925402,
      "grad_norm": 1.5012474547377062,
      "learning_rate": 1.6744101280831948e-06,
      "loss": 0.6763,
      "step": 10081
    },
    {
      "epoch": 0.7397122077826793,
      "grad_norm": 1.5402999828719757,
      "learning_rate": 1.6735229498531946e-06,
      "loss": 0.7269,
      "step": 10082
    },
    {
      "epoch": 0.7397855773728184,
      "grad_norm": 0.5596784817808472,
      "learning_rate": 1.672635959477591e-06,
      "loss": 0.4932,
      "step": 10083
    },
    {
      "epoch": 0.7398589469629575,
      "grad_norm": 1.4854613000355077,
      "learning_rate": 1.6717491570064753e-06,
      "loss": 0.7304,
      "step": 10084
    },
    {
      "epoch": 0.7399323165530967,
      "grad_norm": 0.5604976833429881,
      "learning_rate": 1.6708625424899266e-06,
      "loss": 0.4766,
      "step": 10085
    },
    {
      "epoch": 0.7400056861432358,
      "grad_norm": 1.5158234430383721,
      "learning_rate": 1.6699761159780147e-06,
      "loss": 0.7501,
      "step": 10086
    },
    {
      "epoch": 0.740079055733375,
      "grad_norm": 1.6544805434733094,
      "learning_rate": 1.6690898775207964e-06,
      "loss": 0.807,
      "step": 10087
    },
    {
      "epoch": 0.740152425323514,
      "grad_norm": 1.3910103724874683,
      "learning_rate": 1.6682038271683232e-06,
      "loss": 0.7193,
      "step": 10088
    },
    {
      "epoch": 0.740152425323514,
      "lm_loss": 1.2411,
      "step": 10088,
      "vm_loss": 0.0352
    },
    {
      "epoch": 0.740152425323514,
      "lm_loss": 1.0953,
      "step": 10088,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.740152425323514,
      "lm_loss": 0.8177,
      "step": 10088,
      "vm_loss": 0.045
    },
    {
      "epoch": 0.740152425323514,
      "lm_loss": 0.1457,
      "step": 10088,
      "vm_loss": 0.1151
    },
    {
      "epoch": 0.740152425323514,
      "lm_loss": 0.8997,
      "step": 10088,
      "vm_loss": 0.1195
    },
    {
      "epoch": 0.740152425323514,
      "lm_loss": 0.4811,
      "step": 10088,
      "vm_loss": 0.1104
    },
    {
      "epoch": 0.740152425323514,
      "lm_loss": 0.019,
      "step": 10088,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.740152425323514,
      "lm_loss": 0.2582,
      "step": 10088,
      "vm_loss": 0.1737
    },
    {
      "epoch": 0.7402257949136531,
      "grad_norm": 1.6920465702888452,
      "learning_rate": 1.6673179649706312e-06,
      "loss": 0.7868,
      "step": 10089
    },
    {
      "epoch": 0.7402991645037923,
      "grad_norm": 0.5556000089348951,
      "learning_rate": 1.666432290977747e-06,
      "loss": 0.4784,
      "step": 10090
    },
    {
      "epoch": 0.7403725340939314,
      "grad_norm": 1.6753154897604425,
      "learning_rate": 1.6655468052396862e-06,
      "loss": 0.6508,
      "step": 10091
    },
    {
      "epoch": 0.7404459036840706,
      "grad_norm": 0.5409187468726813,
      "learning_rate": 1.6646615078064537e-06,
      "loss": 0.4775,
      "step": 10092
    },
    {
      "epoch": 0.7405192732742096,
      "grad_norm": 0.5690092137705873,
      "learning_rate": 1.6637763987280475e-06,
      "loss": 0.5046,
      "step": 10093
    },
    {
      "epoch": 0.7405926428643488,
      "grad_norm": 1.5247503894289802,
      "learning_rate": 1.6628914780544498e-06,
      "loss": 0.7226,
      "step": 10094
    },
    {
      "epoch": 0.7406660124544879,
      "grad_norm": 1.595282824811283,
      "learning_rate": 1.6620067458356348e-06,
      "loss": 0.6854,
      "step": 10095
    },
    {
      "epoch": 0.7407393820446271,
      "grad_norm": 1.642679630345062,
      "learning_rate": 1.661122202121565e-06,
      "loss": 0.7262,
      "step": 10096
    },
    {
      "epoch": 0.7407393820446271,
      "lm_loss": 1.3182,
      "step": 10096,
      "vm_loss": 0.0609
    },
    {
      "epoch": 0.7407393820446271,
      "lm_loss": 0.4418,
      "step": 10096,
      "vm_loss": 0.094
    },
    {
      "epoch": 0.7407393820446271,
      "lm_loss": 0.6775,
      "step": 10096,
      "vm_loss": 0.0628
    },
    {
      "epoch": 0.7407393820446271,
      "lm_loss": 1.2079,
      "step": 10096,
      "vm_loss": 0.1389
    },
    {
      "epoch": 0.7407393820446271,
      "lm_loss": 0.8209,
      "step": 10096,
      "vm_loss": 0.1556
    },
    {
      "epoch": 0.7407393820446271,
      "lm_loss": 0.3417,
      "step": 10096,
      "vm_loss": 0.0947
    },
    {
      "epoch": 0.7407393820446271,
      "lm_loss": 0.7168,
      "step": 10096,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.7407393820446271,
      "lm_loss": 0.7965,
      "step": 10096,
      "vm_loss": 0.0796
    },
    {
      "epoch": 0.7408127516347662,
      "grad_norm": 1.585928514231845,
      "learning_rate": 1.6602378469621916e-06,
      "loss": 0.7754,
      "step": 10097
    },
    {
      "epoch": 0.7408861212249053,
      "grad_norm": 1.828491546728102,
      "learning_rate": 1.6593536804074595e-06,
      "loss": 0.7,
      "step": 10098
    },
    {
      "epoch": 0.7409594908150444,
      "grad_norm": 0.5397322600692812,
      "learning_rate": 1.6584697025072993e-06,
      "loss": 0.4844,
      "step": 10099
    },
    {
      "epoch": 0.7410328604051836,
      "grad_norm": 1.446744954141733,
      "learning_rate": 1.6575859133116284e-06,
      "loss": 0.6582,
      "step": 10100
    },
    {
      "epoch": 0.7411062299953227,
      "grad_norm": 1.554013982121055,
      "learning_rate": 1.656702312870359e-06,
      "loss": 0.7365,
      "step": 10101
    },
    {
      "epoch": 0.7411795995854619,
      "grad_norm": 1.4317859786018876,
      "learning_rate": 1.6558189012333876e-06,
      "loss": 0.7246,
      "step": 10102
    },
    {
      "epoch": 0.7412529691756009,
      "grad_norm": 1.3576235368468248,
      "learning_rate": 1.654935678450606e-06,
      "loss": 0.7021,
      "step": 10103
    },
    {
      "epoch": 0.7413263387657401,
      "grad_norm": 1.5114936419247458,
      "learning_rate": 1.6540526445718907e-06,
      "loss": 0.6924,
      "step": 10104
    },
    {
      "epoch": 0.7413263387657401,
      "lm_loss": 1.4416,
      "step": 10104,
      "vm_loss": 0.0589
    },
    {
      "epoch": 0.7413263387657401,
      "lm_loss": 1.1545,
      "step": 10104,
      "vm_loss": 0.0978
    },
    {
      "epoch": 0.7413263387657401,
      "lm_loss": 0.5915,
      "step": 10104,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.7413263387657401,
      "lm_loss": 0.2114,
      "step": 10104,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.7413263387657401,
      "lm_loss": 0.3879,
      "step": 10104,
      "vm_loss": 0.0945
    },
    {
      "epoch": 0.7413263387657401,
      "lm_loss": 1.0354,
      "step": 10104,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.7413263387657401,
      "lm_loss": 1.1157,
      "step": 10104,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.7413263387657401,
      "lm_loss": 0.2647,
      "step": 10104,
      "vm_loss": 0.1453
    },
    {
      "epoch": 0.7413997083558792,
      "grad_norm": 1.4917466459293725,
      "learning_rate": 1.6531697996471085e-06,
      "loss": 0.7291,
      "step": 10105
    },
    {
      "epoch": 0.7414730779460184,
      "grad_norm": 1.5481379849115584,
      "learning_rate": 1.6522871437261161e-06,
      "loss": 0.7286,
      "step": 10106
    },
    {
      "epoch": 0.7415464475361575,
      "grad_norm": 0.54573676065603,
      "learning_rate": 1.6514046768587578e-06,
      "loss": 0.474,
      "step": 10107
    },
    {
      "epoch": 0.7416198171262965,
      "grad_norm": 0.5544217683651147,
      "learning_rate": 1.6505223990948715e-06,
      "loss": 0.4775,
      "step": 10108
    },
    {
      "epoch": 0.7416931867164357,
      "grad_norm": 0.5512526852662568,
      "learning_rate": 1.64964031048428e-06,
      "loss": 0.4839,
      "step": 10109
    },
    {
      "epoch": 0.7417665563065748,
      "grad_norm": 1.5314263627978635,
      "learning_rate": 1.6487584110767973e-06,
      "loss": 0.8101,
      "step": 10110
    },
    {
      "epoch": 0.741839925896714,
      "grad_norm": 0.5366691800470815,
      "learning_rate": 1.6478767009222263e-06,
      "loss": 0.4855,
      "step": 10111
    },
    {
      "epoch": 0.7419132954868531,
      "grad_norm": 1.4199441913866098,
      "learning_rate": 1.6469951800703577e-06,
      "loss": 0.7161,
      "step": 10112
    },
    {
      "epoch": 0.7419132954868531,
      "lm_loss": 1.1714,
      "step": 10112,
      "vm_loss": 0.0652
    },
    {
      "epoch": 0.7419132954868531,
      "lm_loss": 1.3986,
      "step": 10112,
      "vm_loss": 0.1158
    },
    {
      "epoch": 0.7419132954868531,
      "lm_loss": 0.8738,
      "step": 10112,
      "vm_loss": 0.1924
    },
    {
      "epoch": 0.7419132954868531,
      "lm_loss": 0.7729,
      "step": 10112,
      "vm_loss": 0.165
    },
    {
      "epoch": 0.7419132954868531,
      "lm_loss": 0.3149,
      "step": 10112,
      "vm_loss": 0.0781
    },
    {
      "epoch": 0.7419132954868531,
      "lm_loss": 0.0318,
      "step": 10112,
      "vm_loss": 0.0661
    },
    {
      "epoch": 0.7419132954868531,
      "lm_loss": 0.5927,
      "step": 10112,
      "vm_loss": 0.1591
    },
    {
      "epoch": 0.7419132954868531,
      "lm_loss": 0.8085,
      "step": 10112,
      "vm_loss": 0.115
    },
    {
      "epoch": 0.7419866650769922,
      "grad_norm": 1.470039384224791,
      "learning_rate": 1.6461138485709765e-06,
      "loss": 0.7943,
      "step": 10113
    },
    {
      "epoch": 0.7420600346671313,
      "grad_norm": 1.635503172086332,
      "learning_rate": 1.645232706473852e-06,
      "loss": 0.8289,
      "step": 10114
    },
    {
      "epoch": 0.7421334042572705,
      "grad_norm": 1.5809633145261477,
      "learning_rate": 1.6443517538287452e-06,
      "loss": 0.8396,
      "step": 10115
    },
    {
      "epoch": 0.7422067738474096,
      "grad_norm": 0.5568262741883137,
      "learning_rate": 1.643470990685404e-06,
      "loss": 0.4823,
      "step": 10116
    },
    {
      "epoch": 0.7422801434375487,
      "grad_norm": 1.5511132093127504,
      "learning_rate": 1.6425904170935692e-06,
      "loss": 0.7688,
      "step": 10117
    },
    {
      "epoch": 0.7423535130276878,
      "grad_norm": 1.432840331817781,
      "learning_rate": 1.6417100331029657e-06,
      "loss": 0.7251,
      "step": 10118
    },
    {
      "epoch": 0.742426882617827,
      "grad_norm": 1.4774087286119753,
      "learning_rate": 1.6408298387633148e-06,
      "loss": 0.7025,
      "step": 10119
    },
    {
      "epoch": 0.7425002522079661,
      "grad_norm": 0.5807670807199634,
      "learning_rate": 1.6399498341243225e-06,
      "loss": 0.4924,
      "step": 10120
    },
    {
      "epoch": 0.7425002522079661,
      "lm_loss": 1.2332,
      "step": 10120,
      "vm_loss": 0.0465
    },
    {
      "epoch": 0.7425002522079661,
      "lm_loss": 0.8213,
      "step": 10120,
      "vm_loss": 0.1463
    },
    {
      "epoch": 0.7425002522079661,
      "lm_loss": 0.0942,
      "step": 10120,
      "vm_loss": 0.1711
    },
    {
      "epoch": 0.7425002522079661,
      "lm_loss": 1.0394,
      "step": 10120,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.7425002522079661,
      "lm_loss": 0.3807,
      "step": 10120,
      "vm_loss": 0.0934
    },
    {
      "epoch": 0.7425002522079661,
      "lm_loss": 0.985,
      "step": 10120,
      "vm_loss": 0.1562
    },
    {
      "epoch": 0.7425002522079661,
      "lm_loss": 0.4441,
      "step": 10120,
      "vm_loss": 0.1392
    },
    {
      "epoch": 0.7425002522079661,
      "lm_loss": 0.0948,
      "step": 10120,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.7425736217981053,
      "grad_norm": 1.3052140795375777,
      "learning_rate": 1.6390700192356834e-06,
      "loss": 0.6694,
      "step": 10121
    },
    {
      "epoch": 0.7426469913882443,
      "grad_norm": 1.8710560445470752,
      "learning_rate": 1.638190394147084e-06,
      "loss": 0.7381,
      "step": 10122
    },
    {
      "epoch": 0.7427203609783835,
      "grad_norm": 1.6532827312126275,
      "learning_rate": 1.6373109589081965e-06,
      "loss": 0.7038,
      "step": 10123
    },
    {
      "epoch": 0.7427937305685226,
      "grad_norm": 1.5849302560270833,
      "learning_rate": 1.636431713568688e-06,
      "loss": 0.7017,
      "step": 10124
    },
    {
      "epoch": 0.7428671001586618,
      "grad_norm": 2.0003165082279026,
      "learning_rate": 1.6355526581782111e-06,
      "loss": 0.825,
      "step": 10125
    },
    {
      "epoch": 0.7429404697488009,
      "grad_norm": 1.561410695960244,
      "learning_rate": 1.634673792786407e-06,
      "loss": 0.7582,
      "step": 10126
    },
    {
      "epoch": 0.7430138393389399,
      "grad_norm": 0.5437798823723501,
      "learning_rate": 1.6337951174429083e-06,
      "loss": 0.4749,
      "step": 10127
    },
    {
      "epoch": 0.7430872089290791,
      "grad_norm": 0.5659103348588598,
      "learning_rate": 1.632916632197335e-06,
      "loss": 0.4788,
      "step": 10128
    },
    {
      "epoch": 0.7430872089290791,
      "lm_loss": 1.2542,
      "step": 10128,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.7430872089290791,
      "lm_loss": 0.1379,
      "step": 10128,
      "vm_loss": 0.0833
    },
    {
      "epoch": 0.7430872089290791,
      "lm_loss": 0.9602,
      "step": 10128,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.7430872089290791,
      "lm_loss": 0.3348,
      "step": 10128,
      "vm_loss": 0.1318
    },
    {
      "epoch": 0.7430872089290791,
      "lm_loss": 0.8466,
      "step": 10128,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.7430872089290791,
      "lm_loss": 0.3861,
      "step": 10128,
      "vm_loss": 0.0925
    },
    {
      "epoch": 0.7430872089290791,
      "lm_loss": 0.2537,
      "step": 10128,
      "vm_loss": 0.1319
    },
    {
      "epoch": 0.7430872089290791,
      "lm_loss": 1.0166,
      "step": 10128,
      "vm_loss": 0.1484
    },
    {
      "epoch": 0.7431605785192182,
      "grad_norm": 1.709173663853846,
      "learning_rate": 1.632038337099297e-06,
      "loss": 0.7627,
      "step": 10129
    },
    {
      "epoch": 0.7432339481093574,
      "grad_norm": 0.5616238330335829,
      "learning_rate": 1.631160232198396e-06,
      "loss": 0.4731,
      "step": 10130
    },
    {
      "epoch": 0.7433073176994965,
      "grad_norm": 1.5266005862152043,
      "learning_rate": 1.6302823175442195e-06,
      "loss": 0.7603,
      "step": 10131
    },
    {
      "epoch": 0.7433806872896356,
      "grad_norm": 1.5059974442742947,
      "learning_rate": 1.6294045931863456e-06,
      "loss": 0.7896,
      "step": 10132
    },
    {
      "epoch": 0.7434540568797747,
      "grad_norm": 1.4918100942858623,
      "learning_rate": 1.6285270591743414e-06,
      "loss": 0.7799,
      "step": 10133
    },
    {
      "epoch": 0.7435274264699139,
      "grad_norm": 1.6169543735984693,
      "learning_rate": 1.6276497155577614e-06,
      "loss": 0.7269,
      "step": 10134
    },
    {
      "epoch": 0.743600796060053,
      "grad_norm": 0.5594309206706513,
      "learning_rate": 1.6267725623861558e-06,
      "loss": 0.4733,
      "step": 10135
    },
    {
      "epoch": 0.7436741656501922,
      "grad_norm": 1.389277205523162,
      "learning_rate": 1.625895599709057e-06,
      "loss": 0.8068,
      "step": 10136
    },
    {
      "epoch": 0.7436741656501922,
      "lm_loss": 0.2383,
      "step": 10136,
      "vm_loss": 0.0601
    },
    {
      "epoch": 0.7436741656501922,
      "lm_loss": 1.1843,
      "step": 10136,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.7436741656501922,
      "lm_loss": 0.7887,
      "step": 10136,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.7436741656501922,
      "lm_loss": 0.8478,
      "step": 10136,
      "vm_loss": 0.1983
    },
    {
      "epoch": 0.7436741656501922,
      "lm_loss": 0.734,
      "step": 10136,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.7436741656501922,
      "lm_loss": 1.0263,
      "step": 10136,
      "vm_loss": 0.2067
    },
    {
      "epoch": 0.7436741656501922,
      "lm_loss": 0.6699,
      "step": 10136,
      "vm_loss": 0.1705
    },
    {
      "epoch": 0.7436741656501922,
      "lm_loss": 0.9992,
      "step": 10136,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.7437475352403312,
      "grad_norm": 1.6300190450511913,
      "learning_rate": 1.6250188275759893e-06,
      "loss": 0.8166,
      "step": 10137
    },
    {
      "epoch": 0.7438209048304704,
      "grad_norm": 1.5204758369373388,
      "learning_rate": 1.6241422460364664e-06,
      "loss": 0.7635,
      "step": 10138
    },
    {
      "epoch": 0.7438942744206095,
      "grad_norm": 1.4854766472757077,
      "learning_rate": 1.6232658551399895e-06,
      "loss": 0.7585,
      "step": 10139
    },
    {
      "epoch": 0.7439676440107487,
      "grad_norm": 0.5463153572095104,
      "learning_rate": 1.622389654936054e-06,
      "loss": 0.4545,
      "step": 10140
    },
    {
      "epoch": 0.7440410136008878,
      "grad_norm": 1.556710450493138,
      "learning_rate": 1.6215136454741387e-06,
      "loss": 0.7628,
      "step": 10141
    },
    {
      "epoch": 0.7441143831910269,
      "grad_norm": 1.3631231554391354,
      "learning_rate": 1.6206378268037148e-06,
      "loss": 0.7364,
      "step": 10142
    },
    {
      "epoch": 0.744187752781166,
      "grad_norm": 1.7579385089795412,
      "learning_rate": 1.6197621989742419e-06,
      "loss": 0.7111,
      "step": 10143
    },
    {
      "epoch": 0.7442611223713052,
      "grad_norm": 1.8758612442576474,
      "learning_rate": 1.6188867620351683e-06,
      "loss": 0.8499,
      "step": 10144
    },
    {
      "epoch": 0.7442611223713052,
      "lm_loss": 0.6281,
      "step": 10144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7442611223713052,
      "lm_loss": 0.4681,
      "step": 10144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7442611223713052,
      "lm_loss": 0.5043,
      "step": 10144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7442611223713052,
      "lm_loss": 0.4974,
      "step": 10144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7442611223713052,
      "lm_loss": 0.7098,
      "step": 10144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7442611223713052,
      "lm_loss": 0.428,
      "step": 10144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7442611223713052,
      "lm_loss": 0.4904,
      "step": 10144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7442611223713052,
      "lm_loss": 0.4835,
      "step": 10144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7443344919614443,
      "grad_norm": 0.5692270845450768,
      "learning_rate": 1.618011516035931e-06,
      "loss": 0.4905,
      "step": 10145
    },
    {
      "epoch": 0.7444078615515834,
      "grad_norm": 1.651361755378365,
      "learning_rate": 1.6171364610259605e-06,
      "loss": 0.7747,
      "step": 10146
    },
    {
      "epoch": 0.7444812311417225,
      "grad_norm": 1.4644135596150023,
      "learning_rate": 1.6162615970546713e-06,
      "loss": 0.7784,
      "step": 10147
    },
    {
      "epoch": 0.7445546007318616,
      "grad_norm": 0.5665114019416635,
      "learning_rate": 1.6153869241714694e-06,
      "loss": 0.4935,
      "step": 10148
    },
    {
      "epoch": 0.7446279703220008,
      "grad_norm": 1.862276104469278,
      "learning_rate": 1.6145124424257497e-06,
      "loss": 0.7202,
      "step": 10149
    },
    {
      "epoch": 0.7447013399121399,
      "grad_norm": 0.5494830269637715,
      "learning_rate": 1.6136381518668947e-06,
      "loss": 0.4871,
      "step": 10150
    },
    {
      "epoch": 0.7447747095022791,
      "grad_norm": 1.5252249153258357,
      "learning_rate": 1.6127640525442816e-06,
      "loss": 0.7282,
      "step": 10151
    },
    {
      "epoch": 0.7448480790924181,
      "grad_norm": 1.4252802040766914,
      "learning_rate": 1.6118901445072705e-06,
      "loss": 0.701,
      "step": 10152
    },
    {
      "epoch": 0.7448480790924181,
      "lm_loss": 1.582,
      "step": 10152,
      "vm_loss": 0.0269
    },
    {
      "epoch": 0.7448480790924181,
      "lm_loss": 0.0327,
      "step": 10152,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.7448480790924181,
      "lm_loss": 0.2352,
      "step": 10152,
      "vm_loss": 0.1554
    },
    {
      "epoch": 0.7448480790924181,
      "lm_loss": 1.58,
      "step": 10152,
      "vm_loss": 0.2535
    },
    {
      "epoch": 0.7448480790924181,
      "lm_loss": 0.6368,
      "step": 10152,
      "vm_loss": 0.1767
    },
    {
      "epoch": 0.7448480790924181,
      "lm_loss": 0.3984,
      "step": 10152,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.7448480790924181,
      "lm_loss": 0.5363,
      "step": 10152,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.7448480790924181,
      "lm_loss": 0.6898,
      "step": 10152,
      "vm_loss": 0.1585
    },
    {
      "epoch": 0.7449214486825573,
      "grad_norm": 1.5905025403538764,
      "learning_rate": 1.6110164278052131e-06,
      "loss": 0.7411,
      "step": 10153
    },
    {
      "epoch": 0.7449948182726964,
      "grad_norm": 0.5650114155127164,
      "learning_rate": 1.610142902487451e-06,
      "loss": 0.4804,
      "step": 10154
    },
    {
      "epoch": 0.7450681878628356,
      "grad_norm": 1.6362392637924141,
      "learning_rate": 1.6092695686033144e-06,
      "loss": 0.7145,
      "step": 10155
    },
    {
      "epoch": 0.7451415574529747,
      "grad_norm": 0.541231597544142,
      "learning_rate": 1.6083964262021201e-06,
      "loss": 0.4573,
      "step": 10156
    },
    {
      "epoch": 0.7452149270431138,
      "grad_norm": 1.5891597923450154,
      "learning_rate": 1.6075234753331804e-06,
      "loss": 0.8296,
      "step": 10157
    },
    {
      "epoch": 0.7452882966332529,
      "grad_norm": 1.4416837499121287,
      "learning_rate": 1.6066507160457916e-06,
      "loss": 0.6476,
      "step": 10158
    },
    {
      "epoch": 0.7453616662233921,
      "grad_norm": 1.7961389032773916,
      "learning_rate": 1.6057781483892403e-06,
      "loss": 0.8105,
      "step": 10159
    },
    {
      "epoch": 0.7454350358135312,
      "grad_norm": 1.3997662818986987,
      "learning_rate": 1.6049057724128026e-06,
      "loss": 0.6436,
      "step": 10160
    },
    {
      "epoch": 0.7454350358135312,
      "lm_loss": 0.3884,
      "step": 10160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7454350358135312,
      "lm_loss": 0.6609,
      "step": 10160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7454350358135312,
      "lm_loss": 0.4641,
      "step": 10160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7454350358135312,
      "lm_loss": 0.3473,
      "step": 10160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7454350358135312,
      "lm_loss": 0.4691,
      "step": 10160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7454350358135312,
      "lm_loss": 0.5625,
      "step": 10160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7454350358135312,
      "lm_loss": 0.5247,
      "step": 10160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7454350358135312,
      "lm_loss": 0.4875,
      "step": 10160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7455084054036704,
      "grad_norm": 0.5353766180840849,
      "learning_rate": 1.604033588165742e-06,
      "loss": 0.4774,
      "step": 10161
    },
    {
      "epoch": 0.7455817749938094,
      "grad_norm": 0.5338137508252213,
      "learning_rate": 1.6031615956973168e-06,
      "loss": 0.4632,
      "step": 10162
    },
    {
      "epoch": 0.7456551445839485,
      "grad_norm": 1.334864139525949,
      "learning_rate": 1.6022897950567678e-06,
      "loss": 0.6907,
      "step": 10163
    },
    {
      "epoch": 0.7457285141740877,
      "grad_norm": 1.4901392096236843,
      "learning_rate": 1.6014181862933287e-06,
      "loss": 0.7424,
      "step": 10164
    },
    {
      "epoch": 0.7458018837642268,
      "grad_norm": 0.5553097638185626,
      "learning_rate": 1.600546769456221e-06,
      "loss": 0.4908,
      "step": 10165
    },
    {
      "epoch": 0.745875253354366,
      "grad_norm": 1.4599986143100334,
      "learning_rate": 1.5996755445946539e-06,
      "loss": 0.7749,
      "step": 10166
    },
    {
      "epoch": 0.745948622944505,
      "grad_norm": 0.5412584727046174,
      "learning_rate": 1.5988045117578315e-06,
      "loss": 0.488,
      "step": 10167
    },
    {
      "epoch": 0.7460219925346442,
      "grad_norm": 1.419779868457701,
      "learning_rate": 1.5979336709949411e-06,
      "loss": 0.7021,
      "step": 10168
    },
    {
      "epoch": 0.7460219925346442,
      "lm_loss": 0.6578,
      "step": 10168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7460219925346442,
      "lm_loss": 0.2027,
      "step": 10168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7460219925346442,
      "lm_loss": 0.4562,
      "step": 10168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7460219925346442,
      "lm_loss": 0.3836,
      "step": 10168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7460219925346442,
      "lm_loss": 0.3742,
      "step": 10168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7460219925346442,
      "lm_loss": 0.28,
      "step": 10168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7460219925346442,
      "lm_loss": 0.6594,
      "step": 10168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7460219925346442,
      "lm_loss": 0.5952,
      "step": 10168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7460953621247833,
      "grad_norm": 0.5622160132241121,
      "learning_rate": 1.5970630223551614e-06,
      "loss": 0.4772,
      "step": 10169
    },
    {
      "epoch": 0.7461687317149225,
      "grad_norm": 0.5543532705914206,
      "learning_rate": 1.5961925658876603e-06,
      "loss": 0.4797,
      "step": 10170
    },
    {
      "epoch": 0.7462421013050616,
      "grad_norm": 1.7347856762474891,
      "learning_rate": 1.5953223016415941e-06,
      "loss": 0.6775,
      "step": 10171
    },
    {
      "epoch": 0.7463154708952007,
      "grad_norm": 0.5718546766650362,
      "learning_rate": 1.5944522296661075e-06,
      "loss": 0.484,
      "step": 10172
    },
    {
      "epoch": 0.7463888404853398,
      "grad_norm": 2.0390034348142807,
      "learning_rate": 1.5935823500103387e-06,
      "loss": 0.7791,
      "step": 10173
    },
    {
      "epoch": 0.746462210075479,
      "grad_norm": 1.448142794062296,
      "learning_rate": 1.5927126627234108e-06,
      "loss": 0.7149,
      "step": 10174
    },
    {
      "epoch": 0.7465355796656181,
      "grad_norm": 1.40019386825327,
      "learning_rate": 1.5918431678544366e-06,
      "loss": 0.7833,
      "step": 10175
    },
    {
      "epoch": 0.7466089492557573,
      "grad_norm": 0.5616291570774221,
      "learning_rate": 1.5909738654525191e-06,
      "loss": 0.4836,
      "step": 10176
    },
    {
      "epoch": 0.7466089492557573,
      "lm_loss": 0.553,
      "step": 10176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7466089492557573,
      "lm_loss": 0.4762,
      "step": 10176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7466089492557573,
      "lm_loss": 0.6344,
      "step": 10176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7466089492557573,
      "lm_loss": 0.5506,
      "step": 10176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7466089492557573,
      "lm_loss": 0.3484,
      "step": 10176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7466089492557573,
      "lm_loss": 0.5688,
      "step": 10176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7466089492557573,
      "lm_loss": 0.5818,
      "step": 10176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7466089492557573,
      "lm_loss": 0.4275,
      "step": 10176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7466823188458963,
      "grad_norm": 0.5553320472619078,
      "learning_rate": 1.5901047555667476e-06,
      "loss": 0.4772,
      "step": 10177
    },
    {
      "epoch": 0.7467556884360355,
      "grad_norm": 1.3535046155907116,
      "learning_rate": 1.589235838246207e-06,
      "loss": 0.7066,
      "step": 10178
    },
    {
      "epoch": 0.7468290580261746,
      "grad_norm": 1.5497311319556033,
      "learning_rate": 1.5883671135399652e-06,
      "loss": 0.8573,
      "step": 10179
    },
    {
      "epoch": 0.7469024276163138,
      "grad_norm": 0.5429203733860904,
      "learning_rate": 1.5874985814970812e-06,
      "loss": 0.468,
      "step": 10180
    },
    {
      "epoch": 0.7469757972064529,
      "grad_norm": 0.5661828711057991,
      "learning_rate": 1.5866302421666035e-06,
      "loss": 0.4748,
      "step": 10181
    },
    {
      "epoch": 0.747049166796592,
      "grad_norm": 1.4672431302210376,
      "learning_rate": 1.5857620955975688e-06,
      "loss": 0.6618,
      "step": 10182
    },
    {
      "epoch": 0.7471225363867311,
      "grad_norm": 1.4628446660182954,
      "learning_rate": 1.5848941418390023e-06,
      "loss": 0.8125,
      "step": 10183
    },
    {
      "epoch": 0.7471959059768702,
      "grad_norm": 1.3285308579439996,
      "learning_rate": 1.5840263809399226e-06,
      "loss": 0.7245,
      "step": 10184
    },
    {
      "epoch": 0.7471959059768702,
      "lm_loss": 0.7725,
      "step": 10184,
      "vm_loss": 0.0545
    },
    {
      "epoch": 0.7471959059768702,
      "lm_loss": 0.6754,
      "step": 10184,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.7471959059768702,
      "lm_loss": 0.6438,
      "step": 10184,
      "vm_loss": 0.1288
    },
    {
      "epoch": 0.7471959059768702,
      "lm_loss": 1.2555,
      "step": 10184,
      "vm_loss": 0.1439
    },
    {
      "epoch": 0.7471959059768702,
      "lm_loss": 0.9066,
      "step": 10184,
      "vm_loss": 0.1376
    },
    {
      "epoch": 0.7471959059768702,
      "lm_loss": 0.7889,
      "step": 10184,
      "vm_loss": 0.103
    },
    {
      "epoch": 0.7471959059768702,
      "lm_loss": 0.8291,
      "step": 10184,
      "vm_loss": 0.1397
    },
    {
      "epoch": 0.7471959059768702,
      "lm_loss": 0.1545,
      "step": 10184,
      "vm_loss": 0.0776
    },
    {
      "epoch": 0.7472692755670094,
      "grad_norm": 1.6040056763402528,
      "learning_rate": 1.5831588129493325e-06,
      "loss": 0.744,
      "step": 10185
    },
    {
      "epoch": 0.7473426451571485,
      "grad_norm": 1.539793777721206,
      "learning_rate": 1.5822914379162258e-06,
      "loss": 0.7117,
      "step": 10186
    },
    {
      "epoch": 0.7474160147472876,
      "grad_norm": 0.5587706438756229,
      "learning_rate": 1.581424255889586e-06,
      "loss": 0.4865,
      "step": 10187
    },
    {
      "epoch": 0.7474893843374267,
      "grad_norm": 1.5009828230142666,
      "learning_rate": 1.5805572669183823e-06,
      "loss": 0.7637,
      "step": 10188
    },
    {
      "epoch": 0.7475627539275659,
      "grad_norm": 0.5538814865072954,
      "learning_rate": 1.5796904710515792e-06,
      "loss": 0.4868,
      "step": 10189
    },
    {
      "epoch": 0.747636123517705,
      "grad_norm": 0.5531866253876361,
      "learning_rate": 1.5788238683381262e-06,
      "loss": 0.4592,
      "step": 10190
    },
    {
      "epoch": 0.7477094931078442,
      "grad_norm": 1.3657950907999192,
      "learning_rate": 1.5779574588269615e-06,
      "loss": 0.693,
      "step": 10191
    },
    {
      "epoch": 0.7477828626979832,
      "grad_norm": 0.55466302591192,
      "learning_rate": 1.5770912425670133e-06,
      "loss": 0.4712,
      "step": 10192
    },
    {
      "epoch": 0.7477828626979832,
      "lm_loss": 1.3604,
      "step": 10192,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.7477828626979832,
      "lm_loss": 0.4745,
      "step": 10192,
      "vm_loss": 0.0655
    },
    {
      "epoch": 0.7477828626979832,
      "lm_loss": 0.5238,
      "step": 10192,
      "vm_loss": 0.1448
    },
    {
      "epoch": 0.7477828626979832,
      "lm_loss": 0.1059,
      "step": 10192,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.7477828626979832,
      "lm_loss": 0.5013,
      "step": 10192,
      "vm_loss": 0.1568
    },
    {
      "epoch": 0.7477828626979832,
      "lm_loss": 0.3833,
      "step": 10192,
      "vm_loss": 0.1388
    },
    {
      "epoch": 0.7477828626979832,
      "lm_loss": 0.451,
      "step": 10192,
      "vm_loss": 0.1624
    },
    {
      "epoch": 0.7477828626979832,
      "lm_loss": 0.1446,
      "step": 10192,
      "vm_loss": 0.1333
    },
    {
      "epoch": 0.7478562322881224,
      "grad_norm": 1.474991166287039,
      "learning_rate": 1.576225219607198e-06,
      "loss": 0.6123,
      "step": 10193
    },
    {
      "epoch": 0.7479296018782615,
      "grad_norm": 1.599796636644515,
      "learning_rate": 1.5753593899964247e-06,
      "loss": 0.6779,
      "step": 10194
    },
    {
      "epoch": 0.7480029714684007,
      "grad_norm": 0.5379559745813194,
      "learning_rate": 1.574493753783588e-06,
      "loss": 0.4858,
      "step": 10195
    },
    {
      "epoch": 0.7480763410585397,
      "grad_norm": 1.5468702940664447,
      "learning_rate": 1.573628311017572e-06,
      "loss": 0.7436,
      "step": 10196
    },
    {
      "epoch": 0.7481497106486789,
      "grad_norm": 1.753453742740486,
      "learning_rate": 1.5727630617472505e-06,
      "loss": 0.7207,
      "step": 10197
    },
    {
      "epoch": 0.748223080238818,
      "grad_norm": 1.6891138688365264,
      "learning_rate": 1.5718980060214866e-06,
      "loss": 0.7281,
      "step": 10198
    },
    {
      "epoch": 0.7482964498289572,
      "grad_norm": 1.4294241028629542,
      "learning_rate": 1.5710331438891302e-06,
      "loss": 0.7426,
      "step": 10199
    },
    {
      "epoch": 0.7483698194190963,
      "grad_norm": 1.2674234078873405,
      "learning_rate": 1.5701684753990248e-06,
      "loss": 0.8178,
      "step": 10200
    },
    {
      "epoch": 0.7483698194190963,
      "lm_loss": 1.2001,
      "step": 10200,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.7483698194190963,
      "lm_loss": 1.564,
      "step": 10200,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.7483698194190963,
      "lm_loss": 0.2808,
      "step": 10200,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.7483698194190963,
      "lm_loss": 0.5403,
      "step": 10200,
      "vm_loss": 0.2502
    },
    {
      "epoch": 0.7483698194190963,
      "lm_loss": 0.2148,
      "step": 10200,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.7483698194190963,
      "lm_loss": 0.5108,
      "step": 10200,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.7483698194190963,
      "lm_loss": 0.6211,
      "step": 10200,
      "vm_loss": 0.2004
    },
    {
      "epoch": 0.7483698194190963,
      "lm_loss": 0.59,
      "step": 10200,
      "vm_loss": 0.1294
    },
    {
      "epoch": 0.7484431890092353,
      "grad_norm": 1.3319587268397135,
      "learning_rate": 1.5693040006000005e-06,
      "loss": 0.762,
      "step": 10201
    },
    {
      "epoch": 0.7485165585993745,
      "grad_norm": 1.4169377231277516,
      "learning_rate": 1.5684397195408747e-06,
      "loss": 0.752,
      "step": 10202
    },
    {
      "epoch": 0.7485899281895136,
      "grad_norm": 1.3985639736445494,
      "learning_rate": 1.5675756322704554e-06,
      "loss": 0.6307,
      "step": 10203
    },
    {
      "epoch": 0.7486632977796528,
      "grad_norm": 1.6369695639945196,
      "learning_rate": 1.5667117388375396e-06,
      "loss": 0.734,
      "step": 10204
    },
    {
      "epoch": 0.7487366673697919,
      "grad_norm": 1.3734092151875328,
      "learning_rate": 1.5658480392909158e-06,
      "loss": 0.665,
      "step": 10205
    },
    {
      "epoch": 0.748810036959931,
      "grad_norm": 1.7567026534159944,
      "learning_rate": 1.5649845336793579e-06,
      "loss": 0.7867,
      "step": 10206
    },
    {
      "epoch": 0.7488834065500701,
      "grad_norm": 1.4975969488519165,
      "learning_rate": 1.5641212220516295e-06,
      "loss": 0.7528,
      "step": 10207
    },
    {
      "epoch": 0.7489567761402093,
      "grad_norm": 0.5538308486859115,
      "learning_rate": 1.5632581044564849e-06,
      "loss": 0.4693,
      "step": 10208
    },
    {
      "epoch": 0.7489567761402093,
      "lm_loss": 0.8506,
      "step": 10208,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.7489567761402093,
      "lm_loss": 0.0198,
      "step": 10208,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.7489567761402093,
      "lm_loss": 0.5175,
      "step": 10208,
      "vm_loss": 0.168
    },
    {
      "epoch": 0.7489567761402093,
      "lm_loss": 0.8031,
      "step": 10208,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.7489567761402093,
      "lm_loss": 0.905,
      "step": 10208,
      "vm_loss": 0.0796
    },
    {
      "epoch": 0.7489567761402093,
      "lm_loss": 0.6425,
      "step": 10208,
      "vm_loss": 0.1843
    },
    {
      "epoch": 0.7489567761402093,
      "lm_loss": 0.1656,
      "step": 10208,
      "vm_loss": 0.1795
    },
    {
      "epoch": 0.7489567761402093,
      "lm_loss": 0.8445,
      "step": 10208,
      "vm_loss": 0.1364
    },
    {
      "epoch": 0.7490301457303484,
      "grad_norm": 1.4345160960397931,
      "learning_rate": 1.5623951809426663e-06,
      "loss": 0.7438,
      "step": 10209
    },
    {
      "epoch": 0.7491035153204876,
      "grad_norm": 1.8127238100025185,
      "learning_rate": 1.5615324515589031e-06,
      "loss": 0.7079,
      "step": 10210
    },
    {
      "epoch": 0.7491768849106266,
      "grad_norm": 1.4858999618247377,
      "learning_rate": 1.560669916353919e-06,
      "loss": 0.702,
      "step": 10211
    },
    {
      "epoch": 0.7492502545007658,
      "grad_norm": 0.5824137818409482,
      "learning_rate": 1.559807575376422e-06,
      "loss": 0.4793,
      "step": 10212
    },
    {
      "epoch": 0.7493236240909049,
      "grad_norm": 0.5554063854986668,
      "learning_rate": 1.558945428675111e-06,
      "loss": 0.4871,
      "step": 10213
    },
    {
      "epoch": 0.7493969936810441,
      "grad_norm": 1.617238238950329,
      "learning_rate": 1.5580834762986735e-06,
      "loss": 0.7116,
      "step": 10214
    },
    {
      "epoch": 0.7494703632711832,
      "grad_norm": 1.5567623316140924,
      "learning_rate": 1.5572217182957838e-06,
      "loss": 0.7235,
      "step": 10215
    },
    {
      "epoch": 0.7495437328613223,
      "grad_norm": 0.5684032939199933,
      "learning_rate": 1.5563601547151113e-06,
      "loss": 0.4892,
      "step": 10216
    },
    {
      "epoch": 0.7495437328613223,
      "lm_loss": 1.307,
      "step": 10216,
      "vm_loss": 0.0402
    },
    {
      "epoch": 0.7495437328613223,
      "lm_loss": 0.0066,
      "step": 10216,
      "vm_loss": 0.1392
    },
    {
      "epoch": 0.7495437328613223,
      "lm_loss": 0.1184,
      "step": 10216,
      "vm_loss": 0.1116
    },
    {
      "epoch": 0.7495437328613223,
      "lm_loss": 0.6856,
      "step": 10216,
      "vm_loss": 0.0701
    },
    {
      "epoch": 0.7495437328613223,
      "lm_loss": 0.0638,
      "step": 10216,
      "vm_loss": 0.168
    },
    {
      "epoch": 0.7495437328613223,
      "lm_loss": 0.7399,
      "step": 10216,
      "vm_loss": 0.1079
    },
    {
      "epoch": 0.7495437328613223,
      "lm_loss": 1.9875,
      "step": 10216,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.7495437328613223,
      "lm_loss": 0.0761,
      "step": 10216,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.7496171024514614,
      "grad_norm": 1.4427340497034895,
      "learning_rate": 1.555498785605309e-06,
      "loss": 0.7746,
      "step": 10217
    },
    {
      "epoch": 0.7496904720416006,
      "grad_norm": 1.556647573669554,
      "learning_rate": 1.5546376110150203e-06,
      "loss": 0.8136,
      "step": 10218
    },
    {
      "epoch": 0.7497638416317397,
      "grad_norm": 1.4954831676286229,
      "learning_rate": 1.553776630992878e-06,
      "loss": 0.7088,
      "step": 10219
    },
    {
      "epoch": 0.7498372112218789,
      "grad_norm": 1.4626428777881573,
      "learning_rate": 1.552915845587502e-06,
      "loss": 0.7792,
      "step": 10220
    },
    {
      "epoch": 0.7499105808120179,
      "grad_norm": 1.6000729196941743,
      "learning_rate": 1.5520552548475065e-06,
      "loss": 0.694,
      "step": 10221
    },
    {
      "epoch": 0.749983950402157,
      "grad_norm": 0.5542897466613648,
      "learning_rate": 1.5511948588214899e-06,
      "loss": 0.4698,
      "step": 10222
    },
    {
      "epoch": 0.7500573199922962,
      "grad_norm": 0.5521905150969216,
      "learning_rate": 1.5503346575580398e-06,
      "loss": 0.4658,
      "step": 10223
    },
    {
      "epoch": 0.7501306895824353,
      "grad_norm": 1.6619700203108014,
      "learning_rate": 1.5494746511057351e-06,
      "loss": 0.7468,
      "step": 10224
    },
    {
      "epoch": 0.7501306895824353,
      "lm_loss": 1.3317,
      "step": 10224,
      "vm_loss": 0.0451
    },
    {
      "epoch": 0.7501306895824353,
      "lm_loss": 0.3202,
      "step": 10224,
      "vm_loss": 0.143
    },
    {
      "epoch": 0.7501306895824353,
      "lm_loss": 0.537,
      "step": 10224,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.7501306895824353,
      "lm_loss": 0.007,
      "step": 10224,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.7501306895824353,
      "lm_loss": 0.0937,
      "step": 10224,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.7501306895824353,
      "lm_loss": 0.5567,
      "step": 10224,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.7501306895824353,
      "lm_loss": 0.6523,
      "step": 10224,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.7501306895824353,
      "lm_loss": 0.0099,
      "step": 10224,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.7502040591725745,
      "grad_norm": 1.6481651215805624,
      "learning_rate": 1.5486148395131417e-06,
      "loss": 0.7405,
      "step": 10225
    },
    {
      "epoch": 0.7502774287627135,
      "grad_norm": 1.5472219330988668,
      "learning_rate": 1.5477552228288135e-06,
      "loss": 0.7678,
      "step": 10226
    },
    {
      "epoch": 0.7503507983528527,
      "grad_norm": 1.590581666703968,
      "learning_rate": 1.546895801101299e-06,
      "loss": 0.7196,
      "step": 10227
    },
    {
      "epoch": 0.7504241679429918,
      "grad_norm": 1.3739957684882338,
      "learning_rate": 1.5460365743791305e-06,
      "loss": 0.7316,
      "step": 10228
    },
    {
      "epoch": 0.750497537533131,
      "grad_norm": 1.5331796199797378,
      "learning_rate": 1.5451775427108302e-06,
      "loss": 0.6721,
      "step": 10229
    },
    {
      "epoch": 0.7505709071232701,
      "grad_norm": 1.4688582404053123,
      "learning_rate": 1.5443187061449094e-06,
      "loss": 0.6286,
      "step": 10230
    },
    {
      "epoch": 0.7506442767134092,
      "grad_norm": 1.4658306193941977,
      "learning_rate": 1.543460064729868e-06,
      "loss": 0.6775,
      "step": 10231
    },
    {
      "epoch": 0.7507176463035483,
      "grad_norm": 1.4463233472086319,
      "learning_rate": 1.542601618514198e-06,
      "loss": 0.7259,
      "step": 10232
    },
    {
      "epoch": 0.7507176463035483,
      "lm_loss": 1.1018,
      "step": 10232,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.7507176463035483,
      "lm_loss": 0.1983,
      "step": 10232,
      "vm_loss": 0.066
    },
    {
      "epoch": 0.7507176463035483,
      "lm_loss": 0.6994,
      "step": 10232,
      "vm_loss": 0.1499
    },
    {
      "epoch": 0.7507176463035483,
      "lm_loss": 0.9689,
      "step": 10232,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.7507176463035483,
      "lm_loss": 1.3229,
      "step": 10232,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.7507176463035483,
      "lm_loss": 0.1465,
      "step": 10232,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.7507176463035483,
      "lm_loss": 0.9026,
      "step": 10232,
      "vm_loss": 0.1118
    },
    {
      "epoch": 0.7507176463035483,
      "lm_loss": 0.099,
      "step": 10232,
      "vm_loss": 0.0729
    },
    {
      "epoch": 0.7507910158936875,
      "grad_norm": 1.4191881899232905,
      "learning_rate": 1.5417433675463766e-06,
      "loss": 0.6592,
      "step": 10233
    },
    {
      "epoch": 0.7508643854838266,
      "grad_norm": 1.5780450634652676,
      "learning_rate": 1.5408853118748719e-06,
      "loss": 0.7526,
      "step": 10234
    },
    {
      "epoch": 0.7509377550739658,
      "grad_norm": 1.4047913262952565,
      "learning_rate": 1.5400274515481394e-06,
      "loss": 0.7556,
      "step": 10235
    },
    {
      "epoch": 0.7510111246641048,
      "grad_norm": 0.5567530842999786,
      "learning_rate": 1.5391697866146254e-06,
      "loss": 0.4711,
      "step": 10236
    },
    {
      "epoch": 0.751084494254244,
      "grad_norm": 1.5041226493255286,
      "learning_rate": 1.5383123171227627e-06,
      "loss": 0.5632,
      "step": 10237
    },
    {
      "epoch": 0.7511578638443831,
      "grad_norm": 1.633107937243017,
      "learning_rate": 1.5374550431209779e-06,
      "loss": 0.755,
      "step": 10238
    },
    {
      "epoch": 0.7512312334345222,
      "grad_norm": 1.4582267359160823,
      "learning_rate": 1.5365979646576811e-06,
      "loss": 0.7256,
      "step": 10239
    },
    {
      "epoch": 0.7513046030246614,
      "grad_norm": 1.5729978773287283,
      "learning_rate": 1.5357410817812746e-06,
      "loss": 0.6883,
      "step": 10240
    },
    {
      "epoch": 0.7513046030246614,
      "lm_loss": 0.4469,
      "step": 10240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7513046030246614,
      "lm_loss": 0.5228,
      "step": 10240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7513046030246614,
      "lm_loss": 0.4502,
      "step": 10240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7513046030246614,
      "lm_loss": 0.5133,
      "step": 10240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7513046030246614,
      "lm_loss": 0.5288,
      "step": 10240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7513046030246614,
      "lm_loss": 0.5318,
      "step": 10240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7513046030246614,
      "lm_loss": 0.4948,
      "step": 10240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7513046030246614,
      "lm_loss": 0.4815,
      "step": 10240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7513779726148004,
      "grad_norm": 0.5669505585179405,
      "learning_rate": 1.534884394540148e-06,
      "loss": 0.477,
      "step": 10241
    },
    {
      "epoch": 0.7514513422049396,
      "grad_norm": 1.5698252628608669,
      "learning_rate": 1.5340279029826798e-06,
      "loss": 0.7327,
      "step": 10242
    },
    {
      "epoch": 0.7515247117950787,
      "grad_norm": 1.5363491318505447,
      "learning_rate": 1.5331716071572405e-06,
      "loss": 0.8066,
      "step": 10243
    },
    {
      "epoch": 0.7515980813852179,
      "grad_norm": 0.5628460567924923,
      "learning_rate": 1.5323155071121865e-06,
      "loss": 0.4676,
      "step": 10244
    },
    {
      "epoch": 0.751671450975357,
      "grad_norm": 1.6043133108561651,
      "learning_rate": 1.5314596028958628e-06,
      "loss": 0.6753,
      "step": 10245
    },
    {
      "epoch": 0.7517448205654961,
      "grad_norm": 0.5727085205040384,
      "learning_rate": 1.5306038945566048e-06,
      "loss": 0.4794,
      "step": 10246
    },
    {
      "epoch": 0.7518181901556352,
      "grad_norm": 1.6265279891412463,
      "learning_rate": 1.5297483821427355e-06,
      "loss": 0.7542,
      "step": 10247
    },
    {
      "epoch": 0.7518915597457744,
      "grad_norm": 0.5506390141819416,
      "learning_rate": 1.5288930657025709e-06,
      "loss": 0.4757,
      "step": 10248
    },
    {
      "epoch": 0.7518915597457744,
      "lm_loss": 0.4571,
      "step": 10248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7518915597457744,
      "lm_loss": 0.3721,
      "step": 10248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7518915597457744,
      "lm_loss": 0.5805,
      "step": 10248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7518915597457744,
      "lm_loss": 0.6314,
      "step": 10248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7518915597457744,
      "lm_loss": 0.4957,
      "step": 10248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7518915597457744,
      "lm_loss": 0.4283,
      "step": 10248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7518915597457744,
      "lm_loss": 0.5212,
      "step": 10248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7518915597457744,
      "lm_loss": 0.5249,
      "step": 10248,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7519649293359135,
      "grad_norm": 0.5696082557630886,
      "learning_rate": 1.5280379452844124e-06,
      "loss": 0.5082,
      "step": 10249
    },
    {
      "epoch": 0.7520382989260527,
      "grad_norm": 1.437127711200617,
      "learning_rate": 1.5271830209365474e-06,
      "loss": 0.668,
      "step": 10250
    },
    {
      "epoch": 0.7521116685161917,
      "grad_norm": 1.468334286367026,
      "learning_rate": 1.5263282927072575e-06,
      "loss": 0.7551,
      "step": 10251
    },
    {
      "epoch": 0.7521850381063309,
      "grad_norm": 1.6863505532706005,
      "learning_rate": 1.52547376064481e-06,
      "loss": 0.6465,
      "step": 10252
    },
    {
      "epoch": 0.75225840769647,
      "grad_norm": 0.5692256786742468,
      "learning_rate": 1.524619424797465e-06,
      "loss": 0.4963,
      "step": 10253
    },
    {
      "epoch": 0.7523317772866092,
      "grad_norm": 1.5541946107820623,
      "learning_rate": 1.5237652852134676e-06,
      "loss": 0.8521,
      "step": 10254
    },
    {
      "epoch": 0.7524051468767483,
      "grad_norm": 1.5627382055709378,
      "learning_rate": 1.522911341941054e-06,
      "loss": 0.6655,
      "step": 10255
    },
    {
      "epoch": 0.7524785164668873,
      "grad_norm": 1.5378142760594413,
      "learning_rate": 1.5220575950284472e-06,
      "loss": 0.7802,
      "step": 10256
    },
    {
      "epoch": 0.7524785164668873,
      "lm_loss": 1.4553,
      "step": 10256,
      "vm_loss": 0.0482
    },
    {
      "epoch": 0.7524785164668873,
      "lm_loss": 0.8009,
      "step": 10256,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.7524785164668873,
      "lm_loss": 0.0847,
      "step": 10256,
      "vm_loss": 0.0831
    },
    {
      "epoch": 0.7524785164668873,
      "lm_loss": 0.692,
      "step": 10256,
      "vm_loss": 0.1331
    },
    {
      "epoch": 0.7524785164668873,
      "lm_loss": 0.9179,
      "step": 10256,
      "vm_loss": 0.1463
    },
    {
      "epoch": 0.7524785164668873,
      "lm_loss": 0.887,
      "step": 10256,
      "vm_loss": 0.1549
    },
    {
      "epoch": 0.7524785164668873,
      "lm_loss": 0.3437,
      "step": 10256,
      "vm_loss": 0.1891
    },
    {
      "epoch": 0.7524785164668873,
      "lm_loss": 0.4952,
      "step": 10256,
      "vm_loss": 0.0514
    },
    {
      "epoch": 0.7525518860570265,
      "grad_norm": 1.5985896655212417,
      "learning_rate": 1.5212040445238596e-06,
      "loss": 0.7103,
      "step": 10257
    },
    {
      "epoch": 0.7526252556471656,
      "grad_norm": 1.549476954021431,
      "learning_rate": 1.520350690475496e-06,
      "loss": 0.7755,
      "step": 10258
    },
    {
      "epoch": 0.7526986252373048,
      "grad_norm": 0.5721604141441528,
      "learning_rate": 1.5194975329315465e-06,
      "loss": 0.4855,
      "step": 10259
    },
    {
      "epoch": 0.752771994827444,
      "grad_norm": 1.484236357000599,
      "learning_rate": 1.5186445719401904e-06,
      "loss": 0.7665,
      "step": 10260
    },
    {
      "epoch": 0.752845364417583,
      "grad_norm": 1.602231464135084,
      "learning_rate": 1.5177918075495968e-06,
      "loss": 0.7427,
      "step": 10261
    },
    {
      "epoch": 0.7529187340077221,
      "grad_norm": 1.729938028758312,
      "learning_rate": 1.5169392398079218e-06,
      "loss": 0.7539,
      "step": 10262
    },
    {
      "epoch": 0.7529921035978613,
      "grad_norm": 1.4942654394787713,
      "learning_rate": 1.5160868687633156e-06,
      "loss": 0.7832,
      "step": 10263
    },
    {
      "epoch": 0.7530654731880004,
      "grad_norm": 1.5680930753165747,
      "learning_rate": 1.5152346944639112e-06,
      "loss": 0.7689,
      "step": 10264
    },
    {
      "epoch": 0.7530654731880004,
      "lm_loss": 1.6859,
      "step": 10264,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.7530654731880004,
      "lm_loss": 0.6658,
      "step": 10264,
      "vm_loss": 0.0554
    },
    {
      "epoch": 0.7530654731880004,
      "lm_loss": 0.5904,
      "step": 10264,
      "vm_loss": 0.1123
    },
    {
      "epoch": 0.7530654731880004,
      "lm_loss": 0.113,
      "step": 10264,
      "vm_loss": 0.0551
    },
    {
      "epoch": 0.7530654731880004,
      "lm_loss": 0.6572,
      "step": 10264,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.7530654731880004,
      "lm_loss": 0.1365,
      "step": 10264,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.7530654731880004,
      "lm_loss": 0.1431,
      "step": 10264,
      "vm_loss": 0.1113
    },
    {
      "epoch": 0.7530654731880004,
      "lm_loss": 0.5005,
      "step": 10264,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.7531388427781396,
      "grad_norm": 1.460042922850332,
      "learning_rate": 1.5143827169578334e-06,
      "loss": 0.6659,
      "step": 10265
    },
    {
      "epoch": 0.7532122123682786,
      "grad_norm": 0.5403441539893629,
      "learning_rate": 1.513530936293196e-06,
      "loss": 0.4851,
      "step": 10266
    },
    {
      "epoch": 0.7532855819584178,
      "grad_norm": 1.5606549005679344,
      "learning_rate": 1.5126793525181004e-06,
      "loss": 0.7649,
      "step": 10267
    },
    {
      "epoch": 0.7533589515485569,
      "grad_norm": 1.76903714908455,
      "learning_rate": 1.5118279656806363e-06,
      "loss": 0.7074,
      "step": 10268
    },
    {
      "epoch": 0.7534323211386961,
      "grad_norm": 0.5575148871279904,
      "learning_rate": 1.510976775828887e-06,
      "loss": 0.48,
      "step": 10269
    },
    {
      "epoch": 0.7535056907288352,
      "grad_norm": 1.6084488240201218,
      "learning_rate": 1.5101257830109195e-06,
      "loss": 0.7103,
      "step": 10270
    },
    {
      "epoch": 0.7535790603189743,
      "grad_norm": 0.5378193740583788,
      "learning_rate": 1.509274987274792e-06,
      "loss": 0.4775,
      "step": 10271
    },
    {
      "epoch": 0.7536524299091134,
      "grad_norm": 0.5554894576376271,
      "learning_rate": 1.5084243886685502e-06,
      "loss": 0.4645,
      "step": 10272
    },
    {
      "epoch": 0.7536524299091134,
      "lm_loss": 1.0629,
      "step": 10272,
      "vm_loss": 0.0525
    },
    {
      "epoch": 0.7536524299091134,
      "lm_loss": 0.1198,
      "step": 10272,
      "vm_loss": 0.0917
    },
    {
      "epoch": 0.7536524299091134,
      "lm_loss": 0.6824,
      "step": 10272,
      "vm_loss": 0.0358
    },
    {
      "epoch": 0.7536524299091134,
      "lm_loss": 0.2971,
      "step": 10272,
      "vm_loss": 0.2372
    },
    {
      "epoch": 0.7536524299091134,
      "lm_loss": 0.3344,
      "step": 10272,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.7536524299091134,
      "lm_loss": 0.6937,
      "step": 10272,
      "vm_loss": 0.1899
    },
    {
      "epoch": 0.7536524299091134,
      "lm_loss": 0.6769,
      "step": 10272,
      "vm_loss": 0.1736
    },
    {
      "epoch": 0.7536524299091134,
      "lm_loss": 0.7364,
      "step": 10272,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.7537257994992526,
      "grad_norm": 1.558037208874153,
      "learning_rate": 1.5075739872402284e-06,
      "loss": 0.672,
      "step": 10273
    },
    {
      "epoch": 0.7537991690893917,
      "grad_norm": 0.5310306754853095,
      "learning_rate": 1.506723783037854e-06,
      "loss": 0.4629,
      "step": 10274
    },
    {
      "epoch": 0.7538725386795309,
      "grad_norm": 1.5926525195448422,
      "learning_rate": 1.5058737761094384e-06,
      "loss": 0.8413,
      "step": 10275
    },
    {
      "epoch": 0.7539459082696699,
      "grad_norm": 1.6421326639880536,
      "learning_rate": 1.5050239665029837e-06,
      "loss": 0.7215,
      "step": 10276
    },
    {
      "epoch": 0.754019277859809,
      "grad_norm": 1.525918603613496,
      "learning_rate": 1.504174354266481e-06,
      "loss": 0.8438,
      "step": 10277
    },
    {
      "epoch": 0.7540926474499482,
      "grad_norm": 0.5402987973485113,
      "learning_rate": 1.5033249394479093e-06,
      "loss": 0.4949,
      "step": 10278
    },
    {
      "epoch": 0.7541660170400873,
      "grad_norm": 1.5988308017870476,
      "learning_rate": 1.5024757220952368e-06,
      "loss": 0.8119,
      "step": 10279
    },
    {
      "epoch": 0.7542393866302264,
      "grad_norm": 1.4521595541561523,
      "learning_rate": 1.501626702256423e-06,
      "loss": 0.7584,
      "step": 10280
    },
    {
      "epoch": 0.7542393866302264,
      "lm_loss": 1.0703,
      "step": 10280,
      "vm_loss": 0.1136
    },
    {
      "epoch": 0.7542393866302264,
      "lm_loss": 0.7469,
      "step": 10280,
      "vm_loss": 0.1784
    },
    {
      "epoch": 0.7542393866302264,
      "lm_loss": 0.9335,
      "step": 10280,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.7542393866302264,
      "lm_loss": 0.7376,
      "step": 10280,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.7542393866302264,
      "lm_loss": 0.6951,
      "step": 10280,
      "vm_loss": 0.1041
    },
    {
      "epoch": 0.7542393866302264,
      "lm_loss": 0.0337,
      "step": 10280,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.7542393866302264,
      "lm_loss": 0.6662,
      "step": 10280,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.7542393866302264,
      "lm_loss": 0.6813,
      "step": 10280,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.7543127562203655,
      "grad_norm": 1.4557054486400904,
      "learning_rate": 1.5007778799794126e-06,
      "loss": 0.7183,
      "step": 10281
    },
    {
      "epoch": 0.7543861258105047,
      "grad_norm": 0.551139418823052,
      "learning_rate": 1.4999292553121413e-06,
      "loss": 0.4883,
      "step": 10282
    },
    {
      "epoch": 0.7544594954006438,
      "grad_norm": 1.506941774510868,
      "learning_rate": 1.4990808283025327e-06,
      "loss": 0.7022,
      "step": 10283
    },
    {
      "epoch": 0.754532864990783,
      "grad_norm": 0.5463293554472854,
      "learning_rate": 1.498232598998498e-06,
      "loss": 0.4743,
      "step": 10284
    },
    {
      "epoch": 0.754606234580922,
      "grad_norm": 1.4969638842720816,
      "learning_rate": 1.497384567447942e-06,
      "loss": 0.7782,
      "step": 10285
    },
    {
      "epoch": 0.7546796041710612,
      "grad_norm": 1.5780463829709481,
      "learning_rate": 1.4965367336987536e-06,
      "loss": 0.6411,
      "step": 10286
    },
    {
      "epoch": 0.7547529737612003,
      "grad_norm": 0.5513533081001596,
      "learning_rate": 1.495689097798812e-06,
      "loss": 0.4797,
      "step": 10287
    },
    {
      "epoch": 0.7548263433513395,
      "grad_norm": 1.415157208920038,
      "learning_rate": 1.4948416597959853e-06,
      "loss": 0.7291,
      "step": 10288
    },
    {
      "epoch": 0.7548263433513395,
      "lm_loss": 1.2792,
      "step": 10288,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.7548263433513395,
      "lm_loss": 0.0162,
      "step": 10288,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.7548263433513395,
      "lm_loss": 0.0993,
      "step": 10288,
      "vm_loss": 0.1049
    },
    {
      "epoch": 0.7548263433513395,
      "lm_loss": 0.2658,
      "step": 10288,
      "vm_loss": 0.0978
    },
    {
      "epoch": 0.7548263433513395,
      "lm_loss": 0.0741,
      "step": 10288,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.7548263433513395,
      "lm_loss": 0.9729,
      "step": 10288,
      "vm_loss": 0.0457
    },
    {
      "epoch": 0.7548263433513395,
      "lm_loss": 0.2872,
      "step": 10288,
      "vm_loss": 0.1574
    },
    {
      "epoch": 0.7548263433513395,
      "lm_loss": 0.5878,
      "step": 10288,
      "vm_loss": 0.2009
    },
    {
      "epoch": 0.7548997129414786,
      "grad_norm": 1.4806575063348384,
      "learning_rate": 1.493994419738129e-06,
      "loss": 0.6618,
      "step": 10289
    },
    {
      "epoch": 0.7549730825316177,
      "grad_norm": 1.431869308704483,
      "learning_rate": 1.4931473776730915e-06,
      "loss": 0.7271,
      "step": 10290
    },
    {
      "epoch": 0.7550464521217568,
      "grad_norm": 0.533186984118,
      "learning_rate": 1.4923005336487061e-06,
      "loss": 0.4832,
      "step": 10291
    },
    {
      "epoch": 0.755119821711896,
      "grad_norm": 1.4398670252641825,
      "learning_rate": 1.491453887712797e-06,
      "loss": 0.81,
      "step": 10292
    },
    {
      "epoch": 0.7551931913020351,
      "grad_norm": 1.5426946224325209,
      "learning_rate": 1.4906074399131747e-06,
      "loss": 0.7887,
      "step": 10293
    },
    {
      "epoch": 0.7552665608921743,
      "grad_norm": 0.538038214346399,
      "learning_rate": 1.4897611902976422e-06,
      "loss": 0.4701,
      "step": 10294
    },
    {
      "epoch": 0.7553399304823133,
      "grad_norm": 1.4531653999366354,
      "learning_rate": 1.4889151389139862e-06,
      "loss": 0.6692,
      "step": 10295
    },
    {
      "epoch": 0.7554133000724524,
      "grad_norm": 1.5754410336236484,
      "learning_rate": 1.4880692858099882e-06,
      "loss": 0.7976,
      "step": 10296
    },
    {
      "epoch": 0.7554133000724524,
      "lm_loss": 0.4975,
      "step": 10296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7554133000724524,
      "lm_loss": 0.4095,
      "step": 10296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7554133000724524,
      "lm_loss": 0.4138,
      "step": 10296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7554133000724524,
      "lm_loss": 0.6879,
      "step": 10296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7554133000724524,
      "lm_loss": 0.6813,
      "step": 10296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7554133000724524,
      "lm_loss": 0.4692,
      "step": 10296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7554133000724524,
      "lm_loss": 0.615,
      "step": 10296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7554133000724524,
      "lm_loss": 0.4884,
      "step": 10296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7554866696625916,
      "grad_norm": 0.5489414283189611,
      "learning_rate": 1.4872236310334154e-06,
      "loss": 0.48,
      "step": 10297
    },
    {
      "epoch": 0.7555600392527307,
      "grad_norm": 1.2984044864167472,
      "learning_rate": 1.486378174632024e-06,
      "loss": 0.7042,
      "step": 10298
    },
    {
      "epoch": 0.7556334088428699,
      "grad_norm": 1.3993270804594604,
      "learning_rate": 1.4855329166535577e-06,
      "loss": 0.6963,
      "step": 10299
    },
    {
      "epoch": 0.7557067784330089,
      "grad_norm": 1.5867204909116146,
      "learning_rate": 1.484687857145749e-06,
      "loss": 0.7926,
      "step": 10300
    },
    {
      "epoch": 0.7557801480231481,
      "grad_norm": 0.5453752458780671,
      "learning_rate": 1.483842996156325e-06,
      "loss": 0.4805,
      "step": 10301
    },
    {
      "epoch": 0.7558535176132872,
      "grad_norm": 1.5485412941022405,
      "learning_rate": 1.482998333732994e-06,
      "loss": 0.679,
      "step": 10302
    },
    {
      "epoch": 0.7559268872034264,
      "grad_norm": 1.459808711988636,
      "learning_rate": 1.4821538699234572e-06,
      "loss": 0.6977,
      "step": 10303
    },
    {
      "epoch": 0.7560002567935655,
      "grad_norm": 1.4002664739515138,
      "learning_rate": 1.4813096047754034e-06,
      "loss": 0.6351,
      "step": 10304
    },
    {
      "epoch": 0.7560002567935655,
      "lm_loss": 1.3764,
      "step": 10304,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.7560002567935655,
      "lm_loss": 0.0438,
      "step": 10304,
      "vm_loss": 0.1187
    },
    {
      "epoch": 0.7560002567935655,
      "lm_loss": 0.4582,
      "step": 10304,
      "vm_loss": 0.1833
    },
    {
      "epoch": 0.7560002567935655,
      "lm_loss": 0.437,
      "step": 10304,
      "vm_loss": 0.1663
    },
    {
      "epoch": 0.7560002567935655,
      "lm_loss": 0.4329,
      "step": 10304,
      "vm_loss": 0.1866
    },
    {
      "epoch": 0.7560002567935655,
      "lm_loss": 1.0374,
      "step": 10304,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.7560002567935655,
      "lm_loss": 0.798,
      "step": 10304,
      "vm_loss": 0.1821
    },
    {
      "epoch": 0.7560002567935655,
      "lm_loss": 0.4699,
      "step": 10304,
      "vm_loss": 0.1815
    },
    {
      "epoch": 0.7560736263837046,
      "grad_norm": 1.4259404554660988,
      "learning_rate": 1.4804655383365097e-06,
      "loss": 0.664,
      "step": 10305
    },
    {
      "epoch": 0.7561469959738437,
      "grad_norm": 1.6163348855899988,
      "learning_rate": 1.4796216706544414e-06,
      "loss": 0.8144,
      "step": 10306
    },
    {
      "epoch": 0.7562203655639829,
      "grad_norm": 1.3944178219721963,
      "learning_rate": 1.4787780017768572e-06,
      "loss": 0.6751,
      "step": 10307
    },
    {
      "epoch": 0.756293735154122,
      "grad_norm": 1.492410210437213,
      "learning_rate": 1.4779345317513994e-06,
      "loss": 0.7758,
      "step": 10308
    },
    {
      "epoch": 0.7563671047442612,
      "grad_norm": 1.7294765389805264,
      "learning_rate": 1.4770912606257003e-06,
      "loss": 0.7885,
      "step": 10309
    },
    {
      "epoch": 0.7564404743344002,
      "grad_norm": 1.6693131219901298,
      "learning_rate": 1.4762481884473827e-06,
      "loss": 0.7663,
      "step": 10310
    },
    {
      "epoch": 0.7565138439245394,
      "grad_norm": 1.5142000342330277,
      "learning_rate": 1.4754053152640547e-06,
      "loss": 0.7319,
      "step": 10311
    },
    {
      "epoch": 0.7565872135146785,
      "grad_norm": 1.5300989093654607,
      "learning_rate": 1.4745626411233182e-06,
      "loss": 0.7277,
      "step": 10312
    },
    {
      "epoch": 0.7565872135146785,
      "lm_loss": 1.6238,
      "step": 10312,
      "vm_loss": 0.0478
    },
    {
      "epoch": 0.7565872135146785,
      "lm_loss": 1.5606,
      "step": 10312,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.7565872135146785,
      "lm_loss": 0.0151,
      "step": 10312,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.7565872135146785,
      "lm_loss": 0.1386,
      "step": 10312,
      "vm_loss": 0.1301
    },
    {
      "epoch": 0.7565872135146785,
      "lm_loss": 0.8306,
      "step": 10312,
      "vm_loss": 0.2598
    },
    {
      "epoch": 0.7565872135146785,
      "lm_loss": 0.5438,
      "step": 10312,
      "vm_loss": 0.0988
    },
    {
      "epoch": 0.7565872135146785,
      "lm_loss": 0.8,
      "step": 10312,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.7565872135146785,
      "lm_loss": 0.8545,
      "step": 10312,
      "vm_loss": 0.156
    },
    {
      "epoch": 0.7566605831048177,
      "grad_norm": 1.5179458648923536,
      "learning_rate": 1.4737201660727602e-06,
      "loss": 0.7522,
      "step": 10313
    },
    {
      "epoch": 0.7567339526949568,
      "grad_norm": 1.6906775857088898,
      "learning_rate": 1.4728778901599573e-06,
      "loss": 0.768,
      "step": 10314
    },
    {
      "epoch": 0.7568073222850958,
      "grad_norm": 1.4438177294421002,
      "learning_rate": 1.4720358134324747e-06,
      "loss": 0.6975,
      "step": 10315
    },
    {
      "epoch": 0.756880691875235,
      "grad_norm": 0.5682195122247878,
      "learning_rate": 1.4711939359378646e-06,
      "loss": 0.4958,
      "step": 10316
    },
    {
      "epoch": 0.7569540614653741,
      "grad_norm": 0.5520377017967821,
      "learning_rate": 1.4703522577236739e-06,
      "loss": 0.481,
      "step": 10317
    },
    {
      "epoch": 0.7570274310555133,
      "grad_norm": 1.5049036191403282,
      "learning_rate": 1.4695107788374318e-06,
      "loss": 0.756,
      "step": 10318
    },
    {
      "epoch": 0.7571008006456524,
      "grad_norm": 1.5365327364413999,
      "learning_rate": 1.4686694993266598e-06,
      "loss": 0.8023,
      "step": 10319
    },
    {
      "epoch": 0.7571741702357915,
      "grad_norm": 0.5527101330917352,
      "learning_rate": 1.4678284192388658e-06,
      "loss": 0.4869,
      "step": 10320
    },
    {
      "epoch": 0.7571741702357915,
      "lm_loss": 1.5885,
      "step": 10320,
      "vm_loss": 0.0458
    },
    {
      "epoch": 0.7571741702357915,
      "lm_loss": 0.9429,
      "step": 10320,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.7571741702357915,
      "lm_loss": 0.8171,
      "step": 10320,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.7571741702357915,
      "lm_loss": 1.0603,
      "step": 10320,
      "vm_loss": 0.1588
    },
    {
      "epoch": 0.7571741702357915,
      "lm_loss": 0.0424,
      "step": 10320,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.7571741702357915,
      "lm_loss": 0.0301,
      "step": 10320,
      "vm_loss": 0.035
    },
    {
      "epoch": 0.7571741702357915,
      "lm_loss": 0.5039,
      "step": 10320,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.7571741702357915,
      "lm_loss": 0.437,
      "step": 10320,
      "vm_loss": 0.1131
    },
    {
      "epoch": 0.7572475398259306,
      "grad_norm": 1.5983842561196906,
      "learning_rate": 1.4669875386215481e-06,
      "loss": 0.7344,
      "step": 10321
    },
    {
      "epoch": 0.7573209094160698,
      "grad_norm": 0.5443359263588726,
      "learning_rate": 1.466146857522192e-06,
      "loss": 0.4732,
      "step": 10322
    },
    {
      "epoch": 0.7573942790062089,
      "grad_norm": 1.4409780143942585,
      "learning_rate": 1.4653063759882762e-06,
      "loss": 0.7328,
      "step": 10323
    },
    {
      "epoch": 0.7574676485963481,
      "grad_norm": 1.4261346211017416,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.7351,
      "step": 10324
    },
    {
      "epoch": 0.7575410181864871,
      "grad_norm": 1.5722332529528376,
      "learning_rate": 1.4636260118066048e-06,
      "loss": 0.7499,
      "step": 10325
    },
    {
      "epoch": 0.7576143877766263,
      "grad_norm": 1.597917203879488,
      "learning_rate": 1.462786129253743e-06,
      "loss": 0.7552,
      "step": 10326
    },
    {
      "epoch": 0.7576877573667654,
      "grad_norm": 1.5257089594410347,
      "learning_rate": 1.4619464464561072e-06,
      "loss": 0.728,
      "step": 10327
    },
    {
      "epoch": 0.7577611269569046,
      "grad_norm": 1.5204578991462157,
      "learning_rate": 1.4611069634611186e-06,
      "loss": 0.7571,
      "step": 10328
    },
    {
      "epoch": 0.7577611269569046,
      "lm_loss": 0.231,
      "step": 10328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7577611269569046,
      "lm_loss": 0.5004,
      "step": 10328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7577611269569046,
      "lm_loss": 0.4344,
      "step": 10328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7577611269569046,
      "lm_loss": 0.3845,
      "step": 10328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7577611269569046,
      "lm_loss": 0.5269,
      "step": 10328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7577611269569046,
      "lm_loss": 0.5039,
      "step": 10328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7577611269569046,
      "lm_loss": 0.4804,
      "step": 10328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7577611269569046,
      "lm_loss": 0.4886,
      "step": 10328,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7578344965470437,
      "grad_norm": 0.5350430695045129,
      "learning_rate": 1.4602676803161842e-06,
      "loss": 0.4814,
      "step": 10329
    },
    {
      "epoch": 0.7579078661371828,
      "grad_norm": 1.5129268958594575,
      "learning_rate": 1.4594285970686995e-06,
      "loss": 0.7404,
      "step": 10330
    },
    {
      "epoch": 0.7579812357273219,
      "grad_norm": 1.382377796537705,
      "learning_rate": 1.4585897137660498e-06,
      "loss": 0.701,
      "step": 10331
    },
    {
      "epoch": 0.758054605317461,
      "grad_norm": 1.4903688070269743,
      "learning_rate": 1.4577510304556098e-06,
      "loss": 0.6631,
      "step": 10332
    },
    {
      "epoch": 0.7581279749076002,
      "grad_norm": 1.4419532485637183,
      "learning_rate": 1.4569125471847394e-06,
      "loss": 0.6773,
      "step": 10333
    },
    {
      "epoch": 0.7582013444977393,
      "grad_norm": 1.4643427572146173,
      "learning_rate": 1.456074264000793e-06,
      "loss": 0.6831,
      "step": 10334
    },
    {
      "epoch": 0.7582747140878784,
      "grad_norm": 1.6153475335778722,
      "learning_rate": 1.45523618095111e-06,
      "loss": 0.7361,
      "step": 10335
    },
    {
      "epoch": 0.7583480836780175,
      "grad_norm": 1.7023352177674074,
      "learning_rate": 1.4543982980830179e-06,
      "loss": 0.81,
      "step": 10336
    },
    {
      "epoch": 0.7583480836780175,
      "lm_loss": 0.3598,
      "step": 10336,
      "vm_loss": 0.0379
    },
    {
      "epoch": 0.7583480836780175,
      "lm_loss": 2.1,
      "step": 10336,
      "vm_loss": 0.1073
    },
    {
      "epoch": 0.7583480836780175,
      "lm_loss": 0.8563,
      "step": 10336,
      "vm_loss": 0.2001
    },
    {
      "epoch": 0.7583480836780175,
      "lm_loss": 0.2719,
      "step": 10336,
      "vm_loss": 0.1608
    },
    {
      "epoch": 0.7583480836780175,
      "lm_loss": 0.047,
      "step": 10336,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.7583480836780175,
      "lm_loss": 0.4798,
      "step": 10336,
      "vm_loss": 0.1377
    },
    {
      "epoch": 0.7583480836780175,
      "lm_loss": 1.1771,
      "step": 10336,
      "vm_loss": 0.1464
    },
    {
      "epoch": 0.7583480836780175,
      "lm_loss": 0.891,
      "step": 10336,
      "vm_loss": 0.1101
    },
    {
      "epoch": 0.7584214532681567,
      "grad_norm": 1.7205016274457932,
      "learning_rate": 1.4535606154438342e-06,
      "loss": 0.8353,
      "step": 10337
    },
    {
      "epoch": 0.7584948228582958,
      "grad_norm": 0.5667299883859552,
      "learning_rate": 1.452723133080864e-06,
      "loss": 0.4686,
      "step": 10338
    },
    {
      "epoch": 0.758568192448435,
      "grad_norm": 1.515783453477156,
      "learning_rate": 1.4518858510414047e-06,
      "loss": 0.7043,
      "step": 10339
    },
    {
      "epoch": 0.758641562038574,
      "grad_norm": 1.5402765035039598,
      "learning_rate": 1.4510487693727383e-06,
      "loss": 0.7464,
      "step": 10340
    },
    {
      "epoch": 0.7587149316287132,
      "grad_norm": 1.4313108179854288,
      "learning_rate": 1.450211888122136e-06,
      "loss": 0.7105,
      "step": 10341
    },
    {
      "epoch": 0.7587883012188523,
      "grad_norm": 1.4926727623674323,
      "learning_rate": 1.44937520733686e-06,
      "loss": 0.8256,
      "step": 10342
    },
    {
      "epoch": 0.7588616708089915,
      "grad_norm": 0.5420600467481841,
      "learning_rate": 1.4485387270641572e-06,
      "loss": 0.4958,
      "step": 10343
    },
    {
      "epoch": 0.7589350403991306,
      "grad_norm": 1.5831648655637562,
      "learning_rate": 1.447702447351269e-06,
      "loss": 0.7862,
      "step": 10344
    },
    {
      "epoch": 0.7589350403991306,
      "lm_loss": 0.317,
      "step": 10344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7589350403991306,
      "lm_loss": 0.4475,
      "step": 10344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7589350403991306,
      "lm_loss": 0.5157,
      "step": 10344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7589350403991306,
      "lm_loss": 0.482,
      "step": 10344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7589350403991306,
      "lm_loss": 0.3874,
      "step": 10344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7589350403991306,
      "lm_loss": 0.5411,
      "step": 10344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7589350403991306,
      "lm_loss": 0.4772,
      "step": 10344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7589350403991306,
      "lm_loss": 0.53,
      "step": 10344,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7590084099892697,
      "grad_norm": 0.5522410729933438,
      "learning_rate": 1.4468663682454209e-06,
      "loss": 0.4774,
      "step": 10345
    },
    {
      "epoch": 0.7590817795794088,
      "grad_norm": 1.4783953359901405,
      "learning_rate": 1.4460304897938282e-06,
      "loss": 0.7197,
      "step": 10346
    },
    {
      "epoch": 0.759155149169548,
      "grad_norm": 1.3141999533043323,
      "learning_rate": 1.4451948120436948e-06,
      "loss": 0.7325,
      "step": 10347
    },
    {
      "epoch": 0.7592285187596871,
      "grad_norm": 1.5283114733273218,
      "learning_rate": 1.4443593350422137e-06,
      "loss": 0.7143,
      "step": 10348
    },
    {
      "epoch": 0.7593018883498263,
      "grad_norm": 1.553717366355748,
      "learning_rate": 1.4435240588365645e-06,
      "loss": 0.6671,
      "step": 10349
    },
    {
      "epoch": 0.7593752579399653,
      "grad_norm": 1.3261315753170235,
      "learning_rate": 1.442688983473921e-06,
      "loss": 0.5629,
      "step": 10350
    },
    {
      "epoch": 0.7594486275301044,
      "grad_norm": 1.4575508764777418,
      "learning_rate": 1.4418541090014404e-06,
      "loss": 0.7402,
      "step": 10351
    },
    {
      "epoch": 0.7595219971202436,
      "grad_norm": 1.450076077140106,
      "learning_rate": 1.44101943546627e-06,
      "loss": 0.7924,
      "step": 10352
    },
    {
      "epoch": 0.7595219971202436,
      "lm_loss": 0.7592,
      "step": 10352,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.7595219971202436,
      "lm_loss": 0.8793,
      "step": 10352,
      "vm_loss": 0.073
    },
    {
      "epoch": 0.7595219971202436,
      "lm_loss": 1.1827,
      "step": 10352,
      "vm_loss": 0.1681
    },
    {
      "epoch": 0.7595219971202436,
      "lm_loss": 0.3773,
      "step": 10352,
      "vm_loss": 0.2284
    },
    {
      "epoch": 0.7595219971202436,
      "lm_loss": 1.0623,
      "step": 10352,
      "vm_loss": 0.1553
    },
    {
      "epoch": 0.7595219971202436,
      "lm_loss": 0.1194,
      "step": 10352,
      "vm_loss": 0.1945
    },
    {
      "epoch": 0.7595219971202436,
      "lm_loss": 0.4758,
      "step": 10352,
      "vm_loss": 0.1027
    },
    {
      "epoch": 0.7595219971202436,
      "lm_loss": 0.11,
      "step": 10352,
      "vm_loss": 0.1413
    },
    {
      "epoch": 0.7595953667103827,
      "grad_norm": 1.5328289507012485,
      "learning_rate": 1.440184962915545e-06,
      "loss": 0.6851,
      "step": 10353
    },
    {
      "epoch": 0.7596687363005219,
      "grad_norm": 1.5659026105248757,
      "learning_rate": 1.4393506913963901e-06,
      "loss": 0.7921,
      "step": 10354
    },
    {
      "epoch": 0.7597421058906609,
      "grad_norm": 0.5539869643306301,
      "learning_rate": 1.4385166209559203e-06,
      "loss": 0.4889,
      "step": 10355
    },
    {
      "epoch": 0.7598154754808001,
      "grad_norm": 1.7134104974207163,
      "learning_rate": 1.4376827516412372e-06,
      "loss": 0.7675,
      "step": 10356
    },
    {
      "epoch": 0.7598888450709392,
      "grad_norm": 1.4354438570253742,
      "learning_rate": 1.436849083499431e-06,
      "loss": 0.7503,
      "step": 10357
    },
    {
      "epoch": 0.7599622146610784,
      "grad_norm": 1.5452568449781825,
      "learning_rate": 1.4360156165775812e-06,
      "loss": 0.8207,
      "step": 10358
    },
    {
      "epoch": 0.7600355842512174,
      "grad_norm": 0.5723966384077593,
      "learning_rate": 1.4351823509227553e-06,
      "loss": 0.493,
      "step": 10359
    },
    {
      "epoch": 0.7601089538413566,
      "grad_norm": 1.4454813199073548,
      "learning_rate": 1.4343492865820087e-06,
      "loss": 0.662,
      "step": 10360
    },
    {
      "epoch": 0.7601089538413566,
      "lm_loss": 0.0772,
      "step": 10360,
      "vm_loss": 0.0349
    },
    {
      "epoch": 0.7601089538413566,
      "lm_loss": 0.3397,
      "step": 10360,
      "vm_loss": 0.0935
    },
    {
      "epoch": 0.7601089538413566,
      "lm_loss": 0.3192,
      "step": 10360,
      "vm_loss": 0.1774
    },
    {
      "epoch": 0.7601089538413566,
      "lm_loss": 0.8519,
      "step": 10360,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.7601089538413566,
      "lm_loss": 0.6207,
      "step": 10360,
      "vm_loss": 0.114
    },
    {
      "epoch": 0.7601089538413566,
      "lm_loss": 0.4161,
      "step": 10360,
      "vm_loss": 0.0529
    },
    {
      "epoch": 0.7601089538413566,
      "lm_loss": 0.0887,
      "step": 10360,
      "vm_loss": 0.0568
    },
    {
      "epoch": 0.7601089538413566,
      "lm_loss": 0.7323,
      "step": 10360,
      "vm_loss": 0.153
    },
    {
      "epoch": 0.7601823234314957,
      "grad_norm": 1.6402030176907167,
      "learning_rate": 1.4335164236023896e-06,
      "loss": 0.7644,
      "step": 10361
    },
    {
      "epoch": 0.7602556930216349,
      "grad_norm": 1.4988034564543629,
      "learning_rate": 1.4326837620309303e-06,
      "loss": 0.7592,
      "step": 10362
    },
    {
      "epoch": 0.760329062611774,
      "grad_norm": 1.4322193026357792,
      "learning_rate": 1.4318513019146536e-06,
      "loss": 0.6422,
      "step": 10363
    },
    {
      "epoch": 0.7604024322019131,
      "grad_norm": 1.4308245473536942,
      "learning_rate": 1.43101904330057e-06,
      "loss": 0.7492,
      "step": 10364
    },
    {
      "epoch": 0.7604758017920522,
      "grad_norm": 1.5216291298510303,
      "learning_rate": 1.4301869862356777e-06,
      "loss": 0.6931,
      "step": 10365
    },
    {
      "epoch": 0.7605491713821914,
      "grad_norm": 1.4690727045201186,
      "learning_rate": 1.4293551307669683e-06,
      "loss": 0.678,
      "step": 10366
    },
    {
      "epoch": 0.7606225409723305,
      "grad_norm": 1.408877671034974,
      "learning_rate": 1.4285234769414174e-06,
      "loss": 0.7178,
      "step": 10367
    },
    {
      "epoch": 0.7606959105624697,
      "grad_norm": 1.6541894607620145,
      "learning_rate": 1.4276920248059906e-06,
      "loss": 0.7391,
      "step": 10368
    },
    {
      "epoch": 0.7606959105624697,
      "lm_loss": 0.4586,
      "step": 10368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7606959105624697,
      "lm_loss": 0.5683,
      "step": 10368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7606959105624697,
      "lm_loss": 0.4038,
      "step": 10368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7606959105624697,
      "lm_loss": 0.3847,
      "step": 10368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7606959105624697,
      "lm_loss": 0.5177,
      "step": 10368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7606959105624697,
      "lm_loss": 0.545,
      "step": 10368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7606959105624697,
      "lm_loss": 0.4798,
      "step": 10368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7606959105624697,
      "lm_loss": 0.5296,
      "step": 10368,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7607692801526087,
      "grad_norm": 0.5528064445977726,
      "learning_rate": 1.4268607744076419e-06,
      "loss": 0.4868,
      "step": 10369
    },
    {
      "epoch": 0.7608426497427478,
      "grad_norm": 1.7483294606805928,
      "learning_rate": 1.426029725793312e-06,
      "loss": 0.6592,
      "step": 10370
    },
    {
      "epoch": 0.760916019332887,
      "grad_norm": 1.568966677360102,
      "learning_rate": 1.425198879009936e-06,
      "loss": 0.7924,
      "step": 10371
    },
    {
      "epoch": 0.7609893889230261,
      "grad_norm": 1.3703562653398356,
      "learning_rate": 1.4243682341044324e-06,
      "loss": 0.7447,
      "step": 10372
    },
    {
      "epoch": 0.7610627585131653,
      "grad_norm": 1.4350599426354373,
      "learning_rate": 1.4235377911237091e-06,
      "loss": 0.6891,
      "step": 10373
    },
    {
      "epoch": 0.7611361281033043,
      "grad_norm": 0.5411326213289733,
      "learning_rate": 1.4227075501146637e-06,
      "loss": 0.4686,
      "step": 10374
    },
    {
      "epoch": 0.7612094976934435,
      "grad_norm": 1.5573808206839186,
      "learning_rate": 1.4218775111241822e-06,
      "loss": 0.7466,
      "step": 10375
    },
    {
      "epoch": 0.7612828672835826,
      "grad_norm": 1.4815345703206313,
      "learning_rate": 1.421047674199137e-06,
      "loss": 0.7758,
      "step": 10376
    },
    {
      "epoch": 0.7612828672835826,
      "lm_loss": 0.2813,
      "step": 10376,
      "vm_loss": 0.0809
    },
    {
      "epoch": 0.7612828672835826,
      "lm_loss": 0.7955,
      "step": 10376,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.7612828672835826,
      "lm_loss": 0.2542,
      "step": 10376,
      "vm_loss": 0.1265
    },
    {
      "epoch": 0.7612828672835826,
      "lm_loss": 0.3508,
      "step": 10376,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.7612828672835826,
      "lm_loss": 0.131,
      "step": 10376,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.7612828672835826,
      "lm_loss": 0.2913,
      "step": 10376,
      "vm_loss": 0.1271
    },
    {
      "epoch": 0.7612828672835826,
      "lm_loss": 0.8828,
      "step": 10376,
      "vm_loss": 0.1402
    },
    {
      "epoch": 0.7612828672835826,
      "lm_loss": 0.9756,
      "step": 10376,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.7613562368737218,
      "grad_norm": 1.4448359672670765,
      "learning_rate": 1.4202180393863945e-06,
      "loss": 0.6932,
      "step": 10377
    },
    {
      "epoch": 0.7614296064638609,
      "grad_norm": 1.6388127593790667,
      "learning_rate": 1.4193886067328045e-06,
      "loss": 0.6603,
      "step": 10378
    },
    {
      "epoch": 0.761502976054,
      "grad_norm": 1.687951556171239,
      "learning_rate": 1.4185593762852069e-06,
      "loss": 0.756,
      "step": 10379
    },
    {
      "epoch": 0.7615763456441391,
      "grad_norm": 1.496024303913451,
      "learning_rate": 1.417730348090431e-06,
      "loss": 0.7812,
      "step": 10380
    },
    {
      "epoch": 0.7616497152342783,
      "grad_norm": 0.5471202482782663,
      "learning_rate": 1.416901522195292e-06,
      "loss": 0.4769,
      "step": 10381
    },
    {
      "epoch": 0.7617230848244174,
      "grad_norm": 0.5486546496321311,
      "learning_rate": 1.4160728986465987e-06,
      "loss": 0.4693,
      "step": 10382
    },
    {
      "epoch": 0.7617964544145566,
      "grad_norm": 1.5047121328169684,
      "learning_rate": 1.4152444774911445e-06,
      "loss": 0.8183,
      "step": 10383
    },
    {
      "epoch": 0.7618698240046956,
      "grad_norm": 1.6220827245347191,
      "learning_rate": 1.414416258775712e-06,
      "loss": 0.7923,
      "step": 10384
    },
    {
      "epoch": 0.7618698240046956,
      "lm_loss": 1.1195,
      "step": 10384,
      "vm_loss": 0.0439
    },
    {
      "epoch": 0.7618698240046956,
      "lm_loss": 0.751,
      "step": 10384,
      "vm_loss": 0.1151
    },
    {
      "epoch": 0.7618698240046956,
      "lm_loss": 0.5149,
      "step": 10384,
      "vm_loss": 0.0944
    },
    {
      "epoch": 0.7618698240046956,
      "lm_loss": 0.9004,
      "step": 10384,
      "vm_loss": 0.0755
    },
    {
      "epoch": 0.7618698240046956,
      "lm_loss": 0.1413,
      "step": 10384,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.7618698240046956,
      "lm_loss": 0.0798,
      "step": 10384,
      "vm_loss": 0.2027
    },
    {
      "epoch": 0.7618698240046956,
      "lm_loss": 0.2985,
      "step": 10384,
      "vm_loss": 0.0976
    },
    {
      "epoch": 0.7618698240046956,
      "lm_loss": 0.7589,
      "step": 10384,
      "vm_loss": 0.1891
    },
    {
      "epoch": 0.7619431935948348,
      "grad_norm": 1.6878388810894684,
      "learning_rate": 1.4135882425470726e-06,
      "loss": 0.6868,
      "step": 10385
    },
    {
      "epoch": 0.7620165631849739,
      "grad_norm": 0.5453322125463609,
      "learning_rate": 1.4127604288519875e-06,
      "loss": 0.4976,
      "step": 10386
    },
    {
      "epoch": 0.762089932775113,
      "grad_norm": 0.5380896800269673,
      "learning_rate": 1.4119328177372028e-06,
      "loss": 0.4755,
      "step": 10387
    },
    {
      "epoch": 0.7621633023652522,
      "grad_norm": 1.5364093604050355,
      "learning_rate": 1.4111054092494586e-06,
      "loss": 0.7484,
      "step": 10388
    },
    {
      "epoch": 0.7622366719553912,
      "grad_norm": 0.5461194041955187,
      "learning_rate": 1.41027820343548e-06,
      "loss": 0.4875,
      "step": 10389
    },
    {
      "epoch": 0.7623100415455304,
      "grad_norm": 0.5356698607942418,
      "learning_rate": 1.4094512003419813e-06,
      "loss": 0.4807,
      "step": 10390
    },
    {
      "epoch": 0.7623834111356695,
      "grad_norm": 0.5467044228418503,
      "learning_rate": 1.4086244000156652e-06,
      "loss": 0.4899,
      "step": 10391
    },
    {
      "epoch": 0.7624567807258087,
      "grad_norm": 1.509093898917834,
      "learning_rate": 1.4077978025032214e-06,
      "loss": 0.7466,
      "step": 10392
    },
    {
      "epoch": 0.7624567807258087,
      "lm_loss": 1.3624,
      "step": 10392,
      "vm_loss": 0.0337
    },
    {
      "epoch": 0.7624567807258087,
      "lm_loss": 0.0057,
      "step": 10392,
      "vm_loss": 0.0431
    },
    {
      "epoch": 0.7624567807258087,
      "lm_loss": 0.191,
      "step": 10392,
      "vm_loss": 0.0588
    },
    {
      "epoch": 0.7624567807258087,
      "lm_loss": 1.0794,
      "step": 10392,
      "vm_loss": 0.1495
    },
    {
      "epoch": 0.7624567807258087,
      "lm_loss": 0.5009,
      "step": 10392,
      "vm_loss": 0.2035
    },
    {
      "epoch": 0.7624567807258087,
      "lm_loss": 0.8578,
      "step": 10392,
      "vm_loss": 0.1437
    },
    {
      "epoch": 0.7624567807258087,
      "lm_loss": 0.8892,
      "step": 10392,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.7624567807258087,
      "lm_loss": 0.0841,
      "step": 10392,
      "vm_loss": 0.139
    },
    {
      "epoch": 0.7625301503159478,
      "grad_norm": 1.3712322486671733,
      "learning_rate": 1.406971407851333e-06,
      "loss": 0.7763,
      "step": 10393
    },
    {
      "epoch": 0.7626035199060869,
      "grad_norm": 1.53702710036632,
      "learning_rate": 1.4061452161066674e-06,
      "loss": 0.7162,
      "step": 10394
    },
    {
      "epoch": 0.762676889496226,
      "grad_norm": 1.5725041974338383,
      "learning_rate": 1.4053192273158818e-06,
      "loss": 0.7371,
      "step": 10395
    },
    {
      "epoch": 0.7627502590863652,
      "grad_norm": 0.5662991750850672,
      "learning_rate": 1.404493441525621e-06,
      "loss": 0.4909,
      "step": 10396
    },
    {
      "epoch": 0.7628236286765043,
      "grad_norm": 1.433867251094015,
      "learning_rate": 1.4036678587825181e-06,
      "loss": 0.676,
      "step": 10397
    },
    {
      "epoch": 0.7628969982666435,
      "grad_norm": 1.4058099796697636,
      "learning_rate": 1.402842479133199e-06,
      "loss": 0.7773,
      "step": 10398
    },
    {
      "epoch": 0.7629703678567825,
      "grad_norm": 0.5521313349965445,
      "learning_rate": 1.402017302624274e-06,
      "loss": 0.487,
      "step": 10399
    },
    {
      "epoch": 0.7630437374469217,
      "grad_norm": 1.6650747697745714,
      "learning_rate": 1.4011923293023427e-06,
      "loss": 0.7535,
      "step": 10400
    },
    {
      "epoch": 0.7630437374469217,
      "lm_loss": 0.8612,
      "step": 10400,
      "vm_loss": 0.0495
    },
    {
      "epoch": 0.7630437374469217,
      "lm_loss": 0.5506,
      "step": 10400,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.7630437374469217,
      "lm_loss": 0.8284,
      "step": 10400,
      "vm_loss": 0.1701
    },
    {
      "epoch": 0.7630437374469217,
      "lm_loss": 1.006,
      "step": 10400,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.7630437374469217,
      "lm_loss": 0.0628,
      "step": 10400,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.7630437374469217,
      "lm_loss": 0.8046,
      "step": 10400,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.7630437374469217,
      "lm_loss": 0.3626,
      "step": 10400,
      "vm_loss": 0.251
    },
    {
      "epoch": 0.7630437374469217,
      "lm_loss": 0.1754,
      "step": 10400,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.7631171070370608,
      "grad_norm": 1.6265646124030375,
      "learning_rate": 1.400367559213992e-06,
      "loss": 0.7451,
      "step": 10401
    },
    {
      "epoch": 0.7631904766272,
      "grad_norm": 1.5858542990172653,
      "learning_rate": 1.399542992405798e-06,
      "loss": 0.631,
      "step": 10402
    },
    {
      "epoch": 0.7632638462173391,
      "grad_norm": 1.4745458033083212,
      "learning_rate": 1.3987186289243288e-06,
      "loss": 0.6649,
      "step": 10403
    },
    {
      "epoch": 0.7633372158074782,
      "grad_norm": 1.5659775755345764,
      "learning_rate": 1.3978944688161372e-06,
      "loss": 0.6912,
      "step": 10404
    },
    {
      "epoch": 0.7634105853976173,
      "grad_norm": 1.672588882287244,
      "learning_rate": 1.397070512127765e-06,
      "loss": 0.6875,
      "step": 10405
    },
    {
      "epoch": 0.7634839549877565,
      "grad_norm": 1.5373928559914465,
      "learning_rate": 1.3962467589057438e-06,
      "loss": 0.746,
      "step": 10406
    },
    {
      "epoch": 0.7635573245778956,
      "grad_norm": 1.4513571884399121,
      "learning_rate": 1.3954232091965913e-06,
      "loss": 0.684,
      "step": 10407
    },
    {
      "epoch": 0.7636306941680348,
      "grad_norm": 1.5470046045090164,
      "learning_rate": 1.3945998630468183e-06,
      "loss": 0.7171,
      "step": 10408
    },
    {
      "epoch": 0.7636306941680348,
      "lm_loss": 1.1126,
      "step": 10408,
      "vm_loss": 0.046
    },
    {
      "epoch": 0.7636306941680348,
      "lm_loss": 1.0624,
      "step": 10408,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.7636306941680348,
      "lm_loss": 0.3837,
      "step": 10408,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.7636306941680348,
      "lm_loss": 0.3249,
      "step": 10408,
      "vm_loss": 0.1384
    },
    {
      "epoch": 0.7636306941680348,
      "lm_loss": 0.9314,
      "step": 10408,
      "vm_loss": 0.09
    },
    {
      "epoch": 0.7636306941680348,
      "lm_loss": 0.3806,
      "step": 10408,
      "vm_loss": 0.0864
    },
    {
      "epoch": 0.7636306941680348,
      "lm_loss": 0.7928,
      "step": 10408,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.7636306941680348,
      "lm_loss": 0.5101,
      "step": 10408,
      "vm_loss": 0.0609
    },
    {
      "epoch": 0.7637040637581738,
      "grad_norm": 1.3562014069222057,
      "learning_rate": 1.3937767205029196e-06,
      "loss": 0.7028,
      "step": 10409
    },
    {
      "epoch": 0.7637774333483129,
      "grad_norm": 1.5706958917619105,
      "learning_rate": 1.3929537816113803e-06,
      "loss": 0.7759,
      "step": 10410
    },
    {
      "epoch": 0.7638508029384521,
      "grad_norm": 1.617379355464857,
      "learning_rate": 1.3921310464186733e-06,
      "loss": 0.7386,
      "step": 10411
    },
    {
      "epoch": 0.7639241725285912,
      "grad_norm": 1.4729156333802333,
      "learning_rate": 1.3913085149712595e-06,
      "loss": 0.7438,
      "step": 10412
    },
    {
      "epoch": 0.7639975421187304,
      "grad_norm": 0.5564645019108142,
      "learning_rate": 1.3904861873155922e-06,
      "loss": 0.4553,
      "step": 10413
    },
    {
      "epoch": 0.7640709117088694,
      "grad_norm": 1.533407116368548,
      "learning_rate": 1.3896640634981085e-06,
      "loss": 0.7862,
      "step": 10414
    },
    {
      "epoch": 0.7641442812990086,
      "grad_norm": 1.4289218606299563,
      "learning_rate": 1.3888421435652361e-06,
      "loss": 0.7536,
      "step": 10415
    },
    {
      "epoch": 0.7642176508891477,
      "grad_norm": 1.3987839083472655,
      "learning_rate": 1.3880204275633908e-06,
      "loss": 0.6361,
      "step": 10416
    },
    {
      "epoch": 0.7642176508891477,
      "lm_loss": 0.3266,
      "step": 10416,
      "vm_loss": 0.0764
    },
    {
      "epoch": 0.7642176508891477,
      "lm_loss": 0.1012,
      "step": 10416,
      "vm_loss": 0.2421
    },
    {
      "epoch": 0.7642176508891477,
      "lm_loss": 0.9538,
      "step": 10416,
      "vm_loss": 0.1172
    },
    {
      "epoch": 0.7642176508891477,
      "lm_loss": 0.0579,
      "step": 10416,
      "vm_loss": 0.0918
    },
    {
      "epoch": 0.7642176508891477,
      "lm_loss": 0.0979,
      "step": 10416,
      "vm_loss": 0.0708
    },
    {
      "epoch": 0.7642176508891477,
      "lm_loss": 0.4153,
      "step": 10416,
      "vm_loss": 0.1585
    },
    {
      "epoch": 0.7642176508891477,
      "lm_loss": 0.9199,
      "step": 10416,
      "vm_loss": 0.2034
    },
    {
      "epoch": 0.7642176508891477,
      "lm_loss": 0.1846,
      "step": 10416,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.7642910204792869,
      "grad_norm": 1.4591117721618423,
      "learning_rate": 1.387198915538977e-06,
      "loss": 0.7033,
      "step": 10417
    },
    {
      "epoch": 0.764364390069426,
      "grad_norm": 0.5395829892574989,
      "learning_rate": 1.3863776075383855e-06,
      "loss": 0.4642,
      "step": 10418
    },
    {
      "epoch": 0.7644377596595651,
      "grad_norm": 1.5594245203196506,
      "learning_rate": 1.3855565036080015e-06,
      "loss": 0.7656,
      "step": 10419
    },
    {
      "epoch": 0.7645111292497042,
      "grad_norm": 1.452920073088418,
      "learning_rate": 1.3847356037941922e-06,
      "loss": 0.8198,
      "step": 10420
    },
    {
      "epoch": 0.7645844988398434,
      "grad_norm": 1.4003284753647958,
      "learning_rate": 1.3839149081433168e-06,
      "loss": 0.7536,
      "step": 10421
    },
    {
      "epoch": 0.7646578684299825,
      "grad_norm": 1.6063730244585903,
      "learning_rate": 1.3830944167017213e-06,
      "loss": 0.6751,
      "step": 10422
    },
    {
      "epoch": 0.7647312380201217,
      "grad_norm": 1.475747905862235,
      "learning_rate": 1.3822741295157399e-06,
      "loss": 0.8019,
      "step": 10423
    },
    {
      "epoch": 0.7648046076102607,
      "grad_norm": 0.5701069214659651,
      "learning_rate": 1.3814540466316994e-06,
      "loss": 0.4781,
      "step": 10424
    },
    {
      "epoch": 0.7648046076102607,
      "lm_loss": 1.3192,
      "step": 10424,
      "vm_loss": 0.065
    },
    {
      "epoch": 0.7648046076102607,
      "lm_loss": 1.2958,
      "step": 10424,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.7648046076102607,
      "lm_loss": 0.006,
      "step": 10424,
      "vm_loss": 0.0895
    },
    {
      "epoch": 0.7648046076102607,
      "lm_loss": 1.0368,
      "step": 10424,
      "vm_loss": 0.119
    },
    {
      "epoch": 0.7648046076102607,
      "lm_loss": 1.1148,
      "step": 10424,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.7648046076102607,
      "lm_loss": 0.6996,
      "step": 10424,
      "vm_loss": 0.0671
    },
    {
      "epoch": 0.7648046076102607,
      "lm_loss": 0.8303,
      "step": 10424,
      "vm_loss": 0.0893
    },
    {
      "epoch": 0.7648046076102607,
      "lm_loss": 0.3251,
      "step": 10424,
      "vm_loss": 0.0599
    },
    {
      "epoch": 0.7648779772003999,
      "grad_norm": 1.2565692164395015,
      "learning_rate": 1.3806341680959096e-06,
      "loss": 0.663,
      "step": 10425
    },
    {
      "epoch": 0.764951346790539,
      "grad_norm": 1.5947724727202643,
      "learning_rate": 1.3798144939546714e-06,
      "loss": 0.796,
      "step": 10426
    },
    {
      "epoch": 0.7650247163806781,
      "grad_norm": 1.4638104389012343,
      "learning_rate": 1.3789950242542744e-06,
      "loss": 0.6743,
      "step": 10427
    },
    {
      "epoch": 0.7650980859708173,
      "grad_norm": 1.4729015979023925,
      "learning_rate": 1.3781757590409956e-06,
      "loss": 0.7943,
      "step": 10428
    },
    {
      "epoch": 0.7651714555609563,
      "grad_norm": 1.3674673511377606,
      "learning_rate": 1.3773566983610992e-06,
      "loss": 0.7542,
      "step": 10429
    },
    {
      "epoch": 0.7652448251510955,
      "grad_norm": 1.4572854559092494,
      "learning_rate": 1.3765378422608433e-06,
      "loss": 0.6772,
      "step": 10430
    },
    {
      "epoch": 0.7653181947412346,
      "grad_norm": 0.5572050959998482,
      "learning_rate": 1.3757191907864686e-06,
      "loss": 0.4838,
      "step": 10431
    },
    {
      "epoch": 0.7653915643313738,
      "grad_norm": 1.4260551432114084,
      "learning_rate": 1.3749007439842065e-06,
      "loss": 0.7537,
      "step": 10432
    },
    {
      "epoch": 0.7653915643313738,
      "lm_loss": 1.2651,
      "step": 10432,
      "vm_loss": 0.0838
    },
    {
      "epoch": 0.7653915643313738,
      "lm_loss": 0.1312,
      "step": 10432,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.7653915643313738,
      "lm_loss": 1.232,
      "step": 10432,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.7653915643313738,
      "lm_loss": 0.4121,
      "step": 10432,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.7653915643313738,
      "lm_loss": 0.4516,
      "step": 10432,
      "vm_loss": 0.1845
    },
    {
      "epoch": 0.7653915643313738,
      "lm_loss": 0.0399,
      "step": 10432,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.7653915643313738,
      "lm_loss": 0.4437,
      "step": 10432,
      "vm_loss": 0.2039
    },
    {
      "epoch": 0.7653915643313738,
      "lm_loss": 0.0221,
      "step": 10432,
      "vm_loss": 0.1468
    },
    {
      "epoch": 0.7654649339215129,
      "grad_norm": 1.5619977246254622,
      "learning_rate": 1.3740825019002769e-06,
      "loss": 0.6933,
      "step": 10433
    },
    {
      "epoch": 0.765538303511652,
      "grad_norm": 1.4297773838243846,
      "learning_rate": 1.3732644645808858e-06,
      "loss": 0.7399,
      "step": 10434
    },
    {
      "epoch": 0.7656116731017911,
      "grad_norm": 1.5534036033430625,
      "learning_rate": 1.372446632072234e-06,
      "loss": 0.6725,
      "step": 10435
    },
    {
      "epoch": 0.7656850426919303,
      "grad_norm": 1.6541132791587128,
      "learning_rate": 1.3716290044205039e-06,
      "loss": 0.8156,
      "step": 10436
    },
    {
      "epoch": 0.7657584122820694,
      "grad_norm": 1.5179513943886893,
      "learning_rate": 1.3708115816718698e-06,
      "loss": 0.7343,
      "step": 10437
    },
    {
      "epoch": 0.7658317818722085,
      "grad_norm": 1.5659106975100667,
      "learning_rate": 1.3699943638724933e-06,
      "loss": 0.7488,
      "step": 10438
    },
    {
      "epoch": 0.7659051514623476,
      "grad_norm": 0.5216011846858132,
      "learning_rate": 1.369177351068523e-06,
      "loss": 0.4669,
      "step": 10439
    },
    {
      "epoch": 0.7659785210524868,
      "grad_norm": 0.5576697911759261,
      "learning_rate": 1.3683605433061004e-06,
      "loss": 0.4938,
      "step": 10440
    },
    {
      "epoch": 0.7659785210524868,
      "lm_loss": 1.2024,
      "step": 10440,
      "vm_loss": 0.0267
    },
    {
      "epoch": 0.7659785210524868,
      "lm_loss": 0.5302,
      "step": 10440,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.7659785210524868,
      "lm_loss": 0.0098,
      "step": 10440,
      "vm_loss": 0.1667
    },
    {
      "epoch": 0.7659785210524868,
      "lm_loss": 0.3862,
      "step": 10440,
      "vm_loss": 0.1498
    },
    {
      "epoch": 0.7659785210524868,
      "lm_loss": 0.7065,
      "step": 10440,
      "vm_loss": 0.1608
    },
    {
      "epoch": 0.7659785210524868,
      "lm_loss": 0.7147,
      "step": 10440,
      "vm_loss": 0.0601
    },
    {
      "epoch": 0.7659785210524868,
      "lm_loss": 0.1676,
      "step": 10440,
      "vm_loss": 0.0652
    },
    {
      "epoch": 0.7659785210524868,
      "lm_loss": 1.2222,
      "step": 10440,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.7660518906426259,
      "grad_norm": 1.4612637670585567,
      "learning_rate": 1.3675439406313519e-06,
      "loss": 0.7322,
      "step": 10441
    },
    {
      "epoch": 0.7661252602327651,
      "grad_norm": 0.5271375395359312,
      "learning_rate": 1.3667275430903926e-06,
      "loss": 0.4557,
      "step": 10442
    },
    {
      "epoch": 0.7661986298229041,
      "grad_norm": 1.4460807059719811,
      "learning_rate": 1.3659113507293265e-06,
      "loss": 0.788,
      "step": 10443
    },
    {
      "epoch": 0.7662719994130432,
      "grad_norm": 0.5208000590300759,
      "learning_rate": 1.3650953635942466e-06,
      "loss": 0.4578,
      "step": 10444
    },
    {
      "epoch": 0.7663453690031824,
      "grad_norm": 1.6236541140722254,
      "learning_rate": 1.364279581731231e-06,
      "loss": 0.8264,
      "step": 10445
    },
    {
      "epoch": 0.7664187385933215,
      "grad_norm": 1.5722972702417644,
      "learning_rate": 1.3634640051863529e-06,
      "loss": 0.6517,
      "step": 10446
    },
    {
      "epoch": 0.7664921081834607,
      "grad_norm": 1.5621767328516527,
      "learning_rate": 1.3626486340056682e-06,
      "loss": 0.7956,
      "step": 10447
    },
    {
      "epoch": 0.7665654777735997,
      "grad_norm": 1.5143803343613509,
      "learning_rate": 1.361833468235223e-06,
      "loss": 0.7696,
      "step": 10448
    },
    {
      "epoch": 0.7665654777735997,
      "lm_loss": 0.4176,
      "step": 10448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7665654777735997,
      "lm_loss": 0.4526,
      "step": 10448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7665654777735997,
      "lm_loss": 0.5062,
      "step": 10448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7665654777735997,
      "lm_loss": 0.4019,
      "step": 10448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7665654777735997,
      "lm_loss": 0.3913,
      "step": 10448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7665654777735997,
      "lm_loss": 0.389,
      "step": 10448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7665654777735997,
      "lm_loss": 0.4109,
      "step": 10448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7665654777735997,
      "lm_loss": 0.4596,
      "step": 10448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7666388473637389,
      "grad_norm": 0.5344155337719397,
      "learning_rate": 1.3610185079210514e-06,
      "loss": 0.4673,
      "step": 10449
    },
    {
      "epoch": 0.766712216953878,
      "grad_norm": 1.6795895548678905,
      "learning_rate": 1.360203753109175e-06,
      "loss": 0.7874,
      "step": 10450
    },
    {
      "epoch": 0.7667855865440172,
      "grad_norm": 0.5523721411909139,
      "learning_rate": 1.359389203845608e-06,
      "loss": 0.4986,
      "step": 10451
    },
    {
      "epoch": 0.7668589561341563,
      "grad_norm": 0.5339403555251988,
      "learning_rate": 1.3585748601763487e-06,
      "loss": 0.4878,
      "step": 10452
    },
    {
      "epoch": 0.7669323257242954,
      "grad_norm": 1.513861347471066,
      "learning_rate": 1.3577607221473848e-06,
      "loss": 0.7533,
      "step": 10453
    },
    {
      "epoch": 0.7670056953144345,
      "grad_norm": 1.3975854485401502,
      "learning_rate": 1.3569467898046924e-06,
      "loss": 0.7855,
      "step": 10454
    },
    {
      "epoch": 0.7670790649045737,
      "grad_norm": 0.5452623181011566,
      "learning_rate": 1.3561330631942371e-06,
      "loss": 0.4868,
      "step": 10455
    },
    {
      "epoch": 0.7671524344947128,
      "grad_norm": 1.4088048978016174,
      "learning_rate": 1.3553195423619697e-06,
      "loss": 0.6823,
      "step": 10456
    },
    {
      "epoch": 0.7671524344947128,
      "lm_loss": 1.3604,
      "step": 10456,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.7671524344947128,
      "lm_loss": 0.4418,
      "step": 10456,
      "vm_loss": 0.1624
    },
    {
      "epoch": 0.7671524344947128,
      "lm_loss": 0.3308,
      "step": 10456,
      "vm_loss": 0.0606
    },
    {
      "epoch": 0.7671524344947128,
      "lm_loss": 0.9305,
      "step": 10456,
      "vm_loss": 0.138
    },
    {
      "epoch": 0.7671524344947128,
      "lm_loss": 0.3846,
      "step": 10456,
      "vm_loss": 0.1209
    },
    {
      "epoch": 0.7671524344947128,
      "lm_loss": 0.0817,
      "step": 10456,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.7671524344947128,
      "lm_loss": 0.3979,
      "step": 10456,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.7671524344947128,
      "lm_loss": 0.9223,
      "step": 10456,
      "vm_loss": 0.1387
    },
    {
      "epoch": 0.767225804084852,
      "grad_norm": 1.4602567722244952,
      "learning_rate": 1.3545062273538357e-06,
      "loss": 0.6816,
      "step": 10457
    },
    {
      "epoch": 0.767299173674991,
      "grad_norm": 1.5950603207017842,
      "learning_rate": 1.3536931182157625e-06,
      "loss": 0.7206,
      "step": 10458
    },
    {
      "epoch": 0.7673725432651302,
      "grad_norm": 1.4225010407255492,
      "learning_rate": 1.3528802149936688e-06,
      "loss": 0.701,
      "step": 10459
    },
    {
      "epoch": 0.7674459128552693,
      "grad_norm": 0.5412333775757082,
      "learning_rate": 1.3520675177334614e-06,
      "loss": 0.4859,
      "step": 10460
    },
    {
      "epoch": 0.7675192824454085,
      "grad_norm": 1.6177261976984445,
      "learning_rate": 1.3512550264810336e-06,
      "loss": 0.7539,
      "step": 10461
    },
    {
      "epoch": 0.7675926520355476,
      "grad_norm": 1.692693364002185,
      "learning_rate": 1.3504427412822724e-06,
      "loss": 0.8123,
      "step": 10462
    },
    {
      "epoch": 0.7676660216256866,
      "grad_norm": 0.5608870205074049,
      "learning_rate": 1.3496306621830474e-06,
      "loss": 0.4783,
      "step": 10463
    },
    {
      "epoch": 0.7677393912158258,
      "grad_norm": 1.2976181487937897,
      "learning_rate": 1.3488187892292187e-06,
      "loss": 0.5884,
      "step": 10464
    },
    {
      "epoch": 0.7677393912158258,
      "lm_loss": 1.3898,
      "step": 10464,
      "vm_loss": 0.0699
    },
    {
      "epoch": 0.7677393912158258,
      "lm_loss": 0.8478,
      "step": 10464,
      "vm_loss": 0.0872
    },
    {
      "epoch": 0.7677393912158258,
      "lm_loss": 0.0671,
      "step": 10464,
      "vm_loss": 0.1439
    },
    {
      "epoch": 0.7677393912158258,
      "lm_loss": 0.5766,
      "step": 10464,
      "vm_loss": 0.1922
    },
    {
      "epoch": 0.7677393912158258,
      "lm_loss": 0.5296,
      "step": 10464,
      "vm_loss": 0.145
    },
    {
      "epoch": 0.7677393912158258,
      "lm_loss": 0.1621,
      "step": 10464,
      "vm_loss": 0.1444
    },
    {
      "epoch": 0.7677393912158258,
      "lm_loss": 0.1685,
      "step": 10464,
      "vm_loss": 0.1063
    },
    {
      "epoch": 0.7677393912158258,
      "lm_loss": 0.1055,
      "step": 10464,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.7678127608059649,
      "grad_norm": 1.4192149807321923,
      "learning_rate": 1.3480071224666347e-06,
      "loss": 0.6958,
      "step": 10465
    },
    {
      "epoch": 0.7678861303961041,
      "grad_norm": 1.340239353502846,
      "learning_rate": 1.3471956619411313e-06,
      "loss": 0.889,
      "step": 10466
    },
    {
      "epoch": 0.7679594999862432,
      "grad_norm": 0.5614341222536435,
      "learning_rate": 1.3463844076985367e-06,
      "loss": 0.4913,
      "step": 10467
    },
    {
      "epoch": 0.7680328695763823,
      "grad_norm": 1.6036790422522849,
      "learning_rate": 1.345573359784662e-06,
      "loss": 0.6717,
      "step": 10468
    },
    {
      "epoch": 0.7681062391665214,
      "grad_norm": 1.3500841156444627,
      "learning_rate": 1.34476251824531e-06,
      "loss": 0.7176,
      "step": 10469
    },
    {
      "epoch": 0.7681796087566606,
      "grad_norm": 1.3295002360937926,
      "learning_rate": 1.343951883126271e-06,
      "loss": 0.7702,
      "step": 10470
    },
    {
      "epoch": 0.7682529783467997,
      "grad_norm": 0.5516037249846091,
      "learning_rate": 1.343141454473323e-06,
      "loss": 0.4848,
      "step": 10471
    },
    {
      "epoch": 0.7683263479369389,
      "grad_norm": 0.5585284475895503,
      "learning_rate": 1.3423312323322314e-06,
      "loss": 0.496,
      "step": 10472
    },
    {
      "epoch": 0.7683263479369389,
      "lm_loss": 0.6105,
      "step": 10472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7683263479369389,
      "lm_loss": 0.4101,
      "step": 10472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7683263479369389,
      "lm_loss": 0.4276,
      "step": 10472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7683263479369389,
      "lm_loss": 0.4362,
      "step": 10472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7683263479369389,
      "lm_loss": 0.4573,
      "step": 10472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7683263479369389,
      "lm_loss": 0.5438,
      "step": 10472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7683263479369389,
      "lm_loss": 0.5171,
      "step": 10472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7683263479369389,
      "lm_loss": 0.5245,
      "step": 10472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7683997175270779,
      "grad_norm": 0.5407260548651033,
      "learning_rate": 1.3415212167487545e-06,
      "loss": 0.4685,
      "step": 10473
    },
    {
      "epoch": 0.7684730871172171,
      "grad_norm": 1.4887044013701383,
      "learning_rate": 1.340711407768635e-06,
      "loss": 0.7616,
      "step": 10474
    },
    {
      "epoch": 0.7685464567073562,
      "grad_norm": 0.5420297320182085,
      "learning_rate": 1.3399018054376045e-06,
      "loss": 0.4762,
      "step": 10475
    },
    {
      "epoch": 0.7686198262974954,
      "grad_norm": 0.5511370937385672,
      "learning_rate": 1.3390924098013824e-06,
      "loss": 0.474,
      "step": 10476
    },
    {
      "epoch": 0.7686931958876345,
      "grad_norm": 0.5508183575411779,
      "learning_rate": 1.3382832209056774e-06,
      "loss": 0.4823,
      "step": 10477
    },
    {
      "epoch": 0.7687665654777736,
      "grad_norm": 1.587636724285062,
      "learning_rate": 1.3374742387961881e-06,
      "loss": 0.7331,
      "step": 10478
    },
    {
      "epoch": 0.7688399350679127,
      "grad_norm": 1.5400394639403803,
      "learning_rate": 1.3366654635185983e-06,
      "loss": 0.8042,
      "step": 10479
    },
    {
      "epoch": 0.7689133046580519,
      "grad_norm": 1.4344569057691026,
      "learning_rate": 1.3358568951185825e-06,
      "loss": 0.7327,
      "step": 10480
    },
    {
      "epoch": 0.7689133046580519,
      "lm_loss": 1.5297,
      "step": 10480,
      "vm_loss": 0.0381
    },
    {
      "epoch": 0.7689133046580519,
      "lm_loss": 0.8184,
      "step": 10480,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.7689133046580519,
      "lm_loss": 0.8225,
      "step": 10480,
      "vm_loss": 0.1707
    },
    {
      "epoch": 0.7689133046580519,
      "lm_loss": 0.2267,
      "step": 10480,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.7689133046580519,
      "lm_loss": 0.7026,
      "step": 10480,
      "vm_loss": 0.1461
    },
    {
      "epoch": 0.7689133046580519,
      "lm_loss": 0.6608,
      "step": 10480,
      "vm_loss": 0.0925
    },
    {
      "epoch": 0.7689133046580519,
      "lm_loss": 0.565,
      "step": 10480,
      "vm_loss": 0.1567
    },
    {
      "epoch": 0.7689133046580519,
      "lm_loss": 0.8633,
      "step": 10480,
      "vm_loss": 0.1041
    },
    {
      "epoch": 0.768986674248191,
      "grad_norm": 1.535356212037089,
      "learning_rate": 1.3350485336418012e-06,
      "loss": 0.6768,
      "step": 10481
    },
    {
      "epoch": 0.7690600438383302,
      "grad_norm": 1.5306405988856977,
      "learning_rate": 1.3342403791339059e-06,
      "loss": 0.7302,
      "step": 10482
    },
    {
      "epoch": 0.7691334134284692,
      "grad_norm": 0.522469181793351,
      "learning_rate": 1.333432431640533e-06,
      "loss": 0.4834,
      "step": 10483
    },
    {
      "epoch": 0.7692067830186083,
      "grad_norm": 1.5407953365138825,
      "learning_rate": 1.3326246912073116e-06,
      "loss": 0.6927,
      "step": 10484
    },
    {
      "epoch": 0.7692801526087475,
      "grad_norm": 0.5514341996855562,
      "learning_rate": 1.3318171578798567e-06,
      "loss": 0.4938,
      "step": 10485
    },
    {
      "epoch": 0.7693535221988866,
      "grad_norm": 0.5430590262231977,
      "learning_rate": 1.3310098317037707e-06,
      "loss": 0.478,
      "step": 10486
    },
    {
      "epoch": 0.7694268917890258,
      "grad_norm": 1.6221521281892275,
      "learning_rate": 1.3302027127246459e-06,
      "loss": 0.7798,
      "step": 10487
    },
    {
      "epoch": 0.7695002613791648,
      "grad_norm": 1.4586730434269035,
      "learning_rate": 1.3293958009880598e-06,
      "loss": 0.7435,
      "step": 10488
    },
    {
      "epoch": 0.7695002613791648,
      "lm_loss": 0.4642,
      "step": 10488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7695002613791648,
      "lm_loss": 0.4926,
      "step": 10488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7695002613791648,
      "lm_loss": 0.563,
      "step": 10488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7695002613791648,
      "lm_loss": 0.3665,
      "step": 10488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7695002613791648,
      "lm_loss": 0.531,
      "step": 10488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7695002613791648,
      "lm_loss": 0.4545,
      "step": 10488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7695002613791648,
      "lm_loss": 0.4913,
      "step": 10488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7695002613791648,
      "lm_loss": 0.465,
      "step": 10488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.769573630969304,
      "grad_norm": 0.5613067910263742,
      "learning_rate": 1.3285890965395853e-06,
      "loss": 0.4863,
      "step": 10489
    },
    {
      "epoch": 0.7696470005594431,
      "grad_norm": 0.5613843963013745,
      "learning_rate": 1.3277825994247761e-06,
      "loss": 0.501,
      "step": 10490
    },
    {
      "epoch": 0.7697203701495823,
      "grad_norm": 1.4138006954512983,
      "learning_rate": 1.3269763096891774e-06,
      "loss": 0.8468,
      "step": 10491
    },
    {
      "epoch": 0.7697937397397214,
      "grad_norm": 1.8046760397253025,
      "learning_rate": 1.3261702273783235e-06,
      "loss": 0.8241,
      "step": 10492
    },
    {
      "epoch": 0.7698671093298605,
      "grad_norm": 1.7130111087491944,
      "learning_rate": 1.3253643525377324e-06,
      "loss": 0.7839,
      "step": 10493
    },
    {
      "epoch": 0.7699404789199996,
      "grad_norm": 0.5684735798016068,
      "learning_rate": 1.3245586852129184e-06,
      "loss": 0.4836,
      "step": 10494
    },
    {
      "epoch": 0.7700138485101388,
      "grad_norm": 1.308169248276317,
      "learning_rate": 1.3237532254493774e-06,
      "loss": 0.6993,
      "step": 10495
    },
    {
      "epoch": 0.7700872181002779,
      "grad_norm": 1.4898634730571327,
      "learning_rate": 1.322947973292596e-06,
      "loss": 0.7168,
      "step": 10496
    },
    {
      "epoch": 0.7700872181002779,
      "lm_loss": 1.1994,
      "step": 10496,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.7700872181002779,
      "lm_loss": 0.3622,
      "step": 10496,
      "vm_loss": 0.1878
    },
    {
      "epoch": 0.7700872181002779,
      "lm_loss": 0.7747,
      "step": 10496,
      "vm_loss": 0.131
    },
    {
      "epoch": 0.7700872181002779,
      "lm_loss": 0.988,
      "step": 10496,
      "vm_loss": 0.1578
    },
    {
      "epoch": 0.7700872181002779,
      "lm_loss": 0.9658,
      "step": 10496,
      "vm_loss": 0.1681
    },
    {
      "epoch": 0.7700872181002779,
      "lm_loss": 0.1908,
      "step": 10496,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.7700872181002779,
      "lm_loss": 0.949,
      "step": 10496,
      "vm_loss": 0.1135
    },
    {
      "epoch": 0.7700872181002779,
      "lm_loss": 1.5575,
      "step": 10496,
      "vm_loss": 0.0701
    },
    {
      "epoch": 0.7701605876904171,
      "grad_norm": 1.5242457700165262,
      "learning_rate": 1.3221429287880482e-06,
      "loss": 0.7309,
      "step": 10497
    },
    {
      "epoch": 0.7702339572805561,
      "grad_norm": 1.5958623337005304,
      "learning_rate": 1.321338091981198e-06,
      "loss": 0.8334,
      "step": 10498
    },
    {
      "epoch": 0.7703073268706953,
      "grad_norm": 1.493248321189614,
      "learning_rate": 1.3205334629174937e-06,
      "loss": 0.7175,
      "step": 10499
    },
    {
      "epoch": 0.7703806964608344,
      "grad_norm": 1.685517702794351,
      "learning_rate": 1.3197290416423786e-06,
      "loss": 0.8016,
      "step": 10500
    },
    {
      "epoch": 0.7704540660509736,
      "grad_norm": 1.527240513562527,
      "learning_rate": 1.3189248282012784e-06,
      "loss": 0.8385,
      "step": 10501
    },
    {
      "epoch": 0.7705274356411127,
      "grad_norm": 1.3842428898104089,
      "learning_rate": 1.3181208226396093e-06,
      "loss": 0.6879,
      "step": 10502
    },
    {
      "epoch": 0.7706008052312517,
      "grad_norm": 0.5474292507903372,
      "learning_rate": 1.317317025002775e-06,
      "loss": 0.4773,
      "step": 10503
    },
    {
      "epoch": 0.7706741748213909,
      "grad_norm": 0.5475112633110097,
      "learning_rate": 1.316513435336167e-06,
      "loss": 0.4905,
      "step": 10504
    },
    {
      "epoch": 0.7706741748213909,
      "lm_loss": 0.6385,
      "step": 10504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7706741748213909,
      "lm_loss": 0.4715,
      "step": 10504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7706741748213909,
      "lm_loss": 0.4664,
      "step": 10504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7706741748213909,
      "lm_loss": 0.5783,
      "step": 10504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7706741748213909,
      "lm_loss": 0.3242,
      "step": 10504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7706741748213909,
      "lm_loss": 0.4388,
      "step": 10504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7706741748213909,
      "lm_loss": 0.4569,
      "step": 10504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7706741748213909,
      "lm_loss": 0.3784,
      "step": 10504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.77074754441153,
      "grad_norm": 0.5593972098257562,
      "learning_rate": 1.315710053685169e-06,
      "loss": 0.4797,
      "step": 10505
    },
    {
      "epoch": 0.7708209140016692,
      "grad_norm": 1.3827638082007472,
      "learning_rate": 1.3149068800951481e-06,
      "loss": 0.7311,
      "step": 10506
    },
    {
      "epoch": 0.7708942835918083,
      "grad_norm": 1.4699334564772675,
      "learning_rate": 1.3141039146114615e-06,
      "loss": 0.6896,
      "step": 10507
    },
    {
      "epoch": 0.7709676531819474,
      "grad_norm": 0.5596771992033933,
      "learning_rate": 1.313301157279455e-06,
      "loss": 0.4814,
      "step": 10508
    },
    {
      "epoch": 0.7710410227720865,
      "grad_norm": 1.511241707387344,
      "learning_rate": 1.3124986081444625e-06,
      "loss": 0.6612,
      "step": 10509
    },
    {
      "epoch": 0.7711143923622257,
      "grad_norm": 0.56557089668528,
      "learning_rate": 1.3116962672518036e-06,
      "loss": 0.4778,
      "step": 10510
    },
    {
      "epoch": 0.7711877619523648,
      "grad_norm": 1.6025057890868142,
      "learning_rate": 1.3108941346467917e-06,
      "loss": 0.7422,
      "step": 10511
    },
    {
      "epoch": 0.771261131542504,
      "grad_norm": 1.4844861729806724,
      "learning_rate": 1.3100922103747243e-06,
      "loss": 0.7852,
      "step": 10512
    },
    {
      "epoch": 0.771261131542504,
      "lm_loss": 1.3859,
      "step": 10512,
      "vm_loss": 0.1028
    },
    {
      "epoch": 0.771261131542504,
      "lm_loss": 1.1621,
      "step": 10512,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.771261131542504,
      "lm_loss": 1.0064,
      "step": 10512,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.771261131542504,
      "lm_loss": 0.0358,
      "step": 10512,
      "vm_loss": 0.0454
    },
    {
      "epoch": 0.771261131542504,
      "lm_loss": 0.5538,
      "step": 10512,
      "vm_loss": 0.0889
    },
    {
      "epoch": 0.771261131542504,
      "lm_loss": 1.3377,
      "step": 10512,
      "vm_loss": 0.1868
    },
    {
      "epoch": 0.771261131542504,
      "lm_loss": 0.1614,
      "step": 10512,
      "vm_loss": 0.1137
    },
    {
      "epoch": 0.771261131542504,
      "lm_loss": 0.7429,
      "step": 10512,
      "vm_loss": 0.1857
    },
    {
      "epoch": 0.771334501132643,
      "grad_norm": 1.7323645647475632,
      "learning_rate": 1.3092904944808876e-06,
      "loss": 0.7241,
      "step": 10513
    },
    {
      "epoch": 0.7714078707227822,
      "grad_norm": 1.50858036704412,
      "learning_rate": 1.3084889870105566e-06,
      "loss": 0.6558,
      "step": 10514
    },
    {
      "epoch": 0.7714812403129213,
      "grad_norm": 0.5453221114174707,
      "learning_rate": 1.3076876880089927e-06,
      "loss": 0.4651,
      "step": 10515
    },
    {
      "epoch": 0.7715546099030605,
      "grad_norm": 1.750155285933563,
      "learning_rate": 1.3068865975214505e-06,
      "loss": 0.696,
      "step": 10516
    },
    {
      "epoch": 0.7716279794931995,
      "grad_norm": 1.5422166563857538,
      "learning_rate": 1.3060857155931683e-06,
      "loss": 0.7159,
      "step": 10517
    },
    {
      "epoch": 0.7717013490833387,
      "grad_norm": 1.4244963521350555,
      "learning_rate": 1.305285042269373e-06,
      "loss": 0.7757,
      "step": 10518
    },
    {
      "epoch": 0.7717747186734778,
      "grad_norm": 0.5531498775958702,
      "learning_rate": 1.3044845775952814e-06,
      "loss": 0.48,
      "step": 10519
    },
    {
      "epoch": 0.771848088263617,
      "grad_norm": 1.3385032837542048,
      "learning_rate": 1.303684321616096e-06,
      "loss": 0.8367,
      "step": 10520
    },
    {
      "epoch": 0.771848088263617,
      "lm_loss": 0.4445,
      "step": 10520,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.771848088263617,
      "lm_loss": 0.4512,
      "step": 10520,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.771848088263617,
      "lm_loss": 0.534,
      "step": 10520,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.771848088263617,
      "lm_loss": 0.3716,
      "step": 10520,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.771848088263617,
      "lm_loss": 0.4662,
      "step": 10520,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.771848088263617,
      "lm_loss": 0.5098,
      "step": 10520,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.771848088263617,
      "lm_loss": 0.5392,
      "step": 10520,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.771848088263617,
      "lm_loss": 0.5408,
      "step": 10520,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7719214578537561,
      "grad_norm": 0.5445850626294712,
      "learning_rate": 1.3028842743770126e-06,
      "loss": 0.4865,
      "step": 10521
    },
    {
      "epoch": 0.7719948274438951,
      "grad_norm": 1.6093211367691944,
      "learning_rate": 1.3020844359232093e-06,
      "loss": 0.7243,
      "step": 10522
    },
    {
      "epoch": 0.7720681970340343,
      "grad_norm": 0.5450008732604984,
      "learning_rate": 1.301284806299856e-06,
      "loss": 0.4581,
      "step": 10523
    },
    {
      "epoch": 0.7721415666241734,
      "grad_norm": 0.548749306648556,
      "learning_rate": 1.30048538555211e-06,
      "loss": 0.4723,
      "step": 10524
    },
    {
      "epoch": 0.7722149362143126,
      "grad_norm": 1.5091808475252166,
      "learning_rate": 1.2996861737251148e-06,
      "loss": 0.7471,
      "step": 10525
    },
    {
      "epoch": 0.7722883058044517,
      "grad_norm": 1.5295190921671773,
      "learning_rate": 1.2988871708640043e-06,
      "loss": 0.695,
      "step": 10526
    },
    {
      "epoch": 0.7723616753945908,
      "grad_norm": 1.635772888714161,
      "learning_rate": 1.2980883770139026e-06,
      "loss": 0.6917,
      "step": 10527
    },
    {
      "epoch": 0.7724350449847299,
      "grad_norm": 0.534083572460533,
      "learning_rate": 1.2972897922199174e-06,
      "loss": 0.4586,
      "step": 10528
    },
    {
      "epoch": 0.7724350449847299,
      "lm_loss": 1.4708,
      "step": 10528,
      "vm_loss": 0.0235
    },
    {
      "epoch": 0.7724350449847299,
      "lm_loss": 0.1505,
      "step": 10528,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.7724350449847299,
      "lm_loss": 0.7999,
      "step": 10528,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.7724350449847299,
      "lm_loss": 0.796,
      "step": 10528,
      "vm_loss": 0.2223
    },
    {
      "epoch": 0.7724350449847299,
      "lm_loss": 0.6288,
      "step": 10528,
      "vm_loss": 0.177
    },
    {
      "epoch": 0.7724350449847299,
      "lm_loss": 0.1994,
      "step": 10528,
      "vm_loss": 0.1667
    },
    {
      "epoch": 0.7724350449847299,
      "lm_loss": 0.1325,
      "step": 10528,
      "vm_loss": 0.0907
    },
    {
      "epoch": 0.7724350449847299,
      "lm_loss": 0.8613,
      "step": 10528,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.7725084145748691,
      "grad_norm": 1.520045935934509,
      "learning_rate": 1.296491416527147e-06,
      "loss": 0.713,
      "step": 10529
    },
    {
      "epoch": 0.7725817841650082,
      "grad_norm": 0.5538176655005681,
      "learning_rate": 1.2956932499806786e-06,
      "loss": 0.4914,
      "step": 10530
    },
    {
      "epoch": 0.7726551537551474,
      "grad_norm": 1.6768078727422413,
      "learning_rate": 1.2948952926255838e-06,
      "loss": 0.8238,
      "step": 10531
    },
    {
      "epoch": 0.7727285233452864,
      "grad_norm": 1.46789320571786,
      "learning_rate": 1.2940975445069293e-06,
      "loss": 0.7715,
      "step": 10532
    },
    {
      "epoch": 0.7728018929354256,
      "grad_norm": 1.7860905545176091,
      "learning_rate": 1.2933000056697637e-06,
      "loss": 0.763,
      "step": 10533
    },
    {
      "epoch": 0.7728752625255647,
      "grad_norm": 0.5356292415762925,
      "learning_rate": 1.2925026761591264e-06,
      "loss": 0.4686,
      "step": 10534
    },
    {
      "epoch": 0.7729486321157039,
      "grad_norm": 1.7818076508178977,
      "learning_rate": 1.2917055560200443e-06,
      "loss": 0.7489,
      "step": 10535
    },
    {
      "epoch": 0.773022001705843,
      "grad_norm": 1.6958788154014894,
      "learning_rate": 1.2909086452975327e-06,
      "loss": 0.6916,
      "step": 10536
    },
    {
      "epoch": 0.773022001705843,
      "lm_loss": 0.6456,
      "step": 10536,
      "vm_loss": 0.0724
    },
    {
      "epoch": 0.773022001705843,
      "lm_loss": 1.316,
      "step": 10536,
      "vm_loss": 0.0569
    },
    {
      "epoch": 0.773022001705843,
      "lm_loss": 1.1596,
      "step": 10536,
      "vm_loss": 0.1813
    },
    {
      "epoch": 0.773022001705843,
      "lm_loss": 0.4701,
      "step": 10536,
      "vm_loss": 0.1588
    },
    {
      "epoch": 0.773022001705843,
      "lm_loss": 0.6779,
      "step": 10536,
      "vm_loss": 0.0644
    },
    {
      "epoch": 0.773022001705843,
      "lm_loss": 0.3603,
      "step": 10536,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.773022001705843,
      "lm_loss": 0.6831,
      "step": 10536,
      "vm_loss": 0.1057
    },
    {
      "epoch": 0.773022001705843,
      "lm_loss": 0.2243,
      "step": 10536,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.773095371295982,
      "grad_norm": 1.5016192780113893,
      "learning_rate": 1.2901119440365944e-06,
      "loss": 0.7058,
      "step": 10537
    },
    {
      "epoch": 0.7731687408861212,
      "grad_norm": 0.5441538470845916,
      "learning_rate": 1.2893154522822231e-06,
      "loss": 0.4667,
      "step": 10538
    },
    {
      "epoch": 0.7732421104762603,
      "grad_norm": 1.5217010927657897,
      "learning_rate": 1.288519170079398e-06,
      "loss": 0.7945,
      "step": 10539
    },
    {
      "epoch": 0.7733154800663995,
      "grad_norm": 0.5566142590944344,
      "learning_rate": 1.2877230974730865e-06,
      "loss": 0.4981,
      "step": 10540
    },
    {
      "epoch": 0.7733888496565386,
      "grad_norm": 1.5377972535528412,
      "learning_rate": 1.2869272345082446e-06,
      "loss": 0.7258,
      "step": 10541
    },
    {
      "epoch": 0.7734622192466777,
      "grad_norm": 1.5147453756620177,
      "learning_rate": 1.2861315812298159e-06,
      "loss": 0.7626,
      "step": 10542
    },
    {
      "epoch": 0.7735355888368168,
      "grad_norm": 1.5904594457751893,
      "learning_rate": 1.2853361376827356e-06,
      "loss": 0.737,
      "step": 10543
    },
    {
      "epoch": 0.773608958426956,
      "grad_norm": 1.5467612707746952,
      "learning_rate": 1.2845409039119222e-06,
      "loss": 0.7515,
      "step": 10544
    },
    {
      "epoch": 0.773608958426956,
      "lm_loss": 0.4148,
      "step": 10544,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.773608958426956,
      "lm_loss": 0.4586,
      "step": 10544,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.773608958426956,
      "lm_loss": 0.3906,
      "step": 10544,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.773608958426956,
      "lm_loss": 0.6015,
      "step": 10544,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.773608958426956,
      "lm_loss": 0.3848,
      "step": 10544,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.773608958426956,
      "lm_loss": 0.4999,
      "step": 10544,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.773608958426956,
      "lm_loss": 0.5032,
      "step": 10544,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.773608958426956,
      "lm_loss": 0.4615,
      "step": 10544,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7736823280170951,
      "grad_norm": 0.539873598937569,
      "learning_rate": 1.283745879962286e-06,
      "loss": 0.4646,
      "step": 10545
    },
    {
      "epoch": 0.7737556976072343,
      "grad_norm": 1.5716738278028364,
      "learning_rate": 1.2829510658787224e-06,
      "loss": 0.7392,
      "step": 10546
    },
    {
      "epoch": 0.7738290671973733,
      "grad_norm": 0.5531467467357607,
      "learning_rate": 1.282156461706116e-06,
      "loss": 0.471,
      "step": 10547
    },
    {
      "epoch": 0.7739024367875125,
      "grad_norm": 1.5452871133382702,
      "learning_rate": 1.2813620674893429e-06,
      "loss": 0.6894,
      "step": 10548
    },
    {
      "epoch": 0.7739758063776516,
      "grad_norm": 0.5613038956056324,
      "learning_rate": 1.2805678832732627e-06,
      "loss": 0.4855,
      "step": 10549
    },
    {
      "epoch": 0.7740491759677908,
      "grad_norm": 0.5288016888266365,
      "learning_rate": 1.2797739091027268e-06,
      "loss": 0.4636,
      "step": 10550
    },
    {
      "epoch": 0.7741225455579299,
      "grad_norm": 0.5629670362044028,
      "learning_rate": 1.2789801450225696e-06,
      "loss": 0.49,
      "step": 10551
    },
    {
      "epoch": 0.774195915148069,
      "grad_norm": 1.3948113045957613,
      "learning_rate": 1.2781865910776175e-06,
      "loss": 0.6565,
      "step": 10552
    },
    {
      "epoch": 0.774195915148069,
      "lm_loss": 0.4103,
      "step": 10552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774195915148069,
      "lm_loss": 0.4245,
      "step": 10552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774195915148069,
      "lm_loss": 0.4508,
      "step": 10552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774195915148069,
      "lm_loss": 0.4745,
      "step": 10552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774195915148069,
      "lm_loss": 0.5425,
      "step": 10552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774195915148069,
      "lm_loss": 0.6212,
      "step": 10552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774195915148069,
      "lm_loss": 0.5246,
      "step": 10552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774195915148069,
      "lm_loss": 0.6062,
      "step": 10552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7742692847382081,
      "grad_norm": 0.5442927943298025,
      "learning_rate": 1.2773932473126865e-06,
      "loss": 0.4969,
      "step": 10553
    },
    {
      "epoch": 0.7743426543283473,
      "grad_norm": 1.3891947366786468,
      "learning_rate": 1.2766001137725782e-06,
      "loss": 0.5709,
      "step": 10554
    },
    {
      "epoch": 0.7744160239184864,
      "grad_norm": 1.5378182038771946,
      "learning_rate": 1.275807190502082e-06,
      "loss": 0.7567,
      "step": 10555
    },
    {
      "epoch": 0.7744893935086256,
      "grad_norm": 1.446745766874692,
      "learning_rate": 1.275014477545976e-06,
      "loss": 0.754,
      "step": 10556
    },
    {
      "epoch": 0.7745627630987646,
      "grad_norm": 1.5503231998918494,
      "learning_rate": 1.2742219749490264e-06,
      "loss": 0.7115,
      "step": 10557
    },
    {
      "epoch": 0.7746361326889037,
      "grad_norm": 1.6864224360683484,
      "learning_rate": 1.27342968275599e-06,
      "loss": 0.8359,
      "step": 10558
    },
    {
      "epoch": 0.7747095022790429,
      "grad_norm": 0.5451909676355295,
      "learning_rate": 1.2726376010116082e-06,
      "loss": 0.4723,
      "step": 10559
    },
    {
      "epoch": 0.774782871869182,
      "grad_norm": 1.5680365648167063,
      "learning_rate": 1.2718457297606119e-06,
      "loss": 0.7884,
      "step": 10560
    },
    {
      "epoch": 0.774782871869182,
      "lm_loss": 0.5241,
      "step": 10560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774782871869182,
      "lm_loss": 0.4384,
      "step": 10560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774782871869182,
      "lm_loss": 0.5753,
      "step": 10560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774782871869182,
      "lm_loss": 0.7899,
      "step": 10560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774782871869182,
      "lm_loss": 0.5543,
      "step": 10560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774782871869182,
      "lm_loss": 0.395,
      "step": 10560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774782871869182,
      "lm_loss": 0.572,
      "step": 10560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.774782871869182,
      "lm_loss": 0.4765,
      "step": 10560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7748562414593212,
      "grad_norm": 0.5697050646289449,
      "learning_rate": 1.2710540690477196e-06,
      "loss": 0.4975,
      "step": 10561
    },
    {
      "epoch": 0.7749296110494602,
      "grad_norm": 1.5649422267003756,
      "learning_rate": 1.270262618917637e-06,
      "loss": 0.7671,
      "step": 10562
    },
    {
      "epoch": 0.7750029806395994,
      "grad_norm": 1.3506749812342136,
      "learning_rate": 1.2694713794150626e-06,
      "loss": 0.5598,
      "step": 10563
    },
    {
      "epoch": 0.7750763502297385,
      "grad_norm": 1.5304433465203453,
      "learning_rate": 1.2686803505846773e-06,
      "loss": 0.7349,
      "step": 10564
    },
    {
      "epoch": 0.7751497198198777,
      "grad_norm": 0.5343190983693705,
      "learning_rate": 1.2678895324711537e-06,
      "loss": 0.4666,
      "step": 10565
    },
    {
      "epoch": 0.7752230894100168,
      "grad_norm": 0.5495463565320458,
      "learning_rate": 1.2670989251191506e-06,
      "loss": 0.4675,
      "step": 10566
    },
    {
      "epoch": 0.7752964590001559,
      "grad_norm": 0.5581350579713942,
      "learning_rate": 1.2663085285733152e-06,
      "loss": 0.475,
      "step": 10567
    },
    {
      "epoch": 0.775369828590295,
      "grad_norm": 1.5827010857722548,
      "learning_rate": 1.2655183428782824e-06,
      "loss": 0.7273,
      "step": 10568
    },
    {
      "epoch": 0.775369828590295,
      "lm_loss": 1.1979,
      "step": 10568,
      "vm_loss": 0.0272
    },
    {
      "epoch": 0.775369828590295,
      "lm_loss": 1.0411,
      "step": 10568,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.775369828590295,
      "lm_loss": 0.24,
      "step": 10568,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.775369828590295,
      "lm_loss": 0.1514,
      "step": 10568,
      "vm_loss": 0.0943
    },
    {
      "epoch": 0.775369828590295,
      "lm_loss": 0.3537,
      "step": 10568,
      "vm_loss": 0.0928
    },
    {
      "epoch": 0.775369828590295,
      "lm_loss": 0.6567,
      "step": 10568,
      "vm_loss": 0.1899
    },
    {
      "epoch": 0.775369828590295,
      "lm_loss": 0.3313,
      "step": 10568,
      "vm_loss": 0.1467
    },
    {
      "epoch": 0.775369828590295,
      "lm_loss": 0.0396,
      "step": 10568,
      "vm_loss": 0.1148
    },
    {
      "epoch": 0.7754431981804342,
      "grad_norm": 1.5619009235655066,
      "learning_rate": 1.264728368078678e-06,
      "loss": 0.6954,
      "step": 10569
    },
    {
      "epoch": 0.7755165677705733,
      "grad_norm": 1.5421336169768025,
      "learning_rate": 1.2639386042191138e-06,
      "loss": 0.8857,
      "step": 10570
    },
    {
      "epoch": 0.7755899373607125,
      "grad_norm": 1.4879974229293484,
      "learning_rate": 1.263149051344188e-06,
      "loss": 0.7065,
      "step": 10571
    },
    {
      "epoch": 0.7756633069508515,
      "grad_norm": 0.5635905422932818,
      "learning_rate": 1.2623597094984897e-06,
      "loss": 0.495,
      "step": 10572
    },
    {
      "epoch": 0.7757366765409907,
      "grad_norm": 1.6349123515002326,
      "learning_rate": 1.261570578726593e-06,
      "loss": 0.7178,
      "step": 10573
    },
    {
      "epoch": 0.7758100461311298,
      "grad_norm": 1.4963351694803275,
      "learning_rate": 1.2607816590730658e-06,
      "loss": 0.6874,
      "step": 10574
    },
    {
      "epoch": 0.775883415721269,
      "grad_norm": 1.4660538418282951,
      "learning_rate": 1.2599929505824582e-06,
      "loss": 0.7292,
      "step": 10575
    },
    {
      "epoch": 0.7759567853114081,
      "grad_norm": 0.5321259096932686,
      "learning_rate": 1.2592044532993102e-06,
      "loss": 0.4728,
      "step": 10576
    },
    {
      "epoch": 0.7759567853114081,
      "lm_loss": 1.4019,
      "step": 10576,
      "vm_loss": 0.0639
    },
    {
      "epoch": 0.7759567853114081,
      "lm_loss": 1.0177,
      "step": 10576,
      "vm_loss": 0.0475
    },
    {
      "epoch": 0.7759567853114081,
      "lm_loss": 0.0205,
      "step": 10576,
      "vm_loss": 0.1918
    },
    {
      "epoch": 0.7759567853114081,
      "lm_loss": 0.8384,
      "step": 10576,
      "vm_loss": 0.273
    },
    {
      "epoch": 0.7759567853114081,
      "lm_loss": 0.6733,
      "step": 10576,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.7759567853114081,
      "lm_loss": 0.4017,
      "step": 10576,
      "vm_loss": 0.1449
    },
    {
      "epoch": 0.7759567853114081,
      "lm_loss": 0.2602,
      "step": 10576,
      "vm_loss": 0.1313
    },
    {
      "epoch": 0.7759567853114081,
      "lm_loss": 0.3706,
      "step": 10576,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.7760301549015471,
      "grad_norm": 1.6348455123656722,
      "learning_rate": 1.2584161672681511e-06,
      "loss": 0.7776,
      "step": 10577
    },
    {
      "epoch": 0.7761035244916863,
      "grad_norm": 0.5260812033810928,
      "learning_rate": 1.2576280925334966e-06,
      "loss": 0.4511,
      "step": 10578
    },
    {
      "epoch": 0.7761768940818254,
      "grad_norm": 0.5592230155490268,
      "learning_rate": 1.2568402291398501e-06,
      "loss": 0.4773,
      "step": 10579
    },
    {
      "epoch": 0.7762502636719646,
      "grad_norm": 0.5459950026016085,
      "learning_rate": 1.2560525771317079e-06,
      "loss": 0.4847,
      "step": 10580
    },
    {
      "epoch": 0.7763236332621037,
      "grad_norm": 1.517131450938998,
      "learning_rate": 1.2552651365535473e-06,
      "loss": 0.7416,
      "step": 10581
    },
    {
      "epoch": 0.7763970028522428,
      "grad_norm": 1.571904298478329,
      "learning_rate": 1.2544779074498387e-06,
      "loss": 0.7945,
      "step": 10582
    },
    {
      "epoch": 0.7764703724423819,
      "grad_norm": 1.6998283962886833,
      "learning_rate": 1.253690889865038e-06,
      "loss": 0.7654,
      "step": 10583
    },
    {
      "epoch": 0.7765437420325211,
      "grad_norm": 0.5367530609230278,
      "learning_rate": 1.2529040838435891e-06,
      "loss": 0.4748,
      "step": 10584
    },
    {
      "epoch": 0.7765437420325211,
      "lm_loss": 0.9124,
      "step": 10584,
      "vm_loss": 0.0547
    },
    {
      "epoch": 0.7765437420325211,
      "lm_loss": 0.75,
      "step": 10584,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.7765437420325211,
      "lm_loss": 0.5797,
      "step": 10584,
      "vm_loss": 0.1817
    },
    {
      "epoch": 0.7765437420325211,
      "lm_loss": 1.1536,
      "step": 10584,
      "vm_loss": 0.1748
    },
    {
      "epoch": 0.7765437420325211,
      "lm_loss": 0.6615,
      "step": 10584,
      "vm_loss": 0.11
    },
    {
      "epoch": 0.7765437420325211,
      "lm_loss": 0.9422,
      "step": 10584,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.7765437420325211,
      "lm_loss": 0.5081,
      "step": 10584,
      "vm_loss": 0.2197
    },
    {
      "epoch": 0.7765437420325211,
      "lm_loss": 0.0197,
      "step": 10584,
      "vm_loss": 0.1643
    },
    {
      "epoch": 0.7766171116226602,
      "grad_norm": 1.543909448158555,
      "learning_rate": 1.2521174894299275e-06,
      "loss": 0.741,
      "step": 10585
    },
    {
      "epoch": 0.7766904812127994,
      "grad_norm": 0.566140569085249,
      "learning_rate": 1.251331106668473e-06,
      "loss": 0.471,
      "step": 10586
    },
    {
      "epoch": 0.7767638508029384,
      "grad_norm": 1.4973473798438126,
      "learning_rate": 1.2505449356036336e-06,
      "loss": 0.7582,
      "step": 10587
    },
    {
      "epoch": 0.7768372203930776,
      "grad_norm": 1.5298786378529952,
      "learning_rate": 1.249758976279808e-06,
      "loss": 0.7468,
      "step": 10588
    },
    {
      "epoch": 0.7769105899832167,
      "grad_norm": 0.5563087357996683,
      "learning_rate": 1.248973228741378e-06,
      "loss": 0.4897,
      "step": 10589
    },
    {
      "epoch": 0.7769839595733559,
      "grad_norm": 1.3696089328858778,
      "learning_rate": 1.2481876930327208e-06,
      "loss": 0.684,
      "step": 10590
    },
    {
      "epoch": 0.777057329163495,
      "grad_norm": 1.444177897520498,
      "learning_rate": 1.2474023691981963e-06,
      "loss": 0.702,
      "step": 10591
    },
    {
      "epoch": 0.777130698753634,
      "grad_norm": 1.41652624819431,
      "learning_rate": 1.2466172572821527e-06,
      "loss": 0.8529,
      "step": 10592
    },
    {
      "epoch": 0.777130698753634,
      "lm_loss": 1.044,
      "step": 10592,
      "vm_loss": 0.1112
    },
    {
      "epoch": 0.777130698753634,
      "lm_loss": 1.5483,
      "step": 10592,
      "vm_loss": 0.0678
    },
    {
      "epoch": 0.777130698753634,
      "lm_loss": 0.2388,
      "step": 10592,
      "vm_loss": 0.2503
    },
    {
      "epoch": 0.777130698753634,
      "lm_loss": 1.4028,
      "step": 10592,
      "vm_loss": 0.1644
    },
    {
      "epoch": 0.777130698753634,
      "lm_loss": 0.8055,
      "step": 10592,
      "vm_loss": 0.1644
    },
    {
      "epoch": 0.777130698753634,
      "lm_loss": 0.4484,
      "step": 10592,
      "vm_loss": 0.0711
    },
    {
      "epoch": 0.777130698753634,
      "lm_loss": 0.4112,
      "step": 10592,
      "vm_loss": 0.1417
    },
    {
      "epoch": 0.777130698753634,
      "lm_loss": 1.3437,
      "step": 10592,
      "vm_loss": 0.2199
    },
    {
      "epoch": 0.7772040683437732,
      "grad_norm": 1.4516118742486697,
      "learning_rate": 1.2458323573289276e-06,
      "loss": 0.8016,
      "step": 10593
    },
    {
      "epoch": 0.7772774379339124,
      "grad_norm": 0.5559601336546668,
      "learning_rate": 1.2450476693828461e-06,
      "loss": 0.4702,
      "step": 10594
    },
    {
      "epoch": 0.7773508075240515,
      "grad_norm": 1.6630457426919512,
      "learning_rate": 1.244263193488221e-06,
      "loss": 0.7869,
      "step": 10595
    },
    {
      "epoch": 0.7774241771141907,
      "grad_norm": 1.3889717409592228,
      "learning_rate": 1.2434789296893552e-06,
      "loss": 0.7325,
      "step": 10596
    },
    {
      "epoch": 0.7774975467043297,
      "grad_norm": 1.7566938631701734,
      "learning_rate": 1.242694878030537e-06,
      "loss": 0.6911,
      "step": 10597
    },
    {
      "epoch": 0.7775709162944688,
      "grad_norm": 1.5462369079123264,
      "learning_rate": 1.241911038556044e-06,
      "loss": 0.7198,
      "step": 10598
    },
    {
      "epoch": 0.777644285884608,
      "grad_norm": 1.5273209745303766,
      "learning_rate": 1.2411274113101418e-06,
      "loss": 0.6446,
      "step": 10599
    },
    {
      "epoch": 0.7777176554747471,
      "grad_norm": 1.5604266255921895,
      "learning_rate": 1.2403439963370811e-06,
      "loss": 0.7178,
      "step": 10600
    },
    {
      "epoch": 0.7777176554747471,
      "lm_loss": 0.974,
      "step": 10600,
      "vm_loss": 0.1376
    },
    {
      "epoch": 0.7777176554747471,
      "lm_loss": 1.366,
      "step": 10600,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.7777176554747471,
      "lm_loss": 0.1246,
      "step": 10600,
      "vm_loss": 0.1974
    },
    {
      "epoch": 0.7777176554747471,
      "lm_loss": 0.7455,
      "step": 10600,
      "vm_loss": 0.1299
    },
    {
      "epoch": 0.7777176554747471,
      "lm_loss": 0.8669,
      "step": 10600,
      "vm_loss": 0.1187
    },
    {
      "epoch": 0.7777176554747471,
      "lm_loss": 0.3284,
      "step": 10600,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.7777176554747471,
      "lm_loss": 0.3149,
      "step": 10600,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.7777176554747471,
      "lm_loss": 0.7837,
      "step": 10600,
      "vm_loss": 0.17
    },
    {
      "epoch": 0.7777910250648862,
      "grad_norm": 1.684698276697018,
      "learning_rate": 1.2395607936811078e-06,
      "loss": 0.9336,
      "step": 10601
    },
    {
      "epoch": 0.7778643946550253,
      "grad_norm": 1.79610173284946,
      "learning_rate": 1.238777803386449e-06,
      "loss": 0.8163,
      "step": 10602
    },
    {
      "epoch": 0.7779377642451645,
      "grad_norm": 1.5590885567029618,
      "learning_rate": 1.2379950254973217e-06,
      "loss": 0.7217,
      "step": 10603
    },
    {
      "epoch": 0.7780111338353036,
      "grad_norm": 1.4445800766571926,
      "learning_rate": 1.2372124600579323e-06,
      "loss": 0.6911,
      "step": 10604
    },
    {
      "epoch": 0.7780845034254428,
      "grad_norm": 1.4118128088412247,
      "learning_rate": 1.2364301071124728e-06,
      "loss": 0.7296,
      "step": 10605
    },
    {
      "epoch": 0.7781578730155818,
      "grad_norm": 0.5350708065689977,
      "learning_rate": 1.2356479667051247e-06,
      "loss": 0.4682,
      "step": 10606
    },
    {
      "epoch": 0.778231242605721,
      "grad_norm": 1.4790932158799128,
      "learning_rate": 1.2348660388800593e-06,
      "loss": 0.7365,
      "step": 10607
    },
    {
      "epoch": 0.7783046121958601,
      "grad_norm": 0.5723483287241253,
      "learning_rate": 1.234084323681433e-06,
      "loss": 0.4933,
      "step": 10608
    },
    {
      "epoch": 0.7783046121958601,
      "lm_loss": 0.1547,
      "step": 10608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7783046121958601,
      "lm_loss": 0.3885,
      "step": 10608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7783046121958601,
      "lm_loss": 0.5126,
      "step": 10608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7783046121958601,
      "lm_loss": 0.418,
      "step": 10608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7783046121958601,
      "lm_loss": 0.415,
      "step": 10608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7783046121958601,
      "lm_loss": 0.516,
      "step": 10608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7783046121958601,
      "lm_loss": 0.4396,
      "step": 10608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7783046121958601,
      "lm_loss": 0.454,
      "step": 10608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7783779817859993,
      "grad_norm": 0.5563932097822645,
      "learning_rate": 1.2333028211533916e-06,
      "loss": 0.4778,
      "step": 10609
    },
    {
      "epoch": 0.7784513513761384,
      "grad_norm": 1.6557686981316042,
      "learning_rate": 1.2325215313400673e-06,
      "loss": 0.8012,
      "step": 10610
    },
    {
      "epoch": 0.7785247209662775,
      "grad_norm": 1.722365497511698,
      "learning_rate": 1.231740454285581e-06,
      "loss": 0.851,
      "step": 10611
    },
    {
      "epoch": 0.7785980905564166,
      "grad_norm": 1.5629581185560657,
      "learning_rate": 1.2309595900340443e-06,
      "loss": 0.7221,
      "step": 10612
    },
    {
      "epoch": 0.7786714601465557,
      "grad_norm": 1.5911442863353948,
      "learning_rate": 1.2301789386295538e-06,
      "loss": 0.7395,
      "step": 10613
    },
    {
      "epoch": 0.7787448297366949,
      "grad_norm": 0.547212220894518,
      "learning_rate": 1.229398500116194e-06,
      "loss": 0.4847,
      "step": 10614
    },
    {
      "epoch": 0.778818199326834,
      "grad_norm": 1.435049464090869,
      "learning_rate": 1.2286182745380387e-06,
      "loss": 0.6834,
      "step": 10615
    },
    {
      "epoch": 0.7788915689169731,
      "grad_norm": 1.4721690014428466,
      "learning_rate": 1.2278382619391483e-06,
      "loss": 0.739,
      "step": 10616
    },
    {
      "epoch": 0.7788915689169731,
      "lm_loss": 0.5476,
      "step": 10616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7788915689169731,
      "lm_loss": 0.4471,
      "step": 10616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7788915689169731,
      "lm_loss": 0.5745,
      "step": 10616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7788915689169731,
      "lm_loss": 0.4289,
      "step": 10616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7788915689169731,
      "lm_loss": 0.3433,
      "step": 10616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7788915689169731,
      "lm_loss": 0.5561,
      "step": 10616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7788915689169731,
      "lm_loss": 0.4839,
      "step": 10616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7788915689169731,
      "lm_loss": 0.5635,
      "step": 10616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7789649385071122,
      "grad_norm": 0.5453202739907191,
      "learning_rate": 1.227058462363574e-06,
      "loss": 0.4658,
      "step": 10617
    },
    {
      "epoch": 0.7790383080972514,
      "grad_norm": 1.505985250355043,
      "learning_rate": 1.2262788758553512e-06,
      "loss": 0.6864,
      "step": 10618
    },
    {
      "epoch": 0.7791116776873905,
      "grad_norm": 1.4763928662114132,
      "learning_rate": 1.2254995024585064e-06,
      "loss": 0.7117,
      "step": 10619
    },
    {
      "epoch": 0.7791850472775297,
      "grad_norm": 0.547331597272606,
      "learning_rate": 1.2247203422170522e-06,
      "loss": 0.4818,
      "step": 10620
    },
    {
      "epoch": 0.7792584168676687,
      "grad_norm": 1.4291054043536522,
      "learning_rate": 1.223941395174989e-06,
      "loss": 0.7557,
      "step": 10621
    },
    {
      "epoch": 0.7793317864578079,
      "grad_norm": 1.4234382896743647,
      "learning_rate": 1.2231626613763054e-06,
      "loss": 0.747,
      "step": 10622
    },
    {
      "epoch": 0.779405156047947,
      "grad_norm": 1.5184704047749447,
      "learning_rate": 1.2223841408649812e-06,
      "loss": 0.6688,
      "step": 10623
    },
    {
      "epoch": 0.7794785256380862,
      "grad_norm": 1.3382554549648422,
      "learning_rate": 1.221605833684979e-06,
      "loss": 0.6086,
      "step": 10624
    },
    {
      "epoch": 0.7794785256380862,
      "lm_loss": 0.9747,
      "step": 10624,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.7794785256380862,
      "lm_loss": 1.7398,
      "step": 10624,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.7794785256380862,
      "lm_loss": 0.0558,
      "step": 10624,
      "vm_loss": 0.134
    },
    {
      "epoch": 0.7794785256380862,
      "lm_loss": 0.8,
      "step": 10624,
      "vm_loss": 0.0952
    },
    {
      "epoch": 0.7794785256380862,
      "lm_loss": 0.6247,
      "step": 10624,
      "vm_loss": 0.1435
    },
    {
      "epoch": 0.7794785256380862,
      "lm_loss": 0.9693,
      "step": 10624,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.7794785256380862,
      "lm_loss": 1.0859,
      "step": 10624,
      "vm_loss": 0.068
    },
    {
      "epoch": 0.7794785256380862,
      "lm_loss": 0.6911,
      "step": 10624,
      "vm_loss": 0.1864
    },
    {
      "epoch": 0.7795518952282253,
      "grad_norm": 1.4937094414003695,
      "learning_rate": 1.2208277398802531e-06,
      "loss": 0.8314,
      "step": 10625
    },
    {
      "epoch": 0.7796252648183644,
      "grad_norm": 1.3906975934962482,
      "learning_rate": 1.2200498594947436e-06,
      "loss": 0.7482,
      "step": 10626
    },
    {
      "epoch": 0.7796986344085035,
      "grad_norm": 1.4194441706271568,
      "learning_rate": 1.2192721925723773e-06,
      "loss": 0.785,
      "step": 10627
    },
    {
      "epoch": 0.7797720039986427,
      "grad_norm": 1.5112916579663978,
      "learning_rate": 1.218494739157075e-06,
      "loss": 0.8268,
      "step": 10628
    },
    {
      "epoch": 0.7798453735887818,
      "grad_norm": 1.5169674324001634,
      "learning_rate": 1.21771749929274e-06,
      "loss": 0.8105,
      "step": 10629
    },
    {
      "epoch": 0.779918743178921,
      "grad_norm": 0.5547526168553638,
      "learning_rate": 1.2169404730232637e-06,
      "loss": 0.4757,
      "step": 10630
    },
    {
      "epoch": 0.77999211276906,
      "grad_norm": 0.5388209335521228,
      "learning_rate": 1.2161636603925276e-06,
      "loss": 0.4745,
      "step": 10631
    },
    {
      "epoch": 0.7800654823591991,
      "grad_norm": 1.4235137230089085,
      "learning_rate": 1.2153870614444003e-06,
      "loss": 0.7054,
      "step": 10632
    },
    {
      "epoch": 0.7800654823591991,
      "lm_loss": 1.0523,
      "step": 10632,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.7800654823591991,
      "lm_loss": 0.233,
      "step": 10632,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.7800654823591991,
      "lm_loss": 0.9453,
      "step": 10632,
      "vm_loss": 0.1008
    },
    {
      "epoch": 0.7800654823591991,
      "lm_loss": 0.5145,
      "step": 10632,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.7800654823591991,
      "lm_loss": 0.067,
      "step": 10632,
      "vm_loss": 0.074
    },
    {
      "epoch": 0.7800654823591991,
      "lm_loss": 0.0494,
      "step": 10632,
      "vm_loss": 0.0965
    },
    {
      "epoch": 0.7800654823591991,
      "lm_loss": 0.0866,
      "step": 10632,
      "vm_loss": 0.0681
    },
    {
      "epoch": 0.7800654823591991,
      "lm_loss": 0.5809,
      "step": 10632,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.7801388519493383,
      "grad_norm": 1.3392938521160127,
      "learning_rate": 1.2146106762227366e-06,
      "loss": 0.659,
      "step": 10633
    },
    {
      "epoch": 0.7802122215394774,
      "grad_norm": 1.574685604553896,
      "learning_rate": 1.213834504771384e-06,
      "loss": 0.7536,
      "step": 10634
    },
    {
      "epoch": 0.7802855911296166,
      "grad_norm": 0.552204204869422,
      "learning_rate": 1.2130585471341727e-06,
      "loss": 0.457,
      "step": 10635
    },
    {
      "epoch": 0.7803589607197556,
      "grad_norm": 0.5589589500196414,
      "learning_rate": 1.212282803354924e-06,
      "loss": 0.4797,
      "step": 10636
    },
    {
      "epoch": 0.7804323303098948,
      "grad_norm": 1.7111592179664468,
      "learning_rate": 1.2115072734774451e-06,
      "loss": 0.6563,
      "step": 10637
    },
    {
      "epoch": 0.7805056999000339,
      "grad_norm": 0.560920819052624,
      "learning_rate": 1.2107319575455317e-06,
      "loss": 0.4798,
      "step": 10638
    },
    {
      "epoch": 0.7805790694901731,
      "grad_norm": 0.5352776082969918,
      "learning_rate": 1.2099568556029694e-06,
      "loss": 0.471,
      "step": 10639
    },
    {
      "epoch": 0.7806524390803122,
      "grad_norm": 0.5413116938648606,
      "learning_rate": 1.2091819676935295e-06,
      "loss": 0.4818,
      "step": 10640
    },
    {
      "epoch": 0.7806524390803122,
      "lm_loss": 0.9494,
      "step": 10640,
      "vm_loss": 0.0701
    },
    {
      "epoch": 0.7806524390803122,
      "lm_loss": 1.1682,
      "step": 10640,
      "vm_loss": 0.0412
    },
    {
      "epoch": 0.7806524390803122,
      "lm_loss": 0.0999,
      "step": 10640,
      "vm_loss": 0.1069
    },
    {
      "epoch": 0.7806524390803122,
      "lm_loss": 0.8112,
      "step": 10640,
      "vm_loss": 0.0826
    },
    {
      "epoch": 0.7806524390803122,
      "lm_loss": 0.35,
      "step": 10640,
      "vm_loss": 0.1565
    },
    {
      "epoch": 0.7806524390803122,
      "lm_loss": 0.0644,
      "step": 10640,
      "vm_loss": 0.1811
    },
    {
      "epoch": 0.7806524390803122,
      "lm_loss": 0.3551,
      "step": 10640,
      "vm_loss": 0.1298
    },
    {
      "epoch": 0.7806524390803122,
      "lm_loss": 0.4539,
      "step": 10640,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.7807258086704513,
      "grad_norm": 1.4742531613224517,
      "learning_rate": 1.208407293860972e-06,
      "loss": 0.7609,
      "step": 10641
    },
    {
      "epoch": 0.7807991782605904,
      "grad_norm": 0.5493920183924995,
      "learning_rate": 1.207632834149044e-06,
      "loss": 0.4715,
      "step": 10642
    },
    {
      "epoch": 0.7808725478507296,
      "grad_norm": 1.4568918848349874,
      "learning_rate": 1.20685858860148e-06,
      "loss": 0.6588,
      "step": 10643
    },
    {
      "epoch": 0.7809459174408687,
      "grad_norm": 0.5430226305603313,
      "learning_rate": 1.2060845572620067e-06,
      "loss": 0.4725,
      "step": 10644
    },
    {
      "epoch": 0.7810192870310079,
      "grad_norm": 0.5455496195717455,
      "learning_rate": 1.2053107401743341e-06,
      "loss": 0.4731,
      "step": 10645
    },
    {
      "epoch": 0.7810926566211469,
      "grad_norm": 1.4288735773964931,
      "learning_rate": 1.2045371373821614e-06,
      "loss": 0.7551,
      "step": 10646
    },
    {
      "epoch": 0.7811660262112861,
      "grad_norm": 1.5001037902886873,
      "learning_rate": 1.2037637489291753e-06,
      "loss": 0.718,
      "step": 10647
    },
    {
      "epoch": 0.7812393958014252,
      "grad_norm": 1.5194988952563815,
      "learning_rate": 1.202990574859052e-06,
      "loss": 0.789,
      "step": 10648
    },
    {
      "epoch": 0.7812393958014252,
      "lm_loss": 1.5393,
      "step": 10648,
      "vm_loss": 0.0651
    },
    {
      "epoch": 0.7812393958014252,
      "lm_loss": 0.9694,
      "step": 10648,
      "vm_loss": 0.1838
    },
    {
      "epoch": 0.7812393958014252,
      "lm_loss": 0.0605,
      "step": 10648,
      "vm_loss": 0.101
    },
    {
      "epoch": 0.7812393958014252,
      "lm_loss": 0.8359,
      "step": 10648,
      "vm_loss": 0.1103
    },
    {
      "epoch": 0.7812393958014252,
      "lm_loss": 0.8268,
      "step": 10648,
      "vm_loss": 0.1465
    },
    {
      "epoch": 0.7812393958014252,
      "lm_loss": 0.7742,
      "step": 10648,
      "vm_loss": 0.2212
    },
    {
      "epoch": 0.7812393958014252,
      "lm_loss": 2.0545,
      "step": 10648,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.7812393958014252,
      "lm_loss": 0.1959,
      "step": 10648,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.7813127653915644,
      "grad_norm": 1.4297577817578624,
      "learning_rate": 1.2022176152154525e-06,
      "loss": 0.7763,
      "step": 10649
    },
    {
      "epoch": 0.7813861349817035,
      "grad_norm": 1.5007468511236868,
      "learning_rate": 1.2014448700420306e-06,
      "loss": 0.773,
      "step": 10650
    },
    {
      "epoch": 0.7814595045718425,
      "grad_norm": 1.5999092144918619,
      "learning_rate": 1.2006723393824239e-06,
      "loss": 0.6925,
      "step": 10651
    },
    {
      "epoch": 0.7815328741619817,
      "grad_norm": 1.4563774771063493,
      "learning_rate": 1.1999000232802594e-06,
      "loss": 0.6892,
      "step": 10652
    },
    {
      "epoch": 0.7816062437521208,
      "grad_norm": 1.575015055613423,
      "learning_rate": 1.1991279217791507e-06,
      "loss": 0.8848,
      "step": 10653
    },
    {
      "epoch": 0.78167961334226,
      "grad_norm": 1.5820174168490297,
      "learning_rate": 1.1983560349227003e-06,
      "loss": 0.7692,
      "step": 10654
    },
    {
      "epoch": 0.7817529829323991,
      "grad_norm": 0.5431665892197055,
      "learning_rate": 1.1975843627545003e-06,
      "loss": 0.4749,
      "step": 10655
    },
    {
      "epoch": 0.7818263525225382,
      "grad_norm": 1.445686147159843,
      "learning_rate": 1.1968129053181276e-06,
      "loss": 0.6479,
      "step": 10656
    },
    {
      "epoch": 0.7818263525225382,
      "lm_loss": 1.6634,
      "step": 10656,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.7818263525225382,
      "lm_loss": 0.2089,
      "step": 10656,
      "vm_loss": 0.0634
    },
    {
      "epoch": 0.7818263525225382,
      "lm_loss": 0.7622,
      "step": 10656,
      "vm_loss": 0.2142
    },
    {
      "epoch": 0.7818263525225382,
      "lm_loss": 0.6353,
      "step": 10656,
      "vm_loss": 0.1051
    },
    {
      "epoch": 0.7818263525225382,
      "lm_loss": 1.1844,
      "step": 10656,
      "vm_loss": 0.2521
    },
    {
      "epoch": 0.7818263525225382,
      "lm_loss": 0.7003,
      "step": 10656,
      "vm_loss": 0.1258
    },
    {
      "epoch": 0.7818263525225382,
      "lm_loss": 0.7623,
      "step": 10656,
      "vm_loss": 0.1066
    },
    {
      "epoch": 0.7818263525225382,
      "lm_loss": 0.855,
      "step": 10656,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.7818997221126773,
      "grad_norm": 1.5117619841203316,
      "learning_rate": 1.1960416626571491e-06,
      "loss": 0.7796,
      "step": 10657
    },
    {
      "epoch": 0.7819730917028165,
      "grad_norm": 0.550788245554648,
      "learning_rate": 1.1952706348151178e-06,
      "loss": 0.4835,
      "step": 10658
    },
    {
      "epoch": 0.7820464612929556,
      "grad_norm": 0.5551797426129401,
      "learning_rate": 1.1944998218355762e-06,
      "loss": 0.4927,
      "step": 10659
    },
    {
      "epoch": 0.7821198308830948,
      "grad_norm": 0.576768827186557,
      "learning_rate": 1.1937292237620524e-06,
      "loss": 0.4929,
      "step": 10660
    },
    {
      "epoch": 0.7821932004732338,
      "grad_norm": 1.4250179564075551,
      "learning_rate": 1.1929588406380666e-06,
      "loss": 0.6799,
      "step": 10661
    },
    {
      "epoch": 0.782266570063373,
      "grad_norm": 1.5281362222662764,
      "learning_rate": 1.1921886725071235e-06,
      "loss": 0.7035,
      "step": 10662
    },
    {
      "epoch": 0.7823399396535121,
      "grad_norm": 1.4363448433896486,
      "learning_rate": 1.1914187194127157e-06,
      "loss": 0.6936,
      "step": 10663
    },
    {
      "epoch": 0.7824133092436513,
      "grad_norm": 1.6955252115172337,
      "learning_rate": 1.1906489813983246e-06,
      "loss": 0.7604,
      "step": 10664
    },
    {
      "epoch": 0.7824133092436513,
      "lm_loss": 1.2264,
      "step": 10664,
      "vm_loss": 0.0724
    },
    {
      "epoch": 0.7824133092436513,
      "lm_loss": 0.1378,
      "step": 10664,
      "vm_loss": 0.0524
    },
    {
      "epoch": 0.7824133092436513,
      "lm_loss": 0.7664,
      "step": 10664,
      "vm_loss": 0.1276
    },
    {
      "epoch": 0.7824133092436513,
      "lm_loss": 0.4854,
      "step": 10664,
      "vm_loss": 0.155
    },
    {
      "epoch": 0.7824133092436513,
      "lm_loss": 0.5734,
      "step": 10664,
      "vm_loss": 0.1778
    },
    {
      "epoch": 0.7824133092436513,
      "lm_loss": 0.9037,
      "step": 10664,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.7824133092436513,
      "lm_loss": 0.2251,
      "step": 10664,
      "vm_loss": 0.107
    },
    {
      "epoch": 0.7824133092436513,
      "lm_loss": 1.1971,
      "step": 10664,
      "vm_loss": 0.1812
    },
    {
      "epoch": 0.7824866788337904,
      "grad_norm": 1.6462828166517784,
      "learning_rate": 1.1898794585074181e-06,
      "loss": 0.7889,
      "step": 10665
    },
    {
      "epoch": 0.7825600484239295,
      "grad_norm": 0.5484310800313936,
      "learning_rate": 1.1891101507834552e-06,
      "loss": 0.489,
      "step": 10666
    },
    {
      "epoch": 0.7826334180140686,
      "grad_norm": 1.6744215929941202,
      "learning_rate": 1.1883410582698795e-06,
      "loss": 0.7545,
      "step": 10667
    },
    {
      "epoch": 0.7827067876042078,
      "grad_norm": 1.540183192901431,
      "learning_rate": 1.1875721810101238e-06,
      "loss": 0.734,
      "step": 10668
    },
    {
      "epoch": 0.7827801571943469,
      "grad_norm": 1.5544327029195577,
      "learning_rate": 1.1868035190476085e-06,
      "loss": 0.7469,
      "step": 10669
    },
    {
      "epoch": 0.782853526784486,
      "grad_norm": 1.3166837851275923,
      "learning_rate": 1.1860350724257403e-06,
      "loss": 0.656,
      "step": 10670
    },
    {
      "epoch": 0.7829268963746251,
      "grad_norm": 0.5567953535603266,
      "learning_rate": 1.185266841187918e-06,
      "loss": 0.4897,
      "step": 10671
    },
    {
      "epoch": 0.7830002659647642,
      "grad_norm": 1.567011121867884,
      "learning_rate": 1.1844988253775242e-06,
      "loss": 0.7998,
      "step": 10672
    },
    {
      "epoch": 0.7830002659647642,
      "lm_loss": 0.3444,
      "step": 10672,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.7830002659647642,
      "lm_loss": 1.0608,
      "step": 10672,
      "vm_loss": 0.1236
    },
    {
      "epoch": 0.7830002659647642,
      "lm_loss": 0.3088,
      "step": 10672,
      "vm_loss": 0.0857
    },
    {
      "epoch": 0.7830002659647642,
      "lm_loss": 0.0134,
      "step": 10672,
      "vm_loss": 0.1629
    },
    {
      "epoch": 0.7830002659647642,
      "lm_loss": 0.6838,
      "step": 10672,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.7830002659647642,
      "lm_loss": 0.8325,
      "step": 10672,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.7830002659647642,
      "lm_loss": 0.0068,
      "step": 10672,
      "vm_loss": 0.0706
    },
    {
      "epoch": 0.7830002659647642,
      "lm_loss": 0.0966,
      "step": 10672,
      "vm_loss": 0.1909
    },
    {
      "epoch": 0.7830736355549034,
      "grad_norm": 1.4808218745203627,
      "learning_rate": 1.1837310250379308e-06,
      "loss": 0.6546,
      "step": 10673
    },
    {
      "epoch": 0.7831470051450425,
      "grad_norm": 0.5514829310933587,
      "learning_rate": 1.1829634402124974e-06,
      "loss": 0.4864,
      "step": 10674
    },
    {
      "epoch": 0.7832203747351817,
      "grad_norm": 2.2500485747620926,
      "learning_rate": 1.1821960709445707e-06,
      "loss": 0.7149,
      "step": 10675
    },
    {
      "epoch": 0.7832937443253207,
      "grad_norm": 1.4038770823172464,
      "learning_rate": 1.181428917277485e-06,
      "loss": 0.7283,
      "step": 10676
    },
    {
      "epoch": 0.7833671139154599,
      "grad_norm": 1.5577317871343852,
      "learning_rate": 1.180661979254567e-06,
      "loss": 0.681,
      "step": 10677
    },
    {
      "epoch": 0.783440483505599,
      "grad_norm": 0.5604732721182617,
      "learning_rate": 1.1798952569191253e-06,
      "loss": 0.4947,
      "step": 10678
    },
    {
      "epoch": 0.7835138530957382,
      "grad_norm": 1.5634464808341917,
      "learning_rate": 1.1791287503144582e-06,
      "loss": 0.7402,
      "step": 10679
    },
    {
      "epoch": 0.7835872226858772,
      "grad_norm": 0.5446184400274529,
      "learning_rate": 1.1783624594838533e-06,
      "loss": 0.4727,
      "step": 10680
    },
    {
      "epoch": 0.7835872226858772,
      "lm_loss": 1.4908,
      "step": 10680,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.7835872226858772,
      "lm_loss": 0.7074,
      "step": 10680,
      "vm_loss": 0.0851
    },
    {
      "epoch": 0.7835872226858772,
      "lm_loss": 0.0769,
      "step": 10680,
      "vm_loss": 0.078
    },
    {
      "epoch": 0.7835872226858772,
      "lm_loss": 0.5062,
      "step": 10680,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.7835872226858772,
      "lm_loss": 0.9322,
      "step": 10680,
      "vm_loss": 0.1522
    },
    {
      "epoch": 0.7835872226858772,
      "lm_loss": 0.0431,
      "step": 10680,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.7835872226858772,
      "lm_loss": 0.1165,
      "step": 10680,
      "vm_loss": 0.0658
    },
    {
      "epoch": 0.7835872226858772,
      "lm_loss": 0.0872,
      "step": 10680,
      "vm_loss": 0.0666
    },
    {
      "epoch": 0.7836605922760164,
      "grad_norm": 1.481468246023687,
      "learning_rate": 1.1775963844705829e-06,
      "loss": 0.5997,
      "step": 10681
    },
    {
      "epoch": 0.7837339618661555,
      "grad_norm": 1.5455551059848383,
      "learning_rate": 1.1768305253179124e-06,
      "loss": 0.7393,
      "step": 10682
    },
    {
      "epoch": 0.7838073314562947,
      "grad_norm": 1.5405459612351187,
      "learning_rate": 1.1760648820690894e-06,
      "loss": 0.7544,
      "step": 10683
    },
    {
      "epoch": 0.7838807010464338,
      "grad_norm": 1.5873217558494581,
      "learning_rate": 1.175299454767353e-06,
      "loss": 0.7254,
      "step": 10684
    },
    {
      "epoch": 0.7839540706365729,
      "grad_norm": 1.6455579911096623,
      "learning_rate": 1.1745342434559275e-06,
      "loss": 0.7666,
      "step": 10685
    },
    {
      "epoch": 0.784027440226712,
      "grad_norm": 1.492116700575627,
      "learning_rate": 1.1737692481780278e-06,
      "loss": 0.7874,
      "step": 10686
    },
    {
      "epoch": 0.7841008098168512,
      "grad_norm": 1.8799354513027045,
      "learning_rate": 1.1730044689768517e-06,
      "loss": 0.7415,
      "step": 10687
    },
    {
      "epoch": 0.7841741794069903,
      "grad_norm": 1.6355473666479248,
      "learning_rate": 1.1722399058955925e-06,
      "loss": 0.7762,
      "step": 10688
    },
    {
      "epoch": 0.7841741794069903,
      "lm_loss": 1.4398,
      "step": 10688,
      "vm_loss": 0.0382
    },
    {
      "epoch": 0.7841741794069903,
      "lm_loss": 0.108,
      "step": 10688,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.7841741794069903,
      "lm_loss": 0.0577,
      "step": 10688,
      "vm_loss": 0.1559
    },
    {
      "epoch": 0.7841741794069903,
      "lm_loss": 0.1383,
      "step": 10688,
      "vm_loss": 0.1813
    },
    {
      "epoch": 0.7841741794069903,
      "lm_loss": 0.4545,
      "step": 10688,
      "vm_loss": 0.1786
    },
    {
      "epoch": 0.7841741794069903,
      "lm_loss": 0.0588,
      "step": 10688,
      "vm_loss": 0.1041
    },
    {
      "epoch": 0.7841741794069903,
      "lm_loss": 0.9588,
      "step": 10688,
      "vm_loss": 0.1794
    },
    {
      "epoch": 0.7841741794069903,
      "lm_loss": 0.7224,
      "step": 10688,
      "vm_loss": 0.1158
    },
    {
      "epoch": 0.7842475489971295,
      "grad_norm": 1.7323460906904051,
      "learning_rate": 1.1714755589774252e-06,
      "loss": 0.7526,
      "step": 10689
    },
    {
      "epoch": 0.7843209185872685,
      "grad_norm": 1.4999038323439944,
      "learning_rate": 1.1707114282655141e-06,
      "loss": 0.727,
      "step": 10690
    },
    {
      "epoch": 0.7843942881774076,
      "grad_norm": 1.6533556368941524,
      "learning_rate": 1.169947513803012e-06,
      "loss": 0.8017,
      "step": 10691
    },
    {
      "epoch": 0.7844676577675468,
      "grad_norm": 1.499139770348275,
      "learning_rate": 1.169183815633057e-06,
      "loss": 0.8231,
      "step": 10692
    },
    {
      "epoch": 0.7845410273576859,
      "grad_norm": 1.542609727300506,
      "learning_rate": 1.1684203337987803e-06,
      "loss": 0.7769,
      "step": 10693
    },
    {
      "epoch": 0.7846143969478251,
      "grad_norm": 1.547319189755097,
      "learning_rate": 1.1676570683432959e-06,
      "loss": 0.8125,
      "step": 10694
    },
    {
      "epoch": 0.7846877665379641,
      "grad_norm": 1.5615384223901823,
      "learning_rate": 1.1668940193097079e-06,
      "loss": 0.7524,
      "step": 10695
    },
    {
      "epoch": 0.7847611361281033,
      "grad_norm": 0.5584548193154169,
      "learning_rate": 1.1661311867411068e-06,
      "loss": 0.4806,
      "step": 10696
    },
    {
      "epoch": 0.7847611361281033,
      "lm_loss": 1.5744,
      "step": 10696,
      "vm_loss": 0.0503
    },
    {
      "epoch": 0.7847611361281033,
      "lm_loss": 0.7947,
      "step": 10696,
      "vm_loss": 0.123
    },
    {
      "epoch": 0.7847611361281033,
      "lm_loss": 0.1173,
      "step": 10696,
      "vm_loss": 0.0741
    },
    {
      "epoch": 0.7847611361281033,
      "lm_loss": 1.1386,
      "step": 10696,
      "vm_loss": 0.1988
    },
    {
      "epoch": 0.7847611361281033,
      "lm_loss": 0.5309,
      "step": 10696,
      "vm_loss": 0.1577
    },
    {
      "epoch": 0.7847611361281033,
      "lm_loss": 0.3131,
      "step": 10696,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.7847611361281033,
      "lm_loss": 0.6754,
      "step": 10696,
      "vm_loss": 0.1317
    },
    {
      "epoch": 0.7847611361281033,
      "lm_loss": 0.1532,
      "step": 10696,
      "vm_loss": 0.1256
    },
    {
      "epoch": 0.7848345057182424,
      "grad_norm": 1.469610515188696,
      "learning_rate": 1.1653685706805707e-06,
      "loss": 0.6676,
      "step": 10697
    },
    {
      "epoch": 0.7849078753083816,
      "grad_norm": 0.5774315573883962,
      "learning_rate": 1.1646061711711693e-06,
      "loss": 0.4797,
      "step": 10698
    },
    {
      "epoch": 0.7849812448985207,
      "grad_norm": 1.4186948186761033,
      "learning_rate": 1.1638439882559554e-06,
      "loss": 0.6095,
      "step": 10699
    },
    {
      "epoch": 0.7850546144886598,
      "grad_norm": 1.3880685340943657,
      "learning_rate": 1.163082021977973e-06,
      "loss": 0.7492,
      "step": 10700
    },
    {
      "epoch": 0.7851279840787989,
      "grad_norm": 1.6464808614677686,
      "learning_rate": 1.1623202723802495e-06,
      "loss": 0.6816,
      "step": 10701
    },
    {
      "epoch": 0.7852013536689381,
      "grad_norm": 1.4647464425235133,
      "learning_rate": 1.1615587395058025e-06,
      "loss": 0.8288,
      "step": 10702
    },
    {
      "epoch": 0.7852747232590772,
      "grad_norm": 1.496665001956023,
      "learning_rate": 1.160797423397641e-06,
      "loss": 0.8111,
      "step": 10703
    },
    {
      "epoch": 0.7853480928492164,
      "grad_norm": 1.4868646297936745,
      "learning_rate": 1.1600363240987562e-06,
      "loss": 0.7094,
      "step": 10704
    },
    {
      "epoch": 0.7853480928492164,
      "lm_loss": 1.2554,
      "step": 10704,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.7853480928492164,
      "lm_loss": 0.9745,
      "step": 10704,
      "vm_loss": 0.1929
    },
    {
      "epoch": 0.7853480928492164,
      "lm_loss": 0.7152,
      "step": 10704,
      "vm_loss": 0.1263
    },
    {
      "epoch": 0.7853480928492164,
      "lm_loss": 0.2476,
      "step": 10704,
      "vm_loss": 0.1945
    },
    {
      "epoch": 0.7853480928492164,
      "lm_loss": 0.5597,
      "step": 10704,
      "vm_loss": 0.0743
    },
    {
      "epoch": 0.7853480928492164,
      "lm_loss": 0.2512,
      "step": 10704,
      "vm_loss": 0.0823
    },
    {
      "epoch": 0.7853480928492164,
      "lm_loss": 0.2317,
      "step": 10704,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.7853480928492164,
      "lm_loss": 0.4364,
      "step": 10704,
      "vm_loss": 0.131
    },
    {
      "epoch": 0.7854214624393554,
      "grad_norm": 1.444865233252626,
      "learning_rate": 1.15927544165213e-06,
      "loss": 0.7272,
      "step": 10705
    },
    {
      "epoch": 0.7854948320294945,
      "grad_norm": 0.5421221902661989,
      "learning_rate": 1.1585147761007309e-06,
      "loss": 0.4862,
      "step": 10706
    },
    {
      "epoch": 0.7855682016196337,
      "grad_norm": 1.7019240497833894,
      "learning_rate": 1.1577543274875142e-06,
      "loss": 0.7908,
      "step": 10707
    },
    {
      "epoch": 0.7856415712097728,
      "grad_norm": 1.5593378789963428,
      "learning_rate": 1.156994095855427e-06,
      "loss": 0.8008,
      "step": 10708
    },
    {
      "epoch": 0.785714940799912,
      "grad_norm": 0.5601746796478665,
      "learning_rate": 1.1562340812474004e-06,
      "loss": 0.4595,
      "step": 10709
    },
    {
      "epoch": 0.785788310390051,
      "grad_norm": 1.6250689799656275,
      "learning_rate": 1.1554742837063543e-06,
      "loss": 0.7075,
      "step": 10710
    },
    {
      "epoch": 0.7858616799801902,
      "grad_norm": 1.4430471815611408,
      "learning_rate": 1.154714703275196e-06,
      "loss": 0.7301,
      "step": 10711
    },
    {
      "epoch": 0.7859350495703293,
      "grad_norm": 1.3315685810449551,
      "learning_rate": 1.1539553399968201e-06,
      "loss": 0.7252,
      "step": 10712
    },
    {
      "epoch": 0.7859350495703293,
      "lm_loss": 0.6958,
      "step": 10712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7859350495703293,
      "lm_loss": 0.5571,
      "step": 10712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7859350495703293,
      "lm_loss": 0.4505,
      "step": 10712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7859350495703293,
      "lm_loss": 0.4586,
      "step": 10712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7859350495703293,
      "lm_loss": 0.568,
      "step": 10712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7859350495703293,
      "lm_loss": 0.4814,
      "step": 10712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7859350495703293,
      "lm_loss": 0.5207,
      "step": 10712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7859350495703293,
      "lm_loss": 0.411,
      "step": 10712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7860084191604685,
      "grad_norm": 0.5610064267682933,
      "learning_rate": 1.1531961939141117e-06,
      "loss": 0.4819,
      "step": 10713
    },
    {
      "epoch": 0.7860817887506076,
      "grad_norm": 0.5488316195325208,
      "learning_rate": 1.1524372650699405e-06,
      "loss": 0.4622,
      "step": 10714
    },
    {
      "epoch": 0.7861551583407467,
      "grad_norm": 1.5652133686185552,
      "learning_rate": 1.1516785535071656e-06,
      "loss": 0.7525,
      "step": 10715
    },
    {
      "epoch": 0.7862285279308858,
      "grad_norm": 1.4458521876012735,
      "learning_rate": 1.1509200592686332e-06,
      "loss": 0.8565,
      "step": 10716
    },
    {
      "epoch": 0.786301897521025,
      "grad_norm": 1.5869971958916145,
      "learning_rate": 1.1501617823971766e-06,
      "loss": 0.6734,
      "step": 10717
    },
    {
      "epoch": 0.7863752671111641,
      "grad_norm": 1.5260693468548236,
      "learning_rate": 1.149403722935617e-06,
      "loss": 0.8191,
      "step": 10718
    },
    {
      "epoch": 0.7864486367013033,
      "grad_norm": 1.4643593907025985,
      "learning_rate": 1.1486458809267664e-06,
      "loss": 0.712,
      "step": 10719
    },
    {
      "epoch": 0.7865220062914423,
      "grad_norm": 1.3663335156976142,
      "learning_rate": 1.147888256413421e-06,
      "loss": 0.6731,
      "step": 10720
    },
    {
      "epoch": 0.7865220062914423,
      "lm_loss": 1.4993,
      "step": 10720,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.7865220062914423,
      "lm_loss": 0.9821,
      "step": 10720,
      "vm_loss": 0.0743
    },
    {
      "epoch": 0.7865220062914423,
      "lm_loss": 1.2939,
      "step": 10720,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.7865220062914423,
      "lm_loss": 0.751,
      "step": 10720,
      "vm_loss": 0.215
    },
    {
      "epoch": 0.7865220062914423,
      "lm_loss": 1.09,
      "step": 10720,
      "vm_loss": 0.2208
    },
    {
      "epoch": 0.7865220062914423,
      "lm_loss": 0.8146,
      "step": 10720,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.7865220062914423,
      "lm_loss": 1.1325,
      "step": 10720,
      "vm_loss": 0.3062
    },
    {
      "epoch": 0.7865220062914423,
      "lm_loss": 1.7791,
      "step": 10720,
      "vm_loss": 0.1202
    },
    {
      "epoch": 0.7865953758815815,
      "grad_norm": 1.6009979761863231,
      "learning_rate": 1.1471308494383647e-06,
      "loss": 0.8296,
      "step": 10721
    },
    {
      "epoch": 0.7866687454717206,
      "grad_norm": 1.448603979687984,
      "learning_rate": 1.1463736600443715e-06,
      "loss": 0.6854,
      "step": 10722
    },
    {
      "epoch": 0.7867421150618598,
      "grad_norm": 0.578076605198988,
      "learning_rate": 1.1456166882741992e-06,
      "loss": 0.4808,
      "step": 10723
    },
    {
      "epoch": 0.7868154846519989,
      "grad_norm": 1.472320411063,
      "learning_rate": 1.144859934170599e-06,
      "loss": 0.579,
      "step": 10724
    },
    {
      "epoch": 0.786888854242138,
      "grad_norm": 1.3795144169659672,
      "learning_rate": 1.1441033977763056e-06,
      "loss": 0.6397,
      "step": 10725
    },
    {
      "epoch": 0.7869622238322771,
      "grad_norm": 1.7257665216947469,
      "learning_rate": 1.1433470791340422e-06,
      "loss": 0.87,
      "step": 10726
    },
    {
      "epoch": 0.7870355934224162,
      "grad_norm": 1.59462805440011,
      "learning_rate": 1.1425909782865203e-06,
      "loss": 0.7944,
      "step": 10727
    },
    {
      "epoch": 0.7871089630125554,
      "grad_norm": 1.9261488454849136,
      "learning_rate": 1.1418350952764385e-06,
      "loss": 0.8319,
      "step": 10728
    },
    {
      "epoch": 0.7871089630125554,
      "lm_loss": 1.4455,
      "step": 10728,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.7871089630125554,
      "lm_loss": 1.3412,
      "step": 10728,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.7871089630125554,
      "lm_loss": 0.0017,
      "step": 10728,
      "vm_loss": 0.0418
    },
    {
      "epoch": 0.7871089630125554,
      "lm_loss": 0.3156,
      "step": 10728,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.7871089630125554,
      "lm_loss": 0.0074,
      "step": 10728,
      "vm_loss": 0.0616
    },
    {
      "epoch": 0.7871089630125554,
      "lm_loss": 0.5804,
      "step": 10728,
      "vm_loss": 0.0971
    },
    {
      "epoch": 0.7871089630125554,
      "lm_loss": 0.5482,
      "step": 10728,
      "vm_loss": 0.1315
    },
    {
      "epoch": 0.7871089630125554,
      "lm_loss": 0.1899,
      "step": 10728,
      "vm_loss": 0.138
    },
    {
      "epoch": 0.7871823326026945,
      "grad_norm": 1.3902466266559583,
      "learning_rate": 1.1410794301464817e-06,
      "loss": 0.7133,
      "step": 10729
    },
    {
      "epoch": 0.7872557021928336,
      "grad_norm": 1.7104476116661764,
      "learning_rate": 1.140323982939328e-06,
      "loss": 0.785,
      "step": 10730
    },
    {
      "epoch": 0.7873290717829727,
      "grad_norm": 0.5549020255291167,
      "learning_rate": 1.1395687536976368e-06,
      "loss": 0.4659,
      "step": 10731
    },
    {
      "epoch": 0.7874024413731119,
      "grad_norm": 0.5679987538203151,
      "learning_rate": 1.1388137424640583e-06,
      "loss": 0.4981,
      "step": 10732
    },
    {
      "epoch": 0.787475810963251,
      "grad_norm": 0.5681541706384045,
      "learning_rate": 1.13805894928123e-06,
      "loss": 0.4538,
      "step": 10733
    },
    {
      "epoch": 0.7875491805533902,
      "grad_norm": 1.512149529209804,
      "learning_rate": 1.1373043741917755e-06,
      "loss": 0.6974,
      "step": 10734
    },
    {
      "epoch": 0.7876225501435292,
      "grad_norm": 0.5595736919711142,
      "learning_rate": 1.1365500172383099e-06,
      "loss": 0.4894,
      "step": 10735
    },
    {
      "epoch": 0.7876959197336684,
      "grad_norm": 0.5438511667872642,
      "learning_rate": 1.1357958784634326e-06,
      "loss": 0.4802,
      "step": 10736
    },
    {
      "epoch": 0.7876959197336684,
      "lm_loss": 0.451,
      "step": 10736,
      "vm_loss": 0.0549
    },
    {
      "epoch": 0.7876959197336684,
      "lm_loss": 0.7592,
      "step": 10736,
      "vm_loss": 0.1743
    },
    {
      "epoch": 0.7876959197336684,
      "lm_loss": 0.3832,
      "step": 10736,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.7876959197336684,
      "lm_loss": 0.2509,
      "step": 10736,
      "vm_loss": 0.1748
    },
    {
      "epoch": 0.7876959197336684,
      "lm_loss": 0.2499,
      "step": 10736,
      "vm_loss": 0.0946
    },
    {
      "epoch": 0.7876959197336684,
      "lm_loss": 0.7971,
      "step": 10736,
      "vm_loss": 0.1547
    },
    {
      "epoch": 0.7876959197336684,
      "lm_loss": 0.9702,
      "step": 10736,
      "vm_loss": 0.1383
    },
    {
      "epoch": 0.7876959197336684,
      "lm_loss": 0.3366,
      "step": 10736,
      "vm_loss": 0.1439
    },
    {
      "epoch": 0.7877692893238075,
      "grad_norm": 1.395210735322097,
      "learning_rate": 1.1350419579097316e-06,
      "loss": 0.6795,
      "step": 10737
    },
    {
      "epoch": 0.7878426589139467,
      "grad_norm": 1.555691969173257,
      "learning_rate": 1.1342882556197826e-06,
      "loss": 0.801,
      "step": 10738
    },
    {
      "epoch": 0.7879160285040858,
      "grad_norm": 1.3802529289066066,
      "learning_rate": 1.1335347716361479e-06,
      "loss": 0.6797,
      "step": 10739
    },
    {
      "epoch": 0.7879893980942249,
      "grad_norm": 0.5582752200607831,
      "learning_rate": 1.132781506001381e-06,
      "loss": 0.4818,
      "step": 10740
    },
    {
      "epoch": 0.788062767684364,
      "grad_norm": 1.465455190018778,
      "learning_rate": 1.1320284587580193e-06,
      "loss": 0.6888,
      "step": 10741
    },
    {
      "epoch": 0.7881361372745032,
      "grad_norm": 0.5489672750276045,
      "learning_rate": 1.1312756299485893e-06,
      "loss": 0.4782,
      "step": 10742
    },
    {
      "epoch": 0.7882095068646423,
      "grad_norm": 1.5167433950160458,
      "learning_rate": 1.1305230196156057e-06,
      "loss": 0.72,
      "step": 10743
    },
    {
      "epoch": 0.7882828764547815,
      "grad_norm": 1.38702037083849,
      "learning_rate": 1.1297706278015695e-06,
      "loss": 0.7442,
      "step": 10744
    },
    {
      "epoch": 0.7882828764547815,
      "lm_loss": 0.4739,
      "step": 10744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7882828764547815,
      "lm_loss": 0.3611,
      "step": 10744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7882828764547815,
      "lm_loss": 0.5538,
      "step": 10744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7882828764547815,
      "lm_loss": 0.3735,
      "step": 10744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7882828764547815,
      "lm_loss": 0.3254,
      "step": 10744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7882828764547815,
      "lm_loss": 0.484,
      "step": 10744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7882828764547815,
      "lm_loss": 0.4136,
      "step": 10744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7882828764547815,
      "lm_loss": 0.6033,
      "step": 10744,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7883562460449205,
      "grad_norm": 0.5173357466287413,
      "learning_rate": 1.129018454548969e-06,
      "loss": 0.4477,
      "step": 10745
    },
    {
      "epoch": 0.7884296156350596,
      "grad_norm": 1.5352207375396187,
      "learning_rate": 1.1282664999002846e-06,
      "loss": 0.7402,
      "step": 10746
    },
    {
      "epoch": 0.7885029852251988,
      "grad_norm": 1.51410193418896,
      "learning_rate": 1.1275147638979783e-06,
      "loss": 0.7317,
      "step": 10747
    },
    {
      "epoch": 0.7885763548153379,
      "grad_norm": 1.433719186204663,
      "learning_rate": 1.126763246584504e-06,
      "loss": 0.6774,
      "step": 10748
    },
    {
      "epoch": 0.7886497244054771,
      "grad_norm": 1.6134820778748071,
      "learning_rate": 1.1260119480023008e-06,
      "loss": 0.6992,
      "step": 10749
    },
    {
      "epoch": 0.7887230939956161,
      "grad_norm": 1.313089427373861,
      "learning_rate": 1.1252608681937954e-06,
      "loss": 0.6668,
      "step": 10750
    },
    {
      "epoch": 0.7887964635857553,
      "grad_norm": 1.466943528322339,
      "learning_rate": 1.1245100072014054e-06,
      "loss": 0.7207,
      "step": 10751
    },
    {
      "epoch": 0.7888698331758944,
      "grad_norm": 1.4387491408300155,
      "learning_rate": 1.1237593650675333e-06,
      "loss": 0.6632,
      "step": 10752
    },
    {
      "epoch": 0.7888698331758944,
      "lm_loss": 1.5277,
      "step": 10752,
      "vm_loss": 0.0413
    },
    {
      "epoch": 0.7888698331758944,
      "lm_loss": 0.3435,
      "step": 10752,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.7888698331758944,
      "lm_loss": 0.7858,
      "step": 10752,
      "vm_loss": 0.1271
    },
    {
      "epoch": 0.7888698331758944,
      "lm_loss": 0.4095,
      "step": 10752,
      "vm_loss": 0.1099
    },
    {
      "epoch": 0.7888698331758944,
      "lm_loss": 0.983,
      "step": 10752,
      "vm_loss": 0.1455
    },
    {
      "epoch": 0.7888698331758944,
      "lm_loss": 0.5754,
      "step": 10752,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.7888698331758944,
      "lm_loss": 0.0841,
      "step": 10752,
      "vm_loss": 0.0639
    },
    {
      "epoch": 0.7888698331758944,
      "lm_loss": 0.7581,
      "step": 10752,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.7889432027660336,
      "grad_norm": 1.4431265920117597,
      "learning_rate": 1.123008941834569e-06,
      "loss": 0.6525,
      "step": 10753
    },
    {
      "epoch": 0.7890165723561727,
      "grad_norm": 1.5087021099578892,
      "learning_rate": 1.1222587375448908e-06,
      "loss": 0.6981,
      "step": 10754
    },
    {
      "epoch": 0.7890899419463118,
      "grad_norm": 1.5575767630818074,
      "learning_rate": 1.1215087522408646e-06,
      "loss": 0.7943,
      "step": 10755
    },
    {
      "epoch": 0.7891633115364509,
      "grad_norm": 1.5651649461436097,
      "learning_rate": 1.1207589859648433e-06,
      "loss": 0.6857,
      "step": 10756
    },
    {
      "epoch": 0.7892366811265901,
      "grad_norm": 1.411235661972426,
      "learning_rate": 1.1200094387591698e-06,
      "loss": 0.6277,
      "step": 10757
    },
    {
      "epoch": 0.7893100507167292,
      "grad_norm": 1.5942073705408506,
      "learning_rate": 1.119260110666172e-06,
      "loss": 0.7294,
      "step": 10758
    },
    {
      "epoch": 0.7893834203068683,
      "grad_norm": 0.5365521612610994,
      "learning_rate": 1.1185110017281664e-06,
      "loss": 0.4705,
      "step": 10759
    },
    {
      "epoch": 0.7894567898970074,
      "grad_norm": 0.5288725933959322,
      "learning_rate": 1.1177621119874566e-06,
      "loss": 0.4828,
      "step": 10760
    },
    {
      "epoch": 0.7894567898970074,
      "lm_loss": 0.844,
      "step": 10760,
      "vm_loss": 0.0732
    },
    {
      "epoch": 0.7894567898970074,
      "lm_loss": 1.6769,
      "step": 10760,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.7894567898970074,
      "lm_loss": 0.1196,
      "step": 10760,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.7894567898970074,
      "lm_loss": 0.1744,
      "step": 10760,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.7894567898970074,
      "lm_loss": 0.9523,
      "step": 10760,
      "vm_loss": 0.2296
    },
    {
      "epoch": 0.7894567898970074,
      "lm_loss": 0.6285,
      "step": 10760,
      "vm_loss": 0.1335
    },
    {
      "epoch": 0.7894567898970074,
      "lm_loss": 0.4381,
      "step": 10760,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.7894567898970074,
      "lm_loss": 0.0529,
      "step": 10760,
      "vm_loss": 0.2053
    },
    {
      "epoch": 0.7895301594871466,
      "grad_norm": 1.437233855874972,
      "learning_rate": 1.117013441486333e-06,
      "loss": 0.8141,
      "step": 10761
    },
    {
      "epoch": 0.7896035290772857,
      "grad_norm": 1.2217444498326349,
      "learning_rate": 1.1162649902670787e-06,
      "loss": 0.6121,
      "step": 10762
    },
    {
      "epoch": 0.7896768986674249,
      "grad_norm": 1.5689145433118112,
      "learning_rate": 1.1155167583719573e-06,
      "loss": 0.6921,
      "step": 10763
    },
    {
      "epoch": 0.7897502682575639,
      "grad_norm": 0.5476620164687294,
      "learning_rate": 1.1147687458432248e-06,
      "loss": 0.4758,
      "step": 10764
    },
    {
      "epoch": 0.789823637847703,
      "grad_norm": 0.5482589883645742,
      "learning_rate": 1.1140209527231226e-06,
      "loss": 0.4731,
      "step": 10765
    },
    {
      "epoch": 0.7898970074378422,
      "grad_norm": 1.4394589400958842,
      "learning_rate": 1.1132733790538797e-06,
      "loss": 0.6649,
      "step": 10766
    },
    {
      "epoch": 0.7899703770279813,
      "grad_norm": 1.5413042939487716,
      "learning_rate": 1.1125260248777153e-06,
      "loss": 0.8219,
      "step": 10767
    },
    {
      "epoch": 0.7900437466181205,
      "grad_norm": 1.600476917972403,
      "learning_rate": 1.1117788902368338e-06,
      "loss": 0.7374,
      "step": 10768
    },
    {
      "epoch": 0.7900437466181205,
      "lm_loss": 1.6865,
      "step": 10768,
      "vm_loss": 0.061
    },
    {
      "epoch": 0.7900437466181205,
      "lm_loss": 0.0913,
      "step": 10768,
      "vm_loss": 0.1003
    },
    {
      "epoch": 0.7900437466181205,
      "lm_loss": 0.2164,
      "step": 10768,
      "vm_loss": 0.0931
    },
    {
      "epoch": 0.7900437466181205,
      "lm_loss": 0.0752,
      "step": 10768,
      "vm_loss": 0.0876
    },
    {
      "epoch": 0.7900437466181205,
      "lm_loss": 0.6998,
      "step": 10768,
      "vm_loss": 0.1309
    },
    {
      "epoch": 0.7900437466181205,
      "lm_loss": 0.5637,
      "step": 10768,
      "vm_loss": 0.0975
    },
    {
      "epoch": 0.7900437466181205,
      "lm_loss": 0.0719,
      "step": 10768,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.7900437466181205,
      "lm_loss": 0.3781,
      "step": 10768,
      "vm_loss": 0.1009
    },
    {
      "epoch": 0.7901171162082595,
      "grad_norm": 1.417546483690175,
      "learning_rate": 1.1110319751734271e-06,
      "loss": 0.6896,
      "step": 10769
    },
    {
      "epoch": 0.7901904857983987,
      "grad_norm": 0.5418387816195359,
      "learning_rate": 1.110285279729676e-06,
      "loss": 0.4887,
      "step": 10770
    },
    {
      "epoch": 0.7902638553885378,
      "grad_norm": 1.5448302089010653,
      "learning_rate": 1.1095388039477479e-06,
      "loss": 0.7791,
      "step": 10771
    },
    {
      "epoch": 0.790337224978677,
      "grad_norm": 0.5464051280843787,
      "learning_rate": 1.1087925478697964e-06,
      "loss": 0.4766,
      "step": 10772
    },
    {
      "epoch": 0.7904105945688161,
      "grad_norm": 1.5116919246354787,
      "learning_rate": 1.1080465115379674e-06,
      "loss": 0.6689,
      "step": 10773
    },
    {
      "epoch": 0.7904839641589552,
      "grad_norm": 1.5439213824867386,
      "learning_rate": 1.1073006949943904e-06,
      "loss": 0.7726,
      "step": 10774
    },
    {
      "epoch": 0.7905573337490943,
      "grad_norm": 1.6262564528413248,
      "learning_rate": 1.1065550982811828e-06,
      "loss": 0.6633,
      "step": 10775
    },
    {
      "epoch": 0.7906307033392335,
      "grad_norm": 1.5550738944704596,
      "learning_rate": 1.1058097214404517e-06,
      "loss": 0.7082,
      "step": 10776
    },
    {
      "epoch": 0.7906307033392335,
      "lm_loss": 0.2276,
      "step": 10776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7906307033392335,
      "lm_loss": 0.5091,
      "step": 10776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7906307033392335,
      "lm_loss": 0.5737,
      "step": 10776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7906307033392335,
      "lm_loss": 0.574,
      "step": 10776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7906307033392335,
      "lm_loss": 0.3963,
      "step": 10776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7906307033392335,
      "lm_loss": 0.4277,
      "step": 10776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7906307033392335,
      "lm_loss": 0.4807,
      "step": 10776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7906307033392335,
      "lm_loss": 0.4335,
      "step": 10776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7907040729293726,
      "grad_norm": 0.5832506445807977,
      "learning_rate": 1.105064564514287e-06,
      "loss": 0.4705,
      "step": 10777
    },
    {
      "epoch": 0.7907774425195118,
      "grad_norm": 1.5895756256495082,
      "learning_rate": 1.1043196275447742e-06,
      "loss": 0.7819,
      "step": 10778
    },
    {
      "epoch": 0.7908508121096508,
      "grad_norm": 1.6047186596712835,
      "learning_rate": 1.1035749105739791e-06,
      "loss": 0.6515,
      "step": 10779
    },
    {
      "epoch": 0.79092418169979,
      "grad_norm": 1.612731158064979,
      "learning_rate": 1.1028304136439582e-06,
      "loss": 0.725,
      "step": 10780
    },
    {
      "epoch": 0.7909975512899291,
      "grad_norm": 1.4193877011217173,
      "learning_rate": 1.102086136796755e-06,
      "loss": 0.6291,
      "step": 10781
    },
    {
      "epoch": 0.7910709208800683,
      "grad_norm": 0.5506138659293744,
      "learning_rate": 1.1013420800744007e-06,
      "loss": 0.4797,
      "step": 10782
    },
    {
      "epoch": 0.7911442904702074,
      "grad_norm": 1.4223848096821212,
      "learning_rate": 1.1005982435189127e-06,
      "loss": 0.7518,
      "step": 10783
    },
    {
      "epoch": 0.7912176600603464,
      "grad_norm": 0.5413153583694171,
      "learning_rate": 1.0998546271723e-06,
      "loss": 0.4835,
      "step": 10784
    },
    {
      "epoch": 0.7912176600603464,
      "lm_loss": 1.4143,
      "step": 10784,
      "vm_loss": 0.0768
    },
    {
      "epoch": 0.7912176600603464,
      "lm_loss": 0.1282,
      "step": 10784,
      "vm_loss": 0.102
    },
    {
      "epoch": 0.7912176600603464,
      "lm_loss": 0.86,
      "step": 10784,
      "vm_loss": 0.0944
    },
    {
      "epoch": 0.7912176600603464,
      "lm_loss": 0.5758,
      "step": 10784,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.7912176600603464,
      "lm_loss": 0.192,
      "step": 10784,
      "vm_loss": 0.0983
    },
    {
      "epoch": 0.7912176600603464,
      "lm_loss": 0.0626,
      "step": 10784,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.7912176600603464,
      "lm_loss": 0.0672,
      "step": 10784,
      "vm_loss": 0.0623
    },
    {
      "epoch": 0.7912176600603464,
      "lm_loss": 0.5875,
      "step": 10784,
      "vm_loss": 0.1297
    },
    {
      "epoch": 0.7912910296504856,
      "grad_norm": 1.5994132738388673,
      "learning_rate": 1.099111231076555e-06,
      "loss": 0.6917,
      "step": 10785
    },
    {
      "epoch": 0.7913643992406247,
      "grad_norm": 0.5630301926313328,
      "learning_rate": 1.0983680552736592e-06,
      "loss": 0.4899,
      "step": 10786
    },
    {
      "epoch": 0.7914377688307639,
      "grad_norm": 0.576109911739212,
      "learning_rate": 1.0976250998055815e-06,
      "loss": 0.4849,
      "step": 10787
    },
    {
      "epoch": 0.791511138420903,
      "grad_norm": 0.5181830236746625,
      "learning_rate": 1.0968823647142768e-06,
      "loss": 0.4634,
      "step": 10788
    },
    {
      "epoch": 0.7915845080110421,
      "grad_norm": 1.387745451513157,
      "learning_rate": 1.0961398500416926e-06,
      "loss": 0.7014,
      "step": 10789
    },
    {
      "epoch": 0.7916578776011812,
      "grad_norm": 1.4644512967457228,
      "learning_rate": 1.0953975558297586e-06,
      "loss": 0.7414,
      "step": 10790
    },
    {
      "epoch": 0.7917312471913204,
      "grad_norm": 0.5633059048114955,
      "learning_rate": 1.0946554821203947e-06,
      "loss": 0.4751,
      "step": 10791
    },
    {
      "epoch": 0.7918046167814595,
      "grad_norm": 1.6811633282564586,
      "learning_rate": 1.0939136289555069e-06,
      "loss": 0.7879,
      "step": 10792
    },
    {
      "epoch": 0.7918046167814595,
      "lm_loss": 0.4185,
      "step": 10792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7918046167814595,
      "lm_loss": 0.3052,
      "step": 10792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7918046167814595,
      "lm_loss": 0.5667,
      "step": 10792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7918046167814595,
      "lm_loss": 0.4075,
      "step": 10792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7918046167814595,
      "lm_loss": 0.4937,
      "step": 10792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7918046167814595,
      "lm_loss": 0.4322,
      "step": 10792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7918046167814595,
      "lm_loss": 0.6361,
      "step": 10792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7918046167814595,
      "lm_loss": 0.6076,
      "step": 10792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7918779863715987,
      "grad_norm": 0.5733398814639019,
      "learning_rate": 1.0931719963769881e-06,
      "loss": 0.4747,
      "step": 10793
    },
    {
      "epoch": 0.7919513559617377,
      "grad_norm": 1.4692220816147679,
      "learning_rate": 1.092430584426723e-06,
      "loss": 0.6665,
      "step": 10794
    },
    {
      "epoch": 0.7920247255518769,
      "grad_norm": 0.5443590508375797,
      "learning_rate": 1.0916893931465806e-06,
      "loss": 0.4674,
      "step": 10795
    },
    {
      "epoch": 0.792098095142016,
      "grad_norm": 0.5578654907611174,
      "learning_rate": 1.0909484225784163e-06,
      "loss": 0.4842,
      "step": 10796
    },
    {
      "epoch": 0.7921714647321552,
      "grad_norm": 1.47112038918817,
      "learning_rate": 1.0902076727640748e-06,
      "loss": 0.7104,
      "step": 10797
    },
    {
      "epoch": 0.7922448343222943,
      "grad_norm": 1.4803877130695102,
      "learning_rate": 1.089467143745389e-06,
      "loss": 0.7105,
      "step": 10798
    },
    {
      "epoch": 0.7923182039124333,
      "grad_norm": 1.610916743614867,
      "learning_rate": 1.0887268355641768e-06,
      "loss": 0.7303,
      "step": 10799
    },
    {
      "epoch": 0.7923915735025725,
      "grad_norm": 1.4643986246857084,
      "learning_rate": 1.087986748262247e-06,
      "loss": 0.7627,
      "step": 10800
    },
    {
      "epoch": 0.7923915735025725,
      "lm_loss": 1.1468,
      "step": 10800,
      "vm_loss": 0.0711
    },
    {
      "epoch": 0.7923915735025725,
      "lm_loss": 0.6283,
      "step": 10800,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.7923915735025725,
      "lm_loss": 0.7225,
      "step": 10800,
      "vm_loss": 0.2054
    },
    {
      "epoch": 0.7923915735025725,
      "lm_loss": 0.5649,
      "step": 10800,
      "vm_loss": 0.0781
    },
    {
      "epoch": 0.7923915735025725,
      "lm_loss": 0.1011,
      "step": 10800,
      "vm_loss": 0.056
    },
    {
      "epoch": 0.7923915735025725,
      "lm_loss": 0.6534,
      "step": 10800,
      "vm_loss": 0.1704
    },
    {
      "epoch": 0.7923915735025725,
      "lm_loss": 0.1494,
      "step": 10800,
      "vm_loss": 0.07
    },
    {
      "epoch": 0.7923915735025725,
      "lm_loss": 1.0089,
      "step": 10800,
      "vm_loss": 0.2066
    },
    {
      "epoch": 0.7924649430927116,
      "grad_norm": 1.6500590157898058,
      "learning_rate": 1.0872468818813935e-06,
      "loss": 0.6591,
      "step": 10801
    },
    {
      "epoch": 0.7925383126828508,
      "grad_norm": 1.491824793264064,
      "learning_rate": 1.086507236463399e-06,
      "loss": 0.7613,
      "step": 10802
    },
    {
      "epoch": 0.79261168227299,
      "grad_norm": 1.6227909688700257,
      "learning_rate": 1.085767812050032e-06,
      "loss": 0.7512,
      "step": 10803
    },
    {
      "epoch": 0.792685051863129,
      "grad_norm": 0.5609162713652056,
      "learning_rate": 1.0850286086830486e-06,
      "loss": 0.4725,
      "step": 10804
    },
    {
      "epoch": 0.7927584214532681,
      "grad_norm": 1.621868224820646,
      "learning_rate": 1.0842896264041963e-06,
      "loss": 0.661,
      "step": 10805
    },
    {
      "epoch": 0.7928317910434073,
      "grad_norm": 1.6181212096475703,
      "learning_rate": 1.0835508652552062e-06,
      "loss": 0.7206,
      "step": 10806
    },
    {
      "epoch": 0.7929051606335464,
      "grad_norm": 0.5389171606472836,
      "learning_rate": 1.0828123252777973e-06,
      "loss": 0.4844,
      "step": 10807
    },
    {
      "epoch": 0.7929785302236856,
      "grad_norm": 1.5432866882682728,
      "learning_rate": 1.0820740065136775e-06,
      "loss": 0.7468,
      "step": 10808
    },
    {
      "epoch": 0.7929785302236856,
      "lm_loss": 1.061,
      "step": 10808,
      "vm_loss": 0.0383
    },
    {
      "epoch": 0.7929785302236856,
      "lm_loss": 0.222,
      "step": 10808,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.7929785302236856,
      "lm_loss": 0.2636,
      "step": 10808,
      "vm_loss": 0.1503
    },
    {
      "epoch": 0.7929785302236856,
      "lm_loss": 0.7693,
      "step": 10808,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.7929785302236856,
      "lm_loss": 0.0665,
      "step": 10808,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.7929785302236856,
      "lm_loss": 0.7144,
      "step": 10808,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.7929785302236856,
      "lm_loss": 0.8085,
      "step": 10808,
      "vm_loss": 0.095
    },
    {
      "epoch": 0.7929785302236856,
      "lm_loss": 0.7159,
      "step": 10808,
      "vm_loss": 0.1857
    },
    {
      "epoch": 0.7930518998138246,
      "grad_norm": 1.6163261191869776,
      "learning_rate": 1.0813359090045412e-06,
      "loss": 0.8275,
      "step": 10809
    },
    {
      "epoch": 0.7931252694039638,
      "grad_norm": 1.650995398967651,
      "learning_rate": 1.080598032792069e-06,
      "loss": 0.6524,
      "step": 10810
    },
    {
      "epoch": 0.7931986389941029,
      "grad_norm": 1.5393304590155803,
      "learning_rate": 1.0798603779179334e-06,
      "loss": 0.7539,
      "step": 10811
    },
    {
      "epoch": 0.7932720085842421,
      "grad_norm": 1.5125175879774206,
      "learning_rate": 1.07912294442379e-06,
      "loss": 0.7592,
      "step": 10812
    },
    {
      "epoch": 0.7933453781743812,
      "grad_norm": 1.4700673272224947,
      "learning_rate": 1.078385732351284e-06,
      "loss": 0.8367,
      "step": 10813
    },
    {
      "epoch": 0.7934187477645203,
      "grad_norm": 1.6099958885810282,
      "learning_rate": 1.0776487417420478e-06,
      "loss": 0.7487,
      "step": 10814
    },
    {
      "epoch": 0.7934921173546594,
      "grad_norm": 1.3580375958571076,
      "learning_rate": 1.0769119726376992e-06,
      "loss": 0.8238,
      "step": 10815
    },
    {
      "epoch": 0.7935654869447986,
      "grad_norm": 1.52462772452967,
      "learning_rate": 1.0761754250798478e-06,
      "loss": 0.7499,
      "step": 10816
    },
    {
      "epoch": 0.7935654869447986,
      "lm_loss": 0.975,
      "step": 10816,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.7935654869447986,
      "lm_loss": 1.0126,
      "step": 10816,
      "vm_loss": 0.1335
    },
    {
      "epoch": 0.7935654869447986,
      "lm_loss": 0.5799,
      "step": 10816,
      "vm_loss": 0.2584
    },
    {
      "epoch": 0.7935654869447986,
      "lm_loss": 1.0331,
      "step": 10816,
      "vm_loss": 0.1717
    },
    {
      "epoch": 0.7935654869447986,
      "lm_loss": 1.4189,
      "step": 10816,
      "vm_loss": 0.0781
    },
    {
      "epoch": 0.7935654869447986,
      "lm_loss": 0.0603,
      "step": 10816,
      "vm_loss": 0.1328
    },
    {
      "epoch": 0.7935654869447986,
      "lm_loss": 0.5695,
      "step": 10816,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.7935654869447986,
      "lm_loss": 0.1775,
      "step": 10816,
      "vm_loss": 0.1141
    },
    {
      "epoch": 0.7936388565349377,
      "grad_norm": 1.5944443808065707,
      "learning_rate": 1.0754390991100882e-06,
      "loss": 0.6405,
      "step": 10817
    },
    {
      "epoch": 0.7937122261250769,
      "grad_norm": 1.5474033411516643,
      "learning_rate": 1.0747029947700011e-06,
      "loss": 0.7515,
      "step": 10818
    },
    {
      "epoch": 0.7937855957152159,
      "grad_norm": 1.505870891060034,
      "learning_rate": 1.0739671121011568e-06,
      "loss": 0.7188,
      "step": 10819
    },
    {
      "epoch": 0.793858965305355,
      "grad_norm": 1.4089309821084222,
      "learning_rate": 1.073231451145111e-06,
      "loss": 0.6589,
      "step": 10820
    },
    {
      "epoch": 0.7939323348954942,
      "grad_norm": 0.5419406996068462,
      "learning_rate": 1.0724960119434108e-06,
      "loss": 0.4746,
      "step": 10821
    },
    {
      "epoch": 0.7940057044856333,
      "grad_norm": 1.4148785251640006,
      "learning_rate": 1.0717607945375874e-06,
      "loss": 0.6996,
      "step": 10822
    },
    {
      "epoch": 0.7940790740757725,
      "grad_norm": 1.4925987977273183,
      "learning_rate": 1.0710257989691596e-06,
      "loss": 0.7214,
      "step": 10823
    },
    {
      "epoch": 0.7941524436659115,
      "grad_norm": 0.5598456194562134,
      "learning_rate": 1.0702910252796351e-06,
      "loss": 0.4904,
      "step": 10824
    },
    {
      "epoch": 0.7941524436659115,
      "lm_loss": 0.4345,
      "step": 10824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7941524436659115,
      "lm_loss": 0.4192,
      "step": 10824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7941524436659115,
      "lm_loss": 0.424,
      "step": 10824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7941524436659115,
      "lm_loss": 0.5371,
      "step": 10824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7941524436659115,
      "lm_loss": 0.486,
      "step": 10824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7941524436659115,
      "lm_loss": 0.4255,
      "step": 10824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7941524436659115,
      "lm_loss": 0.5083,
      "step": 10824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7941524436659115,
      "lm_loss": 0.5246,
      "step": 10824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7942258132560507,
      "grad_norm": 0.5330145828731983,
      "learning_rate": 1.0695564735105085e-06,
      "loss": 0.4892,
      "step": 10825
    },
    {
      "epoch": 0.7942991828461898,
      "grad_norm": 1.6048077060236527,
      "learning_rate": 1.0688221437032597e-06,
      "loss": 0.8417,
      "step": 10826
    },
    {
      "epoch": 0.794372552436329,
      "grad_norm": 1.4602621812337346,
      "learning_rate": 1.0680880358993613e-06,
      "loss": 0.6995,
      "step": 10827
    },
    {
      "epoch": 0.7944459220264681,
      "grad_norm": 1.4803880334532005,
      "learning_rate": 1.0673541501402685e-06,
      "loss": 0.7665,
      "step": 10828
    },
    {
      "epoch": 0.7945192916166072,
      "grad_norm": 1.525864916088818,
      "learning_rate": 1.0666204864674263e-06,
      "loss": 0.7141,
      "step": 10829
    },
    {
      "epoch": 0.7945926612067463,
      "grad_norm": 1.4954055725761652,
      "learning_rate": 1.0658870449222658e-06,
      "loss": 0.7466,
      "step": 10830
    },
    {
      "epoch": 0.7946660307968855,
      "grad_norm": 1.7248557980280572,
      "learning_rate": 1.0651538255462053e-06,
      "loss": 0.8008,
      "step": 10831
    },
    {
      "epoch": 0.7947394003870246,
      "grad_norm": 1.3948992765834956,
      "learning_rate": 1.064420828380654e-06,
      "loss": 0.7726,
      "step": 10832
    },
    {
      "epoch": 0.7947394003870246,
      "lm_loss": 0.938,
      "step": 10832,
      "vm_loss": 0.0358
    },
    {
      "epoch": 0.7947394003870246,
      "lm_loss": 1.3102,
      "step": 10832,
      "vm_loss": 0.1052
    },
    {
      "epoch": 0.7947394003870246,
      "lm_loss": 0.2965,
      "step": 10832,
      "vm_loss": 0.1838
    },
    {
      "epoch": 0.7947394003870246,
      "lm_loss": 0.8183,
      "step": 10832,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.7947394003870246,
      "lm_loss": 0.6358,
      "step": 10832,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.7947394003870246,
      "lm_loss": 0.6074,
      "step": 10832,
      "vm_loss": 0.0996
    },
    {
      "epoch": 0.7947394003870246,
      "lm_loss": 0.6585,
      "step": 10832,
      "vm_loss": 0.1482
    },
    {
      "epoch": 0.7947394003870246,
      "lm_loss": 0.2517,
      "step": 10832,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.7948127699771638,
      "grad_norm": 1.4525546638621907,
      "learning_rate": 1.0636880534670053e-06,
      "loss": 0.7378,
      "step": 10833
    },
    {
      "epoch": 0.7948861395673028,
      "grad_norm": 1.454756422464794,
      "learning_rate": 1.0629555008466397e-06,
      "loss": 0.8313,
      "step": 10834
    },
    {
      "epoch": 0.794959509157442,
      "grad_norm": 0.5484022300690389,
      "learning_rate": 1.062223170560927e-06,
      "loss": 0.4687,
      "step": 10835
    },
    {
      "epoch": 0.7950328787475811,
      "grad_norm": 1.3774604785115951,
      "learning_rate": 1.0614910626512236e-06,
      "loss": 0.7604,
      "step": 10836
    },
    {
      "epoch": 0.7951062483377203,
      "grad_norm": 1.4934498291402059,
      "learning_rate": 1.0607591771588722e-06,
      "loss": 0.7826,
      "step": 10837
    },
    {
      "epoch": 0.7951796179278594,
      "grad_norm": 1.6232690031875838,
      "learning_rate": 1.060027514125207e-06,
      "loss": 0.6738,
      "step": 10838
    },
    {
      "epoch": 0.7952529875179984,
      "grad_norm": 0.5545704360120858,
      "learning_rate": 1.0592960735915447e-06,
      "loss": 0.4795,
      "step": 10839
    },
    {
      "epoch": 0.7953263571081376,
      "grad_norm": 0.5594719850388076,
      "learning_rate": 1.0585648555991928e-06,
      "loss": 0.4951,
      "step": 10840
    },
    {
      "epoch": 0.7953263571081376,
      "lm_loss": 0.091,
      "step": 10840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7953263571081376,
      "lm_loss": 0.4206,
      "step": 10840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7953263571081376,
      "lm_loss": 0.4083,
      "step": 10840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7953263571081376,
      "lm_loss": 0.4928,
      "step": 10840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7953263571081376,
      "lm_loss": 0.5164,
      "step": 10840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7953263571081376,
      "lm_loss": 0.5163,
      "step": 10840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7953263571081376,
      "lm_loss": 0.4674,
      "step": 10840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7953263571081376,
      "lm_loss": 0.5588,
      "step": 10840,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7953997266982767,
      "grad_norm": 0.5495004444495597,
      "learning_rate": 1.0578338601894439e-06,
      "loss": 0.4823,
      "step": 10841
    },
    {
      "epoch": 0.7954730962884159,
      "grad_norm": 0.5494649339363693,
      "learning_rate": 1.057103087403578e-06,
      "loss": 0.4785,
      "step": 10842
    },
    {
      "epoch": 0.7955464658785549,
      "grad_norm": 0.5521147466471706,
      "learning_rate": 1.056372537282867e-06,
      "loss": 0.4566,
      "step": 10843
    },
    {
      "epoch": 0.7956198354686941,
      "grad_norm": 1.4586421142328603,
      "learning_rate": 1.0556422098685648e-06,
      "loss": 0.7536,
      "step": 10844
    },
    {
      "epoch": 0.7956932050588332,
      "grad_norm": 1.8307942158797885,
      "learning_rate": 1.054912105201915e-06,
      "loss": 0.7217,
      "step": 10845
    },
    {
      "epoch": 0.7957665746489724,
      "grad_norm": 0.5298492705252489,
      "learning_rate": 1.054182223324149e-06,
      "loss": 0.4749,
      "step": 10846
    },
    {
      "epoch": 0.7958399442391115,
      "grad_norm": 0.5643906774113325,
      "learning_rate": 1.053452564276482e-06,
      "loss": 0.4778,
      "step": 10847
    },
    {
      "epoch": 0.7959133138292506,
      "grad_norm": 1.5702511403924302,
      "learning_rate": 1.0527231281001243e-06,
      "loss": 0.7148,
      "step": 10848
    },
    {
      "epoch": 0.7959133138292506,
      "lm_loss": 0.7822,
      "step": 10848,
      "vm_loss": 0.0711
    },
    {
      "epoch": 0.7959133138292506,
      "lm_loss": 1.2128,
      "step": 10848,
      "vm_loss": 0.1109
    },
    {
      "epoch": 0.7959133138292506,
      "lm_loss": 0.5201,
      "step": 10848,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.7959133138292506,
      "lm_loss": 0.5407,
      "step": 10848,
      "vm_loss": 0.2713
    },
    {
      "epoch": 0.7959133138292506,
      "lm_loss": 0.7622,
      "step": 10848,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.7959133138292506,
      "lm_loss": 0.6833,
      "step": 10848,
      "vm_loss": 0.1966
    },
    {
      "epoch": 0.7959133138292506,
      "lm_loss": 0.1208,
      "step": 10848,
      "vm_loss": 0.139
    },
    {
      "epoch": 0.7959133138292506,
      "lm_loss": 0.7888,
      "step": 10848,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.7959866834193897,
      "grad_norm": 1.5834605183928427,
      "learning_rate": 1.0519939148362667e-06,
      "loss": 0.719,
      "step": 10849
    },
    {
      "epoch": 0.7960600530095289,
      "grad_norm": 1.538222265528965,
      "learning_rate": 1.0512649245260913e-06,
      "loss": 0.7176,
      "step": 10850
    },
    {
      "epoch": 0.796133422599668,
      "grad_norm": 2.0121999945550413,
      "learning_rate": 1.050536157210763e-06,
      "loss": 0.7072,
      "step": 10851
    },
    {
      "epoch": 0.7962067921898072,
      "grad_norm": 1.7038183897224903,
      "learning_rate": 1.0498076129314372e-06,
      "loss": 0.7767,
      "step": 10852
    },
    {
      "epoch": 0.7962801617799462,
      "grad_norm": 1.47791729766752,
      "learning_rate": 1.0490792917292591e-06,
      "loss": 0.6893,
      "step": 10853
    },
    {
      "epoch": 0.7963535313700854,
      "grad_norm": 1.4361801834334673,
      "learning_rate": 1.0483511936453583e-06,
      "loss": 0.6979,
      "step": 10854
    },
    {
      "epoch": 0.7964269009602245,
      "grad_norm": 1.397186934147125,
      "learning_rate": 1.0476233187208511e-06,
      "loss": 0.6747,
      "step": 10855
    },
    {
      "epoch": 0.7965002705503637,
      "grad_norm": 1.3235658735677076,
      "learning_rate": 1.0468956669968433e-06,
      "loss": 0.7714,
      "step": 10856
    },
    {
      "epoch": 0.7965002705503637,
      "lm_loss": 0.6158,
      "step": 10856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7965002705503637,
      "lm_loss": 0.3363,
      "step": 10856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7965002705503637,
      "lm_loss": 0.5833,
      "step": 10856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7965002705503637,
      "lm_loss": 0.4367,
      "step": 10856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7965002705503637,
      "lm_loss": 0.555,
      "step": 10856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7965002705503637,
      "lm_loss": 0.5232,
      "step": 10856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7965002705503637,
      "lm_loss": 0.5676,
      "step": 10856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7965002705503637,
      "lm_loss": 0.4321,
      "step": 10856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7965736401405028,
      "grad_norm": 0.5424455941840336,
      "learning_rate": 1.0461682385144256e-06,
      "loss": 0.4763,
      "step": 10857
    },
    {
      "epoch": 0.7966470097306418,
      "grad_norm": 1.4410204182151103,
      "learning_rate": 1.0454410333146802e-06,
      "loss": 0.7159,
      "step": 10858
    },
    {
      "epoch": 0.796720379320781,
      "grad_norm": 0.5449300962034432,
      "learning_rate": 1.0447140514386728e-06,
      "loss": 0.4708,
      "step": 10859
    },
    {
      "epoch": 0.7967937489109201,
      "grad_norm": 1.6867312930146683,
      "learning_rate": 1.043987292927458e-06,
      "loss": 0.6986,
      "step": 10860
    },
    {
      "epoch": 0.7968671185010593,
      "grad_norm": 0.5352720438137036,
      "learning_rate": 1.0432607578220782e-06,
      "loss": 0.4829,
      "step": 10861
    },
    {
      "epoch": 0.7969404880911984,
      "grad_norm": 1.477702673021614,
      "learning_rate": 1.04253444616356e-06,
      "loss": 0.7694,
      "step": 10862
    },
    {
      "epoch": 0.7970138576813375,
      "grad_norm": 0.5631985691032805,
      "learning_rate": 1.0418083579929238e-06,
      "loss": 0.4952,
      "step": 10863
    },
    {
      "epoch": 0.7970872272714766,
      "grad_norm": 1.2972879354480715,
      "learning_rate": 1.041082493351172e-06,
      "loss": 0.7032,
      "step": 10864
    },
    {
      "epoch": 0.7970872272714766,
      "lm_loss": 0.1124,
      "step": 10864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7970872272714766,
      "lm_loss": 0.5545,
      "step": 10864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7970872272714766,
      "lm_loss": 0.3694,
      "step": 10864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7970872272714766,
      "lm_loss": 0.5669,
      "step": 10864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7970872272714766,
      "lm_loss": 0.4738,
      "step": 10864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7970872272714766,
      "lm_loss": 0.4688,
      "step": 10864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7970872272714766,
      "lm_loss": 0.4972,
      "step": 10864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7970872272714766,
      "lm_loss": 0.4808,
      "step": 10864,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7971605968616158,
      "grad_norm": 0.5318841441411432,
      "learning_rate": 1.0403568522792962e-06,
      "loss": 0.4687,
      "step": 10865
    },
    {
      "epoch": 0.7972339664517549,
      "grad_norm": 0.5338330485004535,
      "learning_rate": 1.0396314348182746e-06,
      "loss": 0.4733,
      "step": 10866
    },
    {
      "epoch": 0.7973073360418941,
      "grad_norm": 1.907719582624045,
      "learning_rate": 1.0389062410090734e-06,
      "loss": 0.6714,
      "step": 10867
    },
    {
      "epoch": 0.7973807056320331,
      "grad_norm": 1.4118712372420659,
      "learning_rate": 1.0381812708926447e-06,
      "loss": 0.83,
      "step": 10868
    },
    {
      "epoch": 0.7974540752221723,
      "grad_norm": 1.624279090137353,
      "learning_rate": 1.0374565245099328e-06,
      "loss": 0.7665,
      "step": 10869
    },
    {
      "epoch": 0.7975274448123114,
      "grad_norm": 1.4107366993880899,
      "learning_rate": 1.0367320019018634e-06,
      "loss": 0.7427,
      "step": 10870
    },
    {
      "epoch": 0.7976008144024506,
      "grad_norm": 1.4687020505245323,
      "learning_rate": 1.036007703109353e-06,
      "loss": 0.7151,
      "step": 10871
    },
    {
      "epoch": 0.7976741839925897,
      "grad_norm": 1.3660301763294114,
      "learning_rate": 1.0352836281733048e-06,
      "loss": 0.7394,
      "step": 10872
    },
    {
      "epoch": 0.7976741839925897,
      "lm_loss": 0.9968,
      "step": 10872,
      "vm_loss": 0.1305
    },
    {
      "epoch": 0.7976741839925897,
      "lm_loss": 1.2908,
      "step": 10872,
      "vm_loss": 0.0342
    },
    {
      "epoch": 0.7976741839925897,
      "lm_loss": 0.4019,
      "step": 10872,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.7976741839925897,
      "lm_loss": 0.9703,
      "step": 10872,
      "vm_loss": 0.1822
    },
    {
      "epoch": 0.7976741839925897,
      "lm_loss": 0.8683,
      "step": 10872,
      "vm_loss": 0.1274
    },
    {
      "epoch": 0.7976741839925897,
      "lm_loss": 0.8743,
      "step": 10872,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.7976741839925897,
      "lm_loss": 0.6215,
      "step": 10872,
      "vm_loss": 0.1806
    },
    {
      "epoch": 0.7976741839925897,
      "lm_loss": 0.5131,
      "step": 10872,
      "vm_loss": 0.1328
    },
    {
      "epoch": 0.7977475535827288,
      "grad_norm": 1.4772074292557615,
      "learning_rate": 1.0345597771346066e-06,
      "loss": 0.7984,
      "step": 10873
    },
    {
      "epoch": 0.7978209231728679,
      "grad_norm": 0.5512105025476072,
      "learning_rate": 1.0338361500341392e-06,
      "loss": 0.4877,
      "step": 10874
    },
    {
      "epoch": 0.797894292763007,
      "grad_norm": 0.5528577442926224,
      "learning_rate": 1.0331127469127671e-06,
      "loss": 0.4808,
      "step": 10875
    },
    {
      "epoch": 0.7979676623531462,
      "grad_norm": 1.7396557933056924,
      "learning_rate": 1.0323895678113415e-06,
      "loss": 0.7874,
      "step": 10876
    },
    {
      "epoch": 0.7980410319432854,
      "grad_norm": 1.5340004742663738,
      "learning_rate": 1.0316666127707032e-06,
      "loss": 0.6843,
      "step": 10877
    },
    {
      "epoch": 0.7981144015334244,
      "grad_norm": 1.4562795076603194,
      "learning_rate": 1.0309438818316781e-06,
      "loss": 0.6426,
      "step": 10878
    },
    {
      "epoch": 0.7981877711235635,
      "grad_norm": 0.5608962306800158,
      "learning_rate": 1.0302213750350797e-06,
      "loss": 0.4939,
      "step": 10879
    },
    {
      "epoch": 0.7982611407137027,
      "grad_norm": 0.5442749345237357,
      "learning_rate": 1.0294990924217135e-06,
      "loss": 0.4742,
      "step": 10880
    },
    {
      "epoch": 0.7982611407137027,
      "lm_loss": 0.4191,
      "step": 10880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7982611407137027,
      "lm_loss": 0.445,
      "step": 10880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7982611407137027,
      "lm_loss": 0.3018,
      "step": 10880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7982611407137027,
      "lm_loss": 0.5534,
      "step": 10880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7982611407137027,
      "lm_loss": 0.4794,
      "step": 10880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7982611407137027,
      "lm_loss": 0.4583,
      "step": 10880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7982611407137027,
      "lm_loss": 0.3502,
      "step": 10880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7982611407137027,
      "lm_loss": 0.4877,
      "step": 10880,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7983345103038418,
      "grad_norm": 0.5437206104169537,
      "learning_rate": 1.0287770340323655e-06,
      "loss": 0.4737,
      "step": 10881
    },
    {
      "epoch": 0.798407879893981,
      "grad_norm": 0.5537454650707008,
      "learning_rate": 1.028055199907813e-06,
      "loss": 0.4796,
      "step": 10882
    },
    {
      "epoch": 0.79848124948412,
      "grad_norm": 0.5702947246536644,
      "learning_rate": 1.0273335900888205e-06,
      "loss": 0.4868,
      "step": 10883
    },
    {
      "epoch": 0.7985546190742592,
      "grad_norm": 1.5318961175971717,
      "learning_rate": 1.0266122046161365e-06,
      "loss": 0.8465,
      "step": 10884
    },
    {
      "epoch": 0.7986279886643983,
      "grad_norm": 1.3901587114222254,
      "learning_rate": 1.0258910435305025e-06,
      "loss": 0.7083,
      "step": 10885
    },
    {
      "epoch": 0.7987013582545375,
      "grad_norm": 0.5423280571808938,
      "learning_rate": 1.0251701068726427e-06,
      "loss": 0.4856,
      "step": 10886
    },
    {
      "epoch": 0.7987747278446766,
      "grad_norm": 1.6507061043987121,
      "learning_rate": 1.0244493946832707e-06,
      "loss": 0.7329,
      "step": 10887
    },
    {
      "epoch": 0.7988480974348157,
      "grad_norm": 1.3679347391940455,
      "learning_rate": 1.023728907003087e-06,
      "loss": 0.7741,
      "step": 10888
    },
    {
      "epoch": 0.7988480974348157,
      "lm_loss": 0.4249,
      "step": 10888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7988480974348157,
      "lm_loss": 0.4643,
      "step": 10888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7988480974348157,
      "lm_loss": 0.6579,
      "step": 10888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7988480974348157,
      "lm_loss": 0.6047,
      "step": 10888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7988480974348157,
      "lm_loss": 0.5169,
      "step": 10888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7988480974348157,
      "lm_loss": 0.3303,
      "step": 10888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7988480974348157,
      "lm_loss": 0.5298,
      "step": 10888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7988480974348157,
      "lm_loss": 0.5065,
      "step": 10888,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.7989214670249548,
      "grad_norm": 0.5370561826322102,
      "learning_rate": 1.0230086438727771e-06,
      "loss": 0.4917,
      "step": 10889
    },
    {
      "epoch": 0.798994836615094,
      "grad_norm": 0.519606663913395,
      "learning_rate": 1.0222886053330195e-06,
      "loss": 0.4644,
      "step": 10890
    },
    {
      "epoch": 0.7990682062052331,
      "grad_norm": 0.5425273013267136,
      "learning_rate": 1.0215687914244754e-06,
      "loss": 0.475,
      "step": 10891
    },
    {
      "epoch": 0.7991415757953723,
      "grad_norm": 1.5474586663876813,
      "learning_rate": 1.0208492021877936e-06,
      "loss": 0.6122,
      "step": 10892
    },
    {
      "epoch": 0.7992149453855113,
      "grad_norm": 1.5376896854292341,
      "learning_rate": 1.0201298376636115e-06,
      "loss": 0.8339,
      "step": 10893
    },
    {
      "epoch": 0.7992883149756504,
      "grad_norm": 1.6337114392762062,
      "learning_rate": 1.0194106978925539e-06,
      "loss": 0.6097,
      "step": 10894
    },
    {
      "epoch": 0.7993616845657896,
      "grad_norm": 0.5443540196625488,
      "learning_rate": 1.0186917829152298e-06,
      "loss": 0.4718,
      "step": 10895
    },
    {
      "epoch": 0.7994350541559287,
      "grad_norm": 0.5318187020429522,
      "learning_rate": 1.0179730927722425e-06,
      "loss": 0.4594,
      "step": 10896
    },
    {
      "epoch": 0.7994350541559287,
      "lm_loss": 0.1113,
      "step": 10896,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.7994350541559287,
      "lm_loss": 0.5592,
      "step": 10896,
      "vm_loss": 0.0253
    },
    {
      "epoch": 0.7994350541559287,
      "lm_loss": 0.2742,
      "step": 10896,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.7994350541559287,
      "lm_loss": 0.7574,
      "step": 10896,
      "vm_loss": 0.1659
    },
    {
      "epoch": 0.7994350541559287,
      "lm_loss": 0.3098,
      "step": 10896,
      "vm_loss": 0.2139
    },
    {
      "epoch": 0.7994350541559287,
      "lm_loss": 0.4078,
      "step": 10896,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.7994350541559287,
      "lm_loss": 0.9725,
      "step": 10896,
      "vm_loss": 0.1813
    },
    {
      "epoch": 0.7994350541559287,
      "lm_loss": 0.1174,
      "step": 10896,
      "vm_loss": 0.0477
    },
    {
      "epoch": 0.7995084237460679,
      "grad_norm": 1.511922079991459,
      "learning_rate": 1.0172546275041756e-06,
      "loss": 0.7141,
      "step": 10897
    },
    {
      "epoch": 0.7995817933362069,
      "grad_norm": 1.4481860660842463,
      "learning_rate": 1.0165363871516027e-06,
      "loss": 0.8449,
      "step": 10898
    },
    {
      "epoch": 0.7996551629263461,
      "grad_norm": 0.5409449741030663,
      "learning_rate": 1.015818371755085e-06,
      "loss": 0.5016,
      "step": 10899
    },
    {
      "epoch": 0.7997285325164852,
      "grad_norm": 0.5427473658538794,
      "learning_rate": 1.0151005813551695e-06,
      "loss": 0.488,
      "step": 10900
    },
    {
      "epoch": 0.7998019021066244,
      "grad_norm": 1.484459372319529,
      "learning_rate": 1.014383015992394e-06,
      "loss": 0.827,
      "step": 10901
    },
    {
      "epoch": 0.7998752716967635,
      "grad_norm": 0.5349045589664718,
      "learning_rate": 1.01366567570728e-06,
      "loss": 0.4557,
      "step": 10902
    },
    {
      "epoch": 0.7999486412869026,
      "grad_norm": 1.5565384825784156,
      "learning_rate": 1.0129485605403365e-06,
      "loss": 0.8118,
      "step": 10903
    },
    {
      "epoch": 0.8000220108770417,
      "grad_norm": 0.5453481476878752,
      "learning_rate": 1.0122316705320618e-06,
      "loss": 0.49,
      "step": 10904
    },
    {
      "epoch": 0.8000220108770417,
      "lm_loss": 1.0289,
      "step": 10904,
      "vm_loss": 0.0824
    },
    {
      "epoch": 0.8000220108770417,
      "lm_loss": 1.1916,
      "step": 10904,
      "vm_loss": 0.087
    },
    {
      "epoch": 0.8000220108770417,
      "lm_loss": 0.7301,
      "step": 10904,
      "vm_loss": 0.1595
    },
    {
      "epoch": 0.8000220108770417,
      "lm_loss": 0.5304,
      "step": 10904,
      "vm_loss": 0.1331
    },
    {
      "epoch": 0.8000220108770417,
      "lm_loss": 0.9334,
      "step": 10904,
      "vm_loss": 0.1393
    },
    {
      "epoch": 0.8000220108770417,
      "lm_loss": 0.5211,
      "step": 10904,
      "vm_loss": 0.073
    },
    {
      "epoch": 0.8000220108770417,
      "lm_loss": 0.4651,
      "step": 10904,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.8000220108770417,
      "lm_loss": 0.0014,
      "step": 10904,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.8000953804671809,
      "grad_norm": 1.437130095893086,
      "learning_rate": 1.0115150057229406e-06,
      "loss": 0.7718,
      "step": 10905
    },
    {
      "epoch": 0.80016875005732,
      "grad_norm": 1.5464580355175221,
      "learning_rate": 1.0107985661534426e-06,
      "loss": 0.6972,
      "step": 10906
    },
    {
      "epoch": 0.8002421196474592,
      "grad_norm": 1.4502897992999904,
      "learning_rate": 1.01008235186403e-06,
      "loss": 0.7534,
      "step": 10907
    },
    {
      "epoch": 0.8003154892375982,
      "grad_norm": 0.5580702080839093,
      "learning_rate": 1.0093663628951484e-06,
      "loss": 0.4839,
      "step": 10908
    },
    {
      "epoch": 0.8003888588277374,
      "grad_norm": 1.54536707404335,
      "learning_rate": 1.0086505992872304e-06,
      "loss": 0.708,
      "step": 10909
    },
    {
      "epoch": 0.8004622284178765,
      "grad_norm": 1.7459976477884307,
      "learning_rate": 1.0079350610806976e-06,
      "loss": 0.6979,
      "step": 10910
    },
    {
      "epoch": 0.8005355980080157,
      "grad_norm": 0.5542919898207015,
      "learning_rate": 1.0072197483159568e-06,
      "loss": 0.4726,
      "step": 10911
    },
    {
      "epoch": 0.8006089675981548,
      "grad_norm": 1.4847451657257766,
      "learning_rate": 1.0065046610334062e-06,
      "loss": 0.7653,
      "step": 10912
    },
    {
      "epoch": 0.8006089675981548,
      "lm_loss": 1.1726,
      "step": 10912,
      "vm_loss": 0.0372
    },
    {
      "epoch": 0.8006089675981548,
      "lm_loss": 0.1821,
      "step": 10912,
      "vm_loss": 0.0557
    },
    {
      "epoch": 0.8006089675981548,
      "lm_loss": 0.1858,
      "step": 10912,
      "vm_loss": 0.0918
    },
    {
      "epoch": 0.8006089675981548,
      "lm_loss": 0.7215,
      "step": 10912,
      "vm_loss": 0.0467
    },
    {
      "epoch": 0.8006089675981548,
      "lm_loss": 0.5434,
      "step": 10912,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.8006089675981548,
      "lm_loss": 0.8891,
      "step": 10912,
      "vm_loss": 0.0628
    },
    {
      "epoch": 0.8006089675981548,
      "lm_loss": 0.1602,
      "step": 10912,
      "vm_loss": 0.1202
    },
    {
      "epoch": 0.8006089675981548,
      "lm_loss": 0.3899,
      "step": 10912,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.8006823371882938,
      "grad_norm": 1.4534593224078673,
      "learning_rate": 1.0057897992734266e-06,
      "loss": 0.7042,
      "step": 10913
    },
    {
      "epoch": 0.800755706778433,
      "grad_norm": 1.3865201715079882,
      "learning_rate": 1.005075163076389e-06,
      "loss": 0.7363,
      "step": 10914
    },
    {
      "epoch": 0.8008290763685721,
      "grad_norm": 1.6031531208752559,
      "learning_rate": 1.00436075248265e-06,
      "loss": 0.7298,
      "step": 10915
    },
    {
      "epoch": 0.8009024459587113,
      "grad_norm": 1.4955551030143854,
      "learning_rate": 1.003646567532553e-06,
      "loss": 0.734,
      "step": 10916
    },
    {
      "epoch": 0.8009758155488504,
      "grad_norm": 1.473261929211557,
      "learning_rate": 1.0029326082664326e-06,
      "loss": 0.7647,
      "step": 10917
    },
    {
      "epoch": 0.8010491851389895,
      "grad_norm": 1.5008160871697347,
      "learning_rate": 1.0022188747246064e-06,
      "loss": 0.7434,
      "step": 10918
    },
    {
      "epoch": 0.8011225547291286,
      "grad_norm": 1.3291229694164406,
      "learning_rate": 1.0015053669473806e-06,
      "loss": 0.6812,
      "step": 10919
    },
    {
      "epoch": 0.8011959243192678,
      "grad_norm": 1.57051565892918,
      "learning_rate": 1.0007920849750492e-06,
      "loss": 0.6969,
      "step": 10920
    },
    {
      "epoch": 0.8011959243192678,
      "lm_loss": 0.4288,
      "step": 10920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8011959243192678,
      "lm_loss": 0.377,
      "step": 10920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8011959243192678,
      "lm_loss": 0.5945,
      "step": 10920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8011959243192678,
      "lm_loss": 0.4889,
      "step": 10920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8011959243192678,
      "lm_loss": 0.3445,
      "step": 10920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8011959243192678,
      "lm_loss": 0.553,
      "step": 10920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8011959243192678,
      "lm_loss": 0.4373,
      "step": 10920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8011959243192678,
      "lm_loss": 0.4222,
      "step": 10920,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8012692939094069,
      "grad_norm": 0.5453468045464028,
      "learning_rate": 1.0000790288478924e-06,
      "loss": 0.4705,
      "step": 10921
    },
    {
      "epoch": 0.801342663499546,
      "grad_norm": 1.525974113150099,
      "learning_rate": 9.99366198606177e-07,
      "loss": 0.6866,
      "step": 10922
    },
    {
      "epoch": 0.8014160330896851,
      "grad_norm": 1.4475192162431576,
      "learning_rate": 9.98653594290161e-07,
      "loss": 0.6262,
      "step": 10923
    },
    {
      "epoch": 0.8014894026798243,
      "grad_norm": 1.3749356364196657,
      "learning_rate": 9.979412159400858e-07,
      "loss": 0.6693,
      "step": 10924
    },
    {
      "epoch": 0.8015627722699634,
      "grad_norm": 1.4933689412214206,
      "learning_rate": 9.972290635961812e-07,
      "loss": 0.7814,
      "step": 10925
    },
    {
      "epoch": 0.8016361418601026,
      "grad_norm": 1.5106106053228734,
      "learning_rate": 9.965171372986637e-07,
      "loss": 0.7501,
      "step": 10926
    },
    {
      "epoch": 0.8017095114502416,
      "grad_norm": 0.5545951388014779,
      "learning_rate": 9.958054370877367e-07,
      "loss": 0.4834,
      "step": 10927
    },
    {
      "epoch": 0.8017828810403808,
      "grad_norm": 1.425310557755222,
      "learning_rate": 9.950939630035944e-07,
      "loss": 0.7055,
      "step": 10928
    },
    {
      "epoch": 0.8017828810403808,
      "lm_loss": 0.3995,
      "step": 10928,
      "vm_loss": 0.0384
    },
    {
      "epoch": 0.8017828810403808,
      "lm_loss": 1.6678,
      "step": 10928,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.8017828810403808,
      "lm_loss": 0.0099,
      "step": 10928,
      "vm_loss": 0.1304
    },
    {
      "epoch": 0.8017828810403808,
      "lm_loss": 0.5099,
      "step": 10928,
      "vm_loss": 0.0601
    },
    {
      "epoch": 0.8017828810403808,
      "lm_loss": 0.2589,
      "step": 10928,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.8017828810403808,
      "lm_loss": 0.6622,
      "step": 10928,
      "vm_loss": 0.1005
    },
    {
      "epoch": 0.8017828810403808,
      "lm_loss": 0.3761,
      "step": 10928,
      "vm_loss": 0.1979
    },
    {
      "epoch": 0.8017828810403808,
      "lm_loss": 0.3736,
      "step": 10928,
      "vm_loss": 0.1057
    },
    {
      "epoch": 0.8018562506305199,
      "grad_norm": 1.48696160956672,
      "learning_rate": 9.943827150864143e-07,
      "loss": 0.6829,
      "step": 10929
    },
    {
      "epoch": 0.8019296202206591,
      "grad_norm": 0.5400369622689568,
      "learning_rate": 9.936716933763618e-07,
      "loss": 0.4756,
      "step": 10930
    },
    {
      "epoch": 0.8020029898107982,
      "grad_norm": 1.6800366819447368,
      "learning_rate": 9.929608979135902e-07,
      "loss": 0.7939,
      "step": 10931
    },
    {
      "epoch": 0.8020763594009372,
      "grad_norm": 1.5024456939791637,
      "learning_rate": 9.922503287382396e-07,
      "loss": 0.6372,
      "step": 10932
    },
    {
      "epoch": 0.8021497289910764,
      "grad_norm": 0.5523708844954094,
      "learning_rate": 9.915399858904373e-07,
      "loss": 0.4937,
      "step": 10933
    },
    {
      "epoch": 0.8022230985812155,
      "grad_norm": 1.4738539175864749,
      "learning_rate": 9.908298694102997e-07,
      "loss": 0.7322,
      "step": 10934
    },
    {
      "epoch": 0.8022964681713547,
      "grad_norm": 0.5472660196852965,
      "learning_rate": 9.901199793379278e-07,
      "loss": 0.4795,
      "step": 10935
    },
    {
      "epoch": 0.8023698377614938,
      "grad_norm": 1.4628461867796445,
      "learning_rate": 9.89410315713411e-07,
      "loss": 0.6544,
      "step": 10936
    },
    {
      "epoch": 0.8023698377614938,
      "lm_loss": 1.2407,
      "step": 10936,
      "vm_loss": 0.0481
    },
    {
      "epoch": 0.8023698377614938,
      "lm_loss": 1.4046,
      "step": 10936,
      "vm_loss": 0.1802
    },
    {
      "epoch": 0.8023698377614938,
      "lm_loss": 0.1409,
      "step": 10936,
      "vm_loss": 0.1055
    },
    {
      "epoch": 0.8023698377614938,
      "lm_loss": 0.6741,
      "step": 10936,
      "vm_loss": 0.1755
    },
    {
      "epoch": 0.8023698377614938,
      "lm_loss": 0.6107,
      "step": 10936,
      "vm_loss": 0.1733
    },
    {
      "epoch": 0.8023698377614938,
      "lm_loss": 0.392,
      "step": 10936,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.8023698377614938,
      "lm_loss": 0.3036,
      "step": 10936,
      "vm_loss": 0.0645
    },
    {
      "epoch": 0.8023698377614938,
      "lm_loss": 0.8653,
      "step": 10936,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.8024432073516329,
      "grad_norm": 1.586189230011531,
      "learning_rate": 9.88700878576826e-07,
      "loss": 0.7906,
      "step": 10937
    },
    {
      "epoch": 0.802516576941772,
      "grad_norm": 1.4985701035367247,
      "learning_rate": 9.87991667968235e-07,
      "loss": 0.5993,
      "step": 10938
    },
    {
      "epoch": 0.8025899465319112,
      "grad_norm": 1.3990141125400484,
      "learning_rate": 9.87282683927691e-07,
      "loss": 0.6858,
      "step": 10939
    },
    {
      "epoch": 0.8026633161220503,
      "grad_norm": 1.524190659945796,
      "learning_rate": 9.865739264952312e-07,
      "loss": 0.7492,
      "step": 10940
    },
    {
      "epoch": 0.8027366857121895,
      "grad_norm": 1.3468613499659583,
      "learning_rate": 9.85865395710881e-07,
      "loss": 0.8568,
      "step": 10941
    },
    {
      "epoch": 0.8028100553023285,
      "grad_norm": 1.512534566159963,
      "learning_rate": 9.85157091614653e-07,
      "loss": 0.6544,
      "step": 10942
    },
    {
      "epoch": 0.8028834248924677,
      "grad_norm": 1.5141449078096492,
      "learning_rate": 9.844490142465452e-07,
      "loss": 0.6984,
      "step": 10943
    },
    {
      "epoch": 0.8029567944826068,
      "grad_norm": 1.463541291335195,
      "learning_rate": 9.837411636465472e-07,
      "loss": 0.6219,
      "step": 10944
    },
    {
      "epoch": 0.8029567944826068,
      "lm_loss": 1.2562,
      "step": 10944,
      "vm_loss": 0.1009
    },
    {
      "epoch": 0.8029567944826068,
      "lm_loss": 0.5036,
      "step": 10944,
      "vm_loss": 0.0846
    },
    {
      "epoch": 0.8029567944826068,
      "lm_loss": 0.6243,
      "step": 10944,
      "vm_loss": 0.0903
    },
    {
      "epoch": 0.8029567944826068,
      "lm_loss": 0.6186,
      "step": 10944,
      "vm_loss": 0.1117
    },
    {
      "epoch": 0.8029567944826068,
      "lm_loss": 0.1001,
      "step": 10944,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.8029567944826068,
      "lm_loss": 0.5346,
      "step": 10944,
      "vm_loss": 0.1405
    },
    {
      "epoch": 0.8029567944826068,
      "lm_loss": 0.649,
      "step": 10944,
      "vm_loss": 0.135
    },
    {
      "epoch": 0.8029567944826068,
      "lm_loss": 1.1249,
      "step": 10944,
      "vm_loss": 0.0848
    },
    {
      "epoch": 0.803030164072746,
      "grad_norm": 1.4049619940471798,
      "learning_rate": 9.830335398546314e-07,
      "loss": 0.7694,
      "step": 10945
    },
    {
      "epoch": 0.8031035336628851,
      "grad_norm": 1.5844494021621285,
      "learning_rate": 9.823261429107595e-07,
      "loss": 0.7079,
      "step": 10946
    },
    {
      "epoch": 0.8031769032530242,
      "grad_norm": 0.5659797989125157,
      "learning_rate": 9.816189728548803e-07,
      "loss": 0.4885,
      "step": 10947
    },
    {
      "epoch": 0.8032502728431633,
      "grad_norm": 0.5439366147753185,
      "learning_rate": 9.809120297269287e-07,
      "loss": 0.4642,
      "step": 10948
    },
    {
      "epoch": 0.8033236424333025,
      "grad_norm": 1.4258448912140078,
      "learning_rate": 9.80205313566827e-07,
      "loss": 0.6585,
      "step": 10949
    },
    {
      "epoch": 0.8033970120234416,
      "grad_norm": 1.3707716659146605,
      "learning_rate": 9.79498824414487e-07,
      "loss": 0.7117,
      "step": 10950
    },
    {
      "epoch": 0.8034703816135808,
      "grad_norm": 1.6857252604168655,
      "learning_rate": 9.787925623098054e-07,
      "loss": 0.7336,
      "step": 10951
    },
    {
      "epoch": 0.8035437512037198,
      "grad_norm": 1.6259380086216404,
      "learning_rate": 9.78086527292666e-07,
      "loss": 0.7317,
      "step": 10952
    },
    {
      "epoch": 0.8035437512037198,
      "lm_loss": 0.5141,
      "step": 10952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8035437512037198,
      "lm_loss": 0.5946,
      "step": 10952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8035437512037198,
      "lm_loss": 0.4278,
      "step": 10952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8035437512037198,
      "lm_loss": 0.449,
      "step": 10952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8035437512037198,
      "lm_loss": 0.4096,
      "step": 10952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8035437512037198,
      "lm_loss": 0.5161,
      "step": 10952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8035437512037198,
      "lm_loss": 0.5666,
      "step": 10952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8035437512037198,
      "lm_loss": 0.5346,
      "step": 10952,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8036171207938589,
      "grad_norm": 0.5425826645233421,
      "learning_rate": 9.7738071940294e-07,
      "loss": 0.4657,
      "step": 10953
    },
    {
      "epoch": 0.8036904903839981,
      "grad_norm": 1.40257357808579,
      "learning_rate": 9.766751386804856e-07,
      "loss": 0.7811,
      "step": 10954
    },
    {
      "epoch": 0.8037638599741372,
      "grad_norm": 1.6753003560483954,
      "learning_rate": 9.759697851651511e-07,
      "loss": 0.7621,
      "step": 10955
    },
    {
      "epoch": 0.8038372295642764,
      "grad_norm": 0.5495498099917895,
      "learning_rate": 9.752646588967685e-07,
      "loss": 0.4902,
      "step": 10956
    },
    {
      "epoch": 0.8039105991544154,
      "grad_norm": 1.4585329871194697,
      "learning_rate": 9.745597599151575e-07,
      "loss": 0.7771,
      "step": 10957
    },
    {
      "epoch": 0.8039839687445546,
      "grad_norm": 1.5212807051621096,
      "learning_rate": 9.738550882601255e-07,
      "loss": 0.8045,
      "step": 10958
    },
    {
      "epoch": 0.8040573383346937,
      "grad_norm": 1.6237080348909292,
      "learning_rate": 9.731506439714672e-07,
      "loss": 0.6987,
      "step": 10959
    },
    {
      "epoch": 0.8041307079248329,
      "grad_norm": 1.6140348193309175,
      "learning_rate": 9.724464270889634e-07,
      "loss": 0.7267,
      "step": 10960
    },
    {
      "epoch": 0.8041307079248329,
      "lm_loss": 1.2087,
      "step": 10960,
      "vm_loss": 0.0394
    },
    {
      "epoch": 0.8041307079248329,
      "lm_loss": 0.3739,
      "step": 10960,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.8041307079248329,
      "lm_loss": 0.1766,
      "step": 10960,
      "vm_loss": 0.0563
    },
    {
      "epoch": 0.8041307079248329,
      "lm_loss": 0.0491,
      "step": 10960,
      "vm_loss": 0.0826
    },
    {
      "epoch": 0.8041307079248329,
      "lm_loss": 1.1415,
      "step": 10960,
      "vm_loss": 0.1719
    },
    {
      "epoch": 0.8041307079248329,
      "lm_loss": 0.8675,
      "step": 10960,
      "vm_loss": 0.1278
    },
    {
      "epoch": 0.8041307079248329,
      "lm_loss": 0.0285,
      "step": 10960,
      "vm_loss": 0.0404
    },
    {
      "epoch": 0.8041307079248329,
      "lm_loss": 0.4491,
      "step": 10960,
      "vm_loss": 0.1599
    },
    {
      "epoch": 0.804204077514972,
      "grad_norm": 1.6422005318444786,
      "learning_rate": 9.717424376523848e-07,
      "loss": 0.6495,
      "step": 10961
    },
    {
      "epoch": 0.8042774471051111,
      "grad_norm": 0.5575157076861695,
      "learning_rate": 9.710386757014872e-07,
      "loss": 0.4721,
      "step": 10962
    },
    {
      "epoch": 0.8043508166952502,
      "grad_norm": 1.3703105471588526,
      "learning_rate": 9.703351412760126e-07,
      "loss": 0.6819,
      "step": 10963
    },
    {
      "epoch": 0.8044241862853894,
      "grad_norm": 1.4459203097645064,
      "learning_rate": 9.696318344156919e-07,
      "loss": 0.8283,
      "step": 10964
    },
    {
      "epoch": 0.8044975558755285,
      "grad_norm": 0.5427379557603583,
      "learning_rate": 9.689287551602417e-07,
      "loss": 0.4682,
      "step": 10965
    },
    {
      "epoch": 0.8045709254656677,
      "grad_norm": 0.5567663345237391,
      "learning_rate": 9.682259035493686e-07,
      "loss": 0.4889,
      "step": 10966
    },
    {
      "epoch": 0.8046442950558067,
      "grad_norm": 1.5497873746403004,
      "learning_rate": 9.675232796227629e-07,
      "loss": 0.7588,
      "step": 10967
    },
    {
      "epoch": 0.8047176646459459,
      "grad_norm": 0.5470964375609079,
      "learning_rate": 9.668208834201036e-07,
      "loss": 0.4596,
      "step": 10968
    },
    {
      "epoch": 0.8047176646459459,
      "lm_loss": 0.6341,
      "step": 10968,
      "vm_loss": 0.0304
    },
    {
      "epoch": 0.8047176646459459,
      "lm_loss": 1.2341,
      "step": 10968,
      "vm_loss": 0.1121
    },
    {
      "epoch": 0.8047176646459459,
      "lm_loss": 0.2649,
      "step": 10968,
      "vm_loss": 0.0976
    },
    {
      "epoch": 0.8047176646459459,
      "lm_loss": 0.0349,
      "step": 10968,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.8047176646459459,
      "lm_loss": 0.4512,
      "step": 10968,
      "vm_loss": 0.2439
    },
    {
      "epoch": 0.8047176646459459,
      "lm_loss": 0.9076,
      "step": 10968,
      "vm_loss": 0.204
    },
    {
      "epoch": 0.8047176646459459,
      "lm_loss": 0.8003,
      "step": 10968,
      "vm_loss": 0.098
    },
    {
      "epoch": 0.8047176646459459,
      "lm_loss": 0.2399,
      "step": 10968,
      "vm_loss": 0.0752
    },
    {
      "epoch": 0.804791034236085,
      "grad_norm": 1.4649508703301555,
      "learning_rate": 9.66118714981058e-07,
      "loss": 0.7632,
      "step": 10969
    },
    {
      "epoch": 0.8048644038262242,
      "grad_norm": 1.6150555656741075,
      "learning_rate": 9.654167743452764e-07,
      "loss": 0.7168,
      "step": 10970
    },
    {
      "epoch": 0.8049377734163633,
      "grad_norm": 1.5980550017542094,
      "learning_rate": 9.64715061552402e-07,
      "loss": 0.7379,
      "step": 10971
    },
    {
      "epoch": 0.8050111430065023,
      "grad_norm": 0.5536239908935844,
      "learning_rate": 9.640135766420621e-07,
      "loss": 0.4696,
      "step": 10972
    },
    {
      "epoch": 0.8050845125966415,
      "grad_norm": 0.5560302294242189,
      "learning_rate": 9.633123196538697e-07,
      "loss": 0.4917,
      "step": 10973
    },
    {
      "epoch": 0.8051578821867806,
      "grad_norm": 1.4483286002140263,
      "learning_rate": 9.626112906274276e-07,
      "loss": 0.7147,
      "step": 10974
    },
    {
      "epoch": 0.8052312517769198,
      "grad_norm": 1.4925136697199912,
      "learning_rate": 9.619104896023246e-07,
      "loss": 0.7034,
      "step": 10975
    },
    {
      "epoch": 0.8053046213670589,
      "grad_norm": 0.5520555122213868,
      "learning_rate": 9.61209916618135e-07,
      "loss": 0.501,
      "step": 10976
    },
    {
      "epoch": 0.8053046213670589,
      "lm_loss": 1.3048,
      "step": 10976,
      "vm_loss": 0.019
    },
    {
      "epoch": 0.8053046213670589,
      "lm_loss": 0.8926,
      "step": 10976,
      "vm_loss": 0.098
    },
    {
      "epoch": 0.8053046213670589,
      "lm_loss": 0.7692,
      "step": 10976,
      "vm_loss": 0.1023
    },
    {
      "epoch": 0.8053046213670589,
      "lm_loss": 0.3347,
      "step": 10976,
      "vm_loss": 0.1267
    },
    {
      "epoch": 0.8053046213670589,
      "lm_loss": 0.8999,
      "step": 10976,
      "vm_loss": 0.046
    },
    {
      "epoch": 0.8053046213670589,
      "lm_loss": 0.3485,
      "step": 10976,
      "vm_loss": 0.1432
    },
    {
      "epoch": 0.8053046213670589,
      "lm_loss": 0.7555,
      "step": 10976,
      "vm_loss": 0.2923
    },
    {
      "epoch": 0.8053046213670589,
      "lm_loss": 1.3862,
      "step": 10976,
      "vm_loss": 0.1904
    },
    {
      "epoch": 0.805377990957198,
      "grad_norm": 1.5914404594755887,
      "learning_rate": 9.605095717144247e-07,
      "loss": 0.7816,
      "step": 10977
    },
    {
      "epoch": 0.8054513605473371,
      "grad_norm": 1.7035585147916634,
      "learning_rate": 9.59809454930742e-07,
      "loss": 0.7158,
      "step": 10978
    },
    {
      "epoch": 0.8055247301374763,
      "grad_norm": 0.5486193234102451,
      "learning_rate": 9.591095663066253e-07,
      "loss": 0.4789,
      "step": 10979
    },
    {
      "epoch": 0.8055980997276154,
      "grad_norm": 1.307756256638083,
      "learning_rate": 9.584099058815983e-07,
      "loss": 0.7277,
      "step": 10980
    },
    {
      "epoch": 0.8056714693177546,
      "grad_norm": 1.4590367861654925,
      "learning_rate": 9.577104736951715e-07,
      "loss": 0.7782,
      "step": 10981
    },
    {
      "epoch": 0.8057448389078936,
      "grad_norm": 1.6854651711965938,
      "learning_rate": 9.570112697868456e-07,
      "loss": 0.7839,
      "step": 10982
    },
    {
      "epoch": 0.8058182084980328,
      "grad_norm": 1.5304495618145144,
      "learning_rate": 9.56312294196106e-07,
      "loss": 0.7848,
      "step": 10983
    },
    {
      "epoch": 0.8058915780881719,
      "grad_norm": 0.5613415399673296,
      "learning_rate": 9.556135469624255e-07,
      "loss": 0.4907,
      "step": 10984
    },
    {
      "epoch": 0.8058915780881719,
      "lm_loss": 1.2458,
      "step": 10984,
      "vm_loss": 0.063
    },
    {
      "epoch": 0.8058915780881719,
      "lm_loss": 1.2463,
      "step": 10984,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.8058915780881719,
      "lm_loss": 0.4335,
      "step": 10984,
      "vm_loss": 0.0826
    },
    {
      "epoch": 0.8058915780881719,
      "lm_loss": 0.0272,
      "step": 10984,
      "vm_loss": 0.1795
    },
    {
      "epoch": 0.8058915780881719,
      "lm_loss": 0.7797,
      "step": 10984,
      "vm_loss": 0.0841
    },
    {
      "epoch": 0.8058915780881719,
      "lm_loss": 0.1479,
      "step": 10984,
      "vm_loss": 0.165
    },
    {
      "epoch": 0.8058915780881719,
      "lm_loss": 0.442,
      "step": 10984,
      "vm_loss": 0.0669
    },
    {
      "epoch": 0.8058915780881719,
      "lm_loss": 0.8204,
      "step": 10984,
      "vm_loss": 0.1712
    },
    {
      "epoch": 0.8059649476783111,
      "grad_norm": 1.494629235423458,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.7805,
      "step": 10985
    },
    {
      "epoch": 0.8060383172684502,
      "grad_norm": 0.540665876223091,
      "learning_rate": 9.542167377240668e-07,
      "loss": 0.4668,
      "step": 10986
    },
    {
      "epoch": 0.8061116868585892,
      "grad_norm": 1.476139587087873,
      "learning_rate": 9.535186757982695e-07,
      "loss": 0.6754,
      "step": 10987
    },
    {
      "epoch": 0.8061850564487284,
      "grad_norm": 1.4499731473765987,
      "learning_rate": 9.528208423872948e-07,
      "loss": 0.8053,
      "step": 10988
    },
    {
      "epoch": 0.8062584260388675,
      "grad_norm": 1.5498644893033882,
      "learning_rate": 9.521232375305494e-07,
      "loss": 0.7297,
      "step": 10989
    },
    {
      "epoch": 0.8063317956290067,
      "grad_norm": 0.549194123501172,
      "learning_rate": 9.514258612674293e-07,
      "loss": 0.48,
      "step": 10990
    },
    {
      "epoch": 0.8064051652191458,
      "grad_norm": 1.4802383663851664,
      "learning_rate": 9.507287136373173e-07,
      "loss": 0.6826,
      "step": 10991
    },
    {
      "epoch": 0.8064785348092849,
      "grad_norm": 1.457036628475445,
      "learning_rate": 9.500317946795811e-07,
      "loss": 0.6865,
      "step": 10992
    },
    {
      "epoch": 0.8064785348092849,
      "lm_loss": 1.3288,
      "step": 10992,
      "vm_loss": 0.04
    },
    {
      "epoch": 0.8064785348092849,
      "lm_loss": 0.182,
      "step": 10992,
      "vm_loss": 0.0453
    },
    {
      "epoch": 0.8064785348092849,
      "lm_loss": 0.8231,
      "step": 10992,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.8064785348092849,
      "lm_loss": 0.3194,
      "step": 10992,
      "vm_loss": 0.0552
    },
    {
      "epoch": 0.8064785348092849,
      "lm_loss": 0.1381,
      "step": 10992,
      "vm_loss": 0.1739
    },
    {
      "epoch": 0.8064785348092849,
      "lm_loss": 0.2572,
      "step": 10992,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.8064785348092849,
      "lm_loss": 0.1798,
      "step": 10992,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.8064785348092849,
      "lm_loss": 0.3465,
      "step": 10992,
      "vm_loss": 0.16
    },
    {
      "epoch": 0.806551904399424,
      "grad_norm": 1.4853319484749605,
      "learning_rate": 9.493351044335802e-07,
      "loss": 0.601,
      "step": 10993
    },
    {
      "epoch": 0.8066252739895632,
      "grad_norm": 1.5598171503470084,
      "learning_rate": 9.486386429386573e-07,
      "loss": 0.841,
      "step": 10994
    },
    {
      "epoch": 0.8066986435797023,
      "grad_norm": 1.4845401145099453,
      "learning_rate": 9.479424102341439e-07,
      "loss": 0.7187,
      "step": 10995
    },
    {
      "epoch": 0.8067720131698415,
      "grad_norm": 1.6677503542629026,
      "learning_rate": 9.472464063593567e-07,
      "loss": 0.7113,
      "step": 10996
    },
    {
      "epoch": 0.8068453827599805,
      "grad_norm": 1.6817206901048147,
      "learning_rate": 9.465506313536e-07,
      "loss": 0.6718,
      "step": 10997
    },
    {
      "epoch": 0.8069187523501197,
      "grad_norm": 1.5896589014326576,
      "learning_rate": 9.458550852561687e-07,
      "loss": 0.7843,
      "step": 10998
    },
    {
      "epoch": 0.8069921219402588,
      "grad_norm": 1.4813005701871966,
      "learning_rate": 9.451597681063412e-07,
      "loss": 0.8224,
      "step": 10999
    },
    {
      "epoch": 0.807065491530398,
      "grad_norm": 1.3341249322745439,
      "learning_rate": 9.444646799433826e-07,
      "loss": 0.7599,
      "step": 11000
    },
    {
      "epoch": 0.807065491530398,
      "lm_loss": 1.5814,
      "step": 11000,
      "vm_loss": 0.0684
    },
    {
      "epoch": 0.807065491530398,
      "lm_loss": 0.0226,
      "step": 11000,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.807065491530398,
      "lm_loss": 2.1446,
      "step": 11000,
      "vm_loss": 0.1731
    },
    {
      "epoch": 0.807065491530398,
      "lm_loss": 0.5512,
      "step": 11000,
      "vm_loss": 0.2303
    },
    {
      "epoch": 0.807065491530398,
      "lm_loss": 0.2534,
      "step": 11000,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.807065491530398,
      "lm_loss": 0.4025,
      "step": 11000,
      "vm_loss": 0.1099
    },
    {
      "epoch": 0.807065491530398,
      "lm_loss": 0.1487,
      "step": 11000,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.807065491530398,
      "lm_loss": 0.1677,
      "step": 11000,
      "vm_loss": 0.1001
    },
    {
      "epoch": 0.807138861120537,
      "grad_norm": 1.3942929299707723,
      "learning_rate": 9.437698208065488e-07,
      "loss": 0.7063,
      "step": 11001
    },
    {
      "epoch": 0.8072122307106762,
      "grad_norm": 1.528009588486119,
      "learning_rate": 9.430751907350754e-07,
      "loss": 0.7337,
      "step": 11002
    },
    {
      "epoch": 0.8072856003008153,
      "grad_norm": 1.6702210827731334,
      "learning_rate": 9.423807897681941e-07,
      "loss": 0.751,
      "step": 11003
    },
    {
      "epoch": 0.8073589698909545,
      "grad_norm": 1.5638602446090262,
      "learning_rate": 9.416866179451178e-07,
      "loss": 0.7159,
      "step": 11004
    },
    {
      "epoch": 0.8074323394810936,
      "grad_norm": 1.4656514629602913,
      "learning_rate": 9.409926753050485e-07,
      "loss": 0.7729,
      "step": 11005
    },
    {
      "epoch": 0.8075057090712326,
      "grad_norm": 1.3739153648738582,
      "learning_rate": 9.402989618871755e-07,
      "loss": 0.7033,
      "step": 11006
    },
    {
      "epoch": 0.8075790786613718,
      "grad_norm": 1.4658219569106279,
      "learning_rate": 9.396054777306718e-07,
      "loss": 0.721,
      "step": 11007
    },
    {
      "epoch": 0.807652448251511,
      "grad_norm": 1.344302774162539,
      "learning_rate": 9.389122228747038e-07,
      "loss": 0.8256,
      "step": 11008
    },
    {
      "epoch": 0.807652448251511,
      "lm_loss": 0.5215,
      "step": 11008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.807652448251511,
      "lm_loss": 0.7075,
      "step": 11008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.807652448251511,
      "lm_loss": 0.3749,
      "step": 11008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.807652448251511,
      "lm_loss": 0.393,
      "step": 11008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.807652448251511,
      "lm_loss": 0.402,
      "step": 11008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.807652448251511,
      "lm_loss": 0.5359,
      "step": 11008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.807652448251511,
      "lm_loss": 0.4747,
      "step": 11008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.807652448251511,
      "lm_loss": 0.5167,
      "step": 11008,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8077258178416501,
      "grad_norm": 0.5578235225960411,
      "learning_rate": 9.382191973584193e-07,
      "loss": 0.4981,
      "step": 11009
    },
    {
      "epoch": 0.8077991874317892,
      "grad_norm": 1.6436857369951137,
      "learning_rate": 9.375264012209556e-07,
      "loss": 0.6593,
      "step": 11010
    },
    {
      "epoch": 0.8078725570219283,
      "grad_norm": 1.5242023345409028,
      "learning_rate": 9.368338345014371e-07,
      "loss": 0.7623,
      "step": 11011
    },
    {
      "epoch": 0.8079459266120674,
      "grad_norm": 1.5555017862999874,
      "learning_rate": 9.361414972389727e-07,
      "loss": 0.7511,
      "step": 11012
    },
    {
      "epoch": 0.8080192962022066,
      "grad_norm": 1.4061125171144548,
      "learning_rate": 9.35449389472663e-07,
      "loss": 0.8777,
      "step": 11013
    },
    {
      "epoch": 0.8080926657923457,
      "grad_norm": 1.4531331586715672,
      "learning_rate": 9.347575112415919e-07,
      "loss": 0.7293,
      "step": 11014
    },
    {
      "epoch": 0.8081660353824849,
      "grad_norm": 1.5295258225091144,
      "learning_rate": 9.34065862584832e-07,
      "loss": 0.7005,
      "step": 11015
    },
    {
      "epoch": 0.8082394049726239,
      "grad_norm": 0.5495111096552637,
      "learning_rate": 9.333744435414415e-07,
      "loss": 0.483,
      "step": 11016
    },
    {
      "epoch": 0.8082394049726239,
      "lm_loss": 1.2698,
      "step": 11016,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.8082394049726239,
      "lm_loss": 0.8166,
      "step": 11016,
      "vm_loss": 0.1675
    },
    {
      "epoch": 0.8082394049726239,
      "lm_loss": 0.2603,
      "step": 11016,
      "vm_loss": 0.0544
    },
    {
      "epoch": 0.8082394049726239,
      "lm_loss": 0.9657,
      "step": 11016,
      "vm_loss": 0.2016
    },
    {
      "epoch": 0.8082394049726239,
      "lm_loss": 0.3081,
      "step": 11016,
      "vm_loss": 0.0502
    },
    {
      "epoch": 0.8082394049726239,
      "lm_loss": 0.4707,
      "step": 11016,
      "vm_loss": 0.145
    },
    {
      "epoch": 0.8082394049726239,
      "lm_loss": 0.4773,
      "step": 11016,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.8082394049726239,
      "lm_loss": 0.9049,
      "step": 11016,
      "vm_loss": 0.2511
    },
    {
      "epoch": 0.8083127745627631,
      "grad_norm": 1.435544007871035,
      "learning_rate": 9.326832541504672e-07,
      "loss": 0.6776,
      "step": 11017
    },
    {
      "epoch": 0.8083861441529022,
      "grad_norm": 1.4921003490105942,
      "learning_rate": 9.319922944509408e-07,
      "loss": 0.6968,
      "step": 11018
    },
    {
      "epoch": 0.8084595137430414,
      "grad_norm": 2.1207421039987038,
      "learning_rate": 9.313015644818851e-07,
      "loss": 0.8077,
      "step": 11019
    },
    {
      "epoch": 0.8085328833331805,
      "grad_norm": 1.5602010926058503,
      "learning_rate": 9.306110642823063e-07,
      "loss": 0.6966,
      "step": 11020
    },
    {
      "epoch": 0.8086062529233196,
      "grad_norm": 1.490758156897666,
      "learning_rate": 9.29920793891198e-07,
      "loss": 0.7767,
      "step": 11021
    },
    {
      "epoch": 0.8086796225134587,
      "grad_norm": 1.5852785914519738,
      "learning_rate": 9.292307533475414e-07,
      "loss": 0.7628,
      "step": 11022
    },
    {
      "epoch": 0.8087529921035979,
      "grad_norm": 1.4203567160713284,
      "learning_rate": 9.285409426903047e-07,
      "loss": 0.6474,
      "step": 11023
    },
    {
      "epoch": 0.808826361693737,
      "grad_norm": 1.720753504661206,
      "learning_rate": 9.278513619584451e-07,
      "loss": 0.7605,
      "step": 11024
    },
    {
      "epoch": 0.808826361693737,
      "lm_loss": 0.1515,
      "step": 11024,
      "vm_loss": 0.0336
    },
    {
      "epoch": 0.808826361693737,
      "lm_loss": 1.3624,
      "step": 11024,
      "vm_loss": 0.096
    },
    {
      "epoch": 0.808826361693737,
      "lm_loss": 0.3164,
      "step": 11024,
      "vm_loss": 0.0931
    },
    {
      "epoch": 0.808826361693737,
      "lm_loss": 0.5172,
      "step": 11024,
      "vm_loss": 0.1927
    },
    {
      "epoch": 0.808826361693737,
      "lm_loss": 0.6608,
      "step": 11024,
      "vm_loss": 0.187
    },
    {
      "epoch": 0.808826361693737,
      "lm_loss": 0.3602,
      "step": 11024,
      "vm_loss": 0.1964
    },
    {
      "epoch": 0.808826361693737,
      "lm_loss": 0.8768,
      "step": 11024,
      "vm_loss": 0.1211
    },
    {
      "epoch": 0.808826361693737,
      "lm_loss": 0.8382,
      "step": 11024,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.8088997312838762,
      "grad_norm": 1.4756271239250935,
      "learning_rate": 9.27162011190903e-07,
      "loss": 0.7744,
      "step": 11025
    },
    {
      "epoch": 0.8089731008740152,
      "grad_norm": 1.800871257557437,
      "learning_rate": 9.264728904266085e-07,
      "loss": 0.7855,
      "step": 11026
    },
    {
      "epoch": 0.8090464704641543,
      "grad_norm": 1.4625218541721532,
      "learning_rate": 9.257839997044782e-07,
      "loss": 0.6313,
      "step": 11027
    },
    {
      "epoch": 0.8091198400542935,
      "grad_norm": 1.6818243440389902,
      "learning_rate": 9.250953390634149e-07,
      "loss": 0.6958,
      "step": 11028
    },
    {
      "epoch": 0.8091932096444326,
      "grad_norm": 0.5554545846009435,
      "learning_rate": 9.244069085423074e-07,
      "loss": 0.4968,
      "step": 11029
    },
    {
      "epoch": 0.8092665792345718,
      "grad_norm": 0.549547673192703,
      "learning_rate": 9.237187081800364e-07,
      "loss": 0.4811,
      "step": 11030
    },
    {
      "epoch": 0.8093399488247108,
      "grad_norm": 1.5709790193957474,
      "learning_rate": 9.230307380154651e-07,
      "loss": 0.6687,
      "step": 11031
    },
    {
      "epoch": 0.80941331841485,
      "grad_norm": 1.6324783947976604,
      "learning_rate": 9.223429980874437e-07,
      "loss": 0.7913,
      "step": 11032
    },
    {
      "epoch": 0.80941331841485,
      "lm_loss": 1.484,
      "step": 11032,
      "vm_loss": 0.081
    },
    {
      "epoch": 0.80941331841485,
      "lm_loss": 0.2906,
      "step": 11032,
      "vm_loss": 0.0547
    },
    {
      "epoch": 0.80941331841485,
      "lm_loss": 0.8631,
      "step": 11032,
      "vm_loss": 0.2514
    },
    {
      "epoch": 0.80941331841485,
      "lm_loss": 0.7251,
      "step": 11032,
      "vm_loss": 0.0833
    },
    {
      "epoch": 0.80941331841485,
      "lm_loss": 0.1337,
      "step": 11032,
      "vm_loss": 0.0737
    },
    {
      "epoch": 0.80941331841485,
      "lm_loss": 0.1856,
      "step": 11032,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.80941331841485,
      "lm_loss": 0.3627,
      "step": 11032,
      "vm_loss": 0.1372
    },
    {
      "epoch": 0.80941331841485,
      "lm_loss": 0.063,
      "step": 11032,
      "vm_loss": 0.0579
    },
    {
      "epoch": 0.8094866880049891,
      "grad_norm": 1.4337777654943074,
      "learning_rate": 9.216554884348122e-07,
      "loss": 0.6703,
      "step": 11033
    },
    {
      "epoch": 0.8095600575951283,
      "grad_norm": 1.80726208085467,
      "learning_rate": 9.209682090963934e-07,
      "loss": 0.7192,
      "step": 11034
    },
    {
      "epoch": 0.8096334271852674,
      "grad_norm": 1.6585576088478773,
      "learning_rate": 9.202811601110023e-07,
      "loss": 0.7144,
      "step": 11035
    },
    {
      "epoch": 0.8097067967754065,
      "grad_norm": 1.4373918469764468,
      "learning_rate": 9.19594341517438e-07,
      "loss": 0.6899,
      "step": 11036
    },
    {
      "epoch": 0.8097801663655456,
      "grad_norm": 1.5421027196134094,
      "learning_rate": 9.18907753354486e-07,
      "loss": 0.804,
      "step": 11037
    },
    {
      "epoch": 0.8098535359556848,
      "grad_norm": 1.607443927002543,
      "learning_rate": 9.182213956609193e-07,
      "loss": 0.7428,
      "step": 11038
    },
    {
      "epoch": 0.8099269055458239,
      "grad_norm": 0.5375763413201791,
      "learning_rate": 9.175352684754979e-07,
      "loss": 0.4716,
      "step": 11039
    },
    {
      "epoch": 0.8100002751359631,
      "grad_norm": 1.5766579704552854,
      "learning_rate": 9.168493718369715e-07,
      "loss": 0.7613,
      "step": 11040
    },
    {
      "epoch": 0.8100002751359631,
      "lm_loss": 1.1809,
      "step": 11040,
      "vm_loss": 0.078
    },
    {
      "epoch": 0.8100002751359631,
      "lm_loss": 0.1266,
      "step": 11040,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.8100002751359631,
      "lm_loss": 0.0625,
      "step": 11040,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.8100002751359631,
      "lm_loss": 0.7537,
      "step": 11040,
      "vm_loss": 0.1274
    },
    {
      "epoch": 0.8100002751359631,
      "lm_loss": 0.0238,
      "step": 11040,
      "vm_loss": 0.1238
    },
    {
      "epoch": 0.8100002751359631,
      "lm_loss": 0.5557,
      "step": 11040,
      "vm_loss": 0.0777
    },
    {
      "epoch": 0.8100002751359631,
      "lm_loss": 0.6117,
      "step": 11040,
      "vm_loss": 0.1135
    },
    {
      "epoch": 0.8100002751359631,
      "lm_loss": 0.4403,
      "step": 11040,
      "vm_loss": 0.1059
    },
    {
      "epoch": 0.8100736447261021,
      "grad_norm": 1.455670665217289,
      "learning_rate": 9.161637057840722e-07,
      "loss": 0.624,
      "step": 11041
    },
    {
      "epoch": 0.8101470143162413,
      "grad_norm": 1.448374343396915,
      "learning_rate": 9.154782703555226e-07,
      "loss": 0.8294,
      "step": 11042
    },
    {
      "epoch": 0.8102203839063804,
      "grad_norm": 1.5751515248317085,
      "learning_rate": 9.147930655900294e-07,
      "loss": 0.7394,
      "step": 11043
    },
    {
      "epoch": 0.8102937534965196,
      "grad_norm": 1.495261888848657,
      "learning_rate": 9.14108091526289e-07,
      "loss": 0.7763,
      "step": 11044
    },
    {
      "epoch": 0.8103671230866587,
      "grad_norm": 1.5905965362572272,
      "learning_rate": 9.134233482029826e-07,
      "loss": 0.7189,
      "step": 11045
    },
    {
      "epoch": 0.8104404926767977,
      "grad_norm": 1.7744883189875222,
      "learning_rate": 9.127388356587802e-07,
      "loss": 0.6771,
      "step": 11046
    },
    {
      "epoch": 0.8105138622669369,
      "grad_norm": 1.5388922959548992,
      "learning_rate": 9.120545539323383e-07,
      "loss": 0.7476,
      "step": 11047
    },
    {
      "epoch": 0.810587231857076,
      "grad_norm": 0.5442031884111627,
      "learning_rate": 9.113705030622988e-07,
      "loss": 0.4713,
      "step": 11048
    },
    {
      "epoch": 0.810587231857076,
      "lm_loss": 0.6306,
      "step": 11048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.810587231857076,
      "lm_loss": 0.4955,
      "step": 11048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.810587231857076,
      "lm_loss": 0.6449,
      "step": 11048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.810587231857076,
      "lm_loss": 0.4468,
      "step": 11048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.810587231857076,
      "lm_loss": 0.4048,
      "step": 11048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.810587231857076,
      "lm_loss": 0.6278,
      "step": 11048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.810587231857076,
      "lm_loss": 0.5652,
      "step": 11048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.810587231857076,
      "lm_loss": 0.4195,
      "step": 11048,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8106606014472152,
      "grad_norm": 0.571695501867604,
      "learning_rate": 9.106866830872929e-07,
      "loss": 0.4927,
      "step": 11049
    },
    {
      "epoch": 0.8107339710373543,
      "grad_norm": 1.5110491796670897,
      "learning_rate": 9.100030940459353e-07,
      "loss": 0.7554,
      "step": 11050
    },
    {
      "epoch": 0.8108073406274934,
      "grad_norm": 1.4560172919858163,
      "learning_rate": 9.093197359768336e-07,
      "loss": 0.6937,
      "step": 11051
    },
    {
      "epoch": 0.8108807102176325,
      "grad_norm": 1.525320852831117,
      "learning_rate": 9.086366089185761e-07,
      "loss": 0.7167,
      "step": 11052
    },
    {
      "epoch": 0.8109540798077717,
      "grad_norm": 1.689268682673821,
      "learning_rate": 9.079537129097421e-07,
      "loss": 0.7825,
      "step": 11053
    },
    {
      "epoch": 0.8110274493979108,
      "grad_norm": 1.5731801477038638,
      "learning_rate": 9.072710479888952e-07,
      "loss": 0.7603,
      "step": 11054
    },
    {
      "epoch": 0.81110081898805,
      "grad_norm": 1.5210050595117997,
      "learning_rate": 9.065886141945878e-07,
      "loss": 0.8174,
      "step": 11055
    },
    {
      "epoch": 0.811174188578189,
      "grad_norm": 1.4994194736584658,
      "learning_rate": 9.059064115653571e-07,
      "loss": 0.8529,
      "step": 11056
    },
    {
      "epoch": 0.811174188578189,
      "lm_loss": 0.577,
      "step": 11056,
      "vm_loss": 0.0299
    },
    {
      "epoch": 0.811174188578189,
      "lm_loss": 1.3736,
      "step": 11056,
      "vm_loss": 0.0791
    },
    {
      "epoch": 0.811174188578189,
      "lm_loss": 0.7982,
      "step": 11056,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.811174188578189,
      "lm_loss": 0.5787,
      "step": 11056,
      "vm_loss": 0.079
    },
    {
      "epoch": 0.811174188578189,
      "lm_loss": 0.3816,
      "step": 11056,
      "vm_loss": 0.1794
    },
    {
      "epoch": 0.811174188578189,
      "lm_loss": 0.6998,
      "step": 11056,
      "vm_loss": 0.218
    },
    {
      "epoch": 0.811174188578189,
      "lm_loss": 0.4487,
      "step": 11056,
      "vm_loss": 0.1548
    },
    {
      "epoch": 0.811174188578189,
      "lm_loss": 0.0878,
      "step": 11056,
      "vm_loss": 0.1922
    },
    {
      "epoch": 0.8112475581683282,
      "grad_norm": 1.4669061102920535,
      "learning_rate": 9.052244401397314e-07,
      "loss": 0.6935,
      "step": 11057
    },
    {
      "epoch": 0.8113209277584673,
      "grad_norm": 1.468577769365713,
      "learning_rate": 9.045426999562223e-07,
      "loss": 0.773,
      "step": 11058
    },
    {
      "epoch": 0.8113942973486065,
      "grad_norm": 1.5897727682346832,
      "learning_rate": 9.038611910533296e-07,
      "loss": 0.7832,
      "step": 11059
    },
    {
      "epoch": 0.8114676669387456,
      "grad_norm": 1.51184924798624,
      "learning_rate": 9.031799134695385e-07,
      "loss": 0.7087,
      "step": 11060
    },
    {
      "epoch": 0.8115410365288847,
      "grad_norm": 1.7266732796451085,
      "learning_rate": 9.024988672433227e-07,
      "loss": 0.7931,
      "step": 11061
    },
    {
      "epoch": 0.8116144061190238,
      "grad_norm": 0.5494891766234614,
      "learning_rate": 9.018180524131443e-07,
      "loss": 0.4842,
      "step": 11062
    },
    {
      "epoch": 0.811687775709163,
      "grad_norm": 0.5520355697830731,
      "learning_rate": 9.01137469017449e-07,
      "loss": 0.4687,
      "step": 11063
    },
    {
      "epoch": 0.8117611452993021,
      "grad_norm": 0.5294617227223763,
      "learning_rate": 9.004571170946719e-07,
      "loss": 0.4628,
      "step": 11064
    },
    {
      "epoch": 0.8117611452993021,
      "lm_loss": 0.6074,
      "step": 11064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8117611452993021,
      "lm_loss": 0.5343,
      "step": 11064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8117611452993021,
      "lm_loss": 0.538,
      "step": 11064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8117611452993021,
      "lm_loss": 0.4401,
      "step": 11064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8117611452993021,
      "lm_loss": 0.5215,
      "step": 11064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8117611452993021,
      "lm_loss": 0.5183,
      "step": 11064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8117611452993021,
      "lm_loss": 0.412,
      "step": 11064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8117611452993021,
      "lm_loss": 0.5336,
      "step": 11064,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8118345148894412,
      "grad_norm": 0.5464972464475838,
      "learning_rate": 8.997769966832338e-07,
      "loss": 0.492,
      "step": 11065
    },
    {
      "epoch": 0.8119078844795803,
      "grad_norm": 1.567565021787443,
      "learning_rate": 8.99097107821541e-07,
      "loss": 0.7323,
      "step": 11066
    },
    {
      "epoch": 0.8119812540697194,
      "grad_norm": 1.5890328365971793,
      "learning_rate": 8.98417450547992e-07,
      "loss": 0.7704,
      "step": 11067
    },
    {
      "epoch": 0.8120546236598586,
      "grad_norm": 0.5378794935134932,
      "learning_rate": 8.977380249009671e-07,
      "loss": 0.4678,
      "step": 11068
    },
    {
      "epoch": 0.8121279932499977,
      "grad_norm": 1.4337121920028597,
      "learning_rate": 8.970588309188343e-07,
      "loss": 0.7074,
      "step": 11069
    },
    {
      "epoch": 0.8122013628401369,
      "grad_norm": 1.467491304641095,
      "learning_rate": 8.963798686399505e-07,
      "loss": 0.6743,
      "step": 11070
    },
    {
      "epoch": 0.8122747324302759,
      "grad_norm": 0.5499859589451808,
      "learning_rate": 8.957011381026581e-07,
      "loss": 0.4812,
      "step": 11071
    },
    {
      "epoch": 0.8123481020204151,
      "grad_norm": 1.473563215782143,
      "learning_rate": 8.950226393452849e-07,
      "loss": 0.6777,
      "step": 11072
    },
    {
      "epoch": 0.8123481020204151,
      "lm_loss": 0.5347,
      "step": 11072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8123481020204151,
      "lm_loss": 0.6297,
      "step": 11072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8123481020204151,
      "lm_loss": 0.4959,
      "step": 11072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8123481020204151,
      "lm_loss": 0.5095,
      "step": 11072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8123481020204151,
      "lm_loss": 0.434,
      "step": 11072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8123481020204151,
      "lm_loss": 0.6045,
      "step": 11072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8123481020204151,
      "lm_loss": 0.479,
      "step": 11072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8123481020204151,
      "lm_loss": 0.4328,
      "step": 11072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8124214716105542,
      "grad_norm": 0.5444049978703543,
      "learning_rate": 8.943443724061507e-07,
      "loss": 0.4981,
      "step": 11073
    },
    {
      "epoch": 0.8124948412006934,
      "grad_norm": 1.700762318134068,
      "learning_rate": 8.936663373235571e-07,
      "loss": 0.7141,
      "step": 11074
    },
    {
      "epoch": 0.8125682107908325,
      "grad_norm": 1.5340240555652436,
      "learning_rate": 8.929885341357947e-07,
      "loss": 0.7877,
      "step": 11075
    },
    {
      "epoch": 0.8126415803809716,
      "grad_norm": 1.5030658776549934,
      "learning_rate": 8.923109628811405e-07,
      "loss": 0.6614,
      "step": 11076
    },
    {
      "epoch": 0.8127149499711107,
      "grad_norm": 0.5412508676297827,
      "learning_rate": 8.916336235978573e-07,
      "loss": 0.4797,
      "step": 11077
    },
    {
      "epoch": 0.8127883195612499,
      "grad_norm": 1.476486064059799,
      "learning_rate": 8.909565163241996e-07,
      "loss": 0.7633,
      "step": 11078
    },
    {
      "epoch": 0.812861689151389,
      "grad_norm": 1.4198839125573115,
      "learning_rate": 8.902796410984027e-07,
      "loss": 0.7238,
      "step": 11079
    },
    {
      "epoch": 0.8129350587415282,
      "grad_norm": 0.5531644451459238,
      "learning_rate": 8.896029979586918e-07,
      "loss": 0.4972,
      "step": 11080
    },
    {
      "epoch": 0.8129350587415282,
      "lm_loss": 1.2517,
      "step": 11080,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.8129350587415282,
      "lm_loss": 0.0238,
      "step": 11080,
      "vm_loss": 0.2225
    },
    {
      "epoch": 0.8129350587415282,
      "lm_loss": 0.0974,
      "step": 11080,
      "vm_loss": 0.0735
    },
    {
      "epoch": 0.8129350587415282,
      "lm_loss": 0.7524,
      "step": 11080,
      "vm_loss": 0.1834
    },
    {
      "epoch": 0.8129350587415282,
      "lm_loss": 0.5956,
      "step": 11080,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.8129350587415282,
      "lm_loss": 0.0468,
      "step": 11080,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.8129350587415282,
      "lm_loss": 0.0664,
      "step": 11080,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.8129350587415282,
      "lm_loss": 0.883,
      "step": 11080,
      "vm_loss": 0.151
    },
    {
      "epoch": 0.8130084283316672,
      "grad_norm": 2.0998542748276914,
      "learning_rate": 8.889265869432795e-07,
      "loss": 0.7308,
      "step": 11081
    },
    {
      "epoch": 0.8130817979218063,
      "grad_norm": 1.5374891329256524,
      "learning_rate": 8.882504080903631e-07,
      "loss": 0.7734,
      "step": 11082
    },
    {
      "epoch": 0.8131551675119455,
      "grad_norm": 1.342852500181063,
      "learning_rate": 8.875744614381277e-07,
      "loss": 0.6868,
      "step": 11083
    },
    {
      "epoch": 0.8132285371020846,
      "grad_norm": 1.4830228007363286,
      "learning_rate": 8.868987470247481e-07,
      "loss": 0.751,
      "step": 11084
    },
    {
      "epoch": 0.8133019066922237,
      "grad_norm": 1.7107917981183307,
      "learning_rate": 8.862232648883817e-07,
      "loss": 0.7512,
      "step": 11085
    },
    {
      "epoch": 0.8133752762823628,
      "grad_norm": 0.5654676333308581,
      "learning_rate": 8.855480150671758e-07,
      "loss": 0.5058,
      "step": 11086
    },
    {
      "epoch": 0.813448645872502,
      "grad_norm": 0.5528279679030375,
      "learning_rate": 8.848729975992625e-07,
      "loss": 0.47,
      "step": 11087
    },
    {
      "epoch": 0.8135220154626411,
      "grad_norm": 0.5458071986352385,
      "learning_rate": 8.841982125227605e-07,
      "loss": 0.4759,
      "step": 11088
    },
    {
      "epoch": 0.8135220154626411,
      "lm_loss": 1.2767,
      "step": 11088,
      "vm_loss": 0.0534
    },
    {
      "epoch": 0.8135220154626411,
      "lm_loss": 0.4937,
      "step": 11088,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.8135220154626411,
      "lm_loss": 0.0411,
      "step": 11088,
      "vm_loss": 0.152
    },
    {
      "epoch": 0.8135220154626411,
      "lm_loss": 0.8147,
      "step": 11088,
      "vm_loss": 0.121
    },
    {
      "epoch": 0.8135220154626411,
      "lm_loss": 0.6474,
      "step": 11088,
      "vm_loss": 0.1783
    },
    {
      "epoch": 0.8135220154626411,
      "lm_loss": 1.1047,
      "step": 11088,
      "vm_loss": 0.163
    },
    {
      "epoch": 0.8135220154626411,
      "lm_loss": 0.7196,
      "step": 11088,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.8135220154626411,
      "lm_loss": 0.0739,
      "step": 11088,
      "vm_loss": 0.1093
    },
    {
      "epoch": 0.8135953850527803,
      "grad_norm": 1.4983914534731355,
      "learning_rate": 8.835236598757796e-07,
      "loss": 0.6487,
      "step": 11089
    },
    {
      "epoch": 0.8136687546429193,
      "grad_norm": 1.5826305251016848,
      "learning_rate": 8.828493396964116e-07,
      "loss": 0.7481,
      "step": 11090
    },
    {
      "epoch": 0.8137421242330585,
      "grad_norm": 1.8104770523368965,
      "learning_rate": 8.821752520227378e-07,
      "loss": 0.7805,
      "step": 11091
    },
    {
      "epoch": 0.8138154938231976,
      "grad_norm": 1.3942746062874374,
      "learning_rate": 8.815013968928249e-07,
      "loss": 0.7002,
      "step": 11092
    },
    {
      "epoch": 0.8138888634133368,
      "grad_norm": 1.4476178980546652,
      "learning_rate": 8.808277743447257e-07,
      "loss": 0.7382,
      "step": 11093
    },
    {
      "epoch": 0.8139622330034759,
      "grad_norm": 0.5571388295461847,
      "learning_rate": 8.80154384416485e-07,
      "loss": 0.4763,
      "step": 11094
    },
    {
      "epoch": 0.814035602593615,
      "grad_norm": 0.5360967502539999,
      "learning_rate": 8.794812271461284e-07,
      "loss": 0.455,
      "step": 11095
    },
    {
      "epoch": 0.8141089721837541,
      "grad_norm": 1.4106279768161805,
      "learning_rate": 8.788083025716715e-07,
      "loss": 0.6014,
      "step": 11096
    },
    {
      "epoch": 0.8141089721837541,
      "lm_loss": 0.5894,
      "step": 11096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8141089721837541,
      "lm_loss": 0.4176,
      "step": 11096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8141089721837541,
      "lm_loss": 0.4025,
      "step": 11096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8141089721837541,
      "lm_loss": 0.3317,
      "step": 11096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8141089721837541,
      "lm_loss": 0.4745,
      "step": 11096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8141089721837541,
      "lm_loss": 0.575,
      "step": 11096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8141089721837541,
      "lm_loss": 0.5664,
      "step": 11096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8141089721837541,
      "lm_loss": 0.5881,
      "step": 11096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8141823417738933,
      "grad_norm": 0.5659147994535892,
      "learning_rate": 8.781356107311157e-07,
      "loss": 0.4911,
      "step": 11097
    },
    {
      "epoch": 0.8142557113640324,
      "grad_norm": 1.4910979982619808,
      "learning_rate": 8.774631516624493e-07,
      "loss": 0.7664,
      "step": 11098
    },
    {
      "epoch": 0.8143290809541716,
      "grad_norm": 1.567902951708359,
      "learning_rate": 8.767909254036472e-07,
      "loss": 0.7489,
      "step": 11099
    },
    {
      "epoch": 0.8144024505443106,
      "grad_norm": 1.5340013660531204,
      "learning_rate": 8.761189319926733e-07,
      "loss": 0.8177,
      "step": 11100
    },
    {
      "epoch": 0.8144758201344497,
      "grad_norm": 1.7036675735869282,
      "learning_rate": 8.754471714674761e-07,
      "loss": 0.764,
      "step": 11101
    },
    {
      "epoch": 0.8145491897245889,
      "grad_norm": 0.5350891589035778,
      "learning_rate": 8.747756438659916e-07,
      "loss": 0.4642,
      "step": 11102
    },
    {
      "epoch": 0.814622559314728,
      "grad_norm": 1.5015481427763515,
      "learning_rate": 8.741043492261419e-07,
      "loss": 0.7675,
      "step": 11103
    },
    {
      "epoch": 0.8146959289048672,
      "grad_norm": 1.579560858082562,
      "learning_rate": 8.734332875858365e-07,
      "loss": 0.7527,
      "step": 11104
    },
    {
      "epoch": 0.8146959289048672,
      "lm_loss": 1.359,
      "step": 11104,
      "vm_loss": 0.0641
    },
    {
      "epoch": 0.8146959289048672,
      "lm_loss": 0.2491,
      "step": 11104,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.8146959289048672,
      "lm_loss": 0.0518,
      "step": 11104,
      "vm_loss": 0.1362
    },
    {
      "epoch": 0.8146959289048672,
      "lm_loss": 0.7965,
      "step": 11104,
      "vm_loss": 0.17
    },
    {
      "epoch": 0.8146959289048672,
      "lm_loss": 0.303,
      "step": 11104,
      "vm_loss": 0.196
    },
    {
      "epoch": 0.8146959289048672,
      "lm_loss": 1.0751,
      "step": 11104,
      "vm_loss": 0.2356
    },
    {
      "epoch": 0.8146959289048672,
      "lm_loss": 0.0463,
      "step": 11104,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.8146959289048672,
      "lm_loss": 0.0826,
      "step": 11104,
      "vm_loss": 0.0383
    },
    {
      "epoch": 0.8147692984950062,
      "grad_norm": 1.5105655433391076,
      "learning_rate": 8.727624589829736e-07,
      "loss": 0.7941,
      "step": 11105
    },
    {
      "epoch": 0.8148426680851454,
      "grad_norm": 0.5366261146328103,
      "learning_rate": 8.720918634554354e-07,
      "loss": 0.4739,
      "step": 11106
    },
    {
      "epoch": 0.8149160376752845,
      "grad_norm": 1.4210979944659925,
      "learning_rate": 8.714215010410925e-07,
      "loss": 0.713,
      "step": 11107
    },
    {
      "epoch": 0.8149894072654237,
      "grad_norm": 0.5357677270011227,
      "learning_rate": 8.707513717778015e-07,
      "loss": 0.4806,
      "step": 11108
    },
    {
      "epoch": 0.8150627768555628,
      "grad_norm": 1.5729437876223566,
      "learning_rate": 8.70081475703406e-07,
      "loss": 0.7118,
      "step": 11109
    },
    {
      "epoch": 0.8151361464457019,
      "grad_norm": 1.3687362255976663,
      "learning_rate": 8.694118128557378e-07,
      "loss": 0.748,
      "step": 11110
    },
    {
      "epoch": 0.815209516035841,
      "grad_norm": 0.5479230190479315,
      "learning_rate": 8.687423832726133e-07,
      "loss": 0.4709,
      "step": 11111
    },
    {
      "epoch": 0.8152828856259802,
      "grad_norm": 1.2793777950640914,
      "learning_rate": 8.680731869918379e-07,
      "loss": 0.7036,
      "step": 11112
    },
    {
      "epoch": 0.8152828856259802,
      "lm_loss": 1.349,
      "step": 11112,
      "vm_loss": 0.0743
    },
    {
      "epoch": 0.8152828856259802,
      "lm_loss": 0.0147,
      "step": 11112,
      "vm_loss": 0.1521
    },
    {
      "epoch": 0.8152828856259802,
      "lm_loss": 0.0257,
      "step": 11112,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.8152828856259802,
      "lm_loss": 0.2995,
      "step": 11112,
      "vm_loss": 0.1047
    },
    {
      "epoch": 0.8152828856259802,
      "lm_loss": 0.3626,
      "step": 11112,
      "vm_loss": 0.0789
    },
    {
      "epoch": 0.8152828856259802,
      "lm_loss": 0.9203,
      "step": 11112,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.8152828856259802,
      "lm_loss": 0.5746,
      "step": 11112,
      "vm_loss": 0.1403
    },
    {
      "epoch": 0.8152828856259802,
      "lm_loss": 0.2637,
      "step": 11112,
      "vm_loss": 0.0838
    },
    {
      "epoch": 0.8153562552161193,
      "grad_norm": 1.5586656433157904,
      "learning_rate": 8.674042240512021e-07,
      "loss": 0.6656,
      "step": 11113
    },
    {
      "epoch": 0.8154296248062585,
      "grad_norm": 0.5362733223023256,
      "learning_rate": 8.667354944884837e-07,
      "loss": 0.4803,
      "step": 11114
    },
    {
      "epoch": 0.8155029943963975,
      "grad_norm": 1.560836579792026,
      "learning_rate": 8.660669983414466e-07,
      "loss": 0.7258,
      "step": 11115
    },
    {
      "epoch": 0.8155763639865367,
      "grad_norm": 0.5548159157789154,
      "learning_rate": 8.65398735647845e-07,
      "loss": 0.4886,
      "step": 11116
    },
    {
      "epoch": 0.8156497335766758,
      "grad_norm": 1.4128858685140284,
      "learning_rate": 8.647307064454158e-07,
      "loss": 0.6757,
      "step": 11117
    },
    {
      "epoch": 0.815723103166815,
      "grad_norm": 1.717329560133763,
      "learning_rate": 8.640629107718845e-07,
      "loss": 0.7577,
      "step": 11118
    },
    {
      "epoch": 0.8157964727569541,
      "grad_norm": 1.6819643499059416,
      "learning_rate": 8.633953486649632e-07,
      "loss": 0.8429,
      "step": 11119
    },
    {
      "epoch": 0.8158698423470931,
      "grad_norm": 1.36105167930261,
      "learning_rate": 8.627280201623494e-07,
      "loss": 0.6514,
      "step": 11120
    },
    {
      "epoch": 0.8158698423470931,
      "lm_loss": 0.9442,
      "step": 11120,
      "vm_loss": 0.0666
    },
    {
      "epoch": 0.8158698423470931,
      "lm_loss": 1.2462,
      "step": 11120,
      "vm_loss": 0.2266
    },
    {
      "epoch": 0.8158698423470931,
      "lm_loss": 0.9244,
      "step": 11120,
      "vm_loss": 0.1782
    },
    {
      "epoch": 0.8158698423470931,
      "lm_loss": 0.3284,
      "step": 11120,
      "vm_loss": 0.1047
    },
    {
      "epoch": 0.8158698423470931,
      "lm_loss": 0.1936,
      "step": 11120,
      "vm_loss": 0.1667
    },
    {
      "epoch": 0.8158698423470931,
      "lm_loss": 1.0116,
      "step": 11120,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.8158698423470931,
      "lm_loss": 0.0848,
      "step": 11120,
      "vm_loss": 0.1018
    },
    {
      "epoch": 0.8158698423470931,
      "lm_loss": 0.0912,
      "step": 11120,
      "vm_loss": 0.1523
    },
    {
      "epoch": 0.8159432119372323,
      "grad_norm": 1.3957464441563594,
      "learning_rate": 8.620609253017315e-07,
      "loss": 0.7082,
      "step": 11121
    },
    {
      "epoch": 0.8160165815273714,
      "grad_norm": 1.684459556681822,
      "learning_rate": 8.613940641207807e-07,
      "loss": 0.8063,
      "step": 11122
    },
    {
      "epoch": 0.8160899511175106,
      "grad_norm": 1.5607940818886301,
      "learning_rate": 8.607274366571556e-07,
      "loss": 0.7464,
      "step": 11123
    },
    {
      "epoch": 0.8161633207076497,
      "grad_norm": 1.5185677713003476,
      "learning_rate": 8.600610429485034e-07,
      "loss": 0.7447,
      "step": 11124
    },
    {
      "epoch": 0.8162366902977888,
      "grad_norm": 1.5284942553092429,
      "learning_rate": 8.593948830324561e-07,
      "loss": 0.7275,
      "step": 11125
    },
    {
      "epoch": 0.8163100598879279,
      "grad_norm": 0.5590444082916333,
      "learning_rate": 8.587289569466328e-07,
      "loss": 0.4756,
      "step": 11126
    },
    {
      "epoch": 0.8163834294780671,
      "grad_norm": 1.4495728638974976,
      "learning_rate": 8.580632647286424e-07,
      "loss": 0.7623,
      "step": 11127
    },
    {
      "epoch": 0.8164567990682062,
      "grad_norm": 1.6037196541480454,
      "learning_rate": 8.573978064160765e-07,
      "loss": 0.7941,
      "step": 11128
    },
    {
      "epoch": 0.8164567990682062,
      "lm_loss": 1.2305,
      "step": 11128,
      "vm_loss": 0.0969
    },
    {
      "epoch": 0.8164567990682062,
      "lm_loss": 0.2664,
      "step": 11128,
      "vm_loss": 0.09
    },
    {
      "epoch": 0.8164567990682062,
      "lm_loss": 0.0321,
      "step": 11128,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.8164567990682062,
      "lm_loss": 0.2289,
      "step": 11128,
      "vm_loss": 0.0807
    },
    {
      "epoch": 0.8164567990682062,
      "lm_loss": 0.9414,
      "step": 11128,
      "vm_loss": 0.0688
    },
    {
      "epoch": 0.8164567990682062,
      "lm_loss": 0.7392,
      "step": 11128,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.8164567990682062,
      "lm_loss": 0.7518,
      "step": 11128,
      "vm_loss": 0.2157
    },
    {
      "epoch": 0.8164567990682062,
      "lm_loss": 1.1667,
      "step": 11128,
      "vm_loss": 0.1302
    },
    {
      "epoch": 0.8165301686583454,
      "grad_norm": 1.4210430586180556,
      "learning_rate": 8.567325820465156e-07,
      "loss": 0.7085,
      "step": 11129
    },
    {
      "epoch": 0.8166035382484844,
      "grad_norm": 1.5548054134955744,
      "learning_rate": 8.560675916575262e-07,
      "loss": 0.8401,
      "step": 11130
    },
    {
      "epoch": 0.8166769078386236,
      "grad_norm": 0.5488467350900652,
      "learning_rate": 8.554028352866612e-07,
      "loss": 0.4772,
      "step": 11131
    },
    {
      "epoch": 0.8167502774287627,
      "grad_norm": 1.462749024884501,
      "learning_rate": 8.547383129714626e-07,
      "loss": 0.7345,
      "step": 11132
    },
    {
      "epoch": 0.8168236470189019,
      "grad_norm": 0.542103596492375,
      "learning_rate": 8.540740247494577e-07,
      "loss": 0.4803,
      "step": 11133
    },
    {
      "epoch": 0.816897016609041,
      "grad_norm": 0.5433539978112145,
      "learning_rate": 8.5340997065816e-07,
      "loss": 0.4788,
      "step": 11134
    },
    {
      "epoch": 0.81697038619918,
      "grad_norm": 1.6068172310883422,
      "learning_rate": 8.527461507350693e-07,
      "loss": 0.7608,
      "step": 11135
    },
    {
      "epoch": 0.8170437557893192,
      "grad_norm": 1.399880779795872,
      "learning_rate": 8.520825650176734e-07,
      "loss": 0.7384,
      "step": 11136
    },
    {
      "epoch": 0.8170437557893192,
      "lm_loss": 1.1828,
      "step": 11136,
      "vm_loss": 0.0728
    },
    {
      "epoch": 0.8170437557893192,
      "lm_loss": 1.074,
      "step": 11136,
      "vm_loss": 0.235
    },
    {
      "epoch": 0.8170437557893192,
      "lm_loss": 0.4334,
      "step": 11136,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.8170437557893192,
      "lm_loss": 0.2519,
      "step": 11136,
      "vm_loss": 0.0955
    },
    {
      "epoch": 0.8170437557893192,
      "lm_loss": 0.1505,
      "step": 11136,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.8170437557893192,
      "lm_loss": 0.8286,
      "step": 11136,
      "vm_loss": 0.042
    },
    {
      "epoch": 0.8170437557893192,
      "lm_loss": 0.3807,
      "step": 11136,
      "vm_loss": 0.1409
    },
    {
      "epoch": 0.8170437557893192,
      "lm_loss": 0.7946,
      "step": 11136,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.8171171253794584,
      "grad_norm": 1.4483136700884383,
      "learning_rate": 8.514192135434484e-07,
      "loss": 0.722,
      "step": 11137
    },
    {
      "epoch": 0.8171904949695975,
      "grad_norm": 0.5656199436522795,
      "learning_rate": 8.50756096349854e-07,
      "loss": 0.4829,
      "step": 11138
    },
    {
      "epoch": 0.8172638645597367,
      "grad_norm": 1.5854884107082374,
      "learning_rate": 8.500932134743384e-07,
      "loss": 0.7332,
      "step": 11139
    },
    {
      "epoch": 0.8173372341498757,
      "grad_norm": 0.5578534110135903,
      "learning_rate": 8.49430564954336e-07,
      "loss": 0.4689,
      "step": 11140
    },
    {
      "epoch": 0.8174106037400148,
      "grad_norm": 0.5562845354628893,
      "learning_rate": 8.487681508272683e-07,
      "loss": 0.4792,
      "step": 11141
    },
    {
      "epoch": 0.817483973330154,
      "grad_norm": 1.4841888852999856,
      "learning_rate": 8.481059711305423e-07,
      "loss": 0.7177,
      "step": 11142
    },
    {
      "epoch": 0.8175573429202931,
      "grad_norm": 1.702284921445018,
      "learning_rate": 8.474440259015554e-07,
      "loss": 0.7216,
      "step": 11143
    },
    {
      "epoch": 0.8176307125104323,
      "grad_norm": 0.5410322471349178,
      "learning_rate": 8.467823151776877e-07,
      "loss": 0.4735,
      "step": 11144
    },
    {
      "epoch": 0.8176307125104323,
      "lm_loss": 0.1941,
      "step": 11144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8176307125104323,
      "lm_loss": 0.5206,
      "step": 11144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8176307125104323,
      "lm_loss": 0.5747,
      "step": 11144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8176307125104323,
      "lm_loss": 0.5335,
      "step": 11144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8176307125104323,
      "lm_loss": 0.5589,
      "step": 11144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8176307125104323,
      "lm_loss": 0.5125,
      "step": 11144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8176307125104323,
      "lm_loss": 0.3652,
      "step": 11144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8176307125104323,
      "lm_loss": 0.5311,
      "step": 11144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8177040821005713,
      "grad_norm": 0.5445370354865964,
      "learning_rate": 8.461208389963082e-07,
      "loss": 0.4731,
      "step": 11145
    },
    {
      "epoch": 0.8177774516907105,
      "grad_norm": 1.5986919788158627,
      "learning_rate": 8.454595973947715e-07,
      "loss": 0.7411,
      "step": 11146
    },
    {
      "epoch": 0.8178508212808496,
      "grad_norm": 0.5472242502382918,
      "learning_rate": 8.447985904104189e-07,
      "loss": 0.486,
      "step": 11147
    },
    {
      "epoch": 0.8179241908709888,
      "grad_norm": 0.5343183909555259,
      "learning_rate": 8.441378180805815e-07,
      "loss": 0.4726,
      "step": 11148
    },
    {
      "epoch": 0.8179975604611279,
      "grad_norm": 1.438816130589131,
      "learning_rate": 8.434772804425734e-07,
      "loss": 0.7529,
      "step": 11149
    },
    {
      "epoch": 0.818070930051267,
      "grad_norm": 1.3227452932278736,
      "learning_rate": 8.428169775336964e-07,
      "loss": 0.8069,
      "step": 11150
    },
    {
      "epoch": 0.8181442996414061,
      "grad_norm": 1.5241428329842641,
      "learning_rate": 8.421569093912407e-07,
      "loss": 0.6847,
      "step": 11151
    },
    {
      "epoch": 0.8182176692315453,
      "grad_norm": 1.5017414931982573,
      "learning_rate": 8.414970760524793e-07,
      "loss": 0.7741,
      "step": 11152
    },
    {
      "epoch": 0.8182176692315453,
      "lm_loss": 1.3968,
      "step": 11152,
      "vm_loss": 0.0545
    },
    {
      "epoch": 0.8182176692315453,
      "lm_loss": 0.0605,
      "step": 11152,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.8182176692315453,
      "lm_loss": 0.0668,
      "step": 11152,
      "vm_loss": 0.112
    },
    {
      "epoch": 0.8182176692315453,
      "lm_loss": 0.7592,
      "step": 11152,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.8182176692315453,
      "lm_loss": 0.9085,
      "step": 11152,
      "vm_loss": 0.1908
    },
    {
      "epoch": 0.8182176692315453,
      "lm_loss": 0.5059,
      "step": 11152,
      "vm_loss": 0.2129
    },
    {
      "epoch": 0.8182176692315453,
      "lm_loss": 0.5113,
      "step": 11152,
      "vm_loss": 0.2203
    },
    {
      "epoch": 0.8182176692315453,
      "lm_loss": 0.1327,
      "step": 11152,
      "vm_loss": 0.1118
    },
    {
      "epoch": 0.8182910388216844,
      "grad_norm": 1.6194267198131613,
      "learning_rate": 8.408374775546774e-07,
      "loss": 0.7369,
      "step": 11153
    },
    {
      "epoch": 0.8183644084118236,
      "grad_norm": 1.4922947520782492,
      "learning_rate": 8.401781139350829e-07,
      "loss": 0.6536,
      "step": 11154
    },
    {
      "epoch": 0.8184377780019626,
      "grad_norm": 1.4349001051617811,
      "learning_rate": 8.395189852309321e-07,
      "loss": 0.7243,
      "step": 11155
    },
    {
      "epoch": 0.8185111475921018,
      "grad_norm": 0.531881297412374,
      "learning_rate": 8.38860091479447e-07,
      "loss": 0.4559,
      "step": 11156
    },
    {
      "epoch": 0.8185845171822409,
      "grad_norm": 1.5262952551977966,
      "learning_rate": 8.382014327178367e-07,
      "loss": 0.6656,
      "step": 11157
    },
    {
      "epoch": 0.81865788677238,
      "grad_norm": 1.4531790342297035,
      "learning_rate": 8.375430089832992e-07,
      "loss": 0.7556,
      "step": 11158
    },
    {
      "epoch": 0.8187312563625192,
      "grad_norm": 1.3985772801134146,
      "learning_rate": 8.368848203130159e-07,
      "loss": 0.6287,
      "step": 11159
    },
    {
      "epoch": 0.8188046259526582,
      "grad_norm": 0.544925394353497,
      "learning_rate": 8.36226866744157e-07,
      "loss": 0.4763,
      "step": 11160
    },
    {
      "epoch": 0.8188046259526582,
      "lm_loss": 0.5258,
      "step": 11160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8188046259526582,
      "lm_loss": 0.4132,
      "step": 11160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8188046259526582,
      "lm_loss": 0.3662,
      "step": 11160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8188046259526582,
      "lm_loss": 0.4719,
      "step": 11160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8188046259526582,
      "lm_loss": 0.3356,
      "step": 11160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8188046259526582,
      "lm_loss": 0.4632,
      "step": 11160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8188046259526582,
      "lm_loss": 0.4822,
      "step": 11160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8188046259526582,
      "lm_loss": 0.4052,
      "step": 11160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8188779955427974,
      "grad_norm": 0.5460163830891963,
      "learning_rate": 8.355691483138778e-07,
      "loss": 0.482,
      "step": 11161
    },
    {
      "epoch": 0.8189513651329365,
      "grad_norm": 1.6458427470739214,
      "learning_rate": 8.349116650593209e-07,
      "loss": 0.736,
      "step": 11162
    },
    {
      "epoch": 0.8190247347230757,
      "grad_norm": 1.7572465883592474,
      "learning_rate": 8.342544170176186e-07,
      "loss": 0.775,
      "step": 11163
    },
    {
      "epoch": 0.8190981043132147,
      "grad_norm": 0.5741009515021008,
      "learning_rate": 8.335974042258849e-07,
      "loss": 0.4739,
      "step": 11164
    },
    {
      "epoch": 0.8191714739033539,
      "grad_norm": 1.409821954954181,
      "learning_rate": 8.329406267212242e-07,
      "loss": 0.7624,
      "step": 11165
    },
    {
      "epoch": 0.819244843493493,
      "grad_norm": 1.6130100011055288,
      "learning_rate": 8.32284084540726e-07,
      "loss": 0.7426,
      "step": 11166
    },
    {
      "epoch": 0.8193182130836322,
      "grad_norm": 0.5351395302943627,
      "learning_rate": 8.31627777721466e-07,
      "loss": 0.4611,
      "step": 11167
    },
    {
      "epoch": 0.8193915826737713,
      "grad_norm": 1.7243127718498343,
      "learning_rate": 8.309717063005074e-07,
      "loss": 0.7347,
      "step": 11168
    },
    {
      "epoch": 0.8193915826737713,
      "lm_loss": 0.2609,
      "step": 11168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8193915826737713,
      "lm_loss": 0.3874,
      "step": 11168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8193915826737713,
      "lm_loss": 0.5227,
      "step": 11168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8193915826737713,
      "lm_loss": 0.3712,
      "step": 11168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8193915826737713,
      "lm_loss": 0.4869,
      "step": 11168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8193915826737713,
      "lm_loss": 0.5051,
      "step": 11168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8193915826737713,
      "lm_loss": 0.4709,
      "step": 11168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8193915826737713,
      "lm_loss": 0.4876,
      "step": 11168,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8194649522639104,
      "grad_norm": 0.5298264020542226,
      "learning_rate": 8.303158703149023e-07,
      "loss": 0.4533,
      "step": 11169
    },
    {
      "epoch": 0.8195383218540495,
      "grad_norm": 0.5354824278382173,
      "learning_rate": 8.296602698016859e-07,
      "loss": 0.469,
      "step": 11170
    },
    {
      "epoch": 0.8196116914441887,
      "grad_norm": 1.3783348239570128,
      "learning_rate": 8.290049047978816e-07,
      "loss": 0.6802,
      "step": 11171
    },
    {
      "epoch": 0.8196850610343278,
      "grad_norm": 1.5388780581566621,
      "learning_rate": 8.283497753405e-07,
      "loss": 0.6703,
      "step": 11172
    },
    {
      "epoch": 0.819758430624467,
      "grad_norm": 1.4150445246178727,
      "learning_rate": 8.276948814665359e-07,
      "loss": 0.7146,
      "step": 11173
    },
    {
      "epoch": 0.819831800214606,
      "grad_norm": 1.581679099145745,
      "learning_rate": 8.270402232129759e-07,
      "loss": 0.7671,
      "step": 11174
    },
    {
      "epoch": 0.8199051698047451,
      "grad_norm": 1.5184886425378121,
      "learning_rate": 8.263858006167885e-07,
      "loss": 0.716,
      "step": 11175
    },
    {
      "epoch": 0.8199785393948843,
      "grad_norm": 1.4922086154810152,
      "learning_rate": 8.257316137149307e-07,
      "loss": 0.7684,
      "step": 11176
    },
    {
      "epoch": 0.8199785393948843,
      "lm_loss": 1.2438,
      "step": 11176,
      "vm_loss": 0.0526
    },
    {
      "epoch": 0.8199785393948843,
      "lm_loss": 1.5594,
      "step": 11176,
      "vm_loss": 0.1359
    },
    {
      "epoch": 0.8199785393948843,
      "lm_loss": 0.2082,
      "step": 11176,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.8199785393948843,
      "lm_loss": 0.5246,
      "step": 11176,
      "vm_loss": 0.1844
    },
    {
      "epoch": 0.8199785393948843,
      "lm_loss": 0.5861,
      "step": 11176,
      "vm_loss": 0.1566
    },
    {
      "epoch": 0.8199785393948843,
      "lm_loss": 0.0151,
      "step": 11176,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.8199785393948843,
      "lm_loss": 0.6871,
      "step": 11176,
      "vm_loss": 0.1187
    },
    {
      "epoch": 0.8199785393948843,
      "lm_loss": 0.1756,
      "step": 11176,
      "vm_loss": 0.0874
    },
    {
      "epoch": 0.8200519089850234,
      "grad_norm": 1.393859970613914,
      "learning_rate": 8.250776625443463e-07,
      "loss": 0.7381,
      "step": 11177
    },
    {
      "epoch": 0.8201252785751626,
      "grad_norm": 1.4355652363533384,
      "learning_rate": 8.244239471419652e-07,
      "loss": 0.7472,
      "step": 11178
    },
    {
      "epoch": 0.8201986481653016,
      "grad_norm": 1.4059610497242225,
      "learning_rate": 8.237704675447028e-07,
      "loss": 0.7202,
      "step": 11179
    },
    {
      "epoch": 0.8202720177554408,
      "grad_norm": 1.4993881933968998,
      "learning_rate": 8.231172237894664e-07,
      "loss": 0.7314,
      "step": 11180
    },
    {
      "epoch": 0.8203453873455799,
      "grad_norm": 0.5683022983220439,
      "learning_rate": 8.224642159131435e-07,
      "loss": 0.4908,
      "step": 11181
    },
    {
      "epoch": 0.8204187569357191,
      "grad_norm": 0.563078962983909,
      "learning_rate": 8.21811443952612e-07,
      "loss": 0.4806,
      "step": 11182
    },
    {
      "epoch": 0.8204921265258582,
      "grad_norm": 1.3379073761073652,
      "learning_rate": 8.211589079447357e-07,
      "loss": 0.7241,
      "step": 11183
    },
    {
      "epoch": 0.8205654961159973,
      "grad_norm": 1.6128289112093583,
      "learning_rate": 8.205066079263629e-07,
      "loss": 0.6909,
      "step": 11184
    },
    {
      "epoch": 0.8205654961159973,
      "lm_loss": 1.2992,
      "step": 11184,
      "vm_loss": 0.0689
    },
    {
      "epoch": 0.8205654961159973,
      "lm_loss": 0.6794,
      "step": 11184,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.8205654961159973,
      "lm_loss": 0.0797,
      "step": 11184,
      "vm_loss": 0.1332
    },
    {
      "epoch": 0.8205654961159973,
      "lm_loss": 0.6512,
      "step": 11184,
      "vm_loss": 0.2285
    },
    {
      "epoch": 0.8205654961159973,
      "lm_loss": 0.7587,
      "step": 11184,
      "vm_loss": 0.1265
    },
    {
      "epoch": 0.8205654961159973,
      "lm_loss": 0.2585,
      "step": 11184,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.8205654961159973,
      "lm_loss": 0.768,
      "step": 11184,
      "vm_loss": 0.1833
    },
    {
      "epoch": 0.8205654961159973,
      "lm_loss": 0.1401,
      "step": 11184,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.8206388657061364,
      "grad_norm": 1.541385678363937,
      "learning_rate": 8.198545439343341e-07,
      "loss": 0.7201,
      "step": 11185
    },
    {
      "epoch": 0.8207122352962756,
      "grad_norm": 0.5615903141380569,
      "learning_rate": 8.192027160054705e-07,
      "loss": 0.4769,
      "step": 11186
    },
    {
      "epoch": 0.8207856048864147,
      "grad_norm": 1.6462514204633576,
      "learning_rate": 8.185511241765836e-07,
      "loss": 0.758,
      "step": 11187
    },
    {
      "epoch": 0.8208589744765539,
      "grad_norm": 0.5373555312600443,
      "learning_rate": 8.178997684844697e-07,
      "loss": 0.4804,
      "step": 11188
    },
    {
      "epoch": 0.8209323440666929,
      "grad_norm": 1.4664538465793215,
      "learning_rate": 8.172486489659115e-07,
      "loss": 0.69,
      "step": 11189
    },
    {
      "epoch": 0.8210057136568321,
      "grad_norm": 1.3957078684336646,
      "learning_rate": 8.165977656576818e-07,
      "loss": 0.7277,
      "step": 11190
    },
    {
      "epoch": 0.8210790832469712,
      "grad_norm": 1.4181446213759694,
      "learning_rate": 8.159471185965362e-07,
      "loss": 0.6688,
      "step": 11191
    },
    {
      "epoch": 0.8211524528371104,
      "grad_norm": 1.5975011492268099,
      "learning_rate": 8.152967078192187e-07,
      "loss": 0.8139,
      "step": 11192
    },
    {
      "epoch": 0.8211524528371104,
      "lm_loss": 1.4487,
      "step": 11192,
      "vm_loss": 0.0437
    },
    {
      "epoch": 0.8211524528371104,
      "lm_loss": 0.8443,
      "step": 11192,
      "vm_loss": 0.1851
    },
    {
      "epoch": 0.8211524528371104,
      "lm_loss": 0.1228,
      "step": 11192,
      "vm_loss": 0.1182
    },
    {
      "epoch": 0.8211524528371104,
      "lm_loss": 1.0799,
      "step": 11192,
      "vm_loss": 0.1676
    },
    {
      "epoch": 0.8211524528371104,
      "lm_loss": 0.7903,
      "step": 11192,
      "vm_loss": 0.1537
    },
    {
      "epoch": 0.8211524528371104,
      "lm_loss": 0.375,
      "step": 11192,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.8211524528371104,
      "lm_loss": 1.3385,
      "step": 11192,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.8211524528371104,
      "lm_loss": 0.4798,
      "step": 11192,
      "vm_loss": 0.2032
    },
    {
      "epoch": 0.8212258224272495,
      "grad_norm": 1.6521501712055215,
      "learning_rate": 8.146465333624587e-07,
      "loss": 0.8048,
      "step": 11193
    },
    {
      "epoch": 0.8212991920173885,
      "grad_norm": 1.608887763011693,
      "learning_rate": 8.139965952629742e-07,
      "loss": 0.7221,
      "step": 11194
    },
    {
      "epoch": 0.8213725616075277,
      "grad_norm": 0.5508421626866133,
      "learning_rate": 8.133468935574673e-07,
      "loss": 0.4737,
      "step": 11195
    },
    {
      "epoch": 0.8214459311976668,
      "grad_norm": 1.509337689805714,
      "learning_rate": 8.126974282826305e-07,
      "loss": 0.6716,
      "step": 11196
    },
    {
      "epoch": 0.821519300787806,
      "grad_norm": 1.6037461584269568,
      "learning_rate": 8.120481994751395e-07,
      "loss": 0.8048,
      "step": 11197
    },
    {
      "epoch": 0.8215926703779451,
      "grad_norm": 1.7165790376711343,
      "learning_rate": 8.113992071716581e-07,
      "loss": 0.6814,
      "step": 11198
    },
    {
      "epoch": 0.8216660399680842,
      "grad_norm": 1.5644668265066453,
      "learning_rate": 8.107504514088365e-07,
      "loss": 0.743,
      "step": 11199
    },
    {
      "epoch": 0.8217394095582233,
      "grad_norm": 0.5459102418365247,
      "learning_rate": 8.1010193222331e-07,
      "loss": 0.4653,
      "step": 11200
    },
    {
      "epoch": 0.8217394095582233,
      "lm_loss": 1.191,
      "step": 11200,
      "vm_loss": 0.0591
    },
    {
      "epoch": 0.8217394095582233,
      "lm_loss": 0.3238,
      "step": 11200,
      "vm_loss": 0.1358
    },
    {
      "epoch": 0.8217394095582233,
      "lm_loss": 0.7678,
      "step": 11200,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.8217394095582233,
      "lm_loss": 0.7575,
      "step": 11200,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.8217394095582233,
      "lm_loss": 1.1098,
      "step": 11200,
      "vm_loss": 0.2002
    },
    {
      "epoch": 0.8217394095582233,
      "lm_loss": 0.3501,
      "step": 11200,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.8217394095582233,
      "lm_loss": 0.6693,
      "step": 11200,
      "vm_loss": 0.1375
    },
    {
      "epoch": 0.8217394095582233,
      "lm_loss": 0.9387,
      "step": 11200,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.8218127791483625,
      "grad_norm": 1.5542929211366068,
      "learning_rate": 8.094536496517046e-07,
      "loss": 0.6685,
      "step": 11201
    },
    {
      "epoch": 0.8218861487385016,
      "grad_norm": 1.4793072316370761,
      "learning_rate": 8.088056037306296e-07,
      "loss": 0.677,
      "step": 11202
    },
    {
      "epoch": 0.8219595183286408,
      "grad_norm": 0.5569970604581722,
      "learning_rate": 8.081577944966812e-07,
      "loss": 0.4894,
      "step": 11203
    },
    {
      "epoch": 0.8220328879187798,
      "grad_norm": 1.6392963606056112,
      "learning_rate": 8.075102219864434e-07,
      "loss": 0.7911,
      "step": 11204
    },
    {
      "epoch": 0.822106257508919,
      "grad_norm": 1.5799306378559002,
      "learning_rate": 8.068628862364847e-07,
      "loss": 0.6929,
      "step": 11205
    },
    {
      "epoch": 0.8221796270990581,
      "grad_norm": 0.5532594264597275,
      "learning_rate": 8.062157872833636e-07,
      "loss": 0.4852,
      "step": 11206
    },
    {
      "epoch": 0.8222529966891973,
      "grad_norm": 1.4170470761474814,
      "learning_rate": 8.055689251636234e-07,
      "loss": 0.7327,
      "step": 11207
    },
    {
      "epoch": 0.8223263662793364,
      "grad_norm": 1.6332000835303007,
      "learning_rate": 8.049222999137929e-07,
      "loss": 0.803,
      "step": 11208
    },
    {
      "epoch": 0.8223263662793364,
      "lm_loss": 1.0355,
      "step": 11208,
      "vm_loss": 0.0337
    },
    {
      "epoch": 0.8223263662793364,
      "lm_loss": 1.0145,
      "step": 11208,
      "vm_loss": 0.1901
    },
    {
      "epoch": 0.8223263662793364,
      "lm_loss": 0.6911,
      "step": 11208,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.8223263662793364,
      "lm_loss": 0.8632,
      "step": 11208,
      "vm_loss": 0.1886
    },
    {
      "epoch": 0.8223263662793364,
      "lm_loss": 0.6798,
      "step": 11208,
      "vm_loss": 0.0753
    },
    {
      "epoch": 0.8223263662793364,
      "lm_loss": 0.4019,
      "step": 11208,
      "vm_loss": 0.1963
    },
    {
      "epoch": 0.8223263662793364,
      "lm_loss": 0.297,
      "step": 11208,
      "vm_loss": 0.0953
    },
    {
      "epoch": 0.8223263662793364,
      "lm_loss": 0.8289,
      "step": 11208,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.8223997358694755,
      "grad_norm": 1.5683635129374172,
      "learning_rate": 8.042759115703891e-07,
      "loss": 0.791,
      "step": 11209
    },
    {
      "epoch": 0.8224731054596146,
      "grad_norm": 1.4737537402744683,
      "learning_rate": 8.03629760169915e-07,
      "loss": 0.7307,
      "step": 11210
    },
    {
      "epoch": 0.8225464750497538,
      "grad_norm": 1.4632496817709404,
      "learning_rate": 8.029838457488597e-07,
      "loss": 0.7084,
      "step": 11211
    },
    {
      "epoch": 0.8226198446398929,
      "grad_norm": 1.5338965520439103,
      "learning_rate": 8.023381683437015e-07,
      "loss": 0.7333,
      "step": 11212
    },
    {
      "epoch": 0.822693214230032,
      "grad_norm": 0.5548158131895725,
      "learning_rate": 8.01692727990902e-07,
      "loss": 0.4767,
      "step": 11213
    },
    {
      "epoch": 0.8227665838201711,
      "grad_norm": 0.5474121419464336,
      "learning_rate": 8.010475247269111e-07,
      "loss": 0.4657,
      "step": 11214
    },
    {
      "epoch": 0.8228399534103102,
      "grad_norm": 0.5591623381855343,
      "learning_rate": 8.004025585881653e-07,
      "loss": 0.4778,
      "step": 11215
    },
    {
      "epoch": 0.8229133230004494,
      "grad_norm": 1.5886117306879726,
      "learning_rate": 7.997578296110853e-07,
      "loss": 0.6764,
      "step": 11216
    },
    {
      "epoch": 0.8229133230004494,
      "lm_loss": 1.2663,
      "step": 11216,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.8229133230004494,
      "lm_loss": 1.8793,
      "step": 11216,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.8229133230004494,
      "lm_loss": 0.5756,
      "step": 11216,
      "vm_loss": 0.0556
    },
    {
      "epoch": 0.8229133230004494,
      "lm_loss": 0.8557,
      "step": 11216,
      "vm_loss": 0.1437
    },
    {
      "epoch": 0.8229133230004494,
      "lm_loss": 1.069,
      "step": 11216,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.8229133230004494,
      "lm_loss": 0.6313,
      "step": 11216,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.8229133230004494,
      "lm_loss": 0.9966,
      "step": 11216,
      "vm_loss": 0.0779
    },
    {
      "epoch": 0.8229133230004494,
      "lm_loss": 0.7833,
      "step": 11216,
      "vm_loss": 0.0992
    },
    {
      "epoch": 0.8229866925905885,
      "grad_norm": 1.6079982350769135,
      "learning_rate": 7.991133378320842e-07,
      "loss": 0.8259,
      "step": 11217
    },
    {
      "epoch": 0.8230600621807277,
      "grad_norm": 1.4572827401441668,
      "learning_rate": 7.984690832875558e-07,
      "loss": 0.7126,
      "step": 11218
    },
    {
      "epoch": 0.8231334317708667,
      "grad_norm": 0.5564937036794819,
      "learning_rate": 7.978250660138837e-07,
      "loss": 0.478,
      "step": 11219
    },
    {
      "epoch": 0.8232068013610059,
      "grad_norm": 0.5580235831442455,
      "learning_rate": 7.971812860474365e-07,
      "loss": 0.4962,
      "step": 11220
    },
    {
      "epoch": 0.823280170951145,
      "grad_norm": 1.3738992552659892,
      "learning_rate": 7.965377434245697e-07,
      "loss": 0.6773,
      "step": 11221
    },
    {
      "epoch": 0.8233535405412842,
      "grad_norm": 1.4675035086253514,
      "learning_rate": 7.958944381816253e-07,
      "loss": 0.7416,
      "step": 11222
    },
    {
      "epoch": 0.8234269101314233,
      "grad_norm": 0.5464169682344858,
      "learning_rate": 7.952513703549347e-07,
      "loss": 0.4767,
      "step": 11223
    },
    {
      "epoch": 0.8235002797215624,
      "grad_norm": 1.5091860626375442,
      "learning_rate": 7.946085399808118e-07,
      "loss": 0.7255,
      "step": 11224
    },
    {
      "epoch": 0.8235002797215624,
      "lm_loss": 0.2993,
      "step": 11224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8235002797215624,
      "lm_loss": 0.3878,
      "step": 11224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8235002797215624,
      "lm_loss": 0.3572,
      "step": 11224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8235002797215624,
      "lm_loss": 0.4629,
      "step": 11224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8235002797215624,
      "lm_loss": 0.1773,
      "step": 11224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8235002797215624,
      "lm_loss": 0.5202,
      "step": 11224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8235002797215624,
      "lm_loss": 0.5039,
      "step": 11224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8235002797215624,
      "lm_loss": 0.3789,
      "step": 11224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8235736493117015,
      "grad_norm": 0.5321297805251156,
      "learning_rate": 7.939659470955596e-07,
      "loss": 0.4738,
      "step": 11225
    },
    {
      "epoch": 0.8236470189018407,
      "grad_norm": 1.4568705376764264,
      "learning_rate": 7.933235917354659e-07,
      "loss": 0.7321,
      "step": 11226
    },
    {
      "epoch": 0.8237203884919798,
      "grad_norm": 1.6087120898984084,
      "learning_rate": 7.926814739368055e-07,
      "loss": 0.7324,
      "step": 11227
    },
    {
      "epoch": 0.823793758082119,
      "grad_norm": 0.562898994210371,
      "learning_rate": 7.920395937358433e-07,
      "loss": 0.486,
      "step": 11228
    },
    {
      "epoch": 0.823867127672258,
      "grad_norm": 1.505413613362434,
      "learning_rate": 7.913979511688252e-07,
      "loss": 0.7431,
      "step": 11229
    },
    {
      "epoch": 0.8239404972623972,
      "grad_norm": 1.4695285413950547,
      "learning_rate": 7.907565462719879e-07,
      "loss": 0.7275,
      "step": 11230
    },
    {
      "epoch": 0.8240138668525363,
      "grad_norm": 1.5758456414055488,
      "learning_rate": 7.901153790815519e-07,
      "loss": 0.7085,
      "step": 11231
    },
    {
      "epoch": 0.8240872364426755,
      "grad_norm": 0.5367077697062256,
      "learning_rate": 7.894744496337254e-07,
      "loss": 0.4639,
      "step": 11232
    },
    {
      "epoch": 0.8240872364426755,
      "lm_loss": 0.3087,
      "step": 11232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8240872364426755,
      "lm_loss": 0.3073,
      "step": 11232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8240872364426755,
      "lm_loss": 0.5568,
      "step": 11232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8240872364426755,
      "lm_loss": 0.5417,
      "step": 11232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8240872364426755,
      "lm_loss": 0.4,
      "step": 11232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8240872364426755,
      "lm_loss": 0.5361,
      "step": 11232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8240872364426755,
      "lm_loss": 0.5912,
      "step": 11232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8240872364426755,
      "lm_loss": 0.521,
      "step": 11232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8241606060328146,
      "grad_norm": 0.5466053147599457,
      "learning_rate": 7.888337579647048e-07,
      "loss": 0.4722,
      "step": 11233
    },
    {
      "epoch": 0.8242339756229536,
      "grad_norm": 1.5088995510085177,
      "learning_rate": 7.88193304110671e-07,
      "loss": 0.6981,
      "step": 11234
    },
    {
      "epoch": 0.8243073452130928,
      "grad_norm": 1.585845031189674,
      "learning_rate": 7.875530881077908e-07,
      "loss": 0.7636,
      "step": 11235
    },
    {
      "epoch": 0.8243807148032319,
      "grad_norm": 1.5256153301178497,
      "learning_rate": 7.869131099922206e-07,
      "loss": 0.7918,
      "step": 11236
    },
    {
      "epoch": 0.8244540843933711,
      "grad_norm": 1.4403761163620157,
      "learning_rate": 7.862733698001002e-07,
      "loss": 0.6714,
      "step": 11237
    },
    {
      "epoch": 0.8245274539835102,
      "grad_norm": 0.5497362318165111,
      "learning_rate": 7.856338675675567e-07,
      "loss": 0.4693,
      "step": 11238
    },
    {
      "epoch": 0.8246008235736493,
      "grad_norm": 1.637782882080292,
      "learning_rate": 7.849946033307066e-07,
      "loss": 0.6536,
      "step": 11239
    },
    {
      "epoch": 0.8246741931637884,
      "grad_norm": 0.5675817608952531,
      "learning_rate": 7.843555771256495e-07,
      "loss": 0.4906,
      "step": 11240
    },
    {
      "epoch": 0.8246741931637884,
      "lm_loss": 0.4888,
      "step": 11240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8246741931637884,
      "lm_loss": 0.5229,
      "step": 11240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8246741931637884,
      "lm_loss": 0.4175,
      "step": 11240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8246741931637884,
      "lm_loss": 0.4935,
      "step": 11240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8246741931637884,
      "lm_loss": 0.5428,
      "step": 11240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8246741931637884,
      "lm_loss": 0.4354,
      "step": 11240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8246741931637884,
      "lm_loss": 0.3458,
      "step": 11240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8246741931637884,
      "lm_loss": 0.5077,
      "step": 11240,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8247475627539276,
      "grad_norm": 0.5488291864390209,
      "learning_rate": 7.837167889884728e-07,
      "loss": 0.4614,
      "step": 11241
    },
    {
      "epoch": 0.8248209323440667,
      "grad_norm": 1.5104150504232676,
      "learning_rate": 7.830782389552505e-07,
      "loss": 0.75,
      "step": 11242
    },
    {
      "epoch": 0.8248943019342058,
      "grad_norm": 0.5520303440961745,
      "learning_rate": 7.824399270620415e-07,
      "loss": 0.485,
      "step": 11243
    },
    {
      "epoch": 0.8249676715243449,
      "grad_norm": 1.8038708084765531,
      "learning_rate": 7.818018533448962e-07,
      "loss": 0.7218,
      "step": 11244
    },
    {
      "epoch": 0.8250410411144841,
      "grad_norm": 0.5635099925365232,
      "learning_rate": 7.81164017839846e-07,
      "loss": 0.4892,
      "step": 11245
    },
    {
      "epoch": 0.8251144107046232,
      "grad_norm": 0.5462404577475578,
      "learning_rate": 7.805264205829117e-07,
      "loss": 0.4819,
      "step": 11246
    },
    {
      "epoch": 0.8251877802947624,
      "grad_norm": 1.4464980830153105,
      "learning_rate": 7.798890616100996e-07,
      "loss": 0.6956,
      "step": 11247
    },
    {
      "epoch": 0.8252611498849014,
      "grad_norm": 1.3724719384562298,
      "learning_rate": 7.792519409574034e-07,
      "loss": 0.6452,
      "step": 11248
    },
    {
      "epoch": 0.8252611498849014,
      "lm_loss": 1.0603,
      "step": 11248,
      "vm_loss": 0.0322
    },
    {
      "epoch": 0.8252611498849014,
      "lm_loss": 0.8972,
      "step": 11248,
      "vm_loss": 0.2071
    },
    {
      "epoch": 0.8252611498849014,
      "lm_loss": 0.5158,
      "step": 11248,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.8252611498849014,
      "lm_loss": 0.4054,
      "step": 11248,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.8252611498849014,
      "lm_loss": 0.6757,
      "step": 11248,
      "vm_loss": 0.1527
    },
    {
      "epoch": 0.8252611498849014,
      "lm_loss": 1.0423,
      "step": 11248,
      "vm_loss": 0.1356
    },
    {
      "epoch": 0.8252611498849014,
      "lm_loss": 0.1633,
      "step": 11248,
      "vm_loss": 0.0953
    },
    {
      "epoch": 0.8252611498849014,
      "lm_loss": 0.9505,
      "step": 11248,
      "vm_loss": 0.1656
    },
    {
      "epoch": 0.8253345194750406,
      "grad_norm": 1.573413347343433,
      "learning_rate": 7.78615058660801e-07,
      "loss": 0.7777,
      "step": 11249
    },
    {
      "epoch": 0.8254078890651797,
      "grad_norm": 1.5401540645182608,
      "learning_rate": 7.779784147562619e-07,
      "loss": 0.7135,
      "step": 11250
    },
    {
      "epoch": 0.8254812586553189,
      "grad_norm": 0.5543869665914167,
      "learning_rate": 7.773420092797374e-07,
      "loss": 0.4768,
      "step": 11251
    },
    {
      "epoch": 0.825554628245458,
      "grad_norm": 1.5974239019593497,
      "learning_rate": 7.767058422671669e-07,
      "loss": 0.6774,
      "step": 11252
    },
    {
      "epoch": 0.825627997835597,
      "grad_norm": 0.5350059386906729,
      "learning_rate": 7.760699137544758e-07,
      "loss": 0.4659,
      "step": 11253
    },
    {
      "epoch": 0.8257013674257362,
      "grad_norm": 1.4365225659807699,
      "learning_rate": 7.754342237775764e-07,
      "loss": 0.6821,
      "step": 11254
    },
    {
      "epoch": 0.8257747370158753,
      "grad_norm": 0.5620676521193532,
      "learning_rate": 7.747987723723693e-07,
      "loss": 0.4825,
      "step": 11255
    },
    {
      "epoch": 0.8258481066060145,
      "grad_norm": 0.5475115600751028,
      "learning_rate": 7.741635595747394e-07,
      "loss": 0.4838,
      "step": 11256
    },
    {
      "epoch": 0.8258481066060145,
      "lm_loss": 0.5315,
      "step": 11256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8258481066060145,
      "lm_loss": 0.4644,
      "step": 11256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8258481066060145,
      "lm_loss": 0.1989,
      "step": 11256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8258481066060145,
      "lm_loss": 0.3542,
      "step": 11256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8258481066060145,
      "lm_loss": 0.5597,
      "step": 11256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8258481066060145,
      "lm_loss": 0.3489,
      "step": 11256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8258481066060145,
      "lm_loss": 0.5736,
      "step": 11256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8258481066060145,
      "lm_loss": 0.5004,
      "step": 11256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8259214761961536,
      "grad_norm": 0.5298388116395478,
      "learning_rate": 7.735285854205576e-07,
      "loss": 0.4555,
      "step": 11257
    },
    {
      "epoch": 0.8259948457862927,
      "grad_norm": 1.4707016885154205,
      "learning_rate": 7.72893849945684e-07,
      "loss": 0.7792,
      "step": 11258
    },
    {
      "epoch": 0.8260682153764318,
      "grad_norm": 1.5913868797269102,
      "learning_rate": 7.722593531859613e-07,
      "loss": 0.626,
      "step": 11259
    },
    {
      "epoch": 0.826141584966571,
      "grad_norm": 1.56610885117586,
      "learning_rate": 7.716250951772242e-07,
      "loss": 0.6255,
      "step": 11260
    },
    {
      "epoch": 0.8262149545567101,
      "grad_norm": 0.5484421008088319,
      "learning_rate": 7.709910759552896e-07,
      "loss": 0.4828,
      "step": 11261
    },
    {
      "epoch": 0.8262883241468493,
      "grad_norm": 0.5399538692368819,
      "learning_rate": 7.703572955559613e-07,
      "loss": 0.4827,
      "step": 11262
    },
    {
      "epoch": 0.8263616937369883,
      "grad_norm": 1.3416104847884802,
      "learning_rate": 7.697237540150316e-07,
      "loss": 0.6915,
      "step": 11263
    },
    {
      "epoch": 0.8264350633271275,
      "grad_norm": 0.5542938373204407,
      "learning_rate": 7.690904513682773e-07,
      "loss": 0.4783,
      "step": 11264
    },
    {
      "epoch": 0.8264350633271275,
      "lm_loss": 1.175,
      "step": 11264,
      "vm_loss": 0.0796
    },
    {
      "epoch": 0.8264350633271275,
      "lm_loss": 0.1274,
      "step": 11264,
      "vm_loss": 0.0876
    },
    {
      "epoch": 0.8264350633271275,
      "lm_loss": 0.7696,
      "step": 11264,
      "vm_loss": 0.2584
    },
    {
      "epoch": 0.8264350633271275,
      "lm_loss": 0.8352,
      "step": 11264,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.8264350633271275,
      "lm_loss": 0.834,
      "step": 11264,
      "vm_loss": 0.1379
    },
    {
      "epoch": 0.8264350633271275,
      "lm_loss": 0.8823,
      "step": 11264,
      "vm_loss": 0.1622
    },
    {
      "epoch": 0.8264350633271275,
      "lm_loss": 0.098,
      "step": 11264,
      "vm_loss": 0.1666
    },
    {
      "epoch": 0.8264350633271275,
      "lm_loss": 1.1303,
      "step": 11264,
      "vm_loss": 0.1878
    },
    {
      "epoch": 0.8265084329172666,
      "grad_norm": 1.417215562373943,
      "learning_rate": 7.684573876514618e-07,
      "loss": 0.7957,
      "step": 11265
    },
    {
      "epoch": 0.8265818025074058,
      "grad_norm": 1.5426272714761886,
      "learning_rate": 7.678245629003378e-07,
      "loss": 0.8334,
      "step": 11266
    },
    {
      "epoch": 0.8266551720975449,
      "grad_norm": 1.4499154572854906,
      "learning_rate": 7.67191977150642e-07,
      "loss": 0.8337,
      "step": 11267
    },
    {
      "epoch": 0.826728541687684,
      "grad_norm": 1.5392987806031326,
      "learning_rate": 7.665596304380973e-07,
      "loss": 0.7203,
      "step": 11268
    },
    {
      "epoch": 0.8268019112778231,
      "grad_norm": 1.5777128805184737,
      "learning_rate": 7.659275227984142e-07,
      "loss": 0.6133,
      "step": 11269
    },
    {
      "epoch": 0.8268752808679622,
      "grad_norm": 1.7645574627881766,
      "learning_rate": 7.652956542672885e-07,
      "loss": 0.8746,
      "step": 11270
    },
    {
      "epoch": 0.8269486504581014,
      "grad_norm": 1.4697119362116893,
      "learning_rate": 7.646640248804055e-07,
      "loss": 0.7343,
      "step": 11271
    },
    {
      "epoch": 0.8270220200482405,
      "grad_norm": 1.5702821693170657,
      "learning_rate": 7.640326346734334e-07,
      "loss": 0.6998,
      "step": 11272
    },
    {
      "epoch": 0.8270220200482405,
      "lm_loss": 1.1868,
      "step": 11272,
      "vm_loss": 0.0927
    },
    {
      "epoch": 0.8270220200482405,
      "lm_loss": 0.235,
      "step": 11272,
      "vm_loss": 0.096
    },
    {
      "epoch": 0.8270220200482405,
      "lm_loss": 0.3237,
      "step": 11272,
      "vm_loss": 0.15
    },
    {
      "epoch": 0.8270220200482405,
      "lm_loss": 0.3138,
      "step": 11272,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.8270220200482405,
      "lm_loss": 0.0959,
      "step": 11272,
      "vm_loss": 0.0686
    },
    {
      "epoch": 0.8270220200482405,
      "lm_loss": 0.2314,
      "step": 11272,
      "vm_loss": 0.09
    },
    {
      "epoch": 0.8270220200482405,
      "lm_loss": 1.3776,
      "step": 11272,
      "vm_loss": 0.1254
    },
    {
      "epoch": 0.8270220200482405,
      "lm_loss": 0.8887,
      "step": 11272,
      "vm_loss": 0.0739
    },
    {
      "epoch": 0.8270953896383796,
      "grad_norm": 1.604368836890601,
      "learning_rate": 7.634014836820285e-07,
      "loss": 0.6952,
      "step": 11273
    },
    {
      "epoch": 0.8271687592285187,
      "grad_norm": 0.5586823034891597,
      "learning_rate": 7.627705719418338e-07,
      "loss": 0.4982,
      "step": 11274
    },
    {
      "epoch": 0.8272421288186579,
      "grad_norm": 1.6757647700282974,
      "learning_rate": 7.621398994884776e-07,
      "loss": 0.7805,
      "step": 11275
    },
    {
      "epoch": 0.827315498408797,
      "grad_norm": 0.5639100552070141,
      "learning_rate": 7.615094663575757e-07,
      "loss": 0.4888,
      "step": 11276
    },
    {
      "epoch": 0.8273888679989362,
      "grad_norm": 0.5473169148098704,
      "learning_rate": 7.608792725847319e-07,
      "loss": 0.4673,
      "step": 11277
    },
    {
      "epoch": 0.8274622375890752,
      "grad_norm": 1.4478933441375001,
      "learning_rate": 7.602493182055332e-07,
      "loss": 0.7692,
      "step": 11278
    },
    {
      "epoch": 0.8275356071792144,
      "grad_norm": 0.5503304653008136,
      "learning_rate": 7.596196032555553e-07,
      "loss": 0.4819,
      "step": 11279
    },
    {
      "epoch": 0.8276089767693535,
      "grad_norm": 0.536679572504696,
      "learning_rate": 7.589901277703593e-07,
      "loss": 0.4702,
      "step": 11280
    },
    {
      "epoch": 0.8276089767693535,
      "lm_loss": 0.2747,
      "step": 11280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8276089767693535,
      "lm_loss": 0.59,
      "step": 11280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8276089767693535,
      "lm_loss": 0.4858,
      "step": 11280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8276089767693535,
      "lm_loss": 0.3488,
      "step": 11280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8276089767693535,
      "lm_loss": 0.4027,
      "step": 11280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8276089767693535,
      "lm_loss": 0.4508,
      "step": 11280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8276089767693535,
      "lm_loss": 0.4716,
      "step": 11280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8276089767693535,
      "lm_loss": 0.4659,
      "step": 11280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8276823463594927,
      "grad_norm": 0.5311866501502182,
      "learning_rate": 7.583608917854923e-07,
      "loss": 0.4589,
      "step": 11281
    },
    {
      "epoch": 0.8277557159496318,
      "grad_norm": 1.4379418930499916,
      "learning_rate": 7.577318953364915e-07,
      "loss": 0.739,
      "step": 11282
    },
    {
      "epoch": 0.8278290855397709,
      "grad_norm": 1.6258442120143353,
      "learning_rate": 7.571031384588762e-07,
      "loss": 0.7441,
      "step": 11283
    },
    {
      "epoch": 0.82790245512991,
      "grad_norm": 1.5447927610490795,
      "learning_rate": 7.564746211881541e-07,
      "loss": 0.7139,
      "step": 11284
    },
    {
      "epoch": 0.8279758247200492,
      "grad_norm": 1.6405806619869572,
      "learning_rate": 7.55846343559819e-07,
      "loss": 0.7139,
      "step": 11285
    },
    {
      "epoch": 0.8280491943101883,
      "grad_norm": 0.5458130577439341,
      "learning_rate": 7.552183056093498e-07,
      "loss": 0.4755,
      "step": 11286
    },
    {
      "epoch": 0.8281225639003275,
      "grad_norm": 1.4974706673420968,
      "learning_rate": 7.545905073722165e-07,
      "loss": 0.8214,
      "step": 11287
    },
    {
      "epoch": 0.8281959334904665,
      "grad_norm": 1.5371031089883593,
      "learning_rate": 7.539629488838707e-07,
      "loss": 0.7938,
      "step": 11288
    },
    {
      "epoch": 0.8281959334904665,
      "lm_loss": 1.072,
      "step": 11288,
      "vm_loss": 0.0449
    },
    {
      "epoch": 0.8281959334904665,
      "lm_loss": 1.4862,
      "step": 11288,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.8281959334904665,
      "lm_loss": 0.1794,
      "step": 11288,
      "vm_loss": 0.1312
    },
    {
      "epoch": 0.8281959334904665,
      "lm_loss": 0.516,
      "step": 11288,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.8281959334904665,
      "lm_loss": 0.7841,
      "step": 11288,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.8281959334904665,
      "lm_loss": 0.5132,
      "step": 11288,
      "vm_loss": 0.078
    },
    {
      "epoch": 0.8281959334904665,
      "lm_loss": 1.0032,
      "step": 11288,
      "vm_loss": 0.1101
    },
    {
      "epoch": 0.8281959334904665,
      "lm_loss": 0.5187,
      "step": 11288,
      "vm_loss": 0.0893
    },
    {
      "epoch": 0.8282693030806056,
      "grad_norm": 1.508163806902224,
      "learning_rate": 7.533356301797523e-07,
      "loss": 0.7151,
      "step": 11289
    },
    {
      "epoch": 0.8283426726707448,
      "grad_norm": 1.5554299378216432,
      "learning_rate": 7.527085512952876e-07,
      "loss": 0.6514,
      "step": 11290
    },
    {
      "epoch": 0.828416042260884,
      "grad_norm": 1.6145502390286948,
      "learning_rate": 7.520817122658892e-07,
      "loss": 0.7535,
      "step": 11291
    },
    {
      "epoch": 0.8284894118510231,
      "grad_norm": 1.4623940666342075,
      "learning_rate": 7.514551131269555e-07,
      "loss": 0.7595,
      "step": 11292
    },
    {
      "epoch": 0.8285627814411621,
      "grad_norm": 1.5371984692488567,
      "learning_rate": 7.508287539138737e-07,
      "loss": 0.6431,
      "step": 11293
    },
    {
      "epoch": 0.8286361510313013,
      "grad_norm": 0.557709079492423,
      "learning_rate": 7.502026346620151e-07,
      "loss": 0.4758,
      "step": 11294
    },
    {
      "epoch": 0.8287095206214404,
      "grad_norm": 1.5724100864015416,
      "learning_rate": 7.495767554067379e-07,
      "loss": 0.7454,
      "step": 11295
    },
    {
      "epoch": 0.8287828902115796,
      "grad_norm": 1.624191392710657,
      "learning_rate": 7.489511161833874e-07,
      "loss": 0.7608,
      "step": 11296
    },
    {
      "epoch": 0.8287828902115796,
      "lm_loss": 0.583,
      "step": 11296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8287828902115796,
      "lm_loss": 0.5205,
      "step": 11296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8287828902115796,
      "lm_loss": 0.2429,
      "step": 11296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8287828902115796,
      "lm_loss": 0.5269,
      "step": 11296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8287828902115796,
      "lm_loss": 0.4074,
      "step": 11296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8287828902115796,
      "lm_loss": 0.4626,
      "step": 11296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8287828902115796,
      "lm_loss": 0.4114,
      "step": 11296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8287828902115796,
      "lm_loss": 0.4842,
      "step": 11296,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8288562598017187,
      "grad_norm": 0.5607704214126238,
      "learning_rate": 7.48325717027294e-07,
      "loss": 0.4723,
      "step": 11297
    },
    {
      "epoch": 0.8289296293918578,
      "grad_norm": 1.5485228876499784,
      "learning_rate": 7.477005579737778e-07,
      "loss": 0.7142,
      "step": 11298
    },
    {
      "epoch": 0.8290029989819969,
      "grad_norm": 1.7733960138467637,
      "learning_rate": 7.470756390581412e-07,
      "loss": 0.7734,
      "step": 11299
    },
    {
      "epoch": 0.8290763685721361,
      "grad_norm": 1.6912789215095296,
      "learning_rate": 7.464509603156761e-07,
      "loss": 0.9101,
      "step": 11300
    },
    {
      "epoch": 0.8291497381622752,
      "grad_norm": 1.501911603653523,
      "learning_rate": 7.458265217816596e-07,
      "loss": 0.7414,
      "step": 11301
    },
    {
      "epoch": 0.8292231077524144,
      "grad_norm": 1.6951867263080702,
      "learning_rate": 7.452023234913524e-07,
      "loss": 0.8048,
      "step": 11302
    },
    {
      "epoch": 0.8292964773425534,
      "grad_norm": 1.73198131415757,
      "learning_rate": 7.445783654800082e-07,
      "loss": 0.9097,
      "step": 11303
    },
    {
      "epoch": 0.8293698469326926,
      "grad_norm": 1.319891801808008,
      "learning_rate": 7.439546477828624e-07,
      "loss": 0.6927,
      "step": 11304
    },
    {
      "epoch": 0.8293698469326926,
      "lm_loss": 0.4789,
      "step": 11304,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.8293698469326926,
      "lm_loss": 0.282,
      "step": 11304,
      "vm_loss": 0.0791
    },
    {
      "epoch": 0.8293698469326926,
      "lm_loss": 0.0152,
      "step": 11304,
      "vm_loss": 0.1138
    },
    {
      "epoch": 0.8293698469326926,
      "lm_loss": 0.1663,
      "step": 11304,
      "vm_loss": 0.1144
    },
    {
      "epoch": 0.8293698469326926,
      "lm_loss": 1.201,
      "step": 11304,
      "vm_loss": 0.1043
    },
    {
      "epoch": 0.8293698469326926,
      "lm_loss": 0.8166,
      "step": 11304,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.8293698469326926,
      "lm_loss": 0.7721,
      "step": 11304,
      "vm_loss": 0.1474
    },
    {
      "epoch": 0.8293698469326926,
      "lm_loss": 0.378,
      "step": 11304,
      "vm_loss": 0.0544
    },
    {
      "epoch": 0.8294432165228317,
      "grad_norm": 1.593113647942087,
      "learning_rate": 7.433311704351371e-07,
      "loss": 0.7156,
      "step": 11305
    },
    {
      "epoch": 0.8295165861129709,
      "grad_norm": 1.4579641891960897,
      "learning_rate": 7.427079334720422e-07,
      "loss": 0.7191,
      "step": 11306
    },
    {
      "epoch": 0.82958995570311,
      "grad_norm": 0.5476942281433395,
      "learning_rate": 7.420849369287719e-07,
      "loss": 0.4842,
      "step": 11307
    },
    {
      "epoch": 0.829663325293249,
      "grad_norm": 1.603042865877424,
      "learning_rate": 7.414621808405115e-07,
      "loss": 0.655,
      "step": 11308
    },
    {
      "epoch": 0.8297366948833882,
      "grad_norm": 1.6114163797298526,
      "learning_rate": 7.408396652424271e-07,
      "loss": 0.6647,
      "step": 11309
    },
    {
      "epoch": 0.8298100644735273,
      "grad_norm": 0.5387124059370816,
      "learning_rate": 7.402173901696747e-07,
      "loss": 0.4718,
      "step": 11310
    },
    {
      "epoch": 0.8298834340636665,
      "grad_norm": 0.5511491219815039,
      "learning_rate": 7.395953556573959e-07,
      "loss": 0.4949,
      "step": 11311
    },
    {
      "epoch": 0.8299568036538056,
      "grad_norm": 1.4570795708513804,
      "learning_rate": 7.389735617407162e-07,
      "loss": 0.6853,
      "step": 11312
    },
    {
      "epoch": 0.8299568036538056,
      "lm_loss": 1.3728,
      "step": 11312,
      "vm_loss": 0.0422
    },
    {
      "epoch": 0.8299568036538056,
      "lm_loss": 0.499,
      "step": 11312,
      "vm_loss": 0.1624
    },
    {
      "epoch": 0.8299568036538056,
      "lm_loss": 0.8717,
      "step": 11312,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.8299568036538056,
      "lm_loss": 0.6402,
      "step": 11312,
      "vm_loss": 0.1411
    },
    {
      "epoch": 0.8299568036538056,
      "lm_loss": 0.4032,
      "step": 11312,
      "vm_loss": 0.2489
    },
    {
      "epoch": 0.8299568036538056,
      "lm_loss": 0.657,
      "step": 11312,
      "vm_loss": 0.1415
    },
    {
      "epoch": 0.8299568036538056,
      "lm_loss": 0.8513,
      "step": 11312,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.8299568036538056,
      "lm_loss": 0.4204,
      "step": 11312,
      "vm_loss": 0.1424
    },
    {
      "epoch": 0.8300301732439447,
      "grad_norm": 1.3834298240587017,
      "learning_rate": 7.383520084547535e-07,
      "loss": 0.6811,
      "step": 11313
    },
    {
      "epoch": 0.8301035428340838,
      "grad_norm": 1.4799014791228637,
      "learning_rate": 7.377306958346065e-07,
      "loss": 0.7773,
      "step": 11314
    },
    {
      "epoch": 0.830176912424223,
      "grad_norm": 1.51789506503963,
      "learning_rate": 7.371096239153624e-07,
      "loss": 0.7118,
      "step": 11315
    },
    {
      "epoch": 0.8302502820143621,
      "grad_norm": 0.5456071031294094,
      "learning_rate": 7.364887927320947e-07,
      "loss": 0.4665,
      "step": 11316
    },
    {
      "epoch": 0.8303236516045013,
      "grad_norm": 1.5475935788788755,
      "learning_rate": 7.358682023198632e-07,
      "loss": 0.8021,
      "step": 11317
    },
    {
      "epoch": 0.8303970211946403,
      "grad_norm": 1.5725683737430556,
      "learning_rate": 7.35247852713713e-07,
      "loss": 0.8237,
      "step": 11318
    },
    {
      "epoch": 0.8304703907847795,
      "grad_norm": 0.5755954882854535,
      "learning_rate": 7.346277439486793e-07,
      "loss": 0.4933,
      "step": 11319
    },
    {
      "epoch": 0.8305437603749186,
      "grad_norm": 1.4166609741488116,
      "learning_rate": 7.340078760597802e-07,
      "loss": 0.6073,
      "step": 11320
    },
    {
      "epoch": 0.8305437603749186,
      "lm_loss": 0.9396,
      "step": 11320,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.8305437603749186,
      "lm_loss": 0.5546,
      "step": 11320,
      "vm_loss": 0.1392
    },
    {
      "epoch": 0.8305437603749186,
      "lm_loss": 0.3295,
      "step": 11320,
      "vm_loss": 0.0367
    },
    {
      "epoch": 0.8305437603749186,
      "lm_loss": 0.1077,
      "step": 11320,
      "vm_loss": 0.079
    },
    {
      "epoch": 0.8305437603749186,
      "lm_loss": 0.9702,
      "step": 11320,
      "vm_loss": 0.1568
    },
    {
      "epoch": 0.8305437603749186,
      "lm_loss": 0.3711,
      "step": 11320,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.8305437603749186,
      "lm_loss": 0.6208,
      "step": 11320,
      "vm_loss": 0.3949
    },
    {
      "epoch": 0.8305437603749186,
      "lm_loss": 0.7392,
      "step": 11320,
      "vm_loss": 0.1487
    },
    {
      "epoch": 0.8306171299650578,
      "grad_norm": 1.5785828311861512,
      "learning_rate": 7.333882490820204e-07,
      "loss": 0.6983,
      "step": 11321
    },
    {
      "epoch": 0.8306904995551969,
      "grad_norm": 1.5766853340842495,
      "learning_rate": 7.327688630503927e-07,
      "loss": 0.7648,
      "step": 11322
    },
    {
      "epoch": 0.830763869145336,
      "grad_norm": 1.3583846380605102,
      "learning_rate": 7.321497179998738e-07,
      "loss": 0.6741,
      "step": 11323
    },
    {
      "epoch": 0.8308372387354751,
      "grad_norm": 0.5393756521527255,
      "learning_rate": 7.315308139654303e-07,
      "loss": 0.4897,
      "step": 11324
    },
    {
      "epoch": 0.8309106083256143,
      "grad_norm": 1.5078116840546707,
      "learning_rate": 7.309121509820122e-07,
      "loss": 0.7034,
      "step": 11325
    },
    {
      "epoch": 0.8309839779157534,
      "grad_norm": 1.4822354704282563,
      "learning_rate": 7.30293729084558e-07,
      "loss": 0.683,
      "step": 11326
    },
    {
      "epoch": 0.8310573475058924,
      "grad_norm": 0.552738017415379,
      "learning_rate": 7.296755483079898e-07,
      "loss": 0.4952,
      "step": 11327
    },
    {
      "epoch": 0.8311307170960316,
      "grad_norm": 1.5343195506754521,
      "learning_rate": 7.29057608687218e-07,
      "loss": 0.7969,
      "step": 11328
    },
    {
      "epoch": 0.8311307170960316,
      "lm_loss": 0.5302,
      "step": 11328,
      "vm_loss": 0.0419
    },
    {
      "epoch": 0.8311307170960316,
      "lm_loss": 1.4297,
      "step": 11328,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.8311307170960316,
      "lm_loss": 0.0018,
      "step": 11328,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.8311307170960316,
      "lm_loss": 1.0153,
      "step": 11328,
      "vm_loss": 0.1908
    },
    {
      "epoch": 0.8311307170960316,
      "lm_loss": 1.2752,
      "step": 11328,
      "vm_loss": 0.1859
    },
    {
      "epoch": 0.8311307170960316,
      "lm_loss": 0.7372,
      "step": 11328,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.8311307170960316,
      "lm_loss": 0.0162,
      "step": 11328,
      "vm_loss": 0.1347
    },
    {
      "epoch": 0.8311307170960316,
      "lm_loss": 0.7698,
      "step": 11328,
      "vm_loss": 0.1256
    },
    {
      "epoch": 0.8312040866861707,
      "grad_norm": 1.4769285186401024,
      "learning_rate": 7.28439910257141e-07,
      "loss": 0.7199,
      "step": 11329
    },
    {
      "epoch": 0.8312774562763099,
      "grad_norm": 0.5246793774111358,
      "learning_rate": 7.278224530526401e-07,
      "loss": 0.4565,
      "step": 11330
    },
    {
      "epoch": 0.831350825866449,
      "grad_norm": 1.6933997096591415,
      "learning_rate": 7.272052371085859e-07,
      "loss": 0.7403,
      "step": 11331
    },
    {
      "epoch": 0.8314241954565881,
      "grad_norm": 0.5561850599715771,
      "learning_rate": 7.265882624598331e-07,
      "loss": 0.4776,
      "step": 11332
    },
    {
      "epoch": 0.8314975650467272,
      "grad_norm": 1.39635433164981,
      "learning_rate": 7.25971529141224e-07,
      "loss": 0.7782,
      "step": 11333
    },
    {
      "epoch": 0.8315709346368664,
      "grad_norm": 0.576546743491852,
      "learning_rate": 7.253550371875856e-07,
      "loss": 0.4923,
      "step": 11334
    },
    {
      "epoch": 0.8316443042270055,
      "grad_norm": 0.5595747191888291,
      "learning_rate": 7.247387866337358e-07,
      "loss": 0.4779,
      "step": 11335
    },
    {
      "epoch": 0.8317176738171447,
      "grad_norm": 1.5475923100358757,
      "learning_rate": 7.241227775144737e-07,
      "loss": 0.8267,
      "step": 11336
    },
    {
      "epoch": 0.8317176738171447,
      "lm_loss": 0.9213,
      "step": 11336,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.8317176738171447,
      "lm_loss": 0.022,
      "step": 11336,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.8317176738171447,
      "lm_loss": 0.7748,
      "step": 11336,
      "vm_loss": 0.1693
    },
    {
      "epoch": 0.8317176738171447,
      "lm_loss": 0.5166,
      "step": 11336,
      "vm_loss": 0.1294
    },
    {
      "epoch": 0.8317176738171447,
      "lm_loss": 0.3923,
      "step": 11336,
      "vm_loss": 0.1185
    },
    {
      "epoch": 0.8317176738171447,
      "lm_loss": 0.9564,
      "step": 11336,
      "vm_loss": 0.1591
    },
    {
      "epoch": 0.8317176738171447,
      "lm_loss": 0.013,
      "step": 11336,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.8317176738171447,
      "lm_loss": 0.6331,
      "step": 11336,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.8317910434072837,
      "grad_norm": 1.420974833804464,
      "learning_rate": 7.23507009864588e-07,
      "loss": 0.6995,
      "step": 11337
    },
    {
      "epoch": 0.8318644129974229,
      "grad_norm": 1.5965929045494767,
      "learning_rate": 7.228914837188517e-07,
      "loss": 0.6148,
      "step": 11338
    },
    {
      "epoch": 0.831937782587562,
      "grad_norm": 1.3829407212034235,
      "learning_rate": 7.222761991120241e-07,
      "loss": 0.6284,
      "step": 11339
    },
    {
      "epoch": 0.8320111521777012,
      "grad_norm": 1.624231156136369,
      "learning_rate": 7.21661156078855e-07,
      "loss": 0.8187,
      "step": 11340
    },
    {
      "epoch": 0.8320845217678403,
      "grad_norm": 1.7972632027658602,
      "learning_rate": 7.210463546540747e-07,
      "loss": 0.8146,
      "step": 11341
    },
    {
      "epoch": 0.8321578913579794,
      "grad_norm": 0.5516283146117142,
      "learning_rate": 7.204317948724038e-07,
      "loss": 0.4746,
      "step": 11342
    },
    {
      "epoch": 0.8322312609481185,
      "grad_norm": 1.486085866247166,
      "learning_rate": 7.198174767685479e-07,
      "loss": 0.759,
      "step": 11343
    },
    {
      "epoch": 0.8323046305382577,
      "grad_norm": 1.679927550951602,
      "learning_rate": 7.192034003771986e-07,
      "loss": 0.7751,
      "step": 11344
    },
    {
      "epoch": 0.8323046305382577,
      "lm_loss": 1.8541,
      "step": 11344,
      "vm_loss": 0.0723
    },
    {
      "epoch": 0.8323046305382577,
      "lm_loss": 0.9754,
      "step": 11344,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.8323046305382577,
      "lm_loss": 0.1324,
      "step": 11344,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.8323046305382577,
      "lm_loss": 0.6546,
      "step": 11344,
      "vm_loss": 0.1383
    },
    {
      "epoch": 0.8323046305382577,
      "lm_loss": 0.2495,
      "step": 11344,
      "vm_loss": 0.1592
    },
    {
      "epoch": 0.8323046305382577,
      "lm_loss": 0.7128,
      "step": 11344,
      "vm_loss": 0.1118
    },
    {
      "epoch": 0.8323046305382577,
      "lm_loss": 0.2563,
      "step": 11344,
      "vm_loss": 0.1424
    },
    {
      "epoch": 0.8323046305382577,
      "lm_loss": 0.3188,
      "step": 11344,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.8323780001283968,
      "grad_norm": 1.5089756805733225,
      "learning_rate": 7.185895657330327e-07,
      "loss": 0.7964,
      "step": 11345
    },
    {
      "epoch": 0.832451369718536,
      "grad_norm": 1.7265175814032276,
      "learning_rate": 7.179759728707187e-07,
      "loss": 0.7743,
      "step": 11346
    },
    {
      "epoch": 0.832524739308675,
      "grad_norm": 0.5261102896107788,
      "learning_rate": 7.173626218249052e-07,
      "loss": 0.4647,
      "step": 11347
    },
    {
      "epoch": 0.8325981088988141,
      "grad_norm": 1.4701984808668744,
      "learning_rate": 7.167495126302309e-07,
      "loss": 0.7266,
      "step": 11348
    },
    {
      "epoch": 0.8326714784889533,
      "grad_norm": 1.5628152630469303,
      "learning_rate": 7.161366453213181e-07,
      "loss": 0.7533,
      "step": 11349
    },
    {
      "epoch": 0.8327448480790924,
      "grad_norm": 1.4533429766499675,
      "learning_rate": 7.155240199327773e-07,
      "loss": 0.598,
      "step": 11350
    },
    {
      "epoch": 0.8328182176692316,
      "grad_norm": 1.5674147517933914,
      "learning_rate": 7.149116364992065e-07,
      "loss": 0.6794,
      "step": 11351
    },
    {
      "epoch": 0.8328915872593706,
      "grad_norm": 0.5544481091562373,
      "learning_rate": 7.142994950551868e-07,
      "loss": 0.4566,
      "step": 11352
    },
    {
      "epoch": 0.8328915872593706,
      "lm_loss": 0.5365,
      "step": 11352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8328915872593706,
      "lm_loss": 0.3457,
      "step": 11352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8328915872593706,
      "lm_loss": 0.5157,
      "step": 11352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8328915872593706,
      "lm_loss": 0.4729,
      "step": 11352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8328915872593706,
      "lm_loss": 0.4727,
      "step": 11352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8328915872593706,
      "lm_loss": 0.3488,
      "step": 11352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8328915872593706,
      "lm_loss": 0.5872,
      "step": 11352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8328915872593706,
      "lm_loss": 0.7344,
      "step": 11352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8329649568495098,
      "grad_norm": 0.5441429412784187,
      "learning_rate": 7.136875956352884e-07,
      "loss": 0.4692,
      "step": 11353
    },
    {
      "epoch": 0.8330383264396489,
      "grad_norm": 1.498841885013096,
      "learning_rate": 7.130759382740665e-07,
      "loss": 0.6816,
      "step": 11354
    },
    {
      "epoch": 0.8331116960297881,
      "grad_norm": 1.4909814326689488,
      "learning_rate": 7.124645230060612e-07,
      "loss": 0.7263,
      "step": 11355
    },
    {
      "epoch": 0.8331850656199272,
      "grad_norm": 1.7517437145282324,
      "learning_rate": 7.118533498658037e-07,
      "loss": 0.7361,
      "step": 11356
    },
    {
      "epoch": 0.8332584352100663,
      "grad_norm": 1.396838640273477,
      "learning_rate": 7.11242418887807e-07,
      "loss": 0.6234,
      "step": 11357
    },
    {
      "epoch": 0.8333318048002054,
      "grad_norm": 1.4760178545321012,
      "learning_rate": 7.106317301065713e-07,
      "loss": 0.6228,
      "step": 11358
    },
    {
      "epoch": 0.8334051743903446,
      "grad_norm": 0.5584156135387213,
      "learning_rate": 7.10021283556585e-07,
      "loss": 0.4795,
      "step": 11359
    },
    {
      "epoch": 0.8334785439804837,
      "grad_norm": 1.581543180329199,
      "learning_rate": 7.094110792723202e-07,
      "loss": 0.6625,
      "step": 11360
    },
    {
      "epoch": 0.8334785439804837,
      "lm_loss": 1.2719,
      "step": 11360,
      "vm_loss": 0.1276
    },
    {
      "epoch": 0.8334785439804837,
      "lm_loss": 1.3192,
      "step": 11360,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.8334785439804837,
      "lm_loss": 0.3126,
      "step": 11360,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.8334785439804837,
      "lm_loss": 0.2146,
      "step": 11360,
      "vm_loss": 0.0618
    },
    {
      "epoch": 0.8334785439804837,
      "lm_loss": 0.6926,
      "step": 11360,
      "vm_loss": 0.0797
    },
    {
      "epoch": 0.8334785439804837,
      "lm_loss": 0.4609,
      "step": 11360,
      "vm_loss": 0.0704
    },
    {
      "epoch": 0.8334785439804837,
      "lm_loss": 0.0513,
      "step": 11360,
      "vm_loss": 0.0549
    },
    {
      "epoch": 0.8334785439804837,
      "lm_loss": 0.0034,
      "step": 11360,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.8335519135706229,
      "grad_norm": 1.380362070609617,
      "learning_rate": 7.088011172882364e-07,
      "loss": 0.7012,
      "step": 11361
    },
    {
      "epoch": 0.8336252831607619,
      "grad_norm": 1.5542382265846966,
      "learning_rate": 7.081913976387811e-07,
      "loss": 0.7766,
      "step": 11362
    },
    {
      "epoch": 0.833698652750901,
      "grad_norm": 1.483724837233292,
      "learning_rate": 7.075819203583867e-07,
      "loss": 0.6887,
      "step": 11363
    },
    {
      "epoch": 0.8337720223410402,
      "grad_norm": 1.544762483294114,
      "learning_rate": 7.069726854814707e-07,
      "loss": 0.5808,
      "step": 11364
    },
    {
      "epoch": 0.8338453919311793,
      "grad_norm": 0.5280310433525621,
      "learning_rate": 7.063636930424389e-07,
      "loss": 0.4668,
      "step": 11365
    },
    {
      "epoch": 0.8339187615213185,
      "grad_norm": 1.4269535099916868,
      "learning_rate": 7.057549430756816e-07,
      "loss": 0.7893,
      "step": 11366
    },
    {
      "epoch": 0.8339921311114575,
      "grad_norm": 1.4781603261365595,
      "learning_rate": 7.051464356155779e-07,
      "loss": 0.7283,
      "step": 11367
    },
    {
      "epoch": 0.8340655007015967,
      "grad_norm": 1.596644708514213,
      "learning_rate": 7.04538170696491e-07,
      "loss": 0.7652,
      "step": 11368
    },
    {
      "epoch": 0.8340655007015967,
      "lm_loss": 1.6296,
      "step": 11368,
      "vm_loss": 0.0587
    },
    {
      "epoch": 0.8340655007015967,
      "lm_loss": 0.8643,
      "step": 11368,
      "vm_loss": 0.1581
    },
    {
      "epoch": 0.8340655007015967,
      "lm_loss": 0.7687,
      "step": 11368,
      "vm_loss": 0.1664
    },
    {
      "epoch": 0.8340655007015967,
      "lm_loss": 0.4807,
      "step": 11368,
      "vm_loss": 0.0674
    },
    {
      "epoch": 0.8340655007015967,
      "lm_loss": 0.5395,
      "step": 11368,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.8340655007015967,
      "lm_loss": 0.3818,
      "step": 11368,
      "vm_loss": 0.2204
    },
    {
      "epoch": 0.8340655007015967,
      "lm_loss": 1.0302,
      "step": 11368,
      "vm_loss": 0.226
    },
    {
      "epoch": 0.8340655007015967,
      "lm_loss": 0.7375,
      "step": 11368,
      "vm_loss": 0.2248
    },
    {
      "epoch": 0.8341388702917358,
      "grad_norm": 1.6248321941941517,
      "learning_rate": 7.03930148352771e-07,
      "loss": 0.8357,
      "step": 11369
    },
    {
      "epoch": 0.834212239881875,
      "grad_norm": 1.4857843227214962,
      "learning_rate": 7.033223686187552e-07,
      "loss": 0.7521,
      "step": 11370
    },
    {
      "epoch": 0.8342856094720141,
      "grad_norm": 1.7638940086956687,
      "learning_rate": 7.027148315287652e-07,
      "loss": 0.6863,
      "step": 11371
    },
    {
      "epoch": 0.8343589790621532,
      "grad_norm": 0.5393223670974943,
      "learning_rate": 7.021075371171099e-07,
      "loss": 0.4907,
      "step": 11372
    },
    {
      "epoch": 0.8344323486522923,
      "grad_norm": 0.5582648178413123,
      "learning_rate": 7.015004854180863e-07,
      "loss": 0.4837,
      "step": 11373
    },
    {
      "epoch": 0.8345057182424315,
      "grad_norm": 1.716165075372978,
      "learning_rate": 7.008936764659758e-07,
      "loss": 0.7441,
      "step": 11374
    },
    {
      "epoch": 0.8345790878325706,
      "grad_norm": 0.5322446218082825,
      "learning_rate": 7.002871102950453e-07,
      "loss": 0.4654,
      "step": 11375
    },
    {
      "epoch": 0.8346524574227098,
      "grad_norm": 1.5508530755224932,
      "learning_rate": 6.996807869395506e-07,
      "loss": 0.6602,
      "step": 11376
    },
    {
      "epoch": 0.8346524574227098,
      "lm_loss": 1.464,
      "step": 11376,
      "vm_loss": 0.0333
    },
    {
      "epoch": 0.8346524574227098,
      "lm_loss": 1.6343,
      "step": 11376,
      "vm_loss": 0.1493
    },
    {
      "epoch": 0.8346524574227098,
      "lm_loss": 0.0369,
      "step": 11376,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.8346524574227098,
      "lm_loss": 1.3186,
      "step": 11376,
      "vm_loss": 0.1748
    },
    {
      "epoch": 0.8346524574227098,
      "lm_loss": 0.5903,
      "step": 11376,
      "vm_loss": 0.0844
    },
    {
      "epoch": 0.8346524574227098,
      "lm_loss": 0.8896,
      "step": 11376,
      "vm_loss": 0.1606
    },
    {
      "epoch": 0.8346524574227098,
      "lm_loss": 0.0108,
      "step": 11376,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.8346524574227098,
      "lm_loss": 0.587,
      "step": 11376,
      "vm_loss": 0.2237
    },
    {
      "epoch": 0.8347258270128488,
      "grad_norm": 1.3591389077666072,
      "learning_rate": 6.990747064337294e-07,
      "loss": 0.7423,
      "step": 11377
    },
    {
      "epoch": 0.834799196602988,
      "grad_norm": 1.4367324012701561,
      "learning_rate": 6.984688688118124e-07,
      "loss": 0.778,
      "step": 11378
    },
    {
      "epoch": 0.8348725661931271,
      "grad_norm": 1.5932801015471392,
      "learning_rate": 6.978632741080105e-07,
      "loss": 0.6763,
      "step": 11379
    },
    {
      "epoch": 0.8349459357832663,
      "grad_norm": 1.486271164599169,
      "learning_rate": 6.97257922356524e-07,
      "loss": 0.6971,
      "step": 11380
    },
    {
      "epoch": 0.8350193053734054,
      "grad_norm": 1.4950889540633676,
      "learning_rate": 6.966528135915379e-07,
      "loss": 0.6842,
      "step": 11381
    },
    {
      "epoch": 0.8350926749635444,
      "grad_norm": 1.4609021153944057,
      "learning_rate": 6.960479478472232e-07,
      "loss": 0.7921,
      "step": 11382
    },
    {
      "epoch": 0.8351660445536836,
      "grad_norm": 1.5681058500849403,
      "learning_rate": 6.954433251577408e-07,
      "loss": 0.7674,
      "step": 11383
    },
    {
      "epoch": 0.8352394141438227,
      "grad_norm": 1.5544244054926413,
      "learning_rate": 6.948389455572341e-07,
      "loss": 0.7136,
      "step": 11384
    },
    {
      "epoch": 0.8352394141438227,
      "lm_loss": 0.5084,
      "step": 11384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8352394141438227,
      "lm_loss": 0.4644,
      "step": 11384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8352394141438227,
      "lm_loss": 0.3489,
      "step": 11384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8352394141438227,
      "lm_loss": 0.545,
      "step": 11384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8352394141438227,
      "lm_loss": 0.5345,
      "step": 11384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8352394141438227,
      "lm_loss": 0.5194,
      "step": 11384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8352394141438227,
      "lm_loss": 0.5005,
      "step": 11384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8352394141438227,
      "lm_loss": 0.593,
      "step": 11384,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8353127837339619,
      "grad_norm": 0.5666083107825127,
      "learning_rate": 6.942348090798334e-07,
      "loss": 0.4748,
      "step": 11385
    },
    {
      "epoch": 0.835386153324101,
      "grad_norm": 1.676423143248995,
      "learning_rate": 6.93630915759656e-07,
      "loss": 0.6302,
      "step": 11386
    },
    {
      "epoch": 0.8354595229142401,
      "grad_norm": 1.5786938211814578,
      "learning_rate": 6.930272656308051e-07,
      "loss": 0.6721,
      "step": 11387
    },
    {
      "epoch": 0.8355328925043792,
      "grad_norm": 1.4281942518781565,
      "learning_rate": 6.924238587273701e-07,
      "loss": 0.7183,
      "step": 11388
    },
    {
      "epoch": 0.8356062620945184,
      "grad_norm": 1.4751105372311604,
      "learning_rate": 6.918206950834283e-07,
      "loss": 0.7429,
      "step": 11389
    },
    {
      "epoch": 0.8356796316846575,
      "grad_norm": 0.5662622206736153,
      "learning_rate": 6.912177747330407e-07,
      "loss": 0.502,
      "step": 11390
    },
    {
      "epoch": 0.8357530012747967,
      "grad_norm": 0.5432898381147268,
      "learning_rate": 6.906150977102555e-07,
      "loss": 0.4779,
      "step": 11391
    },
    {
      "epoch": 0.8358263708649357,
      "grad_norm": 1.4818166201570946,
      "learning_rate": 6.900126640491084e-07,
      "loss": 0.6898,
      "step": 11392
    },
    {
      "epoch": 0.8358263708649357,
      "lm_loss": 0.5466,
      "step": 11392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8358263708649357,
      "lm_loss": 0.3394,
      "step": 11392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8358263708649357,
      "lm_loss": 0.4523,
      "step": 11392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8358263708649357,
      "lm_loss": 0.5856,
      "step": 11392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8358263708649357,
      "lm_loss": 0.4829,
      "step": 11392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8358263708649357,
      "lm_loss": 0.5497,
      "step": 11392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8358263708649357,
      "lm_loss": 0.4311,
      "step": 11392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8358263708649357,
      "lm_loss": 0.4128,
      "step": 11392,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8358997404550749,
      "grad_norm": 0.5591128866434678,
      "learning_rate": 6.894104737836182e-07,
      "loss": 0.4899,
      "step": 11393
    },
    {
      "epoch": 0.835973110045214,
      "grad_norm": 1.643944512738185,
      "learning_rate": 6.888085269477945e-07,
      "loss": 0.6784,
      "step": 11394
    },
    {
      "epoch": 0.8360464796353532,
      "grad_norm": 1.5676270177560516,
      "learning_rate": 6.8820682357563e-07,
      "loss": 0.7969,
      "step": 11395
    },
    {
      "epoch": 0.8361198492254923,
      "grad_norm": 1.5084443586184932,
      "learning_rate": 6.876053637011043e-07,
      "loss": 0.5596,
      "step": 11396
    },
    {
      "epoch": 0.8361932188156314,
      "grad_norm": 1.3711089527007891,
      "learning_rate": 6.870041473581829e-07,
      "loss": 0.7452,
      "step": 11397
    },
    {
      "epoch": 0.8362665884057705,
      "grad_norm": 0.5522959129356934,
      "learning_rate": 6.864031745808186e-07,
      "loss": 0.4762,
      "step": 11398
    },
    {
      "epoch": 0.8363399579959097,
      "grad_norm": 0.5619115016759418,
      "learning_rate": 6.858024454029482e-07,
      "loss": 0.4821,
      "step": 11399
    },
    {
      "epoch": 0.8364133275860488,
      "grad_norm": 1.7545426415576257,
      "learning_rate": 6.852019598584986e-07,
      "loss": 0.7035,
      "step": 11400
    },
    {
      "epoch": 0.8364133275860488,
      "lm_loss": 1.6614,
      "step": 11400,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.8364133275860488,
      "lm_loss": 1.0608,
      "step": 11400,
      "vm_loss": 0.2039
    },
    {
      "epoch": 0.8364133275860488,
      "lm_loss": 0.7397,
      "step": 11400,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.8364133275860488,
      "lm_loss": 0.9674,
      "step": 11400,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.8364133275860488,
      "lm_loss": 0.3084,
      "step": 11400,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.8364133275860488,
      "lm_loss": 0.8732,
      "step": 11400,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.8364133275860488,
      "lm_loss": 0.9124,
      "step": 11400,
      "vm_loss": 0.1713
    },
    {
      "epoch": 0.8364133275860488,
      "lm_loss": 1.0762,
      "step": 11400,
      "vm_loss": 0.0682
    },
    {
      "epoch": 0.836486697176188,
      "grad_norm": 1.5117511723161463,
      "learning_rate": 6.846017179813796e-07,
      "loss": 0.8423,
      "step": 11401
    },
    {
      "epoch": 0.836560066766327,
      "grad_norm": 1.47739647674143,
      "learning_rate": 6.840017198054888e-07,
      "loss": 0.6507,
      "step": 11402
    },
    {
      "epoch": 0.8366334363564661,
      "grad_norm": 1.5350584541649426,
      "learning_rate": 6.834019653647094e-07,
      "loss": 0.7809,
      "step": 11403
    },
    {
      "epoch": 0.8367068059466053,
      "grad_norm": 1.4738172628060056,
      "learning_rate": 6.828024546929096e-07,
      "loss": 0.6701,
      "step": 11404
    },
    {
      "epoch": 0.8367801755367444,
      "grad_norm": 1.474824562233463,
      "learning_rate": 6.822031878239477e-07,
      "loss": 0.7384,
      "step": 11405
    },
    {
      "epoch": 0.8368535451268835,
      "grad_norm": 1.5954161728688723,
      "learning_rate": 6.816041647916643e-07,
      "loss": 0.711,
      "step": 11406
    },
    {
      "epoch": 0.8369269147170226,
      "grad_norm": 0.5407370550662282,
      "learning_rate": 6.810053856298882e-07,
      "loss": 0.4696,
      "step": 11407
    },
    {
      "epoch": 0.8370002843071618,
      "grad_norm": 0.5535819720054789,
      "learning_rate": 6.804068503724343e-07,
      "loss": 0.4736,
      "step": 11408
    },
    {
      "epoch": 0.8370002843071618,
      "lm_loss": 0.4783,
      "step": 11408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8370002843071618,
      "lm_loss": 0.3723,
      "step": 11408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8370002843071618,
      "lm_loss": 0.3461,
      "step": 11408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8370002843071618,
      "lm_loss": 0.4352,
      "step": 11408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8370002843071618,
      "lm_loss": 0.5228,
      "step": 11408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8370002843071618,
      "lm_loss": 0.4168,
      "step": 11408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8370002843071618,
      "lm_loss": 0.561,
      "step": 11408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8370002843071618,
      "lm_loss": 0.3601,
      "step": 11408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8370736538973009,
      "grad_norm": 0.5492186975916373,
      "learning_rate": 6.798085590531012e-07,
      "loss": 0.4848,
      "step": 11409
    },
    {
      "epoch": 0.8371470234874401,
      "grad_norm": 1.5095183302275954,
      "learning_rate": 6.792105117056785e-07,
      "loss": 0.7847,
      "step": 11410
    },
    {
      "epoch": 0.8372203930775791,
      "grad_norm": 0.5535103265172068,
      "learning_rate": 6.786127083639388e-07,
      "loss": 0.483,
      "step": 11411
    },
    {
      "epoch": 0.8372937626677183,
      "grad_norm": 1.4005056578852741,
      "learning_rate": 6.78015149061641e-07,
      "loss": 0.6996,
      "step": 11412
    },
    {
      "epoch": 0.8373671322578574,
      "grad_norm": 1.5111696164725756,
      "learning_rate": 6.77417833832531e-07,
      "loss": 0.7087,
      "step": 11413
    },
    {
      "epoch": 0.8374405018479966,
      "grad_norm": 0.5690022199103891,
      "learning_rate": 6.768207627103413e-07,
      "loss": 0.4847,
      "step": 11414
    },
    {
      "epoch": 0.8375138714381357,
      "grad_norm": 1.4188127546042861,
      "learning_rate": 6.762239357287876e-07,
      "loss": 0.6902,
      "step": 11415
    },
    {
      "epoch": 0.8375872410282748,
      "grad_norm": 0.5489408097858153,
      "learning_rate": 6.756273529215768e-07,
      "loss": 0.4798,
      "step": 11416
    },
    {
      "epoch": 0.8375872410282748,
      "lm_loss": 0.4225,
      "step": 11416,
      "vm_loss": 0.0706
    },
    {
      "epoch": 0.8375872410282748,
      "lm_loss": 0.6886,
      "step": 11416,
      "vm_loss": 0.0952
    },
    {
      "epoch": 0.8375872410282748,
      "lm_loss": 0.0057,
      "step": 11416,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.8375872410282748,
      "lm_loss": 0.9214,
      "step": 11416,
      "vm_loss": 0.1333
    },
    {
      "epoch": 0.8375872410282748,
      "lm_loss": 0.9328,
      "step": 11416,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.8375872410282748,
      "lm_loss": 0.4582,
      "step": 11416,
      "vm_loss": 0.1459
    },
    {
      "epoch": 0.8375872410282748,
      "lm_loss": 0.6536,
      "step": 11416,
      "vm_loss": 0.1653
    },
    {
      "epoch": 0.8375872410282748,
      "lm_loss": 0.2546,
      "step": 11416,
      "vm_loss": 0.0883
    },
    {
      "epoch": 0.8376606106184139,
      "grad_norm": 1.4685070869317203,
      "learning_rate": 6.750310143223987e-07,
      "loss": 0.7163,
      "step": 11417
    },
    {
      "epoch": 0.837733980208553,
      "grad_norm": 0.5562387663391792,
      "learning_rate": 6.744349199649298e-07,
      "loss": 0.4709,
      "step": 11418
    },
    {
      "epoch": 0.8378073497986922,
      "grad_norm": 0.547551796981705,
      "learning_rate": 6.738390698828329e-07,
      "loss": 0.4811,
      "step": 11419
    },
    {
      "epoch": 0.8378807193888314,
      "grad_norm": 1.4456574396833937,
      "learning_rate": 6.732434641097563e-07,
      "loss": 0.6531,
      "step": 11420
    },
    {
      "epoch": 0.8379540889789704,
      "grad_norm": 1.5121700327300158,
      "learning_rate": 6.726481026793374e-07,
      "loss": 0.7659,
      "step": 11421
    },
    {
      "epoch": 0.8380274585691095,
      "grad_norm": 1.3992293975479992,
      "learning_rate": 6.720529856251967e-07,
      "loss": 0.6859,
      "step": 11422
    },
    {
      "epoch": 0.8381008281592487,
      "grad_norm": 0.5707520531439332,
      "learning_rate": 6.714581129809422e-07,
      "loss": 0.4749,
      "step": 11423
    },
    {
      "epoch": 0.8381741977493878,
      "grad_norm": 0.5452613656854529,
      "learning_rate": 6.70863484780167e-07,
      "loss": 0.465,
      "step": 11424
    },
    {
      "epoch": 0.8381741977493878,
      "lm_loss": 1.2831,
      "step": 11424,
      "vm_loss": 0.0586
    },
    {
      "epoch": 0.8381741977493878,
      "lm_loss": 0.5954,
      "step": 11424,
      "vm_loss": 0.1936
    },
    {
      "epoch": 0.8381741977493878,
      "lm_loss": 0.0367,
      "step": 11424,
      "vm_loss": 0.1029
    },
    {
      "epoch": 0.8381741977493878,
      "lm_loss": 0.0796,
      "step": 11424,
      "vm_loss": 0.0531
    },
    {
      "epoch": 0.8381741977493878,
      "lm_loss": 0.8829,
      "step": 11424,
      "vm_loss": 0.1293
    },
    {
      "epoch": 0.8381741977493878,
      "lm_loss": 1.026,
      "step": 11424,
      "vm_loss": 0.0704
    },
    {
      "epoch": 0.8381741977493878,
      "lm_loss": 0.7247,
      "step": 11424,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.8381741977493878,
      "lm_loss": 0.0753,
      "step": 11424,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.838247567339527,
      "grad_norm": 1.3338611644890737,
      "learning_rate": 6.702691010564522e-07,
      "loss": 0.666,
      "step": 11425
    },
    {
      "epoch": 0.838320936929666,
      "grad_norm": 1.4448380452221392,
      "learning_rate": 6.696749618433624e-07,
      "loss": 0.6511,
      "step": 11426
    },
    {
      "epoch": 0.8383943065198052,
      "grad_norm": 1.5570722994066084,
      "learning_rate": 6.690810671744524e-07,
      "loss": 0.6023,
      "step": 11427
    },
    {
      "epoch": 0.8384676761099443,
      "grad_norm": 1.572336755449659,
      "learning_rate": 6.684874170832594e-07,
      "loss": 0.7155,
      "step": 11428
    },
    {
      "epoch": 0.8385410457000835,
      "grad_norm": 1.433081239683621,
      "learning_rate": 6.678940116033095e-07,
      "loss": 0.7122,
      "step": 11429
    },
    {
      "epoch": 0.8386144152902226,
      "grad_norm": 1.3965437020384963,
      "learning_rate": 6.673008507681128e-07,
      "loss": 0.6482,
      "step": 11430
    },
    {
      "epoch": 0.8386877848803617,
      "grad_norm": 1.5604740111480815,
      "learning_rate": 6.667079346111655e-07,
      "loss": 0.7285,
      "step": 11431
    },
    {
      "epoch": 0.8387611544705008,
      "grad_norm": 0.5489735260634132,
      "learning_rate": 6.66115263165954e-07,
      "loss": 0.4726,
      "step": 11432
    },
    {
      "epoch": 0.8387611544705008,
      "lm_loss": 0.5118,
      "step": 11432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8387611544705008,
      "lm_loss": 0.2683,
      "step": 11432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8387611544705008,
      "lm_loss": 0.4727,
      "step": 11432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8387611544705008,
      "lm_loss": 0.4504,
      "step": 11432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8387611544705008,
      "lm_loss": 0.4184,
      "step": 11432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8387611544705008,
      "lm_loss": 0.7238,
      "step": 11432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8387611544705008,
      "lm_loss": 0.3526,
      "step": 11432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8387611544705008,
      "lm_loss": 0.3788,
      "step": 11432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.83883452406064,
      "grad_norm": 0.5391857276611346,
      "learning_rate": 6.655228364659455e-07,
      "loss": 0.4779,
      "step": 11433
    },
    {
      "epoch": 0.8389078936507791,
      "grad_norm": 1.6348514297661159,
      "learning_rate": 6.649306545445972e-07,
      "loss": 0.7836,
      "step": 11434
    },
    {
      "epoch": 0.8389812632409183,
      "grad_norm": 0.5370601757692273,
      "learning_rate": 6.643387174353499e-07,
      "loss": 0.4858,
      "step": 11435
    },
    {
      "epoch": 0.8390546328310573,
      "grad_norm": 1.4822786981579252,
      "learning_rate": 6.637470251716322e-07,
      "loss": 0.7797,
      "step": 11436
    },
    {
      "epoch": 0.8391280024211965,
      "grad_norm": 1.5307979945667767,
      "learning_rate": 6.631555777868587e-07,
      "loss": 0.7626,
      "step": 11437
    },
    {
      "epoch": 0.8392013720113356,
      "grad_norm": 0.5675311090320958,
      "learning_rate": 6.6256437531443e-07,
      "loss": 0.486,
      "step": 11438
    },
    {
      "epoch": 0.8392747416014747,
      "grad_norm": 1.4603115561913644,
      "learning_rate": 6.619734177877324e-07,
      "loss": 0.8075,
      "step": 11439
    },
    {
      "epoch": 0.8393481111916139,
      "grad_norm": 1.607389879033581,
      "learning_rate": 6.613827052401389e-07,
      "loss": 0.7045,
      "step": 11440
    },
    {
      "epoch": 0.8393481111916139,
      "lm_loss": 1.6387,
      "step": 11440,
      "vm_loss": 0.1032
    },
    {
      "epoch": 0.8393481111916139,
      "lm_loss": 1.7844,
      "step": 11440,
      "vm_loss": 0.1978
    },
    {
      "epoch": 0.8393481111916139,
      "lm_loss": 0.1325,
      "step": 11440,
      "vm_loss": 0.1989
    },
    {
      "epoch": 0.8393481111916139,
      "lm_loss": 0.6186,
      "step": 11440,
      "vm_loss": 0.1698
    },
    {
      "epoch": 0.8393481111916139,
      "lm_loss": 0.0056,
      "step": 11440,
      "vm_loss": 0.1176
    },
    {
      "epoch": 0.8393481111916139,
      "lm_loss": 0.9798,
      "step": 11440,
      "vm_loss": 0.1185
    },
    {
      "epoch": 0.8393481111916139,
      "lm_loss": 0.5631,
      "step": 11440,
      "vm_loss": 0.2368
    },
    {
      "epoch": 0.8393481111916139,
      "lm_loss": 0.8166,
      "step": 11440,
      "vm_loss": 0.1977
    },
    {
      "epoch": 0.8394214807817529,
      "grad_norm": 1.6212134457443903,
      "learning_rate": 6.607922377050086e-07,
      "loss": 0.8471,
      "step": 11441
    },
    {
      "epoch": 0.8394948503718921,
      "grad_norm": 1.6239324651730291,
      "learning_rate": 6.602020152156851e-07,
      "loss": 0.8244,
      "step": 11442
    },
    {
      "epoch": 0.8395682199620312,
      "grad_norm": 1.4286172823494143,
      "learning_rate": 6.596120378055026e-07,
      "loss": 0.7781,
      "step": 11443
    },
    {
      "epoch": 0.8396415895521704,
      "grad_norm": 1.71993551219483,
      "learning_rate": 6.590223055077772e-07,
      "loss": 0.7069,
      "step": 11444
    },
    {
      "epoch": 0.8397149591423095,
      "grad_norm": 1.412438387787481,
      "learning_rate": 6.584328183558119e-07,
      "loss": 0.6809,
      "step": 11445
    },
    {
      "epoch": 0.8397883287324486,
      "grad_norm": 1.725350854969627,
      "learning_rate": 6.578435763828977e-07,
      "loss": 0.743,
      "step": 11446
    },
    {
      "epoch": 0.8398616983225877,
      "grad_norm": 1.2885239776912278,
      "learning_rate": 6.572545796223085e-07,
      "loss": 0.7358,
      "step": 11447
    },
    {
      "epoch": 0.8399350679127269,
      "grad_norm": 0.5602554961998335,
      "learning_rate": 6.566658281073085e-07,
      "loss": 0.5034,
      "step": 11448
    },
    {
      "epoch": 0.8399350679127269,
      "lm_loss": 1.1047,
      "step": 11448,
      "vm_loss": 0.0745
    },
    {
      "epoch": 0.8399350679127269,
      "lm_loss": 1.0604,
      "step": 11448,
      "vm_loss": 0.0638
    },
    {
      "epoch": 0.8399350679127269,
      "lm_loss": 0.6385,
      "step": 11448,
      "vm_loss": 0.0697
    },
    {
      "epoch": 0.8399350679127269,
      "lm_loss": 0.2617,
      "step": 11448,
      "vm_loss": 0.0889
    },
    {
      "epoch": 0.8399350679127269,
      "lm_loss": 0.6529,
      "step": 11448,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.8399350679127269,
      "lm_loss": 0.002,
      "step": 11448,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.8399350679127269,
      "lm_loss": 0.1249,
      "step": 11448,
      "vm_loss": 0.1158
    },
    {
      "epoch": 0.8399350679127269,
      "lm_loss": 0.9613,
      "step": 11448,
      "vm_loss": 0.1294
    },
    {
      "epoch": 0.840008437502866,
      "grad_norm": 1.57405525026819,
      "learning_rate": 6.560773218711458e-07,
      "loss": 0.8012,
      "step": 11449
    },
    {
      "epoch": 0.8400818070930052,
      "grad_norm": 0.5441750298155819,
      "learning_rate": 6.554890609470543e-07,
      "loss": 0.4898,
      "step": 11450
    },
    {
      "epoch": 0.8401551766831442,
      "grad_norm": 0.5444142835765664,
      "learning_rate": 6.549010453682542e-07,
      "loss": 0.4601,
      "step": 11451
    },
    {
      "epoch": 0.8402285462732834,
      "grad_norm": 0.544963749717017,
      "learning_rate": 6.543132751679527e-07,
      "loss": 0.4727,
      "step": 11452
    },
    {
      "epoch": 0.8403019158634225,
      "grad_norm": 0.5518623662002973,
      "learning_rate": 6.537257503793427e-07,
      "loss": 0.4764,
      "step": 11453
    },
    {
      "epoch": 0.8403752854535617,
      "grad_norm": 1.4899113488865434,
      "learning_rate": 6.531384710356025e-07,
      "loss": 0.6903,
      "step": 11454
    },
    {
      "epoch": 0.8404486550437008,
      "grad_norm": 1.4692871364197728,
      "learning_rate": 6.525514371698982e-07,
      "loss": 0.7925,
      "step": 11455
    },
    {
      "epoch": 0.8405220246338398,
      "grad_norm": 1.5852106297462971,
      "learning_rate": 6.519646488153797e-07,
      "loss": 0.7857,
      "step": 11456
    },
    {
      "epoch": 0.8405220246338398,
      "lm_loss": 2.6597,
      "step": 11456,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.8405220246338398,
      "lm_loss": 0.2473,
      "step": 11456,
      "vm_loss": 0.1089
    },
    {
      "epoch": 0.8405220246338398,
      "lm_loss": 0.1966,
      "step": 11456,
      "vm_loss": 0.1198
    },
    {
      "epoch": 0.8405220246338398,
      "lm_loss": 0.4925,
      "step": 11456,
      "vm_loss": 0.0768
    },
    {
      "epoch": 0.8405220246338398,
      "lm_loss": 0.9759,
      "step": 11456,
      "vm_loss": 0.138
    },
    {
      "epoch": 0.8405220246338398,
      "lm_loss": 0.3906,
      "step": 11456,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.8405220246338398,
      "lm_loss": 0.4835,
      "step": 11456,
      "vm_loss": 0.1369
    },
    {
      "epoch": 0.8405220246338398,
      "lm_loss": 0.5083,
      "step": 11456,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.840595394223979,
      "grad_norm": 1.4820069340232835,
      "learning_rate": 6.513781060051849e-07,
      "loss": 0.7012,
      "step": 11457
    },
    {
      "epoch": 0.8406687638141181,
      "grad_norm": 1.5795838713484878,
      "learning_rate": 6.507918087724385e-07,
      "loss": 0.797,
      "step": 11458
    },
    {
      "epoch": 0.8407421334042573,
      "grad_norm": 1.4646019586249677,
      "learning_rate": 6.50205757150249e-07,
      "loss": 0.7419,
      "step": 11459
    },
    {
      "epoch": 0.8408155029943964,
      "grad_norm": 1.5247243380260183,
      "learning_rate": 6.496199511717127e-07,
      "loss": 0.7056,
      "step": 11460
    },
    {
      "epoch": 0.8408888725845355,
      "grad_norm": 0.5543860086666603,
      "learning_rate": 6.490343908699115e-07,
      "loss": 0.4756,
      "step": 11461
    },
    {
      "epoch": 0.8409622421746746,
      "grad_norm": 1.3344582814928427,
      "learning_rate": 6.484490762779117e-07,
      "loss": 0.7006,
      "step": 11462
    },
    {
      "epoch": 0.8410356117648138,
      "grad_norm": 1.5726277436958656,
      "learning_rate": 6.478640074287707e-07,
      "loss": 0.6713,
      "step": 11463
    },
    {
      "epoch": 0.8411089813549529,
      "grad_norm": 0.5282271379387015,
      "learning_rate": 6.472791843555265e-07,
      "loss": 0.4566,
      "step": 11464
    },
    {
      "epoch": 0.8411089813549529,
      "lm_loss": 1.4314,
      "step": 11464,
      "vm_loss": 0.1395
    },
    {
      "epoch": 0.8411089813549529,
      "lm_loss": 1.4193,
      "step": 11464,
      "vm_loss": 0.1826
    },
    {
      "epoch": 0.8411089813549529,
      "lm_loss": 0.0579,
      "step": 11464,
      "vm_loss": 0.0924
    },
    {
      "epoch": 0.8411089813549529,
      "lm_loss": 0.1828,
      "step": 11464,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.8411089813549529,
      "lm_loss": 0.9868,
      "step": 11464,
      "vm_loss": 0.1964
    },
    {
      "epoch": 0.8411089813549529,
      "lm_loss": 0.8414,
      "step": 11464,
      "vm_loss": 0.2282
    },
    {
      "epoch": 0.8411089813549529,
      "lm_loss": 0.3876,
      "step": 11464,
      "vm_loss": 0.1868
    },
    {
      "epoch": 0.8411089813549529,
      "lm_loss": 0.0688,
      "step": 11464,
      "vm_loss": 0.0592
    },
    {
      "epoch": 0.8411823509450921,
      "grad_norm": 1.5121909471842925,
      "learning_rate": 6.466946070912061e-07,
      "loss": 0.7933,
      "step": 11465
    },
    {
      "epoch": 0.8412557205352311,
      "grad_norm": 1.5797539135223204,
      "learning_rate": 6.461102756688226e-07,
      "loss": 0.7579,
      "step": 11466
    },
    {
      "epoch": 0.8413290901253703,
      "grad_norm": 0.5515097265162966,
      "learning_rate": 6.455261901213733e-07,
      "loss": 0.4725,
      "step": 11467
    },
    {
      "epoch": 0.8414024597155094,
      "grad_norm": 1.6212382472118596,
      "learning_rate": 6.44942350481843e-07,
      "loss": 0.8089,
      "step": 11468
    },
    {
      "epoch": 0.8414758293056486,
      "grad_norm": 1.5966457594821712,
      "learning_rate": 6.443587567832044e-07,
      "loss": 0.7454,
      "step": 11469
    },
    {
      "epoch": 0.8415491988957877,
      "grad_norm": 1.530430885585768,
      "learning_rate": 6.437754090584136e-07,
      "loss": 0.7425,
      "step": 11470
    },
    {
      "epoch": 0.8416225684859268,
      "grad_norm": 0.539011534854462,
      "learning_rate": 6.431923073404128e-07,
      "loss": 0.4693,
      "step": 11471
    },
    {
      "epoch": 0.8416959380760659,
      "grad_norm": 1.5475525399075736,
      "learning_rate": 6.426094516621323e-07,
      "loss": 0.7214,
      "step": 11472
    },
    {
      "epoch": 0.8416959380760659,
      "lm_loss": 0.6092,
      "step": 11472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8416959380760659,
      "lm_loss": 0.5204,
      "step": 11472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8416959380760659,
      "lm_loss": 0.4271,
      "step": 11472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8416959380760659,
      "lm_loss": 0.4685,
      "step": 11472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8416959380760659,
      "lm_loss": 0.3795,
      "step": 11472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8416959380760659,
      "lm_loss": 0.4755,
      "step": 11472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8416959380760659,
      "lm_loss": 0.3592,
      "step": 11472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8416959380760659,
      "lm_loss": 0.3889,
      "step": 11472,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8417693076662051,
      "grad_norm": 0.5628470731322315,
      "learning_rate": 6.420268420564857e-07,
      "loss": 0.4692,
      "step": 11473
    },
    {
      "epoch": 0.8418426772563442,
      "grad_norm": 0.5406538636091475,
      "learning_rate": 6.414444785563773e-07,
      "loss": 0.4792,
      "step": 11474
    },
    {
      "epoch": 0.8419160468464834,
      "grad_norm": 1.4398207723407759,
      "learning_rate": 6.408623611946923e-07,
      "loss": 0.7101,
      "step": 11475
    },
    {
      "epoch": 0.8419894164366224,
      "grad_norm": 1.7137200149362048,
      "learning_rate": 6.402804900043052e-07,
      "loss": 0.6663,
      "step": 11476
    },
    {
      "epoch": 0.8420627860267615,
      "grad_norm": 0.5287050384608136,
      "learning_rate": 6.396988650180758e-07,
      "loss": 0.4635,
      "step": 11477
    },
    {
      "epoch": 0.8421361556169007,
      "grad_norm": 1.4117073322560272,
      "learning_rate": 6.391174862688482e-07,
      "loss": 0.6023,
      "step": 11478
    },
    {
      "epoch": 0.8422095252070398,
      "grad_norm": 1.5206379814902706,
      "learning_rate": 6.385363537894568e-07,
      "loss": 0.7231,
      "step": 11479
    },
    {
      "epoch": 0.842282894797179,
      "grad_norm": 1.521958299297348,
      "learning_rate": 6.379554676127186e-07,
      "loss": 0.7592,
      "step": 11480
    },
    {
      "epoch": 0.842282894797179,
      "lm_loss": 0.257,
      "step": 11480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.842282894797179,
      "lm_loss": 0.5352,
      "step": 11480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.842282894797179,
      "lm_loss": 0.4299,
      "step": 11480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.842282894797179,
      "lm_loss": 0.3879,
      "step": 11480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.842282894797179,
      "lm_loss": 0.4496,
      "step": 11480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.842282894797179,
      "lm_loss": 0.5883,
      "step": 11480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.842282894797179,
      "lm_loss": 0.6455,
      "step": 11480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.842282894797179,
      "lm_loss": 0.3402,
      "step": 11480,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.842356264387318,
      "grad_norm": 0.5472900766654495,
      "learning_rate": 6.373748277714376e-07,
      "loss": 0.4766,
      "step": 11481
    },
    {
      "epoch": 0.8424296339774572,
      "grad_norm": 0.5507663175019011,
      "learning_rate": 6.367944342984034e-07,
      "loss": 0.4784,
      "step": 11482
    },
    {
      "epoch": 0.8425030035675963,
      "grad_norm": 1.6679036641501483,
      "learning_rate": 6.362142872263933e-07,
      "loss": 0.7259,
      "step": 11483
    },
    {
      "epoch": 0.8425763731577355,
      "grad_norm": 1.4753892065080279,
      "learning_rate": 6.356343865881681e-07,
      "loss": 0.6925,
      "step": 11484
    },
    {
      "epoch": 0.8426497427478745,
      "grad_norm": 1.4812912207126245,
      "learning_rate": 6.350547324164779e-07,
      "loss": 0.7678,
      "step": 11485
    },
    {
      "epoch": 0.8427231123380137,
      "grad_norm": 1.5958045122377607,
      "learning_rate": 6.344753247440571e-07,
      "loss": 0.7437,
      "step": 11486
    },
    {
      "epoch": 0.8427964819281528,
      "grad_norm": 1.7462793947784074,
      "learning_rate": 6.338961636036256e-07,
      "loss": 0.6737,
      "step": 11487
    },
    {
      "epoch": 0.842869851518292,
      "grad_norm": 1.5101340736371094,
      "learning_rate": 6.3331724902789e-07,
      "loss": 0.8246,
      "step": 11488
    },
    {
      "epoch": 0.842869851518292,
      "lm_loss": 0.8982,
      "step": 11488,
      "vm_loss": 0.1759
    },
    {
      "epoch": 0.842869851518292,
      "lm_loss": 0.7541,
      "step": 11488,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.842869851518292,
      "lm_loss": 1.8628,
      "step": 11488,
      "vm_loss": 0.0894
    },
    {
      "epoch": 0.842869851518292,
      "lm_loss": 0.4481,
      "step": 11488,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.842869851518292,
      "lm_loss": 0.6635,
      "step": 11488,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.842869851518292,
      "lm_loss": 1.682,
      "step": 11488,
      "vm_loss": 0.0987
    },
    {
      "epoch": 0.842869851518292,
      "lm_loss": 0.297,
      "step": 11488,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.842869851518292,
      "lm_loss": 0.0949,
      "step": 11488,
      "vm_loss": 0.0917
    },
    {
      "epoch": 0.8429432211084311,
      "grad_norm": 1.7058820016815206,
      "learning_rate": 6.327385810495423e-07,
      "loss": 0.7716,
      "step": 11489
    },
    {
      "epoch": 0.8430165906985702,
      "grad_norm": 1.6851423968191748,
      "learning_rate": 6.321601597012633e-07,
      "loss": 0.7907,
      "step": 11490
    },
    {
      "epoch": 0.8430899602887093,
      "grad_norm": 0.5363985961268162,
      "learning_rate": 6.315819850157168e-07,
      "loss": 0.4805,
      "step": 11491
    },
    {
      "epoch": 0.8431633298788485,
      "grad_norm": 1.4926124443836553,
      "learning_rate": 6.310040570255538e-07,
      "loss": 0.6928,
      "step": 11492
    },
    {
      "epoch": 0.8432366994689876,
      "grad_norm": 0.5487358180445641,
      "learning_rate": 6.304263757634116e-07,
      "loss": 0.4761,
      "step": 11493
    },
    {
      "epoch": 0.8433100690591268,
      "grad_norm": 1.5047757300375069,
      "learning_rate": 6.29848941261913e-07,
      "loss": 0.7665,
      "step": 11494
    },
    {
      "epoch": 0.8433834386492658,
      "grad_norm": 1.6794484208653833,
      "learning_rate": 6.292717535536658e-07,
      "loss": 0.673,
      "step": 11495
    },
    {
      "epoch": 0.8434568082394049,
      "grad_norm": 0.5452316791288587,
      "learning_rate": 6.286948126712677e-07,
      "loss": 0.4703,
      "step": 11496
    },
    {
      "epoch": 0.8434568082394049,
      "lm_loss": 0.466,
      "step": 11496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8434568082394049,
      "lm_loss": 0.45,
      "step": 11496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8434568082394049,
      "lm_loss": 0.5213,
      "step": 11496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8434568082394049,
      "lm_loss": 0.2654,
      "step": 11496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8434568082394049,
      "lm_loss": 0.4831,
      "step": 11496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8434568082394049,
      "lm_loss": 0.3063,
      "step": 11496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8434568082394049,
      "lm_loss": 0.5405,
      "step": 11496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8434568082394049,
      "lm_loss": 0.4353,
      "step": 11496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8435301778295441,
      "grad_norm": 0.5582176779870615,
      "learning_rate": 6.28118118647299e-07,
      "loss": 0.4742,
      "step": 11497
    },
    {
      "epoch": 0.8436035474196832,
      "grad_norm": 1.4359867385571488,
      "learning_rate": 6.275416715143273e-07,
      "loss": 0.6988,
      "step": 11498
    },
    {
      "epoch": 0.8436769170098224,
      "grad_norm": 0.549642073749554,
      "learning_rate": 6.269654713049051e-07,
      "loss": 0.4529,
      "step": 11499
    },
    {
      "epoch": 0.8437502865999614,
      "grad_norm": 1.501278627084606,
      "learning_rate": 6.263895180515716e-07,
      "loss": 0.799,
      "step": 11500
    },
    {
      "epoch": 0.8438236561901006,
      "grad_norm": 1.5697316644522108,
      "learning_rate": 6.258138117868539e-07,
      "loss": 0.8132,
      "step": 11501
    },
    {
      "epoch": 0.8438970257802397,
      "grad_norm": 1.4329562780952014,
      "learning_rate": 6.252383525432632e-07,
      "loss": 0.7103,
      "step": 11502
    },
    {
      "epoch": 0.8439703953703789,
      "grad_norm": 0.5391450365119932,
      "learning_rate": 6.24663140353296e-07,
      "loss": 0.473,
      "step": 11503
    },
    {
      "epoch": 0.844043764960518,
      "grad_norm": 1.4635785465313558,
      "learning_rate": 6.240881752494371e-07,
      "loss": 0.7363,
      "step": 11504
    },
    {
      "epoch": 0.844043764960518,
      "lm_loss": 0.6828,
      "step": 11504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844043764960518,
      "lm_loss": 0.3213,
      "step": 11504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844043764960518,
      "lm_loss": 0.5562,
      "step": 11504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844043764960518,
      "lm_loss": 0.5079,
      "step": 11504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844043764960518,
      "lm_loss": 0.6352,
      "step": 11504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844043764960518,
      "lm_loss": 0.5611,
      "step": 11504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844043764960518,
      "lm_loss": 0.5505,
      "step": 11504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844043764960518,
      "lm_loss": 0.3705,
      "step": 11504,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8441171345506571,
      "grad_norm": 0.5247843152983029,
      "learning_rate": 6.235134572641544e-07,
      "loss": 0.4534,
      "step": 11505
    },
    {
      "epoch": 0.8441905041407962,
      "grad_norm": 0.5433134383382319,
      "learning_rate": 6.229389864299062e-07,
      "loss": 0.4647,
      "step": 11506
    },
    {
      "epoch": 0.8442638737309354,
      "grad_norm": 1.5883216730787633,
      "learning_rate": 6.223647627791329e-07,
      "loss": 0.663,
      "step": 11507
    },
    {
      "epoch": 0.8443372433210745,
      "grad_norm": 1.6296915756064143,
      "learning_rate": 6.217907863442624e-07,
      "loss": 0.8006,
      "step": 11508
    },
    {
      "epoch": 0.8444106129112137,
      "grad_norm": 0.5510438540414089,
      "learning_rate": 6.212170571577087e-07,
      "loss": 0.48,
      "step": 11509
    },
    {
      "epoch": 0.8444839825013527,
      "grad_norm": 0.5703836233658065,
      "learning_rate": 6.206435752518719e-07,
      "loss": 0.4917,
      "step": 11510
    },
    {
      "epoch": 0.8445573520914919,
      "grad_norm": 1.6162051054962752,
      "learning_rate": 6.200703406591362e-07,
      "loss": 0.7313,
      "step": 11511
    },
    {
      "epoch": 0.844630721681631,
      "grad_norm": 1.5541997727427133,
      "learning_rate": 6.194973534118764e-07,
      "loss": 0.7341,
      "step": 11512
    },
    {
      "epoch": 0.844630721681631,
      "lm_loss": 0.5867,
      "step": 11512,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844630721681631,
      "lm_loss": 0.5709,
      "step": 11512,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844630721681631,
      "lm_loss": 0.5069,
      "step": 11512,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844630721681631,
      "lm_loss": 0.3672,
      "step": 11512,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844630721681631,
      "lm_loss": 0.4145,
      "step": 11512,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844630721681631,
      "lm_loss": 0.3993,
      "step": 11512,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844630721681631,
      "lm_loss": 0.5732,
      "step": 11512,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.844630721681631,
      "lm_loss": 0.4267,
      "step": 11512,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8447040912717702,
      "grad_norm": 0.5472204934478421,
      "learning_rate": 6.189246135424493e-07,
      "loss": 0.4758,
      "step": 11513
    },
    {
      "epoch": 0.8447774608619093,
      "grad_norm": 0.5548570130236852,
      "learning_rate": 6.183521210831989e-07,
      "loss": 0.4903,
      "step": 11514
    },
    {
      "epoch": 0.8448508304520483,
      "grad_norm": 1.3737302880394255,
      "learning_rate": 6.177798760664549e-07,
      "loss": 0.6749,
      "step": 11515
    },
    {
      "epoch": 0.8449242000421875,
      "grad_norm": 1.6816276271029733,
      "learning_rate": 6.172078785245322e-07,
      "loss": 0.7453,
      "step": 11516
    },
    {
      "epoch": 0.8449975696323266,
      "grad_norm": 0.5296008034045601,
      "learning_rate": 6.166361284897365e-07,
      "loss": 0.465,
      "step": 11517
    },
    {
      "epoch": 0.8450709392224658,
      "grad_norm": 1.5552135495075619,
      "learning_rate": 6.160646259943532e-07,
      "loss": 0.7416,
      "step": 11518
    },
    {
      "epoch": 0.8451443088126049,
      "grad_norm": 1.4419189318396375,
      "learning_rate": 6.154933710706573e-07,
      "loss": 0.7284,
      "step": 11519
    },
    {
      "epoch": 0.845217678402744,
      "grad_norm": 0.5588144819105214,
      "learning_rate": 6.149223637509083e-07,
      "loss": 0.4806,
      "step": 11520
    },
    {
      "epoch": 0.845217678402744,
      "lm_loss": 0.8633,
      "step": 11520,
      "vm_loss": 0.0537
    },
    {
      "epoch": 0.845217678402744,
      "lm_loss": 0.5929,
      "step": 11520,
      "vm_loss": 0.0819
    },
    {
      "epoch": 0.845217678402744,
      "lm_loss": 0.6047,
      "step": 11520,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.845217678402744,
      "lm_loss": 0.4791,
      "step": 11520,
      "vm_loss": 0.2316
    },
    {
      "epoch": 0.845217678402744,
      "lm_loss": 0.1898,
      "step": 11520,
      "vm_loss": 0.1524
    },
    {
      "epoch": 0.845217678402744,
      "lm_loss": 0.2225,
      "step": 11520,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.845217678402744,
      "lm_loss": 0.5657,
      "step": 11520,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.845217678402744,
      "lm_loss": 0.0332,
      "step": 11520,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.8452910479928831,
      "grad_norm": 1.3224371281071412,
      "learning_rate": 6.143516040673536e-07,
      "loss": 0.7034,
      "step": 11521
    },
    {
      "epoch": 0.8453644175830223,
      "grad_norm": 1.7667614416077,
      "learning_rate": 6.13781092052223e-07,
      "loss": 0.6435,
      "step": 11522
    },
    {
      "epoch": 0.8454377871731614,
      "grad_norm": 1.439625495378273,
      "learning_rate": 6.13210827737738e-07,
      "loss": 0.7787,
      "step": 11523
    },
    {
      "epoch": 0.8455111567633006,
      "grad_norm": 1.4331307928336963,
      "learning_rate": 6.126408111561011e-07,
      "loss": 0.6882,
      "step": 11524
    },
    {
      "epoch": 0.8455845263534396,
      "grad_norm": 1.5780697454025119,
      "learning_rate": 6.120710423395027e-07,
      "loss": 0.778,
      "step": 11525
    },
    {
      "epoch": 0.8456578959435788,
      "grad_norm": 1.6494052603194662,
      "learning_rate": 6.115015213201186e-07,
      "loss": 0.6915,
      "step": 11526
    },
    {
      "epoch": 0.8457312655337179,
      "grad_norm": 1.5237055311392844,
      "learning_rate": 6.109322481301111e-07,
      "loss": 0.7091,
      "step": 11527
    },
    {
      "epoch": 0.8458046351238571,
      "grad_norm": 1.5225464437479999,
      "learning_rate": 6.103632228016299e-07,
      "loss": 0.6625,
      "step": 11528
    },
    {
      "epoch": 0.8458046351238571,
      "lm_loss": 1.2746,
      "step": 11528,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.8458046351238571,
      "lm_loss": 1.8202,
      "step": 11528,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.8458046351238571,
      "lm_loss": 1.0688,
      "step": 11528,
      "vm_loss": 0.1487
    },
    {
      "epoch": 0.8458046351238571,
      "lm_loss": 0.8397,
      "step": 11528,
      "vm_loss": 0.1477
    },
    {
      "epoch": 0.8458046351238571,
      "lm_loss": 1.3142,
      "step": 11528,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.8458046351238571,
      "lm_loss": 0.117,
      "step": 11528,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.8458046351238571,
      "lm_loss": 0.2859,
      "step": 11528,
      "vm_loss": 0.141
    },
    {
      "epoch": 0.8458046351238571,
      "lm_loss": 0.6172,
      "step": 11528,
      "vm_loss": 0.1081
    },
    {
      "epoch": 0.8458780047139962,
      "grad_norm": 1.7401660410858675,
      "learning_rate": 6.097944453668081e-07,
      "loss": 0.8062,
      "step": 11529
    },
    {
      "epoch": 0.8459513743041353,
      "grad_norm": 0.5508035390141515,
      "learning_rate": 6.092259158577657e-07,
      "loss": 0.4935,
      "step": 11530
    },
    {
      "epoch": 0.8460247438942744,
      "grad_norm": 1.6496739741579847,
      "learning_rate": 6.086576343066103e-07,
      "loss": 0.7722,
      "step": 11531
    },
    {
      "epoch": 0.8460981134844135,
      "grad_norm": 1.44608351829088,
      "learning_rate": 6.080896007454313e-07,
      "loss": 0.7195,
      "step": 11532
    },
    {
      "epoch": 0.8461714830745527,
      "grad_norm": 0.557236034578997,
      "learning_rate": 6.075218152063106e-07,
      "loss": 0.4814,
      "step": 11533
    },
    {
      "epoch": 0.8462448526646918,
      "grad_norm": 1.3982689835186244,
      "learning_rate": 6.069542777213105e-07,
      "loss": 0.6769,
      "step": 11534
    },
    {
      "epoch": 0.8463182222548309,
      "grad_norm": 1.443584498131484,
      "learning_rate": 6.063869883224816e-07,
      "loss": 0.7304,
      "step": 11535
    },
    {
      "epoch": 0.84639159184497,
      "grad_norm": 0.5312295255077576,
      "learning_rate": 6.058199470418602e-07,
      "loss": 0.4558,
      "step": 11536
    },
    {
      "epoch": 0.84639159184497,
      "lm_loss": 0.1295,
      "step": 11536,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.84639159184497,
      "lm_loss": 0.5406,
      "step": 11536,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.84639159184497,
      "lm_loss": 0.7718,
      "step": 11536,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.84639159184497,
      "lm_loss": 0.0457,
      "step": 11536,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.84639159184497,
      "lm_loss": 1.0235,
      "step": 11536,
      "vm_loss": 0.1406
    },
    {
      "epoch": 0.84639159184497,
      "lm_loss": 0.1053,
      "step": 11536,
      "vm_loss": 0.196
    },
    {
      "epoch": 0.84639159184497,
      "lm_loss": 0.6459,
      "step": 11536,
      "vm_loss": 0.2303
    },
    {
      "epoch": 0.84639159184497,
      "lm_loss": 0.1199,
      "step": 11536,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.8464649614351092,
      "grad_norm": 1.5621044281689163,
      "learning_rate": 6.052531539114681e-07,
      "loss": 0.7443,
      "step": 11537
    },
    {
      "epoch": 0.8465383310252483,
      "grad_norm": 1.5296162495307122,
      "learning_rate": 6.04686608963313e-07,
      "loss": 0.7718,
      "step": 11538
    },
    {
      "epoch": 0.8466117006153875,
      "grad_norm": 1.6491100259667448,
      "learning_rate": 6.04120312229391e-07,
      "loss": 0.7902,
      "step": 11539
    },
    {
      "epoch": 0.8466850702055265,
      "grad_norm": 0.5569134877388832,
      "learning_rate": 6.035542637416814e-07,
      "loss": 0.4913,
      "step": 11540
    },
    {
      "epoch": 0.8467584397956657,
      "grad_norm": 0.5487315796378798,
      "learning_rate": 6.029884635321498e-07,
      "loss": 0.4788,
      "step": 11541
    },
    {
      "epoch": 0.8468318093858048,
      "grad_norm": 1.4785252646416158,
      "learning_rate": 6.024229116327485e-07,
      "loss": 0.7036,
      "step": 11542
    },
    {
      "epoch": 0.846905178975944,
      "grad_norm": 0.5563437340839777,
      "learning_rate": 6.018576080754151e-07,
      "loss": 0.4756,
      "step": 11543
    },
    {
      "epoch": 0.8469785485660831,
      "grad_norm": 1.469082444288728,
      "learning_rate": 6.012925528920749e-07,
      "loss": 0.7002,
      "step": 11544
    },
    {
      "epoch": 0.8469785485660831,
      "lm_loss": 0.5793,
      "step": 11544,
      "vm_loss": 0.0568
    },
    {
      "epoch": 0.8469785485660831,
      "lm_loss": 0.8691,
      "step": 11544,
      "vm_loss": 0.128
    },
    {
      "epoch": 0.8469785485660831,
      "lm_loss": 1.013,
      "step": 11544,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.8469785485660831,
      "lm_loss": 0.2453,
      "step": 11544,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.8469785485660831,
      "lm_loss": 0.3375,
      "step": 11544,
      "vm_loss": 0.1166
    },
    {
      "epoch": 0.8469785485660831,
      "lm_loss": 0.244,
      "step": 11544,
      "vm_loss": 0.0491
    },
    {
      "epoch": 0.8469785485660831,
      "lm_loss": 0.9202,
      "step": 11544,
      "vm_loss": 0.1507
    },
    {
      "epoch": 0.8469785485660831,
      "lm_loss": 0.147,
      "step": 11544,
      "vm_loss": 0.1422
    },
    {
      "epoch": 0.8470519181562222,
      "grad_norm": 1.5681327103826126,
      "learning_rate": 6.007277461146383e-07,
      "loss": 0.8096,
      "step": 11545
    },
    {
      "epoch": 0.8471252877463613,
      "grad_norm": 1.4626385383901057,
      "learning_rate": 6.001631877749997e-07,
      "loss": 0.7698,
      "step": 11546
    },
    {
      "epoch": 0.8471986573365005,
      "grad_norm": 0.5176992294221863,
      "learning_rate": 5.995988779050421e-07,
      "loss": 0.4638,
      "step": 11547
    },
    {
      "epoch": 0.8472720269266396,
      "grad_norm": 1.4059319732463822,
      "learning_rate": 5.990348165366333e-07,
      "loss": 0.6958,
      "step": 11548
    },
    {
      "epoch": 0.8473453965167788,
      "grad_norm": 0.5537902619372359,
      "learning_rate": 5.984710037016267e-07,
      "loss": 0.4622,
      "step": 11549
    },
    {
      "epoch": 0.8474187661069178,
      "grad_norm": 1.507507171824293,
      "learning_rate": 5.979074394318629e-07,
      "loss": 0.7232,
      "step": 11550
    },
    {
      "epoch": 0.847492135697057,
      "grad_norm": 1.4905247290531367,
      "learning_rate": 5.973441237591682e-07,
      "loss": 0.7272,
      "step": 11551
    },
    {
      "epoch": 0.8475655052871961,
      "grad_norm": 1.5342837216018852,
      "learning_rate": 5.967810567153537e-07,
      "loss": 0.7836,
      "step": 11552
    },
    {
      "epoch": 0.8475655052871961,
      "lm_loss": 0.8293,
      "step": 11552,
      "vm_loss": 0.1044
    },
    {
      "epoch": 0.8475655052871961,
      "lm_loss": 1.2989,
      "step": 11552,
      "vm_loss": 0.11
    },
    {
      "epoch": 0.8475655052871961,
      "lm_loss": 0.7216,
      "step": 11552,
      "vm_loss": 0.1081
    },
    {
      "epoch": 0.8475655052871961,
      "lm_loss": 0.9013,
      "step": 11552,
      "vm_loss": 0.1117
    },
    {
      "epoch": 0.8475655052871961,
      "lm_loss": 0.244,
      "step": 11552,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.8475655052871961,
      "lm_loss": 1.1649,
      "step": 11552,
      "vm_loss": 0.1978
    },
    {
      "epoch": 0.8475655052871961,
      "lm_loss": 0.6639,
      "step": 11552,
      "vm_loss": 0.1451
    },
    {
      "epoch": 0.8475655052871961,
      "lm_loss": 0.2609,
      "step": 11552,
      "vm_loss": 0.1056
    },
    {
      "epoch": 0.8476388748773352,
      "grad_norm": 1.6002162445678465,
      "learning_rate": 5.962182383322173e-07,
      "loss": 0.8462,
      "step": 11553
    },
    {
      "epoch": 0.8477122444674744,
      "grad_norm": 1.5963829380758952,
      "learning_rate": 5.956556686415416e-07,
      "loss": 0.776,
      "step": 11554
    },
    {
      "epoch": 0.8477856140576134,
      "grad_norm": 1.4664465903386856,
      "learning_rate": 5.950933476750986e-07,
      "loss": 0.7007,
      "step": 11555
    },
    {
      "epoch": 0.8478589836477526,
      "grad_norm": 0.5645508874219869,
      "learning_rate": 5.945312754646431e-07,
      "loss": 0.499,
      "step": 11556
    },
    {
      "epoch": 0.8479323532378917,
      "grad_norm": 1.49479959422697,
      "learning_rate": 5.939694520419159e-07,
      "loss": 0.7051,
      "step": 11557
    },
    {
      "epoch": 0.8480057228280309,
      "grad_norm": 1.4803971022909914,
      "learning_rate": 5.934078774386453e-07,
      "loss": 0.777,
      "step": 11558
    },
    {
      "epoch": 0.84807909241817,
      "grad_norm": 1.5290532176369571,
      "learning_rate": 5.92846551686544e-07,
      "loss": 0.7881,
      "step": 11559
    },
    {
      "epoch": 0.8481524620083091,
      "grad_norm": 1.4547950556343185,
      "learning_rate": 5.922854748173123e-07,
      "loss": 0.7775,
      "step": 11560
    },
    {
      "epoch": 0.8481524620083091,
      "lm_loss": 0.641,
      "step": 11560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8481524620083091,
      "lm_loss": 0.5579,
      "step": 11560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8481524620083091,
      "lm_loss": 0.3332,
      "step": 11560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8481524620083091,
      "lm_loss": 0.4983,
      "step": 11560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8481524620083091,
      "lm_loss": 0.4224,
      "step": 11560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8481524620083091,
      "lm_loss": 0.4637,
      "step": 11560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8481524620083091,
      "lm_loss": 0.4309,
      "step": 11560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8481524620083091,
      "lm_loss": 0.5274,
      "step": 11560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8482258315984482,
      "grad_norm": 0.5393019823180755,
      "learning_rate": 5.917246468626358e-07,
      "loss": 0.4579,
      "step": 11561
    },
    {
      "epoch": 0.8482992011885874,
      "grad_norm": 0.5424469740964982,
      "learning_rate": 5.911640678541858e-07,
      "loss": 0.4691,
      "step": 11562
    },
    {
      "epoch": 0.8483725707787265,
      "grad_norm": 1.4390664833161937,
      "learning_rate": 5.906037378236184e-07,
      "loss": 0.6112,
      "step": 11563
    },
    {
      "epoch": 0.8484459403688657,
      "grad_norm": 1.583363281399812,
      "learning_rate": 5.90043656802578e-07,
      "loss": 0.7426,
      "step": 11564
    },
    {
      "epoch": 0.8485193099590047,
      "grad_norm": 1.4658317430779217,
      "learning_rate": 5.894838248226925e-07,
      "loss": 0.7046,
      "step": 11565
    },
    {
      "epoch": 0.8485926795491439,
      "grad_norm": 1.4201737615841146,
      "learning_rate": 5.889242419155783e-07,
      "loss": 0.7703,
      "step": 11566
    },
    {
      "epoch": 0.848666049139283,
      "grad_norm": 0.5440690112850545,
      "learning_rate": 5.883649081128368e-07,
      "loss": 0.4743,
      "step": 11567
    },
    {
      "epoch": 0.8487394187294222,
      "grad_norm": 1.577083452547811,
      "learning_rate": 5.878058234460538e-07,
      "loss": 0.692,
      "step": 11568
    },
    {
      "epoch": 0.8487394187294222,
      "lm_loss": 1.0516,
      "step": 11568,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.8487394187294222,
      "lm_loss": 0.2137,
      "step": 11568,
      "vm_loss": 0.0629
    },
    {
      "epoch": 0.8487394187294222,
      "lm_loss": 0.1313,
      "step": 11568,
      "vm_loss": 0.1556
    },
    {
      "epoch": 0.8487394187294222,
      "lm_loss": 0.3272,
      "step": 11568,
      "vm_loss": 0.1316
    },
    {
      "epoch": 0.8487394187294222,
      "lm_loss": 0.3142,
      "step": 11568,
      "vm_loss": 0.1389
    },
    {
      "epoch": 0.8487394187294222,
      "lm_loss": 1.0826,
      "step": 11568,
      "vm_loss": 0.1202
    },
    {
      "epoch": 0.8487394187294222,
      "lm_loss": 0.8608,
      "step": 11568,
      "vm_loss": 0.2221
    },
    {
      "epoch": 0.8487394187294222,
      "lm_loss": 1.153,
      "step": 11568,
      "vm_loss": 0.1681
    },
    {
      "epoch": 0.8488127883195612,
      "grad_norm": 1.690531403143715,
      "learning_rate": 5.872469879468024e-07,
      "loss": 0.7165,
      "step": 11569
    },
    {
      "epoch": 0.8488861579097003,
      "grad_norm": 1.5317393435934679,
      "learning_rate": 5.866884016466406e-07,
      "loss": 0.6754,
      "step": 11570
    },
    {
      "epoch": 0.8489595274998395,
      "grad_norm": 1.538071962737554,
      "learning_rate": 5.861300645771156e-07,
      "loss": 0.7966,
      "step": 11571
    },
    {
      "epoch": 0.8490328970899786,
      "grad_norm": 1.6190768956435144,
      "learning_rate": 5.85571976769756e-07,
      "loss": 0.697,
      "step": 11572
    },
    {
      "epoch": 0.8491062666801178,
      "grad_norm": 1.4097230691734857,
      "learning_rate": 5.850141382560792e-07,
      "loss": 0.7315,
      "step": 11573
    },
    {
      "epoch": 0.8491796362702568,
      "grad_norm": 0.5548318119921032,
      "learning_rate": 5.844565490675879e-07,
      "loss": 0.4783,
      "step": 11574
    },
    {
      "epoch": 0.849253005860396,
      "grad_norm": 1.4781024924964623,
      "learning_rate": 5.838992092357698e-07,
      "loss": 0.8148,
      "step": 11575
    },
    {
      "epoch": 0.8493263754505351,
      "grad_norm": 1.414131550528575,
      "learning_rate": 5.83342118792099e-07,
      "loss": 0.7389,
      "step": 11576
    },
    {
      "epoch": 0.8493263754505351,
      "lm_loss": 1.2117,
      "step": 11576,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.8493263754505351,
      "lm_loss": 1.3029,
      "step": 11576,
      "vm_loss": 0.1322
    },
    {
      "epoch": 0.8493263754505351,
      "lm_loss": 0.456,
      "step": 11576,
      "vm_loss": 0.1308
    },
    {
      "epoch": 0.8493263754505351,
      "lm_loss": 0.5203,
      "step": 11576,
      "vm_loss": 0.1578
    },
    {
      "epoch": 0.8493263754505351,
      "lm_loss": 1.0981,
      "step": 11576,
      "vm_loss": 0.1871
    },
    {
      "epoch": 0.8493263754505351,
      "lm_loss": 0.1294,
      "step": 11576,
      "vm_loss": 0.0913
    },
    {
      "epoch": 0.8493263754505351,
      "lm_loss": 0.5346,
      "step": 11576,
      "vm_loss": 0.1068
    },
    {
      "epoch": 0.8493263754505351,
      "lm_loss": 1.1304,
      "step": 11576,
      "vm_loss": 0.233
    },
    {
      "epoch": 0.8493997450406743,
      "grad_norm": 1.518775163085797,
      "learning_rate": 5.827852777680371e-07,
      "loss": 0.7543,
      "step": 11577
    },
    {
      "epoch": 0.8494731146308134,
      "grad_norm": 1.4041334129313425,
      "learning_rate": 5.8222868619503e-07,
      "loss": 0.6539,
      "step": 11578
    },
    {
      "epoch": 0.8495464842209525,
      "grad_norm": 1.6076300058527138,
      "learning_rate": 5.816723441045085e-07,
      "loss": 0.7168,
      "step": 11579
    },
    {
      "epoch": 0.8496198538110916,
      "grad_norm": 0.5629806041576744,
      "learning_rate": 5.811162515278923e-07,
      "loss": 0.4792,
      "step": 11580
    },
    {
      "epoch": 0.8496932234012308,
      "grad_norm": 1.4727202111182625,
      "learning_rate": 5.805604084965838e-07,
      "loss": 0.7283,
      "step": 11581
    },
    {
      "epoch": 0.8497665929913699,
      "grad_norm": 1.6797470779360693,
      "learning_rate": 5.80004815041974e-07,
      "loss": 0.7153,
      "step": 11582
    },
    {
      "epoch": 0.8498399625815091,
      "grad_norm": 0.5398767501353972,
      "learning_rate": 5.794494711954385e-07,
      "loss": 0.466,
      "step": 11583
    },
    {
      "epoch": 0.8499133321716481,
      "grad_norm": 0.5531502539650214,
      "learning_rate": 5.788943769883387e-07,
      "loss": 0.4968,
      "step": 11584
    },
    {
      "epoch": 0.8499133321716481,
      "lm_loss": 1.5348,
      "step": 11584,
      "vm_loss": 0.0309
    },
    {
      "epoch": 0.8499133321716481,
      "lm_loss": 0.2913,
      "step": 11584,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.8499133321716481,
      "lm_loss": 0.6163,
      "step": 11584,
      "vm_loss": 0.1874
    },
    {
      "epoch": 0.8499133321716481,
      "lm_loss": 0.9307,
      "step": 11584,
      "vm_loss": 0.083
    },
    {
      "epoch": 0.8499133321716481,
      "lm_loss": 1.1345,
      "step": 11584,
      "vm_loss": 0.0945
    },
    {
      "epoch": 0.8499133321716481,
      "lm_loss": 0.269,
      "step": 11584,
      "vm_loss": 0.1275
    },
    {
      "epoch": 0.8499133321716481,
      "lm_loss": 0.0143,
      "step": 11584,
      "vm_loss": 0.1521
    },
    {
      "epoch": 0.8499133321716481,
      "lm_loss": 0.3865,
      "step": 11584,
      "vm_loss": 0.1165
    },
    {
      "epoch": 0.8499867017617873,
      "grad_norm": 1.6234508124942668,
      "learning_rate": 5.783395324520219e-07,
      "loss": 0.6866,
      "step": 11585
    },
    {
      "epoch": 0.8500600713519264,
      "grad_norm": 1.5129615087532222,
      "learning_rate": 5.777849376178207e-07,
      "loss": 0.6747,
      "step": 11586
    },
    {
      "epoch": 0.8501334409420656,
      "grad_norm": 1.429206056159223,
      "learning_rate": 5.772305925170568e-07,
      "loss": 0.6836,
      "step": 11587
    },
    {
      "epoch": 0.8502068105322047,
      "grad_norm": 1.5458157700859274,
      "learning_rate": 5.76676497181034e-07,
      "loss": 0.7601,
      "step": 11588
    },
    {
      "epoch": 0.8502801801223437,
      "grad_norm": 1.477208135348362,
      "learning_rate": 5.761226516410434e-07,
      "loss": 0.6867,
      "step": 11589
    },
    {
      "epoch": 0.8503535497124829,
      "grad_norm": 0.5391589876818191,
      "learning_rate": 5.755690559283627e-07,
      "loss": 0.4728,
      "step": 11590
    },
    {
      "epoch": 0.850426919302622,
      "grad_norm": 1.3476495860780116,
      "learning_rate": 5.750157100742543e-07,
      "loss": 0.7218,
      "step": 11591
    },
    {
      "epoch": 0.8505002888927612,
      "grad_norm": 1.5278927094484125,
      "learning_rate": 5.744626141099657e-07,
      "loss": 0.7303,
      "step": 11592
    },
    {
      "epoch": 0.8505002888927612,
      "lm_loss": 1.0182,
      "step": 11592,
      "vm_loss": 0.0581
    },
    {
      "epoch": 0.8505002888927612,
      "lm_loss": 0.7565,
      "step": 11592,
      "vm_loss": 0.1772
    },
    {
      "epoch": 0.8505002888927612,
      "lm_loss": 0.2463,
      "step": 11592,
      "vm_loss": 0.0613
    },
    {
      "epoch": 0.8505002888927612,
      "lm_loss": 0.7496,
      "step": 11592,
      "vm_loss": 0.1452
    },
    {
      "epoch": 0.8505002888927612,
      "lm_loss": 0.8006,
      "step": 11592,
      "vm_loss": 0.1603
    },
    {
      "epoch": 0.8505002888927612,
      "lm_loss": 0.7216,
      "step": 11592,
      "vm_loss": 0.0854
    },
    {
      "epoch": 0.8505002888927612,
      "lm_loss": 0.5901,
      "step": 11592,
      "vm_loss": 0.1396
    },
    {
      "epoch": 0.8505002888927612,
      "lm_loss": 0.9752,
      "step": 11592,
      "vm_loss": 0.1153
    },
    {
      "epoch": 0.8505736584829003,
      "grad_norm": 1.5391873812410866,
      "learning_rate": 5.739097680667338e-07,
      "loss": 0.7316,
      "step": 11593
    },
    {
      "epoch": 0.8506470280730394,
      "grad_norm": 1.7410415986517584,
      "learning_rate": 5.733571719757785e-07,
      "loss": 0.6688,
      "step": 11594
    },
    {
      "epoch": 0.8507203976631785,
      "grad_norm": 1.5875116682598176,
      "learning_rate": 5.728048258683062e-07,
      "loss": 0.84,
      "step": 11595
    },
    {
      "epoch": 0.8507937672533177,
      "grad_norm": 1.3562032499981802,
      "learning_rate": 5.722527297755092e-07,
      "loss": 0.695,
      "step": 11596
    },
    {
      "epoch": 0.8508671368434568,
      "grad_norm": 1.4913214082527764,
      "learning_rate": 5.717008837285643e-07,
      "loss": 0.7439,
      "step": 11597
    },
    {
      "epoch": 0.850940506433596,
      "grad_norm": 1.7625260698557843,
      "learning_rate": 5.711492877586383e-07,
      "loss": 0.7625,
      "step": 11598
    },
    {
      "epoch": 0.851013876023735,
      "grad_norm": 1.5544053722655797,
      "learning_rate": 5.705979418968799e-07,
      "loss": 0.7248,
      "step": 11599
    },
    {
      "epoch": 0.8510872456138742,
      "grad_norm": 0.5483088714925055,
      "learning_rate": 5.700468461744252e-07,
      "loss": 0.4891,
      "step": 11600
    },
    {
      "epoch": 0.8510872456138742,
      "lm_loss": 1.1122,
      "step": 11600,
      "vm_loss": 0.0948
    },
    {
      "epoch": 0.8510872456138742,
      "lm_loss": 0.6084,
      "step": 11600,
      "vm_loss": 0.1168
    },
    {
      "epoch": 0.8510872456138742,
      "lm_loss": 0.7555,
      "step": 11600,
      "vm_loss": 0.1238
    },
    {
      "epoch": 0.8510872456138742,
      "lm_loss": 0.3921,
      "step": 11600,
      "vm_loss": 0.1996
    },
    {
      "epoch": 0.8510872456138742,
      "lm_loss": 0.0956,
      "step": 11600,
      "vm_loss": 0.0656
    },
    {
      "epoch": 0.8510872456138742,
      "lm_loss": 0.1882,
      "step": 11600,
      "vm_loss": 0.0825
    },
    {
      "epoch": 0.8510872456138742,
      "lm_loss": 0.1846,
      "step": 11600,
      "vm_loss": 0.0895
    },
    {
      "epoch": 0.8510872456138742,
      "lm_loss": 0.0468,
      "step": 11600,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.8511606152040133,
      "grad_norm": 1.4618626300727249,
      "learning_rate": 5.694960006223954e-07,
      "loss": 0.7187,
      "step": 11601
    },
    {
      "epoch": 0.8512339847941525,
      "grad_norm": 1.417389217801499,
      "learning_rate": 5.689454052718985e-07,
      "loss": 0.6847,
      "step": 11602
    },
    {
      "epoch": 0.8513073543842916,
      "grad_norm": 0.5460300463175844,
      "learning_rate": 5.683950601540272e-07,
      "loss": 0.4715,
      "step": 11603
    },
    {
      "epoch": 0.8513807239744307,
      "grad_norm": 1.5672034336925413,
      "learning_rate": 5.678449652998624e-07,
      "loss": 0.6599,
      "step": 11604
    },
    {
      "epoch": 0.8514540935645698,
      "grad_norm": 1.5969177396814453,
      "learning_rate": 5.672951207404681e-07,
      "loss": 0.8125,
      "step": 11605
    },
    {
      "epoch": 0.851527463154709,
      "grad_norm": 1.6814914783378956,
      "learning_rate": 5.667455265068961e-07,
      "loss": 0.6802,
      "step": 11606
    },
    {
      "epoch": 0.8516008327448481,
      "grad_norm": 0.5623839134784349,
      "learning_rate": 5.661961826301815e-07,
      "loss": 0.493,
      "step": 11607
    },
    {
      "epoch": 0.8516742023349873,
      "grad_norm": 1.674674705350492,
      "learning_rate": 5.656470891413495e-07,
      "loss": 0.6489,
      "step": 11608
    },
    {
      "epoch": 0.8516742023349873,
      "lm_loss": 1.0393,
      "step": 11608,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.8516742023349873,
      "lm_loss": 0.8118,
      "step": 11608,
      "vm_loss": 0.0556
    },
    {
      "epoch": 0.8516742023349873,
      "lm_loss": 1.0914,
      "step": 11608,
      "vm_loss": 0.1609
    },
    {
      "epoch": 0.8516742023349873,
      "lm_loss": 1.0843,
      "step": 11608,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.8516742023349873,
      "lm_loss": 0.0032,
      "step": 11608,
      "vm_loss": 0.0381
    },
    {
      "epoch": 0.8516742023349873,
      "lm_loss": 0.237,
      "step": 11608,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.8516742023349873,
      "lm_loss": 0.1198,
      "step": 11608,
      "vm_loss": 0.0886
    },
    {
      "epoch": 0.8516742023349873,
      "lm_loss": 0.1098,
      "step": 11608,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.8517475719251263,
      "grad_norm": 1.4478859030953342,
      "learning_rate": 5.650982460714083e-07,
      "loss": 0.7061,
      "step": 11609
    },
    {
      "epoch": 0.8518209415152654,
      "grad_norm": 1.4784014779824535,
      "learning_rate": 5.645496534513523e-07,
      "loss": 0.6485,
      "step": 11610
    },
    {
      "epoch": 0.8518943111054046,
      "grad_norm": 0.5489765714626152,
      "learning_rate": 5.640013113121607e-07,
      "loss": 0.4639,
      "step": 11611
    },
    {
      "epoch": 0.8519676806955437,
      "grad_norm": 1.5818008242697066,
      "learning_rate": 5.634532196848003e-07,
      "loss": 0.7131,
      "step": 11612
    },
    {
      "epoch": 0.8520410502856829,
      "grad_norm": 1.5874267153343404,
      "learning_rate": 5.629053786002242e-07,
      "loss": 0.7974,
      "step": 11613
    },
    {
      "epoch": 0.8521144198758219,
      "grad_norm": 1.482749720705279,
      "learning_rate": 5.623577880893699e-07,
      "loss": 0.7177,
      "step": 11614
    },
    {
      "epoch": 0.8521877894659611,
      "grad_norm": 1.4674967611082996,
      "learning_rate": 5.618104481831605e-07,
      "loss": 0.7122,
      "step": 11615
    },
    {
      "epoch": 0.8522611590561002,
      "grad_norm": 1.5287173709968263,
      "learning_rate": 5.612633589125066e-07,
      "loss": 0.682,
      "step": 11616
    },
    {
      "epoch": 0.8522611590561002,
      "lm_loss": 0.3582,
      "step": 11616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8522611590561002,
      "lm_loss": 0.4753,
      "step": 11616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8522611590561002,
      "lm_loss": 0.4722,
      "step": 11616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8522611590561002,
      "lm_loss": 0.4897,
      "step": 11616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8522611590561002,
      "lm_loss": 0.3318,
      "step": 11616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8522611590561002,
      "lm_loss": 0.5106,
      "step": 11616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8522611590561002,
      "lm_loss": 0.6061,
      "step": 11616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8522611590561002,
      "lm_loss": 0.5844,
      "step": 11616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8523345286462394,
      "grad_norm": 0.5625976838229996,
      "learning_rate": 5.607165203083026e-07,
      "loss": 0.4947,
      "step": 11617
    },
    {
      "epoch": 0.8524078982363785,
      "grad_norm": 1.378577095486922,
      "learning_rate": 5.601699324014298e-07,
      "loss": 0.7153,
      "step": 11618
    },
    {
      "epoch": 0.8524812678265176,
      "grad_norm": 1.6429818096888122,
      "learning_rate": 5.596235952227569e-07,
      "loss": 0.7814,
      "step": 11619
    },
    {
      "epoch": 0.8525546374166567,
      "grad_norm": 1.4301815128563853,
      "learning_rate": 5.590775088031358e-07,
      "loss": 0.7057,
      "step": 11620
    },
    {
      "epoch": 0.8526280070067959,
      "grad_norm": 0.5498193794105571,
      "learning_rate": 5.585316731734058e-07,
      "loss": 0.4711,
      "step": 11621
    },
    {
      "epoch": 0.852701376596935,
      "grad_norm": 0.550667482440183,
      "learning_rate": 5.579860883643912e-07,
      "loss": 0.4861,
      "step": 11622
    },
    {
      "epoch": 0.8527747461870742,
      "grad_norm": 1.7068149090694298,
      "learning_rate": 5.574407544069016e-07,
      "loss": 0.7831,
      "step": 11623
    },
    {
      "epoch": 0.8528481157772132,
      "grad_norm": 0.5425890468566933,
      "learning_rate": 5.568956713317352e-07,
      "loss": 0.4572,
      "step": 11624
    },
    {
      "epoch": 0.8528481157772132,
      "lm_loss": 1.2626,
      "step": 11624,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.8528481157772132,
      "lm_loss": 1.6122,
      "step": 11624,
      "vm_loss": 0.0998
    },
    {
      "epoch": 0.8528481157772132,
      "lm_loss": 0.1258,
      "step": 11624,
      "vm_loss": 0.0756
    },
    {
      "epoch": 0.8528481157772132,
      "lm_loss": 0.2429,
      "step": 11624,
      "vm_loss": 0.1915
    },
    {
      "epoch": 0.8528481157772132,
      "lm_loss": 0.9937,
      "step": 11624,
      "vm_loss": 0.0751
    },
    {
      "epoch": 0.8528481157772132,
      "lm_loss": 0.5358,
      "step": 11624,
      "vm_loss": 0.2104
    },
    {
      "epoch": 0.8528481157772132,
      "lm_loss": 0.316,
      "step": 11624,
      "vm_loss": 0.067
    },
    {
      "epoch": 0.8528481157772132,
      "lm_loss": 0.8747,
      "step": 11624,
      "vm_loss": 0.1093
    },
    {
      "epoch": 0.8529214853673523,
      "grad_norm": 1.6081607979356514,
      "learning_rate": 5.563508391696742e-07,
      "loss": 0.7598,
      "step": 11625
    },
    {
      "epoch": 0.8529948549574915,
      "grad_norm": 1.4619968613956573,
      "learning_rate": 5.55806257951485e-07,
      "loss": 0.7181,
      "step": 11626
    },
    {
      "epoch": 0.8530682245476306,
      "grad_norm": 1.3841130964354575,
      "learning_rate": 5.55261927707923e-07,
      "loss": 0.6725,
      "step": 11627
    },
    {
      "epoch": 0.8531415941377698,
      "grad_norm": 1.4501869618111984,
      "learning_rate": 5.547178484697252e-07,
      "loss": 0.7099,
      "step": 11628
    },
    {
      "epoch": 0.8532149637279088,
      "grad_norm": 1.3830753006268226,
      "learning_rate": 5.54174020267621e-07,
      "loss": 0.603,
      "step": 11629
    },
    {
      "epoch": 0.853288333318048,
      "grad_norm": 1.4679236228963144,
      "learning_rate": 5.536304431323191e-07,
      "loss": 0.7274,
      "step": 11630
    },
    {
      "epoch": 0.8533617029081871,
      "grad_norm": 1.4779508684420606,
      "learning_rate": 5.530871170945179e-07,
      "loss": 0.7264,
      "step": 11631
    },
    {
      "epoch": 0.8534350724983263,
      "grad_norm": 0.5260006537594831,
      "learning_rate": 5.525440421848993e-07,
      "loss": 0.4708,
      "step": 11632
    },
    {
      "epoch": 0.8534350724983263,
      "lm_loss": 0.5898,
      "step": 11632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8534350724983263,
      "lm_loss": 0.4509,
      "step": 11632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8534350724983263,
      "lm_loss": 0.3311,
      "step": 11632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8534350724983263,
      "lm_loss": 0.5138,
      "step": 11632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8534350724983263,
      "lm_loss": 0.4535,
      "step": 11632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8534350724983263,
      "lm_loss": 0.5049,
      "step": 11632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8534350724983263,
      "lm_loss": 0.4852,
      "step": 11632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8534350724983263,
      "lm_loss": 0.4072,
      "step": 11632,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8535084420884654,
      "grad_norm": 0.5403260035074764,
      "learning_rate": 5.520012184341328e-07,
      "loss": 0.4763,
      "step": 11633
    },
    {
      "epoch": 0.8535818116786045,
      "grad_norm": 0.5425184550131927,
      "learning_rate": 5.514586458728715e-07,
      "loss": 0.4844,
      "step": 11634
    },
    {
      "epoch": 0.8536551812687436,
      "grad_norm": 0.5515211837060457,
      "learning_rate": 5.509163245317584e-07,
      "loss": 0.4584,
      "step": 11635
    },
    {
      "epoch": 0.8537285508588828,
      "grad_norm": 0.5521805109058494,
      "learning_rate": 5.503742544414187e-07,
      "loss": 0.4775,
      "step": 11636
    },
    {
      "epoch": 0.8538019204490219,
      "grad_norm": 1.4893716029512325,
      "learning_rate": 5.498324356324636e-07,
      "loss": 0.7575,
      "step": 11637
    },
    {
      "epoch": 0.8538752900391611,
      "grad_norm": 1.3259306833396676,
      "learning_rate": 5.492908681354919e-07,
      "loss": 0.6144,
      "step": 11638
    },
    {
      "epoch": 0.8539486596293001,
      "grad_norm": 0.5372416869198152,
      "learning_rate": 5.487495519810853e-07,
      "loss": 0.4628,
      "step": 11639
    },
    {
      "epoch": 0.8540220292194393,
      "grad_norm": 0.5589038013626357,
      "learning_rate": 5.482084871998166e-07,
      "loss": 0.4564,
      "step": 11640
    },
    {
      "epoch": 0.8540220292194393,
      "lm_loss": 0.2051,
      "step": 11640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8540220292194393,
      "lm_loss": 0.3919,
      "step": 11640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8540220292194393,
      "lm_loss": 0.4665,
      "step": 11640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8540220292194393,
      "lm_loss": 0.4805,
      "step": 11640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8540220292194393,
      "lm_loss": 0.3952,
      "step": 11640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8540220292194393,
      "lm_loss": 0.4816,
      "step": 11640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8540220292194393,
      "lm_loss": 0.409,
      "step": 11640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8540220292194393,
      "lm_loss": 0.6292,
      "step": 11640,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8540953988095784,
      "grad_norm": 0.5374022033926092,
      "learning_rate": 5.476676738222386e-07,
      "loss": 0.4611,
      "step": 11641
    },
    {
      "epoch": 0.8541687683997176,
      "grad_norm": 1.5108889303278752,
      "learning_rate": 5.471271118788935e-07,
      "loss": 0.6959,
      "step": 11642
    },
    {
      "epoch": 0.8542421379898567,
      "grad_norm": 1.7403485581686515,
      "learning_rate": 5.465868014003078e-07,
      "loss": 0.7191,
      "step": 11643
    },
    {
      "epoch": 0.8543155075799957,
      "grad_norm": 0.5508784716725073,
      "learning_rate": 5.460467424169941e-07,
      "loss": 0.4914,
      "step": 11644
    },
    {
      "epoch": 0.8543888771701349,
      "grad_norm": 1.6142606918242988,
      "learning_rate": 5.455069349594494e-07,
      "loss": 0.6383,
      "step": 11645
    },
    {
      "epoch": 0.854462246760274,
      "grad_norm": 0.539069726089898,
      "learning_rate": 5.449673790581611e-07,
      "loss": 0.4691,
      "step": 11646
    },
    {
      "epoch": 0.8545356163504132,
      "grad_norm": 1.5324911963986454,
      "learning_rate": 5.44428074743597e-07,
      "loss": 0.6107,
      "step": 11647
    },
    {
      "epoch": 0.8546089859405522,
      "grad_norm": 0.5306485377237415,
      "learning_rate": 5.43889022046214e-07,
      "loss": 0.4707,
      "step": 11648
    },
    {
      "epoch": 0.8546089859405522,
      "lm_loss": 1.2171,
      "step": 11648,
      "vm_loss": 0.0761
    },
    {
      "epoch": 0.8546089859405522,
      "lm_loss": 0.01,
      "step": 11648,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.8546089859405522,
      "lm_loss": 0.1056,
      "step": 11648,
      "vm_loss": 0.0831
    },
    {
      "epoch": 0.8546089859405522,
      "lm_loss": 1.1398,
      "step": 11648,
      "vm_loss": 0.1591
    },
    {
      "epoch": 0.8546089859405522,
      "lm_loss": 0.0902,
      "step": 11648,
      "vm_loss": 0.1823
    },
    {
      "epoch": 0.8546089859405522,
      "lm_loss": 1.4628,
      "step": 11648,
      "vm_loss": 0.2182
    },
    {
      "epoch": 0.8546089859405522,
      "lm_loss": 0.9434,
      "step": 11648,
      "vm_loss": 0.2099
    },
    {
      "epoch": 0.8546089859405522,
      "lm_loss": 1.8227,
      "step": 11648,
      "vm_loss": 0.2333
    },
    {
      "epoch": 0.8546823555306914,
      "grad_norm": 1.5549417610780516,
      "learning_rate": 5.433502209964531e-07,
      "loss": 0.7974,
      "step": 11649
    },
    {
      "epoch": 0.8547557251208305,
      "grad_norm": 1.431720948221461,
      "learning_rate": 5.428116716247411e-07,
      "loss": 0.7104,
      "step": 11650
    },
    {
      "epoch": 0.8548290947109697,
      "grad_norm": 1.3851193899187544,
      "learning_rate": 5.422733739614927e-07,
      "loss": 0.7207,
      "step": 11651
    },
    {
      "epoch": 0.8549024643011088,
      "grad_norm": 1.4318803334771497,
      "learning_rate": 5.417353280371068e-07,
      "loss": 0.8192,
      "step": 11652
    },
    {
      "epoch": 0.8549758338912479,
      "grad_norm": 1.4287955790640052,
      "learning_rate": 5.411975338819675e-07,
      "loss": 0.8293,
      "step": 11653
    },
    {
      "epoch": 0.855049203481387,
      "grad_norm": 1.6659135885726672,
      "learning_rate": 5.406599915264448e-07,
      "loss": 0.6722,
      "step": 11654
    },
    {
      "epoch": 0.8551225730715262,
      "grad_norm": 0.5453771924723986,
      "learning_rate": 5.401227010008958e-07,
      "loss": 0.4682,
      "step": 11655
    },
    {
      "epoch": 0.8551959426616653,
      "grad_norm": 0.5487168638060885,
      "learning_rate": 5.395856623356627e-07,
      "loss": 0.4834,
      "step": 11656
    },
    {
      "epoch": 0.8551959426616653,
      "lm_loss": 1.0783,
      "step": 11656,
      "vm_loss": 0.027
    },
    {
      "epoch": 0.8551959426616653,
      "lm_loss": 1.0273,
      "step": 11656,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.8551959426616653,
      "lm_loss": 0.9132,
      "step": 11656,
      "vm_loss": 0.1084
    },
    {
      "epoch": 0.8551959426616653,
      "lm_loss": 0.7134,
      "step": 11656,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.8551959426616653,
      "lm_loss": 0.5634,
      "step": 11656,
      "vm_loss": 0.152
    },
    {
      "epoch": 0.8551959426616653,
      "lm_loss": 0.0478,
      "step": 11656,
      "vm_loss": 0.0707
    },
    {
      "epoch": 0.8551959426616653,
      "lm_loss": 0.0987,
      "step": 11656,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.8551959426616653,
      "lm_loss": 0.8092,
      "step": 11656,
      "vm_loss": 0.2005
    },
    {
      "epoch": 0.8552693122518045,
      "grad_norm": 1.4942823221083146,
      "learning_rate": 5.39048875561074e-07,
      "loss": 0.6821,
      "step": 11657
    },
    {
      "epoch": 0.8553426818419435,
      "grad_norm": 0.5563739615433382,
      "learning_rate": 5.385123407074422e-07,
      "loss": 0.4887,
      "step": 11658
    },
    {
      "epoch": 0.8554160514320827,
      "grad_norm": 1.5695016957243604,
      "learning_rate": 5.379760578050669e-07,
      "loss": 0.6412,
      "step": 11659
    },
    {
      "epoch": 0.8554894210222218,
      "grad_norm": 1.4458064317449284,
      "learning_rate": 5.374400268842345e-07,
      "loss": 0.7774,
      "step": 11660
    },
    {
      "epoch": 0.855562790612361,
      "grad_norm": 1.503941761794528,
      "learning_rate": 5.369042479752134e-07,
      "loss": 0.7645,
      "step": 11661
    },
    {
      "epoch": 0.8556361602025001,
      "grad_norm": 1.5627095571373009,
      "learning_rate": 5.363687211082635e-07,
      "loss": 0.7203,
      "step": 11662
    },
    {
      "epoch": 0.8557095297926391,
      "grad_norm": 1.649155947968017,
      "learning_rate": 5.358334463136255e-07,
      "loss": 0.6807,
      "step": 11663
    },
    {
      "epoch": 0.8557828993827783,
      "grad_norm": 1.5416834524631327,
      "learning_rate": 5.352984236215286e-07,
      "loss": 0.7282,
      "step": 11664
    },
    {
      "epoch": 0.8557828993827783,
      "lm_loss": 1.1415,
      "step": 11664,
      "vm_loss": 0.0369
    },
    {
      "epoch": 0.8557828993827783,
      "lm_loss": 1.2164,
      "step": 11664,
      "vm_loss": 0.1236
    },
    {
      "epoch": 0.8557828993827783,
      "lm_loss": 0.2182,
      "step": 11664,
      "vm_loss": 0.1625
    },
    {
      "epoch": 0.8557828993827783,
      "lm_loss": 0.7267,
      "step": 11664,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.8557828993827783,
      "lm_loss": 0.4369,
      "step": 11664,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.8557828993827783,
      "lm_loss": 0.7912,
      "step": 11664,
      "vm_loss": 0.2352
    },
    {
      "epoch": 0.8557828993827783,
      "lm_loss": 0.3024,
      "step": 11664,
      "vm_loss": 0.1036
    },
    {
      "epoch": 0.8557828993827783,
      "lm_loss": 0.184,
      "step": 11664,
      "vm_loss": 0.1286
    },
    {
      "epoch": 0.8558562689729174,
      "grad_norm": 1.435975607145077,
      "learning_rate": 5.347636530621858e-07,
      "loss": 0.8537,
      "step": 11665
    },
    {
      "epoch": 0.8559296385630566,
      "grad_norm": 1.580628791961724,
      "learning_rate": 5.342291346657968e-07,
      "loss": 0.8214,
      "step": 11666
    },
    {
      "epoch": 0.8560030081531957,
      "grad_norm": 1.5871199231247932,
      "learning_rate": 5.336948684625492e-07,
      "loss": 0.7688,
      "step": 11667
    },
    {
      "epoch": 0.8560763777433348,
      "grad_norm": 1.496585195797711,
      "learning_rate": 5.331608544826122e-07,
      "loss": 0.6353,
      "step": 11668
    },
    {
      "epoch": 0.8561497473334739,
      "grad_norm": 1.7373626113088942,
      "learning_rate": 5.326270927561444e-07,
      "loss": 0.7028,
      "step": 11669
    },
    {
      "epoch": 0.8562231169236131,
      "grad_norm": 1.3966964732301053,
      "learning_rate": 5.320935833132879e-07,
      "loss": 0.7066,
      "step": 11670
    },
    {
      "epoch": 0.8562964865137522,
      "grad_norm": 1.376116555909956,
      "learning_rate": 5.315603261841707e-07,
      "loss": 0.6849,
      "step": 11671
    },
    {
      "epoch": 0.8563698561038914,
      "grad_norm": 1.5127670764006278,
      "learning_rate": 5.310273213989076e-07,
      "loss": 0.6304,
      "step": 11672
    },
    {
      "epoch": 0.8563698561038914,
      "lm_loss": 0.3965,
      "step": 11672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8563698561038914,
      "lm_loss": 0.5118,
      "step": 11672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8563698561038914,
      "lm_loss": 0.4505,
      "step": 11672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8563698561038914,
      "lm_loss": 0.4411,
      "step": 11672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8563698561038914,
      "lm_loss": 0.4918,
      "step": 11672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8563698561038914,
      "lm_loss": 0.4387,
      "step": 11672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8563698561038914,
      "lm_loss": 0.4567,
      "step": 11672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8563698561038914,
      "lm_loss": 0.2394,
      "step": 11672,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8564432256940304,
      "grad_norm": 0.5425662174366022,
      "learning_rate": 5.304945689875989e-07,
      "loss": 0.4491,
      "step": 11673
    },
    {
      "epoch": 0.8565165952841696,
      "grad_norm": 1.524669699400642,
      "learning_rate": 5.299620689803314e-07,
      "loss": 0.6625,
      "step": 11674
    },
    {
      "epoch": 0.8565899648743087,
      "grad_norm": 1.636511846106637,
      "learning_rate": 5.294298214071752e-07,
      "loss": 0.7782,
      "step": 11675
    },
    {
      "epoch": 0.8566633344644479,
      "grad_norm": 1.5228080878179489,
      "learning_rate": 5.288978262981886e-07,
      "loss": 0.7197,
      "step": 11676
    },
    {
      "epoch": 0.856736704054587,
      "grad_norm": 1.5106535456143162,
      "learning_rate": 5.283660836834125e-07,
      "loss": 0.7134,
      "step": 11677
    },
    {
      "epoch": 0.8568100736447261,
      "grad_norm": 1.4686170976980013,
      "learning_rate": 5.278345935928791e-07,
      "loss": 0.7313,
      "step": 11678
    },
    {
      "epoch": 0.8568834432348652,
      "grad_norm": 1.5405901168346114,
      "learning_rate": 5.273033560566015e-07,
      "loss": 0.6468,
      "step": 11679
    },
    {
      "epoch": 0.8569568128250044,
      "grad_norm": 1.6170102379320217,
      "learning_rate": 5.267723711045792e-07,
      "loss": 0.6781,
      "step": 11680
    },
    {
      "epoch": 0.8569568128250044,
      "lm_loss": 0.418,
      "step": 11680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8569568128250044,
      "lm_loss": 0.4249,
      "step": 11680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8569568128250044,
      "lm_loss": 0.3787,
      "step": 11680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8569568128250044,
      "lm_loss": 0.5651,
      "step": 11680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8569568128250044,
      "lm_loss": 0.4543,
      "step": 11680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8569568128250044,
      "lm_loss": 0.5242,
      "step": 11680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8569568128250044,
      "lm_loss": 0.3867,
      "step": 11680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8569568128250044,
      "lm_loss": 0.3864,
      "step": 11680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8570301824151435,
      "grad_norm": 0.5609842026612679,
      "learning_rate": 5.26241638766799e-07,
      "loss": 0.4623,
      "step": 11681
    },
    {
      "epoch": 0.8571035520052827,
      "grad_norm": 0.551825098349909,
      "learning_rate": 5.257111590732316e-07,
      "loss": 0.4761,
      "step": 11682
    },
    {
      "epoch": 0.8571769215954217,
      "grad_norm": 1.603627924130668,
      "learning_rate": 5.25180932053837e-07,
      "loss": 0.7836,
      "step": 11683
    },
    {
      "epoch": 0.8572502911855608,
      "grad_norm": 1.359862711029353,
      "learning_rate": 5.246509577385561e-07,
      "loss": 0.6477,
      "step": 11684
    },
    {
      "epoch": 0.8573236607757,
      "grad_norm": 0.5318951935202997,
      "learning_rate": 5.24121236157319e-07,
      "loss": 0.4676,
      "step": 11685
    },
    {
      "epoch": 0.8573970303658391,
      "grad_norm": 0.5396267935944705,
      "learning_rate": 5.235917673400398e-07,
      "loss": 0.4718,
      "step": 11686
    },
    {
      "epoch": 0.8574703999559783,
      "grad_norm": 0.5422574039658424,
      "learning_rate": 5.230625513166193e-07,
      "loss": 0.4796,
      "step": 11687
    },
    {
      "epoch": 0.8575437695461173,
      "grad_norm": 1.484826644462086,
      "learning_rate": 5.225335881169425e-07,
      "loss": 0.7734,
      "step": 11688
    },
    {
      "epoch": 0.8575437695461173,
      "lm_loss": 0.5547,
      "step": 11688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8575437695461173,
      "lm_loss": 0.6671,
      "step": 11688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8575437695461173,
      "lm_loss": 0.3111,
      "step": 11688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8575437695461173,
      "lm_loss": 0.4876,
      "step": 11688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8575437695461173,
      "lm_loss": 0.3898,
      "step": 11688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8575437695461173,
      "lm_loss": 0.635,
      "step": 11688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8575437695461173,
      "lm_loss": 0.4844,
      "step": 11688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8575437695461173,
      "lm_loss": 0.4288,
      "step": 11688,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8576171391362565,
      "grad_norm": 0.5573405760415969,
      "learning_rate": 5.22004877770883e-07,
      "loss": 0.4809,
      "step": 11689
    },
    {
      "epoch": 0.8576905087263956,
      "grad_norm": 1.4940270272997427,
      "learning_rate": 5.214764203082984e-07,
      "loss": 0.7117,
      "step": 11690
    },
    {
      "epoch": 0.8577638783165348,
      "grad_norm": 1.554385211380578,
      "learning_rate": 5.209482157590301e-07,
      "loss": 0.76,
      "step": 11691
    },
    {
      "epoch": 0.8578372479066739,
      "grad_norm": 1.6396983209015723,
      "learning_rate": 5.204202641529088e-07,
      "loss": 0.7653,
      "step": 11692
    },
    {
      "epoch": 0.857910617496813,
      "grad_norm": 1.553345214223332,
      "learning_rate": 5.198925655197479e-07,
      "loss": 0.6825,
      "step": 11693
    },
    {
      "epoch": 0.8579839870869521,
      "grad_norm": 1.5197672426817568,
      "learning_rate": 5.193651198893496e-07,
      "loss": 0.7592,
      "step": 11694
    },
    {
      "epoch": 0.8580573566770913,
      "grad_norm": 1.4433042969479846,
      "learning_rate": 5.188379272914984e-07,
      "loss": 0.6595,
      "step": 11695
    },
    {
      "epoch": 0.8581307262672304,
      "grad_norm": 1.6444321124211123,
      "learning_rate": 5.183109877559678e-07,
      "loss": 0.7716,
      "step": 11696
    },
    {
      "epoch": 0.8581307262672304,
      "lm_loss": 0.9703,
      "step": 11696,
      "vm_loss": 0.162
    },
    {
      "epoch": 0.8581307262672304,
      "lm_loss": 0.2002,
      "step": 11696,
      "vm_loss": 0.0952
    },
    {
      "epoch": 0.8581307262672304,
      "lm_loss": 0.3224,
      "step": 11696,
      "vm_loss": 0.2681
    },
    {
      "epoch": 0.8581307262672304,
      "lm_loss": 0.8249,
      "step": 11696,
      "vm_loss": 0.177
    },
    {
      "epoch": 0.8581307262672304,
      "lm_loss": 0.6903,
      "step": 11696,
      "vm_loss": 0.1171
    },
    {
      "epoch": 0.8581307262672304,
      "lm_loss": 0.2738,
      "step": 11696,
      "vm_loss": 0.1132
    },
    {
      "epoch": 0.8581307262672304,
      "lm_loss": 0.8305,
      "step": 11696,
      "vm_loss": 0.1945
    },
    {
      "epoch": 0.8581307262672304,
      "lm_loss": 0.1313,
      "step": 11696,
      "vm_loss": 0.1557
    },
    {
      "epoch": 0.8582040958573696,
      "grad_norm": 1.6423083252353226,
      "learning_rate": 5.177843013125139e-07,
      "loss": 0.7242,
      "step": 11697
    },
    {
      "epoch": 0.8582774654475086,
      "grad_norm": 1.5328465212465279,
      "learning_rate": 5.172578679908802e-07,
      "loss": 0.7613,
      "step": 11698
    },
    {
      "epoch": 0.8583508350376478,
      "grad_norm": 1.6376439588480234,
      "learning_rate": 5.167316878207956e-07,
      "loss": 0.8295,
      "step": 11699
    },
    {
      "epoch": 0.8584242046277869,
      "grad_norm": 1.558188259141495,
      "learning_rate": 5.162057608319759e-07,
      "loss": 0.8474,
      "step": 11700
    },
    {
      "epoch": 0.858497574217926,
      "grad_norm": 1.6321654705830604,
      "learning_rate": 5.156800870541212e-07,
      "loss": 0.6298,
      "step": 11701
    },
    {
      "epoch": 0.8585709438080652,
      "grad_norm": 0.5602534432363683,
      "learning_rate": 5.151546665169166e-07,
      "loss": 0.4747,
      "step": 11702
    },
    {
      "epoch": 0.8586443133982042,
      "grad_norm": 0.5520622781584169,
      "learning_rate": 5.14629499250035e-07,
      "loss": 0.4665,
      "step": 11703
    },
    {
      "epoch": 0.8587176829883434,
      "grad_norm": 1.5273107098772083,
      "learning_rate": 5.141045852831322e-07,
      "loss": 0.7092,
      "step": 11704
    },
    {
      "epoch": 0.8587176829883434,
      "lm_loss": 0.7419,
      "step": 11704,
      "vm_loss": 0.0404
    },
    {
      "epoch": 0.8587176829883434,
      "lm_loss": 1.4044,
      "step": 11704,
      "vm_loss": 0.2327
    },
    {
      "epoch": 0.8587176829883434,
      "lm_loss": 1.0025,
      "step": 11704,
      "vm_loss": 0.1425
    },
    {
      "epoch": 0.8587176829883434,
      "lm_loss": 0.4657,
      "step": 11704,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.8587176829883434,
      "lm_loss": 0.7061,
      "step": 11704,
      "vm_loss": 0.1437
    },
    {
      "epoch": 0.8587176829883434,
      "lm_loss": 0.2795,
      "step": 11704,
      "vm_loss": 0.0943
    },
    {
      "epoch": 0.8587176829883434,
      "lm_loss": 0.8064,
      "step": 11704,
      "vm_loss": 0.1641
    },
    {
      "epoch": 0.8587176829883434,
      "lm_loss": 0.711,
      "step": 11704,
      "vm_loss": 0.0861
    },
    {
      "epoch": 0.8587910525784825,
      "grad_norm": 1.6527275579840872,
      "learning_rate": 5.135799246458539e-07,
      "loss": 0.7021,
      "step": 11705
    },
    {
      "epoch": 0.8588644221686217,
      "grad_norm": 1.5959075690823328,
      "learning_rate": 5.130555173678276e-07,
      "loss": 0.7344,
      "step": 11706
    },
    {
      "epoch": 0.8589377917587608,
      "grad_norm": 0.5466324225250067,
      "learning_rate": 5.125313634786678e-07,
      "loss": 0.4727,
      "step": 11707
    },
    {
      "epoch": 0.8590111613488999,
      "grad_norm": 1.5372112281826709,
      "learning_rate": 5.120074630079747e-07,
      "loss": 0.7658,
      "step": 11708
    },
    {
      "epoch": 0.859084530939039,
      "grad_norm": 0.5261479901130457,
      "learning_rate": 5.114838159853336e-07,
      "loss": 0.4624,
      "step": 11709
    },
    {
      "epoch": 0.8591579005291782,
      "grad_norm": 0.5540835587003208,
      "learning_rate": 5.109604224403181e-07,
      "loss": 0.4851,
      "step": 11710
    },
    {
      "epoch": 0.8592312701193173,
      "grad_norm": 0.5659827965494765,
      "learning_rate": 5.104372824024845e-07,
      "loss": 0.4787,
      "step": 11711
    },
    {
      "epoch": 0.8593046397094565,
      "grad_norm": 0.5558640642946843,
      "learning_rate": 5.099143959013758e-07,
      "loss": 0.4961,
      "step": 11712
    },
    {
      "epoch": 0.8593046397094565,
      "lm_loss": 1.8151,
      "step": 11712,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.8593046397094565,
      "lm_loss": 0.0248,
      "step": 11712,
      "vm_loss": 0.0717
    },
    {
      "epoch": 0.8593046397094565,
      "lm_loss": 0.1485,
      "step": 11712,
      "vm_loss": 0.2571
    },
    {
      "epoch": 0.8593046397094565,
      "lm_loss": 1.0114,
      "step": 11712,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.8593046397094565,
      "lm_loss": 0.6994,
      "step": 11712,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.8593046397094565,
      "lm_loss": 0.1412,
      "step": 11712,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.8593046397094565,
      "lm_loss": 0.3178,
      "step": 11712,
      "vm_loss": 0.1629
    },
    {
      "epoch": 0.8593046397094565,
      "lm_loss": 0.4244,
      "step": 11712,
      "vm_loss": 0.1339
    },
    {
      "epoch": 0.8593780092995955,
      "grad_norm": 1.4398998450736977,
      "learning_rate": 5.093917629665207e-07,
      "loss": 0.6616,
      "step": 11713
    },
    {
      "epoch": 0.8594513788897347,
      "grad_norm": 1.5071424899357624,
      "learning_rate": 5.088693836274338e-07,
      "loss": 0.7864,
      "step": 11714
    },
    {
      "epoch": 0.8595247484798738,
      "grad_norm": 1.5528876136696472,
      "learning_rate": 5.083472579136134e-07,
      "loss": 0.756,
      "step": 11715
    },
    {
      "epoch": 0.859598118070013,
      "grad_norm": 1.64641583281396,
      "learning_rate": 5.078253858545485e-07,
      "loss": 0.7233,
      "step": 11716
    },
    {
      "epoch": 0.8596714876601521,
      "grad_norm": 0.5579539010927327,
      "learning_rate": 5.073037674797077e-07,
      "loss": 0.4862,
      "step": 11717
    },
    {
      "epoch": 0.8597448572502912,
      "grad_norm": 1.5392596347620906,
      "learning_rate": 5.067824028185497e-07,
      "loss": 0.6733,
      "step": 11718
    },
    {
      "epoch": 0.8598182268404303,
      "grad_norm": 1.5576888737207701,
      "learning_rate": 5.062612919005166e-07,
      "loss": 0.6728,
      "step": 11719
    },
    {
      "epoch": 0.8598915964305694,
      "grad_norm": 0.5421463002607855,
      "learning_rate": 5.05740434755036e-07,
      "loss": 0.479,
      "step": 11720
    },
    {
      "epoch": 0.8598915964305694,
      "lm_loss": 0.6653,
      "step": 11720,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.8598915964305694,
      "lm_loss": 0.0526,
      "step": 11720,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.8598915964305694,
      "lm_loss": 0.8503,
      "step": 11720,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.8598915964305694,
      "lm_loss": 0.4745,
      "step": 11720,
      "vm_loss": 0.0525
    },
    {
      "epoch": 0.8598915964305694,
      "lm_loss": 0.1621,
      "step": 11720,
      "vm_loss": 0.1502
    },
    {
      "epoch": 0.8598915964305694,
      "lm_loss": 0.7807,
      "step": 11720,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.8598915964305694,
      "lm_loss": 0.6216,
      "step": 11720,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.8598915964305694,
      "lm_loss": 0.1533,
      "step": 11720,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.8599649660207086,
      "grad_norm": 1.4752605339052656,
      "learning_rate": 5.052198314115242e-07,
      "loss": 0.7067,
      "step": 11721
    },
    {
      "epoch": 0.8600383356108477,
      "grad_norm": 1.6211259925918944,
      "learning_rate": 5.046994818993789e-07,
      "loss": 0.7374,
      "step": 11722
    },
    {
      "epoch": 0.8601117052009868,
      "grad_norm": 1.5909889398551424,
      "learning_rate": 5.041793862479865e-07,
      "loss": 0.7723,
      "step": 11723
    },
    {
      "epoch": 0.8601850747911259,
      "grad_norm": 1.4605744545247403,
      "learning_rate": 5.036595444867182e-07,
      "loss": 0.699,
      "step": 11724
    },
    {
      "epoch": 0.8602584443812651,
      "grad_norm": 1.643372231613353,
      "learning_rate": 5.031399566449302e-07,
      "loss": 0.7521,
      "step": 11725
    },
    {
      "epoch": 0.8603318139714042,
      "grad_norm": 1.4679153006686187,
      "learning_rate": 5.026206227519642e-07,
      "loss": 0.7518,
      "step": 11726
    },
    {
      "epoch": 0.8604051835615433,
      "grad_norm": 1.5372134601199392,
      "learning_rate": 5.021015428371506e-07,
      "loss": 0.7034,
      "step": 11727
    },
    {
      "epoch": 0.8604785531516824,
      "grad_norm": 1.522388666845073,
      "learning_rate": 5.01582716929801e-07,
      "loss": 0.7522,
      "step": 11728
    },
    {
      "epoch": 0.8604785531516824,
      "lm_loss": 1.4177,
      "step": 11728,
      "vm_loss": 0.1389
    },
    {
      "epoch": 0.8604785531516824,
      "lm_loss": 0.3507,
      "step": 11728,
      "vm_loss": 0.1335
    },
    {
      "epoch": 0.8604785531516824,
      "lm_loss": 0.1797,
      "step": 11728,
      "vm_loss": 0.2258
    },
    {
      "epoch": 0.8604785531516824,
      "lm_loss": 0.5396,
      "step": 11728,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.8604785531516824,
      "lm_loss": 0.7115,
      "step": 11728,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.8604785531516824,
      "lm_loss": 0.1445,
      "step": 11728,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.8604785531516824,
      "lm_loss": 0.4204,
      "step": 11728,
      "vm_loss": 0.1951
    },
    {
      "epoch": 0.8604785531516824,
      "lm_loss": 0.7092,
      "step": 11728,
      "vm_loss": 0.157
    },
    {
      "epoch": 0.8605519227418216,
      "grad_norm": 1.619640904632228,
      "learning_rate": 5.010641450592158e-07,
      "loss": 0.6706,
      "step": 11729
    },
    {
      "epoch": 0.8606252923319607,
      "grad_norm": 1.385543792385535,
      "learning_rate": 5.005458272546799e-07,
      "loss": 0.6861,
      "step": 11730
    },
    {
      "epoch": 0.8606986619220999,
      "grad_norm": 1.351138585528223,
      "learning_rate": 5.000277635454625e-07,
      "loss": 0.6772,
      "step": 11731
    },
    {
      "epoch": 0.8607720315122389,
      "grad_norm": 1.4944042370379336,
      "learning_rate": 4.995099539608228e-07,
      "loss": 0.7613,
      "step": 11732
    },
    {
      "epoch": 0.8608454011023781,
      "grad_norm": 1.645335136062516,
      "learning_rate": 4.989923985300011e-07,
      "loss": 0.7225,
      "step": 11733
    },
    {
      "epoch": 0.8609187706925172,
      "grad_norm": 1.4453839147824534,
      "learning_rate": 4.98475097282225e-07,
      "loss": 0.7522,
      "step": 11734
    },
    {
      "epoch": 0.8609921402826564,
      "grad_norm": 1.7016935755191878,
      "learning_rate": 4.97958050246708e-07,
      "loss": 0.6596,
      "step": 11735
    },
    {
      "epoch": 0.8610655098727955,
      "grad_norm": 1.282235817184489,
      "learning_rate": 4.974412574526483e-07,
      "loss": 0.723,
      "step": 11736
    },
    {
      "epoch": 0.8610655098727955,
      "lm_loss": 1.2104,
      "step": 11736,
      "vm_loss": 0.034
    },
    {
      "epoch": 0.8610655098727955,
      "lm_loss": 1.2605,
      "step": 11736,
      "vm_loss": 0.1576
    },
    {
      "epoch": 0.8610655098727955,
      "lm_loss": 0.6109,
      "step": 11736,
      "vm_loss": 0.0593
    },
    {
      "epoch": 0.8610655098727955,
      "lm_loss": 0.4594,
      "step": 11736,
      "vm_loss": 0.1012
    },
    {
      "epoch": 0.8610655098727955,
      "lm_loss": 0.4324,
      "step": 11736,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.8610655098727955,
      "lm_loss": 0.3943,
      "step": 11736,
      "vm_loss": 0.1792
    },
    {
      "epoch": 0.8610655098727955,
      "lm_loss": 0.8116,
      "step": 11736,
      "vm_loss": 0.2221
    },
    {
      "epoch": 0.8610655098727955,
      "lm_loss": 0.2043,
      "step": 11736,
      "vm_loss": 0.1228
    },
    {
      "epoch": 0.8611388794629345,
      "grad_norm": 1.6963162837972268,
      "learning_rate": 4.96924718929232e-07,
      "loss": 0.7789,
      "step": 11737
    },
    {
      "epoch": 0.8612122490530737,
      "grad_norm": 1.634577700039696,
      "learning_rate": 4.964084347056287e-07,
      "loss": 0.7991,
      "step": 11738
    },
    {
      "epoch": 0.8612856186432128,
      "grad_norm": 0.5392247524584817,
      "learning_rate": 4.958924048109937e-07,
      "loss": 0.48,
      "step": 11739
    },
    {
      "epoch": 0.861358988233352,
      "grad_norm": 0.5440254397393389,
      "learning_rate": 4.953766292744688e-07,
      "loss": 0.4816,
      "step": 11740
    },
    {
      "epoch": 0.8614323578234911,
      "grad_norm": 0.528914560323641,
      "learning_rate": 4.948611081251814e-07,
      "loss": 0.4633,
      "step": 11741
    },
    {
      "epoch": 0.8615057274136302,
      "grad_norm": 0.5653371771517303,
      "learning_rate": 4.943458413922431e-07,
      "loss": 0.4959,
      "step": 11742
    },
    {
      "epoch": 0.8615790970037693,
      "grad_norm": 1.5378430547175794,
      "learning_rate": 4.938308291047544e-07,
      "loss": 0.7169,
      "step": 11743
    },
    {
      "epoch": 0.8616524665939085,
      "grad_norm": 1.5239690217983075,
      "learning_rate": 4.933160712917978e-07,
      "loss": 0.7243,
      "step": 11744
    },
    {
      "epoch": 0.8616524665939085,
      "lm_loss": 0.7725,
      "step": 11744,
      "vm_loss": 0.1135
    },
    {
      "epoch": 0.8616524665939085,
      "lm_loss": 0.8795,
      "step": 11744,
      "vm_loss": 0.1291
    },
    {
      "epoch": 0.8616524665939085,
      "lm_loss": 0.2973,
      "step": 11744,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.8616524665939085,
      "lm_loss": 0.4631,
      "step": 11744,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.8616524665939085,
      "lm_loss": 1.4317,
      "step": 11744,
      "vm_loss": 0.1949
    },
    {
      "epoch": 0.8616524665939085,
      "lm_loss": 0.8247,
      "step": 11744,
      "vm_loss": 0.0744
    },
    {
      "epoch": 0.8616524665939085,
      "lm_loss": 0.7028,
      "step": 11744,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.8616524665939085,
      "lm_loss": 0.9245,
      "step": 11744,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.8617258361840476,
      "grad_norm": 1.4881300774256072,
      "learning_rate": 4.928015679824433e-07,
      "loss": 0.6663,
      "step": 11745
    },
    {
      "epoch": 0.8617992057741868,
      "grad_norm": 1.4361017129420148,
      "learning_rate": 4.922873192057459e-07,
      "loss": 0.8025,
      "step": 11746
    },
    {
      "epoch": 0.8618725753643258,
      "grad_norm": 1.4337242690933567,
      "learning_rate": 4.917733249907459e-07,
      "loss": 0.7772,
      "step": 11747
    },
    {
      "epoch": 0.861945944954465,
      "grad_norm": 1.4720064474222687,
      "learning_rate": 4.91259585366472e-07,
      "loss": 0.6926,
      "step": 11748
    },
    {
      "epoch": 0.8620193145446041,
      "grad_norm": 0.5461350356207164,
      "learning_rate": 4.907461003619346e-07,
      "loss": 0.474,
      "step": 11749
    },
    {
      "epoch": 0.8620926841347433,
      "grad_norm": 1.5956612083673112,
      "learning_rate": 4.902328700061321e-07,
      "loss": 0.7241,
      "step": 11750
    },
    {
      "epoch": 0.8621660537248824,
      "grad_norm": 1.705216561197778,
      "learning_rate": 4.897198943280473e-07,
      "loss": 0.7793,
      "step": 11751
    },
    {
      "epoch": 0.8622394233150215,
      "grad_norm": 0.5373630126607569,
      "learning_rate": 4.8920717335665e-07,
      "loss": 0.4769,
      "step": 11752
    },
    {
      "epoch": 0.8622394233150215,
      "lm_loss": 1.2556,
      "step": 11752,
      "vm_loss": 0.0549
    },
    {
      "epoch": 0.8622394233150215,
      "lm_loss": 1.3585,
      "step": 11752,
      "vm_loss": 0.097
    },
    {
      "epoch": 0.8622394233150215,
      "lm_loss": 0.6867,
      "step": 11752,
      "vm_loss": 0.1344
    },
    {
      "epoch": 0.8622394233150215,
      "lm_loss": 0.225,
      "step": 11752,
      "vm_loss": 0.2225
    },
    {
      "epoch": 0.8622394233150215,
      "lm_loss": 0.169,
      "step": 11752,
      "vm_loss": 0.1192
    },
    {
      "epoch": 0.8622394233150215,
      "lm_loss": 0.8262,
      "step": 11752,
      "vm_loss": 0.1047
    },
    {
      "epoch": 0.8622394233150215,
      "lm_loss": 0.285,
      "step": 11752,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.8622394233150215,
      "lm_loss": 0.1671,
      "step": 11752,
      "vm_loss": 0.0837
    },
    {
      "epoch": 0.8623127929051606,
      "grad_norm": 1.4730416738076249,
      "learning_rate": 4.886947071208942e-07,
      "loss": 0.7538,
      "step": 11753
    },
    {
      "epoch": 0.8623861624952998,
      "grad_norm": 1.6888086842487917,
      "learning_rate": 4.881824956497205e-07,
      "loss": 0.647,
      "step": 11754
    },
    {
      "epoch": 0.8624595320854389,
      "grad_norm": 1.4594988795053714,
      "learning_rate": 4.876705389720549e-07,
      "loss": 0.7035,
      "step": 11755
    },
    {
      "epoch": 0.8625329016755781,
      "grad_norm": 1.5233602168738667,
      "learning_rate": 4.871588371168079e-07,
      "loss": 0.779,
      "step": 11756
    },
    {
      "epoch": 0.8626062712657171,
      "grad_norm": 0.5434178752565694,
      "learning_rate": 4.866473901128766e-07,
      "loss": 0.4761,
      "step": 11757
    },
    {
      "epoch": 0.8626796408558562,
      "grad_norm": 1.6460265674385852,
      "learning_rate": 4.861361979891455e-07,
      "loss": 0.7855,
      "step": 11758
    },
    {
      "epoch": 0.8627530104459954,
      "grad_norm": 1.6182232663602918,
      "learning_rate": 4.856252607744816e-07,
      "loss": 0.6956,
      "step": 11759
    },
    {
      "epoch": 0.8628263800361345,
      "grad_norm": 1.6065003283814812,
      "learning_rate": 4.851145784977384e-07,
      "loss": 0.7559,
      "step": 11760
    },
    {
      "epoch": 0.8628263800361345,
      "lm_loss": 0.3556,
      "step": 11760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8628263800361345,
      "lm_loss": 0.5006,
      "step": 11760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8628263800361345,
      "lm_loss": 0.5193,
      "step": 11760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8628263800361345,
      "lm_loss": 0.447,
      "step": 11760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8628263800361345,
      "lm_loss": 0.5381,
      "step": 11760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8628263800361345,
      "lm_loss": 0.3887,
      "step": 11760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8628263800361345,
      "lm_loss": 0.5967,
      "step": 11760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8628263800361345,
      "lm_loss": 0.4629,
      "step": 11760,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8628997496262737,
      "grad_norm": 0.5589605190141917,
      "learning_rate": 4.846041511877564e-07,
      "loss": 0.4733,
      "step": 11761
    },
    {
      "epoch": 0.8629731192164127,
      "grad_norm": 1.5940367613830606,
      "learning_rate": 4.84093978873359e-07,
      "loss": 0.7876,
      "step": 11762
    },
    {
      "epoch": 0.8630464888065519,
      "grad_norm": 0.5589208261434927,
      "learning_rate": 4.835840615833592e-07,
      "loss": 0.4816,
      "step": 11763
    },
    {
      "epoch": 0.863119858396691,
      "grad_norm": 1.3308041888028612,
      "learning_rate": 4.83074399346552e-07,
      "loss": 0.745,
      "step": 11764
    },
    {
      "epoch": 0.8631932279868302,
      "grad_norm": 1.2762070759798896,
      "learning_rate": 4.825649921917191e-07,
      "loss": 0.6728,
      "step": 11765
    },
    {
      "epoch": 0.8632665975769693,
      "grad_norm": 1.6090795357416927,
      "learning_rate": 4.820558401476289e-07,
      "loss": 0.7915,
      "step": 11766
    },
    {
      "epoch": 0.8633399671671084,
      "grad_norm": 1.5206895481137912,
      "learning_rate": 4.815469432430337e-07,
      "loss": 0.8067,
      "step": 11767
    },
    {
      "epoch": 0.8634133367572475,
      "grad_norm": 1.6089435019081135,
      "learning_rate": 4.810383015066717e-07,
      "loss": 0.6231,
      "step": 11768
    },
    {
      "epoch": 0.8634133367572475,
      "lm_loss": 0.8483,
      "step": 11768,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.8634133367572475,
      "lm_loss": 0.5044,
      "step": 11768,
      "vm_loss": 0.1023
    },
    {
      "epoch": 0.8634133367572475,
      "lm_loss": 0.7793,
      "step": 11768,
      "vm_loss": 0.0954
    },
    {
      "epoch": 0.8634133367572475,
      "lm_loss": 0.0499,
      "step": 11768,
      "vm_loss": 0.1047
    },
    {
      "epoch": 0.8634133367572475,
      "lm_loss": 0.865,
      "step": 11768,
      "vm_loss": 0.2095
    },
    {
      "epoch": 0.8634133367572475,
      "lm_loss": 0.6393,
      "step": 11768,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.8634133367572475,
      "lm_loss": 0.0931,
      "step": 11768,
      "vm_loss": 0.0467
    },
    {
      "epoch": 0.8634133367572475,
      "lm_loss": 0.797,
      "step": 11768,
      "vm_loss": 0.1026
    },
    {
      "epoch": 0.8634867063473867,
      "grad_norm": 1.518618735293609,
      "learning_rate": 4.805299149672682e-07,
      "loss": 0.7025,
      "step": 11769
    },
    {
      "epoch": 0.8635600759375258,
      "grad_norm": 1.587277801108271,
      "learning_rate": 4.800217836535331e-07,
      "loss": 0.7405,
      "step": 11770
    },
    {
      "epoch": 0.863633445527665,
      "grad_norm": 0.5474014866517448,
      "learning_rate": 4.795139075941613e-07,
      "loss": 0.4627,
      "step": 11771
    },
    {
      "epoch": 0.863706815117804,
      "grad_norm": 0.5326311359976907,
      "learning_rate": 4.790062868178335e-07,
      "loss": 0.4745,
      "step": 11772
    },
    {
      "epoch": 0.8637801847079432,
      "grad_norm": 1.6071591537666032,
      "learning_rate": 4.784989213532165e-07,
      "loss": 0.7201,
      "step": 11773
    },
    {
      "epoch": 0.8638535542980823,
      "grad_norm": 1.534115369174961,
      "learning_rate": 4.779918112289633e-07,
      "loss": 0.6386,
      "step": 11774
    },
    {
      "epoch": 0.8639269238882215,
      "grad_norm": 1.438618381741647,
      "learning_rate": 4.774849564737111e-07,
      "loss": 0.781,
      "step": 11775
    },
    {
      "epoch": 0.8640002934783606,
      "grad_norm": 1.5526989291277074,
      "learning_rate": 4.769783571160835e-07,
      "loss": 0.6552,
      "step": 11776
    },
    {
      "epoch": 0.8640002934783606,
      "lm_loss": 0.548,
      "step": 11776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8640002934783606,
      "lm_loss": 0.3872,
      "step": 11776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8640002934783606,
      "lm_loss": 0.3294,
      "step": 11776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8640002934783606,
      "lm_loss": 0.5033,
      "step": 11776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8640002934783606,
      "lm_loss": 0.5314,
      "step": 11776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8640002934783606,
      "lm_loss": 0.4743,
      "step": 11776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8640002934783606,
      "lm_loss": 0.4928,
      "step": 11776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8640002934783606,
      "lm_loss": 0.5217,
      "step": 11776,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8640736630684996,
      "grad_norm": 0.5484086887636163,
      "learning_rate": 4.7647201318468894e-07,
      "loss": 0.4794,
      "step": 11777
    },
    {
      "epoch": 0.8641470326586388,
      "grad_norm": 1.4675614174395062,
      "learning_rate": 4.759659247081211e-07,
      "loss": 0.7799,
      "step": 11778
    },
    {
      "epoch": 0.8642204022487779,
      "grad_norm": 0.5316911567257824,
      "learning_rate": 4.754600917149621e-07,
      "loss": 0.4662,
      "step": 11779
    },
    {
      "epoch": 0.8642937718389171,
      "grad_norm": 1.8921860171667018,
      "learning_rate": 4.7495451423377635e-07,
      "loss": 0.7899,
      "step": 11780
    },
    {
      "epoch": 0.8643671414290562,
      "grad_norm": 0.563040316506609,
      "learning_rate": 4.74449192293116e-07,
      "loss": 0.4766,
      "step": 11781
    },
    {
      "epoch": 0.8644405110191953,
      "grad_norm": 1.7372547498061612,
      "learning_rate": 4.7394412592151605e-07,
      "loss": 0.7429,
      "step": 11782
    },
    {
      "epoch": 0.8645138806093344,
      "grad_norm": 1.6039753902594769,
      "learning_rate": 4.734393151475009e-07,
      "loss": 0.7534,
      "step": 11783
    },
    {
      "epoch": 0.8645872501994736,
      "grad_norm": 1.6469662057779764,
      "learning_rate": 4.7293475999957595e-07,
      "loss": 0.8048,
      "step": 11784
    },
    {
      "epoch": 0.8645872501994736,
      "lm_loss": 0.4115,
      "step": 11784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8645872501994736,
      "lm_loss": 0.3811,
      "step": 11784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8645872501994736,
      "lm_loss": 0.455,
      "step": 11784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8645872501994736,
      "lm_loss": 0.4044,
      "step": 11784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8645872501994736,
      "lm_loss": 0.4223,
      "step": 11784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8645872501994736,
      "lm_loss": 0.4944,
      "step": 11784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8645872501994736,
      "lm_loss": 0.5729,
      "step": 11784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8645872501994736,
      "lm_loss": 0.4756,
      "step": 11784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8646606197896127,
      "grad_norm": 0.5549039864229622,
      "learning_rate": 4.7243046050623684e-07,
      "loss": 0.4627,
      "step": 11785
    },
    {
      "epoch": 0.8647339893797519,
      "grad_norm": 1.3750228263529427,
      "learning_rate": 4.719264166959625e-07,
      "loss": 0.6174,
      "step": 11786
    },
    {
      "epoch": 0.8648073589698909,
      "grad_norm": 0.5468541732355758,
      "learning_rate": 4.7142262859721675e-07,
      "loss": 0.4729,
      "step": 11787
    },
    {
      "epoch": 0.8648807285600301,
      "grad_norm": 1.4748762882899482,
      "learning_rate": 4.7091909623845024e-07,
      "loss": 0.6576,
      "step": 11788
    },
    {
      "epoch": 0.8649540981501692,
      "grad_norm": 0.5607318607523256,
      "learning_rate": 4.7041581964809733e-07,
      "loss": 0.4912,
      "step": 11789
    },
    {
      "epoch": 0.8650274677403084,
      "grad_norm": 1.4015293649975618,
      "learning_rate": 4.6991279885458154e-07,
      "loss": 0.7002,
      "step": 11790
    },
    {
      "epoch": 0.8651008373304475,
      "grad_norm": 0.5366801069207199,
      "learning_rate": 4.6941003388630836e-07,
      "loss": 0.4749,
      "step": 11791
    },
    {
      "epoch": 0.8651742069205866,
      "grad_norm": 1.5263597654009984,
      "learning_rate": 4.689075247716701e-07,
      "loss": 0.7119,
      "step": 11792
    },
    {
      "epoch": 0.8651742069205866,
      "lm_loss": 1.0664,
      "step": 11792,
      "vm_loss": 0.0435
    },
    {
      "epoch": 0.8651742069205866,
      "lm_loss": 0.1242,
      "step": 11792,
      "vm_loss": 0.2254
    },
    {
      "epoch": 0.8651742069205866,
      "lm_loss": 0.4913,
      "step": 11792,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.8651742069205866,
      "lm_loss": 0.0452,
      "step": 11792,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.8651742069205866,
      "lm_loss": 0.6734,
      "step": 11792,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.8651742069205866,
      "lm_loss": 1.4607,
      "step": 11792,
      "vm_loss": 0.2823
    },
    {
      "epoch": 0.8651742069205866,
      "lm_loss": 0.4681,
      "step": 11792,
      "vm_loss": 0.1083
    },
    {
      "epoch": 0.8651742069205866,
      "lm_loss": 0.7273,
      "step": 11792,
      "vm_loss": 0.143
    },
    {
      "epoch": 0.8652475765107257,
      "grad_norm": 1.3729964139769997,
      "learning_rate": 4.684052715390458e-07,
      "loss": 0.7567,
      "step": 11793
    },
    {
      "epoch": 0.8653209461008649,
      "grad_norm": 1.3827267787148099,
      "learning_rate": 4.6790327421679715e-07,
      "loss": 0.6409,
      "step": 11794
    },
    {
      "epoch": 0.865394315691004,
      "grad_norm": 1.670451651929554,
      "learning_rate": 4.674015328332737e-07,
      "loss": 0.7603,
      "step": 11795
    },
    {
      "epoch": 0.8654676852811432,
      "grad_norm": 1.4837411522172959,
      "learning_rate": 4.669000474168117e-07,
      "loss": 0.66,
      "step": 11796
    },
    {
      "epoch": 0.8655410548712822,
      "grad_norm": 1.537903920580162,
      "learning_rate": 4.663988179957296e-07,
      "loss": 0.7883,
      "step": 11797
    },
    {
      "epoch": 0.8656144244614213,
      "grad_norm": 1.4445139031172838,
      "learning_rate": 4.6589784459833367e-07,
      "loss": 0.7458,
      "step": 11798
    },
    {
      "epoch": 0.8656877940515605,
      "grad_norm": 0.546568490799774,
      "learning_rate": 4.6539712725291507e-07,
      "loss": 0.4691,
      "step": 11799
    },
    {
      "epoch": 0.8657611636416996,
      "grad_norm": 1.5440384329057046,
      "learning_rate": 4.6489666598774964e-07,
      "loss": 0.6774,
      "step": 11800
    },
    {
      "epoch": 0.8657611636416996,
      "lm_loss": 0.9957,
      "step": 11800,
      "vm_loss": 0.0644
    },
    {
      "epoch": 0.8657611636416996,
      "lm_loss": 0.3039,
      "step": 11800,
      "vm_loss": 0.1978
    },
    {
      "epoch": 0.8657611636416996,
      "lm_loss": 0.8879,
      "step": 11800,
      "vm_loss": 0.113
    },
    {
      "epoch": 0.8657611636416996,
      "lm_loss": 0.4987,
      "step": 11800,
      "vm_loss": 0.075
    },
    {
      "epoch": 0.8657611636416996,
      "lm_loss": 0.575,
      "step": 11800,
      "vm_loss": 0.1163
    },
    {
      "epoch": 0.8657611636416996,
      "lm_loss": 0.1228,
      "step": 11800,
      "vm_loss": 0.1387
    },
    {
      "epoch": 0.8657611636416996,
      "lm_loss": 0.0008,
      "step": 11800,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.8657611636416996,
      "lm_loss": 0.0812,
      "step": 11800,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.8658345332318388,
      "grad_norm": 1.5042670101486342,
      "learning_rate": 4.643964608311008e-07,
      "loss": 0.7144,
      "step": 11801
    },
    {
      "epoch": 0.8659079028219778,
      "grad_norm": 1.7414394954032206,
      "learning_rate": 4.6389651181121654e-07,
      "loss": 0.7507,
      "step": 11802
    },
    {
      "epoch": 0.865981272412117,
      "grad_norm": 0.5559302082890522,
      "learning_rate": 4.633968189563298e-07,
      "loss": 0.4624,
      "step": 11803
    },
    {
      "epoch": 0.8660546420022561,
      "grad_norm": 1.5803729488638343,
      "learning_rate": 4.628973822946592e-07,
      "loss": 0.7238,
      "step": 11804
    },
    {
      "epoch": 0.8661280115923953,
      "grad_norm": 1.357468897542566,
      "learning_rate": 4.6239820185440875e-07,
      "loss": 0.7143,
      "step": 11805
    },
    {
      "epoch": 0.8662013811825344,
      "grad_norm": 0.5490808586867948,
      "learning_rate": 4.618992776637693e-07,
      "loss": 0.4787,
      "step": 11806
    },
    {
      "epoch": 0.8662747507726735,
      "grad_norm": 1.4422175791268044,
      "learning_rate": 4.614006097509166e-07,
      "loss": 0.6369,
      "step": 11807
    },
    {
      "epoch": 0.8663481203628126,
      "grad_norm": 1.8352137866571063,
      "learning_rate": 4.6090219814401094e-07,
      "loss": 0.7666,
      "step": 11808
    },
    {
      "epoch": 0.8663481203628126,
      "lm_loss": 1.2717,
      "step": 11808,
      "vm_loss": 0.1751
    },
    {
      "epoch": 0.8663481203628126,
      "lm_loss": 1.0098,
      "step": 11808,
      "vm_loss": 0.1119
    },
    {
      "epoch": 0.8663481203628126,
      "lm_loss": 0.9,
      "step": 11808,
      "vm_loss": 0.09
    },
    {
      "epoch": 0.8663481203628126,
      "lm_loss": 0.0884,
      "step": 11808,
      "vm_loss": 0.157
    },
    {
      "epoch": 0.8663481203628126,
      "lm_loss": 0.7794,
      "step": 11808,
      "vm_loss": 0.1724
    },
    {
      "epoch": 0.8663481203628126,
      "lm_loss": 0.184,
      "step": 11808,
      "vm_loss": 0.1464
    },
    {
      "epoch": 0.8663481203628126,
      "lm_loss": 0.5601,
      "step": 11808,
      "vm_loss": 0.1315
    },
    {
      "epoch": 0.8663481203628126,
      "lm_loss": 1.1789,
      "step": 11808,
      "vm_loss": 0.1883
    },
    {
      "epoch": 0.8664214899529518,
      "grad_norm": 1.648184952107888,
      "learning_rate": 4.6040404287119924e-07,
      "loss": 0.7487,
      "step": 11809
    },
    {
      "epoch": 0.8664948595430909,
      "grad_norm": 1.4042533548374165,
      "learning_rate": 4.5990614396061284e-07,
      "loss": 0.6934,
      "step": 11810
    },
    {
      "epoch": 0.86656822913323,
      "grad_norm": 1.5453782272844307,
      "learning_rate": 4.594085014403693e-07,
      "loss": 0.7164,
      "step": 11811
    },
    {
      "epoch": 0.8666415987233691,
      "grad_norm": 1.5703610333838283,
      "learning_rate": 4.5891111533857336e-07,
      "loss": 0.6939,
      "step": 11812
    },
    {
      "epoch": 0.8667149683135082,
      "grad_norm": 0.5590269533857563,
      "learning_rate": 4.5841398568331195e-07,
      "loss": 0.4817,
      "step": 11813
    },
    {
      "epoch": 0.8667883379036474,
      "grad_norm": 1.6184771735564767,
      "learning_rate": 4.5791711250266045e-07,
      "loss": 0.7902,
      "step": 11814
    },
    {
      "epoch": 0.8668617074937865,
      "grad_norm": 0.549920005134447,
      "learning_rate": 4.574204958246775e-07,
      "loss": 0.4813,
      "step": 11815
    },
    {
      "epoch": 0.8669350770839256,
      "grad_norm": 1.6106932821680437,
      "learning_rate": 4.569241356774079e-07,
      "loss": 0.8015,
      "step": 11816
    },
    {
      "epoch": 0.8669350770839256,
      "lm_loss": 1.3957,
      "step": 11816,
      "vm_loss": 0.0598
    },
    {
      "epoch": 0.8669350770839256,
      "lm_loss": 0.1382,
      "step": 11816,
      "vm_loss": 0.138
    },
    {
      "epoch": 0.8669350770839256,
      "lm_loss": 0.3327,
      "step": 11816,
      "vm_loss": 0.0722
    },
    {
      "epoch": 0.8669350770839256,
      "lm_loss": 0.226,
      "step": 11816,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.8669350770839256,
      "lm_loss": 0.0092,
      "step": 11816,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.8669350770839256,
      "lm_loss": 0.574,
      "step": 11816,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.8669350770839256,
      "lm_loss": 0.0969,
      "step": 11816,
      "vm_loss": 0.1452
    },
    {
      "epoch": 0.8669350770839256,
      "lm_loss": 0.1507,
      "step": 11816,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.8670084466740647,
      "grad_norm": 1.5807737522316199,
      "learning_rate": 4.564280320888842e-07,
      "loss": 0.6246,
      "step": 11817
    },
    {
      "epoch": 0.8670818162642039,
      "grad_norm": 0.5615317584185999,
      "learning_rate": 4.559321850871212e-07,
      "loss": 0.4856,
      "step": 11818
    },
    {
      "epoch": 0.867155185854343,
      "grad_norm": 1.6004692468065138,
      "learning_rate": 4.5543659470012105e-07,
      "loss": 0.7104,
      "step": 11819
    },
    {
      "epoch": 0.8672285554444822,
      "grad_norm": 1.4489203043079046,
      "learning_rate": 4.5494126095587063e-07,
      "loss": 0.8022,
      "step": 11820
    },
    {
      "epoch": 0.8673019250346212,
      "grad_norm": 0.5485329552759747,
      "learning_rate": 4.544461838823433e-07,
      "loss": 0.4907,
      "step": 11821
    },
    {
      "epoch": 0.8673752946247604,
      "grad_norm": 1.4918241919907047,
      "learning_rate": 4.539513635074955e-07,
      "loss": 0.7783,
      "step": 11822
    },
    {
      "epoch": 0.8674486642148995,
      "grad_norm": 0.5466946343600029,
      "learning_rate": 4.534567998592732e-07,
      "loss": 0.4769,
      "step": 11823
    },
    {
      "epoch": 0.8675220338050387,
      "grad_norm": 1.434256579746089,
      "learning_rate": 4.529624929656046e-07,
      "loss": 0.7193,
      "step": 11824
    },
    {
      "epoch": 0.8675220338050387,
      "lm_loss": 1.0538,
      "step": 11824,
      "vm_loss": 0.0793
    },
    {
      "epoch": 0.8675220338050387,
      "lm_loss": 0.2918,
      "step": 11824,
      "vm_loss": 0.0607
    },
    {
      "epoch": 0.8675220338050387,
      "lm_loss": 0.4501,
      "step": 11824,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.8675220338050387,
      "lm_loss": 0.7686,
      "step": 11824,
      "vm_loss": 0.1754
    },
    {
      "epoch": 0.8675220338050387,
      "lm_loss": 0.0927,
      "step": 11824,
      "vm_loss": 0.1117
    },
    {
      "epoch": 0.8675220338050387,
      "lm_loss": 0.7964,
      "step": 11824,
      "vm_loss": 0.1858
    },
    {
      "epoch": 0.8675220338050387,
      "lm_loss": 0.1132,
      "step": 11824,
      "vm_loss": 0.1477
    },
    {
      "epoch": 0.8675220338050387,
      "lm_loss": 0.4453,
      "step": 11824,
      "vm_loss": 0.051
    },
    {
      "epoch": 0.8675954033951778,
      "grad_norm": 1.5052299688091657,
      "learning_rate": 4.524684428544052e-07,
      "loss": 0.6476,
      "step": 11825
    },
    {
      "epoch": 0.8676687729853169,
      "grad_norm": 0.543459592619633,
      "learning_rate": 4.5197464955357427e-07,
      "loss": 0.4678,
      "step": 11826
    },
    {
      "epoch": 0.867742142575456,
      "grad_norm": 0.5445114820865887,
      "learning_rate": 4.5148111309099683e-07,
      "loss": 0.4798,
      "step": 11827
    },
    {
      "epoch": 0.8678155121655952,
      "grad_norm": 1.4745750616491415,
      "learning_rate": 4.5098783349454613e-07,
      "loss": 0.8364,
      "step": 11828
    },
    {
      "epoch": 0.8678888817557343,
      "grad_norm": 1.6090330939788071,
      "learning_rate": 4.504948107920781e-07,
      "loss": 0.8108,
      "step": 11829
    },
    {
      "epoch": 0.8679622513458735,
      "grad_norm": 1.5051231791070065,
      "learning_rate": 4.500020450114345e-07,
      "loss": 0.7153,
      "step": 11830
    },
    {
      "epoch": 0.8680356209360125,
      "grad_norm": 1.6336376890346809,
      "learning_rate": 4.49509536180443e-07,
      "loss": 0.7706,
      "step": 11831
    },
    {
      "epoch": 0.8681089905261516,
      "grad_norm": 1.558986582498721,
      "learning_rate": 4.4901728432691695e-07,
      "loss": 0.7001,
      "step": 11832
    },
    {
      "epoch": 0.8681089905261516,
      "lm_loss": 0.396,
      "step": 11832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8681089905261516,
      "lm_loss": 0.3874,
      "step": 11832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8681089905261516,
      "lm_loss": 0.4169,
      "step": 11832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8681089905261516,
      "lm_loss": 0.5187,
      "step": 11832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8681089905261516,
      "lm_loss": 0.2856,
      "step": 11832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8681089905261516,
      "lm_loss": 0.4108,
      "step": 11832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8681089905261516,
      "lm_loss": 0.4727,
      "step": 11832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8681089905261516,
      "lm_loss": 0.446,
      "step": 11832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8681823601162908,
      "grad_norm": 0.5442073835482605,
      "learning_rate": 4.485252894786557e-07,
      "loss": 0.4743,
      "step": 11833
    },
    {
      "epoch": 0.86825572970643,
      "grad_norm": 1.9496011570090601,
      "learning_rate": 4.4803355166344267e-07,
      "loss": 0.7578,
      "step": 11834
    },
    {
      "epoch": 0.8683290992965691,
      "grad_norm": 1.6637354933117625,
      "learning_rate": 4.4754207090904835e-07,
      "loss": 0.7425,
      "step": 11835
    },
    {
      "epoch": 0.8684024688867081,
      "grad_norm": 1.4915260037661002,
      "learning_rate": 4.4705084724322665e-07,
      "loss": 0.6679,
      "step": 11836
    },
    {
      "epoch": 0.8684758384768473,
      "grad_norm": 1.6998118131021285,
      "learning_rate": 4.465598806937188e-07,
      "loss": 0.8045,
      "step": 11837
    },
    {
      "epoch": 0.8685492080669864,
      "grad_norm": 1.4977206336429967,
      "learning_rate": 4.460691712882503e-07,
      "loss": 0.8307,
      "step": 11838
    },
    {
      "epoch": 0.8686225776571256,
      "grad_norm": 1.4315858619601611,
      "learning_rate": 4.455787190545341e-07,
      "loss": 0.7658,
      "step": 11839
    },
    {
      "epoch": 0.8686959472472647,
      "grad_norm": 1.563315108659039,
      "learning_rate": 4.450885240202668e-07,
      "loss": 0.7582,
      "step": 11840
    },
    {
      "epoch": 0.8686959472472647,
      "lm_loss": 1.3324,
      "step": 11840,
      "vm_loss": 0.0954
    },
    {
      "epoch": 0.8686959472472647,
      "lm_loss": 0.3161,
      "step": 11840,
      "vm_loss": 0.1606
    },
    {
      "epoch": 0.8686959472472647,
      "lm_loss": 0.8314,
      "step": 11840,
      "vm_loss": 0.1554
    },
    {
      "epoch": 0.8686959472472647,
      "lm_loss": 0.1389,
      "step": 11840,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.8686959472472647,
      "lm_loss": 0.6124,
      "step": 11840,
      "vm_loss": 0.055
    },
    {
      "epoch": 0.8686959472472647,
      "lm_loss": 0.0694,
      "step": 11840,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.8686959472472647,
      "lm_loss": 0.2661,
      "step": 11840,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.8686959472472647,
      "lm_loss": 0.1992,
      "step": 11840,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.8687693168374038,
      "grad_norm": 1.5049073664823707,
      "learning_rate": 4.445985862131302e-07,
      "loss": 0.6622,
      "step": 11841
    },
    {
      "epoch": 0.8688426864275429,
      "grad_norm": 1.6056396765043954,
      "learning_rate": 4.4410890566079277e-07,
      "loss": 0.7206,
      "step": 11842
    },
    {
      "epoch": 0.8689160560176821,
      "grad_norm": 0.5545119111408665,
      "learning_rate": 4.436194823909068e-07,
      "loss": 0.4862,
      "step": 11843
    },
    {
      "epoch": 0.8689894256078212,
      "grad_norm": 1.4701423125264033,
      "learning_rate": 4.4313031643111347e-07,
      "loss": 0.7363,
      "step": 11844
    },
    {
      "epoch": 0.8690627951979604,
      "grad_norm": 1.5382346765769486,
      "learning_rate": 4.426414078090363e-07,
      "loss": 0.5986,
      "step": 11845
    },
    {
      "epoch": 0.8691361647880994,
      "grad_norm": 0.5478873344875745,
      "learning_rate": 4.421527565522848e-07,
      "loss": 0.4667,
      "step": 11846
    },
    {
      "epoch": 0.8692095343782386,
      "grad_norm": 1.7431848057362451,
      "learning_rate": 4.4166436268845424e-07,
      "loss": 0.8464,
      "step": 11847
    },
    {
      "epoch": 0.8692829039683777,
      "grad_norm": 1.5771496852084184,
      "learning_rate": 4.411762262451258e-07,
      "loss": 0.7403,
      "step": 11848
    },
    {
      "epoch": 0.8692829039683777,
      "lm_loss": 1.0788,
      "step": 11848,
      "vm_loss": 0.0454
    },
    {
      "epoch": 0.8692829039683777,
      "lm_loss": 0.0924,
      "step": 11848,
      "vm_loss": 0.0716
    },
    {
      "epoch": 0.8692829039683777,
      "lm_loss": 0.9327,
      "step": 11848,
      "vm_loss": 0.0766
    },
    {
      "epoch": 0.8692829039683777,
      "lm_loss": 0.4318,
      "step": 11848,
      "vm_loss": 0.1791
    },
    {
      "epoch": 0.8692829039683777,
      "lm_loss": 0.3125,
      "step": 11848,
      "vm_loss": 0.1076
    },
    {
      "epoch": 0.8692829039683777,
      "lm_loss": 0.4974,
      "step": 11848,
      "vm_loss": 0.1756
    },
    {
      "epoch": 0.8692829039683777,
      "lm_loss": 0.261,
      "step": 11848,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.8692829039683777,
      "lm_loss": 0.8491,
      "step": 11848,
      "vm_loss": 0.1234
    },
    {
      "epoch": 0.8693562735585169,
      "grad_norm": 1.6606054473202552,
      "learning_rate": 4.4068834724986466e-07,
      "loss": 0.6177,
      "step": 11849
    },
    {
      "epoch": 0.869429643148656,
      "grad_norm": 0.542507681228186,
      "learning_rate": 4.402007257302243e-07,
      "loss": 0.4811,
      "step": 11850
    },
    {
      "epoch": 0.869503012738795,
      "grad_norm": 0.552148466966917,
      "learning_rate": 4.397133617137411e-07,
      "loss": 0.4716,
      "step": 11851
    },
    {
      "epoch": 0.8695763823289342,
      "grad_norm": 1.4362764768980112,
      "learning_rate": 4.3922625522793807e-07,
      "loss": 0.702,
      "step": 11852
    },
    {
      "epoch": 0.8696497519190733,
      "grad_norm": 0.5406041519804154,
      "learning_rate": 4.387394063003225e-07,
      "loss": 0.4683,
      "step": 11853
    },
    {
      "epoch": 0.8697231215092125,
      "grad_norm": 1.8328553690208145,
      "learning_rate": 4.382528149583881e-07,
      "loss": 0.6862,
      "step": 11854
    },
    {
      "epoch": 0.8697964910993516,
      "grad_norm": 1.780241100222714,
      "learning_rate": 4.377664812296145e-07,
      "loss": 0.7834,
      "step": 11855
    },
    {
      "epoch": 0.8698698606894907,
      "grad_norm": 1.5485941014863163,
      "learning_rate": 4.372804051414664e-07,
      "loss": 0.7757,
      "step": 11856
    },
    {
      "epoch": 0.8698698606894907,
      "lm_loss": 0.5516,
      "step": 11856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8698698606894907,
      "lm_loss": 0.3344,
      "step": 11856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8698698606894907,
      "lm_loss": 0.5803,
      "step": 11856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8698698606894907,
      "lm_loss": 0.3557,
      "step": 11856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8698698606894907,
      "lm_loss": 0.4334,
      "step": 11856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8698698606894907,
      "lm_loss": 0.559,
      "step": 11856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8698698606894907,
      "lm_loss": 0.5955,
      "step": 11856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8698698606894907,
      "lm_loss": 0.5172,
      "step": 11856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8699432302796298,
      "grad_norm": 0.552880042203964,
      "learning_rate": 4.3679458672139297e-07,
      "loss": 0.4751,
      "step": 11857
    },
    {
      "epoch": 0.870016599869769,
      "grad_norm": 1.6053640643272065,
      "learning_rate": 4.3630902599683e-07,
      "loss": 0.6505,
      "step": 11858
    },
    {
      "epoch": 0.8700899694599081,
      "grad_norm": 1.5725094299455473,
      "learning_rate": 4.358237229951967e-07,
      "loss": 0.6842,
      "step": 11859
    },
    {
      "epoch": 0.8701633390500473,
      "grad_norm": 1.369941663194813,
      "learning_rate": 4.3533867774390224e-07,
      "loss": 0.6944,
      "step": 11860
    },
    {
      "epoch": 0.8702367086401863,
      "grad_norm": 0.5213986166794212,
      "learning_rate": 4.348538902703364e-07,
      "loss": 0.4736,
      "step": 11861
    },
    {
      "epoch": 0.8703100782303255,
      "grad_norm": 1.526431525947975,
      "learning_rate": 4.343693606018773e-07,
      "loss": 0.6337,
      "step": 11862
    },
    {
      "epoch": 0.8703834478204646,
      "grad_norm": 1.6067400703609342,
      "learning_rate": 4.338850887658863e-07,
      "loss": 0.7871,
      "step": 11863
    },
    {
      "epoch": 0.8704568174106038,
      "grad_norm": 1.557146866709426,
      "learning_rate": 4.3340107478971273e-07,
      "loss": 0.7557,
      "step": 11864
    },
    {
      "epoch": 0.8704568174106038,
      "lm_loss": 1.4728,
      "step": 11864,
      "vm_loss": 0.1221
    },
    {
      "epoch": 0.8704568174106038,
      "lm_loss": 1.334,
      "step": 11864,
      "vm_loss": 0.1433
    },
    {
      "epoch": 0.8704568174106038,
      "lm_loss": 0.6614,
      "step": 11864,
      "vm_loss": 0.1626
    },
    {
      "epoch": 0.8704568174106038,
      "lm_loss": 0.2723,
      "step": 11864,
      "vm_loss": 0.1125
    },
    {
      "epoch": 0.8704568174106038,
      "lm_loss": 0.7067,
      "step": 11864,
      "vm_loss": 0.1322
    },
    {
      "epoch": 0.8704568174106038,
      "lm_loss": 0.1244,
      "step": 11864,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.8704568174106038,
      "lm_loss": 0.7611,
      "step": 11864,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.8704568174106038,
      "lm_loss": 0.2211,
      "step": 11864,
      "vm_loss": 0.1458
    },
    {
      "epoch": 0.8705301870007429,
      "grad_norm": 1.5043475539897084,
      "learning_rate": 4.329173187006885e-07,
      "loss": 0.7275,
      "step": 11865
    },
    {
      "epoch": 0.870603556590882,
      "grad_norm": 1.7312756159958054,
      "learning_rate": 4.3243382052613415e-07,
      "loss": 0.7937,
      "step": 11866
    },
    {
      "epoch": 0.8706769261810211,
      "grad_norm": 1.6502644047798078,
      "learning_rate": 4.319505802933532e-07,
      "loss": 0.8085,
      "step": 11867
    },
    {
      "epoch": 0.8707502957711603,
      "grad_norm": 1.474738570523933,
      "learning_rate": 4.314675980296357e-07,
      "loss": 0.6924,
      "step": 11868
    },
    {
      "epoch": 0.8708236653612994,
      "grad_norm": 1.475403211076605,
      "learning_rate": 4.309848737622568e-07,
      "loss": 0.7321,
      "step": 11869
    },
    {
      "epoch": 0.8708970349514386,
      "grad_norm": 0.5508832060033771,
      "learning_rate": 4.3050240751847606e-07,
      "loss": 0.4903,
      "step": 11870
    },
    {
      "epoch": 0.8709704045415776,
      "grad_norm": 0.5412528172124956,
      "learning_rate": 4.300201993255415e-07,
      "loss": 0.4729,
      "step": 11871
    },
    {
      "epoch": 0.8710437741317167,
      "grad_norm": 1.4700725875970395,
      "learning_rate": 4.2953824921068364e-07,
      "loss": 0.7306,
      "step": 11872
    },
    {
      "epoch": 0.8710437741317167,
      "lm_loss": 1.4167,
      "step": 11872,
      "vm_loss": 0.0633
    },
    {
      "epoch": 0.8710437741317167,
      "lm_loss": 0.8324,
      "step": 11872,
      "vm_loss": 0.1505
    },
    {
      "epoch": 0.8710437741317167,
      "lm_loss": 0.4357,
      "step": 11872,
      "vm_loss": 0.1544
    },
    {
      "epoch": 0.8710437741317167,
      "lm_loss": 0.3279,
      "step": 11872,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.8710437741317167,
      "lm_loss": 0.2658,
      "step": 11872,
      "vm_loss": 0.1844
    },
    {
      "epoch": 0.8710437741317167,
      "lm_loss": 0.2316,
      "step": 11872,
      "vm_loss": 0.0825
    },
    {
      "epoch": 0.8710437741317167,
      "lm_loss": 0.0251,
      "step": 11872,
      "vm_loss": 0.1369
    },
    {
      "epoch": 0.8710437741317167,
      "lm_loss": 0.7418,
      "step": 11872,
      "vm_loss": 0.0948
    },
    {
      "epoch": 0.8711171437218559,
      "grad_norm": 1.4617236829126081,
      "learning_rate": 4.290565572011196e-07,
      "loss": 0.6784,
      "step": 11873
    },
    {
      "epoch": 0.871190513311995,
      "grad_norm": 1.5744568932571368,
      "learning_rate": 4.285751233240515e-07,
      "loss": 0.7336,
      "step": 11874
    },
    {
      "epoch": 0.8712638829021342,
      "grad_norm": 1.2582422931555084,
      "learning_rate": 4.28093947606667e-07,
      "loss": 0.6606,
      "step": 11875
    },
    {
      "epoch": 0.8713372524922732,
      "grad_norm": 0.5528436103987708,
      "learning_rate": 4.276130300761383e-07,
      "loss": 0.4568,
      "step": 11876
    },
    {
      "epoch": 0.8714106220824124,
      "grad_norm": 1.7258606326903072,
      "learning_rate": 4.2713237075962646e-07,
      "loss": 0.8268,
      "step": 11877
    },
    {
      "epoch": 0.8714839916725515,
      "grad_norm": 1.429520819646338,
      "learning_rate": 4.266519696842736e-07,
      "loss": 0.7512,
      "step": 11878
    },
    {
      "epoch": 0.8715573612626907,
      "grad_norm": 0.5548382562100966,
      "learning_rate": 4.2617182687721026e-07,
      "loss": 0.4743,
      "step": 11879
    },
    {
      "epoch": 0.8716307308528298,
      "grad_norm": 1.3367581771781698,
      "learning_rate": 4.2569194236555036e-07,
      "loss": 0.6843,
      "step": 11880
    },
    {
      "epoch": 0.8716307308528298,
      "lm_loss": 0.518,
      "step": 11880,
      "vm_loss": 0.0478
    },
    {
      "epoch": 0.8716307308528298,
      "lm_loss": 1.0166,
      "step": 11880,
      "vm_loss": 0.1378
    },
    {
      "epoch": 0.8716307308528298,
      "lm_loss": 0.941,
      "step": 11880,
      "vm_loss": 0.0698
    },
    {
      "epoch": 0.8716307308528298,
      "lm_loss": 0.8799,
      "step": 11880,
      "vm_loss": 0.1698
    },
    {
      "epoch": 0.8716307308528298,
      "lm_loss": 1.1827,
      "step": 11880,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.8716307308528298,
      "lm_loss": 0.7288,
      "step": 11880,
      "vm_loss": 0.1284
    },
    {
      "epoch": 0.8716307308528298,
      "lm_loss": 0.1409,
      "step": 11880,
      "vm_loss": 0.0987
    },
    {
      "epoch": 0.8716307308528298,
      "lm_loss": 0.316,
      "step": 11880,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.8717041004429689,
      "grad_norm": 1.5004354881175863,
      "learning_rate": 4.252123161763938e-07,
      "loss": 0.6771,
      "step": 11881
    },
    {
      "epoch": 0.871777470033108,
      "grad_norm": 1.4618578476986628,
      "learning_rate": 4.2473294833682786e-07,
      "loss": 0.6216,
      "step": 11882
    },
    {
      "epoch": 0.8718508396232472,
      "grad_norm": 1.527105543883842,
      "learning_rate": 4.242538388739231e-07,
      "loss": 0.8146,
      "step": 11883
    },
    {
      "epoch": 0.8719242092133863,
      "grad_norm": 1.5999940838824855,
      "learning_rate": 4.237749878147357e-07,
      "loss": 0.615,
      "step": 11884
    },
    {
      "epoch": 0.8719975788035255,
      "grad_norm": 1.6061153861078985,
      "learning_rate": 4.2329639518630726e-07,
      "loss": 0.6569,
      "step": 11885
    },
    {
      "epoch": 0.8720709483936645,
      "grad_norm": 1.6824336848105814,
      "learning_rate": 4.228180610156646e-07,
      "loss": 0.7897,
      "step": 11886
    },
    {
      "epoch": 0.8721443179838037,
      "grad_norm": 1.7002592080999208,
      "learning_rate": 4.2233998532982214e-07,
      "loss": 0.7578,
      "step": 11887
    },
    {
      "epoch": 0.8722176875739428,
      "grad_norm": 1.532475394517419,
      "learning_rate": 4.2186216815577663e-07,
      "loss": 0.7715,
      "step": 11888
    },
    {
      "epoch": 0.8722176875739428,
      "lm_loss": 1.2147,
      "step": 11888,
      "vm_loss": 0.0528
    },
    {
      "epoch": 0.8722176875739428,
      "lm_loss": 0.2089,
      "step": 11888,
      "vm_loss": 0.0507
    },
    {
      "epoch": 0.8722176875739428,
      "lm_loss": 0.5936,
      "step": 11888,
      "vm_loss": 0.1465
    },
    {
      "epoch": 0.8722176875739428,
      "lm_loss": 0.5613,
      "step": 11888,
      "vm_loss": 0.2381
    },
    {
      "epoch": 0.8722176875739428,
      "lm_loss": 0.1618,
      "step": 11888,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.8722176875739428,
      "lm_loss": 0.7638,
      "step": 11888,
      "vm_loss": 0.1599
    },
    {
      "epoch": 0.8722176875739428,
      "lm_loss": 0.2069,
      "step": 11888,
      "vm_loss": 0.0878
    },
    {
      "epoch": 0.8722176875739428,
      "lm_loss": 0.8311,
      "step": 11888,
      "vm_loss": 0.2061
    },
    {
      "epoch": 0.872291057164082,
      "grad_norm": 1.7709211460562284,
      "learning_rate": 4.213846095205126e-07,
      "loss": 0.7173,
      "step": 11889
    },
    {
      "epoch": 0.872364426754221,
      "grad_norm": 0.5510302398556228,
      "learning_rate": 4.2090730945099857e-07,
      "loss": 0.4777,
      "step": 11890
    },
    {
      "epoch": 0.8724377963443601,
      "grad_norm": 1.6151954457152378,
      "learning_rate": 4.204302679741884e-07,
      "loss": 0.7624,
      "step": 11891
    },
    {
      "epoch": 0.8725111659344993,
      "grad_norm": 0.549208162127484,
      "learning_rate": 4.199534851170212e-07,
      "loss": 0.4717,
      "step": 11892
    },
    {
      "epoch": 0.8725845355246384,
      "grad_norm": 1.5736661914218553,
      "learning_rate": 4.1947696090642365e-07,
      "loss": 0.7303,
      "step": 11893
    },
    {
      "epoch": 0.8726579051147776,
      "grad_norm": 1.464224591802592,
      "learning_rate": 4.19000695369306e-07,
      "loss": 0.5992,
      "step": 11894
    },
    {
      "epoch": 0.8727312747049166,
      "grad_norm": 0.5310887350666288,
      "learning_rate": 4.185246885325633e-07,
      "loss": 0.4863,
      "step": 11895
    },
    {
      "epoch": 0.8728046442950558,
      "grad_norm": 1.5234906647311373,
      "learning_rate": 4.1804894042307806e-07,
      "loss": 0.7945,
      "step": 11896
    },
    {
      "epoch": 0.8728046442950558,
      "lm_loss": 1.312,
      "step": 11896,
      "vm_loss": 0.1293
    },
    {
      "epoch": 0.8728046442950558,
      "lm_loss": 0.3174,
      "step": 11896,
      "vm_loss": 0.0499
    },
    {
      "epoch": 0.8728046442950558,
      "lm_loss": 0.7837,
      "step": 11896,
      "vm_loss": 0.0499
    },
    {
      "epoch": 0.8728046442950558,
      "lm_loss": 0.0578,
      "step": 11896,
      "vm_loss": 0.1025
    },
    {
      "epoch": 0.8728046442950558,
      "lm_loss": 1.007,
      "step": 11896,
      "vm_loss": 0.1696
    },
    {
      "epoch": 0.8728046442950558,
      "lm_loss": 0.3583,
      "step": 11896,
      "vm_loss": 0.1377
    },
    {
      "epoch": 0.8728046442950558,
      "lm_loss": 0.4687,
      "step": 11896,
      "vm_loss": 0.1537
    },
    {
      "epoch": 0.8728046442950558,
      "lm_loss": 0.7685,
      "step": 11896,
      "vm_loss": 0.1638
    },
    {
      "epoch": 0.8728780138851949,
      "grad_norm": 1.8312554828786496,
      "learning_rate": 4.1757345106771486e-07,
      "loss": 0.7455,
      "step": 11897
    },
    {
      "epoch": 0.8729513834753341,
      "grad_norm": 0.5541137691141368,
      "learning_rate": 4.170982204933277e-07,
      "loss": 0.4623,
      "step": 11898
    },
    {
      "epoch": 0.8730247530654732,
      "grad_norm": 1.5315539012387456,
      "learning_rate": 4.1662324872675354e-07,
      "loss": 0.8032,
      "step": 11899
    },
    {
      "epoch": 0.8730981226556123,
      "grad_norm": 1.6713374172022832,
      "learning_rate": 4.161485357948153e-07,
      "loss": 0.7376,
      "step": 11900
    },
    {
      "epoch": 0.8731714922457514,
      "grad_norm": 1.599459205291558,
      "learning_rate": 4.156740817243204e-07,
      "loss": 0.8207,
      "step": 11901
    },
    {
      "epoch": 0.8732448618358906,
      "grad_norm": 0.552917500281653,
      "learning_rate": 4.151998865420637e-07,
      "loss": 0.4721,
      "step": 11902
    },
    {
      "epoch": 0.8733182314260297,
      "grad_norm": 0.541966033734294,
      "learning_rate": 4.14725950274823e-07,
      "loss": 0.4861,
      "step": 11903
    },
    {
      "epoch": 0.8733916010161689,
      "grad_norm": 1.5182293012289803,
      "learning_rate": 4.142522729493631e-07,
      "loss": 0.689,
      "step": 11904
    },
    {
      "epoch": 0.8733916010161689,
      "lm_loss": 1.2644,
      "step": 11904,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.8733916010161689,
      "lm_loss": 1.1518,
      "step": 11904,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.8733916010161689,
      "lm_loss": 0.8392,
      "step": 11904,
      "vm_loss": 0.1911
    },
    {
      "epoch": 0.8733916010161689,
      "lm_loss": 0.0161,
      "step": 11904,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.8733916010161689,
      "lm_loss": 1.0517,
      "step": 11904,
      "vm_loss": 0.2097
    },
    {
      "epoch": 0.8733916010161689,
      "lm_loss": 0.1108,
      "step": 11904,
      "vm_loss": 0.1622
    },
    {
      "epoch": 0.8733916010161689,
      "lm_loss": 0.2878,
      "step": 11904,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.8733916010161689,
      "lm_loss": 0.3402,
      "step": 11904,
      "vm_loss": 0.1397
    },
    {
      "epoch": 0.8734649706063079,
      "grad_norm": 1.2298750707264061,
      "learning_rate": 4.137788545924343e-07,
      "loss": 0.6754,
      "step": 11905
    },
    {
      "epoch": 0.873538340196447,
      "grad_norm": 1.441366152714496,
      "learning_rate": 4.1330569523077067e-07,
      "loss": 0.7541,
      "step": 11906
    },
    {
      "epoch": 0.8736117097865862,
      "grad_norm": 1.4961148584570751,
      "learning_rate": 4.1283279489109205e-07,
      "loss": 0.84,
      "step": 11907
    },
    {
      "epoch": 0.8736850793767253,
      "grad_norm": 1.544507265413445,
      "learning_rate": 4.1236015360010706e-07,
      "loss": 0.8188,
      "step": 11908
    },
    {
      "epoch": 0.8737584489668645,
      "grad_norm": 1.4319449765574561,
      "learning_rate": 4.1188777138450487e-07,
      "loss": 0.7261,
      "step": 11909
    },
    {
      "epoch": 0.8738318185570035,
      "grad_norm": 1.5392080618544752,
      "learning_rate": 4.1141564827096245e-07,
      "loss": 0.6763,
      "step": 11910
    },
    {
      "epoch": 0.8739051881471427,
      "grad_norm": 1.4016085869438846,
      "learning_rate": 4.109437842861419e-07,
      "loss": 0.7886,
      "step": 11911
    },
    {
      "epoch": 0.8739785577372818,
      "grad_norm": 1.6538746574990857,
      "learning_rate": 4.104721794566896e-07,
      "loss": 0.7558,
      "step": 11912
    },
    {
      "epoch": 0.8739785577372818,
      "lm_loss": 0.4304,
      "step": 11912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8739785577372818,
      "lm_loss": 0.4245,
      "step": 11912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8739785577372818,
      "lm_loss": 0.3916,
      "step": 11912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8739785577372818,
      "lm_loss": 0.5439,
      "step": 11912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8739785577372818,
      "lm_loss": 0.4576,
      "step": 11912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8739785577372818,
      "lm_loss": 0.6982,
      "step": 11912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8739785577372818,
      "lm_loss": 0.4749,
      "step": 11912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8739785577372818,
      "lm_loss": 0.4504,
      "step": 11912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.874051927327421,
      "grad_norm": 0.550627280851963,
      "learning_rate": 4.100008338092398e-07,
      "loss": 0.4931,
      "step": 11913
    },
    {
      "epoch": 0.8741252969175601,
      "grad_norm": 1.443596128363635,
      "learning_rate": 4.095297473704102e-07,
      "loss": 0.6431,
      "step": 11914
    },
    {
      "epoch": 0.8741986665076992,
      "grad_norm": 1.560114632420437,
      "learning_rate": 4.090589201668038e-07,
      "loss": 0.7479,
      "step": 11915
    },
    {
      "epoch": 0.8742720360978383,
      "grad_norm": 1.3497933683061962,
      "learning_rate": 4.0858835222500947e-07,
      "loss": 0.7493,
      "step": 11916
    },
    {
      "epoch": 0.8743454056879775,
      "grad_norm": 0.5387379084956897,
      "learning_rate": 4.081180435716009e-07,
      "loss": 0.473,
      "step": 11917
    },
    {
      "epoch": 0.8744187752781166,
      "grad_norm": 1.6103083721367906,
      "learning_rate": 4.076479942331374e-07,
      "loss": 0.6312,
      "step": 11918
    },
    {
      "epoch": 0.8744921448682558,
      "grad_norm": 0.5341753952157311,
      "learning_rate": 4.071782042361655e-07,
      "loss": 0.4664,
      "step": 11919
    },
    {
      "epoch": 0.8745655144583948,
      "grad_norm": 1.579493755568433,
      "learning_rate": 4.06708673607214e-07,
      "loss": 0.8214,
      "step": 11920
    },
    {
      "epoch": 0.8745655144583948,
      "lm_loss": 1.3516,
      "step": 11920,
      "vm_loss": 0.0554
    },
    {
      "epoch": 0.8745655144583948,
      "lm_loss": 0.474,
      "step": 11920,
      "vm_loss": 0.0655
    },
    {
      "epoch": 0.8745655144583948,
      "lm_loss": 0.254,
      "step": 11920,
      "vm_loss": 0.1083
    },
    {
      "epoch": 0.8745655144583948,
      "lm_loss": 0.0738,
      "step": 11920,
      "vm_loss": 0.1597
    },
    {
      "epoch": 0.8745655144583948,
      "lm_loss": 0.8046,
      "step": 11920,
      "vm_loss": 0.1305
    },
    {
      "epoch": 0.8745655144583948,
      "lm_loss": 0.0778,
      "step": 11920,
      "vm_loss": 0.0895
    },
    {
      "epoch": 0.8745655144583948,
      "lm_loss": 0.8498,
      "step": 11920,
      "vm_loss": 0.1182
    },
    {
      "epoch": 0.8745655144583948,
      "lm_loss": 0.3106,
      "step": 11920,
      "vm_loss": 0.1103
    },
    {
      "epoch": 0.874638884048534,
      "grad_norm": 1.5023149524889046,
      "learning_rate": 4.062394023727989e-07,
      "loss": 0.7018,
      "step": 11921
    },
    {
      "epoch": 0.8747122536386731,
      "grad_norm": 1.5389503382343603,
      "learning_rate": 4.0577039055942127e-07,
      "loss": 0.7828,
      "step": 11922
    },
    {
      "epoch": 0.8747856232288123,
      "grad_norm": 1.4682018824761234,
      "learning_rate": 4.0530163819356595e-07,
      "loss": 0.7248,
      "step": 11923
    },
    {
      "epoch": 0.8748589928189514,
      "grad_norm": 1.4215554274363262,
      "learning_rate": 4.048331453017068e-07,
      "loss": 0.7312,
      "step": 11924
    },
    {
      "epoch": 0.8749323624090904,
      "grad_norm": 1.645980455129253,
      "learning_rate": 4.0436491191029935e-07,
      "loss": 0.7977,
      "step": 11925
    },
    {
      "epoch": 0.8750057319992296,
      "grad_norm": 1.5813366268599367,
      "learning_rate": 4.038969380457869e-07,
      "loss": 0.7731,
      "step": 11926
    },
    {
      "epoch": 0.8750791015893687,
      "grad_norm": 1.397742906740296,
      "learning_rate": 4.0342922373459546e-07,
      "loss": 0.7701,
      "step": 11927
    },
    {
      "epoch": 0.8751524711795079,
      "grad_norm": 1.5644334969752958,
      "learning_rate": 4.02961769003139e-07,
      "loss": 0.7493,
      "step": 11928
    },
    {
      "epoch": 0.8751524711795079,
      "lm_loss": 0.6458,
      "step": 11928,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.8751524711795079,
      "lm_loss": 0.0569,
      "step": 11928,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.8751524711795079,
      "lm_loss": 0.5571,
      "step": 11928,
      "vm_loss": 0.1555
    },
    {
      "epoch": 0.8751524711795079,
      "lm_loss": 0.6502,
      "step": 11928,
      "vm_loss": 0.1589
    },
    {
      "epoch": 0.8751524711795079,
      "lm_loss": 0.0184,
      "step": 11928,
      "vm_loss": 0.1993
    },
    {
      "epoch": 0.8751524711795079,
      "lm_loss": 0.5677,
      "step": 11928,
      "vm_loss": 0.1374
    },
    {
      "epoch": 0.8751524711795079,
      "lm_loss": 0.6203,
      "step": 11928,
      "vm_loss": 0.1308
    },
    {
      "epoch": 0.8751524711795079,
      "lm_loss": 0.264,
      "step": 11928,
      "vm_loss": 0.1323
    },
    {
      "epoch": 0.875225840769647,
      "grad_norm": 1.3726907379005844,
      "learning_rate": 4.024945738778163e-07,
      "loss": 0.7529,
      "step": 11929
    },
    {
      "epoch": 0.8752992103597861,
      "grad_norm": 1.7119118593963565,
      "learning_rate": 4.0202763838501027e-07,
      "loss": 0.7576,
      "step": 11930
    },
    {
      "epoch": 0.8753725799499252,
      "grad_norm": 1.4450543694364826,
      "learning_rate": 4.015609625510902e-07,
      "loss": 0.7206,
      "step": 11931
    },
    {
      "epoch": 0.8754459495400644,
      "grad_norm": 1.4809007282463469,
      "learning_rate": 4.0109454640241077e-07,
      "loss": 0.7192,
      "step": 11932
    },
    {
      "epoch": 0.8755193191302035,
      "grad_norm": 1.5070132205003675,
      "learning_rate": 4.006283899653113e-07,
      "loss": 0.6228,
      "step": 11933
    },
    {
      "epoch": 0.8755926887203427,
      "grad_norm": 1.4046381235881775,
      "learning_rate": 4.001624932661158e-07,
      "loss": 0.6487,
      "step": 11934
    },
    {
      "epoch": 0.8756660583104817,
      "grad_norm": 0.5323766895473794,
      "learning_rate": 3.996968563311365e-07,
      "loss": 0.4657,
      "step": 11935
    },
    {
      "epoch": 0.8757394279006209,
      "grad_norm": 1.609737593126447,
      "learning_rate": 3.9923147918666804e-07,
      "loss": 0.7262,
      "step": 11936
    },
    {
      "epoch": 0.8757394279006209,
      "lm_loss": 1.1915,
      "step": 11936,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.8757394279006209,
      "lm_loss": 0.2355,
      "step": 11936,
      "vm_loss": 0.1848
    },
    {
      "epoch": 0.8757394279006209,
      "lm_loss": 0.6999,
      "step": 11936,
      "vm_loss": 0.0725
    },
    {
      "epoch": 0.8757394279006209,
      "lm_loss": 0.4716,
      "step": 11936,
      "vm_loss": 0.1198
    },
    {
      "epoch": 0.8757394279006209,
      "lm_loss": 0.3764,
      "step": 11936,
      "vm_loss": 0.1465
    },
    {
      "epoch": 0.8757394279006209,
      "lm_loss": 0.5508,
      "step": 11936,
      "vm_loss": 0.1371
    },
    {
      "epoch": 0.8757394279006209,
      "lm_loss": 0.042,
      "step": 11936,
      "vm_loss": 0.0862
    },
    {
      "epoch": 0.8757394279006209,
      "lm_loss": 0.0667,
      "step": 11936,
      "vm_loss": 0.0638
    },
    {
      "epoch": 0.87581279749076,
      "grad_norm": 1.4120604788221096,
      "learning_rate": 3.987663618589921e-07,
      "loss": 0.7069,
      "step": 11937
    },
    {
      "epoch": 0.8758861670808992,
      "grad_norm": 1.5406057928266919,
      "learning_rate": 3.9830150437437376e-07,
      "loss": 0.6813,
      "step": 11938
    },
    {
      "epoch": 0.8759595366710383,
      "grad_norm": 1.669144752486457,
      "learning_rate": 3.9783690675906484e-07,
      "loss": 0.6978,
      "step": 11939
    },
    {
      "epoch": 0.8760329062611774,
      "grad_norm": 1.4831892094254209,
      "learning_rate": 3.973725690393038e-07,
      "loss": 0.7307,
      "step": 11940
    },
    {
      "epoch": 0.8761062758513165,
      "grad_norm": 0.5281296568120036,
      "learning_rate": 3.969084912413118e-07,
      "loss": 0.4816,
      "step": 11941
    },
    {
      "epoch": 0.8761796454414557,
      "grad_norm": 1.4947250785020638,
      "learning_rate": 3.964446733912969e-07,
      "loss": 0.8111,
      "step": 11942
    },
    {
      "epoch": 0.8762530150315948,
      "grad_norm": 1.543442395833183,
      "learning_rate": 3.9598111551545136e-07,
      "loss": 0.5754,
      "step": 11943
    },
    {
      "epoch": 0.876326384621734,
      "grad_norm": 1.5452398999189645,
      "learning_rate": 3.9551781763995436e-07,
      "loss": 0.76,
      "step": 11944
    },
    {
      "epoch": 0.876326384621734,
      "lm_loss": 0.4465,
      "step": 11944,
      "vm_loss": 0.0393
    },
    {
      "epoch": 0.876326384621734,
      "lm_loss": 0.9252,
      "step": 11944,
      "vm_loss": 0.0862
    },
    {
      "epoch": 0.876326384621734,
      "lm_loss": 0.1761,
      "step": 11944,
      "vm_loss": 0.1464
    },
    {
      "epoch": 0.876326384621734,
      "lm_loss": 0.2298,
      "step": 11944,
      "vm_loss": 0.1547
    },
    {
      "epoch": 0.876326384621734,
      "lm_loss": 0.0748,
      "step": 11944,
      "vm_loss": 0.1533
    },
    {
      "epoch": 0.876326384621734,
      "lm_loss": 0.4401,
      "step": 11944,
      "vm_loss": 0.1312
    },
    {
      "epoch": 0.876326384621734,
      "lm_loss": 0.3538,
      "step": 11944,
      "vm_loss": 0.1079
    },
    {
      "epoch": 0.876326384621734,
      "lm_loss": 0.459,
      "step": 11944,
      "vm_loss": 0.2046
    },
    {
      "epoch": 0.876399754211873,
      "grad_norm": 1.625240879238589,
      "learning_rate": 3.9505477979096816e-07,
      "loss": 0.8103,
      "step": 11945
    },
    {
      "epoch": 0.8764731238020121,
      "grad_norm": 1.464504529109773,
      "learning_rate": 3.945920019946431e-07,
      "loss": 0.7361,
      "step": 11946
    },
    {
      "epoch": 0.8765464933921513,
      "grad_norm": 1.4962297003348535,
      "learning_rate": 3.9412948427711263e-07,
      "loss": 0.7561,
      "step": 11947
    },
    {
      "epoch": 0.8766198629822904,
      "grad_norm": 1.487845048786186,
      "learning_rate": 3.93667226664497e-07,
      "loss": 0.7763,
      "step": 11948
    },
    {
      "epoch": 0.8766932325724296,
      "grad_norm": 0.5400594020672072,
      "learning_rate": 3.9320522918289973e-07,
      "loss": 0.4673,
      "step": 11949
    },
    {
      "epoch": 0.8767666021625686,
      "grad_norm": 0.5660054880034466,
      "learning_rate": 3.9274349185841164e-07,
      "loss": 0.5,
      "step": 11950
    },
    {
      "epoch": 0.8768399717527078,
      "grad_norm": 1.6208398444234722,
      "learning_rate": 3.922820147171086e-07,
      "loss": 0.7424,
      "step": 11951
    },
    {
      "epoch": 0.8769133413428469,
      "grad_norm": 1.6061712325515662,
      "learning_rate": 3.918207977850513e-07,
      "loss": 0.8137,
      "step": 11952
    },
    {
      "epoch": 0.8769133413428469,
      "lm_loss": 0.0567,
      "step": 11952,
      "vm_loss": 0.0373
    },
    {
      "epoch": 0.8769133413428469,
      "lm_loss": 0.2221,
      "step": 11952,
      "vm_loss": 0.0358
    },
    {
      "epoch": 0.8769133413428469,
      "lm_loss": 0.5114,
      "step": 11952,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.8769133413428469,
      "lm_loss": 1.3212,
      "step": 11952,
      "vm_loss": 0.1307
    },
    {
      "epoch": 0.8769133413428469,
      "lm_loss": 0.7681,
      "step": 11952,
      "vm_loss": 0.0772
    },
    {
      "epoch": 0.8769133413428469,
      "lm_loss": 0.6415,
      "step": 11952,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.8769133413428469,
      "lm_loss": 0.5945,
      "step": 11952,
      "vm_loss": 0.0692
    },
    {
      "epoch": 0.8769133413428469,
      "lm_loss": 0.3291,
      "step": 11952,
      "vm_loss": 0.0921
    },
    {
      "epoch": 0.8769867109329861,
      "grad_norm": 1.5980741667218057,
      "learning_rate": 3.9135984108828517e-07,
      "loss": 0.6946,
      "step": 11953
    },
    {
      "epoch": 0.8770600805231252,
      "grad_norm": 1.5379835500234014,
      "learning_rate": 3.9089914465284207e-07,
      "loss": 0.7177,
      "step": 11954
    },
    {
      "epoch": 0.8771334501132643,
      "grad_norm": 1.3565943525074853,
      "learning_rate": 3.9043870850473785e-07,
      "loss": 0.7355,
      "step": 11955
    },
    {
      "epoch": 0.8772068197034034,
      "grad_norm": 1.4574636743231477,
      "learning_rate": 3.8997853266997566e-07,
      "loss": 0.7833,
      "step": 11956
    },
    {
      "epoch": 0.8772801892935426,
      "grad_norm": 0.538089432140309,
      "learning_rate": 3.8951861717454244e-07,
      "loss": 0.4774,
      "step": 11957
    },
    {
      "epoch": 0.8773535588836817,
      "grad_norm": 1.4580665509234865,
      "learning_rate": 3.8905896204441074e-07,
      "loss": 0.5939,
      "step": 11958
    },
    {
      "epoch": 0.8774269284738209,
      "grad_norm": 1.5904488960631453,
      "learning_rate": 3.885995673055376e-07,
      "loss": 0.7179,
      "step": 11959
    },
    {
      "epoch": 0.8775002980639599,
      "grad_norm": 1.5519462525360948,
      "learning_rate": 3.8814043298386727e-07,
      "loss": 0.8189,
      "step": 11960
    },
    {
      "epoch": 0.8775002980639599,
      "lm_loss": 0.4078,
      "step": 11960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8775002980639599,
      "lm_loss": 0.5779,
      "step": 11960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8775002980639599,
      "lm_loss": 0.5,
      "step": 11960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8775002980639599,
      "lm_loss": 0.3982,
      "step": 11960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8775002980639599,
      "lm_loss": 0.5119,
      "step": 11960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8775002980639599,
      "lm_loss": 0.3465,
      "step": 11960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8775002980639599,
      "lm_loss": 0.3634,
      "step": 11960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8775002980639599,
      "lm_loss": 0.5428,
      "step": 11960,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.877573667654099,
      "grad_norm": 0.5429729743852487,
      "learning_rate": 3.8768155910532736e-07,
      "loss": 0.474,
      "step": 11961
    },
    {
      "epoch": 0.8776470372442382,
      "grad_norm": 1.535740397758802,
      "learning_rate": 3.872229456958321e-07,
      "loss": 0.6954,
      "step": 11962
    },
    {
      "epoch": 0.8777204068343774,
      "grad_norm": 0.5295254559519094,
      "learning_rate": 3.867645927812808e-07,
      "loss": 0.4755,
      "step": 11963
    },
    {
      "epoch": 0.8777937764245165,
      "grad_norm": 0.5527991814918244,
      "learning_rate": 3.863065003875577e-07,
      "loss": 0.4893,
      "step": 11964
    },
    {
      "epoch": 0.8778671460146555,
      "grad_norm": 1.6239396100315655,
      "learning_rate": 3.858486685405316e-07,
      "loss": 0.7443,
      "step": 11965
    },
    {
      "epoch": 0.8779405156047947,
      "grad_norm": 0.5522408700626525,
      "learning_rate": 3.853910972660574e-07,
      "loss": 0.4783,
      "step": 11966
    },
    {
      "epoch": 0.8780138851949338,
      "grad_norm": 1.5242894585361006,
      "learning_rate": 3.849337865899766e-07,
      "loss": 0.7361,
      "step": 11967
    },
    {
      "epoch": 0.878087254785073,
      "grad_norm": 1.7979797882676232,
      "learning_rate": 3.8447673653811413e-07,
      "loss": 0.6958,
      "step": 11968
    },
    {
      "epoch": 0.878087254785073,
      "lm_loss": 1.1754,
      "step": 11968,
      "vm_loss": 0.0513
    },
    {
      "epoch": 0.878087254785073,
      "lm_loss": 1.1291,
      "step": 11968,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.878087254785073,
      "lm_loss": 0.7813,
      "step": 11968,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.878087254785073,
      "lm_loss": 0.5291,
      "step": 11968,
      "vm_loss": 0.147
    },
    {
      "epoch": 0.878087254785073,
      "lm_loss": 0.7247,
      "step": 11968,
      "vm_loss": 0.1564
    },
    {
      "epoch": 0.878087254785073,
      "lm_loss": 0.6928,
      "step": 11968,
      "vm_loss": 0.142
    },
    {
      "epoch": 0.878087254785073,
      "lm_loss": 0.4245,
      "step": 11968,
      "vm_loss": 0.1513
    },
    {
      "epoch": 0.878087254785073,
      "lm_loss": 0.7461,
      "step": 11968,
      "vm_loss": 0.0931
    },
    {
      "epoch": 0.878160624375212,
      "grad_norm": 1.4703881229893088,
      "learning_rate": 3.8401994713628044e-07,
      "loss": 0.737,
      "step": 11969
    },
    {
      "epoch": 0.8782339939653512,
      "grad_norm": 1.5860984969242675,
      "learning_rate": 3.8356341841027156e-07,
      "loss": 0.6776,
      "step": 11970
    },
    {
      "epoch": 0.8783073635554903,
      "grad_norm": 1.540716503939616,
      "learning_rate": 3.8310715038586854e-07,
      "loss": 0.694,
      "step": 11971
    },
    {
      "epoch": 0.8783807331456295,
      "grad_norm": 1.6851798559472784,
      "learning_rate": 3.82651143088838e-07,
      "loss": 0.8544,
      "step": 11972
    },
    {
      "epoch": 0.8784541027357686,
      "grad_norm": 1.4945052412996849,
      "learning_rate": 3.8219539654493266e-07,
      "loss": 0.8112,
      "step": 11973
    },
    {
      "epoch": 0.8785274723259077,
      "grad_norm": 1.5496298143707667,
      "learning_rate": 3.817399107798886e-07,
      "loss": 0.8368,
      "step": 11974
    },
    {
      "epoch": 0.8786008419160468,
      "grad_norm": 0.5264789898330446,
      "learning_rate": 3.8128468581942913e-07,
      "loss": 0.4643,
      "step": 11975
    },
    {
      "epoch": 0.878674211506186,
      "grad_norm": 0.5696262488633397,
      "learning_rate": 3.8082972168926146e-07,
      "loss": 0.4677,
      "step": 11976
    },
    {
      "epoch": 0.878674211506186,
      "lm_loss": 0.9794,
      "step": 11976,
      "vm_loss": 0.0403
    },
    {
      "epoch": 0.878674211506186,
      "lm_loss": 0.2356,
      "step": 11976,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.878674211506186,
      "lm_loss": 0.1589,
      "step": 11976,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.878674211506186,
      "lm_loss": 0.2361,
      "step": 11976,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.878674211506186,
      "lm_loss": 0.7305,
      "step": 11976,
      "vm_loss": 0.1556
    },
    {
      "epoch": 0.878674211506186,
      "lm_loss": 0.8765,
      "step": 11976,
      "vm_loss": 0.1658
    },
    {
      "epoch": 0.878674211506186,
      "lm_loss": 0.2385,
      "step": 11976,
      "vm_loss": 0.1343
    },
    {
      "epoch": 0.878674211506186,
      "lm_loss": 0.0194,
      "step": 11976,
      "vm_loss": 0.0981
    },
    {
      "epoch": 0.8787475810963251,
      "grad_norm": 1.561203183961762,
      "learning_rate": 3.8037501841507786e-07,
      "loss": 0.6514,
      "step": 11977
    },
    {
      "epoch": 0.8788209506864643,
      "grad_norm": 1.4673797170982774,
      "learning_rate": 3.799205760225583e-07,
      "loss": 0.6864,
      "step": 11978
    },
    {
      "epoch": 0.8788943202766033,
      "grad_norm": 1.5785160799468216,
      "learning_rate": 3.79466394537365e-07,
      "loss": 0.7257,
      "step": 11979
    },
    {
      "epoch": 0.8789676898667425,
      "grad_norm": 1.5709492554080344,
      "learning_rate": 3.7901247398514694e-07,
      "loss": 0.8374,
      "step": 11980
    },
    {
      "epoch": 0.8790410594568816,
      "grad_norm": 1.48157969556803,
      "learning_rate": 3.785588143915381e-07,
      "loss": 0.7483,
      "step": 11981
    },
    {
      "epoch": 0.8791144290470208,
      "grad_norm": 1.4238493135482349,
      "learning_rate": 3.7810541578215677e-07,
      "loss": 0.7531,
      "step": 11982
    },
    {
      "epoch": 0.8791877986371599,
      "grad_norm": 1.4203636203620842,
      "learning_rate": 3.7765227818260977e-07,
      "loss": 0.7244,
      "step": 11983
    },
    {
      "epoch": 0.8792611682272989,
      "grad_norm": 1.5683007637848956,
      "learning_rate": 3.7719940161848547e-07,
      "loss": 0.72,
      "step": 11984
    },
    {
      "epoch": 0.8792611682272989,
      "lm_loss": 1.3815,
      "step": 11984,
      "vm_loss": 0.0513
    },
    {
      "epoch": 0.8792611682272989,
      "lm_loss": 1.3166,
      "step": 11984,
      "vm_loss": 0.1654
    },
    {
      "epoch": 0.8792611682272989,
      "lm_loss": 0.1637,
      "step": 11984,
      "vm_loss": 0.1011
    },
    {
      "epoch": 0.8792611682272989,
      "lm_loss": 0.7863,
      "step": 11984,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.8792611682272989,
      "lm_loss": 0.1297,
      "step": 11984,
      "vm_loss": 0.062
    },
    {
      "epoch": 0.8792611682272989,
      "lm_loss": 0.5467,
      "step": 11984,
      "vm_loss": 0.1035
    },
    {
      "epoch": 0.8792611682272989,
      "lm_loss": 0.8157,
      "step": 11984,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.8792611682272989,
      "lm_loss": 0.0739,
      "step": 11984,
      "vm_loss": 0.0868
    },
    {
      "epoch": 0.8793345378174381,
      "grad_norm": 1.3831538705319777,
      "learning_rate": 3.767467861153595e-07,
      "loss": 0.6379,
      "step": 11985
    },
    {
      "epoch": 0.8794079074075772,
      "grad_norm": 1.3812769109905432,
      "learning_rate": 3.7629443169879156e-07,
      "loss": 0.6618,
      "step": 11986
    },
    {
      "epoch": 0.8794812769977164,
      "grad_norm": 1.6119542588588103,
      "learning_rate": 3.7584233839432716e-07,
      "loss": 0.6397,
      "step": 11987
    },
    {
      "epoch": 0.8795546465878555,
      "grad_norm": 1.5894932581649994,
      "learning_rate": 3.7539050622749707e-07,
      "loss": 0.8282,
      "step": 11988
    },
    {
      "epoch": 0.8796280161779946,
      "grad_norm": 1.627689426358295,
      "learning_rate": 3.7493893522381866e-07,
      "loss": 0.7008,
      "step": 11989
    },
    {
      "epoch": 0.8797013857681337,
      "grad_norm": 0.5431395750156109,
      "learning_rate": 3.74487625408792e-07,
      "loss": 0.4828,
      "step": 11990
    },
    {
      "epoch": 0.8797747553582729,
      "grad_norm": 0.5373107522057988,
      "learning_rate": 3.7403657680790405e-07,
      "loss": 0.4807,
      "step": 11991
    },
    {
      "epoch": 0.879848124948412,
      "grad_norm": 1.534441187456324,
      "learning_rate": 3.735857894466266e-07,
      "loss": 0.7692,
      "step": 11992
    },
    {
      "epoch": 0.879848124948412,
      "lm_loss": 0.5455,
      "step": 11992,
      "vm_loss": 0.0512
    },
    {
      "epoch": 0.879848124948412,
      "lm_loss": 0.2049,
      "step": 11992,
      "vm_loss": 0.1021
    },
    {
      "epoch": 0.879848124948412,
      "lm_loss": 0.5385,
      "step": 11992,
      "vm_loss": 0.16
    },
    {
      "epoch": 0.879848124948412,
      "lm_loss": 0.1351,
      "step": 11992,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.879848124948412,
      "lm_loss": 0.0627,
      "step": 11992,
      "vm_loss": 0.1546
    },
    {
      "epoch": 0.879848124948412,
      "lm_loss": 1.0024,
      "step": 11992,
      "vm_loss": 0.1307
    },
    {
      "epoch": 0.879848124948412,
      "lm_loss": 0.5551,
      "step": 11992,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.879848124948412,
      "lm_loss": 0.752,
      "step": 11992,
      "vm_loss": 0.1839
    },
    {
      "epoch": 0.8799214945385512,
      "grad_norm": 1.3064130716641678,
      "learning_rate": 3.7313526335041593e-07,
      "loss": 0.6621,
      "step": 11993
    },
    {
      "epoch": 0.8799948641286902,
      "grad_norm": 1.5590756244658446,
      "learning_rate": 3.726849985447162e-07,
      "loss": 0.7827,
      "step": 11994
    },
    {
      "epoch": 0.8800682337188294,
      "grad_norm": 1.602758586014311,
      "learning_rate": 3.7223499505495365e-07,
      "loss": 0.7367,
      "step": 11995
    },
    {
      "epoch": 0.8801416033089685,
      "grad_norm": 0.5449941740280991,
      "learning_rate": 3.717852529065413e-07,
      "loss": 0.4772,
      "step": 11996
    },
    {
      "epoch": 0.8802149728991077,
      "grad_norm": 1.4594260602720965,
      "learning_rate": 3.713357721248778e-07,
      "loss": 0.7634,
      "step": 11997
    },
    {
      "epoch": 0.8802883424892468,
      "grad_norm": 1.5066177649436763,
      "learning_rate": 3.7088655273534556e-07,
      "loss": 0.6797,
      "step": 11998
    },
    {
      "epoch": 0.8803617120793858,
      "grad_norm": 1.4988203785426983,
      "learning_rate": 3.704375947633132e-07,
      "loss": 0.7623,
      "step": 11999
    },
    {
      "epoch": 0.880435081669525,
      "grad_norm": 0.5262566761847485,
      "learning_rate": 3.6998889823413485e-07,
      "loss": 0.4658,
      "step": 12000
    },
    {
      "epoch": 0.880435081669525,
      "lm_loss": 0.1628,
      "step": 12000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.880435081669525,
      "lm_loss": 0.5543,
      "step": 12000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.880435081669525,
      "lm_loss": 0.6501,
      "step": 12000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.880435081669525,
      "lm_loss": 0.5045,
      "step": 12000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.880435081669525,
      "lm_loss": 0.5479,
      "step": 12000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.880435081669525,
      "lm_loss": 0.4587,
      "step": 12000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.880435081669525,
      "lm_loss": 0.3993,
      "step": 12000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.880435081669525,
      "lm_loss": 0.5781,
      "step": 12000,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8805084512596641,
      "grad_norm": 0.5511590677016566,
      "learning_rate": 3.6954046317314973e-07,
      "loss": 0.49,
      "step": 12001
    },
    {
      "epoch": 0.8805818208498033,
      "grad_norm": 1.526828715675347,
      "learning_rate": 3.69092289605682e-07,
      "loss": 0.7217,
      "step": 12002
    },
    {
      "epoch": 0.8806551904399424,
      "grad_norm": 1.4286256602846936,
      "learning_rate": 3.6864437755704083e-07,
      "loss": 0.7211,
      "step": 12003
    },
    {
      "epoch": 0.8807285600300815,
      "grad_norm": 1.4346380303554083,
      "learning_rate": 3.681967270525205e-07,
      "loss": 0.6931,
      "step": 12004
    },
    {
      "epoch": 0.8808019296202206,
      "grad_norm": 1.538327159613357,
      "learning_rate": 3.6774933811740187e-07,
      "loss": 0.7596,
      "step": 12005
    },
    {
      "epoch": 0.8808752992103598,
      "grad_norm": 0.5482189257161509,
      "learning_rate": 3.6730221077694974e-07,
      "loss": 0.476,
      "step": 12006
    },
    {
      "epoch": 0.8809486688004989,
      "grad_norm": 0.5627914060441702,
      "learning_rate": 3.668553450564144e-07,
      "loss": 0.4712,
      "step": 12007
    },
    {
      "epoch": 0.8810220383906381,
      "grad_norm": 0.5450684015719401,
      "learning_rate": 3.6640874098103186e-07,
      "loss": 0.4883,
      "step": 12008
    },
    {
      "epoch": 0.8810220383906381,
      "lm_loss": 1.2558,
      "step": 12008,
      "vm_loss": 0.0604
    },
    {
      "epoch": 0.8810220383906381,
      "lm_loss": 1.2552,
      "step": 12008,
      "vm_loss": 0.169
    },
    {
      "epoch": 0.8810220383906381,
      "lm_loss": 0.7339,
      "step": 12008,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.8810220383906381,
      "lm_loss": 0.733,
      "step": 12008,
      "vm_loss": 0.1635
    },
    {
      "epoch": 0.8810220383906381,
      "lm_loss": 0.7332,
      "step": 12008,
      "vm_loss": 0.1156
    },
    {
      "epoch": 0.8810220383906381,
      "lm_loss": 0.0264,
      "step": 12008,
      "vm_loss": 0.0636
    },
    {
      "epoch": 0.8810220383906381,
      "lm_loss": 0.1017,
      "step": 12008,
      "vm_loss": 0.0839
    },
    {
      "epoch": 0.8810220383906381,
      "lm_loss": 0.71,
      "step": 12008,
      "vm_loss": 0.1023
    },
    {
      "epoch": 0.8810954079807771,
      "grad_norm": 1.530285045251846,
      "learning_rate": 3.6596239857602136e-07,
      "loss": 0.7537,
      "step": 12009
    },
    {
      "epoch": 0.8811687775709163,
      "grad_norm": 1.48258557844494,
      "learning_rate": 3.6551631786659103e-07,
      "loss": 0.7346,
      "step": 12010
    },
    {
      "epoch": 0.8812421471610554,
      "grad_norm": 0.5488357667374133,
      "learning_rate": 3.650704988779313e-07,
      "loss": 0.4766,
      "step": 12011
    },
    {
      "epoch": 0.8813155167511946,
      "grad_norm": 1.6619474212044083,
      "learning_rate": 3.6462494163521936e-07,
      "loss": 0.8605,
      "step": 12012
    },
    {
      "epoch": 0.8813888863413337,
      "grad_norm": 1.4497744095460017,
      "learning_rate": 3.6417964616361543e-07,
      "loss": 0.7437,
      "step": 12013
    },
    {
      "epoch": 0.8814622559314728,
      "grad_norm": 1.5658374623008215,
      "learning_rate": 3.637346124882679e-07,
      "loss": 0.7755,
      "step": 12014
    },
    {
      "epoch": 0.8815356255216119,
      "grad_norm": 1.5202254135052167,
      "learning_rate": 3.6328984063430716e-07,
      "loss": 0.7533,
      "step": 12015
    },
    {
      "epoch": 0.8816089951117511,
      "grad_norm": 1.458290572369651,
      "learning_rate": 3.6284533062685314e-07,
      "loss": 0.6602,
      "step": 12016
    },
    {
      "epoch": 0.8816089951117511,
      "lm_loss": 0.4879,
      "step": 12016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8816089951117511,
      "lm_loss": 0.4548,
      "step": 12016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8816089951117511,
      "lm_loss": 0.455,
      "step": 12016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8816089951117511,
      "lm_loss": 0.4588,
      "step": 12016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8816089951117511,
      "lm_loss": 0.3698,
      "step": 12016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8816089951117511,
      "lm_loss": 0.3913,
      "step": 12016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8816089951117511,
      "lm_loss": 0.461,
      "step": 12016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8816089951117511,
      "lm_loss": 0.4567,
      "step": 12016,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8816823647018902,
      "grad_norm": 0.549684038160837,
      "learning_rate": 3.6240108249100623e-07,
      "loss": 0.463,
      "step": 12017
    },
    {
      "epoch": 0.8817557342920294,
      "grad_norm": 1.60102457298628,
      "learning_rate": 3.619570962518554e-07,
      "loss": 0.7483,
      "step": 12018
    },
    {
      "epoch": 0.8818291038821684,
      "grad_norm": 0.5446340004911447,
      "learning_rate": 3.6151337193447323e-07,
      "loss": 0.4715,
      "step": 12019
    },
    {
      "epoch": 0.8819024734723075,
      "grad_norm": 0.5209009614677874,
      "learning_rate": 3.61069909563917e-07,
      "loss": 0.4737,
      "step": 12020
    },
    {
      "epoch": 0.8819758430624467,
      "grad_norm": 0.5604295775934757,
      "learning_rate": 3.6062670916523214e-07,
      "loss": 0.4866,
      "step": 12021
    },
    {
      "epoch": 0.8820492126525858,
      "grad_norm": 0.5285715562984623,
      "learning_rate": 3.6018377076344647e-07,
      "loss": 0.4665,
      "step": 12022
    },
    {
      "epoch": 0.882122582242725,
      "grad_norm": 1.7361033714274723,
      "learning_rate": 3.597410943835733e-07,
      "loss": 0.7973,
      "step": 12023
    },
    {
      "epoch": 0.882195951832864,
      "grad_norm": 1.4408398119311108,
      "learning_rate": 3.592986800506115e-07,
      "loss": 0.7118,
      "step": 12024
    },
    {
      "epoch": 0.882195951832864,
      "lm_loss": 0.5511,
      "step": 12024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.882195951832864,
      "lm_loss": 0.4789,
      "step": 12024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.882195951832864,
      "lm_loss": 0.4836,
      "step": 12024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.882195951832864,
      "lm_loss": 0.5317,
      "step": 12024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.882195951832864,
      "lm_loss": 0.49,
      "step": 12024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.882195951832864,
      "lm_loss": 0.5282,
      "step": 12024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.882195951832864,
      "lm_loss": 0.4487,
      "step": 12024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.882195951832864,
      "lm_loss": 0.4701,
      "step": 12024,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8822693214230032,
      "grad_norm": 0.5447290677723361,
      "learning_rate": 3.5885652778954617e-07,
      "loss": 0.4692,
      "step": 12025
    },
    {
      "epoch": 0.8823426910131423,
      "grad_norm": 1.6239361961392107,
      "learning_rate": 3.5841463762534556e-07,
      "loss": 0.7828,
      "step": 12026
    },
    {
      "epoch": 0.8824160606032815,
      "grad_norm": 0.5260228359794614,
      "learning_rate": 3.579730095829659e-07,
      "loss": 0.4633,
      "step": 12027
    },
    {
      "epoch": 0.8824894301934206,
      "grad_norm": 1.427383671290123,
      "learning_rate": 3.5753164368734604e-07,
      "loss": 0.6945,
      "step": 12028
    },
    {
      "epoch": 0.8825627997835597,
      "grad_norm": 0.5589894506686461,
      "learning_rate": 3.570905399634111e-07,
      "loss": 0.4868,
      "step": 12029
    },
    {
      "epoch": 0.8826361693736988,
      "grad_norm": 1.4878586432002465,
      "learning_rate": 3.566496984360712e-07,
      "loss": 0.7827,
      "step": 12030
    },
    {
      "epoch": 0.882709538963838,
      "grad_norm": 0.5458955477261866,
      "learning_rate": 3.562091191302214e-07,
      "loss": 0.4833,
      "step": 12031
    },
    {
      "epoch": 0.8827829085539771,
      "grad_norm": 1.5261612089673435,
      "learning_rate": 3.5576880207074336e-07,
      "loss": 0.7439,
      "step": 12032
    },
    {
      "epoch": 0.8827829085539771,
      "lm_loss": 1.0862,
      "step": 12032,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.8827829085539771,
      "lm_loss": 0.2874,
      "step": 12032,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.8827829085539771,
      "lm_loss": 0.4009,
      "step": 12032,
      "vm_loss": 0.1369
    },
    {
      "epoch": 0.8827829085539771,
      "lm_loss": 0.1107,
      "step": 12032,
      "vm_loss": 0.0423
    },
    {
      "epoch": 0.8827829085539771,
      "lm_loss": 0.2439,
      "step": 12032,
      "vm_loss": 0.0843
    },
    {
      "epoch": 0.8827829085539771,
      "lm_loss": 0.0002,
      "step": 12032,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.8827829085539771,
      "lm_loss": 0.332,
      "step": 12032,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.8827829085539771,
      "lm_loss": 0.1223,
      "step": 12032,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.8828562781441163,
      "grad_norm": 1.459866406860275,
      "learning_rate": 3.553287472825023e-07,
      "loss": 0.6576,
      "step": 12033
    },
    {
      "epoch": 0.8829296477342553,
      "grad_norm": 1.632861460722167,
      "learning_rate": 3.5488895479034944e-07,
      "loss": 0.7836,
      "step": 12034
    },
    {
      "epoch": 0.8830030173243945,
      "grad_norm": 1.4347348698731697,
      "learning_rate": 3.5444942461912044e-07,
      "loss": 0.7216,
      "step": 12035
    },
    {
      "epoch": 0.8830763869145336,
      "grad_norm": 0.5529787930651262,
      "learning_rate": 3.540101567936366e-07,
      "loss": 0.49,
      "step": 12036
    },
    {
      "epoch": 0.8831497565046728,
      "grad_norm": 1.508684234711913,
      "learning_rate": 3.5357115133870524e-07,
      "loss": 0.738,
      "step": 12037
    },
    {
      "epoch": 0.8832231260948119,
      "grad_norm": 1.6828601165840087,
      "learning_rate": 3.5313240827911767e-07,
      "loss": 0.7115,
      "step": 12038
    },
    {
      "epoch": 0.8832964956849509,
      "grad_norm": 1.7505753855439177,
      "learning_rate": 3.526939276396507e-07,
      "loss": 0.6706,
      "step": 12039
    },
    {
      "epoch": 0.8833698652750901,
      "grad_norm": 1.4446066980466985,
      "learning_rate": 3.522557094450668e-07,
      "loss": 0.6521,
      "step": 12040
    },
    {
      "epoch": 0.8833698652750901,
      "lm_loss": 0.0369,
      "step": 12040,
      "vm_loss": 0.0835
    },
    {
      "epoch": 0.8833698652750901,
      "lm_loss": 0.1642,
      "step": 12040,
      "vm_loss": 0.1913
    },
    {
      "epoch": 0.8833698652750901,
      "lm_loss": 0.064,
      "step": 12040,
      "vm_loss": 0.1539
    },
    {
      "epoch": 0.8833698652750901,
      "lm_loss": 0.1835,
      "step": 12040,
      "vm_loss": 0.064
    },
    {
      "epoch": 0.8833698652750901,
      "lm_loss": 0.0084,
      "step": 12040,
      "vm_loss": 0.1443
    },
    {
      "epoch": 0.8833698652750901,
      "lm_loss": 0.5609,
      "step": 12040,
      "vm_loss": 0.1224
    },
    {
      "epoch": 0.8833698652750901,
      "lm_loss": 0.9013,
      "step": 12040,
      "vm_loss": 0.1893
    },
    {
      "epoch": 0.8833698652750901,
      "lm_loss": 0.66,
      "step": 12040,
      "vm_loss": 0.081
    },
    {
      "epoch": 0.8834432348652292,
      "grad_norm": 1.6072119591445073,
      "learning_rate": 3.5181775372011283e-07,
      "loss": 0.6926,
      "step": 12041
    },
    {
      "epoch": 0.8835166044553684,
      "grad_norm": 1.649839032401289,
      "learning_rate": 3.513800604895207e-07,
      "loss": 0.7771,
      "step": 12042
    },
    {
      "epoch": 0.8835899740455075,
      "grad_norm": 0.5556991623885389,
      "learning_rate": 3.5094262977800887e-07,
      "loss": 0.475,
      "step": 12043
    },
    {
      "epoch": 0.8836633436356466,
      "grad_norm": 1.583786156862942,
      "learning_rate": 3.5050546161028044e-07,
      "loss": 0.7328,
      "step": 12044
    },
    {
      "epoch": 0.8837367132257857,
      "grad_norm": 1.5081971477496687,
      "learning_rate": 3.500685560110229e-07,
      "loss": 0.6854,
      "step": 12045
    },
    {
      "epoch": 0.8838100828159249,
      "grad_norm": 1.7296993382302286,
      "learning_rate": 3.496319130049092e-07,
      "loss": 0.6907,
      "step": 12046
    },
    {
      "epoch": 0.883883452406064,
      "grad_norm": 1.5106764951006386,
      "learning_rate": 3.491955326165969e-07,
      "loss": 0.7715,
      "step": 12047
    },
    {
      "epoch": 0.8839568219962032,
      "grad_norm": 1.4640454964816823,
      "learning_rate": 3.487594148707313e-07,
      "loss": 0.7096,
      "step": 12048
    },
    {
      "epoch": 0.8839568219962032,
      "lm_loss": 0.4291,
      "step": 12048,
      "vm_loss": 0.0646
    },
    {
      "epoch": 0.8839568219962032,
      "lm_loss": 0.8815,
      "step": 12048,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.8839568219962032,
      "lm_loss": 0.4213,
      "step": 12048,
      "vm_loss": 0.3282
    },
    {
      "epoch": 0.8839568219962032,
      "lm_loss": 0.6993,
      "step": 12048,
      "vm_loss": 0.1852
    },
    {
      "epoch": 0.8839568219962032,
      "lm_loss": 0.1801,
      "step": 12048,
      "vm_loss": 0.1211
    },
    {
      "epoch": 0.8839568219962032,
      "lm_loss": 0.107,
      "step": 12048,
      "vm_loss": 0.1124
    },
    {
      "epoch": 0.8839568219962032,
      "lm_loss": 0.8763,
      "step": 12048,
      "vm_loss": 0.1448
    },
    {
      "epoch": 0.8839568219962032,
      "lm_loss": 1.0885,
      "step": 12048,
      "vm_loss": 0.1325
    },
    {
      "epoch": 0.8840301915863422,
      "grad_norm": 1.5872971831986178,
      "learning_rate": 3.483235597919404e-07,
      "loss": 0.7807,
      "step": 12049
    },
    {
      "epoch": 0.8841035611764814,
      "grad_norm": 1.6396279004957355,
      "learning_rate": 3.4788796740483745e-07,
      "loss": 0.6775,
      "step": 12050
    },
    {
      "epoch": 0.8841769307666205,
      "grad_norm": 1.5413462795047144,
      "learning_rate": 3.474526377340215e-07,
      "loss": 0.6674,
      "step": 12051
    },
    {
      "epoch": 0.8842503003567597,
      "grad_norm": 1.491285344975742,
      "learning_rate": 3.470175708040774e-07,
      "loss": 0.7229,
      "step": 12052
    },
    {
      "epoch": 0.8843236699468987,
      "grad_norm": 0.5582356709328002,
      "learning_rate": 3.4658276663957377e-07,
      "loss": 0.4673,
      "step": 12053
    },
    {
      "epoch": 0.8843970395370379,
      "grad_norm": 0.5280867266367056,
      "learning_rate": 3.4614822526506555e-07,
      "loss": 0.4682,
      "step": 12054
    },
    {
      "epoch": 0.884470409127177,
      "grad_norm": 0.5457840737069439,
      "learning_rate": 3.4571394670509184e-07,
      "loss": 0.4691,
      "step": 12055
    },
    {
      "epoch": 0.8845437787173162,
      "grad_norm": 0.5432308878857213,
      "learning_rate": 3.452799309841781e-07,
      "loss": 0.4844,
      "step": 12056
    },
    {
      "epoch": 0.8845437787173162,
      "lm_loss": 1.5049,
      "step": 12056,
      "vm_loss": 0.049
    },
    {
      "epoch": 0.8845437787173162,
      "lm_loss": 1.1174,
      "step": 12056,
      "vm_loss": 0.0493
    },
    {
      "epoch": 0.8845437787173162,
      "lm_loss": 0.0386,
      "step": 12056,
      "vm_loss": 0.1195
    },
    {
      "epoch": 0.8845437787173162,
      "lm_loss": 0.9954,
      "step": 12056,
      "vm_loss": 0.1116
    },
    {
      "epoch": 0.8845437787173162,
      "lm_loss": 0.6964,
      "step": 12056,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.8845437787173162,
      "lm_loss": 0.175,
      "step": 12056,
      "vm_loss": 0.1532
    },
    {
      "epoch": 0.8845437787173162,
      "lm_loss": 0.0897,
      "step": 12056,
      "vm_loss": 0.0547
    },
    {
      "epoch": 0.8845437787173162,
      "lm_loss": 1.0117,
      "step": 12056,
      "vm_loss": 0.2068
    },
    {
      "epoch": 0.8846171483074553,
      "grad_norm": 1.615810959223559,
      "learning_rate": 3.448461781268325e-07,
      "loss": 0.7172,
      "step": 12057
    },
    {
      "epoch": 0.8846905178975943,
      "grad_norm": 0.5581309130279558,
      "learning_rate": 3.444126881575527e-07,
      "loss": 0.4915,
      "step": 12058
    },
    {
      "epoch": 0.8847638874877335,
      "grad_norm": 0.5740308264777223,
      "learning_rate": 3.4397946110081793e-07,
      "loss": 0.4844,
      "step": 12059
    },
    {
      "epoch": 0.8848372570778726,
      "grad_norm": 0.5622715887029174,
      "learning_rate": 3.435464969810931e-07,
      "loss": 0.4896,
      "step": 12060
    },
    {
      "epoch": 0.8849106266680118,
      "grad_norm": 0.5690309364208298,
      "learning_rate": 3.431137958228292e-07,
      "loss": 0.5099,
      "step": 12061
    },
    {
      "epoch": 0.8849839962581509,
      "grad_norm": 1.4561027166537832,
      "learning_rate": 3.4268135765046063e-07,
      "loss": 0.6988,
      "step": 12062
    },
    {
      "epoch": 0.88505736584829,
      "grad_norm": 1.37083515922318,
      "learning_rate": 3.4224918248841056e-07,
      "loss": 0.6641,
      "step": 12063
    },
    {
      "epoch": 0.8851307354384291,
      "grad_norm": 1.6064496549000842,
      "learning_rate": 3.418172703610839e-07,
      "loss": 0.745,
      "step": 12064
    },
    {
      "epoch": 0.8851307354384291,
      "lm_loss": 1.1741,
      "step": 12064,
      "vm_loss": 0.0343
    },
    {
      "epoch": 0.8851307354384291,
      "lm_loss": 0.7803,
      "step": 12064,
      "vm_loss": 0.1855
    },
    {
      "epoch": 0.8851307354384291,
      "lm_loss": 0.155,
      "step": 12064,
      "vm_loss": 0.0445
    },
    {
      "epoch": 0.8851307354384291,
      "lm_loss": 0.539,
      "step": 12064,
      "vm_loss": 0.0959
    },
    {
      "epoch": 0.8851307354384291,
      "lm_loss": 0.5042,
      "step": 12064,
      "vm_loss": 0.1387
    },
    {
      "epoch": 0.8851307354384291,
      "lm_loss": 0.5907,
      "step": 12064,
      "vm_loss": 0.0413
    },
    {
      "epoch": 0.8851307354384291,
      "lm_loss": 0.8543,
      "step": 12064,
      "vm_loss": 0.1549
    },
    {
      "epoch": 0.8851307354384291,
      "lm_loss": 1.1454,
      "step": 12064,
      "vm_loss": 0.2234
    },
    {
      "epoch": 0.8852041050285683,
      "grad_norm": 1.4809474625411077,
      "learning_rate": 3.4138562129287235e-07,
      "loss": 0.739,
      "step": 12065
    },
    {
      "epoch": 0.8852774746187074,
      "grad_norm": 1.692272104025732,
      "learning_rate": 3.4095423530815075e-07,
      "loss": 0.7564,
      "step": 12066
    },
    {
      "epoch": 0.8853508442088466,
      "grad_norm": 1.4997739424138075,
      "learning_rate": 3.405231124312819e-07,
      "loss": 0.6624,
      "step": 12067
    },
    {
      "epoch": 0.8854242137989856,
      "grad_norm": 1.4715804944613358,
      "learning_rate": 3.400922526866107e-07,
      "loss": 0.6994,
      "step": 12068
    },
    {
      "epoch": 0.8854975833891248,
      "grad_norm": 1.5561891845307205,
      "learning_rate": 3.396616560984711e-07,
      "loss": 0.7731,
      "step": 12069
    },
    {
      "epoch": 0.8855709529792639,
      "grad_norm": 0.5352382888017782,
      "learning_rate": 3.392313226911792e-07,
      "loss": 0.4744,
      "step": 12070
    },
    {
      "epoch": 0.8856443225694031,
      "grad_norm": 1.8543045600233847,
      "learning_rate": 3.388012524890366e-07,
      "loss": 0.8028,
      "step": 12071
    },
    {
      "epoch": 0.8857176921595422,
      "grad_norm": 1.495744059052279,
      "learning_rate": 3.3837144551633004e-07,
      "loss": 0.7862,
      "step": 12072
    },
    {
      "epoch": 0.8857176921595422,
      "lm_loss": 0.6254,
      "step": 12072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8857176921595422,
      "lm_loss": 0.4527,
      "step": 12072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8857176921595422,
      "lm_loss": 0.6643,
      "step": 12072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8857176921595422,
      "lm_loss": 0.4875,
      "step": 12072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8857176921595422,
      "lm_loss": 0.4789,
      "step": 12072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8857176921595422,
      "lm_loss": 0.4197,
      "step": 12072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8857176921595422,
      "lm_loss": 0.4015,
      "step": 12072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8857176921595422,
      "lm_loss": 0.425,
      "step": 12072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8857910617496813,
      "grad_norm": 0.554141714144028,
      "learning_rate": 3.379419017973318e-07,
      "loss": 0.4837,
      "step": 12073
    },
    {
      "epoch": 0.8858644313398204,
      "grad_norm": 0.5432084088952314,
      "learning_rate": 3.375126213563007e-07,
      "loss": 0.4688,
      "step": 12074
    },
    {
      "epoch": 0.8859378009299596,
      "grad_norm": 1.4505013375138007,
      "learning_rate": 3.37083604217478e-07,
      "loss": 0.6267,
      "step": 12075
    },
    {
      "epoch": 0.8860111705200987,
      "grad_norm": 0.5565882117684763,
      "learning_rate": 3.3665485040509203e-07,
      "loss": 0.4837,
      "step": 12076
    },
    {
      "epoch": 0.8860845401102379,
      "grad_norm": 1.483849128596217,
      "learning_rate": 3.3622635994335517e-07,
      "loss": 0.6316,
      "step": 12077
    },
    {
      "epoch": 0.8861579097003769,
      "grad_norm": 1.5057686101309873,
      "learning_rate": 3.357981328564647e-07,
      "loss": 0.6457,
      "step": 12078
    },
    {
      "epoch": 0.886231279290516,
      "grad_norm": 1.6440293371076524,
      "learning_rate": 3.3537016916860455e-07,
      "loss": 0.8322,
      "step": 12079
    },
    {
      "epoch": 0.8863046488806552,
      "grad_norm": 1.5556969564996068,
      "learning_rate": 3.349424689039432e-07,
      "loss": 0.7713,
      "step": 12080
    },
    {
      "epoch": 0.8863046488806552,
      "lm_loss": 1.5993,
      "step": 12080,
      "vm_loss": 0.0809
    },
    {
      "epoch": 0.8863046488806552,
      "lm_loss": 0.0046,
      "step": 12080,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.8863046488806552,
      "lm_loss": 0.8849,
      "step": 12080,
      "vm_loss": 0.192
    },
    {
      "epoch": 0.8863046488806552,
      "lm_loss": 1.0865,
      "step": 12080,
      "vm_loss": 0.2291
    },
    {
      "epoch": 0.8863046488806552,
      "lm_loss": 0.0525,
      "step": 12080,
      "vm_loss": 0.0927
    },
    {
      "epoch": 0.8863046488806552,
      "lm_loss": 0.833,
      "step": 12080,
      "vm_loss": 0.151
    },
    {
      "epoch": 0.8863046488806552,
      "lm_loss": 0.2189,
      "step": 12080,
      "vm_loss": 0.1278
    },
    {
      "epoch": 0.8863046488806552,
      "lm_loss": 0.1439,
      "step": 12080,
      "vm_loss": 0.1011
    },
    {
      "epoch": 0.8863780184707943,
      "grad_norm": 1.6340915515076675,
      "learning_rate": 3.345150320866336e-07,
      "loss": 0.754,
      "step": 12081
    },
    {
      "epoch": 0.8864513880609335,
      "grad_norm": 1.4974116019240293,
      "learning_rate": 3.340878587408136e-07,
      "loss": 0.6885,
      "step": 12082
    },
    {
      "epoch": 0.8865247576510725,
      "grad_norm": 1.5885616278309869,
      "learning_rate": 3.3366094889060727e-07,
      "loss": 0.792,
      "step": 12083
    },
    {
      "epoch": 0.8865981272412117,
      "grad_norm": 0.5464891193600409,
      "learning_rate": 3.332343025601226e-07,
      "loss": 0.4822,
      "step": 12084
    },
    {
      "epoch": 0.8866714968313508,
      "grad_norm": 1.5749765679163599,
      "learning_rate": 3.3280791977345463e-07,
      "loss": 0.6569,
      "step": 12085
    },
    {
      "epoch": 0.88674486642149,
      "grad_norm": 1.6937413430271473,
      "learning_rate": 3.323818005546814e-07,
      "loss": 0.7299,
      "step": 12086
    },
    {
      "epoch": 0.8868182360116291,
      "grad_norm": 0.5296435097641329,
      "learning_rate": 3.3195594492786696e-07,
      "loss": 0.483,
      "step": 12087
    },
    {
      "epoch": 0.8868916056017682,
      "grad_norm": 1.4919648437750772,
      "learning_rate": 3.3153035291706104e-07,
      "loss": 0.8153,
      "step": 12088
    },
    {
      "epoch": 0.8868916056017682,
      "lm_loss": 1.2898,
      "step": 12088,
      "vm_loss": 0.0366
    },
    {
      "epoch": 0.8868916056017682,
      "lm_loss": 1.294,
      "step": 12088,
      "vm_loss": 0.1205
    },
    {
      "epoch": 0.8868916056017682,
      "lm_loss": 1.7336,
      "step": 12088,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.8868916056017682,
      "lm_loss": 0.8514,
      "step": 12088,
      "vm_loss": 0.1286
    },
    {
      "epoch": 0.8868916056017682,
      "lm_loss": 0.3321,
      "step": 12088,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.8868916056017682,
      "lm_loss": 0.3108,
      "step": 12088,
      "vm_loss": 0.0822
    },
    {
      "epoch": 0.8868916056017682,
      "lm_loss": 1.0261,
      "step": 12088,
      "vm_loss": 0.1996
    },
    {
      "epoch": 0.8868916056017682,
      "lm_loss": 0.2403,
      "step": 12088,
      "vm_loss": 0.0773
    },
    {
      "epoch": 0.8869649751919073,
      "grad_norm": 1.4647663846237382,
      "learning_rate": 3.31105024546296e-07,
      "loss": 0.7635,
      "step": 12089
    },
    {
      "epoch": 0.8870383447820465,
      "grad_norm": 1.539491991639178,
      "learning_rate": 3.306799598395938e-07,
      "loss": 0.834,
      "step": 12090
    },
    {
      "epoch": 0.8871117143721856,
      "grad_norm": 1.5968551216477342,
      "learning_rate": 3.3025515882095736e-07,
      "loss": 0.7327,
      "step": 12091
    },
    {
      "epoch": 0.8871850839623248,
      "grad_norm": 1.5496987618188998,
      "learning_rate": 3.298306215143765e-07,
      "loss": 0.6872,
      "step": 12092
    },
    {
      "epoch": 0.8872584535524638,
      "grad_norm": 1.5187751915665977,
      "learning_rate": 3.2940634794382586e-07,
      "loss": 0.7115,
      "step": 12093
    },
    {
      "epoch": 0.887331823142603,
      "grad_norm": 1.4370137673576702,
      "learning_rate": 3.2898233813326574e-07,
      "loss": 0.7074,
      "step": 12094
    },
    {
      "epoch": 0.8874051927327421,
      "grad_norm": 1.504515136688789,
      "learning_rate": 3.285585921066392e-07,
      "loss": 0.8088,
      "step": 12095
    },
    {
      "epoch": 0.8874785623228812,
      "grad_norm": 0.5501665134982978,
      "learning_rate": 3.281351098878788e-07,
      "loss": 0.4684,
      "step": 12096
    },
    {
      "epoch": 0.8874785623228812,
      "lm_loss": 0.4016,
      "step": 12096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8874785623228812,
      "lm_loss": 0.461,
      "step": 12096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8874785623228812,
      "lm_loss": 0.3915,
      "step": 12096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8874785623228812,
      "lm_loss": 0.5036,
      "step": 12096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8874785623228812,
      "lm_loss": 0.2696,
      "step": 12096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8874785623228812,
      "lm_loss": 0.5013,
      "step": 12096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8874785623228812,
      "lm_loss": 0.5273,
      "step": 12096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8874785623228812,
      "lm_loss": 0.4335,
      "step": 12096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8875519319130204,
      "grad_norm": 0.538030193843356,
      "learning_rate": 3.277118915008981e-07,
      "loss": 0.462,
      "step": 12097
    },
    {
      "epoch": 0.8876253015031594,
      "grad_norm": 1.4866160616763164,
      "learning_rate": 3.2728893696959805e-07,
      "loss": 0.6428,
      "step": 12098
    },
    {
      "epoch": 0.8876986710932986,
      "grad_norm": 0.5402189838213597,
      "learning_rate": 3.268662463178629e-07,
      "loss": 0.4776,
      "step": 12099
    },
    {
      "epoch": 0.8877720406834377,
      "grad_norm": 1.5054061469871878,
      "learning_rate": 3.264438195695635e-07,
      "loss": 0.7456,
      "step": 12100
    },
    {
      "epoch": 0.8878454102735769,
      "grad_norm": 1.6922470591567627,
      "learning_rate": 3.260216567485558e-07,
      "loss": 0.873,
      "step": 12101
    },
    {
      "epoch": 0.887918779863716,
      "grad_norm": 0.5530233819744889,
      "learning_rate": 3.2559975787868013e-07,
      "loss": 0.4809,
      "step": 12102
    },
    {
      "epoch": 0.8879921494538551,
      "grad_norm": 1.5643123440546853,
      "learning_rate": 3.251781229837619e-07,
      "loss": 0.7584,
      "step": 12103
    },
    {
      "epoch": 0.8880655190439942,
      "grad_norm": 1.522969048429209,
      "learning_rate": 3.247567520876127e-07,
      "loss": 0.7633,
      "step": 12104
    },
    {
      "epoch": 0.8880655190439942,
      "lm_loss": 1.272,
      "step": 12104,
      "vm_loss": 0.0355
    },
    {
      "epoch": 0.8880655190439942,
      "lm_loss": 0.8494,
      "step": 12104,
      "vm_loss": 0.1845
    },
    {
      "epoch": 0.8880655190439942,
      "lm_loss": 0.4049,
      "step": 12104,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.8880655190439942,
      "lm_loss": 0.7378,
      "step": 12104,
      "vm_loss": 0.0914
    },
    {
      "epoch": 0.8880655190439942,
      "lm_loss": 0.5293,
      "step": 12104,
      "vm_loss": 0.0873
    },
    {
      "epoch": 0.8880655190439942,
      "lm_loss": 0.0349,
      "step": 12104,
      "vm_loss": 0.0745
    },
    {
      "epoch": 0.8880655190439942,
      "lm_loss": 0.8182,
      "step": 12104,
      "vm_loss": 0.115
    },
    {
      "epoch": 0.8880655190439942,
      "lm_loss": 0.9143,
      "step": 12104,
      "vm_loss": 0.2026
    },
    {
      "epoch": 0.8881388886341334,
      "grad_norm": 1.7132825927760855,
      "learning_rate": 3.2433564521402673e-07,
      "loss": 0.7369,
      "step": 12105
    },
    {
      "epoch": 0.8882122582242725,
      "grad_norm": 1.2924818623849383,
      "learning_rate": 3.2391480238678605e-07,
      "loss": 0.7287,
      "step": 12106
    },
    {
      "epoch": 0.8882856278144117,
      "grad_norm": 1.6175449030144207,
      "learning_rate": 3.234942236296573e-07,
      "loss": 0.7854,
      "step": 12107
    },
    {
      "epoch": 0.8883589974045507,
      "grad_norm": 0.5611500141297356,
      "learning_rate": 3.230739089663909e-07,
      "loss": 0.4979,
      "step": 12108
    },
    {
      "epoch": 0.8884323669946899,
      "grad_norm": 1.5933645539077874,
      "learning_rate": 3.226538584207228e-07,
      "loss": 0.7173,
      "step": 12109
    },
    {
      "epoch": 0.888505736584829,
      "grad_norm": 0.5484803542061305,
      "learning_rate": 3.222340720163747e-07,
      "loss": 0.4681,
      "step": 12110
    },
    {
      "epoch": 0.8885791061749682,
      "grad_norm": 1.5799432499455157,
      "learning_rate": 3.218145497770525e-07,
      "loss": 0.8002,
      "step": 12111
    },
    {
      "epoch": 0.8886524757651073,
      "grad_norm": 1.4362316162905675,
      "learning_rate": 3.213952917264479e-07,
      "loss": 0.6678,
      "step": 12112
    },
    {
      "epoch": 0.8886524757651073,
      "lm_loss": 0.8669,
      "step": 12112,
      "vm_loss": 0.0464
    },
    {
      "epoch": 0.8886524757651073,
      "lm_loss": 0.9754,
      "step": 12112,
      "vm_loss": 0.2537
    },
    {
      "epoch": 0.8886524757651073,
      "lm_loss": 0.2607,
      "step": 12112,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.8886524757651073,
      "lm_loss": 0.6126,
      "step": 12112,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.8886524757651073,
      "lm_loss": 0.6681,
      "step": 12112,
      "vm_loss": 0.0529
    },
    {
      "epoch": 0.8886524757651073,
      "lm_loss": 0.0832,
      "step": 12112,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.8886524757651073,
      "lm_loss": 0.4762,
      "step": 12112,
      "vm_loss": 0.1301
    },
    {
      "epoch": 0.8886524757651073,
      "lm_loss": 0.4421,
      "step": 12112,
      "vm_loss": 0.1991
    },
    {
      "epoch": 0.8887258453552463,
      "grad_norm": 1.4204476502532926,
      "learning_rate": 3.2097629788823805e-07,
      "loss": 0.6705,
      "step": 12113
    },
    {
      "epoch": 0.8887992149453855,
      "grad_norm": 0.5509921261228605,
      "learning_rate": 3.2055756828608456e-07,
      "loss": 0.4696,
      "step": 12114
    },
    {
      "epoch": 0.8888725845355246,
      "grad_norm": 1.3825395042279927,
      "learning_rate": 3.201391029436329e-07,
      "loss": 0.7563,
      "step": 12115
    },
    {
      "epoch": 0.8889459541256638,
      "grad_norm": 1.451754171955099,
      "learning_rate": 3.197209018845149e-07,
      "loss": 0.7848,
      "step": 12116
    },
    {
      "epoch": 0.889019323715803,
      "grad_norm": 1.7697725744825772,
      "learning_rate": 3.193029651323487e-07,
      "loss": 0.6898,
      "step": 12117
    },
    {
      "epoch": 0.889092693305942,
      "grad_norm": 1.5490042417487482,
      "learning_rate": 3.1888529271073607e-07,
      "loss": 0.7698,
      "step": 12118
    },
    {
      "epoch": 0.8891660628960811,
      "grad_norm": 1.644416483566091,
      "learning_rate": 3.1846788464326315e-07,
      "loss": 0.8291,
      "step": 12119
    },
    {
      "epoch": 0.8892394324862203,
      "grad_norm": 1.703586136928932,
      "learning_rate": 3.1805074095350277e-07,
      "loss": 0.6919,
      "step": 12120
    },
    {
      "epoch": 0.8892394324862203,
      "lm_loss": 0.4406,
      "step": 12120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8892394324862203,
      "lm_loss": 0.4598,
      "step": 12120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8892394324862203,
      "lm_loss": 0.4657,
      "step": 12120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8892394324862203,
      "lm_loss": 0.5259,
      "step": 12120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8892394324862203,
      "lm_loss": 0.4545,
      "step": 12120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8892394324862203,
      "lm_loss": 0.3595,
      "step": 12120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8892394324862203,
      "lm_loss": 0.326,
      "step": 12120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8892394324862203,
      "lm_loss": 0.4533,
      "step": 12120,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8893128020763594,
      "grad_norm": 0.5583025877123359,
      "learning_rate": 3.1763386166501155e-07,
      "loss": 0.4618,
      "step": 12121
    },
    {
      "epoch": 0.8893861716664986,
      "grad_norm": 1.3994266717496544,
      "learning_rate": 3.172172468013313e-07,
      "loss": 0.7479,
      "step": 12122
    },
    {
      "epoch": 0.8894595412566376,
      "grad_norm": 1.5098618790131704,
      "learning_rate": 3.168008963859903e-07,
      "loss": 0.7366,
      "step": 12123
    },
    {
      "epoch": 0.8895329108467768,
      "grad_norm": 1.454489885833645,
      "learning_rate": 3.1638481044250045e-07,
      "loss": 0.8154,
      "step": 12124
    },
    {
      "epoch": 0.8896062804369159,
      "grad_norm": 1.5309651983788628,
      "learning_rate": 3.15968988994359e-07,
      "loss": 0.7921,
      "step": 12125
    },
    {
      "epoch": 0.8896796500270551,
      "grad_norm": 1.5828743896791049,
      "learning_rate": 3.1555343206504885e-07,
      "loss": 0.6942,
      "step": 12126
    },
    {
      "epoch": 0.8897530196171942,
      "grad_norm": 0.5380751221523955,
      "learning_rate": 3.151381396780362e-07,
      "loss": 0.4624,
      "step": 12127
    },
    {
      "epoch": 0.8898263892073333,
      "grad_norm": 1.686263104511027,
      "learning_rate": 3.1472311185677564e-07,
      "loss": 0.7228,
      "step": 12128
    },
    {
      "epoch": 0.8898263892073333,
      "lm_loss": 0.3506,
      "step": 12128,
      "vm_loss": 0.0682
    },
    {
      "epoch": 0.8898263892073333,
      "lm_loss": 0.6638,
      "step": 12128,
      "vm_loss": 0.1361
    },
    {
      "epoch": 0.8898263892073333,
      "lm_loss": 0.0189,
      "step": 12128,
      "vm_loss": 0.0753
    },
    {
      "epoch": 0.8898263892073333,
      "lm_loss": 0.0604,
      "step": 12128,
      "vm_loss": 0.1012
    },
    {
      "epoch": 0.8898263892073333,
      "lm_loss": 1.409,
      "step": 12128,
      "vm_loss": 0.2427
    },
    {
      "epoch": 0.8898263892073333,
      "lm_loss": 0.2108,
      "step": 12128,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.8898263892073333,
      "lm_loss": 0.9705,
      "step": 12128,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.8898263892073333,
      "lm_loss": 0.2798,
      "step": 12128,
      "vm_loss": 0.1138
    },
    {
      "epoch": 0.8898997587974724,
      "grad_norm": 1.3565034501543243,
      "learning_rate": 3.1430834862470395e-07,
      "loss": 0.672,
      "step": 12129
    },
    {
      "epoch": 0.8899731283876116,
      "grad_norm": 0.5488073916134657,
      "learning_rate": 3.138938500052435e-07,
      "loss": 0.4744,
      "step": 12130
    },
    {
      "epoch": 0.8900464979777507,
      "grad_norm": 0.5224347620622634,
      "learning_rate": 3.134796160218023e-07,
      "loss": 0.4639,
      "step": 12131
    },
    {
      "epoch": 0.8901198675678897,
      "grad_norm": 1.5005552702564129,
      "learning_rate": 3.1306564669777215e-07,
      "loss": 0.6769,
      "step": 12132
    },
    {
      "epoch": 0.8901932371580289,
      "grad_norm": 0.557570189422005,
      "learning_rate": 3.1265194205653324e-07,
      "loss": 0.4804,
      "step": 12133
    },
    {
      "epoch": 0.890266606748168,
      "grad_norm": 1.5423835929251055,
      "learning_rate": 3.122385021214469e-07,
      "loss": 0.7912,
      "step": 12134
    },
    {
      "epoch": 0.8903399763383072,
      "grad_norm": 1.381909995809366,
      "learning_rate": 3.1182532691586167e-07,
      "loss": 0.6741,
      "step": 12135
    },
    {
      "epoch": 0.8904133459284463,
      "grad_norm": 0.5500204813294067,
      "learning_rate": 3.1141241646310995e-07,
      "loss": 0.4826,
      "step": 12136
    },
    {
      "epoch": 0.8904133459284463,
      "lm_loss": 0.4206,
      "step": 12136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8904133459284463,
      "lm_loss": 0.495,
      "step": 12136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8904133459284463,
      "lm_loss": 0.3291,
      "step": 12136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8904133459284463,
      "lm_loss": 0.3873,
      "step": 12136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8904133459284463,
      "lm_loss": 0.3441,
      "step": 12136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8904133459284463,
      "lm_loss": 0.5894,
      "step": 12136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8904133459284463,
      "lm_loss": 0.6003,
      "step": 12136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8904133459284463,
      "lm_loss": 0.5447,
      "step": 12136,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8904867155185854,
      "grad_norm": 0.5431095983139096,
      "learning_rate": 3.1099977078651036e-07,
      "loss": 0.4825,
      "step": 12137
    },
    {
      "epoch": 0.8905600851087245,
      "grad_norm": 0.5475347291291984,
      "learning_rate": 3.1058738990936476e-07,
      "loss": 0.4889,
      "step": 12138
    },
    {
      "epoch": 0.8906334546988637,
      "grad_norm": 0.5312809923491014,
      "learning_rate": 3.101752738549635e-07,
      "loss": 0.4937,
      "step": 12139
    },
    {
      "epoch": 0.8907068242890028,
      "grad_norm": 0.5397385476083688,
      "learning_rate": 3.0976342264657843e-07,
      "loss": 0.4649,
      "step": 12140
    },
    {
      "epoch": 0.890780193879142,
      "grad_norm": 1.618504004556669,
      "learning_rate": 3.0935183630746824e-07,
      "loss": 0.7887,
      "step": 12141
    },
    {
      "epoch": 0.890853563469281,
      "grad_norm": 0.5439658051924372,
      "learning_rate": 3.089405148608759e-07,
      "loss": 0.4684,
      "step": 12142
    },
    {
      "epoch": 0.8909269330594202,
      "grad_norm": 0.5437900963051238,
      "learning_rate": 3.085294583300291e-07,
      "loss": 0.4866,
      "step": 12143
    },
    {
      "epoch": 0.8910003026495593,
      "grad_norm": 0.5447221839442977,
      "learning_rate": 3.0811866673814306e-07,
      "loss": 0.4763,
      "step": 12144
    },
    {
      "epoch": 0.8910003026495593,
      "lm_loss": 0.6156,
      "step": 12144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8910003026495593,
      "lm_loss": 0.6323,
      "step": 12144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8910003026495593,
      "lm_loss": 0.4492,
      "step": 12144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8910003026495593,
      "lm_loss": 0.5636,
      "step": 12144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8910003026495593,
      "lm_loss": 0.4427,
      "step": 12144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8910003026495593,
      "lm_loss": 0.4776,
      "step": 12144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8910003026495593,
      "lm_loss": 0.5997,
      "step": 12144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8910003026495593,
      "lm_loss": 0.5201,
      "step": 12144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8910736722396985,
      "grad_norm": 0.5510668239581163,
      "learning_rate": 3.077081401084148e-07,
      "loss": 0.4899,
      "step": 12145
    },
    {
      "epoch": 0.8911470418298376,
      "grad_norm": 1.5720573840816263,
      "learning_rate": 3.0729787846402846e-07,
      "loss": 0.6423,
      "step": 12146
    },
    {
      "epoch": 0.8912204114199767,
      "grad_norm": 1.6309661935711808,
      "learning_rate": 3.0688788182815174e-07,
      "loss": 0.6808,
      "step": 12147
    },
    {
      "epoch": 0.8912937810101158,
      "grad_norm": 0.5326106036707933,
      "learning_rate": 3.064781502239389e-07,
      "loss": 0.4682,
      "step": 12148
    },
    {
      "epoch": 0.891367150600255,
      "grad_norm": 1.6876620082587557,
      "learning_rate": 3.0606868367452746e-07,
      "loss": 0.7066,
      "step": 12149
    },
    {
      "epoch": 0.8914405201903941,
      "grad_norm": 1.4544348162178808,
      "learning_rate": 3.056594822030429e-07,
      "loss": 0.7856,
      "step": 12150
    },
    {
      "epoch": 0.8915138897805333,
      "grad_norm": 0.5356140568026267,
      "learning_rate": 3.052505458325922e-07,
      "loss": 0.4623,
      "step": 12151
    },
    {
      "epoch": 0.8915872593706723,
      "grad_norm": 0.5399362234632364,
      "learning_rate": 3.0484187458626966e-07,
      "loss": 0.483,
      "step": 12152
    },
    {
      "epoch": 0.8915872593706723,
      "lm_loss": 1.4164,
      "step": 12152,
      "vm_loss": 0.0722
    },
    {
      "epoch": 0.8915872593706723,
      "lm_loss": 0.9068,
      "step": 12152,
      "vm_loss": 0.0913
    },
    {
      "epoch": 0.8915872593706723,
      "lm_loss": 0.3656,
      "step": 12152,
      "vm_loss": 0.1634
    },
    {
      "epoch": 0.8915872593706723,
      "lm_loss": 0.6738,
      "step": 12152,
      "vm_loss": 0.1716
    },
    {
      "epoch": 0.8915872593706723,
      "lm_loss": 0.7825,
      "step": 12152,
      "vm_loss": 0.1812
    },
    {
      "epoch": 0.8915872593706723,
      "lm_loss": 0.1381,
      "step": 12152,
      "vm_loss": 0.0528
    },
    {
      "epoch": 0.8915872593706723,
      "lm_loss": 0.6811,
      "step": 12152,
      "vm_loss": 0.0983
    },
    {
      "epoch": 0.8915872593706723,
      "lm_loss": 0.0694,
      "step": 12152,
      "vm_loss": 0.1183
    },
    {
      "epoch": 0.8916606289608114,
      "grad_norm": 1.7942602754765693,
      "learning_rate": 3.044334684871542e-07,
      "loss": 0.7798,
      "step": 12153
    },
    {
      "epoch": 0.8917339985509506,
      "grad_norm": 1.4078565822553992,
      "learning_rate": 3.0402532755830826e-07,
      "loss": 0.7188,
      "step": 12154
    },
    {
      "epoch": 0.8918073681410897,
      "grad_norm": 0.5339836075514733,
      "learning_rate": 3.036174518227819e-07,
      "loss": 0.4698,
      "step": 12155
    },
    {
      "epoch": 0.8918807377312289,
      "grad_norm": 1.589609297899701,
      "learning_rate": 3.032098413036083e-07,
      "loss": 0.7218,
      "step": 12156
    },
    {
      "epoch": 0.8919541073213679,
      "grad_norm": 1.6059814567101889,
      "learning_rate": 3.0280249602380685e-07,
      "loss": 0.7063,
      "step": 12157
    },
    {
      "epoch": 0.8920274769115071,
      "grad_norm": 1.4503707583508334,
      "learning_rate": 3.023954160063808e-07,
      "loss": 0.7861,
      "step": 12158
    },
    {
      "epoch": 0.8921008465016462,
      "grad_norm": 1.4339407186170061,
      "learning_rate": 3.0198860127431784e-07,
      "loss": 0.7491,
      "step": 12159
    },
    {
      "epoch": 0.8921742160917854,
      "grad_norm": 1.513417316906302,
      "learning_rate": 3.015820518505941e-07,
      "loss": 0.6482,
      "step": 12160
    },
    {
      "epoch": 0.8921742160917854,
      "lm_loss": 1.2708,
      "step": 12160,
      "vm_loss": 0.1239
    },
    {
      "epoch": 0.8921742160917854,
      "lm_loss": 0.595,
      "step": 12160,
      "vm_loss": 0.1086
    },
    {
      "epoch": 0.8921742160917854,
      "lm_loss": 0.45,
      "step": 12160,
      "vm_loss": 0.1331
    },
    {
      "epoch": 0.8921742160917854,
      "lm_loss": 0.9046,
      "step": 12160,
      "vm_loss": 0.067
    },
    {
      "epoch": 0.8921742160917854,
      "lm_loss": 0.0884,
      "step": 12160,
      "vm_loss": 0.0415
    },
    {
      "epoch": 0.8921742160917854,
      "lm_loss": 0.0349,
      "step": 12160,
      "vm_loss": 0.1232
    },
    {
      "epoch": 0.8921742160917854,
      "lm_loss": 0.0371,
      "step": 12160,
      "vm_loss": 0.1519
    },
    {
      "epoch": 0.8921742160917854,
      "lm_loss": 0.7269,
      "step": 12160,
      "vm_loss": 0.1232
    },
    {
      "epoch": 0.8922475856819245,
      "grad_norm": 1.514664235366341,
      "learning_rate": 3.011757677581673e-07,
      "loss": 0.7373,
      "step": 12161
    },
    {
      "epoch": 0.8923209552720636,
      "grad_norm": 1.478133108131143,
      "learning_rate": 3.0076974901998124e-07,
      "loss": 0.7093,
      "step": 12162
    },
    {
      "epoch": 0.8923943248622027,
      "grad_norm": 1.582166321569125,
      "learning_rate": 3.003639956589649e-07,
      "loss": 0.7206,
      "step": 12163
    },
    {
      "epoch": 0.8924676944523419,
      "grad_norm": 1.4162542572359709,
      "learning_rate": 2.9995850769803215e-07,
      "loss": 0.6867,
      "step": 12164
    },
    {
      "epoch": 0.892541064042481,
      "grad_norm": 1.406211255573256,
      "learning_rate": 2.995532851600813e-07,
      "loss": 0.626,
      "step": 12165
    },
    {
      "epoch": 0.8926144336326202,
      "grad_norm": 1.4718820594388473,
      "learning_rate": 2.99148328067998e-07,
      "loss": 0.6984,
      "step": 12166
    },
    {
      "epoch": 0.8926878032227592,
      "grad_norm": 0.5370402782092624,
      "learning_rate": 2.987436364446494e-07,
      "loss": 0.4896,
      "step": 12167
    },
    {
      "epoch": 0.8927611728128984,
      "grad_norm": 1.5033569089452592,
      "learning_rate": 2.983392103128907e-07,
      "loss": 0.7305,
      "step": 12168
    },
    {
      "epoch": 0.8927611728128984,
      "lm_loss": 1.3173,
      "step": 12168,
      "vm_loss": 0.0624
    },
    {
      "epoch": 0.8927611728128984,
      "lm_loss": 1.291,
      "step": 12168,
      "vm_loss": 0.0824
    },
    {
      "epoch": 0.8927611728128984,
      "lm_loss": 0.2871,
      "step": 12168,
      "vm_loss": 0.1091
    },
    {
      "epoch": 0.8927611728128984,
      "lm_loss": 0.6748,
      "step": 12168,
      "vm_loss": 0.1363
    },
    {
      "epoch": 0.8927611728128984,
      "lm_loss": 0.611,
      "step": 12168,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.8927611728128984,
      "lm_loss": 0.1123,
      "step": 12168,
      "vm_loss": 0.1613
    },
    {
      "epoch": 0.8927611728128984,
      "lm_loss": 0.3396,
      "step": 12168,
      "vm_loss": 0.1723
    },
    {
      "epoch": 0.8927611728128984,
      "lm_loss": 0.9028,
      "step": 12168,
      "vm_loss": 0.1857
    },
    {
      "epoch": 0.8928345424030375,
      "grad_norm": 1.4480109923621443,
      "learning_rate": 2.9793504969555965e-07,
      "loss": 0.6767,
      "step": 12169
    },
    {
      "epoch": 0.8929079119931767,
      "grad_norm": 0.5436837501528098,
      "learning_rate": 2.9753115461548076e-07,
      "loss": 0.4713,
      "step": 12170
    },
    {
      "epoch": 0.8929812815833158,
      "grad_norm": 0.5492452774264197,
      "learning_rate": 2.9712752509546303e-07,
      "loss": 0.4811,
      "step": 12171
    },
    {
      "epoch": 0.8930546511734548,
      "grad_norm": 1.6073081318534066,
      "learning_rate": 2.9672416115830107e-07,
      "loss": 0.6552,
      "step": 12172
    },
    {
      "epoch": 0.893128020763594,
      "grad_norm": 0.5443962511916224,
      "learning_rate": 2.9632106282677264e-07,
      "loss": 0.4741,
      "step": 12173
    },
    {
      "epoch": 0.8932013903537331,
      "grad_norm": 0.5248597031710908,
      "learning_rate": 2.959182301236424e-07,
      "loss": 0.4743,
      "step": 12174
    },
    {
      "epoch": 0.8932747599438723,
      "grad_norm": 1.7175674085961705,
      "learning_rate": 2.9551566307165934e-07,
      "loss": 0.7771,
      "step": 12175
    },
    {
      "epoch": 0.8933481295340114,
      "grad_norm": 1.547704900979027,
      "learning_rate": 2.95113361693557e-07,
      "loss": 0.6575,
      "step": 12176
    },
    {
      "epoch": 0.8933481295340114,
      "lm_loss": 1.037,
      "step": 12176,
      "vm_loss": 0.0713
    },
    {
      "epoch": 0.8933481295340114,
      "lm_loss": 0.0842,
      "step": 12176,
      "vm_loss": 0.1466
    },
    {
      "epoch": 0.8933481295340114,
      "lm_loss": 0.9105,
      "step": 12176,
      "vm_loss": 0.2371
    },
    {
      "epoch": 0.8933481295340114,
      "lm_loss": 0.503,
      "step": 12176,
      "vm_loss": 0.1017
    },
    {
      "epoch": 0.8933481295340114,
      "lm_loss": 0.3006,
      "step": 12176,
      "vm_loss": 0.122
    },
    {
      "epoch": 0.8933481295340114,
      "lm_loss": 0.1422,
      "step": 12176,
      "vm_loss": 0.1713
    },
    {
      "epoch": 0.8933481295340114,
      "lm_loss": 0.7079,
      "step": 12176,
      "vm_loss": 0.1777
    },
    {
      "epoch": 0.8933481295340114,
      "lm_loss": 0.1985,
      "step": 12176,
      "vm_loss": 0.1438
    },
    {
      "epoch": 0.8934214991241505,
      "grad_norm": 1.5104949152318077,
      "learning_rate": 2.947113260120543e-07,
      "loss": 0.718,
      "step": 12177
    },
    {
      "epoch": 0.8934948687142896,
      "grad_norm": 1.5129728706773096,
      "learning_rate": 2.94309556049856e-07,
      "loss": 0.7047,
      "step": 12178
    },
    {
      "epoch": 0.8935682383044288,
      "grad_norm": 1.4893072047397244,
      "learning_rate": 2.9390805182965055e-07,
      "loss": 0.6544,
      "step": 12179
    },
    {
      "epoch": 0.8936416078945679,
      "grad_norm": 1.4440817354982542,
      "learning_rate": 2.93506813374112e-07,
      "loss": 0.6993,
      "step": 12180
    },
    {
      "epoch": 0.8937149774847071,
      "grad_norm": 1.5379376851032027,
      "learning_rate": 2.931058407058979e-07,
      "loss": 0.6883,
      "step": 12181
    },
    {
      "epoch": 0.8937883470748461,
      "grad_norm": 1.483796874113968,
      "learning_rate": 2.927051338476544e-07,
      "loss": 0.6432,
      "step": 12182
    },
    {
      "epoch": 0.8938617166649853,
      "grad_norm": 1.5674454772817517,
      "learning_rate": 2.923046928220097e-07,
      "loss": 0.7835,
      "step": 12183
    },
    {
      "epoch": 0.8939350862551244,
      "grad_norm": 1.516680571669205,
      "learning_rate": 2.919045176515772e-07,
      "loss": 0.7801,
      "step": 12184
    },
    {
      "epoch": 0.8939350862551244,
      "lm_loss": 1.8042,
      "step": 12184,
      "vm_loss": 0.0628
    },
    {
      "epoch": 0.8939350862551244,
      "lm_loss": 0.7406,
      "step": 12184,
      "vm_loss": 0.1246
    },
    {
      "epoch": 0.8939350862551244,
      "lm_loss": 0.7205,
      "step": 12184,
      "vm_loss": 0.1324
    },
    {
      "epoch": 0.8939350862551244,
      "lm_loss": 0.4897,
      "step": 12184,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.8939350862551244,
      "lm_loss": 0.2257,
      "step": 12184,
      "vm_loss": 0.1016
    },
    {
      "epoch": 0.8939350862551244,
      "lm_loss": 0.0276,
      "step": 12184,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.8939350862551244,
      "lm_loss": 0.1231,
      "step": 12184,
      "vm_loss": 0.159
    },
    {
      "epoch": 0.8939350862551244,
      "lm_loss": 0.0387,
      "step": 12184,
      "vm_loss": 0.0805
    },
    {
      "epoch": 0.8940084558452636,
      "grad_norm": 1.584919135234028,
      "learning_rate": 2.9150460835895565e-07,
      "loss": 0.7186,
      "step": 12185
    },
    {
      "epoch": 0.8940818254354027,
      "grad_norm": 1.612624346449036,
      "learning_rate": 2.9110496496672915e-07,
      "loss": 0.7247,
      "step": 12186
    },
    {
      "epoch": 0.8941551950255417,
      "grad_norm": 1.486343948279526,
      "learning_rate": 2.907055874974668e-07,
      "loss": 0.7064,
      "step": 12187
    },
    {
      "epoch": 0.8942285646156809,
      "grad_norm": 1.6006285500309387,
      "learning_rate": 2.903064759737223e-07,
      "loss": 0.7509,
      "step": 12188
    },
    {
      "epoch": 0.89430193420582,
      "grad_norm": 1.3950948087126755,
      "learning_rate": 2.899076304180348e-07,
      "loss": 0.7128,
      "step": 12189
    },
    {
      "epoch": 0.8943753037959592,
      "grad_norm": 1.5063587107778023,
      "learning_rate": 2.8950905085292803e-07,
      "loss": 0.7645,
      "step": 12190
    },
    {
      "epoch": 0.8944486733860983,
      "grad_norm": 0.5367474590148931,
      "learning_rate": 2.8911073730091e-07,
      "loss": 0.4697,
      "step": 12191
    },
    {
      "epoch": 0.8945220429762374,
      "grad_norm": 1.5680365538337389,
      "learning_rate": 2.8871268978447444e-07,
      "loss": 0.9117,
      "step": 12192
    },
    {
      "epoch": 0.8945220429762374,
      "lm_loss": 1.4243,
      "step": 12192,
      "vm_loss": 0.1483
    },
    {
      "epoch": 0.8945220429762374,
      "lm_loss": 0.0798,
      "step": 12192,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.8945220429762374,
      "lm_loss": 0.8841,
      "step": 12192,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.8945220429762374,
      "lm_loss": 0.5189,
      "step": 12192,
      "vm_loss": 0.1723
    },
    {
      "epoch": 0.8945220429762374,
      "lm_loss": 0.6978,
      "step": 12192,
      "vm_loss": 0.1811
    },
    {
      "epoch": 0.8945220429762374,
      "lm_loss": 0.5398,
      "step": 12192,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.8945220429762374,
      "lm_loss": 0.3006,
      "step": 12192,
      "vm_loss": 0.1678
    },
    {
      "epoch": 0.8945220429762374,
      "lm_loss": 0.3082,
      "step": 12192,
      "vm_loss": 0.2002
    },
    {
      "epoch": 0.8945954125663765,
      "grad_norm": 1.3444992911463114,
      "learning_rate": 2.883149083261011e-07,
      "loss": 0.7573,
      "step": 12193
    },
    {
      "epoch": 0.8946687821565157,
      "grad_norm": 1.528922754398874,
      "learning_rate": 2.8791739294825314e-07,
      "loss": 0.7644,
      "step": 12194
    },
    {
      "epoch": 0.8947421517466548,
      "grad_norm": 0.5336511881525389,
      "learning_rate": 2.8752014367337935e-07,
      "loss": 0.4836,
      "step": 12195
    },
    {
      "epoch": 0.894815521336794,
      "grad_norm": 1.5428591435995038,
      "learning_rate": 2.871231605239133e-07,
      "loss": 0.635,
      "step": 12196
    },
    {
      "epoch": 0.894888890926933,
      "grad_norm": 1.5711940257882557,
      "learning_rate": 2.867264435222733e-07,
      "loss": 0.687,
      "step": 12197
    },
    {
      "epoch": 0.8949622605170722,
      "grad_norm": 1.5574028167988128,
      "learning_rate": 2.863299926908636e-07,
      "loss": 0.8232,
      "step": 12198
    },
    {
      "epoch": 0.8950356301072113,
      "grad_norm": 1.4284306401497808,
      "learning_rate": 2.8593380805207237e-07,
      "loss": 0.7695,
      "step": 12199
    },
    {
      "epoch": 0.8951089996973505,
      "grad_norm": 1.4735545394126877,
      "learning_rate": 2.855378896282734e-07,
      "loss": 0.6571,
      "step": 12200
    },
    {
      "epoch": 0.8951089996973505,
      "lm_loss": 1.6577,
      "step": 12200,
      "vm_loss": 0.0925
    },
    {
      "epoch": 0.8951089996973505,
      "lm_loss": 0.5069,
      "step": 12200,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.8951089996973505,
      "lm_loss": 0.9682,
      "step": 12200,
      "vm_loss": 0.0877
    },
    {
      "epoch": 0.8951089996973505,
      "lm_loss": 0.5694,
      "step": 12200,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.8951089996973505,
      "lm_loss": 1.2125,
      "step": 12200,
      "vm_loss": 0.3313
    },
    {
      "epoch": 0.8951089996973505,
      "lm_loss": 0.0218,
      "step": 12200,
      "vm_loss": 0.1041
    },
    {
      "epoch": 0.8951089996973505,
      "lm_loss": 0.8246,
      "step": 12200,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.8951089996973505,
      "lm_loss": 0.7656,
      "step": 12200,
      "vm_loss": 0.0869
    },
    {
      "epoch": 0.8951823692874896,
      "grad_norm": 1.4584961965664214,
      "learning_rate": 2.85142237441825e-07,
      "loss": 0.6871,
      "step": 12201
    },
    {
      "epoch": 0.8952557388776287,
      "grad_norm": 1.405186580954166,
      "learning_rate": 2.847468515150703e-07,
      "loss": 0.6688,
      "step": 12202
    },
    {
      "epoch": 0.8953291084677678,
      "grad_norm": 1.6612225859669456,
      "learning_rate": 2.8435173187033817e-07,
      "loss": 0.7261,
      "step": 12203
    },
    {
      "epoch": 0.895402478057907,
      "grad_norm": 0.5489295534531172,
      "learning_rate": 2.839568785299418e-07,
      "loss": 0.4908,
      "step": 12204
    },
    {
      "epoch": 0.8954758476480461,
      "grad_norm": 1.701899979924943,
      "learning_rate": 2.835622915161801e-07,
      "loss": 0.728,
      "step": 12205
    },
    {
      "epoch": 0.8955492172381853,
      "grad_norm": 1.5735265450285036,
      "learning_rate": 2.831679708513352e-07,
      "loss": 0.732,
      "step": 12206
    },
    {
      "epoch": 0.8956225868283243,
      "grad_norm": 1.520630368388478,
      "learning_rate": 2.827739165576754e-07,
      "loss": 0.6495,
      "step": 12207
    },
    {
      "epoch": 0.8956959564184634,
      "grad_norm": 1.478769506589541,
      "learning_rate": 2.8238012865745565e-07,
      "loss": 0.7001,
      "step": 12208
    },
    {
      "epoch": 0.8956959564184634,
      "lm_loss": 1.085,
      "step": 12208,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.8956959564184634,
      "lm_loss": 0.8576,
      "step": 12208,
      "vm_loss": 0.0936
    },
    {
      "epoch": 0.8956959564184634,
      "lm_loss": 0.5489,
      "step": 12208,
      "vm_loss": 0.144
    },
    {
      "epoch": 0.8956959564184634,
      "lm_loss": 0.1335,
      "step": 12208,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.8956959564184634,
      "lm_loss": 0.6998,
      "step": 12208,
      "vm_loss": 0.1454
    },
    {
      "epoch": 0.8956959564184634,
      "lm_loss": 0.9921,
      "step": 12208,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.8956959564184634,
      "lm_loss": 0.3369,
      "step": 12208,
      "vm_loss": 0.0876
    },
    {
      "epoch": 0.8956959564184634,
      "lm_loss": 0.5367,
      "step": 12208,
      "vm_loss": 0.1658
    },
    {
      "epoch": 0.8957693260086026,
      "grad_norm": 1.5133589816980184,
      "learning_rate": 2.819866071729127e-07,
      "loss": 0.738,
      "step": 12209
    },
    {
      "epoch": 0.8958426955987417,
      "grad_norm": 1.4288366045939158,
      "learning_rate": 2.815933521262698e-07,
      "loss": 0.6387,
      "step": 12210
    },
    {
      "epoch": 0.8959160651888808,
      "grad_norm": 1.5773028323817426,
      "learning_rate": 2.812003635397359e-07,
      "loss": 0.7813,
      "step": 12211
    },
    {
      "epoch": 0.8959894347790199,
      "grad_norm": 1.3954703316511923,
      "learning_rate": 2.808076414355021e-07,
      "loss": 0.6842,
      "step": 12212
    },
    {
      "epoch": 0.8960628043691591,
      "grad_norm": 1.4518117434011026,
      "learning_rate": 2.804151858357484e-07,
      "loss": 0.7515,
      "step": 12213
    },
    {
      "epoch": 0.8961361739592982,
      "grad_norm": 1.7411261034826548,
      "learning_rate": 2.8002299676263724e-07,
      "loss": 0.7512,
      "step": 12214
    },
    {
      "epoch": 0.8962095435494374,
      "grad_norm": 1.5002816665452252,
      "learning_rate": 2.7963107423831627e-07,
      "loss": 0.6513,
      "step": 12215
    },
    {
      "epoch": 0.8962829131395764,
      "grad_norm": 1.4590752583538913,
      "learning_rate": 2.792394182849184e-07,
      "loss": 0.7827,
      "step": 12216
    },
    {
      "epoch": 0.8962829131395764,
      "lm_loss": 1.0002,
      "step": 12216,
      "vm_loss": 0.0505
    },
    {
      "epoch": 0.8962829131395764,
      "lm_loss": 1.3868,
      "step": 12216,
      "vm_loss": 0.0456
    },
    {
      "epoch": 0.8962829131395764,
      "lm_loss": 0.3879,
      "step": 12216,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.8962829131395764,
      "lm_loss": 1.002,
      "step": 12216,
      "vm_loss": 0.2082
    },
    {
      "epoch": 0.8962829131395764,
      "lm_loss": 0.5916,
      "step": 12216,
      "vm_loss": 0.0595
    },
    {
      "epoch": 0.8962829131395764,
      "lm_loss": 0.5319,
      "step": 12216,
      "vm_loss": 0.1773
    },
    {
      "epoch": 0.8962829131395764,
      "lm_loss": 0.1567,
      "step": 12216,
      "vm_loss": 0.0526
    },
    {
      "epoch": 0.8962829131395764,
      "lm_loss": 0.3953,
      "step": 12216,
      "vm_loss": 0.1873
    },
    {
      "epoch": 0.8963562827297156,
      "grad_norm": 1.3994742479282187,
      "learning_rate": 2.7884802892456095e-07,
      "loss": 0.7096,
      "step": 12217
    },
    {
      "epoch": 0.8964296523198547,
      "grad_norm": 1.5132470802709863,
      "learning_rate": 2.7845690617934686e-07,
      "loss": 0.6523,
      "step": 12218
    },
    {
      "epoch": 0.8965030219099939,
      "grad_norm": 1.4980802096629664,
      "learning_rate": 2.7806605007136445e-07,
      "loss": 0.7804,
      "step": 12219
    },
    {
      "epoch": 0.896576391500133,
      "grad_norm": 0.5414757704540614,
      "learning_rate": 2.7767546062268556e-07,
      "loss": 0.4756,
      "step": 12220
    },
    {
      "epoch": 0.8966497610902721,
      "grad_norm": 1.3740508844435788,
      "learning_rate": 2.772851378553687e-07,
      "loss": 0.6825,
      "step": 12221
    },
    {
      "epoch": 0.8967231306804112,
      "grad_norm": 0.5396390584321132,
      "learning_rate": 2.768950817914551e-07,
      "loss": 0.4727,
      "step": 12222
    },
    {
      "epoch": 0.8967965002705504,
      "grad_norm": 1.6932195942249173,
      "learning_rate": 2.7650529245297263e-07,
      "loss": 0.784,
      "step": 12223
    },
    {
      "epoch": 0.8968698698606895,
      "grad_norm": 1.4702574333427558,
      "learning_rate": 2.7611576986193434e-07,
      "loss": 0.6307,
      "step": 12224
    },
    {
      "epoch": 0.8968698698606895,
      "lm_loss": 1.0664,
      "step": 12224,
      "vm_loss": 0.0684
    },
    {
      "epoch": 0.8968698698606895,
      "lm_loss": 0.0194,
      "step": 12224,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.8968698698606895,
      "lm_loss": 0.0973,
      "step": 12224,
      "vm_loss": 0.1396
    },
    {
      "epoch": 0.8968698698606895,
      "lm_loss": 0.5273,
      "step": 12224,
      "vm_loss": 0.1412
    },
    {
      "epoch": 0.8968698698606895,
      "lm_loss": 1.3507,
      "step": 12224,
      "vm_loss": 0.1279
    },
    {
      "epoch": 0.8968698698606895,
      "lm_loss": 0.2651,
      "step": 12224,
      "vm_loss": 0.1004
    },
    {
      "epoch": 0.8968698698606895,
      "lm_loss": 0.0856,
      "step": 12224,
      "vm_loss": 0.0411
    },
    {
      "epoch": 0.8968698698606895,
      "lm_loss": 0.5759,
      "step": 12224,
      "vm_loss": 0.0944
    },
    {
      "epoch": 0.8969432394508287,
      "grad_norm": 1.4605853422211532,
      "learning_rate": 2.7572651404033645e-07,
      "loss": 0.6439,
      "step": 12225
    },
    {
      "epoch": 0.8970166090409677,
      "grad_norm": 0.5450950212132334,
      "learning_rate": 2.7533752501016254e-07,
      "loss": 0.4953,
      "step": 12226
    },
    {
      "epoch": 0.8970899786311068,
      "grad_norm": 0.5515307076909386,
      "learning_rate": 2.7494880279337887e-07,
      "loss": 0.4732,
      "step": 12227
    },
    {
      "epoch": 0.897163348221246,
      "grad_norm": 1.33738990496062,
      "learning_rate": 2.745603474119368e-07,
      "loss": 0.6609,
      "step": 12228
    },
    {
      "epoch": 0.8972367178113851,
      "grad_norm": 0.5430611075295259,
      "learning_rate": 2.7417215888777493e-07,
      "loss": 0.472,
      "step": 12229
    },
    {
      "epoch": 0.8973100874015243,
      "grad_norm": 1.7392836241757943,
      "learning_rate": 2.7378423724281457e-07,
      "loss": 0.682,
      "step": 12230
    },
    {
      "epoch": 0.8973834569916633,
      "grad_norm": 1.5733115304384044,
      "learning_rate": 2.7339658249896317e-07,
      "loss": 0.8213,
      "step": 12231
    },
    {
      "epoch": 0.8974568265818025,
      "grad_norm": 1.5273798809796075,
      "learning_rate": 2.7300919467811104e-07,
      "loss": 0.6983,
      "step": 12232
    },
    {
      "epoch": 0.8974568265818025,
      "lm_loss": 0.3622,
      "step": 12232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8974568265818025,
      "lm_loss": 0.3641,
      "step": 12232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8974568265818025,
      "lm_loss": 0.4836,
      "step": 12232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8974568265818025,
      "lm_loss": 0.4786,
      "step": 12232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8974568265818025,
      "lm_loss": 0.4692,
      "step": 12232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8974568265818025,
      "lm_loss": 0.4653,
      "step": 12232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8974568265818025,
      "lm_loss": 0.4567,
      "step": 12232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8974568265818025,
      "lm_loss": 0.4963,
      "step": 12232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8975301961719416,
      "grad_norm": 0.5561649870438679,
      "learning_rate": 2.726220738021368e-07,
      "loss": 0.4659,
      "step": 12233
    },
    {
      "epoch": 0.8976035657620808,
      "grad_norm": 1.4863783467621445,
      "learning_rate": 2.7223521989290013e-07,
      "loss": 0.7448,
      "step": 12234
    },
    {
      "epoch": 0.8976769353522199,
      "grad_norm": 0.560846324236143,
      "learning_rate": 2.718486329722492e-07,
      "loss": 0.4933,
      "step": 12235
    },
    {
      "epoch": 0.897750304942359,
      "grad_norm": 1.8363195751226482,
      "learning_rate": 2.714623130620153e-07,
      "loss": 0.7905,
      "step": 12236
    },
    {
      "epoch": 0.8978236745324981,
      "grad_norm": 1.4838207725216586,
      "learning_rate": 2.7107626018401447e-07,
      "loss": 0.7157,
      "step": 12237
    },
    {
      "epoch": 0.8978970441226373,
      "grad_norm": 1.3793100040954425,
      "learning_rate": 2.7069047436004805e-07,
      "loss": 0.6808,
      "step": 12238
    },
    {
      "epoch": 0.8979704137127764,
      "grad_norm": 1.445074468049762,
      "learning_rate": 2.7030495561190195e-07,
      "loss": 0.7209,
      "step": 12239
    },
    {
      "epoch": 0.8980437833029156,
      "grad_norm": 1.5757428161399618,
      "learning_rate": 2.6991970396134884e-07,
      "loss": 0.6772,
      "step": 12240
    },
    {
      "epoch": 0.8980437833029156,
      "lm_loss": 1.9552,
      "step": 12240,
      "vm_loss": 0.0473
    },
    {
      "epoch": 0.8980437833029156,
      "lm_loss": 0.2891,
      "step": 12240,
      "vm_loss": 0.1065
    },
    {
      "epoch": 0.8980437833029156,
      "lm_loss": 1.1285,
      "step": 12240,
      "vm_loss": 0.0636
    },
    {
      "epoch": 0.8980437833029156,
      "lm_loss": 0.5804,
      "step": 12240,
      "vm_loss": 0.234
    },
    {
      "epoch": 0.8980437833029156,
      "lm_loss": 0.0754,
      "step": 12240,
      "vm_loss": 0.0842
    },
    {
      "epoch": 0.8980437833029156,
      "lm_loss": 0.3442,
      "step": 12240,
      "vm_loss": 0.0761
    },
    {
      "epoch": 0.8980437833029156,
      "lm_loss": 0.5613,
      "step": 12240,
      "vm_loss": 0.0756
    },
    {
      "epoch": 0.8980437833029156,
      "lm_loss": 0.054,
      "step": 12240,
      "vm_loss": 0.0829
    },
    {
      "epoch": 0.8981171528930546,
      "grad_norm": 1.5574204679236394,
      "learning_rate": 2.695347194301434e-07,
      "loss": 0.8226,
      "step": 12241
    },
    {
      "epoch": 0.8981905224831938,
      "grad_norm": 1.6494395228586982,
      "learning_rate": 2.691500020400278e-07,
      "loss": 0.6804,
      "step": 12242
    },
    {
      "epoch": 0.8982638920733329,
      "grad_norm": 1.6076740743699027,
      "learning_rate": 2.687655518127269e-07,
      "loss": 0.7811,
      "step": 12243
    },
    {
      "epoch": 0.898337261663472,
      "grad_norm": 0.5634732527359303,
      "learning_rate": 2.683813687699521e-07,
      "loss": 0.4833,
      "step": 12244
    },
    {
      "epoch": 0.8984106312536112,
      "grad_norm": 1.4070122699963827,
      "learning_rate": 2.679974529333984e-07,
      "loss": 0.7116,
      "step": 12245
    },
    {
      "epoch": 0.8984840008437502,
      "grad_norm": 1.592573780638375,
      "learning_rate": 2.676138043247473e-07,
      "loss": 0.7003,
      "step": 12246
    },
    {
      "epoch": 0.8985573704338894,
      "grad_norm": 1.5611143026034517,
      "learning_rate": 2.672304229656647e-07,
      "loss": 0.8315,
      "step": 12247
    },
    {
      "epoch": 0.8986307400240285,
      "grad_norm": 0.5436888261601144,
      "learning_rate": 2.6684730887780064e-07,
      "loss": 0.4717,
      "step": 12248
    },
    {
      "epoch": 0.8986307400240285,
      "lm_loss": 0.1619,
      "step": 12248,
      "vm_loss": 0.0393
    },
    {
      "epoch": 0.8986307400240285,
      "lm_loss": 1.4521,
      "step": 12248,
      "vm_loss": 0.1459
    },
    {
      "epoch": 0.8986307400240285,
      "lm_loss": 0.1183,
      "step": 12248,
      "vm_loss": 0.1213
    },
    {
      "epoch": 0.8986307400240285,
      "lm_loss": 0.0948,
      "step": 12248,
      "vm_loss": 0.1222
    },
    {
      "epoch": 0.8986307400240285,
      "lm_loss": 1.0994,
      "step": 12248,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.8986307400240285,
      "lm_loss": 0.9055,
      "step": 12248,
      "vm_loss": 0.181
    },
    {
      "epoch": 0.8986307400240285,
      "lm_loss": 0.6637,
      "step": 12248,
      "vm_loss": 0.0914
    },
    {
      "epoch": 0.8986307400240285,
      "lm_loss": 0.7624,
      "step": 12248,
      "vm_loss": 0.1621
    },
    {
      "epoch": 0.8987041096141677,
      "grad_norm": 1.4041444930436147,
      "learning_rate": 2.6646446208279054e-07,
      "loss": 0.7938,
      "step": 12249
    },
    {
      "epoch": 0.8987774792043068,
      "grad_norm": 1.5933656172582802,
      "learning_rate": 2.660818826022538e-07,
      "loss": 0.8533,
      "step": 12250
    },
    {
      "epoch": 0.8988508487944459,
      "grad_norm": 1.5778331449823702,
      "learning_rate": 2.6569957045779746e-07,
      "loss": 0.7631,
      "step": 12251
    },
    {
      "epoch": 0.898924218384585,
      "grad_norm": 1.6077180240233928,
      "learning_rate": 2.653175256710105e-07,
      "loss": 0.6999,
      "step": 12252
    },
    {
      "epoch": 0.8989975879747242,
      "grad_norm": 1.613510195600362,
      "learning_rate": 2.6493574826346846e-07,
      "loss": 0.7462,
      "step": 12253
    },
    {
      "epoch": 0.8990709575648633,
      "grad_norm": 0.5355281593199435,
      "learning_rate": 2.645542382567301e-07,
      "loss": 0.4801,
      "step": 12254
    },
    {
      "epoch": 0.8991443271550025,
      "grad_norm": 1.3642753216609382,
      "learning_rate": 2.64172995672341e-07,
      "loss": 0.7222,
      "step": 12255
    },
    {
      "epoch": 0.8992176967451415,
      "grad_norm": 1.5096744771476085,
      "learning_rate": 2.6379202053183174e-07,
      "loss": 0.8277,
      "step": 12256
    },
    {
      "epoch": 0.8992176967451415,
      "lm_loss": 0.5786,
      "step": 12256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8992176967451415,
      "lm_loss": 0.5107,
      "step": 12256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8992176967451415,
      "lm_loss": 0.6035,
      "step": 12256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8992176967451415,
      "lm_loss": 0.4344,
      "step": 12256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8992176967451415,
      "lm_loss": 0.4535,
      "step": 12256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8992176967451415,
      "lm_loss": 0.4896,
      "step": 12256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8992176967451415,
      "lm_loss": 0.498,
      "step": 12256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8992176967451415,
      "lm_loss": 0.4175,
      "step": 12256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.8992910663352807,
      "grad_norm": 0.5429747868335795,
      "learning_rate": 2.6341131285671564e-07,
      "loss": 0.4721,
      "step": 12257
    },
    {
      "epoch": 0.8993644359254198,
      "grad_norm": 0.5363680568168231,
      "learning_rate": 2.630308726684927e-07,
      "loss": 0.4671,
      "step": 12258
    },
    {
      "epoch": 0.899437805515559,
      "grad_norm": 1.5241312920415795,
      "learning_rate": 2.6265069998864747e-07,
      "loss": 0.7953,
      "step": 12259
    },
    {
      "epoch": 0.8995111751056981,
      "grad_norm": 1.5728525435835892,
      "learning_rate": 2.622707948386488e-07,
      "loss": 0.6626,
      "step": 12260
    },
    {
      "epoch": 0.8995845446958372,
      "grad_norm": 1.478774402476974,
      "learning_rate": 2.618911572399507e-07,
      "loss": 0.719,
      "step": 12261
    },
    {
      "epoch": 0.8996579142859763,
      "grad_norm": 1.434299926733755,
      "learning_rate": 2.615117872139933e-07,
      "loss": 0.7546,
      "step": 12262
    },
    {
      "epoch": 0.8997312838761155,
      "grad_norm": 1.3856764305492943,
      "learning_rate": 2.611326847821999e-07,
      "loss": 0.722,
      "step": 12263
    },
    {
      "epoch": 0.8998046534662546,
      "grad_norm": 1.562005463984693,
      "learning_rate": 2.607538499659795e-07,
      "loss": 0.8145,
      "step": 12264
    },
    {
      "epoch": 0.8998046534662546,
      "lm_loss": 1.099,
      "step": 12264,
      "vm_loss": 0.0664
    },
    {
      "epoch": 0.8998046534662546,
      "lm_loss": 1.253,
      "step": 12264,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.8998046534662546,
      "lm_loss": 0.0453,
      "step": 12264,
      "vm_loss": 0.0915
    },
    {
      "epoch": 0.8998046534662546,
      "lm_loss": 0.8786,
      "step": 12264,
      "vm_loss": 0.1389
    },
    {
      "epoch": 0.8998046534662546,
      "lm_loss": 0.4077,
      "step": 12264,
      "vm_loss": 0.2217
    },
    {
      "epoch": 0.8998046534662546,
      "lm_loss": 0.1027,
      "step": 12264,
      "vm_loss": 0.1677
    },
    {
      "epoch": 0.8998046534662546,
      "lm_loss": 0.2752,
      "step": 12264,
      "vm_loss": 0.1145
    },
    {
      "epoch": 0.8998046534662546,
      "lm_loss": 0.6584,
      "step": 12264,
      "vm_loss": 0.3283
    },
    {
      "epoch": 0.8998780230563937,
      "grad_norm": 1.4886823254802497,
      "learning_rate": 2.603752827867256e-07,
      "loss": 0.7452,
      "step": 12265
    },
    {
      "epoch": 0.8999513926465328,
      "grad_norm": 1.5897908074098488,
      "learning_rate": 2.599969832658167e-07,
      "loss": 0.6947,
      "step": 12266
    },
    {
      "epoch": 0.9000247622366719,
      "grad_norm": 1.830685979247691,
      "learning_rate": 2.5961895142461666e-07,
      "loss": 0.7807,
      "step": 12267
    },
    {
      "epoch": 0.9000981318268111,
      "grad_norm": 1.7650667029806457,
      "learning_rate": 2.592411872844741e-07,
      "loss": 0.6578,
      "step": 12268
    },
    {
      "epoch": 0.9001715014169502,
      "grad_norm": 1.4970383050793974,
      "learning_rate": 2.5886369086672193e-07,
      "loss": 0.6274,
      "step": 12269
    },
    {
      "epoch": 0.9002448710070894,
      "grad_norm": 1.5120644263432323,
      "learning_rate": 2.5848646219267804e-07,
      "loss": 0.7826,
      "step": 12270
    },
    {
      "epoch": 0.9003182405972284,
      "grad_norm": 1.4012014178308088,
      "learning_rate": 2.581095012836465e-07,
      "loss": 0.8005,
      "step": 12271
    },
    {
      "epoch": 0.9003916101873676,
      "grad_norm": 1.5074042168260282,
      "learning_rate": 2.5773280816091317e-07,
      "loss": 0.7625,
      "step": 12272
    },
    {
      "epoch": 0.9003916101873676,
      "lm_loss": 1.1401,
      "step": 12272,
      "vm_loss": 0.0476
    },
    {
      "epoch": 0.9003916101873676,
      "lm_loss": 0.6906,
      "step": 12272,
      "vm_loss": 0.0888
    },
    {
      "epoch": 0.9003916101873676,
      "lm_loss": 0.6054,
      "step": 12272,
      "vm_loss": 0.0716
    },
    {
      "epoch": 0.9003916101873676,
      "lm_loss": 0.5013,
      "step": 12272,
      "vm_loss": 0.0866
    },
    {
      "epoch": 0.9003916101873676,
      "lm_loss": 1.6637,
      "step": 12272,
      "vm_loss": 0.1256
    },
    {
      "epoch": 0.9003916101873676,
      "lm_loss": 0.1742,
      "step": 12272,
      "vm_loss": 0.1719
    },
    {
      "epoch": 0.9003916101873676,
      "lm_loss": 0.0067,
      "step": 12272,
      "vm_loss": 0.0543
    },
    {
      "epoch": 0.9003916101873676,
      "lm_loss": 0.0666,
      "step": 12272,
      "vm_loss": 0.1628
    },
    {
      "epoch": 0.9004649797775067,
      "grad_norm": 1.6321159590315695,
      "learning_rate": 2.5735638284575316e-07,
      "loss": 0.695,
      "step": 12273
    },
    {
      "epoch": 0.9005383493676459,
      "grad_norm": 1.5101868273931154,
      "learning_rate": 2.5698022535942333e-07,
      "loss": 0.7833,
      "step": 12274
    },
    {
      "epoch": 0.900611718957785,
      "grad_norm": 1.470806516038411,
      "learning_rate": 2.5660433572316567e-07,
      "loss": 0.7573,
      "step": 12275
    },
    {
      "epoch": 0.9006850885479241,
      "grad_norm": 1.6142092903673397,
      "learning_rate": 2.5622871395820814e-07,
      "loss": 0.8265,
      "step": 12276
    },
    {
      "epoch": 0.9007584581380632,
      "grad_norm": 0.5434769466819165,
      "learning_rate": 2.558533600857627e-07,
      "loss": 0.4696,
      "step": 12277
    },
    {
      "epoch": 0.9008318277282024,
      "grad_norm": 1.4502486168380213,
      "learning_rate": 2.5547827412702677e-07,
      "loss": 0.7753,
      "step": 12278
    },
    {
      "epoch": 0.9009051973183415,
      "grad_norm": 1.461089554807328,
      "learning_rate": 2.551034561031823e-07,
      "loss": 0.7187,
      "step": 12279
    },
    {
      "epoch": 0.9009785669084807,
      "grad_norm": 1.6526878570772345,
      "learning_rate": 2.5472890603539635e-07,
      "loss": 0.6763,
      "step": 12280
    },
    {
      "epoch": 0.9009785669084807,
      "lm_loss": 0.5489,
      "step": 12280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9009785669084807,
      "lm_loss": 0.4772,
      "step": 12280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9009785669084807,
      "lm_loss": 0.4935,
      "step": 12280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9009785669084807,
      "lm_loss": 0.447,
      "step": 12280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9009785669084807,
      "lm_loss": 0.5257,
      "step": 12280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9009785669084807,
      "lm_loss": 0.4675,
      "step": 12280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9009785669084807,
      "lm_loss": 0.5359,
      "step": 12280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9009785669084807,
      "lm_loss": 0.5975,
      "step": 12280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9010519364986197,
      "grad_norm": 0.5536137978213767,
      "learning_rate": 2.543546239448208e-07,
      "loss": 0.4907,
      "step": 12281
    },
    {
      "epoch": 0.9011253060887588,
      "grad_norm": 1.5455983899791634,
      "learning_rate": 2.53980609852591e-07,
      "loss": 0.7287,
      "step": 12282
    },
    {
      "epoch": 0.901198675678898,
      "grad_norm": 1.5362410211947444,
      "learning_rate": 2.536068637798306e-07,
      "loss": 0.7387,
      "step": 12283
    },
    {
      "epoch": 0.9012720452690371,
      "grad_norm": 1.57590154139226,
      "learning_rate": 2.532333857476443e-07,
      "loss": 0.7445,
      "step": 12284
    },
    {
      "epoch": 0.9013454148591763,
      "grad_norm": 1.6443552199593865,
      "learning_rate": 2.528601757771243e-07,
      "loss": 0.7557,
      "step": 12285
    },
    {
      "epoch": 0.9014187844493153,
      "grad_norm": 0.5576229391941202,
      "learning_rate": 2.524872338893458e-07,
      "loss": 0.5007,
      "step": 12286
    },
    {
      "epoch": 0.9014921540394545,
      "grad_norm": 0.5507171719629714,
      "learning_rate": 2.521145601053709e-07,
      "loss": 0.4764,
      "step": 12287
    },
    {
      "epoch": 0.9015655236295936,
      "grad_norm": 0.548856030460597,
      "learning_rate": 2.517421544462434e-07,
      "loss": 0.4685,
      "step": 12288
    },
    {
      "epoch": 0.9015655236295936,
      "lm_loss": 0.0831,
      "step": 12288,
      "vm_loss": 0.0433
    },
    {
      "epoch": 0.9015655236295936,
      "lm_loss": 1.7487,
      "step": 12288,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.9015655236295936,
      "lm_loss": 0.2081,
      "step": 12288,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.9015655236295936,
      "lm_loss": 0.1253,
      "step": 12288,
      "vm_loss": 0.1824
    },
    {
      "epoch": 0.9015655236295936,
      "lm_loss": 0.8986,
      "step": 12288,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.9015655236295936,
      "lm_loss": 0.7975,
      "step": 12288,
      "vm_loss": 0.193
    },
    {
      "epoch": 0.9015655236295936,
      "lm_loss": 0.1093,
      "step": 12288,
      "vm_loss": 0.0496
    },
    {
      "epoch": 0.9015655236295936,
      "lm_loss": 0.3892,
      "step": 12288,
      "vm_loss": 0.116
    },
    {
      "epoch": 0.9016388932197328,
      "grad_norm": 1.4386763241780782,
      "learning_rate": 2.513700169329963e-07,
      "loss": 0.6871,
      "step": 12289
    },
    {
      "epoch": 0.9017122628098718,
      "grad_norm": 0.5604604621023018,
      "learning_rate": 2.5099814758664407e-07,
      "loss": 0.4673,
      "step": 12290
    },
    {
      "epoch": 0.901785632400011,
      "grad_norm": 1.513478859889775,
      "learning_rate": 2.5062654642818705e-07,
      "loss": 0.7809,
      "step": 12291
    },
    {
      "epoch": 0.9018590019901501,
      "grad_norm": 0.5428134015555999,
      "learning_rate": 2.502552134786107e-07,
      "loss": 0.4752,
      "step": 12292
    },
    {
      "epoch": 0.9019323715802893,
      "grad_norm": 0.5562736116875647,
      "learning_rate": 2.498841487588843e-07,
      "loss": 0.4912,
      "step": 12293
    },
    {
      "epoch": 0.9020057411704284,
      "grad_norm": 0.5290536378189217,
      "learning_rate": 2.49513352289964e-07,
      "loss": 0.4435,
      "step": 12294
    },
    {
      "epoch": 0.9020791107605675,
      "grad_norm": 1.498704180645535,
      "learning_rate": 2.4914282409278954e-07,
      "loss": 0.694,
      "step": 12295
    },
    {
      "epoch": 0.9021524803507066,
      "grad_norm": 0.5592851911682701,
      "learning_rate": 2.487725641882849e-07,
      "loss": 0.4778,
      "step": 12296
    },
    {
      "epoch": 0.9021524803507066,
      "lm_loss": 1.408,
      "step": 12296,
      "vm_loss": 0.0857
    },
    {
      "epoch": 0.9021524803507066,
      "lm_loss": 0.4865,
      "step": 12296,
      "vm_loss": 0.1362
    },
    {
      "epoch": 0.9021524803507066,
      "lm_loss": 0.0265,
      "step": 12296,
      "vm_loss": 0.1657
    },
    {
      "epoch": 0.9021524803507066,
      "lm_loss": 0.6299,
      "step": 12296,
      "vm_loss": 0.1237
    },
    {
      "epoch": 0.9021524803507066,
      "lm_loss": 0.4002,
      "step": 12296,
      "vm_loss": 0.1311
    },
    {
      "epoch": 0.9021524803507066,
      "lm_loss": 0.8005,
      "step": 12296,
      "vm_loss": 0.1274
    },
    {
      "epoch": 0.9021524803507066,
      "lm_loss": 0.5803,
      "step": 12296,
      "vm_loss": 0.0529
    },
    {
      "epoch": 0.9021524803507066,
      "lm_loss": 0.0562,
      "step": 12296,
      "vm_loss": 0.2172
    },
    {
      "epoch": 0.9022258499408458,
      "grad_norm": 1.4493424018099068,
      "learning_rate": 2.4840257259735933e-07,
      "loss": 0.7995,
      "step": 12297
    },
    {
      "epoch": 0.9022992195309849,
      "grad_norm": 1.637416125142007,
      "learning_rate": 2.480328493409079e-07,
      "loss": 0.8032,
      "step": 12298
    },
    {
      "epoch": 0.9023725891211241,
      "grad_norm": 1.5266294796164621,
      "learning_rate": 2.476633944398088e-07,
      "loss": 0.7699,
      "step": 12299
    },
    {
      "epoch": 0.9024459587112631,
      "grad_norm": 1.4288883425128818,
      "learning_rate": 2.4729420791492767e-07,
      "loss": 0.6412,
      "step": 12300
    },
    {
      "epoch": 0.9025193283014022,
      "grad_norm": 1.562906291389003,
      "learning_rate": 2.4692528978711274e-07,
      "loss": 0.7654,
      "step": 12301
    },
    {
      "epoch": 0.9025926978915414,
      "grad_norm": 0.5341607236350491,
      "learning_rate": 2.4655664007719683e-07,
      "loss": 0.4799,
      "step": 12302
    },
    {
      "epoch": 0.9026660674816805,
      "grad_norm": 1.3688316864727519,
      "learning_rate": 2.461882588059994e-07,
      "loss": 0.625,
      "step": 12303
    },
    {
      "epoch": 0.9027394370718197,
      "grad_norm": 1.6022243185277236,
      "learning_rate": 2.4582014599432325e-07,
      "loss": 0.7371,
      "step": 12304
    },
    {
      "epoch": 0.9027394370718197,
      "lm_loss": 1.4991,
      "step": 12304,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.9027394370718197,
      "lm_loss": 1.1627,
      "step": 12304,
      "vm_loss": 0.0786
    },
    {
      "epoch": 0.9027394370718197,
      "lm_loss": 0.0943,
      "step": 12304,
      "vm_loss": 0.0636
    },
    {
      "epoch": 0.9027394370718197,
      "lm_loss": 0.5038,
      "step": 12304,
      "vm_loss": 0.1599
    },
    {
      "epoch": 0.9027394370718197,
      "lm_loss": 0.0394,
      "step": 12304,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.9027394370718197,
      "lm_loss": 0.8018,
      "step": 12304,
      "vm_loss": 0.1197
    },
    {
      "epoch": 0.9027394370718197,
      "lm_loss": 0.3355,
      "step": 12304,
      "vm_loss": 0.1173
    },
    {
      "epoch": 0.9027394370718197,
      "lm_loss": 0.0015,
      "step": 12304,
      "vm_loss": 0.1663
    },
    {
      "epoch": 0.9028128066619587,
      "grad_norm": 1.4829835471675514,
      "learning_rate": 2.454523016629573e-07,
      "loss": 0.7295,
      "step": 12305
    },
    {
      "epoch": 0.9028861762520979,
      "grad_norm": 1.7037301134230758,
      "learning_rate": 2.4508472583267437e-07,
      "loss": 0.7042,
      "step": 12306
    },
    {
      "epoch": 0.902959545842237,
      "grad_norm": 1.7735797048091937,
      "learning_rate": 2.447174185242324e-07,
      "loss": 0.85,
      "step": 12307
    },
    {
      "epoch": 0.9030329154323762,
      "grad_norm": 1.5039494479134992,
      "learning_rate": 2.4435037975837405e-07,
      "loss": 0.5913,
      "step": 12308
    },
    {
      "epoch": 0.9031062850225153,
      "grad_norm": 1.456704931042611,
      "learning_rate": 2.439836095558262e-07,
      "loss": 0.8103,
      "step": 12309
    },
    {
      "epoch": 0.9031796546126544,
      "grad_norm": 1.5590658320392825,
      "learning_rate": 2.436171079373029e-07,
      "loss": 0.7166,
      "step": 12310
    },
    {
      "epoch": 0.9032530242027935,
      "grad_norm": 1.5947211983689735,
      "learning_rate": 2.432508749235002e-07,
      "loss": 0.7483,
      "step": 12311
    },
    {
      "epoch": 0.9033263937929327,
      "grad_norm": 0.5385163116580395,
      "learning_rate": 2.4288491053510056e-07,
      "loss": 0.47,
      "step": 12312
    },
    {
      "epoch": 0.9033263937929327,
      "lm_loss": 0.5011,
      "step": 12312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9033263937929327,
      "lm_loss": 0.4085,
      "step": 12312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9033263937929327,
      "lm_loss": 0.637,
      "step": 12312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9033263937929327,
      "lm_loss": 0.3608,
      "step": 12312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9033263937929327,
      "lm_loss": 0.3505,
      "step": 12312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9033263937929327,
      "lm_loss": 0.5283,
      "step": 12312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9033263937929327,
      "lm_loss": 0.5127,
      "step": 12312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9033263937929327,
      "lm_loss": 0.529,
      "step": 12312,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9033997633830718,
      "grad_norm": 0.5517912347265712,
      "learning_rate": 2.4251921479277073e-07,
      "loss": 0.4859,
      "step": 12313
    },
    {
      "epoch": 0.903473132973211,
      "grad_norm": 0.55591489414029,
      "learning_rate": 2.421537877171626e-07,
      "loss": 0.4917,
      "step": 12314
    },
    {
      "epoch": 0.90354650256335,
      "grad_norm": 1.3991067974703753,
      "learning_rate": 2.417886293289118e-07,
      "loss": 0.6466,
      "step": 12315
    },
    {
      "epoch": 0.9036198721534892,
      "grad_norm": 1.583334548099821,
      "learning_rate": 2.4142373964864185e-07,
      "loss": 0.7795,
      "step": 12316
    },
    {
      "epoch": 0.9036932417436283,
      "grad_norm": 1.3939090092522655,
      "learning_rate": 2.4105911869695744e-07,
      "loss": 0.7634,
      "step": 12317
    },
    {
      "epoch": 0.9037666113337675,
      "grad_norm": 1.5446662483519218,
      "learning_rate": 2.4069476649444976e-07,
      "loss": 0.6237,
      "step": 12318
    },
    {
      "epoch": 0.9038399809239066,
      "grad_norm": 0.5524358243708863,
      "learning_rate": 2.4033068306169526e-07,
      "loss": 0.4912,
      "step": 12319
    },
    {
      "epoch": 0.9039133505140456,
      "grad_norm": 1.5125046204045869,
      "learning_rate": 2.39966868419253e-07,
      "loss": 0.8296,
      "step": 12320
    },
    {
      "epoch": 0.9039133505140456,
      "lm_loss": 1.2494,
      "step": 12320,
      "vm_loss": 0.0959
    },
    {
      "epoch": 0.9039133505140456,
      "lm_loss": 0.7224,
      "step": 12320,
      "vm_loss": 0.072
    },
    {
      "epoch": 0.9039133505140456,
      "lm_loss": 0.8511,
      "step": 12320,
      "vm_loss": 0.0824
    },
    {
      "epoch": 0.9039133505140456,
      "lm_loss": 0.8175,
      "step": 12320,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.9039133505140456,
      "lm_loss": 0.0619,
      "step": 12320,
      "vm_loss": 0.1109
    },
    {
      "epoch": 0.9039133505140456,
      "lm_loss": 0.7585,
      "step": 12320,
      "vm_loss": 0.0663
    },
    {
      "epoch": 0.9039133505140456,
      "lm_loss": 0.1606,
      "step": 12320,
      "vm_loss": 0.1446
    },
    {
      "epoch": 0.9039133505140456,
      "lm_loss": 0.5141,
      "step": 12320,
      "vm_loss": 0.1784
    },
    {
      "epoch": 0.9039867201041848,
      "grad_norm": 1.3429300754794777,
      "learning_rate": 2.396033225876704e-07,
      "loss": 0.5886,
      "step": 12321
    },
    {
      "epoch": 0.9040600896943239,
      "grad_norm": 0.5482378948747193,
      "learning_rate": 2.392400455874777e-07,
      "loss": 0.4724,
      "step": 12322
    },
    {
      "epoch": 0.9041334592844631,
      "grad_norm": 1.676041666821227,
      "learning_rate": 2.388770374391891e-07,
      "loss": 0.7885,
      "step": 12323
    },
    {
      "epoch": 0.9042068288746022,
      "grad_norm": 1.461134069666452,
      "learning_rate": 2.3851429816330484e-07,
      "loss": 0.6918,
      "step": 12324
    },
    {
      "epoch": 0.9042801984647413,
      "grad_norm": 1.4366318409455934,
      "learning_rate": 2.3815182778030966e-07,
      "loss": 0.5846,
      "step": 12325
    },
    {
      "epoch": 0.9043535680548804,
      "grad_norm": 1.5068655007251863,
      "learning_rate": 2.377896263106727e-07,
      "loss": 0.7474,
      "step": 12326
    },
    {
      "epoch": 0.9044269376450196,
      "grad_norm": 1.5342788921893564,
      "learning_rate": 2.3742769377484986e-07,
      "loss": 0.7486,
      "step": 12327
    },
    {
      "epoch": 0.9045003072351587,
      "grad_norm": 1.4334029478555266,
      "learning_rate": 2.3706603019327922e-07,
      "loss": 0.7298,
      "step": 12328
    },
    {
      "epoch": 0.9045003072351587,
      "lm_loss": 1.2834,
      "step": 12328,
      "vm_loss": 0.1565
    },
    {
      "epoch": 0.9045003072351587,
      "lm_loss": 1.0066,
      "step": 12328,
      "vm_loss": 0.1336
    },
    {
      "epoch": 0.9045003072351587,
      "lm_loss": 0.0007,
      "step": 12328,
      "vm_loss": 0.0761
    },
    {
      "epoch": 0.9045003072351587,
      "lm_loss": 0.4473,
      "step": 12328,
      "vm_loss": 0.1055
    },
    {
      "epoch": 0.9045003072351587,
      "lm_loss": 0.2923,
      "step": 12328,
      "vm_loss": 0.1488
    },
    {
      "epoch": 0.9045003072351587,
      "lm_loss": 0.172,
      "step": 12328,
      "vm_loss": 0.0961
    },
    {
      "epoch": 0.9045003072351587,
      "lm_loss": 0.5588,
      "step": 12328,
      "vm_loss": 0.1652
    },
    {
      "epoch": 0.9045003072351587,
      "lm_loss": 0.1167,
      "step": 12328,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.9045736768252979,
      "grad_norm": 1.4488065094486855,
      "learning_rate": 2.3670463558638556e-07,
      "loss": 0.8105,
      "step": 12329
    },
    {
      "epoch": 0.9046470464154369,
      "grad_norm": 1.4745941104554274,
      "learning_rate": 2.3634350997457645e-07,
      "loss": 0.6985,
      "step": 12330
    },
    {
      "epoch": 0.9047204160055761,
      "grad_norm": 1.5083232542090586,
      "learning_rate": 2.359826533782461e-07,
      "loss": 0.7661,
      "step": 12331
    },
    {
      "epoch": 0.9047937855957152,
      "grad_norm": 1.6200664865465981,
      "learning_rate": 2.356220658177738e-07,
      "loss": 0.7206,
      "step": 12332
    },
    {
      "epoch": 0.9048671551858544,
      "grad_norm": 0.5466043938940027,
      "learning_rate": 2.3526174731352214e-07,
      "loss": 0.4548,
      "step": 12333
    },
    {
      "epoch": 0.9049405247759935,
      "grad_norm": 1.5176331174469506,
      "learning_rate": 2.3490169788583982e-07,
      "loss": 0.7185,
      "step": 12334
    },
    {
      "epoch": 0.9050138943661326,
      "grad_norm": 0.5350442564025316,
      "learning_rate": 2.345419175550584e-07,
      "loss": 0.4812,
      "step": 12335
    },
    {
      "epoch": 0.9050872639562717,
      "grad_norm": 1.3871634046154457,
      "learning_rate": 2.3418240634149604e-07,
      "loss": 0.6864,
      "step": 12336
    },
    {
      "epoch": 0.9050872639562717,
      "lm_loss": 0.8308,
      "step": 12336,
      "vm_loss": 0.0567
    },
    {
      "epoch": 0.9050872639562717,
      "lm_loss": 0.8403,
      "step": 12336,
      "vm_loss": 0.1416
    },
    {
      "epoch": 0.9050872639562717,
      "lm_loss": 0.8497,
      "step": 12336,
      "vm_loss": 0.1457
    },
    {
      "epoch": 0.9050872639562717,
      "lm_loss": 0.6183,
      "step": 12336,
      "vm_loss": 0.0938
    },
    {
      "epoch": 0.9050872639562717,
      "lm_loss": 0.0016,
      "step": 12336,
      "vm_loss": 0.1066
    },
    {
      "epoch": 0.9050872639562717,
      "lm_loss": 0.822,
      "step": 12336,
      "vm_loss": 0.1133
    },
    {
      "epoch": 0.9050872639562717,
      "lm_loss": 0.5948,
      "step": 12336,
      "vm_loss": 0.1734
    },
    {
      "epoch": 0.9050872639562717,
      "lm_loss": 0.5371,
      "step": 12336,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.9051606335464109,
      "grad_norm": 1.4030775039262544,
      "learning_rate": 2.3382316426545647e-07,
      "loss": 0.7135,
      "step": 12337
    },
    {
      "epoch": 0.90523400313655,
      "grad_norm": 1.4119927037462556,
      "learning_rate": 2.3346419134722576e-07,
      "loss": 0.6345,
      "step": 12338
    },
    {
      "epoch": 0.9053073727266892,
      "grad_norm": 0.5358143875941435,
      "learning_rate": 2.3310548760707652e-07,
      "loss": 0.4727,
      "step": 12339
    },
    {
      "epoch": 0.9053807423168282,
      "grad_norm": 0.5307889528651805,
      "learning_rate": 2.3274705306526478e-07,
      "loss": 0.4656,
      "step": 12340
    },
    {
      "epoch": 0.9054541119069673,
      "grad_norm": 1.5745431120871263,
      "learning_rate": 2.3238888774203328e-07,
      "loss": 0.7113,
      "step": 12341
    },
    {
      "epoch": 0.9055274814971065,
      "grad_norm": 1.5496242654195413,
      "learning_rate": 2.3203099165760746e-07,
      "loss": 0.7631,
      "step": 12342
    },
    {
      "epoch": 0.9056008510872456,
      "grad_norm": 1.6063552548347073,
      "learning_rate": 2.3167336483220003e-07,
      "loss": 0.7274,
      "step": 12343
    },
    {
      "epoch": 0.9056742206773848,
      "grad_norm": 1.7395027364396325,
      "learning_rate": 2.31316007286006e-07,
      "loss": 0.856,
      "step": 12344
    },
    {
      "epoch": 0.9056742206773848,
      "lm_loss": 1.1725,
      "step": 12344,
      "vm_loss": 0.0567
    },
    {
      "epoch": 0.9056742206773848,
      "lm_loss": 1.4054,
      "step": 12344,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.9056742206773848,
      "lm_loss": 0.247,
      "step": 12344,
      "vm_loss": 0.0818
    },
    {
      "epoch": 0.9056742206773848,
      "lm_loss": 0.1543,
      "step": 12344,
      "vm_loss": 0.1054
    },
    {
      "epoch": 0.9056742206773848,
      "lm_loss": 1.0362,
      "step": 12344,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.9056742206773848,
      "lm_loss": 0.8829,
      "step": 12344,
      "vm_loss": 0.1936
    },
    {
      "epoch": 0.9056742206773848,
      "lm_loss": 0.4605,
      "step": 12344,
      "vm_loss": 0.1393
    },
    {
      "epoch": 0.9056742206773848,
      "lm_loss": 0.0262,
      "step": 12344,
      "vm_loss": 0.185
    },
    {
      "epoch": 0.9057475902675238,
      "grad_norm": 1.5215476597046484,
      "learning_rate": 2.3095891903920642e-07,
      "loss": 0.7987,
      "step": 12345
    },
    {
      "epoch": 0.905820959857663,
      "grad_norm": 1.6483657652902737,
      "learning_rate": 2.306021001119674e-07,
      "loss": 0.7849,
      "step": 12346
    },
    {
      "epoch": 0.9058943294478021,
      "grad_norm": 1.548045603769425,
      "learning_rate": 2.3024555052443777e-07,
      "loss": 0.7799,
      "step": 12347
    },
    {
      "epoch": 0.9059676990379413,
      "grad_norm": 1.5240880898177662,
      "learning_rate": 2.2988927029675535e-07,
      "loss": 0.7706,
      "step": 12348
    },
    {
      "epoch": 0.9060410686280804,
      "grad_norm": 1.556451343506056,
      "learning_rate": 2.2953325944903848e-07,
      "loss": 0.7577,
      "step": 12349
    },
    {
      "epoch": 0.9061144382182195,
      "grad_norm": 0.5656211851738217,
      "learning_rate": 2.2917751800139222e-07,
      "loss": 0.4726,
      "step": 12350
    },
    {
      "epoch": 0.9061878078083586,
      "grad_norm": 1.3488281548307794,
      "learning_rate": 2.28822045973906e-07,
      "loss": 0.7813,
      "step": 12351
    },
    {
      "epoch": 0.9062611773984978,
      "grad_norm": 0.5407333328617809,
      "learning_rate": 2.284668433866549e-07,
      "loss": 0.4681,
      "step": 12352
    },
    {
      "epoch": 0.9062611773984978,
      "lm_loss": 0.9769,
      "step": 12352,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.9062611773984978,
      "lm_loss": 0.0421,
      "step": 12352,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.9062611773984978,
      "lm_loss": 0.4828,
      "step": 12352,
      "vm_loss": 0.1414
    },
    {
      "epoch": 0.9062611773984978,
      "lm_loss": 0.2556,
      "step": 12352,
      "vm_loss": 0.0782
    },
    {
      "epoch": 0.9062611773984978,
      "lm_loss": 0.071,
      "step": 12352,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.9062611773984978,
      "lm_loss": 0.1042,
      "step": 12352,
      "vm_loss": 0.0964
    },
    {
      "epoch": 0.9062611773984978,
      "lm_loss": 1.2522,
      "step": 12352,
      "vm_loss": 0.2638
    },
    {
      "epoch": 0.9062611773984978,
      "lm_loss": 1.2737,
      "step": 12352,
      "vm_loss": 0.0832
    },
    {
      "epoch": 0.9063345469886369,
      "grad_norm": 1.4405624866561708,
      "learning_rate": 2.2811191025969738e-07,
      "loss": 0.7009,
      "step": 12353
    },
    {
      "epoch": 0.9064079165787761,
      "grad_norm": 1.6566556851088978,
      "learning_rate": 2.2775724661307785e-07,
      "loss": 0.7061,
      "step": 12354
    },
    {
      "epoch": 0.9064812861689151,
      "grad_norm": 1.5595594008523996,
      "learning_rate": 2.274028524668248e-07,
      "loss": 0.7711,
      "step": 12355
    },
    {
      "epoch": 0.9065546557590543,
      "grad_norm": 0.5476135851727831,
      "learning_rate": 2.2704872784095223e-07,
      "loss": 0.4732,
      "step": 12356
    },
    {
      "epoch": 0.9066280253491934,
      "grad_norm": 1.4922894019977104,
      "learning_rate": 2.2669487275545687e-07,
      "loss": 0.6962,
      "step": 12357
    },
    {
      "epoch": 0.9067013949393326,
      "grad_norm": 1.4935439135832664,
      "learning_rate": 2.2634128723032389e-07,
      "loss": 0.73,
      "step": 12358
    },
    {
      "epoch": 0.9067747645294717,
      "grad_norm": 1.5282605753144083,
      "learning_rate": 2.2598797128551953e-07,
      "loss": 0.8224,
      "step": 12359
    },
    {
      "epoch": 0.9068481341196107,
      "grad_norm": 1.4116106081813504,
      "learning_rate": 2.2563492494099782e-07,
      "loss": 0.6314,
      "step": 12360
    },
    {
      "epoch": 0.9068481341196107,
      "lm_loss": 1.2935,
      "step": 12360,
      "vm_loss": 0.0368
    },
    {
      "epoch": 0.9068481341196107,
      "lm_loss": 0.1284,
      "step": 12360,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.9068481341196107,
      "lm_loss": 0.7681,
      "step": 12360,
      "vm_loss": 0.2423
    },
    {
      "epoch": 0.9068481341196107,
      "lm_loss": 0.7869,
      "step": 12360,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.9068481341196107,
      "lm_loss": 0.5146,
      "step": 12360,
      "vm_loss": 0.0664
    },
    {
      "epoch": 0.9068481341196107,
      "lm_loss": 0.8107,
      "step": 12360,
      "vm_loss": 0.2663
    },
    {
      "epoch": 0.9068481341196107,
      "lm_loss": 0.0445,
      "step": 12360,
      "vm_loss": 0.063
    },
    {
      "epoch": 0.9068481341196107,
      "lm_loss": 0.077,
      "step": 12360,
      "vm_loss": 0.0555
    },
    {
      "epoch": 0.9069215037097499,
      "grad_norm": 1.6241478711064732,
      "learning_rate": 2.2528214821669504e-07,
      "loss": 0.731,
      "step": 12361
    },
    {
      "epoch": 0.906994873299889,
      "grad_norm": 0.5436351250389958,
      "learning_rate": 2.24929641132533e-07,
      "loss": 0.4789,
      "step": 12362
    },
    {
      "epoch": 0.9070682428900282,
      "grad_norm": 0.5279716746275405,
      "learning_rate": 2.2457740370842029e-07,
      "loss": 0.4519,
      "step": 12363
    },
    {
      "epoch": 0.9071416124801673,
      "grad_norm": 1.5190168397589867,
      "learning_rate": 2.2422543596424762e-07,
      "loss": 0.7458,
      "step": 12364
    },
    {
      "epoch": 0.9072149820703064,
      "grad_norm": 0.5241329058348757,
      "learning_rate": 2.2387373791989187e-07,
      "loss": 0.464,
      "step": 12365
    },
    {
      "epoch": 0.9072883516604455,
      "grad_norm": 1.5033019446471776,
      "learning_rate": 2.2352230959521383e-07,
      "loss": 0.6976,
      "step": 12366
    },
    {
      "epoch": 0.9073617212505847,
      "grad_norm": 1.5882295271773308,
      "learning_rate": 2.231711510100598e-07,
      "loss": 0.7068,
      "step": 12367
    },
    {
      "epoch": 0.9074350908407238,
      "grad_norm": 1.6989819361653151,
      "learning_rate": 2.2282026218425955e-07,
      "loss": 0.8074,
      "step": 12368
    },
    {
      "epoch": 0.9074350908407238,
      "lm_loss": 1.8037,
      "step": 12368,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.9074350908407238,
      "lm_loss": 0.7793,
      "step": 12368,
      "vm_loss": 0.0962
    },
    {
      "epoch": 0.9074350908407238,
      "lm_loss": 0.0604,
      "step": 12368,
      "vm_loss": 0.0916
    },
    {
      "epoch": 0.9074350908407238,
      "lm_loss": 0.83,
      "step": 12368,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.9074350908407238,
      "lm_loss": 0.0395,
      "step": 12368,
      "vm_loss": 0.1104
    },
    {
      "epoch": 0.9074350908407238,
      "lm_loss": 0.1303,
      "step": 12368,
      "vm_loss": 0.0757
    },
    {
      "epoch": 0.9074350908407238,
      "lm_loss": 1.3735,
      "step": 12368,
      "vm_loss": 0.2001
    },
    {
      "epoch": 0.9074350908407238,
      "lm_loss": 0.895,
      "step": 12368,
      "vm_loss": 0.2017
    },
    {
      "epoch": 0.907508460430863,
      "grad_norm": 1.7935175421961513,
      "learning_rate": 2.2246964313763053e-07,
      "loss": 0.7703,
      "step": 12369
    },
    {
      "epoch": 0.907581830021002,
      "grad_norm": 1.4986828116434534,
      "learning_rate": 2.221192938899719e-07,
      "loss": 0.7071,
      "step": 12370
    },
    {
      "epoch": 0.9076551996111412,
      "grad_norm": 0.5471651561499709,
      "learning_rate": 2.2176921446106893e-07,
      "loss": 0.4589,
      "step": 12371
    },
    {
      "epoch": 0.9077285692012803,
      "grad_norm": 1.5445911898961608,
      "learning_rate": 2.2141940487069136e-07,
      "loss": 0.7557,
      "step": 12372
    },
    {
      "epoch": 0.9078019387914195,
      "grad_norm": 0.5486053332438052,
      "learning_rate": 2.2106986513859396e-07,
      "loss": 0.4772,
      "step": 12373
    },
    {
      "epoch": 0.9078753083815585,
      "grad_norm": 0.54692922809403,
      "learning_rate": 2.2072059528451595e-07,
      "loss": 0.4721,
      "step": 12374
    },
    {
      "epoch": 0.9079486779716976,
      "grad_norm": 1.6124016548655837,
      "learning_rate": 2.203715953281821e-07,
      "loss": 0.6872,
      "step": 12375
    },
    {
      "epoch": 0.9080220475618368,
      "grad_norm": 0.5589406378071076,
      "learning_rate": 2.200228652893005e-07,
      "loss": 0.4717,
      "step": 12376
    },
    {
      "epoch": 0.9080220475618368,
      "lm_loss": 1.3866,
      "step": 12376,
      "vm_loss": 0.0778
    },
    {
      "epoch": 0.9080220475618368,
      "lm_loss": 1.1141,
      "step": 12376,
      "vm_loss": 0.1388
    },
    {
      "epoch": 0.9080220475618368,
      "lm_loss": 0.0069,
      "step": 12376,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.9080220475618368,
      "lm_loss": 0.0529,
      "step": 12376,
      "vm_loss": 0.105
    },
    {
      "epoch": 0.9080220475618368,
      "lm_loss": 0.4427,
      "step": 12376,
      "vm_loss": 0.0708
    },
    {
      "epoch": 0.9080220475618368,
      "lm_loss": 0.349,
      "step": 12376,
      "vm_loss": 0.1314
    },
    {
      "epoch": 0.9080220475618368,
      "lm_loss": 0.7872,
      "step": 12376,
      "vm_loss": 0.0987
    },
    {
      "epoch": 0.9080220475618368,
      "lm_loss": 0.1897,
      "step": 12376,
      "vm_loss": 0.128
    },
    {
      "epoch": 0.908095417151976,
      "grad_norm": 1.4031251708127033,
      "learning_rate": 2.1967440518756544e-07,
      "loss": 0.6908,
      "step": 12377
    },
    {
      "epoch": 0.9081687867421151,
      "grad_norm": 1.4635517344857092,
      "learning_rate": 2.193262150426545e-07,
      "loss": 0.7624,
      "step": 12378
    },
    {
      "epoch": 0.9082421563322541,
      "grad_norm": 1.614991265290854,
      "learning_rate": 2.1897829487423139e-07,
      "loss": 0.6764,
      "step": 12379
    },
    {
      "epoch": 0.9083155259223933,
      "grad_norm": 0.5455091066886032,
      "learning_rate": 2.1863064470194428e-07,
      "loss": 0.477,
      "step": 12380
    },
    {
      "epoch": 0.9083888955125324,
      "grad_norm": 1.428751958165074,
      "learning_rate": 2.182832645454258e-07,
      "loss": 0.6172,
      "step": 12381
    },
    {
      "epoch": 0.9084622651026716,
      "grad_norm": 0.5503609470752918,
      "learning_rate": 2.179361544242925e-07,
      "loss": 0.4763,
      "step": 12382
    },
    {
      "epoch": 0.9085356346928107,
      "grad_norm": 1.4941604158626025,
      "learning_rate": 2.1758931435814756e-07,
      "loss": 0.611,
      "step": 12383
    },
    {
      "epoch": 0.9086090042829498,
      "grad_norm": 0.5486875349311587,
      "learning_rate": 2.17242744366577e-07,
      "loss": 0.4919,
      "step": 12384
    },
    {
      "epoch": 0.9086090042829498,
      "lm_loss": 1.1985,
      "step": 12384,
      "vm_loss": 0.051
    },
    {
      "epoch": 0.9086090042829498,
      "lm_loss": 0.0802,
      "step": 12384,
      "vm_loss": 0.1005
    },
    {
      "epoch": 0.9086090042829498,
      "lm_loss": 0.0502,
      "step": 12384,
      "vm_loss": 0.1014
    },
    {
      "epoch": 0.9086090042829498,
      "lm_loss": 0.0646,
      "step": 12384,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.9086090042829498,
      "lm_loss": 0.8451,
      "step": 12384,
      "vm_loss": 0.1858
    },
    {
      "epoch": 0.9086090042829498,
      "lm_loss": 0.2138,
      "step": 12384,
      "vm_loss": 0.2025
    },
    {
      "epoch": 0.9086090042829498,
      "lm_loss": 0.0629,
      "step": 12384,
      "vm_loss": 0.0923
    },
    {
      "epoch": 0.9086090042829498,
      "lm_loss": 0.627,
      "step": 12384,
      "vm_loss": 0.1482
    },
    {
      "epoch": 0.9086823738730889,
      "grad_norm": 1.3727560843436013,
      "learning_rate": 2.1689644446915292e-07,
      "loss": 0.5629,
      "step": 12385
    },
    {
      "epoch": 0.9087557434632281,
      "grad_norm": 1.645499626296713,
      "learning_rate": 2.1655041468543246e-07,
      "loss": 0.8291,
      "step": 12386
    },
    {
      "epoch": 0.9088291130533672,
      "grad_norm": 1.63538640827698,
      "learning_rate": 2.1620465503495558e-07,
      "loss": 0.7609,
      "step": 12387
    },
    {
      "epoch": 0.9089024826435064,
      "grad_norm": 1.5292436492532833,
      "learning_rate": 2.158591655372494e-07,
      "loss": 0.722,
      "step": 12388
    },
    {
      "epoch": 0.9089758522336454,
      "grad_norm": 1.4197774763623376,
      "learning_rate": 2.1551394621182277e-07,
      "loss": 0.6669,
      "step": 12389
    },
    {
      "epoch": 0.9090492218237846,
      "grad_norm": 1.4319522305949384,
      "learning_rate": 2.1516899707817286e-07,
      "loss": 0.6196,
      "step": 12390
    },
    {
      "epoch": 0.9091225914139237,
      "grad_norm": 1.4321782233397575,
      "learning_rate": 2.1482431815577908e-07,
      "loss": 0.7247,
      "step": 12391
    },
    {
      "epoch": 0.9091959610040629,
      "grad_norm": 0.5599796265982473,
      "learning_rate": 2.144799094641059e-07,
      "loss": 0.4721,
      "step": 12392
    },
    {
      "epoch": 0.9091959610040629,
      "lm_loss": 1.1743,
      "step": 12392,
      "vm_loss": 0.07
    },
    {
      "epoch": 0.9091959610040629,
      "lm_loss": 1.3048,
      "step": 12392,
      "vm_loss": 0.0861
    },
    {
      "epoch": 0.9091959610040629,
      "lm_loss": 0.3574,
      "step": 12392,
      "vm_loss": 0.1368
    },
    {
      "epoch": 0.9091959610040629,
      "lm_loss": 0.6509,
      "step": 12392,
      "vm_loss": 0.1264
    },
    {
      "epoch": 0.9091959610040629,
      "lm_loss": 0.7935,
      "step": 12392,
      "vm_loss": 0.1461
    },
    {
      "epoch": 0.9091959610040629,
      "lm_loss": 0.1208,
      "step": 12392,
      "vm_loss": 0.0486
    },
    {
      "epoch": 0.9091959610040629,
      "lm_loss": 0.912,
      "step": 12392,
      "vm_loss": 0.1012
    },
    {
      "epoch": 0.9091959610040629,
      "lm_loss": 0.639,
      "step": 12392,
      "vm_loss": 0.0817
    },
    {
      "epoch": 0.909269330594202,
      "grad_norm": 1.7060362828887985,
      "learning_rate": 2.1413577102260386e-07,
      "loss": 0.7753,
      "step": 12393
    },
    {
      "epoch": 0.909342700184341,
      "grad_norm": 0.5524100531746486,
      "learning_rate": 2.1379190285070682e-07,
      "loss": 0.4799,
      "step": 12394
    },
    {
      "epoch": 0.9094160697744802,
      "grad_norm": 1.572292416608631,
      "learning_rate": 2.134483049678332e-07,
      "loss": 0.7401,
      "step": 12395
    },
    {
      "epoch": 0.9094894393646193,
      "grad_norm": 1.4701981882547839,
      "learning_rate": 2.1310497739338854e-07,
      "loss": 0.7723,
      "step": 12396
    },
    {
      "epoch": 0.9095628089547585,
      "grad_norm": 1.4340484502992799,
      "learning_rate": 2.1276192014675956e-07,
      "loss": 0.6539,
      "step": 12397
    },
    {
      "epoch": 0.9096361785448976,
      "grad_norm": 1.6936469601755897,
      "learning_rate": 2.1241913324732078e-07,
      "loss": 0.6233,
      "step": 12398
    },
    {
      "epoch": 0.9097095481350367,
      "grad_norm": 0.5451126654132709,
      "learning_rate": 2.1207661671443003e-07,
      "loss": 0.4642,
      "step": 12399
    },
    {
      "epoch": 0.9097829177251758,
      "grad_norm": 1.368649328255379,
      "learning_rate": 2.117343705674285e-07,
      "loss": 0.7274,
      "step": 12400
    },
    {
      "epoch": 0.9097829177251758,
      "lm_loss": 0.5109,
      "step": 12400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9097829177251758,
      "lm_loss": 0.4169,
      "step": 12400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9097829177251758,
      "lm_loss": 0.2788,
      "step": 12400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9097829177251758,
      "lm_loss": 0.5515,
      "step": 12400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9097829177251758,
      "lm_loss": 0.5631,
      "step": 12400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9097829177251758,
      "lm_loss": 0.4973,
      "step": 12400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9097829177251758,
      "lm_loss": 0.4713,
      "step": 12400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9097829177251758,
      "lm_loss": 0.5449,
      "step": 12400,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.909856287315315,
      "grad_norm": 0.5327102991408813,
      "learning_rate": 2.1139239482564634e-07,
      "loss": 0.46,
      "step": 12401
    },
    {
      "epoch": 0.9099296569054541,
      "grad_norm": 1.3995007395243217,
      "learning_rate": 2.1105068950839414e-07,
      "loss": 0.7843,
      "step": 12402
    },
    {
      "epoch": 0.9100030264955933,
      "grad_norm": 1.6089495259164692,
      "learning_rate": 2.1070925463496928e-07,
      "loss": 0.7181,
      "step": 12403
    },
    {
      "epoch": 0.9100763960857323,
      "grad_norm": 1.7316217343043125,
      "learning_rate": 2.10368090224653e-07,
      "loss": 0.764,
      "step": 12404
    },
    {
      "epoch": 0.9101497656758715,
      "grad_norm": 1.4702378709478376,
      "learning_rate": 2.1002719629671154e-07,
      "loss": 0.7044,
      "step": 12405
    },
    {
      "epoch": 0.9102231352660106,
      "grad_norm": 1.6550224934853022,
      "learning_rate": 2.0968657287039674e-07,
      "loss": 0.7975,
      "step": 12406
    },
    {
      "epoch": 0.9102965048561498,
      "grad_norm": 1.4924404548679326,
      "learning_rate": 2.0934621996494486e-07,
      "loss": 0.7232,
      "step": 12407
    },
    {
      "epoch": 0.9103698744462889,
      "grad_norm": 0.5544812015818718,
      "learning_rate": 2.09006137599575e-07,
      "loss": 0.4823,
      "step": 12408
    },
    {
      "epoch": 0.9103698744462889,
      "lm_loss": 0.2766,
      "step": 12408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9103698744462889,
      "lm_loss": 0.2258,
      "step": 12408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9103698744462889,
      "lm_loss": 0.4608,
      "step": 12408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9103698744462889,
      "lm_loss": 0.5773,
      "step": 12408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9103698744462889,
      "lm_loss": 0.3985,
      "step": 12408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9103698744462889,
      "lm_loss": 0.5326,
      "step": 12408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9103698744462889,
      "lm_loss": 0.5184,
      "step": 12408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9103698744462889,
      "lm_loss": 0.3977,
      "step": 12408,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.910443244036428,
      "grad_norm": 0.5563551349958546,
      "learning_rate": 2.08666325793494e-07,
      "loss": 0.4683,
      "step": 12409
    },
    {
      "epoch": 0.9105166136265671,
      "grad_norm": 1.46790269057825,
      "learning_rate": 2.0832678456589095e-07,
      "loss": 0.6717,
      "step": 12410
    },
    {
      "epoch": 0.9105899832167063,
      "grad_norm": 0.5472198368690832,
      "learning_rate": 2.0798751393593996e-07,
      "loss": 0.4714,
      "step": 12411
    },
    {
      "epoch": 0.9106633528068454,
      "grad_norm": 0.5455903150575387,
      "learning_rate": 2.0764851392280183e-07,
      "loss": 0.4906,
      "step": 12412
    },
    {
      "epoch": 0.9107367223969846,
      "grad_norm": 1.4044674845312723,
      "learning_rate": 2.0730978454562012e-07,
      "loss": 0.7287,
      "step": 12413
    },
    {
      "epoch": 0.9108100919871236,
      "grad_norm": 0.5471619465635196,
      "learning_rate": 2.0697132582352397e-07,
      "loss": 0.4613,
      "step": 12414
    },
    {
      "epoch": 0.9108834615772627,
      "grad_norm": 1.5293257808404797,
      "learning_rate": 2.0663313777562644e-07,
      "loss": 0.7841,
      "step": 12415
    },
    {
      "epoch": 0.9109568311674019,
      "grad_norm": 1.5519449486678367,
      "learning_rate": 2.0629522042102613e-07,
      "loss": 0.6602,
      "step": 12416
    },
    {
      "epoch": 0.9109568311674019,
      "lm_loss": 0.2482,
      "step": 12416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9109568311674019,
      "lm_loss": 0.4267,
      "step": 12416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9109568311674019,
      "lm_loss": 0.4646,
      "step": 12416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9109568311674019,
      "lm_loss": 0.5036,
      "step": 12416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9109568311674019,
      "lm_loss": 0.3931,
      "step": 12416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9109568311674019,
      "lm_loss": 0.4232,
      "step": 12416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9109568311674019,
      "lm_loss": 0.3967,
      "step": 12416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9109568311674019,
      "lm_loss": 0.3969,
      "step": 12416,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.911030200757541,
      "grad_norm": 0.5408482861673866,
      "learning_rate": 2.0595757377880665e-07,
      "loss": 0.4764,
      "step": 12417
    },
    {
      "epoch": 0.9111035703476802,
      "grad_norm": 0.5337068650328955,
      "learning_rate": 2.05620197868035e-07,
      "loss": 0.4774,
      "step": 12418
    },
    {
      "epoch": 0.9111769399378192,
      "grad_norm": 0.5422705682465352,
      "learning_rate": 2.052830927077637e-07,
      "loss": 0.4662,
      "step": 12419
    },
    {
      "epoch": 0.9112503095279584,
      "grad_norm": 1.4841995716737986,
      "learning_rate": 2.049462583170303e-07,
      "loss": 0.7774,
      "step": 12420
    },
    {
      "epoch": 0.9113236791180975,
      "grad_norm": 1.70684321820567,
      "learning_rate": 2.0460969471485682e-07,
      "loss": 0.783,
      "step": 12421
    },
    {
      "epoch": 0.9113970487082367,
      "grad_norm": 1.5173875100482044,
      "learning_rate": 2.0427340192024803e-07,
      "loss": 0.7822,
      "step": 12422
    },
    {
      "epoch": 0.9114704182983758,
      "grad_norm": 1.393423710693005,
      "learning_rate": 2.0393737995219765e-07,
      "loss": 0.7151,
      "step": 12423
    },
    {
      "epoch": 0.9115437878885149,
      "grad_norm": 1.4215151312167411,
      "learning_rate": 2.0360162882968105e-07,
      "loss": 0.7537,
      "step": 12424
    },
    {
      "epoch": 0.9115437878885149,
      "lm_loss": 1.2968,
      "step": 12424,
      "vm_loss": 0.1167
    },
    {
      "epoch": 0.9115437878885149,
      "lm_loss": 0.515,
      "step": 12424,
      "vm_loss": 0.0698
    },
    {
      "epoch": 0.9115437878885149,
      "lm_loss": 1.005,
      "step": 12424,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.9115437878885149,
      "lm_loss": 1.0006,
      "step": 12424,
      "vm_loss": 0.2195
    },
    {
      "epoch": 0.9115437878885149,
      "lm_loss": 0.4541,
      "step": 12424,
      "vm_loss": 0.1885
    },
    {
      "epoch": 0.9115437878885149,
      "lm_loss": 0.5109,
      "step": 12424,
      "vm_loss": 0.1186
    },
    {
      "epoch": 0.9115437878885149,
      "lm_loss": 1.256,
      "step": 12424,
      "vm_loss": 0.1061
    },
    {
      "epoch": 0.9115437878885149,
      "lm_loss": 1.0233,
      "step": 12424,
      "vm_loss": 0.1233
    },
    {
      "epoch": 0.911617157478654,
      "grad_norm": 1.5271283339124038,
      "learning_rate": 2.0326614857165806e-07,
      "loss": 0.6253,
      "step": 12425
    },
    {
      "epoch": 0.9116905270687932,
      "grad_norm": 0.5456691526372828,
      "learning_rate": 2.0293093919707463e-07,
      "loss": 0.474,
      "step": 12426
    },
    {
      "epoch": 0.9117638966589323,
      "grad_norm": 1.4179897614497896,
      "learning_rate": 2.0259600072486064e-07,
      "loss": 0.7848,
      "step": 12427
    },
    {
      "epoch": 0.9118372662490715,
      "grad_norm": 1.531750890467328,
      "learning_rate": 2.0226133317393094e-07,
      "loss": 0.6796,
      "step": 12428
    },
    {
      "epoch": 0.9119106358392105,
      "grad_norm": 1.52526076348798,
      "learning_rate": 2.0192693656318597e-07,
      "loss": 0.7284,
      "step": 12429
    },
    {
      "epoch": 0.9119840054293497,
      "grad_norm": 0.5508727376756734,
      "learning_rate": 2.015928109115084e-07,
      "loss": 0.4762,
      "step": 12430
    },
    {
      "epoch": 0.9120573750194888,
      "grad_norm": 1.6992479025894358,
      "learning_rate": 2.0125895623776815e-07,
      "loss": 0.7663,
      "step": 12431
    },
    {
      "epoch": 0.912130744609628,
      "grad_norm": 1.4614560832773793,
      "learning_rate": 2.009253725608179e-07,
      "loss": 0.7186,
      "step": 12432
    },
    {
      "epoch": 0.912130744609628,
      "lm_loss": 1.1559,
      "step": 12432,
      "vm_loss": 0.0755
    },
    {
      "epoch": 0.912130744609628,
      "lm_loss": 1.563,
      "step": 12432,
      "vm_loss": 0.1357
    },
    {
      "epoch": 0.912130744609628,
      "lm_loss": 0.6317,
      "step": 12432,
      "vm_loss": 0.0631
    },
    {
      "epoch": 0.912130744609628,
      "lm_loss": 0.4323,
      "step": 12432,
      "vm_loss": 0.0626
    },
    {
      "epoch": 0.912130744609628,
      "lm_loss": 0.0522,
      "step": 12432,
      "vm_loss": 0.1295
    },
    {
      "epoch": 0.912130744609628,
      "lm_loss": 1.22,
      "step": 12432,
      "vm_loss": 0.1824
    },
    {
      "epoch": 0.912130744609628,
      "lm_loss": 0.4375,
      "step": 12432,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.912130744609628,
      "lm_loss": 0.8168,
      "step": 12432,
      "vm_loss": 0.1193
    },
    {
      "epoch": 0.9122041141997671,
      "grad_norm": 1.7009936690636944,
      "learning_rate": 2.0059205989949703e-07,
      "loss": 0.7789,
      "step": 12433
    },
    {
      "epoch": 0.9122774837899061,
      "grad_norm": 1.4731809728578973,
      "learning_rate": 2.0025901827262828e-07,
      "loss": 0.6756,
      "step": 12434
    },
    {
      "epoch": 0.9123508533800453,
      "grad_norm": 1.435608111535983,
      "learning_rate": 1.9992624769901935e-07,
      "loss": 0.789,
      "step": 12435
    },
    {
      "epoch": 0.9124242229701844,
      "grad_norm": 1.6088189798602894,
      "learning_rate": 1.995937481974619e-07,
      "loss": 0.7298,
      "step": 12436
    },
    {
      "epoch": 0.9124975925603236,
      "grad_norm": 1.657119320008892,
      "learning_rate": 1.9926151978673368e-07,
      "loss": 0.7572,
      "step": 12437
    },
    {
      "epoch": 0.9125709621504627,
      "grad_norm": 1.5790824781278772,
      "learning_rate": 1.989295624855958e-07,
      "loss": 0.7007,
      "step": 12438
    },
    {
      "epoch": 0.9126443317406018,
      "grad_norm": 1.6803179456037365,
      "learning_rate": 1.9859787631279547e-07,
      "loss": 0.7595,
      "step": 12439
    },
    {
      "epoch": 0.9127177013307409,
      "grad_norm": 1.6969556424809766,
      "learning_rate": 1.9826646128706385e-07,
      "loss": 0.7825,
      "step": 12440
    },
    {
      "epoch": 0.9127177013307409,
      "lm_loss": 1.3298,
      "step": 12440,
      "vm_loss": 0.0646
    },
    {
      "epoch": 0.9127177013307409,
      "lm_loss": 0.7072,
      "step": 12440,
      "vm_loss": 0.1092
    },
    {
      "epoch": 0.9127177013307409,
      "lm_loss": 1.0819,
      "step": 12440,
      "vm_loss": 0.1107
    },
    {
      "epoch": 0.9127177013307409,
      "lm_loss": 0.4021,
      "step": 12440,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.9127177013307409,
      "lm_loss": 0.1451,
      "step": 12440,
      "vm_loss": 0.1718
    },
    {
      "epoch": 0.9127177013307409,
      "lm_loss": 1.0673,
      "step": 12440,
      "vm_loss": 0.124
    },
    {
      "epoch": 0.9127177013307409,
      "lm_loss": 0.5321,
      "step": 12440,
      "vm_loss": 0.1328
    },
    {
      "epoch": 0.9127177013307409,
      "lm_loss": 0.9331,
      "step": 12440,
      "vm_loss": 0.1272
    },
    {
      "epoch": 0.9127910709208801,
      "grad_norm": 1.5122503558621279,
      "learning_rate": 1.9793531742711646e-07,
      "loss": 0.7176,
      "step": 12441
    },
    {
      "epoch": 0.9128644405110192,
      "grad_norm": 1.4475311902843977,
      "learning_rate": 1.976044447516534e-07,
      "loss": 0.7141,
      "step": 12442
    },
    {
      "epoch": 0.9129378101011584,
      "grad_norm": 0.5432666152648491,
      "learning_rate": 1.972738432793603e-07,
      "loss": 0.4806,
      "step": 12443
    },
    {
      "epoch": 0.9130111796912974,
      "grad_norm": 1.3990022382774197,
      "learning_rate": 1.9694351302890712e-07,
      "loss": 0.7154,
      "step": 12444
    },
    {
      "epoch": 0.9130845492814366,
      "grad_norm": 1.4921231411812987,
      "learning_rate": 1.966134540189485e-07,
      "loss": 0.7302,
      "step": 12445
    },
    {
      "epoch": 0.9131579188715757,
      "grad_norm": 1.4196092032925407,
      "learning_rate": 1.9628366626812335e-07,
      "loss": 0.7079,
      "step": 12446
    },
    {
      "epoch": 0.9132312884617149,
      "grad_norm": 1.6351637343135002,
      "learning_rate": 1.9595414979505622e-07,
      "loss": 0.7747,
      "step": 12447
    },
    {
      "epoch": 0.913304658051854,
      "grad_norm": 1.5524410374519972,
      "learning_rate": 1.9562490461835504e-07,
      "loss": 0.7343,
      "step": 12448
    },
    {
      "epoch": 0.913304658051854,
      "lm_loss": 0.4925,
      "step": 12448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.913304658051854,
      "lm_loss": 0.4247,
      "step": 12448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.913304658051854,
      "lm_loss": 0.3332,
      "step": 12448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.913304658051854,
      "lm_loss": 0.5818,
      "step": 12448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.913304658051854,
      "lm_loss": 0.454,
      "step": 12448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.913304658051854,
      "lm_loss": 0.5341,
      "step": 12448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.913304658051854,
      "lm_loss": 0.4489,
      "step": 12448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.913304658051854,
      "lm_loss": 0.6016,
      "step": 12448,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.913378027641993,
      "grad_norm": 0.569364057959736,
      "learning_rate": 1.9529593075661267e-07,
      "loss": 0.4742,
      "step": 12449
    },
    {
      "epoch": 0.9134513972321322,
      "grad_norm": 1.4975712941652373,
      "learning_rate": 1.9496722822840819e-07,
      "loss": 0.8193,
      "step": 12450
    },
    {
      "epoch": 0.9135247668222714,
      "grad_norm": 1.545126167491842,
      "learning_rate": 1.9463879705230392e-07,
      "loss": 0.7815,
      "step": 12451
    },
    {
      "epoch": 0.9135981364124105,
      "grad_norm": 1.6073919494933462,
      "learning_rate": 1.943106372468473e-07,
      "loss": 0.7944,
      "step": 12452
    },
    {
      "epoch": 0.9136715060025495,
      "grad_norm": 0.5450291132893299,
      "learning_rate": 1.9398274883056955e-07,
      "loss": 0.4727,
      "step": 12453
    },
    {
      "epoch": 0.9137448755926887,
      "grad_norm": 0.5564882348224489,
      "learning_rate": 1.9365513182198758e-07,
      "loss": 0.4841,
      "step": 12454
    },
    {
      "epoch": 0.9138182451828278,
      "grad_norm": 1.4747847635903484,
      "learning_rate": 1.933277862396038e-07,
      "loss": 0.756,
      "step": 12455
    },
    {
      "epoch": 0.913891614772967,
      "grad_norm": 0.529223699860807,
      "learning_rate": 1.9300071210190342e-07,
      "loss": 0.4523,
      "step": 12456
    },
    {
      "epoch": 0.913891614772967,
      "lm_loss": 1.4933,
      "step": 12456,
      "vm_loss": 0.0638
    },
    {
      "epoch": 0.913891614772967,
      "lm_loss": 0.7272,
      "step": 12456,
      "vm_loss": 0.1281
    },
    {
      "epoch": 0.913891614772967,
      "lm_loss": 0.1437,
      "step": 12456,
      "vm_loss": 0.1274
    },
    {
      "epoch": 0.913891614772967,
      "lm_loss": 0.3456,
      "step": 12456,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.913891614772967,
      "lm_loss": 0.7761,
      "step": 12456,
      "vm_loss": 0.1489
    },
    {
      "epoch": 0.913891614772967,
      "lm_loss": 0.4765,
      "step": 12456,
      "vm_loss": 0.0716
    },
    {
      "epoch": 0.913891614772967,
      "lm_loss": 0.0249,
      "step": 12456,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.913891614772967,
      "lm_loss": 0.0005,
      "step": 12456,
      "vm_loss": 0.1756
    },
    {
      "epoch": 0.9139649843631061,
      "grad_norm": 1.45646996655034,
      "learning_rate": 1.9267390942735664e-07,
      "loss": 0.6531,
      "step": 12457
    },
    {
      "epoch": 0.9140383539532452,
      "grad_norm": 1.7604854374763144,
      "learning_rate": 1.9234737823441985e-07,
      "loss": 0.7971,
      "step": 12458
    },
    {
      "epoch": 0.9141117235433843,
      "grad_norm": 1.5029954460731665,
      "learning_rate": 1.9202111854153217e-07,
      "loss": 0.6961,
      "step": 12459
    },
    {
      "epoch": 0.9141850931335235,
      "grad_norm": 1.553923429407731,
      "learning_rate": 1.9169513036711884e-07,
      "loss": 0.6879,
      "step": 12460
    },
    {
      "epoch": 0.9142584627236626,
      "grad_norm": 1.5183707458697782,
      "learning_rate": 1.9136941372958906e-07,
      "loss": 0.6245,
      "step": 12461
    },
    {
      "epoch": 0.9143318323138018,
      "grad_norm": 1.3620648116876715,
      "learning_rate": 1.9104396864733698e-07,
      "loss": 0.7108,
      "step": 12462
    },
    {
      "epoch": 0.9144052019039408,
      "grad_norm": 1.5611389049928188,
      "learning_rate": 1.9071879513874124e-07,
      "loss": 0.7842,
      "step": 12463
    },
    {
      "epoch": 0.91447857149408,
      "grad_norm": 1.538233006318669,
      "learning_rate": 1.9039389322216495e-07,
      "loss": 0.7627,
      "step": 12464
    },
    {
      "epoch": 0.91447857149408,
      "lm_loss": 1.0546,
      "step": 12464,
      "vm_loss": 0.0533
    },
    {
      "epoch": 0.91447857149408,
      "lm_loss": 1.5321,
      "step": 12464,
      "vm_loss": 0.1502
    },
    {
      "epoch": 0.91447857149408,
      "lm_loss": 1.2437,
      "step": 12464,
      "vm_loss": 0.173
    },
    {
      "epoch": 0.91447857149408,
      "lm_loss": 0.0285,
      "step": 12464,
      "vm_loss": 0.0823
    },
    {
      "epoch": 0.91447857149408,
      "lm_loss": 0.3647,
      "step": 12464,
      "vm_loss": 0.103
    },
    {
      "epoch": 0.91447857149408,
      "lm_loss": 0.088,
      "step": 12464,
      "vm_loss": 0.1123
    },
    {
      "epoch": 0.91447857149408,
      "lm_loss": 0.7427,
      "step": 12464,
      "vm_loss": 0.0996
    },
    {
      "epoch": 0.91447857149408,
      "lm_loss": 0.9737,
      "step": 12464,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.9145519410842191,
      "grad_norm": 1.5769213907588502,
      "learning_rate": 1.9006926291595562e-07,
      "loss": 0.7358,
      "step": 12465
    },
    {
      "epoch": 0.9146253106743583,
      "grad_norm": 1.5062444158637414,
      "learning_rate": 1.8974490423844748e-07,
      "loss": 0.685,
      "step": 12466
    },
    {
      "epoch": 0.9146986802644974,
      "grad_norm": 1.5250103706659186,
      "learning_rate": 1.8942081720795703e-07,
      "loss": 0.7474,
      "step": 12467
    },
    {
      "epoch": 0.9147720498546364,
      "grad_norm": 1.4970533581060628,
      "learning_rate": 1.8909700184278623e-07,
      "loss": 0.6917,
      "step": 12468
    },
    {
      "epoch": 0.9148454194447756,
      "grad_norm": 1.4370894262579588,
      "learning_rate": 1.8877345816122162e-07,
      "loss": 0.734,
      "step": 12469
    },
    {
      "epoch": 0.9149187890349147,
      "grad_norm": 1.567935944517623,
      "learning_rate": 1.884501861815341e-07,
      "loss": 0.8029,
      "step": 12470
    },
    {
      "epoch": 0.9149921586250539,
      "grad_norm": 1.7182186194650262,
      "learning_rate": 1.8812718592198076e-07,
      "loss": 0.682,
      "step": 12471
    },
    {
      "epoch": 0.915065528215193,
      "grad_norm": 1.2880295391647079,
      "learning_rate": 1.8780445740080145e-07,
      "loss": 0.5892,
      "step": 12472
    },
    {
      "epoch": 0.915065528215193,
      "lm_loss": 1.2971,
      "step": 12472,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.915065528215193,
      "lm_loss": 0.5724,
      "step": 12472,
      "vm_loss": 0.1016
    },
    {
      "epoch": 0.915065528215193,
      "lm_loss": 0.439,
      "step": 12472,
      "vm_loss": 0.1872
    },
    {
      "epoch": 0.915065528215193,
      "lm_loss": 1.009,
      "step": 12472,
      "vm_loss": 0.1722
    },
    {
      "epoch": 0.915065528215193,
      "lm_loss": 0.831,
      "step": 12472,
      "vm_loss": 0.1508
    },
    {
      "epoch": 0.915065528215193,
      "lm_loss": 0.5589,
      "step": 12472,
      "vm_loss": 0.1214
    },
    {
      "epoch": 0.915065528215193,
      "lm_loss": 0.0145,
      "step": 12472,
      "vm_loss": 0.1554
    },
    {
      "epoch": 0.915065528215193,
      "lm_loss": 0.5208,
      "step": 12472,
      "vm_loss": 0.1412
    },
    {
      "epoch": 0.9151388978053321,
      "grad_norm": 1.501434703992005,
      "learning_rate": 1.8748200063622157e-07,
      "loss": 0.6892,
      "step": 12473
    },
    {
      "epoch": 0.9152122673954712,
      "grad_norm": 0.5578150296707958,
      "learning_rate": 1.8715981564645103e-07,
      "loss": 0.4903,
      "step": 12474
    },
    {
      "epoch": 0.9152856369856104,
      "grad_norm": 0.542637256348151,
      "learning_rate": 1.8683790244968525e-07,
      "loss": 0.4689,
      "step": 12475
    },
    {
      "epoch": 0.9153590065757495,
      "grad_norm": 1.3985510439483733,
      "learning_rate": 1.8651626106410136e-07,
      "loss": 0.6942,
      "step": 12476
    },
    {
      "epoch": 0.9154323761658887,
      "grad_norm": 1.5666242621524182,
      "learning_rate": 1.861948915078654e-07,
      "loss": 0.7694,
      "step": 12477
    },
    {
      "epoch": 0.9155057457560277,
      "grad_norm": 1.586255712562557,
      "learning_rate": 1.8587379379912506e-07,
      "loss": 0.736,
      "step": 12478
    },
    {
      "epoch": 0.9155791153461669,
      "grad_norm": 1.618841516652542,
      "learning_rate": 1.8555296795601364e-07,
      "loss": 0.661,
      "step": 12479
    },
    {
      "epoch": 0.915652484936306,
      "grad_norm": 1.4382257542187469,
      "learning_rate": 1.8523241399664937e-07,
      "loss": 0.7029,
      "step": 12480
    },
    {
      "epoch": 0.915652484936306,
      "lm_loss": 0.926,
      "step": 12480,
      "vm_loss": 0.1296
    },
    {
      "epoch": 0.915652484936306,
      "lm_loss": 1.1423,
      "step": 12480,
      "vm_loss": 0.1288
    },
    {
      "epoch": 0.915652484936306,
      "lm_loss": 0.0504,
      "step": 12480,
      "vm_loss": 0.1322
    },
    {
      "epoch": 0.915652484936306,
      "lm_loss": 0.2866,
      "step": 12480,
      "vm_loss": 0.2048
    },
    {
      "epoch": 0.915652484936306,
      "lm_loss": 0.9671,
      "step": 12480,
      "vm_loss": 0.1617
    },
    {
      "epoch": 0.915652484936306,
      "lm_loss": 0.9501,
      "step": 12480,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.915652484936306,
      "lm_loss": 0.1715,
      "step": 12480,
      "vm_loss": 0.1873
    },
    {
      "epoch": 0.915652484936306,
      "lm_loss": 1.641,
      "step": 12480,
      "vm_loss": 0.1613
    },
    {
      "epoch": 0.9157258545264452,
      "grad_norm": 1.7473042874393943,
      "learning_rate": 1.8491213193913283e-07,
      "loss": 0.767,
      "step": 12481
    },
    {
      "epoch": 0.9157992241165843,
      "grad_norm": 0.5532302691234858,
      "learning_rate": 1.845921218015534e-07,
      "loss": 0.4822,
      "step": 12482
    },
    {
      "epoch": 0.9158725937067234,
      "grad_norm": 1.539173060208519,
      "learning_rate": 1.842723836019822e-07,
      "loss": 0.7834,
      "step": 12483
    },
    {
      "epoch": 0.9159459632968625,
      "grad_norm": 1.5389322901657945,
      "learning_rate": 1.8395291735847588e-07,
      "loss": 0.6944,
      "step": 12484
    },
    {
      "epoch": 0.9160193328870017,
      "grad_norm": 0.5319236519364569,
      "learning_rate": 1.8363372308907445e-07,
      "loss": 0.469,
      "step": 12485
    },
    {
      "epoch": 0.9160927024771408,
      "grad_norm": 1.4866169014078296,
      "learning_rate": 1.8331480081180354e-07,
      "loss": 0.7687,
      "step": 12486
    },
    {
      "epoch": 0.91616607206728,
      "grad_norm": 0.5583716385401624,
      "learning_rate": 1.829961505446748e-07,
      "loss": 0.4783,
      "step": 12487
    },
    {
      "epoch": 0.916239441657419,
      "grad_norm": 1.7031255793755897,
      "learning_rate": 1.8267777230568274e-07,
      "loss": 0.6202,
      "step": 12488
    },
    {
      "epoch": 0.916239441657419,
      "lm_loss": 0.4295,
      "step": 12488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.916239441657419,
      "lm_loss": 0.4006,
      "step": 12488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.916239441657419,
      "lm_loss": 0.409,
      "step": 12488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.916239441657419,
      "lm_loss": 0.5015,
      "step": 12488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.916239441657419,
      "lm_loss": 0.3968,
      "step": 12488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.916239441657419,
      "lm_loss": 0.4041,
      "step": 12488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.916239441657419,
      "lm_loss": 0.6212,
      "step": 12488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.916239441657419,
      "lm_loss": 0.5354,
      "step": 12488,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9163128112475581,
      "grad_norm": 0.5369896826979124,
      "learning_rate": 1.8235966611280687e-07,
      "loss": 0.4791,
      "step": 12489
    },
    {
      "epoch": 0.9163861808376973,
      "grad_norm": 0.5308870916175206,
      "learning_rate": 1.8204183198401115e-07,
      "loss": 0.4748,
      "step": 12490
    },
    {
      "epoch": 0.9164595504278364,
      "grad_norm": 1.4626277452110292,
      "learning_rate": 1.8172426993724457e-07,
      "loss": 0.6415,
      "step": 12491
    },
    {
      "epoch": 0.9165329200179756,
      "grad_norm": 1.4699333083307768,
      "learning_rate": 1.8140697999044056e-07,
      "loss": 0.7571,
      "step": 12492
    },
    {
      "epoch": 0.9166062896081146,
      "grad_norm": 1.4886310949560728,
      "learning_rate": 1.8108996216151752e-07,
      "loss": 0.6758,
      "step": 12493
    },
    {
      "epoch": 0.9166796591982538,
      "grad_norm": 1.4862672861331476,
      "learning_rate": 1.8077321646837843e-07,
      "loss": 0.7474,
      "step": 12494
    },
    {
      "epoch": 0.9167530287883929,
      "grad_norm": 1.5466267065938755,
      "learning_rate": 1.8045674292891001e-07,
      "loss": 0.8306,
      "step": 12495
    },
    {
      "epoch": 0.9168263983785321,
      "grad_norm": 0.5432552517408622,
      "learning_rate": 1.8014054156098526e-07,
      "loss": 0.486,
      "step": 12496
    },
    {
      "epoch": 0.9168263983785321,
      "lm_loss": 0.5002,
      "step": 12496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9168263983785321,
      "lm_loss": 0.3565,
      "step": 12496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9168263983785321,
      "lm_loss": 0.5048,
      "step": 12496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9168263983785321,
      "lm_loss": 0.4967,
      "step": 12496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9168263983785321,
      "lm_loss": 0.6394,
      "step": 12496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9168263983785321,
      "lm_loss": 0.286,
      "step": 12496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9168263983785321,
      "lm_loss": 0.5866,
      "step": 12496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9168263983785321,
      "lm_loss": 0.5317,
      "step": 12496,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9168997679686712,
      "grad_norm": 0.5506078902696399,
      "learning_rate": 1.7982461238245928e-07,
      "loss": 0.4949,
      "step": 12497
    },
    {
      "epoch": 0.9169731375588103,
      "grad_norm": 1.3829224240213263,
      "learning_rate": 1.7950895541117565e-07,
      "loss": 0.7028,
      "step": 12498
    },
    {
      "epoch": 0.9170465071489494,
      "grad_norm": 1.5532003799679601,
      "learning_rate": 1.7919357066495836e-07,
      "loss": 0.6847,
      "step": 12499
    },
    {
      "epoch": 0.9171198767390886,
      "grad_norm": 1.6690998460184576,
      "learning_rate": 1.7887845816161875e-07,
      "loss": 0.5936,
      "step": 12500
    },
    {
      "epoch": 0.9171932463292277,
      "grad_norm": 1.502490442182407,
      "learning_rate": 1.7856361791895205e-07,
      "loss": 0.7315,
      "step": 12501
    },
    {
      "epoch": 0.9172666159193669,
      "grad_norm": 1.6609697020129564,
      "learning_rate": 1.7824904995473791e-07,
      "loss": 0.6991,
      "step": 12502
    },
    {
      "epoch": 0.9173399855095059,
      "grad_norm": 0.5451626705794789,
      "learning_rate": 1.77934754286741e-07,
      "loss": 0.4665,
      "step": 12503
    },
    {
      "epoch": 0.9174133550996451,
      "grad_norm": 1.580209851395667,
      "learning_rate": 1.7762073093270993e-07,
      "loss": 0.7147,
      "step": 12504
    },
    {
      "epoch": 0.9174133550996451,
      "lm_loss": 1.3786,
      "step": 12504,
      "vm_loss": 0.0728
    },
    {
      "epoch": 0.9174133550996451,
      "lm_loss": 0.8164,
      "step": 12504,
      "vm_loss": 0.0981
    },
    {
      "epoch": 0.9174133550996451,
      "lm_loss": 0.1414,
      "step": 12504,
      "vm_loss": 0.1306
    },
    {
      "epoch": 0.9174133550996451,
      "lm_loss": 0.8447,
      "step": 12504,
      "vm_loss": 0.1975
    },
    {
      "epoch": 0.9174133550996451,
      "lm_loss": 0.079,
      "step": 12504,
      "vm_loss": 0.1048
    },
    {
      "epoch": 0.9174133550996451,
      "lm_loss": 0.5138,
      "step": 12504,
      "vm_loss": 0.1437
    },
    {
      "epoch": 0.9174133550996451,
      "lm_loss": 0.7271,
      "step": 12504,
      "vm_loss": 0.1775
    },
    {
      "epoch": 0.9174133550996451,
      "lm_loss": 0.9756,
      "step": 12504,
      "vm_loss": 0.2531
    },
    {
      "epoch": 0.9174867246897842,
      "grad_norm": 1.401540418657323,
      "learning_rate": 1.7730697991037883e-07,
      "loss": 0.7122,
      "step": 12505
    },
    {
      "epoch": 0.9175600942799234,
      "grad_norm": 0.5427028635659331,
      "learning_rate": 1.7699350123746573e-07,
      "loss": 0.4658,
      "step": 12506
    },
    {
      "epoch": 0.9176334638700625,
      "grad_norm": 1.7228637786626422,
      "learning_rate": 1.766802949316726e-07,
      "loss": 0.6775,
      "step": 12507
    },
    {
      "epoch": 0.9177068334602015,
      "grad_norm": 1.4758970195213825,
      "learning_rate": 1.7636736101068862e-07,
      "loss": 0.8205,
      "step": 12508
    },
    {
      "epoch": 0.9177802030503407,
      "grad_norm": 0.5623490821808261,
      "learning_rate": 1.760546994921858e-07,
      "loss": 0.4841,
      "step": 12509
    },
    {
      "epoch": 0.9178535726404798,
      "grad_norm": 1.5556101519144012,
      "learning_rate": 1.7574231039381995e-07,
      "loss": 0.7723,
      "step": 12510
    },
    {
      "epoch": 0.917926942230619,
      "grad_norm": 1.5720274809571322,
      "learning_rate": 1.754301937332331e-07,
      "loss": 0.6489,
      "step": 12511
    },
    {
      "epoch": 0.9180003118207581,
      "grad_norm": 0.5538492686451593,
      "learning_rate": 1.7511834952805063e-07,
      "loss": 0.4541,
      "step": 12512
    },
    {
      "epoch": 0.9180003118207581,
      "lm_loss": 0.455,
      "step": 12512,
      "vm_loss": 0.0308
    },
    {
      "epoch": 0.9180003118207581,
      "lm_loss": 0.4118,
      "step": 12512,
      "vm_loss": 0.0763
    },
    {
      "epoch": 0.9180003118207581,
      "lm_loss": 0.9972,
      "step": 12512,
      "vm_loss": 0.0954
    },
    {
      "epoch": 0.9180003118207581,
      "lm_loss": 1.0232,
      "step": 12512,
      "vm_loss": 0.085
    },
    {
      "epoch": 0.9180003118207581,
      "lm_loss": 0.4355,
      "step": 12512,
      "vm_loss": 0.0558
    },
    {
      "epoch": 0.9180003118207581,
      "lm_loss": 0.477,
      "step": 12512,
      "vm_loss": 0.0974
    },
    {
      "epoch": 0.9180003118207581,
      "lm_loss": 0.7408,
      "step": 12512,
      "vm_loss": 0.1058
    },
    {
      "epoch": 0.9180003118207581,
      "lm_loss": 0.0876,
      "step": 12512,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.9180736814108972,
      "grad_norm": 1.6511407096407211,
      "learning_rate": 1.7480677779588396e-07,
      "loss": 0.6887,
      "step": 12513
    },
    {
      "epoch": 0.9181470510010363,
      "grad_norm": 1.3206016181832487,
      "learning_rate": 1.7449547855432792e-07,
      "loss": 0.7316,
      "step": 12514
    },
    {
      "epoch": 0.9182204205911755,
      "grad_norm": 1.5137854091163752,
      "learning_rate": 1.7418445182096232e-07,
      "loss": 0.7402,
      "step": 12515
    },
    {
      "epoch": 0.9182937901813146,
      "grad_norm": 1.584517222605203,
      "learning_rate": 1.7387369761335148e-07,
      "loss": 0.9074,
      "step": 12516
    },
    {
      "epoch": 0.9183671597714538,
      "grad_norm": 1.6203503422672538,
      "learning_rate": 1.7356321594904524e-07,
      "loss": 0.6848,
      "step": 12517
    },
    {
      "epoch": 0.9184405293615928,
      "grad_norm": 1.497811687928828,
      "learning_rate": 1.7325300684557567e-07,
      "loss": 0.8023,
      "step": 12518
    },
    {
      "epoch": 0.918513898951732,
      "grad_norm": 1.3238253641935103,
      "learning_rate": 1.7294307032046264e-07,
      "loss": 0.6846,
      "step": 12519
    },
    {
      "epoch": 0.9185872685418711,
      "grad_norm": 0.5359369998384335,
      "learning_rate": 1.7263340639120828e-07,
      "loss": 0.4695,
      "step": 12520
    },
    {
      "epoch": 0.9185872685418711,
      "lm_loss": 1.5049,
      "step": 12520,
      "vm_loss": 0.0951
    },
    {
      "epoch": 0.9185872685418711,
      "lm_loss": 0.2173,
      "step": 12520,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.9185872685418711,
      "lm_loss": 0.4866,
      "step": 12520,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.9185872685418711,
      "lm_loss": 0.7498,
      "step": 12520,
      "vm_loss": 0.0997
    },
    {
      "epoch": 0.9185872685418711,
      "lm_loss": 0.44,
      "step": 12520,
      "vm_loss": 0.1758
    },
    {
      "epoch": 0.9185872685418711,
      "lm_loss": 0.895,
      "step": 12520,
      "vm_loss": 0.1214
    },
    {
      "epoch": 0.9185872685418711,
      "lm_loss": 0.9874,
      "step": 12520,
      "vm_loss": 0.2076
    },
    {
      "epoch": 0.9185872685418711,
      "lm_loss": 0.4074,
      "step": 12520,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.9186606381320103,
      "grad_norm": 1.6432556337402475,
      "learning_rate": 1.7232401507530028e-07,
      "loss": 0.7234,
      "step": 12521
    },
    {
      "epoch": 0.9187340077221494,
      "grad_norm": 1.529005800455387,
      "learning_rate": 1.7201489639021073e-07,
      "loss": 0.6537,
      "step": 12522
    },
    {
      "epoch": 0.9188073773122885,
      "grad_norm": 1.7714189749969311,
      "learning_rate": 1.7170605035339627e-07,
      "loss": 0.7526,
      "step": 12523
    },
    {
      "epoch": 0.9188807469024276,
      "grad_norm": 0.549330193297091,
      "learning_rate": 1.713974769822979e-07,
      "loss": 0.4688,
      "step": 12524
    },
    {
      "epoch": 0.9189541164925668,
      "grad_norm": 0.5465753208185521,
      "learning_rate": 1.710891762943423e-07,
      "loss": 0.473,
      "step": 12525
    },
    {
      "epoch": 0.9190274860827059,
      "grad_norm": 1.3606407143240196,
      "learning_rate": 1.7078114830693936e-07,
      "loss": 0.716,
      "step": 12526
    },
    {
      "epoch": 0.919100855672845,
      "grad_norm": 1.4736879644547514,
      "learning_rate": 1.7047339303748467e-07,
      "loss": 0.7709,
      "step": 12527
    },
    {
      "epoch": 0.9191742252629841,
      "grad_norm": 1.5803761851598068,
      "learning_rate": 1.701659105033565e-07,
      "loss": 0.7201,
      "step": 12528
    },
    {
      "epoch": 0.9191742252629841,
      "lm_loss": 0.629,
      "step": 12528,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.9191742252629841,
      "lm_loss": 0.5039,
      "step": 12528,
      "vm_loss": 0.0539
    },
    {
      "epoch": 0.9191742252629841,
      "lm_loss": 1.0628,
      "step": 12528,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.9191742252629841,
      "lm_loss": 0.0818,
      "step": 12528,
      "vm_loss": 0.0855
    },
    {
      "epoch": 0.9191742252629841,
      "lm_loss": 0.7345,
      "step": 12528,
      "vm_loss": 0.1626
    },
    {
      "epoch": 0.9191742252629841,
      "lm_loss": 0.8586,
      "step": 12528,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.9191742252629841,
      "lm_loss": 0.484,
      "step": 12528,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.9191742252629841,
      "lm_loss": 0.77,
      "step": 12528,
      "vm_loss": 0.0892
    },
    {
      "epoch": 0.9192475948531232,
      "grad_norm": 1.4908518938312592,
      "learning_rate": 1.6985870072192156e-07,
      "loss": 0.7439,
      "step": 12529
    },
    {
      "epoch": 0.9193209644432624,
      "grad_norm": 0.5373211411447102,
      "learning_rate": 1.6955176371052707e-07,
      "loss": 0.4838,
      "step": 12530
    },
    {
      "epoch": 0.9193943340334015,
      "grad_norm": 1.5540067881595367,
      "learning_rate": 1.6924509948650636e-07,
      "loss": 0.6918,
      "step": 12531
    },
    {
      "epoch": 0.9194677036235406,
      "grad_norm": 1.3805031901884763,
      "learning_rate": 1.6893870806717837e-07,
      "loss": 0.7276,
      "step": 12532
    },
    {
      "epoch": 0.9195410732136797,
      "grad_norm": 1.6616805989315384,
      "learning_rate": 1.6863258946984596e-07,
      "loss": 0.7244,
      "step": 12533
    },
    {
      "epoch": 0.9196144428038189,
      "grad_norm": 1.632097698453021,
      "learning_rate": 1.683267437117947e-07,
      "loss": 0.6765,
      "step": 12534
    },
    {
      "epoch": 0.919687812393958,
      "grad_norm": 1.4643982380787526,
      "learning_rate": 1.6802117081029856e-07,
      "loss": 0.6599,
      "step": 12535
    },
    {
      "epoch": 0.9197611819840972,
      "grad_norm": 1.4197047186072704,
      "learning_rate": 1.677158707826132e-07,
      "loss": 0.6892,
      "step": 12536
    },
    {
      "epoch": 0.9197611819840972,
      "lm_loss": 0.5687,
      "step": 12536,
      "vm_loss": 0.0418
    },
    {
      "epoch": 0.9197611819840972,
      "lm_loss": 1.0624,
      "step": 12536,
      "vm_loss": 0.0879
    },
    {
      "epoch": 0.9197611819840972,
      "lm_loss": 0.6846,
      "step": 12536,
      "vm_loss": 0.1244
    },
    {
      "epoch": 0.9197611819840972,
      "lm_loss": 0.5423,
      "step": 12536,
      "vm_loss": 0.1269
    },
    {
      "epoch": 0.9197611819840972,
      "lm_loss": 0.7747,
      "step": 12536,
      "vm_loss": 0.1039
    },
    {
      "epoch": 0.9197611819840972,
      "lm_loss": 0.2214,
      "step": 12536,
      "vm_loss": 0.0644
    },
    {
      "epoch": 0.9197611819840972,
      "lm_loss": 0.0572,
      "step": 12536,
      "vm_loss": 0.1395
    },
    {
      "epoch": 0.9197611819840972,
      "lm_loss": 0.0297,
      "step": 12536,
      "vm_loss": 0.1552
    },
    {
      "epoch": 0.9198345515742362,
      "grad_norm": 1.413219561900677,
      "learning_rate": 1.6741084364597925e-07,
      "loss": 0.6916,
      "step": 12537
    },
    {
      "epoch": 0.9199079211643754,
      "grad_norm": 1.4925944096350718,
      "learning_rate": 1.6710608941762297e-07,
      "loss": 0.6884,
      "step": 12538
    },
    {
      "epoch": 0.9199812907545145,
      "grad_norm": 1.5257255583317082,
      "learning_rate": 1.6680160811475332e-07,
      "loss": 0.6865,
      "step": 12539
    },
    {
      "epoch": 0.9200546603446537,
      "grad_norm": 0.5569938476293953,
      "learning_rate": 1.6649739975456713e-07,
      "loss": 0.4989,
      "step": 12540
    },
    {
      "epoch": 0.9201280299347928,
      "grad_norm": 0.5331098910406838,
      "learning_rate": 1.6619346435424233e-07,
      "loss": 0.4778,
      "step": 12541
    },
    {
      "epoch": 0.9202013995249319,
      "grad_norm": 1.469344195571402,
      "learning_rate": 1.658898019309435e-07,
      "loss": 0.704,
      "step": 12542
    },
    {
      "epoch": 0.920274769115071,
      "grad_norm": 1.6561297409293008,
      "learning_rate": 1.655864125018186e-07,
      "loss": 0.8749,
      "step": 12543
    },
    {
      "epoch": 0.9203481387052102,
      "grad_norm": 1.5684539567984201,
      "learning_rate": 1.6528329608400174e-07,
      "loss": 0.6998,
      "step": 12544
    },
    {
      "epoch": 0.9203481387052102,
      "lm_loss": 0.8589,
      "step": 12544,
      "vm_loss": 0.0704
    },
    {
      "epoch": 0.9203481387052102,
      "lm_loss": 1.3772,
      "step": 12544,
      "vm_loss": 0.1142
    },
    {
      "epoch": 0.9203481387052102,
      "lm_loss": 0.042,
      "step": 12544,
      "vm_loss": 0.076
    },
    {
      "epoch": 0.9203481387052102,
      "lm_loss": 0.9789,
      "step": 12544,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.9203481387052102,
      "lm_loss": 0.7309,
      "step": 12544,
      "vm_loss": 0.0742
    },
    {
      "epoch": 0.9203481387052102,
      "lm_loss": 0.0077,
      "step": 12544,
      "vm_loss": 0.0927
    },
    {
      "epoch": 0.9203481387052102,
      "lm_loss": 0.1614,
      "step": 12544,
      "vm_loss": 0.0981
    },
    {
      "epoch": 0.9203481387052102,
      "lm_loss": 0.9048,
      "step": 12544,
      "vm_loss": 0.1414
    },
    {
      "epoch": 0.9204215082953493,
      "grad_norm": 1.6077340163714504,
      "learning_rate": 1.6498045269460972e-07,
      "loss": 0.8832,
      "step": 12545
    },
    {
      "epoch": 0.9204948778854884,
      "grad_norm": 1.6053741220948103,
      "learning_rate": 1.64677882350745e-07,
      "loss": 0.7907,
      "step": 12546
    },
    {
      "epoch": 0.9205682474756275,
      "grad_norm": 1.5863946837200824,
      "learning_rate": 1.6437558506949558e-07,
      "loss": 0.7252,
      "step": 12547
    },
    {
      "epoch": 0.9206416170657666,
      "grad_norm": 1.4698612623538965,
      "learning_rate": 1.640735608679317e-07,
      "loss": 0.7977,
      "step": 12548
    },
    {
      "epoch": 0.9207149866559058,
      "grad_norm": 0.5416405761098213,
      "learning_rate": 1.6377180976310968e-07,
      "loss": 0.4664,
      "step": 12549
    },
    {
      "epoch": 0.9207883562460449,
      "grad_norm": 1.5565842185677132,
      "learning_rate": 1.6347033177206985e-07,
      "loss": 0.7589,
      "step": 12550
    },
    {
      "epoch": 0.9208617258361841,
      "grad_norm": 0.5421257467718229,
      "learning_rate": 1.6316912691183852e-07,
      "loss": 0.4905,
      "step": 12551
    },
    {
      "epoch": 0.9209350954263231,
      "grad_norm": 1.7009928045830607,
      "learning_rate": 1.6286819519942487e-07,
      "loss": 0.7492,
      "step": 12552
    },
    {
      "epoch": 0.9209350954263231,
      "lm_loss": 0.3587,
      "step": 12552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9209350954263231,
      "lm_loss": 0.5791,
      "step": 12552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9209350954263231,
      "lm_loss": 0.5631,
      "step": 12552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9209350954263231,
      "lm_loss": 0.4312,
      "step": 12552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9209350954263231,
      "lm_loss": 0.6053,
      "step": 12552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9209350954263231,
      "lm_loss": 0.3153,
      "step": 12552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9209350954263231,
      "lm_loss": 0.5622,
      "step": 12552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9209350954263231,
      "lm_loss": 0.4771,
      "step": 12552,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9210084650164623,
      "grad_norm": 0.5467141773767096,
      "learning_rate": 1.6256753665182313e-07,
      "loss": 0.4795,
      "step": 12553
    },
    {
      "epoch": 0.9210818346066014,
      "grad_norm": 1.788397115689007,
      "learning_rate": 1.6226715128601245e-07,
      "loss": 0.74,
      "step": 12554
    },
    {
      "epoch": 0.9211552041967406,
      "grad_norm": 1.551841832634199,
      "learning_rate": 1.619670391189554e-07,
      "loss": 0.7268,
      "step": 12555
    },
    {
      "epoch": 0.9212285737868797,
      "grad_norm": 1.485391127083241,
      "learning_rate": 1.616672001676012e-07,
      "loss": 0.7746,
      "step": 12556
    },
    {
      "epoch": 0.9213019433770188,
      "grad_norm": 1.5707578385862975,
      "learning_rate": 1.6136763444888237e-07,
      "loss": 0.7217,
      "step": 12557
    },
    {
      "epoch": 0.9213753129671579,
      "grad_norm": 0.556931385076715,
      "learning_rate": 1.61068341979716e-07,
      "loss": 0.5076,
      "step": 12558
    },
    {
      "epoch": 0.9214486825572971,
      "grad_norm": 1.6304098046752122,
      "learning_rate": 1.6076932277700352e-07,
      "loss": 0.6435,
      "step": 12559
    },
    {
      "epoch": 0.9215220521474362,
      "grad_norm": 0.5396370880667013,
      "learning_rate": 1.6047057685763145e-07,
      "loss": 0.465,
      "step": 12560
    },
    {
      "epoch": 0.9215220521474362,
      "lm_loss": 0.1938,
      "step": 12560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9215220521474362,
      "lm_loss": 0.3972,
      "step": 12560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9215220521474362,
      "lm_loss": 0.5667,
      "step": 12560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9215220521474362,
      "lm_loss": 0.4347,
      "step": 12560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9215220521474362,
      "lm_loss": 0.4585,
      "step": 12560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9215220521474362,
      "lm_loss": 0.5248,
      "step": 12560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9215220521474362,
      "lm_loss": 0.4243,
      "step": 12560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9215220521474362,
      "lm_loss": 0.4148,
      "step": 12560,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9215954217375754,
      "grad_norm": 0.5299625612947453,
      "learning_rate": 1.6017210423847074e-07,
      "loss": 0.4477,
      "step": 12561
    },
    {
      "epoch": 0.9216687913277144,
      "grad_norm": 1.5219056853816546,
      "learning_rate": 1.5987390493637734e-07,
      "loss": 0.6771,
      "step": 12562
    },
    {
      "epoch": 0.9217421609178535,
      "grad_norm": 1.577996962766196,
      "learning_rate": 1.5957597896819055e-07,
      "loss": 0.8076,
      "step": 12563
    },
    {
      "epoch": 0.9218155305079927,
      "grad_norm": 1.5765598249718817,
      "learning_rate": 1.5927832635073527e-07,
      "loss": 0.7648,
      "step": 12564
    },
    {
      "epoch": 0.9218889000981318,
      "grad_norm": 0.5299615683997116,
      "learning_rate": 1.5898094710082136e-07,
      "loss": 0.4641,
      "step": 12565
    },
    {
      "epoch": 0.921962269688271,
      "grad_norm": 1.484768870612282,
      "learning_rate": 1.5868384123524095e-07,
      "loss": 0.7381,
      "step": 12566
    },
    {
      "epoch": 0.92203563927841,
      "grad_norm": 1.622152297394995,
      "learning_rate": 1.5838700877077397e-07,
      "loss": 0.7048,
      "step": 12567
    },
    {
      "epoch": 0.9221090088685492,
      "grad_norm": 0.5330489932399165,
      "learning_rate": 1.5809044972418251e-07,
      "loss": 0.4577,
      "step": 12568
    },
    {
      "epoch": 0.9221090088685492,
      "lm_loss": 0.3901,
      "step": 12568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9221090088685492,
      "lm_loss": 0.5135,
      "step": 12568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9221090088685492,
      "lm_loss": 0.3775,
      "step": 12568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9221090088685492,
      "lm_loss": 0.4556,
      "step": 12568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9221090088685492,
      "lm_loss": 0.4321,
      "step": 12568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9221090088685492,
      "lm_loss": 0.4117,
      "step": 12568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9221090088685492,
      "lm_loss": 0.3411,
      "step": 12568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9221090088685492,
      "lm_loss": 0.5706,
      "step": 12568,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9221823784586883,
      "grad_norm": 0.5475016319939106,
      "learning_rate": 1.5779416411221437e-07,
      "loss": 0.4965,
      "step": 12569
    },
    {
      "epoch": 0.9222557480488275,
      "grad_norm": 1.5913990659221393,
      "learning_rate": 1.5749815195160111e-07,
      "loss": 0.6482,
      "step": 12570
    },
    {
      "epoch": 0.9223291176389666,
      "grad_norm": 1.4644445199396945,
      "learning_rate": 1.5720241325905994e-07,
      "loss": 0.7842,
      "step": 12571
    },
    {
      "epoch": 0.9224024872291057,
      "grad_norm": 0.5544711818126703,
      "learning_rate": 1.5690694805129024e-07,
      "loss": 0.488,
      "step": 12572
    },
    {
      "epoch": 0.9224758568192448,
      "grad_norm": 1.5371213776752954,
      "learning_rate": 1.5661175634497983e-07,
      "loss": 0.7671,
      "step": 12573
    },
    {
      "epoch": 0.922549226409384,
      "grad_norm": 1.5099261399123747,
      "learning_rate": 1.563168381567981e-07,
      "loss": 0.7543,
      "step": 12574
    },
    {
      "epoch": 0.9226225959995231,
      "grad_norm": 1.5351826900389602,
      "learning_rate": 1.5602219350339952e-07,
      "loss": 0.7493,
      "step": 12575
    },
    {
      "epoch": 0.9226959655896623,
      "grad_norm": 1.6237851187112655,
      "learning_rate": 1.557278224014236e-07,
      "loss": 0.6958,
      "step": 12576
    },
    {
      "epoch": 0.9226959655896623,
      "lm_loss": 1.3894,
      "step": 12576,
      "vm_loss": 0.0675
    },
    {
      "epoch": 0.9226959655896623,
      "lm_loss": 0.5467,
      "step": 12576,
      "vm_loss": 0.0853
    },
    {
      "epoch": 0.9226959655896623,
      "lm_loss": 0.8785,
      "step": 12576,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.9226959655896623,
      "lm_loss": 2.1252,
      "step": 12576,
      "vm_loss": 0.1383
    },
    {
      "epoch": 0.9226959655896623,
      "lm_loss": 0.6031,
      "step": 12576,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.9226959655896623,
      "lm_loss": 0.3068,
      "step": 12576,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.9226959655896623,
      "lm_loss": 0.3996,
      "step": 12576,
      "vm_loss": 0.1113
    },
    {
      "epoch": 0.9226959655896623,
      "lm_loss": 1.1109,
      "step": 12576,
      "vm_loss": 0.1607
    },
    {
      "epoch": 0.9227693351798013,
      "grad_norm": 1.6167832046518194,
      "learning_rate": 1.5543372486749363e-07,
      "loss": 0.6967,
      "step": 12577
    },
    {
      "epoch": 0.9228427047699405,
      "grad_norm": 1.5770048417345481,
      "learning_rate": 1.551399009182192e-07,
      "loss": 0.6083,
      "step": 12578
    },
    {
      "epoch": 0.9229160743600796,
      "grad_norm": 1.4519334289561923,
      "learning_rate": 1.548463505701925e-07,
      "loss": 0.7448,
      "step": 12579
    },
    {
      "epoch": 0.9229894439502188,
      "grad_norm": 1.3599955380875273,
      "learning_rate": 1.545530738399914e-07,
      "loss": 0.7371,
      "step": 12580
    },
    {
      "epoch": 0.9230628135403579,
      "grad_norm": 1.6126156085879009,
      "learning_rate": 1.542600707441777e-07,
      "loss": 0.7708,
      "step": 12581
    },
    {
      "epoch": 0.923136183130497,
      "grad_norm": 1.566570866783389,
      "learning_rate": 1.5396734129929757e-07,
      "loss": 0.7022,
      "step": 12582
    },
    {
      "epoch": 0.9232095527206361,
      "grad_norm": 0.5550893195148309,
      "learning_rate": 1.5367488552188337e-07,
      "loss": 0.4684,
      "step": 12583
    },
    {
      "epoch": 0.9232829223107752,
      "grad_norm": 1.423427346237869,
      "learning_rate": 1.5338270342844962e-07,
      "loss": 0.7532,
      "step": 12584
    },
    {
      "epoch": 0.9232829223107752,
      "lm_loss": 1.0165,
      "step": 12584,
      "vm_loss": 0.0835
    },
    {
      "epoch": 0.9232829223107752,
      "lm_loss": 0.2647,
      "step": 12584,
      "vm_loss": 0.1077
    },
    {
      "epoch": 0.9232829223107752,
      "lm_loss": 0.4282,
      "step": 12584,
      "vm_loss": 0.0397
    },
    {
      "epoch": 0.9232829223107752,
      "lm_loss": 0.3101,
      "step": 12584,
      "vm_loss": 0.0948
    },
    {
      "epoch": 0.9232829223107752,
      "lm_loss": 0.5558,
      "step": 12584,
      "vm_loss": 0.1348
    },
    {
      "epoch": 0.9232829223107752,
      "lm_loss": 0.6392,
      "step": 12584,
      "vm_loss": 0.1495
    },
    {
      "epoch": 0.9232829223107752,
      "lm_loss": 0.2647,
      "step": 12584,
      "vm_loss": 0.0471
    },
    {
      "epoch": 0.9232829223107752,
      "lm_loss": 0.3884,
      "step": 12584,
      "vm_loss": 0.1758
    },
    {
      "epoch": 0.9233562919009144,
      "grad_norm": 1.4379131597499437,
      "learning_rate": 1.5309079503549762e-07,
      "loss": 0.7325,
      "step": 12585
    },
    {
      "epoch": 0.9234296614910535,
      "grad_norm": 1.563510840907479,
      "learning_rate": 1.5279916035951137e-07,
      "loss": 0.7125,
      "step": 12586
    },
    {
      "epoch": 0.9235030310811926,
      "grad_norm": 0.5489614327389483,
      "learning_rate": 1.5250779941696048e-07,
      "loss": 0.46,
      "step": 12587
    },
    {
      "epoch": 0.9235764006713317,
      "grad_norm": 1.6319708575576142,
      "learning_rate": 1.522167122242979e-07,
      "loss": 0.7003,
      "step": 12588
    },
    {
      "epoch": 0.9236497702614709,
      "grad_norm": 0.5323778014538473,
      "learning_rate": 1.5192589879796383e-07,
      "loss": 0.4782,
      "step": 12589
    },
    {
      "epoch": 0.92372313985161,
      "grad_norm": 0.5517915187835676,
      "learning_rate": 1.516353591543801e-07,
      "loss": 0.4831,
      "step": 12590
    },
    {
      "epoch": 0.9237965094417492,
      "grad_norm": 1.4778907705371398,
      "learning_rate": 1.5134509330995416e-07,
      "loss": 0.7064,
      "step": 12591
    },
    {
      "epoch": 0.9238698790318882,
      "grad_norm": 1.479344955245642,
      "learning_rate": 1.5105510128107848e-07,
      "loss": 0.6923,
      "step": 12592
    },
    {
      "epoch": 0.9238698790318882,
      "lm_loss": 2.0421,
      "step": 12592,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.9238698790318882,
      "lm_loss": 0.7759,
      "step": 12592,
      "vm_loss": 0.0479
    },
    {
      "epoch": 0.9238698790318882,
      "lm_loss": 0.2386,
      "step": 12592,
      "vm_loss": 0.0882
    },
    {
      "epoch": 0.9238698790318882,
      "lm_loss": 0.4235,
      "step": 12592,
      "vm_loss": 0.0649
    },
    {
      "epoch": 0.9238698790318882,
      "lm_loss": 0.758,
      "step": 12592,
      "vm_loss": 0.1433
    },
    {
      "epoch": 0.9238698790318882,
      "lm_loss": 0.2215,
      "step": 12592,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.9238698790318882,
      "lm_loss": 0.2959,
      "step": 12592,
      "vm_loss": 0.1036
    },
    {
      "epoch": 0.9238698790318882,
      "lm_loss": 0.6183,
      "step": 12592,
      "vm_loss": 0.0833
    },
    {
      "epoch": 0.9239432486220274,
      "grad_norm": 1.5415749506791117,
      "learning_rate": 1.5076538308412935e-07,
      "loss": 0.7406,
      "step": 12593
    },
    {
      "epoch": 0.9240166182121665,
      "grad_norm": 1.5446895326106436,
      "learning_rate": 1.5047593873546816e-07,
      "loss": 0.595,
      "step": 12594
    },
    {
      "epoch": 0.9240899878023057,
      "grad_norm": 0.5612868715279384,
      "learning_rate": 1.501867682514402e-07,
      "loss": 0.4684,
      "step": 12595
    },
    {
      "epoch": 0.9241633573924448,
      "grad_norm": 1.636547365649467,
      "learning_rate": 1.498978716483762e-07,
      "loss": 0.8407,
      "step": 12596
    },
    {
      "epoch": 0.9242367269825839,
      "grad_norm": 1.4900104773353673,
      "learning_rate": 1.4960924894258988e-07,
      "loss": 0.765,
      "step": 12597
    },
    {
      "epoch": 0.924310096572723,
      "grad_norm": 1.5554315452577037,
      "learning_rate": 1.493209001503815e-07,
      "loss": 0.6598,
      "step": 12598
    },
    {
      "epoch": 0.9243834661628622,
      "grad_norm": 0.5520698329062933,
      "learning_rate": 1.4903282528803354e-07,
      "loss": 0.4829,
      "step": 12599
    },
    {
      "epoch": 0.9244568357530013,
      "grad_norm": 1.5626418000164397,
      "learning_rate": 1.4874502437181637e-07,
      "loss": 0.691,
      "step": 12600
    },
    {
      "epoch": 0.9244568357530013,
      "lm_loss": 0.4326,
      "step": 12600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9244568357530013,
      "lm_loss": 0.5192,
      "step": 12600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9244568357530013,
      "lm_loss": 0.415,
      "step": 12600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9244568357530013,
      "lm_loss": 0.5716,
      "step": 12600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9244568357530013,
      "lm_loss": 0.475,
      "step": 12600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9244568357530013,
      "lm_loss": 0.5138,
      "step": 12600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9244568357530013,
      "lm_loss": 0.5082,
      "step": 12600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9244568357530013,
      "lm_loss": 0.5122,
      "step": 12600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9245302053431405,
      "grad_norm": 0.5798807874744092,
      "learning_rate": 1.4845749741798087e-07,
      "loss": 0.4986,
      "step": 12601
    },
    {
      "epoch": 0.9246035749332795,
      "grad_norm": 1.4975432900938221,
      "learning_rate": 1.4817024444276519e-07,
      "loss": 0.6443,
      "step": 12602
    },
    {
      "epoch": 0.9246769445234186,
      "grad_norm": 1.5802561635175527,
      "learning_rate": 1.4788326546239185e-07,
      "loss": 0.8051,
      "step": 12603
    },
    {
      "epoch": 0.9247503141135578,
      "grad_norm": 1.3935159940783906,
      "learning_rate": 1.4759656049306514e-07,
      "loss": 0.6814,
      "step": 12604
    },
    {
      "epoch": 0.9248236837036969,
      "grad_norm": 1.5302633105447057,
      "learning_rate": 1.473101295509788e-07,
      "loss": 0.7878,
      "step": 12605
    },
    {
      "epoch": 0.9248970532938361,
      "grad_norm": 0.5481254073649539,
      "learning_rate": 1.470239726523065e-07,
      "loss": 0.4787,
      "step": 12606
    },
    {
      "epoch": 0.9249704228839751,
      "grad_norm": 0.5551624192016053,
      "learning_rate": 1.467380898132087e-07,
      "loss": 0.4783,
      "step": 12607
    },
    {
      "epoch": 0.9250437924741143,
      "grad_norm": 1.3805406460054832,
      "learning_rate": 1.4645248104982967e-07,
      "loss": 0.711,
      "step": 12608
    },
    {
      "epoch": 0.9250437924741143,
      "lm_loss": 0.4799,
      "step": 12608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9250437924741143,
      "lm_loss": 0.3686,
      "step": 12608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9250437924741143,
      "lm_loss": 0.4309,
      "step": 12608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9250437924741143,
      "lm_loss": 0.422,
      "step": 12608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9250437924741143,
      "lm_loss": 0.6349,
      "step": 12608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9250437924741143,
      "lm_loss": 0.5009,
      "step": 12608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9250437924741143,
      "lm_loss": 0.6117,
      "step": 12608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9250437924741143,
      "lm_loss": 0.4076,
      "step": 12608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9251171620642534,
      "grad_norm": 0.5416874130261158,
      "learning_rate": 1.4616714637829822e-07,
      "loss": 0.4785,
      "step": 12609
    },
    {
      "epoch": 0.9251905316543926,
      "grad_norm": 0.5383179056789099,
      "learning_rate": 1.4588208581472917e-07,
      "loss": 0.4758,
      "step": 12610
    },
    {
      "epoch": 0.9252639012445317,
      "grad_norm": 1.6116260407651943,
      "learning_rate": 1.4559729937521916e-07,
      "loss": 0.672,
      "step": 12611
    },
    {
      "epoch": 0.9253372708346708,
      "grad_norm": 1.7305226826459335,
      "learning_rate": 1.4531278707585195e-07,
      "loss": 0.7802,
      "step": 12612
    },
    {
      "epoch": 0.9254106404248099,
      "grad_norm": 1.3631685929458368,
      "learning_rate": 1.4502854893269357e-07,
      "loss": 0.7588,
      "step": 12613
    },
    {
      "epoch": 0.9254840100149491,
      "grad_norm": 1.6877591291084204,
      "learning_rate": 1.4474458496179677e-07,
      "loss": 0.7013,
      "step": 12614
    },
    {
      "epoch": 0.9255573796050882,
      "grad_norm": 0.5428943735037853,
      "learning_rate": 1.4446089517919594e-07,
      "loss": 0.4522,
      "step": 12615
    },
    {
      "epoch": 0.9256307491952273,
      "grad_norm": 1.4986096996583382,
      "learning_rate": 1.4417747960091379e-07,
      "loss": 0.6802,
      "step": 12616
    },
    {
      "epoch": 0.9256307491952273,
      "lm_loss": 0.2511,
      "step": 12616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9256307491952273,
      "lm_loss": 0.3177,
      "step": 12616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9256307491952273,
      "lm_loss": 0.5321,
      "step": 12616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9256307491952273,
      "lm_loss": 0.49,
      "step": 12616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9256307491952273,
      "lm_loss": 0.5895,
      "step": 12616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9256307491952273,
      "lm_loss": 0.4768,
      "step": 12616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9256307491952273,
      "lm_loss": 0.5053,
      "step": 12616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9256307491952273,
      "lm_loss": 0.5098,
      "step": 12616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9257041187853664,
      "grad_norm": 0.5320442414922721,
      "learning_rate": 1.4389433824295474e-07,
      "loss": 0.4753,
      "step": 12617
    },
    {
      "epoch": 0.9257774883755056,
      "grad_norm": 0.5408198727797366,
      "learning_rate": 1.4361147112130824e-07,
      "loss": 0.4668,
      "step": 12618
    },
    {
      "epoch": 0.9258508579656447,
      "grad_norm": 1.4780665936686994,
      "learning_rate": 1.4332887825194818e-07,
      "loss": 0.7272,
      "step": 12619
    },
    {
      "epoch": 0.9259242275557839,
      "grad_norm": 1.4922545042005047,
      "learning_rate": 1.4304655965083347e-07,
      "loss": 0.7324,
      "step": 12620
    },
    {
      "epoch": 0.9259975971459229,
      "grad_norm": 1.6488991089994227,
      "learning_rate": 1.4276451533390855e-07,
      "loss": 0.6855,
      "step": 12621
    },
    {
      "epoch": 0.926070966736062,
      "grad_norm": 0.5449536753437652,
      "learning_rate": 1.424827453170996e-07,
      "loss": 0.4554,
      "step": 12622
    },
    {
      "epoch": 0.9261443363262012,
      "grad_norm": 1.5154546017830424,
      "learning_rate": 1.4220124961631999e-07,
      "loss": 0.6597,
      "step": 12623
    },
    {
      "epoch": 0.9262177059163403,
      "grad_norm": 1.5206511092667956,
      "learning_rate": 1.4192002824746532e-07,
      "loss": 0.8289,
      "step": 12624
    },
    {
      "epoch": 0.9262177059163403,
      "lm_loss": 1.064,
      "step": 12624,
      "vm_loss": 0.0759
    },
    {
      "epoch": 0.9262177059163403,
      "lm_loss": 0.7007,
      "step": 12624,
      "vm_loss": 0.0883
    },
    {
      "epoch": 0.9262177059163403,
      "lm_loss": 0.4718,
      "step": 12624,
      "vm_loss": 0.0962
    },
    {
      "epoch": 0.9262177059163403,
      "lm_loss": 0.4789,
      "step": 12624,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.9262177059163403,
      "lm_loss": 0.7659,
      "step": 12624,
      "vm_loss": 0.216
    },
    {
      "epoch": 0.9262177059163403,
      "lm_loss": 0.1029,
      "step": 12624,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.9262177059163403,
      "lm_loss": 0.7609,
      "step": 12624,
      "vm_loss": 0.1721
    },
    {
      "epoch": 0.9262177059163403,
      "lm_loss": 0.0524,
      "step": 12624,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.9262910755064795,
      "grad_norm": 1.6677509542935107,
      "learning_rate": 1.416390812264179e-07,
      "loss": 0.6828,
      "step": 12625
    },
    {
      "epoch": 0.9263644450966185,
      "grad_norm": 0.5338743621758791,
      "learning_rate": 1.4135840856904225e-07,
      "loss": 0.4759,
      "step": 12626
    },
    {
      "epoch": 0.9264378146867577,
      "grad_norm": 1.576349288274782,
      "learning_rate": 1.4107801029119018e-07,
      "loss": 0.7563,
      "step": 12627
    },
    {
      "epoch": 0.9265111842768968,
      "grad_norm": 0.5422714289122388,
      "learning_rate": 1.4079788640869617e-07,
      "loss": 0.4668,
      "step": 12628
    },
    {
      "epoch": 0.926584553867036,
      "grad_norm": 0.5554794127592387,
      "learning_rate": 1.4051803693737876e-07,
      "loss": 0.4662,
      "step": 12629
    },
    {
      "epoch": 0.9266579234571751,
      "grad_norm": 0.5545702119996402,
      "learning_rate": 1.4023846189304192e-07,
      "loss": 0.4908,
      "step": 12630
    },
    {
      "epoch": 0.9267312930473142,
      "grad_norm": 1.5422848181573405,
      "learning_rate": 1.3995916129147414e-07,
      "loss": 0.7283,
      "step": 12631
    },
    {
      "epoch": 0.9268046626374533,
      "grad_norm": 1.5873888668181795,
      "learning_rate": 1.3968013514844835e-07,
      "loss": 0.7439,
      "step": 12632
    },
    {
      "epoch": 0.9268046626374533,
      "lm_loss": 1.5612,
      "step": 12632,
      "vm_loss": 0.0681
    },
    {
      "epoch": 0.9268046626374533,
      "lm_loss": 0.8883,
      "step": 12632,
      "vm_loss": 0.0445
    },
    {
      "epoch": 0.9268046626374533,
      "lm_loss": 0.2461,
      "step": 12632,
      "vm_loss": 0.173
    },
    {
      "epoch": 0.9268046626374533,
      "lm_loss": 0.0901,
      "step": 12632,
      "vm_loss": 0.0721
    },
    {
      "epoch": 0.9268046626374533,
      "lm_loss": 0.92,
      "step": 12632,
      "vm_loss": 0.1764
    },
    {
      "epoch": 0.9268046626374533,
      "lm_loss": 1.1908,
      "step": 12632,
      "vm_loss": 0.138
    },
    {
      "epoch": 0.9268046626374533,
      "lm_loss": 0.604,
      "step": 12632,
      "vm_loss": 0.1131
    },
    {
      "epoch": 0.9268046626374533,
      "lm_loss": 0.3991,
      "step": 12632,
      "vm_loss": 0.1733
    },
    {
      "epoch": 0.9268780322275925,
      "grad_norm": 1.4313228230781603,
      "learning_rate": 1.3940138347972199e-07,
      "loss": 0.7467,
      "step": 12633
    },
    {
      "epoch": 0.9269514018177316,
      "grad_norm": 1.5501181152225219,
      "learning_rate": 1.391229063010363e-07,
      "loss": 0.7632,
      "step": 12634
    },
    {
      "epoch": 0.9270247714078708,
      "grad_norm": 0.5547396606499442,
      "learning_rate": 1.3884470362811763e-07,
      "loss": 0.4624,
      "step": 12635
    },
    {
      "epoch": 0.9270981409980098,
      "grad_norm": 0.5523088574247561,
      "learning_rate": 1.3856677547667674e-07,
      "loss": 0.4622,
      "step": 12636
    },
    {
      "epoch": 0.927171510588149,
      "grad_norm": 1.4899021171574163,
      "learning_rate": 1.3828912186240995e-07,
      "loss": 0.7918,
      "step": 12637
    },
    {
      "epoch": 0.9272448801782881,
      "grad_norm": 0.5638563992625476,
      "learning_rate": 1.3801174280099583e-07,
      "loss": 0.481,
      "step": 12638
    },
    {
      "epoch": 0.9273182497684272,
      "grad_norm": 1.525351451031842,
      "learning_rate": 1.3773463830809962e-07,
      "loss": 0.7399,
      "step": 12639
    },
    {
      "epoch": 0.9273916193585664,
      "grad_norm": 1.4480503056359735,
      "learning_rate": 1.3745780839936884e-07,
      "loss": 0.806,
      "step": 12640
    },
    {
      "epoch": 0.9273916193585664,
      "lm_loss": 1.3043,
      "step": 12640,
      "vm_loss": 0.0636
    },
    {
      "epoch": 0.9273916193585664,
      "lm_loss": 0.8343,
      "step": 12640,
      "vm_loss": 0.1539
    },
    {
      "epoch": 0.9273916193585664,
      "lm_loss": 0.0659,
      "step": 12640,
      "vm_loss": 0.1195
    },
    {
      "epoch": 0.9273916193585664,
      "lm_loss": 0.2302,
      "step": 12640,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.9273916193585664,
      "lm_loss": 0.9522,
      "step": 12640,
      "vm_loss": 0.1958
    },
    {
      "epoch": 0.9273916193585664,
      "lm_loss": 0.0436,
      "step": 12640,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.9273916193585664,
      "lm_loss": 0.8353,
      "step": 12640,
      "vm_loss": 0.1972
    },
    {
      "epoch": 0.9273916193585664,
      "lm_loss": 0.8483,
      "step": 12640,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.9274649889487054,
      "grad_norm": 1.5692458961126132,
      "learning_rate": 1.3718125309043816e-07,
      "loss": 0.7653,
      "step": 12641
    },
    {
      "epoch": 0.9275383585388446,
      "grad_norm": 1.9461918320429095,
      "learning_rate": 1.3690497239692403e-07,
      "loss": 0.7336,
      "step": 12642
    },
    {
      "epoch": 0.9276117281289837,
      "grad_norm": 1.5514526501583155,
      "learning_rate": 1.3662896633442945e-07,
      "loss": 0.7278,
      "step": 12643
    },
    {
      "epoch": 0.9276850977191229,
      "grad_norm": 1.391080818374206,
      "learning_rate": 1.3635323491854146e-07,
      "loss": 0.6942,
      "step": 12644
    },
    {
      "epoch": 0.927758467309262,
      "grad_norm": 0.5685762201194579,
      "learning_rate": 1.3607777816483092e-07,
      "loss": 0.4915,
      "step": 12645
    },
    {
      "epoch": 0.9278318368994011,
      "grad_norm": 1.5205647869319117,
      "learning_rate": 1.358025960888537e-07,
      "loss": 0.7469,
      "step": 12646
    },
    {
      "epoch": 0.9279052064895402,
      "grad_norm": 0.5428823706891519,
      "learning_rate": 1.3552768870614908e-07,
      "loss": 0.4692,
      "step": 12647
    },
    {
      "epoch": 0.9279785760796794,
      "grad_norm": 1.3774469741285713,
      "learning_rate": 1.3525305603224294e-07,
      "loss": 0.838,
      "step": 12648
    },
    {
      "epoch": 0.9279785760796794,
      "lm_loss": 1.0145,
      "step": 12648,
      "vm_loss": 0.0716
    },
    {
      "epoch": 0.9279785760796794,
      "lm_loss": 0.0112,
      "step": 12648,
      "vm_loss": 0.0889
    },
    {
      "epoch": 0.9279785760796794,
      "lm_loss": 0.0845,
      "step": 12648,
      "vm_loss": 0.0964
    },
    {
      "epoch": 0.9279785760796794,
      "lm_loss": 0.8594,
      "step": 12648,
      "vm_loss": 0.0528
    },
    {
      "epoch": 0.9279785760796794,
      "lm_loss": 0.3273,
      "step": 12648,
      "vm_loss": 0.1133
    },
    {
      "epoch": 0.9279785760796794,
      "lm_loss": 0.0988,
      "step": 12648,
      "vm_loss": 0.0907
    },
    {
      "epoch": 0.9279785760796794,
      "lm_loss": 0.7256,
      "step": 12648,
      "vm_loss": 0.0801
    },
    {
      "epoch": 0.9279785760796794,
      "lm_loss": 0.6803,
      "step": 12648,
      "vm_loss": 0.1836
    },
    {
      "epoch": 0.9280519456698185,
      "grad_norm": 1.7161196078064003,
      "learning_rate": 1.3497869808264453e-07,
      "loss": 0.7786,
      "step": 12649
    },
    {
      "epoch": 0.9281253152599577,
      "grad_norm": 1.5905705470119504,
      "learning_rate": 1.3470461487284648e-07,
      "loss": 0.7606,
      "step": 12650
    },
    {
      "epoch": 0.9281986848500967,
      "grad_norm": 1.5357349957045554,
      "learning_rate": 1.3443080641832806e-07,
      "loss": 0.7429,
      "step": 12651
    },
    {
      "epoch": 0.9282720544402359,
      "grad_norm": 1.4743869618207521,
      "learning_rate": 1.3415727273455082e-07,
      "loss": 0.6826,
      "step": 12652
    },
    {
      "epoch": 0.928345424030375,
      "grad_norm": 1.5818811190582167,
      "learning_rate": 1.3388401383696292e-07,
      "loss": 0.7593,
      "step": 12653
    },
    {
      "epoch": 0.9284187936205142,
      "grad_norm": 1.557360796135695,
      "learning_rate": 1.3361102974099537e-07,
      "loss": 0.6607,
      "step": 12654
    },
    {
      "epoch": 0.9284921632106533,
      "grad_norm": 1.4547356997476952,
      "learning_rate": 1.3333832046206475e-07,
      "loss": 0.7202,
      "step": 12655
    },
    {
      "epoch": 0.9285655328007923,
      "grad_norm": 1.578292341715971,
      "learning_rate": 1.3306588601557092e-07,
      "loss": 0.8322,
      "step": 12656
    },
    {
      "epoch": 0.9285655328007923,
      "lm_loss": 0.4124,
      "step": 12656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9285655328007923,
      "lm_loss": 0.5258,
      "step": 12656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9285655328007923,
      "lm_loss": 0.4358,
      "step": 12656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9285655328007923,
      "lm_loss": 0.4382,
      "step": 12656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9285655328007923,
      "lm_loss": 0.3761,
      "step": 12656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9285655328007923,
      "lm_loss": 0.4403,
      "step": 12656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9285655328007923,
      "lm_loss": 0.3945,
      "step": 12656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9285655328007923,
      "lm_loss": 0.4321,
      "step": 12656,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9286389023909315,
      "grad_norm": 0.5422041230079135,
      "learning_rate": 1.3279372641689824e-07,
      "loss": 0.4723,
      "step": 12657
    },
    {
      "epoch": 0.9287122719810706,
      "grad_norm": 0.5290271386021367,
      "learning_rate": 1.3252184168141835e-07,
      "loss": 0.4576,
      "step": 12658
    },
    {
      "epoch": 0.9287856415712098,
      "grad_norm": 4.019118449495366,
      "learning_rate": 1.3225023182448392e-07,
      "loss": 0.7071,
      "step": 12659
    },
    {
      "epoch": 0.928859011161349,
      "grad_norm": 1.5555754188226805,
      "learning_rate": 1.3197889686143384e-07,
      "loss": 0.7492,
      "step": 12660
    },
    {
      "epoch": 0.928932380751488,
      "grad_norm": 1.5564861359598807,
      "learning_rate": 1.317078368075908e-07,
      "loss": 0.7406,
      "step": 12661
    },
    {
      "epoch": 0.9290057503416271,
      "grad_norm": 1.3146242979058005,
      "learning_rate": 1.3143705167826204e-07,
      "loss": 0.5938,
      "step": 12662
    },
    {
      "epoch": 0.9290791199317663,
      "grad_norm": 1.729650000870509,
      "learning_rate": 1.3116654148874032e-07,
      "loss": 0.6624,
      "step": 12663
    },
    {
      "epoch": 0.9291524895219054,
      "grad_norm": 1.5766037879880392,
      "learning_rate": 1.3089630625430116e-07,
      "loss": 0.7995,
      "step": 12664
    },
    {
      "epoch": 0.9291524895219054,
      "lm_loss": 0.5182,
      "step": 12664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9291524895219054,
      "lm_loss": 0.6314,
      "step": 12664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9291524895219054,
      "lm_loss": 0.4499,
      "step": 12664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9291524895219054,
      "lm_loss": 0.7084,
      "step": 12664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9291524895219054,
      "lm_loss": 0.3717,
      "step": 12664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9291524895219054,
      "lm_loss": 0.3905,
      "step": 12664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9291524895219054,
      "lm_loss": 0.6126,
      "step": 12664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9291524895219054,
      "lm_loss": 0.568,
      "step": 12664,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9292258591120446,
      "grad_norm": 0.550949984667654,
      "learning_rate": 1.3062634599020574e-07,
      "loss": 0.4948,
      "step": 12665
    },
    {
      "epoch": 0.9292992287021836,
      "grad_norm": 1.604414786465077,
      "learning_rate": 1.303566607116996e-07,
      "loss": 0.6544,
      "step": 12666
    },
    {
      "epoch": 0.9293725982923228,
      "grad_norm": 1.7233187303226727,
      "learning_rate": 1.3008725043401227e-07,
      "loss": 0.6857,
      "step": 12667
    },
    {
      "epoch": 0.9294459678824619,
      "grad_norm": 1.6379182939370565,
      "learning_rate": 1.298181151723571e-07,
      "loss": 0.7507,
      "step": 12668
    },
    {
      "epoch": 0.9295193374726011,
      "grad_norm": 1.4590595013088343,
      "learning_rate": 1.2954925494193472e-07,
      "loss": 0.7101,
      "step": 12669
    },
    {
      "epoch": 0.9295927070627402,
      "grad_norm": 1.470086085429697,
      "learning_rate": 1.292806697579274e-07,
      "loss": 0.6404,
      "step": 12670
    },
    {
      "epoch": 0.9296660766528793,
      "grad_norm": 1.5647908866989744,
      "learning_rate": 1.290123596355025e-07,
      "loss": 0.7375,
      "step": 12671
    },
    {
      "epoch": 0.9297394462430184,
      "grad_norm": 1.6786889340893418,
      "learning_rate": 1.287443245898129e-07,
      "loss": 0.6937,
      "step": 12672
    },
    {
      "epoch": 0.9297394462430184,
      "lm_loss": 1.0255,
      "step": 12672,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.9297394462430184,
      "lm_loss": 1.3561,
      "step": 12672,
      "vm_loss": 0.0756
    },
    {
      "epoch": 0.9297394462430184,
      "lm_loss": 1.0389,
      "step": 12672,
      "vm_loss": 0.1933
    },
    {
      "epoch": 0.9297394462430184,
      "lm_loss": 0.8403,
      "step": 12672,
      "vm_loss": 0.0712
    },
    {
      "epoch": 0.9297394462430184,
      "lm_loss": 0.1139,
      "step": 12672,
      "vm_loss": 0.1207
    },
    {
      "epoch": 0.9297394462430184,
      "lm_loss": 0.2448,
      "step": 12672,
      "vm_loss": 0.0888
    },
    {
      "epoch": 0.9297394462430184,
      "lm_loss": 0.928,
      "step": 12672,
      "vm_loss": 0.1503
    },
    {
      "epoch": 0.9297394462430184,
      "lm_loss": 0.8535,
      "step": 12672,
      "vm_loss": 0.1654
    },
    {
      "epoch": 0.9298128158331576,
      "grad_norm": 1.4789705783767537,
      "learning_rate": 1.2847656463599368e-07,
      "loss": 0.8119,
      "step": 12673
    },
    {
      "epoch": 0.9298861854232967,
      "grad_norm": 1.5399842592159965,
      "learning_rate": 1.2820907978916773e-07,
      "loss": 0.8051,
      "step": 12674
    },
    {
      "epoch": 0.9299595550134359,
      "grad_norm": 1.5442419455937801,
      "learning_rate": 1.279418700644397e-07,
      "loss": 0.7169,
      "step": 12675
    },
    {
      "epoch": 0.9300329246035749,
      "grad_norm": 0.5394277421697009,
      "learning_rate": 1.2767493547690023e-07,
      "loss": 0.4703,
      "step": 12676
    },
    {
      "epoch": 0.930106294193714,
      "grad_norm": 1.541186406587731,
      "learning_rate": 1.2740827604162288e-07,
      "loss": 0.7344,
      "step": 12677
    },
    {
      "epoch": 0.9301796637838532,
      "grad_norm": 1.7245194286749028,
      "learning_rate": 1.271418917736661e-07,
      "loss": 0.7869,
      "step": 12678
    },
    {
      "epoch": 0.9302530333739923,
      "grad_norm": 1.761870849712848,
      "learning_rate": 1.268757826880751e-07,
      "loss": 0.7113,
      "step": 12679
    },
    {
      "epoch": 0.9303264029641315,
      "grad_norm": 1.5878765360271827,
      "learning_rate": 1.2660994879987676e-07,
      "loss": 0.7263,
      "step": 12680
    },
    {
      "epoch": 0.9303264029641315,
      "lm_loss": 0.1253,
      "step": 12680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9303264029641315,
      "lm_loss": 0.4636,
      "step": 12680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9303264029641315,
      "lm_loss": 0.4472,
      "step": 12680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9303264029641315,
      "lm_loss": 0.2901,
      "step": 12680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9303264029641315,
      "lm_loss": 0.4661,
      "step": 12680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9303264029641315,
      "lm_loss": 0.4191,
      "step": 12680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9303264029641315,
      "lm_loss": 0.4288,
      "step": 12680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9303264029641315,
      "lm_loss": 0.5433,
      "step": 12680,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9303997725542705,
      "grad_norm": 0.5334299626018779,
      "learning_rate": 1.2634439012408294e-07,
      "loss": 0.4754,
      "step": 12681
    },
    {
      "epoch": 0.9304731421444097,
      "grad_norm": 1.6496337210944525,
      "learning_rate": 1.2607910667569111e-07,
      "loss": 0.7466,
      "step": 12682
    },
    {
      "epoch": 0.9305465117345488,
      "grad_norm": 1.3364876375335015,
      "learning_rate": 1.2581409846968263e-07,
      "loss": 0.715,
      "step": 12683
    },
    {
      "epoch": 0.930619881324688,
      "grad_norm": 1.5032212094837802,
      "learning_rate": 1.255493655210216e-07,
      "loss": 0.7712,
      "step": 12684
    },
    {
      "epoch": 0.9306932509148271,
      "grad_norm": 1.4170901989902365,
      "learning_rate": 1.2528490784465996e-07,
      "loss": 0.6687,
      "step": 12685
    },
    {
      "epoch": 0.9307666205049662,
      "grad_norm": 0.5602943085578787,
      "learning_rate": 1.2502072545553135e-07,
      "loss": 0.4514,
      "step": 12686
    },
    {
      "epoch": 0.9308399900951053,
      "grad_norm": 1.4870062119937246,
      "learning_rate": 1.2475681836855546e-07,
      "loss": 0.8034,
      "step": 12687
    },
    {
      "epoch": 0.9309133596852445,
      "grad_norm": 1.6657087103476265,
      "learning_rate": 1.2449318659863542e-07,
      "loss": 0.7297,
      "step": 12688
    },
    {
      "epoch": 0.9309133596852445,
      "lm_loss": 1.4059,
      "step": 12688,
      "vm_loss": 0.0398
    },
    {
      "epoch": 0.9309133596852445,
      "lm_loss": 0.4214,
      "step": 12688,
      "vm_loss": 0.0451
    },
    {
      "epoch": 0.9309133596852445,
      "lm_loss": 0.2026,
      "step": 12688,
      "vm_loss": 0.1514
    },
    {
      "epoch": 0.9309133596852445,
      "lm_loss": 0.7154,
      "step": 12688,
      "vm_loss": 0.1782
    },
    {
      "epoch": 0.9309133596852445,
      "lm_loss": 0.7768,
      "step": 12688,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.9309133596852445,
      "lm_loss": 0.6648,
      "step": 12688,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.9309133596852445,
      "lm_loss": 0.4026,
      "step": 12688,
      "vm_loss": 0.0708
    },
    {
      "epoch": 0.9309133596852445,
      "lm_loss": 0.0873,
      "step": 12688,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.9309867292753836,
      "grad_norm": 1.4216428828035554,
      "learning_rate": 1.2422983016065816e-07,
      "loss": 0.6565,
      "step": 12689
    },
    {
      "epoch": 0.9310600988655228,
      "grad_norm": 1.5707869257723606,
      "learning_rate": 1.239667490694979e-07,
      "loss": 0.6356,
      "step": 12690
    },
    {
      "epoch": 0.9311334684556618,
      "grad_norm": 0.5424155167846069,
      "learning_rate": 1.2370394334001058e-07,
      "loss": 0.4581,
      "step": 12691
    },
    {
      "epoch": 0.931206838045801,
      "grad_norm": 0.5309922970297012,
      "learning_rate": 1.234414129870376e-07,
      "loss": 0.4919,
      "step": 12692
    },
    {
      "epoch": 0.9312802076359401,
      "grad_norm": 1.8158655580320748,
      "learning_rate": 1.2317915802540436e-07,
      "loss": 0.8364,
      "step": 12693
    },
    {
      "epoch": 0.9313535772260793,
      "grad_norm": 1.393280895712565,
      "learning_rate": 1.2291717846992125e-07,
      "loss": 0.6363,
      "step": 12694
    },
    {
      "epoch": 0.9314269468162183,
      "grad_norm": 1.5026704762817464,
      "learning_rate": 1.226554743353825e-07,
      "loss": 0.7215,
      "step": 12695
    },
    {
      "epoch": 0.9315003164063574,
      "grad_norm": 1.5722515446357077,
      "learning_rate": 1.2239404563656854e-07,
      "loss": 0.7345,
      "step": 12696
    },
    {
      "epoch": 0.9315003164063574,
      "lm_loss": 0.4197,
      "step": 12696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9315003164063574,
      "lm_loss": 0.3745,
      "step": 12696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9315003164063574,
      "lm_loss": 0.6521,
      "step": 12696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9315003164063574,
      "lm_loss": 0.6251,
      "step": 12696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9315003164063574,
      "lm_loss": 0.5591,
      "step": 12696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9315003164063574,
      "lm_loss": 0.5392,
      "step": 12696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9315003164063574,
      "lm_loss": 0.5084,
      "step": 12696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9315003164063574,
      "lm_loss": 0.4542,
      "step": 12696,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9315736859964966,
      "grad_norm": 0.5654171492590933,
      "learning_rate": 1.2213289238824145e-07,
      "loss": 0.4953,
      "step": 12697
    },
    {
      "epoch": 0.9316470555866357,
      "grad_norm": 1.603972628045399,
      "learning_rate": 1.2187201460515053e-07,
      "loss": 0.6908,
      "step": 12698
    },
    {
      "epoch": 0.9317204251767749,
      "grad_norm": 1.4094657926126963,
      "learning_rate": 1.2161141230202678e-07,
      "loss": 0.6951,
      "step": 12699
    },
    {
      "epoch": 0.9317937947669139,
      "grad_norm": 1.621175609150493,
      "learning_rate": 1.2135108549358733e-07,
      "loss": 0.7787,
      "step": 12700
    },
    {
      "epoch": 0.9318671643570531,
      "grad_norm": 0.5371511349739038,
      "learning_rate": 1.2109103419453427e-07,
      "loss": 0.4769,
      "step": 12701
    },
    {
      "epoch": 0.9319405339471922,
      "grad_norm": 1.4509920873096451,
      "learning_rate": 1.2083125841955312e-07,
      "loss": 0.6528,
      "step": 12702
    },
    {
      "epoch": 0.9320139035373314,
      "grad_norm": 1.5598835523006216,
      "learning_rate": 1.2057175818331378e-07,
      "loss": 0.7299,
      "step": 12703
    },
    {
      "epoch": 0.9320872731274705,
      "grad_norm": 1.4248964383811862,
      "learning_rate": 1.2031253350047122e-07,
      "loss": 0.6502,
      "step": 12704
    },
    {
      "epoch": 0.9320872731274705,
      "lm_loss": 0.7223,
      "step": 12704,
      "vm_loss": 0.1053
    },
    {
      "epoch": 0.9320872731274705,
      "lm_loss": 0.8773,
      "step": 12704,
      "vm_loss": 0.1114
    },
    {
      "epoch": 0.9320872731274705,
      "lm_loss": 0.7733,
      "step": 12704,
      "vm_loss": 0.1414
    },
    {
      "epoch": 0.9320872731274705,
      "lm_loss": 0.0488,
      "step": 12704,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.9320872731274705,
      "lm_loss": 0.3465,
      "step": 12704,
      "vm_loss": 0.0847
    },
    {
      "epoch": 0.9320872731274705,
      "lm_loss": 0.1572,
      "step": 12704,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.9320872731274705,
      "lm_loss": 0.5762,
      "step": 12704,
      "vm_loss": 0.1004
    },
    {
      "epoch": 0.9320872731274705,
      "lm_loss": 0.329,
      "step": 12704,
      "vm_loss": 0.1031
    },
    {
      "epoch": 0.9321606427176096,
      "grad_norm": 1.4153229959211677,
      "learning_rate": 1.2005358438566318e-07,
      "loss": 0.7183,
      "step": 12705
    },
    {
      "epoch": 0.9322340123077487,
      "grad_norm": 1.4504314265051987,
      "learning_rate": 1.1979491085351514e-07,
      "loss": 0.635,
      "step": 12706
    },
    {
      "epoch": 0.9323073818978879,
      "grad_norm": 0.5310618784511832,
      "learning_rate": 1.1953651291863378e-07,
      "loss": 0.4726,
      "step": 12707
    },
    {
      "epoch": 0.932380751488027,
      "grad_norm": 1.6178372722348453,
      "learning_rate": 1.1927839059561242e-07,
      "loss": 0.7542,
      "step": 12708
    },
    {
      "epoch": 0.9324541210781662,
      "grad_norm": 1.5761862831369715,
      "learning_rate": 1.1902054389902662e-07,
      "loss": 0.768,
      "step": 12709
    },
    {
      "epoch": 0.9325274906683052,
      "grad_norm": 1.4414664581805423,
      "learning_rate": 1.1876297284343862e-07,
      "loss": 0.708,
      "step": 12710
    },
    {
      "epoch": 0.9326008602584444,
      "grad_norm": 1.6061560345221737,
      "learning_rate": 1.1850567744339292e-07,
      "loss": 0.7443,
      "step": 12711
    },
    {
      "epoch": 0.9326742298485835,
      "grad_norm": 1.433855628942457,
      "learning_rate": 1.182486577134212e-07,
      "loss": 0.7515,
      "step": 12712
    },
    {
      "epoch": 0.9326742298485835,
      "lm_loss": 0.4481,
      "step": 12712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9326742298485835,
      "lm_loss": 0.3956,
      "step": 12712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9326742298485835,
      "lm_loss": 0.3185,
      "step": 12712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9326742298485835,
      "lm_loss": 0.3441,
      "step": 12712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9326742298485835,
      "lm_loss": 0.591,
      "step": 12712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9326742298485835,
      "lm_loss": 0.5332,
      "step": 12712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9326742298485835,
      "lm_loss": 0.5335,
      "step": 12712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9326742298485835,
      "lm_loss": 0.5996,
      "step": 12712,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9327475994387227,
      "grad_norm": 0.55317951375266,
      "learning_rate": 1.1799191366803686e-07,
      "loss": 0.4851,
      "step": 12713
    },
    {
      "epoch": 0.9328209690288618,
      "grad_norm": 1.5199475719043871,
      "learning_rate": 1.1773544532173942e-07,
      "loss": 0.7902,
      "step": 12714
    },
    {
      "epoch": 0.9328943386190008,
      "grad_norm": 0.5411829990603816,
      "learning_rate": 1.1747925268901228e-07,
      "loss": 0.4861,
      "step": 12715
    },
    {
      "epoch": 0.93296770820914,
      "grad_norm": 0.5413756692430315,
      "learning_rate": 1.1722333578432277e-07,
      "loss": 0.4673,
      "step": 12716
    },
    {
      "epoch": 0.9330410777992791,
      "grad_norm": 1.5318617735654725,
      "learning_rate": 1.1696769462212376e-07,
      "loss": 0.8143,
      "step": 12717
    },
    {
      "epoch": 0.9331144473894183,
      "grad_norm": 1.5723806378985503,
      "learning_rate": 1.167123292168515e-07,
      "loss": 0.6828,
      "step": 12718
    },
    {
      "epoch": 0.9331878169795574,
      "grad_norm": 1.400632416813204,
      "learning_rate": 1.164572395829272e-07,
      "loss": 0.7606,
      "step": 12719
    },
    {
      "epoch": 0.9332611865696965,
      "grad_norm": 1.6803763308689745,
      "learning_rate": 1.162024257347566e-07,
      "loss": 0.6882,
      "step": 12720
    },
    {
      "epoch": 0.9332611865696965,
      "lm_loss": 0.2969,
      "step": 12720,
      "vm_loss": 0.0538
    },
    {
      "epoch": 0.9332611865696965,
      "lm_loss": 1.5479,
      "step": 12720,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.9332611865696965,
      "lm_loss": 0.6725,
      "step": 12720,
      "vm_loss": 0.1241
    },
    {
      "epoch": 0.9332611865696965,
      "lm_loss": 1.3297,
      "step": 12720,
      "vm_loss": 0.1687
    },
    {
      "epoch": 0.9332611865696965,
      "lm_loss": 0.5856,
      "step": 12720,
      "vm_loss": 0.2079
    },
    {
      "epoch": 0.9332611865696965,
      "lm_loss": 0.7804,
      "step": 12720,
      "vm_loss": 0.1592
    },
    {
      "epoch": 0.9332611865696965,
      "lm_loss": 0.6625,
      "step": 12720,
      "vm_loss": 0.129
    },
    {
      "epoch": 0.9332611865696965,
      "lm_loss": 0.1073,
      "step": 12720,
      "vm_loss": 0.1123
    },
    {
      "epoch": 0.9333345561598356,
      "grad_norm": 1.375394146098306,
      "learning_rate": 1.1594788768672982e-07,
      "loss": 0.8423,
      "step": 12721
    },
    {
      "epoch": 0.9334079257499748,
      "grad_norm": 1.3728754856280003,
      "learning_rate": 1.1569362545322038e-07,
      "loss": 0.606,
      "step": 12722
    },
    {
      "epoch": 0.9334812953401139,
      "grad_norm": 0.5301334693580028,
      "learning_rate": 1.1543963904858791e-07,
      "loss": 0.4607,
      "step": 12723
    },
    {
      "epoch": 0.9335546649302531,
      "grad_norm": 1.536862564437642,
      "learning_rate": 1.1518592848717535e-07,
      "loss": 0.8388,
      "step": 12724
    },
    {
      "epoch": 0.9336280345203921,
      "grad_norm": 0.5317242467837674,
      "learning_rate": 1.1493249378331073e-07,
      "loss": 0.488,
      "step": 12725
    },
    {
      "epoch": 0.9337014041105313,
      "grad_norm": 0.5567632120846575,
      "learning_rate": 1.146793349513059e-07,
      "loss": 0.473,
      "step": 12726
    },
    {
      "epoch": 0.9337747737006704,
      "grad_norm": 1.5649587367017894,
      "learning_rate": 1.1442645200545666e-07,
      "loss": 0.7424,
      "step": 12727
    },
    {
      "epoch": 0.9338481432908096,
      "grad_norm": 1.4874814549876751,
      "learning_rate": 1.1417384496004546e-07,
      "loss": 0.8001,
      "step": 12728
    },
    {
      "epoch": 0.9338481432908096,
      "lm_loss": 0.3439,
      "step": 12728,
      "vm_loss": 0.0745
    },
    {
      "epoch": 0.9338481432908096,
      "lm_loss": 1.0607,
      "step": 12728,
      "vm_loss": 0.1194
    },
    {
      "epoch": 0.9338481432908096,
      "lm_loss": 0.0846,
      "step": 12728,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.9338481432908096,
      "lm_loss": 0.3117,
      "step": 12728,
      "vm_loss": 0.1963
    },
    {
      "epoch": 0.9338481432908096,
      "lm_loss": 0.8364,
      "step": 12728,
      "vm_loss": 0.1096
    },
    {
      "epoch": 0.9338481432908096,
      "lm_loss": 0.2707,
      "step": 12728,
      "vm_loss": 0.1794
    },
    {
      "epoch": 0.9338481432908096,
      "lm_loss": 0.7341,
      "step": 12728,
      "vm_loss": 0.0965
    },
    {
      "epoch": 0.9338481432908096,
      "lm_loss": 0.1669,
      "step": 12728,
      "vm_loss": 0.1182
    },
    {
      "epoch": 0.9339215128809487,
      "grad_norm": 1.4421547331650724,
      "learning_rate": 1.1392151382933647e-07,
      "loss": 0.7799,
      "step": 12729
    },
    {
      "epoch": 0.9339948824710878,
      "grad_norm": 1.4666136267388745,
      "learning_rate": 1.136694586275794e-07,
      "loss": 0.6959,
      "step": 12730
    },
    {
      "epoch": 0.9340682520612269,
      "grad_norm": 0.5521764701928616,
      "learning_rate": 1.134176793690095e-07,
      "loss": 0.4903,
      "step": 12731
    },
    {
      "epoch": 0.934141621651366,
      "grad_norm": 1.594839433559003,
      "learning_rate": 1.1316617606784375e-07,
      "loss": 0.6522,
      "step": 12732
    },
    {
      "epoch": 0.9342149912415052,
      "grad_norm": 1.421170099899371,
      "learning_rate": 1.1291494873828634e-07,
      "loss": 0.6897,
      "step": 12733
    },
    {
      "epoch": 0.9342883608316443,
      "grad_norm": 1.532703158041689,
      "learning_rate": 1.1266399739452482e-07,
      "loss": 0.695,
      "step": 12734
    },
    {
      "epoch": 0.9343617304217834,
      "grad_norm": 0.5499184365547696,
      "learning_rate": 1.1241332205073064e-07,
      "loss": 0.4801,
      "step": 12735
    },
    {
      "epoch": 0.9344351000119225,
      "grad_norm": 1.5154327902990123,
      "learning_rate": 1.1216292272106021e-07,
      "loss": 0.7998,
      "step": 12736
    },
    {
      "epoch": 0.9344351000119225,
      "lm_loss": 1.094,
      "step": 12736,
      "vm_loss": 0.0407
    },
    {
      "epoch": 0.9344351000119225,
      "lm_loss": 1.3879,
      "step": 12736,
      "vm_loss": 0.0834
    },
    {
      "epoch": 0.9344351000119225,
      "lm_loss": 0.6077,
      "step": 12736,
      "vm_loss": 0.1211
    },
    {
      "epoch": 0.9344351000119225,
      "lm_loss": 0.7368,
      "step": 12736,
      "vm_loss": 0.0707
    },
    {
      "epoch": 0.9344351000119225,
      "lm_loss": 0.1858,
      "step": 12736,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.9344351000119225,
      "lm_loss": 0.4421,
      "step": 12736,
      "vm_loss": 0.1045
    },
    {
      "epoch": 0.9344351000119225,
      "lm_loss": 0.3175,
      "step": 12736,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.9344351000119225,
      "lm_loss": 0.5298,
      "step": 12736,
      "vm_loss": 0.1593
    },
    {
      "epoch": 0.9345084696020617,
      "grad_norm": 1.5395673707712565,
      "learning_rate": 1.1191279941965394e-07,
      "loss": 0.7288,
      "step": 12737
    },
    {
      "epoch": 0.9345818391922008,
      "grad_norm": 1.426659492140476,
      "learning_rate": 1.1166295216063605e-07,
      "loss": 0.7564,
      "step": 12738
    },
    {
      "epoch": 0.93465520878234,
      "grad_norm": 0.5456264947905796,
      "learning_rate": 1.1141338095811804e-07,
      "loss": 0.4839,
      "step": 12739
    },
    {
      "epoch": 0.934728578372479,
      "grad_norm": 0.5515785219649333,
      "learning_rate": 1.1116408582619254e-07,
      "loss": 0.4729,
      "step": 12740
    },
    {
      "epoch": 0.9348019479626182,
      "grad_norm": 1.5541959353243866,
      "learning_rate": 1.1091506677893826e-07,
      "loss": 0.7206,
      "step": 12741
    },
    {
      "epoch": 0.9348753175527573,
      "grad_norm": 0.5495407610426385,
      "learning_rate": 1.1066632383041787e-07,
      "loss": 0.462,
      "step": 12742
    },
    {
      "epoch": 0.9349486871428965,
      "grad_norm": 1.459730059246586,
      "learning_rate": 1.1041785699467788e-07,
      "loss": 0.6674,
      "step": 12743
    },
    {
      "epoch": 0.9350220567330356,
      "grad_norm": 0.5647138487554605,
      "learning_rate": 1.1016966628575044e-07,
      "loss": 0.4888,
      "step": 12744
    },
    {
      "epoch": 0.9350220567330356,
      "lm_loss": 1.2942,
      "step": 12744,
      "vm_loss": 0.102
    },
    {
      "epoch": 0.9350220567330356,
      "lm_loss": 0.5826,
      "step": 12744,
      "vm_loss": 0.1042
    },
    {
      "epoch": 0.9350220567330356,
      "lm_loss": 0.0461,
      "step": 12744,
      "vm_loss": 0.0488
    },
    {
      "epoch": 0.9350220567330356,
      "lm_loss": 0.8115,
      "step": 12744,
      "vm_loss": 0.1791
    },
    {
      "epoch": 0.9350220567330356,
      "lm_loss": 0.3838,
      "step": 12744,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.9350220567330356,
      "lm_loss": 1.4503,
      "step": 12744,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.9350220567330356,
      "lm_loss": 0.8652,
      "step": 12744,
      "vm_loss": 0.1649
    },
    {
      "epoch": 0.9350220567330356,
      "lm_loss": 0.3426,
      "step": 12744,
      "vm_loss": 0.0679
    },
    {
      "epoch": 0.9350954263231747,
      "grad_norm": 1.7643170286115537,
      "learning_rate": 1.0992175171765207e-07,
      "loss": 0.6938,
      "step": 12745
    },
    {
      "epoch": 0.9351687959133138,
      "grad_norm": 1.479204267131799,
      "learning_rate": 1.0967411330438216e-07,
      "loss": 0.7479,
      "step": 12746
    },
    {
      "epoch": 0.935242165503453,
      "grad_norm": 1.4460784309870356,
      "learning_rate": 1.0942675105992562e-07,
      "loss": 0.7435,
      "step": 12747
    },
    {
      "epoch": 0.9353155350935921,
      "grad_norm": 0.5433804236430669,
      "learning_rate": 1.0917966499825183e-07,
      "loss": 0.4922,
      "step": 12748
    },
    {
      "epoch": 0.9353889046837313,
      "grad_norm": 1.6638477711626893,
      "learning_rate": 1.0893285513331353e-07,
      "loss": 0.8034,
      "step": 12749
    },
    {
      "epoch": 0.9354622742738703,
      "grad_norm": 1.4750245510024038,
      "learning_rate": 1.0868632147905012e-07,
      "loss": 0.7534,
      "step": 12750
    },
    {
      "epoch": 0.9355356438640094,
      "grad_norm": 1.4547226245652283,
      "learning_rate": 1.0844006404938323e-07,
      "loss": 0.6761,
      "step": 12751
    },
    {
      "epoch": 0.9356090134541486,
      "grad_norm": 1.4378442418835509,
      "learning_rate": 1.0819408285822008e-07,
      "loss": 0.6612,
      "step": 12752
    },
    {
      "epoch": 0.9356090134541486,
      "lm_loss": 0.6542,
      "step": 12752,
      "vm_loss": 0.0469
    },
    {
      "epoch": 0.9356090134541486,
      "lm_loss": 0.1051,
      "step": 12752,
      "vm_loss": 0.0576
    },
    {
      "epoch": 0.9356090134541486,
      "lm_loss": 0.2251,
      "step": 12752,
      "vm_loss": 0.1024
    },
    {
      "epoch": 0.9356090134541486,
      "lm_loss": 0.5772,
      "step": 12752,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.9356090134541486,
      "lm_loss": 0.7771,
      "step": 12752,
      "vm_loss": 0.1838
    },
    {
      "epoch": 0.9356090134541486,
      "lm_loss": 0.1027,
      "step": 12752,
      "vm_loss": 0.1755
    },
    {
      "epoch": 0.9356090134541486,
      "lm_loss": 0.0681,
      "step": 12752,
      "vm_loss": 0.0591
    },
    {
      "epoch": 0.9356090134541486,
      "lm_loss": 0.8341,
      "step": 12752,
      "vm_loss": 0.1897
    },
    {
      "epoch": 0.9356823830442877,
      "grad_norm": 1.555588994968893,
      "learning_rate": 1.0794837791945068e-07,
      "loss": 0.7213,
      "step": 12753
    },
    {
      "epoch": 0.9357557526344269,
      "grad_norm": 1.4766018576584135,
      "learning_rate": 1.0770294924695168e-07,
      "loss": 0.7783,
      "step": 12754
    },
    {
      "epoch": 0.9358291222245659,
      "grad_norm": 1.5847207131971186,
      "learning_rate": 1.0745779685458257e-07,
      "loss": 0.7795,
      "step": 12755
    },
    {
      "epoch": 0.9359024918147051,
      "grad_norm": 0.5584317434200319,
      "learning_rate": 1.0721292075618783e-07,
      "loss": 0.4732,
      "step": 12756
    },
    {
      "epoch": 0.9359758614048442,
      "grad_norm": 1.7041157556683506,
      "learning_rate": 1.0696832096559694e-07,
      "loss": 0.8114,
      "step": 12757
    },
    {
      "epoch": 0.9360492309949834,
      "grad_norm": 1.573181193219616,
      "learning_rate": 1.0672399749662166e-07,
      "loss": 0.6903,
      "step": 12758
    },
    {
      "epoch": 0.9361226005851225,
      "grad_norm": 1.5522513765696127,
      "learning_rate": 1.064799503630598e-07,
      "loss": 0.7718,
      "step": 12759
    },
    {
      "epoch": 0.9361959701752616,
      "grad_norm": 1.5252342428378538,
      "learning_rate": 1.0623617957869482e-07,
      "loss": 0.7478,
      "step": 12760
    },
    {
      "epoch": 0.9361959701752616,
      "lm_loss": 1.0586,
      "step": 12760,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.9361959701752616,
      "lm_loss": 0.5499,
      "step": 12760,
      "vm_loss": 0.0664
    },
    {
      "epoch": 0.9361959701752616,
      "lm_loss": 0.8845,
      "step": 12760,
      "vm_loss": 0.1487
    },
    {
      "epoch": 0.9361959701752616,
      "lm_loss": 0.7978,
      "step": 12760,
      "vm_loss": 0.1586
    },
    {
      "epoch": 0.9361959701752616,
      "lm_loss": 0.7442,
      "step": 12760,
      "vm_loss": 0.0692
    },
    {
      "epoch": 0.9361959701752616,
      "lm_loss": 0.7883,
      "step": 12760,
      "vm_loss": 0.0563
    },
    {
      "epoch": 0.9361959701752616,
      "lm_loss": 0.213,
      "step": 12760,
      "vm_loss": 0.0979
    },
    {
      "epoch": 0.9361959701752616,
      "lm_loss": 0.3554,
      "step": 12760,
      "vm_loss": 0.0323
    },
    {
      "epoch": 0.9362693397654007,
      "grad_norm": 1.3782523500230424,
      "learning_rate": 1.0599268515729122e-07,
      "loss": 0.5933,
      "step": 12761
    },
    {
      "epoch": 0.9363427093555399,
      "grad_norm": 0.5279638079345063,
      "learning_rate": 1.0574946711260082e-07,
      "loss": 0.476,
      "step": 12762
    },
    {
      "epoch": 0.936416078945679,
      "grad_norm": 1.4610805525100214,
      "learning_rate": 1.0550652545835871e-07,
      "loss": 0.6395,
      "step": 12763
    },
    {
      "epoch": 0.9364894485358182,
      "grad_norm": 1.5385409694144656,
      "learning_rate": 1.0526386020828338e-07,
      "loss": 0.779,
      "step": 12764
    },
    {
      "epoch": 0.9365628181259572,
      "grad_norm": 0.54886016944988,
      "learning_rate": 1.0502147137607942e-07,
      "loss": 0.4846,
      "step": 12765
    },
    {
      "epoch": 0.9366361877160964,
      "grad_norm": 1.6587560236933996,
      "learning_rate": 1.0477935897543589e-07,
      "loss": 0.704,
      "step": 12766
    },
    {
      "epoch": 0.9367095573062355,
      "grad_norm": 1.495059995059511,
      "learning_rate": 1.0453752302002462e-07,
      "loss": 0.7308,
      "step": 12767
    },
    {
      "epoch": 0.9367829268963747,
      "grad_norm": 1.5504844235403654,
      "learning_rate": 1.0429596352350246e-07,
      "loss": 0.7793,
      "step": 12768
    },
    {
      "epoch": 0.9367829268963747,
      "lm_loss": 1.1018,
      "step": 12768,
      "vm_loss": 0.0527
    },
    {
      "epoch": 0.9367829268963747,
      "lm_loss": 1.6179,
      "step": 12768,
      "vm_loss": 0.2039
    },
    {
      "epoch": 0.9367829268963747,
      "lm_loss": 0.32,
      "step": 12768,
      "vm_loss": 0.1964
    },
    {
      "epoch": 0.9367829268963747,
      "lm_loss": 0.0647,
      "step": 12768,
      "vm_loss": 0.0626
    },
    {
      "epoch": 0.9367829268963747,
      "lm_loss": 0.8315,
      "step": 12768,
      "vm_loss": 0.1281
    },
    {
      "epoch": 0.9367829268963747,
      "lm_loss": 0.1833,
      "step": 12768,
      "vm_loss": 0.099
    },
    {
      "epoch": 0.9367829268963747,
      "lm_loss": 0.3112,
      "step": 12768,
      "vm_loss": 0.0851
    },
    {
      "epoch": 0.9367829268963747,
      "lm_loss": 0.5744,
      "step": 12768,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.9368562964865138,
      "grad_norm": 1.5153144952814677,
      "learning_rate": 1.0405468049951184e-07,
      "loss": 0.694,
      "step": 12769
    },
    {
      "epoch": 0.9369296660766528,
      "grad_norm": 1.4807672568106158,
      "learning_rate": 1.0381367396167741e-07,
      "loss": 0.7605,
      "step": 12770
    },
    {
      "epoch": 0.937003035666792,
      "grad_norm": 0.5432032746921921,
      "learning_rate": 1.0357294392360994e-07,
      "loss": 0.4638,
      "step": 12771
    },
    {
      "epoch": 0.9370764052569311,
      "grad_norm": 0.5299828089576809,
      "learning_rate": 1.0333249039890469e-07,
      "loss": 0.465,
      "step": 12772
    },
    {
      "epoch": 0.9371497748470703,
      "grad_norm": 1.4733393544674367,
      "learning_rate": 1.0309231340114024e-07,
      "loss": 0.7918,
      "step": 12773
    },
    {
      "epoch": 0.9372231444372093,
      "grad_norm": 1.4507871578216525,
      "learning_rate": 1.0285241294387904e-07,
      "loss": 0.6897,
      "step": 12774
    },
    {
      "epoch": 0.9372965140273485,
      "grad_norm": 1.5999785369340276,
      "learning_rate": 1.0261278904066974e-07,
      "loss": 0.7003,
      "step": 12775
    },
    {
      "epoch": 0.9373698836174876,
      "grad_norm": 1.8206510685738362,
      "learning_rate": 1.0237344170504481e-07,
      "loss": 0.7213,
      "step": 12776
    },
    {
      "epoch": 0.9373698836174876,
      "lm_loss": 1.3666,
      "step": 12776,
      "vm_loss": 0.0464
    },
    {
      "epoch": 0.9373698836174876,
      "lm_loss": 2.7084,
      "step": 12776,
      "vm_loss": 0.0417
    },
    {
      "epoch": 0.9373698836174876,
      "lm_loss": 0.0486,
      "step": 12776,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.9373698836174876,
      "lm_loss": 0.1119,
      "step": 12776,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.9373698836174876,
      "lm_loss": 0.8741,
      "step": 12776,
      "vm_loss": 0.1111
    },
    {
      "epoch": 0.9373698836174876,
      "lm_loss": 1.3956,
      "step": 12776,
      "vm_loss": 0.0333
    },
    {
      "epoch": 0.9373698836174876,
      "lm_loss": 0.3927,
      "step": 12776,
      "vm_loss": 0.1133
    },
    {
      "epoch": 0.9373698836174876,
      "lm_loss": 1.3282,
      "step": 12776,
      "vm_loss": 0.1781
    },
    {
      "epoch": 0.9374432532076268,
      "grad_norm": 1.5306195017575168,
      "learning_rate": 1.021343709505207e-07,
      "loss": 0.8105,
      "step": 12777
    },
    {
      "epoch": 0.9375166227977659,
      "grad_norm": 1.470955842844053,
      "learning_rate": 1.018955767905977e-07,
      "loss": 0.7253,
      "step": 12778
    },
    {
      "epoch": 0.937589992387905,
      "grad_norm": 1.569889407550181,
      "learning_rate": 1.0165705923876113e-07,
      "loss": 0.8041,
      "step": 12779
    },
    {
      "epoch": 0.9376633619780441,
      "grad_norm": 0.5568004035873712,
      "learning_rate": 1.0141881830848133e-07,
      "loss": 0.4692,
      "step": 12780
    },
    {
      "epoch": 0.9377367315681833,
      "grad_norm": 0.5246464913418865,
      "learning_rate": 1.0118085401321198e-07,
      "loss": 0.4612,
      "step": 12781
    },
    {
      "epoch": 0.9378101011583224,
      "grad_norm": 0.5528713616726462,
      "learning_rate": 1.0094316636639123e-07,
      "loss": 0.4711,
      "step": 12782
    },
    {
      "epoch": 0.9378834707484616,
      "grad_norm": 1.69700177895122,
      "learning_rate": 1.0070575538144223e-07,
      "loss": 0.7865,
      "step": 12783
    },
    {
      "epoch": 0.9379568403386006,
      "grad_norm": 0.5382223466824685,
      "learning_rate": 1.0046862107177258e-07,
      "loss": 0.4849,
      "step": 12784
    },
    {
      "epoch": 0.9379568403386006,
      "lm_loss": 0.7025,
      "step": 12784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9379568403386006,
      "lm_loss": 0.4573,
      "step": 12784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9379568403386006,
      "lm_loss": 0.5932,
      "step": 12784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9379568403386006,
      "lm_loss": 0.4059,
      "step": 12784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9379568403386006,
      "lm_loss": 0.4659,
      "step": 12784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9379568403386006,
      "lm_loss": 0.4619,
      "step": 12784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9379568403386006,
      "lm_loss": 0.5339,
      "step": 12784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9379568403386006,
      "lm_loss": 0.3365,
      "step": 12784,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9380302099287398,
      "grad_norm": 0.5434096312541367,
      "learning_rate": 1.0023176345077323e-07,
      "loss": 0.4574,
      "step": 12785
    },
    {
      "epoch": 0.9381035795188789,
      "grad_norm": 1.5215017287175294,
      "learning_rate": 9.999518253181962e-08,
      "loss": 0.7644,
      "step": 12786
    },
    {
      "epoch": 0.938176949109018,
      "grad_norm": 1.3876115794205057,
      "learning_rate": 9.975887832827325e-08,
      "loss": 0.6408,
      "step": 12787
    },
    {
      "epoch": 0.9382503186991572,
      "grad_norm": 1.5613261856587466,
      "learning_rate": 9.952285085347845e-08,
      "loss": 0.7839,
      "step": 12788
    },
    {
      "epoch": 0.9383236882892962,
      "grad_norm": 1.4053298940569074,
      "learning_rate": 9.928710012076404e-08,
      "loss": 0.7126,
      "step": 12789
    },
    {
      "epoch": 0.9383970578794354,
      "grad_norm": 1.5417744825237854,
      "learning_rate": 9.905162614344322e-08,
      "loss": 0.7752,
      "step": 12790
    },
    {
      "epoch": 0.9384704274695745,
      "grad_norm": 1.4238665208963261,
      "learning_rate": 9.881642893481425e-08,
      "loss": 0.6291,
      "step": 12791
    },
    {
      "epoch": 0.9385437970597137,
      "grad_norm": 1.5790490731719018,
      "learning_rate": 9.858150850815873e-08,
      "loss": 0.6679,
      "step": 12792
    },
    {
      "epoch": 0.9385437970597137,
      "lm_loss": 0.5406,
      "step": 12792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9385437970597137,
      "lm_loss": 0.3568,
      "step": 12792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9385437970597137,
      "lm_loss": 0.373,
      "step": 12792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9385437970597137,
      "lm_loss": 0.6236,
      "step": 12792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9385437970597137,
      "lm_loss": 0.435,
      "step": 12792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9385437970597137,
      "lm_loss": 0.4403,
      "step": 12792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9385437970597137,
      "lm_loss": 0.4547,
      "step": 12792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9385437970597137,
      "lm_loss": 0.5557,
      "step": 12792,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9386171666498528,
      "grad_norm": 0.5507864105095186,
      "learning_rate": 9.834686487674383e-08,
      "loss": 0.4894,
      "step": 12793
    },
    {
      "epoch": 0.9386905362399919,
      "grad_norm": 1.4474109688432775,
      "learning_rate": 9.811249805382061e-08,
      "loss": 0.6958,
      "step": 12794
    },
    {
      "epoch": 0.938763905830131,
      "grad_norm": 0.5494345379734056,
      "learning_rate": 9.787840805262349e-08,
      "loss": 0.488,
      "step": 12795
    },
    {
      "epoch": 0.9388372754202702,
      "grad_norm": 0.5431451403046029,
      "learning_rate": 9.764459488637246e-08,
      "loss": 0.4772,
      "step": 12796
    },
    {
      "epoch": 0.9389106450104093,
      "grad_norm": 1.4927109643850665,
      "learning_rate": 9.741105856827193e-08,
      "loss": 0.6603,
      "step": 12797
    },
    {
      "epoch": 0.9389840146005485,
      "grad_norm": 1.6252192194352875,
      "learning_rate": 9.71777991115097e-08,
      "loss": 0.6896,
      "step": 12798
    },
    {
      "epoch": 0.9390573841906875,
      "grad_norm": 1.512021937007574,
      "learning_rate": 9.694481652925913e-08,
      "loss": 0.7187,
      "step": 12799
    },
    {
      "epoch": 0.9391307537808267,
      "grad_norm": 0.5287150768729079,
      "learning_rate": 9.671211083467691e-08,
      "loss": 0.4519,
      "step": 12800
    },
    {
      "epoch": 0.9391307537808267,
      "lm_loss": 1.2122,
      "step": 12800,
      "vm_loss": 0.0505
    },
    {
      "epoch": 0.9391307537808267,
      "lm_loss": 1.5736,
      "step": 12800,
      "vm_loss": 0.1162
    },
    {
      "epoch": 0.9391307537808267,
      "lm_loss": 0.6667,
      "step": 12800,
      "vm_loss": 0.0809
    },
    {
      "epoch": 0.9391307537808267,
      "lm_loss": 0.736,
      "step": 12800,
      "vm_loss": 0.1282
    },
    {
      "epoch": 0.9391307537808267,
      "lm_loss": 0.054,
      "step": 12800,
      "vm_loss": 0.1037
    },
    {
      "epoch": 0.9391307537808267,
      "lm_loss": 0.1788,
      "step": 12800,
      "vm_loss": 0.0836
    },
    {
      "epoch": 0.9391307537808267,
      "lm_loss": 0.7392,
      "step": 12800,
      "vm_loss": 0.2425
    },
    {
      "epoch": 0.9391307537808267,
      "lm_loss": 0.2555,
      "step": 12800,
      "vm_loss": 0.1757
    },
    {
      "epoch": 0.9392041233709658,
      "grad_norm": 1.4997705166630741,
      "learning_rate": 9.647968204090474e-08,
      "loss": 0.7847,
      "step": 12801
    },
    {
      "epoch": 0.939277492961105,
      "grad_norm": 1.4742883310257233,
      "learning_rate": 9.624753016106769e-08,
      "loss": 0.6983,
      "step": 12802
    },
    {
      "epoch": 0.9393508625512441,
      "grad_norm": 1.578380182587256,
      "learning_rate": 9.60156552082775e-08,
      "loss": 0.7768,
      "step": 12803
    },
    {
      "epoch": 0.9394242321413832,
      "grad_norm": 0.5403452150517905,
      "learning_rate": 9.5784057195627e-08,
      "loss": 0.4745,
      "step": 12804
    },
    {
      "epoch": 0.9394976017315223,
      "grad_norm": 0.5600334391473769,
      "learning_rate": 9.555273613619631e-08,
      "loss": 0.4806,
      "step": 12805
    },
    {
      "epoch": 0.9395709713216615,
      "grad_norm": 1.3954703337468033,
      "learning_rate": 9.532169204304776e-08,
      "loss": 0.7502,
      "step": 12806
    },
    {
      "epoch": 0.9396443409118006,
      "grad_norm": 1.5208451614461447,
      "learning_rate": 9.509092492922978e-08,
      "loss": 0.6986,
      "step": 12807
    },
    {
      "epoch": 0.9397177105019398,
      "grad_norm": 1.5413008960462424,
      "learning_rate": 9.486043480777362e-08,
      "loss": 0.7909,
      "step": 12808
    },
    {
      "epoch": 0.9397177105019398,
      "lm_loss": 0.8062,
      "step": 12808,
      "vm_loss": 0.0578
    },
    {
      "epoch": 0.9397177105019398,
      "lm_loss": 0.3022,
      "step": 12808,
      "vm_loss": 0.0709
    },
    {
      "epoch": 0.9397177105019398,
      "lm_loss": 0.5036,
      "step": 12808,
      "vm_loss": 0.0999
    },
    {
      "epoch": 0.9397177105019398,
      "lm_loss": 0.7975,
      "step": 12808,
      "vm_loss": 0.1795
    },
    {
      "epoch": 0.9397177105019398,
      "lm_loss": 0.5,
      "step": 12808,
      "vm_loss": 0.1402
    },
    {
      "epoch": 0.9397177105019398,
      "lm_loss": 0.1507,
      "step": 12808,
      "vm_loss": 0.1453
    },
    {
      "epoch": 0.9397177105019398,
      "lm_loss": 0.3447,
      "step": 12808,
      "vm_loss": 0.1075
    },
    {
      "epoch": 0.9397177105019398,
      "lm_loss": 0.9556,
      "step": 12808,
      "vm_loss": 0.1447
    },
    {
      "epoch": 0.9397910800920788,
      "grad_norm": 1.4453029096715129,
      "learning_rate": 9.463022169169666e-08,
      "loss": 0.6737,
      "step": 12809
    },
    {
      "epoch": 0.9398644496822179,
      "grad_norm": 0.5417607222878416,
      "learning_rate": 9.440028559399905e-08,
      "loss": 0.4628,
      "step": 12810
    },
    {
      "epoch": 0.9399378192723571,
      "grad_norm": 0.5392269576341914,
      "learning_rate": 9.417062652766595e-08,
      "loss": 0.4893,
      "step": 12811
    },
    {
      "epoch": 0.9400111888624962,
      "grad_norm": 1.49500527896397,
      "learning_rate": 9.394124450566589e-08,
      "loss": 0.715,
      "step": 12812
    },
    {
      "epoch": 0.9400845584526354,
      "grad_norm": 1.4663404296601694,
      "learning_rate": 9.371213954095405e-08,
      "loss": 0.6554,
      "step": 12813
    },
    {
      "epoch": 0.9401579280427744,
      "grad_norm": 1.7529888813650936,
      "learning_rate": 9.348331164646785e-08,
      "loss": 0.7757,
      "step": 12814
    },
    {
      "epoch": 0.9402312976329136,
      "grad_norm": 1.6704598139366826,
      "learning_rate": 9.325476083512974e-08,
      "loss": 0.6242,
      "step": 12815
    },
    {
      "epoch": 0.9403046672230527,
      "grad_norm": 1.529142235139966,
      "learning_rate": 9.302648711984663e-08,
      "loss": 0.6827,
      "step": 12816
    },
    {
      "epoch": 0.9403046672230527,
      "lm_loss": 1.1797,
      "step": 12816,
      "vm_loss": 0.0375
    },
    {
      "epoch": 0.9403046672230527,
      "lm_loss": 0.995,
      "step": 12816,
      "vm_loss": 0.0339
    },
    {
      "epoch": 0.9403046672230527,
      "lm_loss": 0.2645,
      "step": 12816,
      "vm_loss": 0.1408
    },
    {
      "epoch": 0.9403046672230527,
      "lm_loss": 0.7504,
      "step": 12816,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.9403046672230527,
      "lm_loss": 0.003,
      "step": 12816,
      "vm_loss": 0.1354
    },
    {
      "epoch": 0.9403046672230527,
      "lm_loss": 0.701,
      "step": 12816,
      "vm_loss": 0.0573
    },
    {
      "epoch": 0.9403046672230527,
      "lm_loss": 1.0351,
      "step": 12816,
      "vm_loss": 0.2219
    },
    {
      "epoch": 0.9403046672230527,
      "lm_loss": 0.674,
      "step": 12816,
      "vm_loss": 0.2287
    },
    {
      "epoch": 0.9403780368131919,
      "grad_norm": 1.7255168723799659,
      "learning_rate": 9.279849051350932e-08,
      "loss": 0.8091,
      "step": 12817
    },
    {
      "epoch": 0.940451406403331,
      "grad_norm": 1.1877979989112397,
      "learning_rate": 9.257077102899414e-08,
      "loss": 0.6395,
      "step": 12818
    },
    {
      "epoch": 0.9405247759934701,
      "grad_norm": 0.5852141371477217,
      "learning_rate": 9.234332867916029e-08,
      "loss": 0.5013,
      "step": 12819
    },
    {
      "epoch": 0.9405981455836092,
      "grad_norm": 0.5521988899717182,
      "learning_rate": 9.211616347685248e-08,
      "loss": 0.4879,
      "step": 12820
    },
    {
      "epoch": 0.9406715151737484,
      "grad_norm": 1.5684740321217465,
      "learning_rate": 9.188927543489934e-08,
      "loss": 0.6761,
      "step": 12821
    },
    {
      "epoch": 0.9407448847638875,
      "grad_norm": 0.5400800246037575,
      "learning_rate": 9.166266456611339e-08,
      "loss": 0.4765,
      "step": 12822
    },
    {
      "epoch": 0.9408182543540267,
      "grad_norm": 0.5390748431586352,
      "learning_rate": 9.143633088329163e-08,
      "loss": 0.4648,
      "step": 12823
    },
    {
      "epoch": 0.9408916239441657,
      "grad_norm": 1.487280544491123,
      "learning_rate": 9.121027439921659e-08,
      "loss": 0.664,
      "step": 12824
    },
    {
      "epoch": 0.9408916239441657,
      "lm_loss": 0.4108,
      "step": 12824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9408916239441657,
      "lm_loss": 0.342,
      "step": 12824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9408916239441657,
      "lm_loss": 0.2989,
      "step": 12824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9408916239441657,
      "lm_loss": 0.5258,
      "step": 12824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9408916239441657,
      "lm_loss": 0.4999,
      "step": 12824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9408916239441657,
      "lm_loss": 0.5608,
      "step": 12824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9408916239441657,
      "lm_loss": 0.5583,
      "step": 12824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9408916239441657,
      "lm_loss": 0.5737,
      "step": 12824,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9409649935343048,
      "grad_norm": 0.5328175208297486,
      "learning_rate": 9.098449512665363e-08,
      "loss": 0.4842,
      "step": 12825
    },
    {
      "epoch": 0.941038363124444,
      "grad_norm": 0.5424421841154741,
      "learning_rate": 9.075899307835312e-08,
      "loss": 0.4606,
      "step": 12826
    },
    {
      "epoch": 0.9411117327145831,
      "grad_norm": 0.5405801629801255,
      "learning_rate": 9.053376826704985e-08,
      "loss": 0.459,
      "step": 12827
    },
    {
      "epoch": 0.9411851023047223,
      "grad_norm": 1.5261480776262926,
      "learning_rate": 9.030882070546255e-08,
      "loss": 0.7926,
      "step": 12828
    },
    {
      "epoch": 0.9412584718948613,
      "grad_norm": 0.5409273278735803,
      "learning_rate": 9.008415040629548e-08,
      "loss": 0.4834,
      "step": 12829
    },
    {
      "epoch": 0.9413318414850005,
      "grad_norm": 1.4402647934331563,
      "learning_rate": 8.985975738223518e-08,
      "loss": 0.6971,
      "step": 12830
    },
    {
      "epoch": 0.9414052110751396,
      "grad_norm": 1.3648883062644526,
      "learning_rate": 8.963564164595429e-08,
      "loss": 0.6364,
      "step": 12831
    },
    {
      "epoch": 0.9414785806652788,
      "grad_norm": 1.5552695565521715,
      "learning_rate": 8.941180321010878e-08,
      "loss": 0.7147,
      "step": 12832
    },
    {
      "epoch": 0.9414785806652788,
      "lm_loss": 0.5286,
      "step": 12832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9414785806652788,
      "lm_loss": 0.478,
      "step": 12832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9414785806652788,
      "lm_loss": 0.4183,
      "step": 12832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9414785806652788,
      "lm_loss": 0.4738,
      "step": 12832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9414785806652788,
      "lm_loss": 0.4288,
      "step": 12832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9414785806652788,
      "lm_loss": 0.5045,
      "step": 12832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9414785806652788,
      "lm_loss": 0.6286,
      "step": 12832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9414785806652788,
      "lm_loss": 0.3572,
      "step": 12832,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9415519502554179,
      "grad_norm": 0.546330040655883,
      "learning_rate": 8.918824208733967e-08,
      "loss": 0.4672,
      "step": 12833
    },
    {
      "epoch": 0.941625319845557,
      "grad_norm": 0.5383420694935732,
      "learning_rate": 8.896495829027185e-08,
      "loss": 0.4628,
      "step": 12834
    },
    {
      "epoch": 0.9416986894356961,
      "grad_norm": 0.5451424445725788,
      "learning_rate": 8.874195183151523e-08,
      "loss": 0.4773,
      "step": 12835
    },
    {
      "epoch": 0.9417720590258353,
      "grad_norm": 1.7569088346611945,
      "learning_rate": 8.851922272366253e-08,
      "loss": 0.7064,
      "step": 12836
    },
    {
      "epoch": 0.9418454286159744,
      "grad_norm": 1.4716493002847129,
      "learning_rate": 8.829677097929312e-08,
      "loss": 0.6719,
      "step": 12837
    },
    {
      "epoch": 0.9419187982061136,
      "grad_norm": 1.4216126551515749,
      "learning_rate": 8.807459661096807e-08,
      "loss": 0.7595,
      "step": 12838
    },
    {
      "epoch": 0.9419921677962526,
      "grad_norm": 0.5523531256005312,
      "learning_rate": 8.785269963123455e-08,
      "loss": 0.4669,
      "step": 12839
    },
    {
      "epoch": 0.9420655373863918,
      "grad_norm": 0.5431865393064849,
      "learning_rate": 8.763108005262422e-08,
      "loss": 0.4853,
      "step": 12840
    },
    {
      "epoch": 0.9420655373863918,
      "lm_loss": 1.5452,
      "step": 12840,
      "vm_loss": 0.1179
    },
    {
      "epoch": 0.9420655373863918,
      "lm_loss": 0.7472,
      "step": 12840,
      "vm_loss": 0.1155
    },
    {
      "epoch": 0.9420655373863918,
      "lm_loss": 0.1469,
      "step": 12840,
      "vm_loss": 0.1128
    },
    {
      "epoch": 0.9420655373863918,
      "lm_loss": 0.0488,
      "step": 12840,
      "vm_loss": 0.088
    },
    {
      "epoch": 0.9420655373863918,
      "lm_loss": 0.9269,
      "step": 12840,
      "vm_loss": 0.0976
    },
    {
      "epoch": 0.9420655373863918,
      "lm_loss": 0.6574,
      "step": 12840,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.9420655373863918,
      "lm_loss": 0.2894,
      "step": 12840,
      "vm_loss": 0.0609
    },
    {
      "epoch": 0.9420655373863918,
      "lm_loss": 0.095,
      "step": 12840,
      "vm_loss": 0.1362
    },
    {
      "epoch": 0.9421389069765309,
      "grad_norm": 1.6004974796165732,
      "learning_rate": 8.740973788765261e-08,
      "loss": 0.7413,
      "step": 12841
    },
    {
      "epoch": 0.9422122765666701,
      "grad_norm": 0.5436906109041704,
      "learning_rate": 8.718867314881807e-08,
      "loss": 0.4751,
      "step": 12842
    },
    {
      "epoch": 0.9422856461568092,
      "grad_norm": 1.5448267976559529,
      "learning_rate": 8.696788584860616e-08,
      "loss": 0.7738,
      "step": 12843
    },
    {
      "epoch": 0.9423590157469482,
      "grad_norm": 1.3809915621918554,
      "learning_rate": 8.674737599948413e-08,
      "loss": 0.6836,
      "step": 12844
    },
    {
      "epoch": 0.9424323853370874,
      "grad_norm": 1.3534679181033746,
      "learning_rate": 8.652714361390535e-08,
      "loss": 0.7547,
      "step": 12845
    },
    {
      "epoch": 0.9425057549272265,
      "grad_norm": 1.3967328776155947,
      "learning_rate": 8.63071887043071e-08,
      "loss": 0.6752,
      "step": 12846
    },
    {
      "epoch": 0.9425791245173657,
      "grad_norm": 0.5304572463514581,
      "learning_rate": 8.608751128311e-08,
      "loss": 0.4569,
      "step": 12847
    },
    {
      "epoch": 0.9426524941075048,
      "grad_norm": 1.5831010793371705,
      "learning_rate": 8.586811136272078e-08,
      "loss": 0.8286,
      "step": 12848
    },
    {
      "epoch": 0.9426524941075048,
      "lm_loss": 1.1182,
      "step": 12848,
      "vm_loss": 0.0583
    },
    {
      "epoch": 0.9426524941075048,
      "lm_loss": 0.801,
      "step": 12848,
      "vm_loss": 0.1909
    },
    {
      "epoch": 0.9426524941075048,
      "lm_loss": 0.6138,
      "step": 12848,
      "vm_loss": 0.1424
    },
    {
      "epoch": 0.9426524941075048,
      "lm_loss": 0.3972,
      "step": 12848,
      "vm_loss": 0.0899
    },
    {
      "epoch": 0.9426524941075048,
      "lm_loss": 0.9847,
      "step": 12848,
      "vm_loss": 0.0914
    },
    {
      "epoch": 0.9426524941075048,
      "lm_loss": 0.7654,
      "step": 12848,
      "vm_loss": 0.1712
    },
    {
      "epoch": 0.9426524941075048,
      "lm_loss": 0.2707,
      "step": 12848,
      "vm_loss": 0.0884
    },
    {
      "epoch": 0.9426524941075048,
      "lm_loss": 0.0842,
      "step": 12848,
      "vm_loss": 0.1427
    },
    {
      "epoch": 0.9427258636976439,
      "grad_norm": 1.5022633416161688,
      "learning_rate": 8.564898895552843e-08,
      "loss": 0.6822,
      "step": 12849
    },
    {
      "epoch": 0.942799233287783,
      "grad_norm": 1.617874060426897,
      "learning_rate": 8.543014407390804e-08,
      "loss": 0.8258,
      "step": 12850
    },
    {
      "epoch": 0.9428726028779222,
      "grad_norm": 1.597567559366944,
      "learning_rate": 8.521157673021807e-08,
      "loss": 0.8158,
      "step": 12851
    },
    {
      "epoch": 0.9429459724680613,
      "grad_norm": 1.7220068670035118,
      "learning_rate": 8.499328693680198e-08,
      "loss": 0.7346,
      "step": 12852
    },
    {
      "epoch": 0.9430193420582005,
      "grad_norm": 1.424132032088346,
      "learning_rate": 8.477527470598656e-08,
      "loss": 0.6971,
      "step": 12853
    },
    {
      "epoch": 0.9430927116483395,
      "grad_norm": 0.536071719982153,
      "learning_rate": 8.455754005008365e-08,
      "loss": 0.4919,
      "step": 12854
    },
    {
      "epoch": 0.9431660812384787,
      "grad_norm": 1.430664477359207,
      "learning_rate": 8.434008298138896e-08,
      "loss": 0.7845,
      "step": 12855
    },
    {
      "epoch": 0.9432394508286178,
      "grad_norm": 1.741448168208781,
      "learning_rate": 8.412290351218377e-08,
      "loss": 0.7734,
      "step": 12856
    },
    {
      "epoch": 0.9432394508286178,
      "lm_loss": 0.5831,
      "step": 12856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9432394508286178,
      "lm_loss": 0.4422,
      "step": 12856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9432394508286178,
      "lm_loss": 0.3359,
      "step": 12856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9432394508286178,
      "lm_loss": 0.5482,
      "step": 12856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9432394508286178,
      "lm_loss": 0.452,
      "step": 12856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9432394508286178,
      "lm_loss": 0.607,
      "step": 12856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9432394508286178,
      "lm_loss": 0.5711,
      "step": 12856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9432394508286178,
      "lm_loss": 0.5256,
      "step": 12856,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.943312820418757,
      "grad_norm": 0.559496744531799,
      "learning_rate": 8.390600165473217e-08,
      "loss": 0.479,
      "step": 12857
    },
    {
      "epoch": 0.943386190008896,
      "grad_norm": 1.4034850060021187,
      "learning_rate": 8.368937742128325e-08,
      "loss": 0.7196,
      "step": 12858
    },
    {
      "epoch": 0.9434595595990352,
      "grad_norm": 1.428585653843964,
      "learning_rate": 8.347303082406999e-08,
      "loss": 0.7974,
      "step": 12859
    },
    {
      "epoch": 0.9435329291891743,
      "grad_norm": 1.6451803190787266,
      "learning_rate": 8.325696187531041e-08,
      "loss": 0.6906,
      "step": 12860
    },
    {
      "epoch": 0.9436062987793135,
      "grad_norm": 1.620858459347799,
      "learning_rate": 8.304117058720585e-08,
      "loss": 0.7676,
      "step": 12861
    },
    {
      "epoch": 0.9436796683694526,
      "grad_norm": 1.4590343490987805,
      "learning_rate": 8.282565697194323e-08,
      "loss": 0.6361,
      "step": 12862
    },
    {
      "epoch": 0.9437530379595916,
      "grad_norm": 1.6729143354297096,
      "learning_rate": 8.261042104169337e-08,
      "loss": 0.7151,
      "step": 12863
    },
    {
      "epoch": 0.9438264075497308,
      "grad_norm": 1.388428072460868,
      "learning_rate": 8.239546280861043e-08,
      "loss": 0.6573,
      "step": 12864
    },
    {
      "epoch": 0.9438264075497308,
      "lm_loss": 1.3441,
      "step": 12864,
      "vm_loss": 0.0472
    },
    {
      "epoch": 0.9438264075497308,
      "lm_loss": 0.2938,
      "step": 12864,
      "vm_loss": 0.1121
    },
    {
      "epoch": 0.9438264075497308,
      "lm_loss": 0.0097,
      "step": 12864,
      "vm_loss": 0.0901
    },
    {
      "epoch": 0.9438264075497308,
      "lm_loss": 0.4183,
      "step": 12864,
      "vm_loss": 0.1369
    },
    {
      "epoch": 0.9438264075497308,
      "lm_loss": 0.7698,
      "step": 12864,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.9438264075497308,
      "lm_loss": 0.0255,
      "step": 12864,
      "vm_loss": 0.1808
    },
    {
      "epoch": 0.9438264075497308,
      "lm_loss": 0.1454,
      "step": 12864,
      "vm_loss": 0.1176
    },
    {
      "epoch": 0.9438264075497308,
      "lm_loss": 0.7817,
      "step": 12864,
      "vm_loss": 0.108
    },
    {
      "epoch": 0.9438997771398699,
      "grad_norm": 1.64976148157906,
      "learning_rate": 8.218078228483417e-08,
      "loss": 0.7564,
      "step": 12865
    },
    {
      "epoch": 0.9439731467300091,
      "grad_norm": 1.436238693282959,
      "learning_rate": 8.196637948248709e-08,
      "loss": 0.6328,
      "step": 12866
    },
    {
      "epoch": 0.9440465163201482,
      "grad_norm": 1.4900562355846991,
      "learning_rate": 8.17522544136784e-08,
      "loss": 0.7006,
      "step": 12867
    },
    {
      "epoch": 0.9441198859102873,
      "grad_norm": 1.4734186272687138,
      "learning_rate": 8.153840709050009e-08,
      "loss": 0.7868,
      "step": 12868
    },
    {
      "epoch": 0.9441932555004264,
      "grad_norm": 0.5364289964073081,
      "learning_rate": 8.132483752502806e-08,
      "loss": 0.4612,
      "step": 12869
    },
    {
      "epoch": 0.9442666250905656,
      "grad_norm": 1.7149960952892502,
      "learning_rate": 8.111154572932322e-08,
      "loss": 0.7709,
      "step": 12870
    },
    {
      "epoch": 0.9443399946807047,
      "grad_norm": 0.5499036651997701,
      "learning_rate": 8.089853171543094e-08,
      "loss": 0.4915,
      "step": 12871
    },
    {
      "epoch": 0.9444133642708439,
      "grad_norm": 1.5535886943156554,
      "learning_rate": 8.068579549538047e-08,
      "loss": 0.7338,
      "step": 12872
    },
    {
      "epoch": 0.9444133642708439,
      "lm_loss": 0.1558,
      "step": 12872,
      "vm_loss": 0.053
    },
    {
      "epoch": 0.9444133642708439,
      "lm_loss": 0.6697,
      "step": 12872,
      "vm_loss": 0.1517
    },
    {
      "epoch": 0.9444133642708439,
      "lm_loss": 0.3207,
      "step": 12872,
      "vm_loss": 0.1445
    },
    {
      "epoch": 0.9444133642708439,
      "lm_loss": 0.2091,
      "step": 12872,
      "vm_loss": 0.1722
    },
    {
      "epoch": 0.9444133642708439,
      "lm_loss": 0.5666,
      "step": 12872,
      "vm_loss": 0.0996
    },
    {
      "epoch": 0.9444133642708439,
      "lm_loss": 0.0811,
      "step": 12872,
      "vm_loss": 0.1199
    },
    {
      "epoch": 0.9444133642708439,
      "lm_loss": 0.1222,
      "step": 12872,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.9444133642708439,
      "lm_loss": 0.1464,
      "step": 12872,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.9444867338609829,
      "grad_norm": 1.4916878452177098,
      "learning_rate": 8.047333708118554e-08,
      "loss": 0.5899,
      "step": 12873
    },
    {
      "epoch": 0.9445601034511221,
      "grad_norm": 0.5618038646523011,
      "learning_rate": 8.026115648484434e-08,
      "loss": 0.4686,
      "step": 12874
    },
    {
      "epoch": 0.9446334730412612,
      "grad_norm": 0.5524124047949945,
      "learning_rate": 8.00492537183395e-08,
      "loss": 0.4664,
      "step": 12875
    },
    {
      "epoch": 0.9447068426314004,
      "grad_norm": 1.4638371676862527,
      "learning_rate": 7.983762879363699e-08,
      "loss": 0.6985,
      "step": 12876
    },
    {
      "epoch": 0.9447802122215395,
      "grad_norm": 0.5334013786309129,
      "learning_rate": 7.962628172268782e-08,
      "loss": 0.4799,
      "step": 12877
    },
    {
      "epoch": 0.9448535818116786,
      "grad_norm": 1.5344225612362294,
      "learning_rate": 7.941521251742801e-08,
      "loss": 0.672,
      "step": 12878
    },
    {
      "epoch": 0.9449269514018177,
      "grad_norm": 1.483947644522596,
      "learning_rate": 7.920442118977689e-08,
      "loss": 0.7191,
      "step": 12879
    },
    {
      "epoch": 0.9450003209919569,
      "grad_norm": 0.534759151768316,
      "learning_rate": 7.899390775163829e-08,
      "loss": 0.4615,
      "step": 12880
    },
    {
      "epoch": 0.9450003209919569,
      "lm_loss": 1.3707,
      "step": 12880,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.9450003209919569,
      "lm_loss": 0.4639,
      "step": 12880,
      "vm_loss": 0.1149
    },
    {
      "epoch": 0.9450003209919569,
      "lm_loss": 0.0682,
      "step": 12880,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.9450003209919569,
      "lm_loss": 0.2874,
      "step": 12880,
      "vm_loss": 0.1126
    },
    {
      "epoch": 0.9450003209919569,
      "lm_loss": 0.305,
      "step": 12880,
      "vm_loss": 0.0344
    },
    {
      "epoch": 0.9450003209919569,
      "lm_loss": 0.5499,
      "step": 12880,
      "vm_loss": 0.1493
    },
    {
      "epoch": 0.9450003209919569,
      "lm_loss": 0.0836,
      "step": 12880,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.9450003209919569,
      "lm_loss": 0.1475,
      "step": 12880,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.945073690582096,
      "grad_norm": 1.4433996504726103,
      "learning_rate": 7.878367221489991e-08,
      "loss": 0.6967,
      "step": 12881
    },
    {
      "epoch": 0.9451470601722352,
      "grad_norm": 1.6273350711515557,
      "learning_rate": 7.857371459143503e-08,
      "loss": 0.7212,
      "step": 12882
    },
    {
      "epoch": 0.9452204297623742,
      "grad_norm": 1.5337289106396836,
      "learning_rate": 7.83640348931003e-08,
      "loss": 0.6387,
      "step": 12883
    },
    {
      "epoch": 0.9452937993525133,
      "grad_norm": 1.384719483897117,
      "learning_rate": 7.815463313173676e-08,
      "loss": 0.7878,
      "step": 12884
    },
    {
      "epoch": 0.9453671689426525,
      "grad_norm": 1.5279115884313752,
      "learning_rate": 7.794550931916945e-08,
      "loss": 0.7934,
      "step": 12885
    },
    {
      "epoch": 0.9454405385327916,
      "grad_norm": 0.5534809586185497,
      "learning_rate": 7.773666346720887e-08,
      "loss": 0.4945,
      "step": 12886
    },
    {
      "epoch": 0.9455139081229308,
      "grad_norm": 1.4409456962226033,
      "learning_rate": 7.752809558764895e-08,
      "loss": 0.7576,
      "step": 12887
    },
    {
      "epoch": 0.9455872777130698,
      "grad_norm": 1.3909482367573454,
      "learning_rate": 7.731980569226694e-08,
      "loss": 0.7049,
      "step": 12888
    },
    {
      "epoch": 0.9455872777130698,
      "lm_loss": 0.8447,
      "step": 12888,
      "vm_loss": 0.1041
    },
    {
      "epoch": 0.9455872777130698,
      "lm_loss": 0.1092,
      "step": 12888,
      "vm_loss": 0.0487
    },
    {
      "epoch": 0.9455872777130698,
      "lm_loss": 1.0274,
      "step": 12888,
      "vm_loss": 0.2325
    },
    {
      "epoch": 0.9455872777130698,
      "lm_loss": 0.8618,
      "step": 12888,
      "vm_loss": 0.068
    },
    {
      "epoch": 0.9455872777130698,
      "lm_loss": 0.0182,
      "step": 12888,
      "vm_loss": 0.0562
    },
    {
      "epoch": 0.9455872777130698,
      "lm_loss": 0.9823,
      "step": 12888,
      "vm_loss": 0.1322
    },
    {
      "epoch": 0.9455872777130698,
      "lm_loss": 0.1264,
      "step": 12888,
      "vm_loss": 0.2529
    },
    {
      "epoch": 0.9455872777130698,
      "lm_loss": 1.0466,
      "step": 12888,
      "vm_loss": 0.1097
    },
    {
      "epoch": 0.945660647303209,
      "grad_norm": 1.606602664935216,
      "learning_rate": 7.711179379282674e-08,
      "loss": 0.7235,
      "step": 12889
    },
    {
      "epoch": 0.9457340168933481,
      "grad_norm": 1.430774679165399,
      "learning_rate": 7.690405990107508e-08,
      "loss": 0.7681,
      "step": 12890
    },
    {
      "epoch": 0.9458073864834873,
      "grad_norm": 1.673583207532069,
      "learning_rate": 7.669660402874258e-08,
      "loss": 0.8058,
      "step": 12891
    },
    {
      "epoch": 0.9458807560736264,
      "grad_norm": 0.5467912366655681,
      "learning_rate": 7.648942618754596e-08,
      "loss": 0.4842,
      "step": 12892
    },
    {
      "epoch": 0.9459541256637655,
      "grad_norm": 1.4202209304854705,
      "learning_rate": 7.62825263891831e-08,
      "loss": 0.7654,
      "step": 12893
    },
    {
      "epoch": 0.9460274952539046,
      "grad_norm": 1.5421459825523351,
      "learning_rate": 7.607590464534021e-08,
      "loss": 0.7009,
      "step": 12894
    },
    {
      "epoch": 0.9461008648440438,
      "grad_norm": 1.4372126955367786,
      "learning_rate": 7.586956096768461e-08,
      "loss": 0.7019,
      "step": 12895
    },
    {
      "epoch": 0.9461742344341829,
      "grad_norm": 0.5606152001848835,
      "learning_rate": 7.566349536786976e-08,
      "loss": 0.4745,
      "step": 12896
    },
    {
      "epoch": 0.9461742344341829,
      "lm_loss": 1.4252,
      "step": 12896,
      "vm_loss": 0.0538
    },
    {
      "epoch": 0.9461742344341829,
      "lm_loss": 0.0903,
      "step": 12896,
      "vm_loss": 0.0649
    },
    {
      "epoch": 0.9461742344341829,
      "lm_loss": 0.8288,
      "step": 12896,
      "vm_loss": 0.1763
    },
    {
      "epoch": 0.9461742344341829,
      "lm_loss": 0.001,
      "step": 12896,
      "vm_loss": 0.0801
    },
    {
      "epoch": 0.9461742344341829,
      "lm_loss": 0.9171,
      "step": 12896,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.9461742344341829,
      "lm_loss": 0.7787,
      "step": 12896,
      "vm_loss": 0.0881
    },
    {
      "epoch": 0.9461742344341829,
      "lm_loss": 0.2129,
      "step": 12896,
      "vm_loss": 0.1177
    },
    {
      "epoch": 0.9461742344341829,
      "lm_loss": 0.735,
      "step": 12896,
      "vm_loss": 0.1317
    },
    {
      "epoch": 0.9462476040243221,
      "grad_norm": 1.5410702119780775,
      "learning_rate": 7.545770785753193e-08,
      "loss": 0.7192,
      "step": 12897
    },
    {
      "epoch": 0.9463209736144611,
      "grad_norm": 1.6311837948264967,
      "learning_rate": 7.525219844829234e-08,
      "loss": 0.7276,
      "step": 12898
    },
    {
      "epoch": 0.9463943432046003,
      "grad_norm": 1.5631237841581131,
      "learning_rate": 7.50469671517573e-08,
      "loss": 0.6595,
      "step": 12899
    },
    {
      "epoch": 0.9464677127947394,
      "grad_norm": 1.4885868149406951,
      "learning_rate": 7.484201397951696e-08,
      "loss": 0.8008,
      "step": 12900
    },
    {
      "epoch": 0.9465410823848786,
      "grad_norm": 0.5449006761211346,
      "learning_rate": 7.463733894314484e-08,
      "loss": 0.4758,
      "step": 12901
    },
    {
      "epoch": 0.9466144519750177,
      "grad_norm": 1.5343521952259371,
      "learning_rate": 7.443294205419948e-08,
      "loss": 0.8206,
      "step": 12902
    },
    {
      "epoch": 0.9466878215651567,
      "grad_norm": 1.5565435229390125,
      "learning_rate": 7.422882332422387e-08,
      "loss": 0.6841,
      "step": 12903
    },
    {
      "epoch": 0.9467611911552959,
      "grad_norm": 0.5274374723298514,
      "learning_rate": 7.402498276474435e-08,
      "loss": 0.4618,
      "step": 12904
    },
    {
      "epoch": 0.9467611911552959,
      "lm_loss": 1.1248,
      "step": 12904,
      "vm_loss": 0.0372
    },
    {
      "epoch": 0.9467611911552959,
      "lm_loss": 0.0435,
      "step": 12904,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.9467611911552959,
      "lm_loss": 0.1989,
      "step": 12904,
      "vm_loss": 0.0668
    },
    {
      "epoch": 0.9467611911552959,
      "lm_loss": 0.2647,
      "step": 12904,
      "vm_loss": 0.1684
    },
    {
      "epoch": 0.9467611911552959,
      "lm_loss": 0.7526,
      "step": 12904,
      "vm_loss": 0.1203
    },
    {
      "epoch": 0.9467611911552959,
      "lm_loss": 0.0901,
      "step": 12904,
      "vm_loss": 0.1686
    },
    {
      "epoch": 0.9467611911552959,
      "lm_loss": 0.6928,
      "step": 12904,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.9467611911552959,
      "lm_loss": 0.4599,
      "step": 12904,
      "vm_loss": 0.1181
    },
    {
      "epoch": 0.946834560745435,
      "grad_norm": 1.3971775236563315,
      "learning_rate": 7.382142038727392e-08,
      "loss": 0.7241,
      "step": 12905
    },
    {
      "epoch": 0.9469079303355742,
      "grad_norm": 1.7358102561340405,
      "learning_rate": 7.361813620330726e-08,
      "loss": 0.7272,
      "step": 12906
    },
    {
      "epoch": 0.9469812999257133,
      "grad_norm": 1.3859207416793287,
      "learning_rate": 7.341513022432467e-08,
      "loss": 0.757,
      "step": 12907
    },
    {
      "epoch": 0.9470546695158524,
      "grad_norm": 1.7032226277008629,
      "learning_rate": 7.321240246178973e-08,
      "loss": 0.721,
      "step": 12908
    },
    {
      "epoch": 0.9471280391059915,
      "grad_norm": 1.4559260738507718,
      "learning_rate": 7.300995292715107e-08,
      "loss": 0.6849,
      "step": 12909
    },
    {
      "epoch": 0.9472014086961307,
      "grad_norm": 0.5494011099241853,
      "learning_rate": 7.280778163184232e-08,
      "loss": 0.4821,
      "step": 12910
    },
    {
      "epoch": 0.9472747782862698,
      "grad_norm": 1.3346836314002695,
      "learning_rate": 7.26058885872799e-08,
      "loss": 0.7199,
      "step": 12911
    },
    {
      "epoch": 0.947348147876409,
      "grad_norm": 1.4994397145786411,
      "learning_rate": 7.240427380486581e-08,
      "loss": 0.6432,
      "step": 12912
    },
    {
      "epoch": 0.947348147876409,
      "lm_loss": 0.5798,
      "step": 12912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.947348147876409,
      "lm_loss": 0.366,
      "step": 12912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.947348147876409,
      "lm_loss": 0.453,
      "step": 12912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.947348147876409,
      "lm_loss": 0.2235,
      "step": 12912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.947348147876409,
      "lm_loss": 0.5794,
      "step": 12912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.947348147876409,
      "lm_loss": 0.2997,
      "step": 12912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.947348147876409,
      "lm_loss": 0.456,
      "step": 12912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.947348147876409,
      "lm_loss": 0.6035,
      "step": 12912,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.947421517466548,
      "grad_norm": 0.5519389418027294,
      "learning_rate": 7.220293729598538e-08,
      "loss": 0.4571,
      "step": 12913
    },
    {
      "epoch": 0.9474948870566872,
      "grad_norm": 1.6622905650136135,
      "learning_rate": 7.20018790720084e-08,
      "loss": 0.761,
      "step": 12914
    },
    {
      "epoch": 0.9475682566468263,
      "grad_norm": 1.6016372260943328,
      "learning_rate": 7.180109914428912e-08,
      "loss": 0.7377,
      "step": 12915
    },
    {
      "epoch": 0.9476416262369655,
      "grad_norm": 1.4914723040510702,
      "learning_rate": 7.160059752416626e-08,
      "loss": 0.7327,
      "step": 12916
    },
    {
      "epoch": 0.9477149958271046,
      "grad_norm": 0.5348942506802672,
      "learning_rate": 7.140037422296297e-08,
      "loss": 0.4688,
      "step": 12917
    },
    {
      "epoch": 0.9477883654172437,
      "grad_norm": 1.383588982692334,
      "learning_rate": 7.120042925198579e-08,
      "loss": 0.7848,
      "step": 12918
    },
    {
      "epoch": 0.9478617350073828,
      "grad_norm": 1.625931891597102,
      "learning_rate": 7.10007626225262e-08,
      "loss": 0.7029,
      "step": 12919
    },
    {
      "epoch": 0.947935104597522,
      "grad_norm": 1.7022677937645392,
      "learning_rate": 7.080137434585965e-08,
      "loss": 0.8795,
      "step": 12920
    },
    {
      "epoch": 0.947935104597522,
      "lm_loss": 1.2102,
      "step": 12920,
      "vm_loss": 0.0571
    },
    {
      "epoch": 0.947935104597522,
      "lm_loss": 0.7837,
      "step": 12920,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.947935104597522,
      "lm_loss": 0.9888,
      "step": 12920,
      "vm_loss": 0.0808
    },
    {
      "epoch": 0.947935104597522,
      "lm_loss": 0.3883,
      "step": 12920,
      "vm_loss": 0.135
    },
    {
      "epoch": 0.947935104597522,
      "lm_loss": 0.8892,
      "step": 12920,
      "vm_loss": 0.0821
    },
    {
      "epoch": 0.947935104597522,
      "lm_loss": 0.5331,
      "step": 12920,
      "vm_loss": 0.0639
    },
    {
      "epoch": 0.947935104597522,
      "lm_loss": 0.0124,
      "step": 12920,
      "vm_loss": 0.118
    },
    {
      "epoch": 0.947935104597522,
      "lm_loss": 0.6312,
      "step": 12920,
      "vm_loss": 0.0465
    },
    {
      "epoch": 0.9480084741876611,
      "grad_norm": 1.6697460052004602,
      "learning_rate": 7.060226443324714e-08,
      "loss": 0.7348,
      "step": 12921
    },
    {
      "epoch": 0.9480818437778002,
      "grad_norm": 1.3986260973154583,
      "learning_rate": 7.040343289593243e-08,
      "loss": 0.6404,
      "step": 12922
    },
    {
      "epoch": 0.9481552133679393,
      "grad_norm": 0.5575811476527452,
      "learning_rate": 7.020487974514323e-08,
      "loss": 0.4668,
      "step": 12923
    },
    {
      "epoch": 0.9482285829580784,
      "grad_norm": 0.5605696356499177,
      "learning_rate": 7.000660499209277e-08,
      "loss": 0.4771,
      "step": 12924
    },
    {
      "epoch": 0.9483019525482176,
      "grad_norm": 1.5060152493662873,
      "learning_rate": 6.980860864797823e-08,
      "loss": 0.7676,
      "step": 12925
    },
    {
      "epoch": 0.9483753221383567,
      "grad_norm": 0.5519561340062124,
      "learning_rate": 6.961089072398176e-08,
      "loss": 0.4878,
      "step": 12926
    },
    {
      "epoch": 0.9484486917284959,
      "grad_norm": 1.5290839527114815,
      "learning_rate": 6.941345123126774e-08,
      "loss": 0.66,
      "step": 12927
    },
    {
      "epoch": 0.9485220613186349,
      "grad_norm": 1.5026992619902064,
      "learning_rate": 6.921629018098619e-08,
      "loss": 0.7643,
      "step": 12928
    },
    {
      "epoch": 0.9485220613186349,
      "lm_loss": 1.1489,
      "step": 12928,
      "vm_loss": 0.0518
    },
    {
      "epoch": 0.9485220613186349,
      "lm_loss": 1.4893,
      "step": 12928,
      "vm_loss": 0.0948
    },
    {
      "epoch": 0.9485220613186349,
      "lm_loss": 0.0826,
      "step": 12928,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.9485220613186349,
      "lm_loss": 0.5677,
      "step": 12928,
      "vm_loss": 0.1407
    },
    {
      "epoch": 0.9485220613186349,
      "lm_loss": 0.7302,
      "step": 12928,
      "vm_loss": 0.1161
    },
    {
      "epoch": 0.9485220613186349,
      "lm_loss": 0.138,
      "step": 12928,
      "vm_loss": 0.1421
    },
    {
      "epoch": 0.9485220613186349,
      "lm_loss": 0.631,
      "step": 12928,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.9485220613186349,
      "lm_loss": 0.0292,
      "step": 12928,
      "vm_loss": 0.1285
    },
    {
      "epoch": 0.9485954309087741,
      "grad_norm": 1.5066915973676975,
      "learning_rate": 6.901940758427206e-08,
      "loss": 0.7393,
      "step": 12929
    },
    {
      "epoch": 0.9486688004989132,
      "grad_norm": 1.6233788888499503,
      "learning_rate": 6.882280345224313e-08,
      "loss": 0.8026,
      "step": 12930
    },
    {
      "epoch": 0.9487421700890524,
      "grad_norm": 1.5553190360784162,
      "learning_rate": 6.862647779600218e-08,
      "loss": 0.729,
      "step": 12931
    },
    {
      "epoch": 0.9488155396791915,
      "grad_norm": 1.5239524362262686,
      "learning_rate": 6.843043062663701e-08,
      "loss": 0.8007,
      "step": 12932
    },
    {
      "epoch": 0.9488889092693306,
      "grad_norm": 1.5820590599261102,
      "learning_rate": 6.823466195521767e-08,
      "loss": 0.6343,
      "step": 12933
    },
    {
      "epoch": 0.9489622788594697,
      "grad_norm": 1.5085276851768672,
      "learning_rate": 6.803917179280085e-08,
      "loss": 0.7374,
      "step": 12934
    },
    {
      "epoch": 0.9490356484496089,
      "grad_norm": 1.587861693278275,
      "learning_rate": 6.784396015042493e-08,
      "loss": 0.8264,
      "step": 12935
    },
    {
      "epoch": 0.949109018039748,
      "grad_norm": 1.558398251985407,
      "learning_rate": 6.764902703911502e-08,
      "loss": 0.7301,
      "step": 12936
    },
    {
      "epoch": 0.949109018039748,
      "lm_loss": 0.2438,
      "step": 12936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.949109018039748,
      "lm_loss": 0.5291,
      "step": 12936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.949109018039748,
      "lm_loss": 0.3445,
      "step": 12936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.949109018039748,
      "lm_loss": 0.4951,
      "step": 12936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.949109018039748,
      "lm_loss": 0.364,
      "step": 12936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.949109018039748,
      "lm_loss": 0.3836,
      "step": 12936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.949109018039748,
      "lm_loss": 0.4835,
      "step": 12936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.949109018039748,
      "lm_loss": 0.509,
      "step": 12936,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.949182387629887,
      "grad_norm": 0.538971431062635,
      "learning_rate": 6.745437246988007e-08,
      "loss": 0.467,
      "step": 12937
    },
    {
      "epoch": 0.9492557572200262,
      "grad_norm": 1.6503514321353516,
      "learning_rate": 6.725999645371129e-08,
      "loss": 0.651,
      "step": 12938
    },
    {
      "epoch": 0.9493291268101653,
      "grad_norm": 1.4937716747852254,
      "learning_rate": 6.706589900158655e-08,
      "loss": 0.8587,
      "step": 12939
    },
    {
      "epoch": 0.9494024964003045,
      "grad_norm": 1.4151415541003425,
      "learning_rate": 6.687208012446656e-08,
      "loss": 0.7253,
      "step": 12940
    },
    {
      "epoch": 0.9494758659904436,
      "grad_norm": 0.5509044059933323,
      "learning_rate": 6.667853983329698e-08,
      "loss": 0.4572,
      "step": 12941
    },
    {
      "epoch": 0.9495492355805827,
      "grad_norm": 1.6966914277957177,
      "learning_rate": 6.64852781390074e-08,
      "loss": 0.6353,
      "step": 12942
    },
    {
      "epoch": 0.9496226051707218,
      "grad_norm": 0.5625072821423094,
      "learning_rate": 6.62922950525119e-08,
      "loss": 0.476,
      "step": 12943
    },
    {
      "epoch": 0.949695974760861,
      "grad_norm": 1.6989920770402063,
      "learning_rate": 6.609959058470839e-08,
      "loss": 0.8147,
      "step": 12944
    },
    {
      "epoch": 0.949695974760861,
      "lm_loss": 2.1604,
      "step": 12944,
      "vm_loss": 0.0738
    },
    {
      "epoch": 0.949695974760861,
      "lm_loss": 0.96,
      "step": 12944,
      "vm_loss": 0.072
    },
    {
      "epoch": 0.949695974760861,
      "lm_loss": 0.2999,
      "step": 12944,
      "vm_loss": 0.0752
    },
    {
      "epoch": 0.949695974760861,
      "lm_loss": 0.9806,
      "step": 12944,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.949695974760861,
      "lm_loss": 0.9583,
      "step": 12944,
      "vm_loss": 0.22
    },
    {
      "epoch": 0.949695974760861,
      "lm_loss": 0.5101,
      "step": 12944,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.949695974760861,
      "lm_loss": 0.4186,
      "step": 12944,
      "vm_loss": 0.1254
    },
    {
      "epoch": 0.949695974760861,
      "lm_loss": 0.5581,
      "step": 12944,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.9497693443510001,
      "grad_norm": 1.5344669620069757,
      "learning_rate": 6.590716474647985e-08,
      "loss": 0.6701,
      "step": 12945
    },
    {
      "epoch": 0.9498427139411393,
      "grad_norm": 0.5524303470798443,
      "learning_rate": 6.571501754869259e-08,
      "loss": 0.4718,
      "step": 12946
    },
    {
      "epoch": 0.9499160835312783,
      "grad_norm": 0.5474848025823921,
      "learning_rate": 6.552314900219736e-08,
      "loss": 0.4564,
      "step": 12947
    },
    {
      "epoch": 0.9499894531214175,
      "grad_norm": 1.474974513059776,
      "learning_rate": 6.533155911783051e-08,
      "loss": 0.6998,
      "step": 12948
    },
    {
      "epoch": 0.9500628227115566,
      "grad_norm": 1.6489858157387174,
      "learning_rate": 6.514024790641116e-08,
      "loss": 0.7376,
      "step": 12949
    },
    {
      "epoch": 0.9501361923016958,
      "grad_norm": 0.5537469213508824,
      "learning_rate": 6.494921537874288e-08,
      "loss": 0.4739,
      "step": 12950
    },
    {
      "epoch": 0.9502095618918349,
      "grad_norm": 1.577541777721199,
      "learning_rate": 6.475846154561371e-08,
      "loss": 0.6723,
      "step": 12951
    },
    {
      "epoch": 0.950282931481974,
      "grad_norm": 1.4573235627989725,
      "learning_rate": 6.456798641779616e-08,
      "loss": 0.7281,
      "step": 12952
    },
    {
      "epoch": 0.950282931481974,
      "lm_loss": 1.3916,
      "step": 12952,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.950282931481974,
      "lm_loss": 1.0826,
      "step": 12952,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.950282931481974,
      "lm_loss": 0.2211,
      "step": 12952,
      "vm_loss": 0.1327
    },
    {
      "epoch": 0.950282931481974,
      "lm_loss": 0.7086,
      "step": 12952,
      "vm_loss": 0.1233
    },
    {
      "epoch": 0.950282931481974,
      "lm_loss": 0.7241,
      "step": 12952,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.950282931481974,
      "lm_loss": 0.9741,
      "step": 12952,
      "vm_loss": 0.1071
    },
    {
      "epoch": 0.950282931481974,
      "lm_loss": 1.2302,
      "step": 12952,
      "vm_loss": 0.1675
    },
    {
      "epoch": 0.950282931481974,
      "lm_loss": 0.5593,
      "step": 12952,
      "vm_loss": 0.0791
    },
    {
      "epoch": 0.9503563010721131,
      "grad_norm": 1.5282568034577686,
      "learning_rate": 6.437779000604716e-08,
      "loss": 0.7158,
      "step": 12953
    },
    {
      "epoch": 0.9504296706622523,
      "grad_norm": 0.5363671044369918,
      "learning_rate": 6.418787232110702e-08,
      "loss": 0.4486,
      "step": 12954
    },
    {
      "epoch": 0.9505030402523914,
      "grad_norm": 1.436773257684224,
      "learning_rate": 6.399823337370104e-08,
      "loss": 0.8043,
      "step": 12955
    },
    {
      "epoch": 0.9505764098425306,
      "grad_norm": 1.56176010213122,
      "learning_rate": 6.380887317453843e-08,
      "loss": 0.7718,
      "step": 12956
    },
    {
      "epoch": 0.9506497794326696,
      "grad_norm": 0.5446690514125413,
      "learning_rate": 6.361979173431288e-08,
      "loss": 0.4749,
      "step": 12957
    },
    {
      "epoch": 0.9507231490228087,
      "grad_norm": 0.5727762808965559,
      "learning_rate": 6.343098906370248e-08,
      "loss": 0.4891,
      "step": 12958
    },
    {
      "epoch": 0.9507965186129479,
      "grad_norm": 0.5290958448025558,
      "learning_rate": 6.324246517336985e-08,
      "loss": 0.4735,
      "step": 12959
    },
    {
      "epoch": 0.950869888203087,
      "grad_norm": 0.5621892650781241,
      "learning_rate": 6.305422007396034e-08,
      "loss": 0.4885,
      "step": 12960
    },
    {
      "epoch": 0.950869888203087,
      "lm_loss": 0.2568,
      "step": 12960,
      "vm_loss": 0.0569
    },
    {
      "epoch": 0.950869888203087,
      "lm_loss": 0.5538,
      "step": 12960,
      "vm_loss": 0.1577
    },
    {
      "epoch": 0.950869888203087,
      "lm_loss": 0.4806,
      "step": 12960,
      "vm_loss": 0.1031
    },
    {
      "epoch": 0.950869888203087,
      "lm_loss": 0.3536,
      "step": 12960,
      "vm_loss": 0.1129
    },
    {
      "epoch": 0.950869888203087,
      "lm_loss": 0.8529,
      "step": 12960,
      "vm_loss": 0.0691
    },
    {
      "epoch": 0.950869888203087,
      "lm_loss": 0.479,
      "step": 12960,
      "vm_loss": 0.1018
    },
    {
      "epoch": 0.950869888203087,
      "lm_loss": 0.6751,
      "step": 12960,
      "vm_loss": 0.2015
    },
    {
      "epoch": 0.950869888203087,
      "lm_loss": 0.3835,
      "step": 12960,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.9509432577932262,
      "grad_norm": 1.5490122069584709,
      "learning_rate": 6.286625377610545e-08,
      "loss": 0.7509,
      "step": 12961
    },
    {
      "epoch": 0.9510166273833652,
      "grad_norm": 1.5265894970882121,
      "learning_rate": 6.267856629041946e-08,
      "loss": 0.6721,
      "step": 12962
    },
    {
      "epoch": 0.9510899969735044,
      "grad_norm": 1.6822774412926527,
      "learning_rate": 6.249115762750224e-08,
      "loss": 0.7708,
      "step": 12963
    },
    {
      "epoch": 0.9511633665636435,
      "grad_norm": 1.5917081449717028,
      "learning_rate": 6.230402779793698e-08,
      "loss": 0.722,
      "step": 12964
    },
    {
      "epoch": 0.9512367361537827,
      "grad_norm": 0.5432790169018132,
      "learning_rate": 6.211717681229135e-08,
      "loss": 0.4638,
      "step": 12965
    },
    {
      "epoch": 0.9513101057439218,
      "grad_norm": 1.5067843769085416,
      "learning_rate": 6.19306046811169e-08,
      "loss": 0.6133,
      "step": 12966
    },
    {
      "epoch": 0.9513834753340609,
      "grad_norm": 0.5492879816131379,
      "learning_rate": 6.174431141495018e-08,
      "loss": 0.4807,
      "step": 12967
    },
    {
      "epoch": 0.9514568449242,
      "grad_norm": 0.537250145558027,
      "learning_rate": 6.15582970243117e-08,
      "loss": 0.4834,
      "step": 12968
    },
    {
      "epoch": 0.9514568449242,
      "lm_loss": 1.2035,
      "step": 12968,
      "vm_loss": 0.054
    },
    {
      "epoch": 0.9514568449242,
      "lm_loss": 0.6811,
      "step": 12968,
      "vm_loss": 0.109
    },
    {
      "epoch": 0.9514568449242,
      "lm_loss": 0.2575,
      "step": 12968,
      "vm_loss": 0.095
    },
    {
      "epoch": 0.9514568449242,
      "lm_loss": 0.6319,
      "step": 12968,
      "vm_loss": 0.0802
    },
    {
      "epoch": 0.9514568449242,
      "lm_loss": 0.4243,
      "step": 12968,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.9514568449242,
      "lm_loss": 0.0071,
      "step": 12968,
      "vm_loss": 0.1147
    },
    {
      "epoch": 0.9514568449242,
      "lm_loss": 0.452,
      "step": 12968,
      "vm_loss": 0.0609
    },
    {
      "epoch": 0.9514568449242,
      "lm_loss": 0.9881,
      "step": 12968,
      "vm_loss": 0.0889
    },
    {
      "epoch": 0.9515302145143392,
      "grad_norm": 1.473481060491374,
      "learning_rate": 6.137256151970583e-08,
      "loss": 0.7077,
      "step": 12969
    },
    {
      "epoch": 0.9516035841044783,
      "grad_norm": 1.5599913569169395,
      "learning_rate": 6.118710491162195e-08,
      "loss": 0.8299,
      "step": 12970
    },
    {
      "epoch": 0.9516769536946175,
      "grad_norm": 1.4438322976838556,
      "learning_rate": 6.100192721053277e-08,
      "loss": 0.7276,
      "step": 12971
    },
    {
      "epoch": 0.9517503232847565,
      "grad_norm": 1.6209110300332308,
      "learning_rate": 6.081702842689607e-08,
      "loss": 0.7595,
      "step": 12972
    },
    {
      "epoch": 0.9518236928748957,
      "grad_norm": 1.4972978552826026,
      "learning_rate": 6.063240857115349e-08,
      "loss": 0.6828,
      "step": 12973
    },
    {
      "epoch": 0.9518970624650348,
      "grad_norm": 0.5508619017883004,
      "learning_rate": 6.044806765373057e-08,
      "loss": 0.4962,
      "step": 12974
    },
    {
      "epoch": 0.951970432055174,
      "grad_norm": 1.4050497461365925,
      "learning_rate": 6.026400568503843e-08,
      "loss": 0.7113,
      "step": 12975
    },
    {
      "epoch": 0.9520438016453131,
      "grad_norm": 1.6606682606779455,
      "learning_rate": 6.008022267547098e-08,
      "loss": 0.6438,
      "step": 12976
    },
    {
      "epoch": 0.9520438016453131,
      "lm_loss": 1.9276,
      "step": 12976,
      "vm_loss": 0.08
    },
    {
      "epoch": 0.9520438016453131,
      "lm_loss": 1.2922,
      "step": 12976,
      "vm_loss": 0.0662
    },
    {
      "epoch": 0.9520438016453131,
      "lm_loss": 0.0771,
      "step": 12976,
      "vm_loss": 0.1641
    },
    {
      "epoch": 0.9520438016453131,
      "lm_loss": 1.0902,
      "step": 12976,
      "vm_loss": 0.1015
    },
    {
      "epoch": 0.9520438016453131,
      "lm_loss": 0.3553,
      "step": 12976,
      "vm_loss": 0.0909
    },
    {
      "epoch": 0.9520438016453131,
      "lm_loss": 0.5747,
      "step": 12976,
      "vm_loss": 0.0991
    },
    {
      "epoch": 0.9520438016453131,
      "lm_loss": 1.3344,
      "step": 12976,
      "vm_loss": 0.0841
    },
    {
      "epoch": 0.9520438016453131,
      "lm_loss": 0.4192,
      "step": 12976,
      "vm_loss": 0.1949
    },
    {
      "epoch": 0.9521171712354521,
      "grad_norm": 1.3594068205999001,
      "learning_rate": 5.989671863540658e-08,
      "loss": 0.7173,
      "step": 12977
    },
    {
      "epoch": 0.9521905408255913,
      "grad_norm": 1.4572741565986744,
      "learning_rate": 5.971349357520806e-08,
      "loss": 0.6614,
      "step": 12978
    },
    {
      "epoch": 0.9522639104157304,
      "grad_norm": 1.5834719584927401,
      "learning_rate": 5.95305475052238e-08,
      "loss": 0.6762,
      "step": 12979
    },
    {
      "epoch": 0.9523372800058696,
      "grad_norm": 1.4584503997371712,
      "learning_rate": 5.934788043578388e-08,
      "loss": 0.6942,
      "step": 12980
    },
    {
      "epoch": 0.9524106495960087,
      "grad_norm": 1.7047612014057025,
      "learning_rate": 5.916549237720448e-08,
      "loss": 0.6543,
      "step": 12981
    },
    {
      "epoch": 0.9524840191861478,
      "grad_norm": 1.5403109704962115,
      "learning_rate": 5.8983383339785704e-08,
      "loss": 0.7223,
      "step": 12982
    },
    {
      "epoch": 0.9525573887762869,
      "grad_norm": 1.5570423299634273,
      "learning_rate": 5.880155333381155e-08,
      "loss": 0.7585,
      "step": 12983
    },
    {
      "epoch": 0.9526307583664261,
      "grad_norm": 1.4114161115220265,
      "learning_rate": 5.86200023695499e-08,
      "loss": 0.6654,
      "step": 12984
    },
    {
      "epoch": 0.9526307583664261,
      "lm_loss": 1.3388,
      "step": 12984,
      "vm_loss": 0.0935
    },
    {
      "epoch": 0.9526307583664261,
      "lm_loss": 0.4891,
      "step": 12984,
      "vm_loss": 0.1082
    },
    {
      "epoch": 0.9526307583664261,
      "lm_loss": 0.1291,
      "step": 12984,
      "vm_loss": 0.1685
    },
    {
      "epoch": 0.9526307583664261,
      "lm_loss": 0.7289,
      "step": 12984,
      "vm_loss": 0.1741
    },
    {
      "epoch": 0.9526307583664261,
      "lm_loss": 0.6708,
      "step": 12984,
      "vm_loss": 0.1018
    },
    {
      "epoch": 0.9526307583664261,
      "lm_loss": 0.1381,
      "step": 12984,
      "vm_loss": 0.1011
    },
    {
      "epoch": 0.9526307583664261,
      "lm_loss": 1.056,
      "step": 12984,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.9526307583664261,
      "lm_loss": 0.9537,
      "step": 12984,
      "vm_loss": 0.1651
    },
    {
      "epoch": 0.9527041279565652,
      "grad_norm": 1.4544517737206053,
      "learning_rate": 5.843873045725423e-08,
      "loss": 0.6555,
      "step": 12985
    },
    {
      "epoch": 0.9527774975467044,
      "grad_norm": 0.5604432878300675,
      "learning_rate": 5.8257737607160804e-08,
      "loss": 0.4674,
      "step": 12986
    },
    {
      "epoch": 0.9528508671368434,
      "grad_norm": 0.5186873135355311,
      "learning_rate": 5.807702382949088e-08,
      "loss": 0.4449,
      "step": 12987
    },
    {
      "epoch": 0.9529242367269826,
      "grad_norm": 1.659622449298387,
      "learning_rate": 5.789658913445018e-08,
      "loss": 0.7042,
      "step": 12988
    },
    {
      "epoch": 0.9529976063171217,
      "grad_norm": 0.5344354173747432,
      "learning_rate": 5.771643353222778e-08,
      "loss": 0.4971,
      "step": 12989
    },
    {
      "epoch": 0.9530709759072609,
      "grad_norm": 1.6231757623307022,
      "learning_rate": 5.7536557032997765e-08,
      "loss": 0.7624,
      "step": 12990
    },
    {
      "epoch": 0.9531443454974,
      "grad_norm": 1.5531509057337576,
      "learning_rate": 5.735695964691812e-08,
      "loss": 0.7349,
      "step": 12991
    },
    {
      "epoch": 0.953217715087539,
      "grad_norm": 1.6101660066107852,
      "learning_rate": 5.717764138413129e-08,
      "loss": 0.7636,
      "step": 12992
    },
    {
      "epoch": 0.953217715087539,
      "lm_loss": 1.3561,
      "step": 12992,
      "vm_loss": 0.1141
    },
    {
      "epoch": 0.953217715087539,
      "lm_loss": 1.1624,
      "step": 12992,
      "vm_loss": 0.0607
    },
    {
      "epoch": 0.953217715087539,
      "lm_loss": 0.5172,
      "step": 12992,
      "vm_loss": 0.1122
    },
    {
      "epoch": 0.953217715087539,
      "lm_loss": 0.4325,
      "step": 12992,
      "vm_loss": 0.1046
    },
    {
      "epoch": 0.953217715087539,
      "lm_loss": 0.4247,
      "step": 12992,
      "vm_loss": 0.154
    },
    {
      "epoch": 0.953217715087539,
      "lm_loss": 0.2903,
      "step": 12992,
      "vm_loss": 0.1874
    },
    {
      "epoch": 0.953217715087539,
      "lm_loss": 0.2361,
      "step": 12992,
      "vm_loss": 0.1062
    },
    {
      "epoch": 0.953217715087539,
      "lm_loss": 0.4561,
      "step": 12992,
      "vm_loss": 0.1587
    },
    {
      "epoch": 0.9532910846776782,
      "grad_norm": 1.4540423695029367,
      "learning_rate": 5.699860225476417e-08,
      "loss": 0.7606,
      "step": 12993
    },
    {
      "epoch": 0.9533644542678174,
      "grad_norm": 1.4325892353980867,
      "learning_rate": 5.6819842268926454e-08,
      "loss": 0.7838,
      "step": 12994
    },
    {
      "epoch": 0.9534378238579565,
      "grad_norm": 0.5395400537756317,
      "learning_rate": 5.664136143671395e-08,
      "loss": 0.4609,
      "step": 12995
    },
    {
      "epoch": 0.9535111934480957,
      "grad_norm": 1.8165031805958045,
      "learning_rate": 5.6463159768206376e-08,
      "loss": 0.7312,
      "step": 12996
    },
    {
      "epoch": 0.9535845630382347,
      "grad_norm": 1.731039436810314,
      "learning_rate": 5.6285237273466245e-08,
      "loss": 0.738,
      "step": 12997
    },
    {
      "epoch": 0.9536579326283738,
      "grad_norm": 1.4300285640454318,
      "learning_rate": 5.610759396254162e-08,
      "loss": 0.8127,
      "step": 12998
    },
    {
      "epoch": 0.953731302218513,
      "grad_norm": 1.6017629316736495,
      "learning_rate": 5.5930229845464476e-08,
      "loss": 0.8066,
      "step": 12999
    },
    {
      "epoch": 0.9538046718086521,
      "grad_norm": 1.5562714107813438,
      "learning_rate": 5.5753144932250705e-08,
      "loss": 0.6783,
      "step": 13000
    },
    {
      "epoch": 0.9538046718086521,
      "lm_loss": 1.3378,
      "step": 13000,
      "vm_loss": 0.0597
    },
    {
      "epoch": 0.9538046718086521,
      "lm_loss": 0.7513,
      "step": 13000,
      "vm_loss": 0.1614
    },
    {
      "epoch": 0.9538046718086521,
      "lm_loss": 0.431,
      "step": 13000,
      "vm_loss": 0.1443
    },
    {
      "epoch": 0.9538046718086521,
      "lm_loss": 0.7308,
      "step": 13000,
      "vm_loss": 0.2263
    },
    {
      "epoch": 0.9538046718086521,
      "lm_loss": 0.313,
      "step": 13000,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.9538046718086521,
      "lm_loss": 0.4968,
      "step": 13000,
      "vm_loss": 0.0956
    },
    {
      "epoch": 0.9538046718086521,
      "lm_loss": 0.1027,
      "step": 13000,
      "vm_loss": 0.1156
    },
    {
      "epoch": 0.9538046718086521,
      "lm_loss": 0.2361,
      "step": 13000,
      "vm_loss": 0.1565
    },
    {
      "epoch": 0.9538780413987913,
      "grad_norm": 1.6541596566075334,
      "learning_rate": 5.5576339232901734e-08,
      "loss": 0.7053,
      "step": 13001
    },
    {
      "epoch": 0.9539514109889303,
      "grad_norm": 1.4221181505089246,
      "learning_rate": 5.539981275740125e-08,
      "loss": 0.6734,
      "step": 13002
    },
    {
      "epoch": 0.9540247805790695,
      "grad_norm": 1.4803868014281116,
      "learning_rate": 5.52235655157185e-08,
      "loss": 0.7349,
      "step": 13003
    },
    {
      "epoch": 0.9540981501692086,
      "grad_norm": 1.4933399087378838,
      "learning_rate": 5.504759751780608e-08,
      "loss": 0.7244,
      "step": 13004
    },
    {
      "epoch": 0.9541715197593478,
      "grad_norm": 1.4212874861071283,
      "learning_rate": 5.487190877360216e-08,
      "loss": 0.6779,
      "step": 13005
    },
    {
      "epoch": 0.9542448893494869,
      "grad_norm": 1.5366424389528068,
      "learning_rate": 5.469649929302823e-08,
      "loss": 0.6824,
      "step": 13006
    },
    {
      "epoch": 0.954318258939626,
      "grad_norm": 1.5654342731484054,
      "learning_rate": 5.452136908598915e-08,
      "loss": 0.6429,
      "step": 13007
    },
    {
      "epoch": 0.9543916285297651,
      "grad_norm": 1.6457291881591625,
      "learning_rate": 5.434651816237646e-08,
      "loss": 0.7313,
      "step": 13008
    },
    {
      "epoch": 0.9543916285297651,
      "lm_loss": 1.1978,
      "step": 13008,
      "vm_loss": 0.0393
    },
    {
      "epoch": 0.9543916285297651,
      "lm_loss": 0.037,
      "step": 13008,
      "vm_loss": 0.0316
    },
    {
      "epoch": 0.9543916285297651,
      "lm_loss": 0.1454,
      "step": 13008,
      "vm_loss": 0.0464
    },
    {
      "epoch": 0.9543916285297651,
      "lm_loss": 0.5908,
      "step": 13008,
      "vm_loss": 0.1353
    },
    {
      "epoch": 0.9543916285297651,
      "lm_loss": 0.1537,
      "step": 13008,
      "vm_loss": 0.2701
    },
    {
      "epoch": 0.9543916285297651,
      "lm_loss": 0.6656,
      "step": 13008,
      "vm_loss": 0.1876
    },
    {
      "epoch": 0.9543916285297651,
      "lm_loss": 0.3653,
      "step": 13008,
      "vm_loss": 0.0864
    },
    {
      "epoch": 0.9543916285297651,
      "lm_loss": 0.6714,
      "step": 13008,
      "vm_loss": 0.0985
    },
    {
      "epoch": 0.9544649981199043,
      "grad_norm": 1.6193957914276054,
      "learning_rate": 5.417194653206337e-08,
      "loss": 0.6786,
      "step": 13009
    },
    {
      "epoch": 0.9545383677100434,
      "grad_norm": 1.3881990731994072,
      "learning_rate": 5.39976542049081e-08,
      "loss": 0.6232,
      "step": 13010
    },
    {
      "epoch": 0.9546117373001826,
      "grad_norm": 0.5180706799628042,
      "learning_rate": 5.382364119075389e-08,
      "loss": 0.4705,
      "step": 13011
    },
    {
      "epoch": 0.9546851068903216,
      "grad_norm": 1.564571232402477,
      "learning_rate": 5.364990749942789e-08,
      "loss": 0.7168,
      "step": 13012
    },
    {
      "epoch": 0.9547584764804607,
      "grad_norm": 0.5397895047888972,
      "learning_rate": 5.3476453140741126e-08,
      "loss": 0.4733,
      "step": 13013
    },
    {
      "epoch": 0.9548318460705999,
      "grad_norm": 1.5901581659522561,
      "learning_rate": 5.33032781244891e-08,
      "loss": 0.7484,
      "step": 13014
    },
    {
      "epoch": 0.954905215660739,
      "grad_norm": 1.5953106176846363,
      "learning_rate": 5.3130382460451216e-08,
      "loss": 0.6981,
      "step": 13015
    },
    {
      "epoch": 0.9549785852508781,
      "grad_norm": 1.3704353699083702,
      "learning_rate": 5.295776615839077e-08,
      "loss": 0.6654,
      "step": 13016
    },
    {
      "epoch": 0.9549785852508781,
      "lm_loss": 1.354,
      "step": 13016,
      "vm_loss": 0.1164
    },
    {
      "epoch": 0.9549785852508781,
      "lm_loss": 0.5721,
      "step": 13016,
      "vm_loss": 0.1077
    },
    {
      "epoch": 0.9549785852508781,
      "lm_loss": 1.5505,
      "step": 13016,
      "vm_loss": 0.1353
    },
    {
      "epoch": 0.9549785852508781,
      "lm_loss": 0.9878,
      "step": 13016,
      "vm_loss": 0.1807
    },
    {
      "epoch": 0.9549785852508781,
      "lm_loss": 0.5903,
      "step": 13016,
      "vm_loss": 0.1165
    },
    {
      "epoch": 0.9549785852508781,
      "lm_loss": 0.0784,
      "step": 13016,
      "vm_loss": 0.1174
    },
    {
      "epoch": 0.9549785852508781,
      "lm_loss": 0.5235,
      "step": 13016,
      "vm_loss": 0.1716
    },
    {
      "epoch": 0.9549785852508781,
      "lm_loss": 0.0773,
      "step": 13016,
      "vm_loss": 0.0646
    },
    {
      "epoch": 0.9550519548410172,
      "grad_norm": 1.4560891870114474,
      "learning_rate": 5.278542922805718e-08,
      "loss": 0.8123,
      "step": 13017
    },
    {
      "epoch": 0.9551253244311564,
      "grad_norm": 1.463626262613512,
      "learning_rate": 5.261337167918157e-08,
      "loss": 0.7459,
      "step": 13018
    },
    {
      "epoch": 0.9551986940212955,
      "grad_norm": 0.56601976583396,
      "learning_rate": 5.244159352148059e-08,
      "loss": 0.4779,
      "step": 13019
    },
    {
      "epoch": 0.9552720636114347,
      "grad_norm": 1.4383524685274076,
      "learning_rate": 5.2270094764655944e-08,
      "loss": 0.6996,
      "step": 13020
    },
    {
      "epoch": 0.9553454332015737,
      "grad_norm": 1.4114301127823865,
      "learning_rate": 5.2098875418390983e-08,
      "loss": 0.6261,
      "step": 13021
    },
    {
      "epoch": 0.9554188027917129,
      "grad_norm": 0.5439870170270524,
      "learning_rate": 5.1927935492356306e-08,
      "loss": 0.482,
      "step": 13022
    },
    {
      "epoch": 0.955492172381852,
      "grad_norm": 1.3580666213037846,
      "learning_rate": 5.175727499620475e-08,
      "loss": 0.7537,
      "step": 13023
    },
    {
      "epoch": 0.9555655419719912,
      "grad_norm": 1.6186891895244977,
      "learning_rate": 5.158689393957417e-08,
      "loss": 0.7364,
      "step": 13024
    },
    {
      "epoch": 0.9555655419719912,
      "lm_loss": 1.0503,
      "step": 13024,
      "vm_loss": 0.0294
    },
    {
      "epoch": 0.9555655419719912,
      "lm_loss": 1.7887,
      "step": 13024,
      "vm_loss": 0.0605
    },
    {
      "epoch": 0.9555655419719912,
      "lm_loss": 0.9015,
      "step": 13024,
      "vm_loss": 0.1683
    },
    {
      "epoch": 0.9555655419719912,
      "lm_loss": 0.7369,
      "step": 13024,
      "vm_loss": 0.1373
    },
    {
      "epoch": 0.9555655419719912,
      "lm_loss": 0.198,
      "step": 13024,
      "vm_loss": 0.0648
    },
    {
      "epoch": 0.9555655419719912,
      "lm_loss": 0.8701,
      "step": 13024,
      "vm_loss": 0.1949
    },
    {
      "epoch": 0.9555655419719912,
      "lm_loss": 0.0978,
      "step": 13024,
      "vm_loss": 0.0903
    },
    {
      "epoch": 0.9555655419719912,
      "lm_loss": 0.4304,
      "step": 13024,
      "vm_loss": 0.0788
    },
    {
      "epoch": 0.9556389115621303,
      "grad_norm": 1.51349006370617,
      "learning_rate": 5.14167923320863e-08,
      "loss": 0.7811,
      "step": 13025
    },
    {
      "epoch": 0.9557122811522694,
      "grad_norm": 0.5483534461525224,
      "learning_rate": 5.124697018334679e-08,
      "loss": 0.4846,
      "step": 13026
    },
    {
      "epoch": 0.9557856507424085,
      "grad_norm": 1.5422476489490624,
      "learning_rate": 5.107742750294631e-08,
      "loss": 0.733,
      "step": 13027
    },
    {
      "epoch": 0.9558590203325477,
      "grad_norm": 1.4940421480867971,
      "learning_rate": 5.0908164300459416e-08,
      "loss": 0.7735,
      "step": 13028
    },
    {
      "epoch": 0.9559323899226868,
      "grad_norm": 1.4378777128158218,
      "learning_rate": 5.073918058544458e-08,
      "loss": 0.8562,
      "step": 13029
    },
    {
      "epoch": 0.956005759512826,
      "grad_norm": 0.5474029266731313,
      "learning_rate": 5.0570476367445277e-08,
      "loss": 0.4641,
      "step": 13030
    },
    {
      "epoch": 0.956079129102965,
      "grad_norm": 1.568976119616098,
      "learning_rate": 5.040205165598777e-08,
      "loss": 0.7495,
      "step": 13031
    },
    {
      "epoch": 0.9561524986931041,
      "grad_norm": 1.5642665132733509,
      "learning_rate": 5.02339064605839e-08,
      "loss": 0.7563,
      "step": 13032
    },
    {
      "epoch": 0.9561524986931041,
      "lm_loss": 1.1023,
      "step": 13032,
      "vm_loss": 0.0636
    },
    {
      "epoch": 0.9561524986931041,
      "lm_loss": 0.2877,
      "step": 13032,
      "vm_loss": 0.0548
    },
    {
      "epoch": 0.9561524986931041,
      "lm_loss": 0.3724,
      "step": 13032,
      "vm_loss": 0.1215
    },
    {
      "epoch": 0.9561524986931041,
      "lm_loss": 0.1024,
      "step": 13032,
      "vm_loss": 0.0763
    },
    {
      "epoch": 0.9561524986931041,
      "lm_loss": 0.4969,
      "step": 13032,
      "vm_loss": 0.128
    },
    {
      "epoch": 0.9561524986931041,
      "lm_loss": 0.6422,
      "step": 13032,
      "vm_loss": 0.1981
    },
    {
      "epoch": 0.9561524986931041,
      "lm_loss": 0.5496,
      "step": 13032,
      "vm_loss": 0.0632
    },
    {
      "epoch": 0.9561524986931041,
      "lm_loss": 0.0833,
      "step": 13032,
      "vm_loss": 0.1565
    },
    {
      "epoch": 0.9562258682832433,
      "grad_norm": 1.4474230558366166,
      "learning_rate": 5.006604079072941e-08,
      "loss": 0.6658,
      "step": 13033
    },
    {
      "epoch": 0.9562992378733824,
      "grad_norm": 1.448809199636102,
      "learning_rate": 4.989845465590393e-08,
      "loss": 0.6951,
      "step": 13034
    },
    {
      "epoch": 0.9563726074635216,
      "grad_norm": 0.5431170078390446,
      "learning_rate": 4.9731148065571e-08,
      "loss": 0.4761,
      "step": 13035
    },
    {
      "epoch": 0.9564459770536606,
      "grad_norm": 1.552507862110845,
      "learning_rate": 4.956412102917974e-08,
      "loss": 0.7365,
      "step": 13036
    },
    {
      "epoch": 0.9565193466437998,
      "grad_norm": 1.5598849388062652,
      "learning_rate": 4.939737355616203e-08,
      "loss": 0.6825,
      "step": 13037
    },
    {
      "epoch": 0.9565927162339389,
      "grad_norm": 1.2298906312012932,
      "learning_rate": 4.9230905655934246e-08,
      "loss": 0.6588,
      "step": 13038
    },
    {
      "epoch": 0.9566660858240781,
      "grad_norm": 1.4948920387679272,
      "learning_rate": 4.906471733789775e-08,
      "loss": 0.6141,
      "step": 13039
    },
    {
      "epoch": 0.9567394554142172,
      "grad_norm": 1.5376522365269563,
      "learning_rate": 4.889880861143781e-08,
      "loss": 0.8063,
      "step": 13040
    },
    {
      "epoch": 0.9567394554142172,
      "lm_loss": 1.2446,
      "step": 13040,
      "vm_loss": 0.0358
    },
    {
      "epoch": 0.9567394554142172,
      "lm_loss": 0.8141,
      "step": 13040,
      "vm_loss": 0.2149
    },
    {
      "epoch": 0.9567394554142172,
      "lm_loss": 0.0135,
      "step": 13040,
      "vm_loss": 0.0904
    },
    {
      "epoch": 0.9567394554142172,
      "lm_loss": 0.0111,
      "step": 13040,
      "vm_loss": 0.1654
    },
    {
      "epoch": 0.9567394554142172,
      "lm_loss": 0.1852,
      "step": 13040,
      "vm_loss": 0.086
    },
    {
      "epoch": 0.9567394554142172,
      "lm_loss": 0.3269,
      "step": 13040,
      "vm_loss": 0.0859
    },
    {
      "epoch": 0.9567394554142172,
      "lm_loss": 0.5269,
      "step": 13040,
      "vm_loss": 0.1098
    },
    {
      "epoch": 0.9567394554142172,
      "lm_loss": 0.4399,
      "step": 13040,
      "vm_loss": 0.157
    },
    {
      "epoch": 0.9568128250043563,
      "grad_norm": 1.7479298447146654,
      "learning_rate": 4.8733179485923047e-08,
      "loss": 0.7631,
      "step": 13041
    },
    {
      "epoch": 0.9568861945944954,
      "grad_norm": 1.505968264138789,
      "learning_rate": 4.8567829970707636e-08,
      "loss": 0.7287,
      "step": 13042
    },
    {
      "epoch": 0.9569595641846346,
      "grad_norm": 1.5609763227526454,
      "learning_rate": 4.840276007512801e-08,
      "loss": 0.6964,
      "step": 13043
    },
    {
      "epoch": 0.9570329337747737,
      "grad_norm": 1.5404673131900368,
      "learning_rate": 4.823796980850781e-08,
      "loss": 0.7626,
      "step": 13044
    },
    {
      "epoch": 0.9571063033649129,
      "grad_norm": 1.513555944179692,
      "learning_rate": 4.8073459180151825e-08,
      "loss": 0.7095,
      "step": 13045
    },
    {
      "epoch": 0.9571796729550519,
      "grad_norm": 1.5602405338636613,
      "learning_rate": 4.7909228199350953e-08,
      "loss": 0.7542,
      "step": 13046
    },
    {
      "epoch": 0.9572530425451911,
      "grad_norm": 1.607763969226244,
      "learning_rate": 4.774527687538e-08,
      "loss": 0.6547,
      "step": 13047
    },
    {
      "epoch": 0.9573264121353302,
      "grad_norm": 1.4086222905856702,
      "learning_rate": 4.758160521749655e-08,
      "loss": 0.6925,
      "step": 13048
    },
    {
      "epoch": 0.9573264121353302,
      "lm_loss": 1.3814,
      "step": 13048,
      "vm_loss": 0.1208
    },
    {
      "epoch": 0.9573264121353302,
      "lm_loss": 0.4944,
      "step": 13048,
      "vm_loss": 0.1113
    },
    {
      "epoch": 0.9573264121353302,
      "lm_loss": 0.7796,
      "step": 13048,
      "vm_loss": 0.0986
    },
    {
      "epoch": 0.9573264121353302,
      "lm_loss": 0.4735,
      "step": 13048,
      "vm_loss": 0.133
    },
    {
      "epoch": 0.9573264121353302,
      "lm_loss": 0.5608,
      "step": 13048,
      "vm_loss": 0.0928
    },
    {
      "epoch": 0.9573264121353302,
      "lm_loss": 0.0866,
      "step": 13048,
      "vm_loss": 0.0866
    },
    {
      "epoch": 0.9573264121353302,
      "lm_loss": 0.6611,
      "step": 13048,
      "vm_loss": 0.1451
    },
    {
      "epoch": 0.9573264121353302,
      "lm_loss": 0.1382,
      "step": 13048,
      "vm_loss": 0.0792
    },
    {
      "epoch": 0.9573997817254694,
      "grad_norm": 1.4067295000635405,
      "learning_rate": 4.741821323494489e-08,
      "loss": 0.6817,
      "step": 13049
    },
    {
      "epoch": 0.9574731513156085,
      "grad_norm": 0.5382504637202652,
      "learning_rate": 4.725510093695152e-08,
      "loss": 0.4786,
      "step": 13050
    },
    {
      "epoch": 0.9575465209057475,
      "grad_norm": 0.5631821967194435,
      "learning_rate": 4.7092268332727395e-08,
      "loss": 0.4667,
      "step": 13051
    },
    {
      "epoch": 0.9576198904958867,
      "grad_norm": 1.4578029792878127,
      "learning_rate": 4.692971543146907e-08,
      "loss": 0.7372,
      "step": 13052
    },
    {
      "epoch": 0.9576932600860258,
      "grad_norm": 1.4368092391257548,
      "learning_rate": 4.676744224235585e-08,
      "loss": 0.7888,
      "step": 13053
    },
    {
      "epoch": 0.957766629676165,
      "grad_norm": 1.535381306217707,
      "learning_rate": 4.660544877455098e-08,
      "loss": 0.6773,
      "step": 13054
    },
    {
      "epoch": 0.9578399992663041,
      "grad_norm": 0.5405235412954377,
      "learning_rate": 4.6443735037203785e-08,
      "loss": 0.4879,
      "step": 13055
    },
    {
      "epoch": 0.9579133688564432,
      "grad_norm": 1.4431216967266678,
      "learning_rate": 4.628230103944642e-08,
      "loss": 0.607,
      "step": 13056
    },
    {
      "epoch": 0.9579133688564432,
      "lm_loss": 1.4147,
      "step": 13056,
      "vm_loss": 0.1346
    },
    {
      "epoch": 0.9579133688564432,
      "lm_loss": 0.7384,
      "step": 13056,
      "vm_loss": 0.1774
    },
    {
      "epoch": 0.9579133688564432,
      "lm_loss": 0.3743,
      "step": 13056,
      "vm_loss": 0.1847
    },
    {
      "epoch": 0.9579133688564432,
      "lm_loss": 0.0225,
      "step": 13056,
      "vm_loss": 0.1237
    },
    {
      "epoch": 0.9579133688564432,
      "lm_loss": 0.2958,
      "step": 13056,
      "vm_loss": 0.0539
    },
    {
      "epoch": 0.9579133688564432,
      "lm_loss": 1.0467,
      "step": 13056,
      "vm_loss": 0.0831
    },
    {
      "epoch": 0.9579133688564432,
      "lm_loss": 0.6811,
      "step": 13056,
      "vm_loss": 0.0989
    },
    {
      "epoch": 0.9579133688564432,
      "lm_loss": 0.0628,
      "step": 13056,
      "vm_loss": 0.1581
    },
    {
      "epoch": 0.9579867384465823,
      "grad_norm": 1.4835853709146374,
      "learning_rate": 4.6121146790394925e-08,
      "loss": 0.6929,
      "step": 13057
    },
    {
      "epoch": 0.9580601080367215,
      "grad_norm": 1.6544420587554534,
      "learning_rate": 4.596027229915034e-08,
      "loss": 0.6041,
      "step": 13058
    },
    {
      "epoch": 0.9581334776268606,
      "grad_norm": 1.5142395609258155,
      "learning_rate": 4.579967757479709e-08,
      "loss": 0.708,
      "step": 13059
    },
    {
      "epoch": 0.9582068472169998,
      "grad_norm": 1.3372806747287767,
      "learning_rate": 4.563936262640567e-08,
      "loss": 0.6823,
      "step": 13060
    },
    {
      "epoch": 0.9582802168071388,
      "grad_norm": 1.4676991696647776,
      "learning_rate": 4.54793274630283e-08,
      "loss": 0.7743,
      "step": 13061
    },
    {
      "epoch": 0.958353586397278,
      "grad_norm": 1.4301806699221782,
      "learning_rate": 4.53195720937033e-08,
      "loss": 0.6939,
      "step": 13062
    },
    {
      "epoch": 0.9584269559874171,
      "grad_norm": 1.4854975754284063,
      "learning_rate": 4.516009652745179e-08,
      "loss": 0.6992,
      "step": 13063
    },
    {
      "epoch": 0.9585003255775563,
      "grad_norm": 1.6313466180624505,
      "learning_rate": 4.5000900773280453e-08,
      "loss": 0.8004,
      "step": 13064
    },
    {
      "epoch": 0.9585003255775563,
      "lm_loss": 0.3774,
      "step": 13064,
      "vm_loss": 0.0492
    },
    {
      "epoch": 0.9585003255775563,
      "lm_loss": 0.9962,
      "step": 13064,
      "vm_loss": 0.0749
    },
    {
      "epoch": 0.9585003255775563,
      "lm_loss": 1.0113,
      "step": 13064,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.9585003255775563,
      "lm_loss": 0.5146,
      "step": 13064,
      "vm_loss": 0.1602
    },
    {
      "epoch": 0.9585003255775563,
      "lm_loss": 0.5125,
      "step": 13064,
      "vm_loss": 0.0991
    },
    {
      "epoch": 0.9585003255775563,
      "lm_loss": 0.4246,
      "step": 13064,
      "vm_loss": 0.1927
    },
    {
      "epoch": 0.9585003255775563,
      "lm_loss": 0.7601,
      "step": 13064,
      "vm_loss": 0.1699
    },
    {
      "epoch": 0.9585003255775563,
      "lm_loss": 1.2768,
      "step": 13064,
      "vm_loss": 0.1358
    },
    {
      "epoch": 0.9585736951676954,
      "grad_norm": 1.5004734154333943,
      "learning_rate": 4.4841984840178764e-08,
      "loss": 0.7667,
      "step": 13065
    },
    {
      "epoch": 0.9586470647578345,
      "grad_norm": 1.8136884878462585,
      "learning_rate": 4.468334873712121e-08,
      "loss": 0.7652,
      "step": 13066
    },
    {
      "epoch": 0.9587204343479736,
      "grad_norm": 1.7877255298182868,
      "learning_rate": 4.45249924730673e-08,
      "loss": 0.7401,
      "step": 13067
    },
    {
      "epoch": 0.9587938039381128,
      "grad_norm": 1.60758516382953,
      "learning_rate": 4.436691605695875e-08,
      "loss": 0.7856,
      "step": 13068
    },
    {
      "epoch": 0.9588671735282519,
      "grad_norm": 1.4657130074731293,
      "learning_rate": 4.4209119497722883e-08,
      "loss": 0.7279,
      "step": 13069
    },
    {
      "epoch": 0.958940543118391,
      "grad_norm": 1.4999957698113495,
      "learning_rate": 4.4051602804270345e-08,
      "loss": 0.7819,
      "step": 13070
    },
    {
      "epoch": 0.9590139127085301,
      "grad_norm": 1.6714693098446898,
      "learning_rate": 4.38943659854979e-08,
      "loss": 0.7417,
      "step": 13071
    },
    {
      "epoch": 0.9590872822986692,
      "grad_norm": 1.3853500476888336,
      "learning_rate": 4.3737409050283454e-08,
      "loss": 0.7005,
      "step": 13072
    },
    {
      "epoch": 0.9590872822986692,
      "lm_loss": 0.523,
      "step": 13072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9590872822986692,
      "lm_loss": 0.5248,
      "step": 13072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9590872822986692,
      "lm_loss": 0.5315,
      "step": 13072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9590872822986692,
      "lm_loss": 0.4951,
      "step": 13072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9590872822986692,
      "lm_loss": 0.4356,
      "step": 13072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9590872822986692,
      "lm_loss": 0.4939,
      "step": 13072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9590872822986692,
      "lm_loss": 0.6506,
      "step": 13072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9590872822986692,
      "lm_loss": 0.699,
      "step": 13072,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9591606518888084,
      "grad_norm": 0.5412143169371938,
      "learning_rate": 4.358073200749213e-08,
      "loss": 0.4777,
      "step": 13073
    },
    {
      "epoch": 0.9592340214789475,
      "grad_norm": 1.4053295564758475,
      "learning_rate": 4.342433486597075e-08,
      "loss": 0.662,
      "step": 13074
    },
    {
      "epoch": 0.9593073910690867,
      "grad_norm": 1.349261638903064,
      "learning_rate": 4.326821763455169e-08,
      "loss": 0.7761,
      "step": 13075
    },
    {
      "epoch": 0.9593807606592257,
      "grad_norm": 1.5754926581584017,
      "learning_rate": 4.311238032205234e-08,
      "loss": 0.7717,
      "step": 13076
    },
    {
      "epoch": 0.9594541302493649,
      "grad_norm": 1.4692623427900866,
      "learning_rate": 4.295682293727177e-08,
      "loss": 0.5608,
      "step": 13077
    },
    {
      "epoch": 0.959527499839504,
      "grad_norm": 1.5721324479494743,
      "learning_rate": 4.280154548899573e-08,
      "loss": 0.7636,
      "step": 13078
    },
    {
      "epoch": 0.9596008694296432,
      "grad_norm": 1.5643903893615727,
      "learning_rate": 4.264654798599277e-08,
      "loss": 0.7223,
      "step": 13079
    },
    {
      "epoch": 0.9596742390197823,
      "grad_norm": 1.5402207768019336,
      "learning_rate": 4.2491830437015326e-08,
      "loss": 0.7736,
      "step": 13080
    },
    {
      "epoch": 0.9596742390197823,
      "lm_loss": 0.9375,
      "step": 13080,
      "vm_loss": 0.0845
    },
    {
      "epoch": 0.9596742390197823,
      "lm_loss": 1.3548,
      "step": 13080,
      "vm_loss": 0.098
    },
    {
      "epoch": 0.9596742390197823,
      "lm_loss": 0.8225,
      "step": 13080,
      "vm_loss": 0.0748
    },
    {
      "epoch": 0.9596742390197823,
      "lm_loss": 0.1483,
      "step": 13080,
      "vm_loss": 0.1247
    },
    {
      "epoch": 0.9596742390197823,
      "lm_loss": 0.8253,
      "step": 13080,
      "vm_loss": 0.05
    },
    {
      "epoch": 0.9596742390197823,
      "lm_loss": 1.1806,
      "step": 13080,
      "vm_loss": 0.1664
    },
    {
      "epoch": 0.9596742390197823,
      "lm_loss": 0.0604,
      "step": 13080,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.9596742390197823,
      "lm_loss": 0.8038,
      "step": 13080,
      "vm_loss": 0.1292
    },
    {
      "epoch": 0.9597476086099214,
      "grad_norm": 1.4738536184802953,
      "learning_rate": 4.2337392850801407e-08,
      "loss": 0.8133,
      "step": 13081
    },
    {
      "epoch": 0.9598209782000605,
      "grad_norm": 1.6352133334149745,
      "learning_rate": 4.2183235236072926e-08,
      "loss": 0.8029,
      "step": 13082
    },
    {
      "epoch": 0.9598943477901997,
      "grad_norm": 1.5389588755703516,
      "learning_rate": 4.202935760153459e-08,
      "loss": 0.6869,
      "step": 13083
    },
    {
      "epoch": 0.9599677173803388,
      "grad_norm": 1.638988523373365,
      "learning_rate": 4.187575995587667e-08,
      "loss": 0.8081,
      "step": 13084
    },
    {
      "epoch": 0.960041086970478,
      "grad_norm": 0.5304158846326206,
      "learning_rate": 4.1722442307773336e-08,
      "loss": 0.4634,
      "step": 13085
    },
    {
      "epoch": 0.960114456560617,
      "grad_norm": 1.386009023803344,
      "learning_rate": 4.156940466588211e-08,
      "loss": 0.6531,
      "step": 13086
    },
    {
      "epoch": 0.9601878261507562,
      "grad_norm": 1.5398967474785343,
      "learning_rate": 4.141664703884607e-08,
      "loss": 0.7739,
      "step": 13087
    },
    {
      "epoch": 0.9602611957408953,
      "grad_norm": 0.5440120248637478,
      "learning_rate": 4.126416943529221e-08,
      "loss": 0.4666,
      "step": 13088
    },
    {
      "epoch": 0.9602611957408953,
      "lm_loss": 1.2595,
      "step": 13088,
      "vm_loss": 0.055
    },
    {
      "epoch": 0.9602611957408953,
      "lm_loss": 0.7216,
      "step": 13088,
      "vm_loss": 0.0968
    },
    {
      "epoch": 0.9602611957408953,
      "lm_loss": 0.5902,
      "step": 13088,
      "vm_loss": 0.138
    },
    {
      "epoch": 0.9602611957408953,
      "lm_loss": 0.1325,
      "step": 13088,
      "vm_loss": 0.1665
    },
    {
      "epoch": 0.9602611957408953,
      "lm_loss": 0.9043,
      "step": 13088,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.9602611957408953,
      "lm_loss": 0.8879,
      "step": 13088,
      "vm_loss": 0.1126
    },
    {
      "epoch": 0.9602611957408953,
      "lm_loss": 0.4703,
      "step": 13088,
      "vm_loss": 0.1861
    },
    {
      "epoch": 0.9602611957408953,
      "lm_loss": 0.6896,
      "step": 13088,
      "vm_loss": 0.2014
    },
    {
      "epoch": 0.9603345653310345,
      "grad_norm": 1.4228736592238995,
      "learning_rate": 4.1111971863830866e-08,
      "loss": 0.8028,
      "step": 13089
    },
    {
      "epoch": 0.9604079349211736,
      "grad_norm": 0.5374386413498552,
      "learning_rate": 4.0960054333056275e-08,
      "loss": 0.4695,
      "step": 13090
    },
    {
      "epoch": 0.9604813045113126,
      "grad_norm": 1.5575152444994278,
      "learning_rate": 4.0808416851548795e-08,
      "loss": 0.6949,
      "step": 13091
    },
    {
      "epoch": 0.9605546741014518,
      "grad_norm": 1.5213178877056674,
      "learning_rate": 4.065705942787102e-08,
      "loss": 0.9451,
      "step": 13092
    },
    {
      "epoch": 0.9606280436915909,
      "grad_norm": 1.6565670381048347,
      "learning_rate": 4.050598207057055e-08,
      "loss": 0.7589,
      "step": 13093
    },
    {
      "epoch": 0.9607014132817301,
      "grad_norm": 1.364836855386764,
      "learning_rate": 4.0355184788179456e-08,
      "loss": 0.7079,
      "step": 13094
    },
    {
      "epoch": 0.9607747828718692,
      "grad_norm": 1.6158215066210095,
      "learning_rate": 4.0204667589213705e-08,
      "loss": 0.702,
      "step": 13095
    },
    {
      "epoch": 0.9608481524620083,
      "grad_norm": 1.6447193055544478,
      "learning_rate": 4.005443048217261e-08,
      "loss": 0.6716,
      "step": 13096
    },
    {
      "epoch": 0.9608481524620083,
      "lm_loss": 0.6267,
      "step": 13096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9608481524620083,
      "lm_loss": 0.5733,
      "step": 13096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9608481524620083,
      "lm_loss": 0.5946,
      "step": 13096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9608481524620083,
      "lm_loss": 0.6318,
      "step": 13096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9608481524620083,
      "lm_loss": 0.4707,
      "step": 13096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9608481524620083,
      "lm_loss": 0.5823,
      "step": 13096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9608481524620083,
      "lm_loss": 0.5798,
      "step": 13096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9608481524620083,
      "lm_loss": 0.3778,
      "step": 13096,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9609215220521474,
      "grad_norm": 0.5387098586731157,
      "learning_rate": 3.990447347554105e-08,
      "loss": 0.4862,
      "step": 13097
    },
    {
      "epoch": 0.9609948916422866,
      "grad_norm": 1.4616892122630223,
      "learning_rate": 3.975479657778725e-08,
      "loss": 0.7201,
      "step": 13098
    },
    {
      "epoch": 0.9610682612324257,
      "grad_norm": 0.5376602068466799,
      "learning_rate": 3.960539979736444e-08,
      "loss": 0.4772,
      "step": 13099
    },
    {
      "epoch": 0.9611416308225648,
      "grad_norm": 1.3524042172128656,
      "learning_rate": 3.945628314270866e-08,
      "loss": 0.7174,
      "step": 13100
    },
    {
      "epoch": 0.9612150004127039,
      "grad_norm": 0.5306987820879971,
      "learning_rate": 3.930744662224095e-08,
      "loss": 0.4539,
      "step": 13101
    },
    {
      "epoch": 0.9612883700028431,
      "grad_norm": 1.507825461302253,
      "learning_rate": 3.9158890244366806e-08,
      "loss": 0.7229,
      "step": 13102
    },
    {
      "epoch": 0.9613617395929822,
      "grad_norm": 0.5422169628515973,
      "learning_rate": 3.9010614017475633e-08,
      "loss": 0.461,
      "step": 13103
    },
    {
      "epoch": 0.9614351091831214,
      "grad_norm": 0.5429376812893333,
      "learning_rate": 3.886261794994073e-08,
      "loss": 0.4617,
      "step": 13104
    },
    {
      "epoch": 0.9614351091831214,
      "lm_loss": 1.6879,
      "step": 13104,
      "vm_loss": 0.057
    },
    {
      "epoch": 0.9614351091831214,
      "lm_loss": 0.1123,
      "step": 13104,
      "vm_loss": 0.1216
    },
    {
      "epoch": 0.9614351091831214,
      "lm_loss": 0.9834,
      "step": 13104,
      "vm_loss": 0.1166
    },
    {
      "epoch": 0.9614351091831214,
      "lm_loss": 0.3919,
      "step": 13104,
      "vm_loss": 0.09
    },
    {
      "epoch": 0.9614351091831214,
      "lm_loss": 0.8923,
      "step": 13104,
      "vm_loss": 0.1342
    },
    {
      "epoch": 0.9614351091831214,
      "lm_loss": 0.5187,
      "step": 13104,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.9614351091831214,
      "lm_loss": 0.1147,
      "step": 13104,
      "vm_loss": 0.1592
    },
    {
      "epoch": 0.9614351091831214,
      "lm_loss": 0.5324,
      "step": 13104,
      "vm_loss": 0.0867
    },
    {
      "epoch": 0.9615084787732604,
      "grad_norm": 1.4316052262142682,
      "learning_rate": 3.8714902050119854e-08,
      "loss": 0.6569,
      "step": 13105
    },
    {
      "epoch": 0.9615818483633995,
      "grad_norm": 1.4801174089615758,
      "learning_rate": 3.856746632635466e-08,
      "loss": 0.8053,
      "step": 13106
    },
    {
      "epoch": 0.9616552179535387,
      "grad_norm": 1.6875828407523936,
      "learning_rate": 3.842031078697128e-08,
      "loss": 0.7516,
      "step": 13107
    },
    {
      "epoch": 0.9617285875436778,
      "grad_norm": 0.5538309342758937,
      "learning_rate": 3.827343544028028e-08,
      "loss": 0.4788,
      "step": 13108
    },
    {
      "epoch": 0.961801957133817,
      "grad_norm": 1.7432034912633774,
      "learning_rate": 3.812684029457614e-08,
      "loss": 0.8665,
      "step": 13109
    },
    {
      "epoch": 0.961875326723956,
      "grad_norm": 1.4173202029673075,
      "learning_rate": 3.798052535813668e-08,
      "loss": 0.7694,
      "step": 13110
    },
    {
      "epoch": 0.9619486963140952,
      "grad_norm": 0.5388808025270124,
      "learning_rate": 3.783449063922584e-08,
      "loss": 0.4789,
      "step": 13111
    },
    {
      "epoch": 0.9620220659042343,
      "grad_norm": 0.5294758047399092,
      "learning_rate": 3.768873614608926e-08,
      "loss": 0.4683,
      "step": 13112
    },
    {
      "epoch": 0.9620220659042343,
      "lm_loss": 1.4139,
      "step": 13112,
      "vm_loss": 0.0368
    },
    {
      "epoch": 0.9620220659042343,
      "lm_loss": 0.921,
      "step": 13112,
      "vm_loss": 0.0637
    },
    {
      "epoch": 0.9620220659042343,
      "lm_loss": 0.013,
      "step": 13112,
      "vm_loss": 0.1253
    },
    {
      "epoch": 0.9620220659042343,
      "lm_loss": 0.8082,
      "step": 13112,
      "vm_loss": 0.1067
    },
    {
      "epoch": 0.9620220659042343,
      "lm_loss": 1.0686,
      "step": 13112,
      "vm_loss": 0.16
    },
    {
      "epoch": 0.9620220659042343,
      "lm_loss": 0.6351,
      "step": 13112,
      "vm_loss": 0.1254
    },
    {
      "epoch": 0.9620220659042343,
      "lm_loss": 0.1797,
      "step": 13112,
      "vm_loss": 0.0787
    },
    {
      "epoch": 0.9620220659042343,
      "lm_loss": 0.6819,
      "step": 13112,
      "vm_loss": 0.1941
    },
    {
      "epoch": 0.9620954354943735,
      "grad_norm": 1.575603725827499,
      "learning_rate": 3.7543261886958674e-08,
      "loss": 0.7634,
      "step": 13113
    },
    {
      "epoch": 0.9621688050845126,
      "grad_norm": 1.433640680233627,
      "learning_rate": 3.739806787004974e-08,
      "loss": 0.7544,
      "step": 13114
    },
    {
      "epoch": 0.9622421746746517,
      "grad_norm": 1.5792205551740754,
      "learning_rate": 3.7253154103562006e-08,
      "loss": 0.6614,
      "step": 13115
    },
    {
      "epoch": 0.9623155442647908,
      "grad_norm": 1.3922808747846784,
      "learning_rate": 3.7108520595677823e-08,
      "loss": 0.7361,
      "step": 13116
    },
    {
      "epoch": 0.96238891385493,
      "grad_norm": 1.6037963816773104,
      "learning_rate": 3.696416735456621e-08,
      "loss": 0.6996,
      "step": 13117
    },
    {
      "epoch": 0.9624622834450691,
      "grad_norm": 1.5107599995963554,
      "learning_rate": 3.682009438837897e-08,
      "loss": 0.7403,
      "step": 13118
    },
    {
      "epoch": 0.9625356530352083,
      "grad_norm": 1.6496878842851017,
      "learning_rate": 3.667630170525238e-08,
      "loss": 0.7081,
      "step": 13119
    },
    {
      "epoch": 0.9626090226253473,
      "grad_norm": 1.5479738050359921,
      "learning_rate": 3.653278931330606e-08,
      "loss": 0.6065,
      "step": 13120
    },
    {
      "epoch": 0.9626090226253473,
      "lm_loss": 0.8405,
      "step": 13120,
      "vm_loss": 0.059
    },
    {
      "epoch": 0.9626090226253473,
      "lm_loss": 1.3145,
      "step": 13120,
      "vm_loss": 0.0812
    },
    {
      "epoch": 0.9626090226253473,
      "lm_loss": 0.1606,
      "step": 13120,
      "vm_loss": 0.0811
    },
    {
      "epoch": 0.9626090226253473,
      "lm_loss": 0.1077,
      "step": 13120,
      "vm_loss": 0.2002
    },
    {
      "epoch": 0.9626090226253473,
      "lm_loss": 0.3067,
      "step": 13120,
      "vm_loss": 0.1106
    },
    {
      "epoch": 0.9626090226253473,
      "lm_loss": 0.6127,
      "step": 13120,
      "vm_loss": 0.0666
    },
    {
      "epoch": 0.9626090226253473,
      "lm_loss": 0.6704,
      "step": 13120,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.9626090226253473,
      "lm_loss": 0.6669,
      "step": 13120,
      "vm_loss": 0.1648
    },
    {
      "epoch": 0.9626823922154865,
      "grad_norm": 1.5951846198248352,
      "learning_rate": 3.6389557220645185e-08,
      "loss": 0.7493,
      "step": 13121
    },
    {
      "epoch": 0.9627557618056256,
      "grad_norm": 1.5147319037054778,
      "learning_rate": 3.624660543535774e-08,
      "loss": 0.7586,
      "step": 13122
    },
    {
      "epoch": 0.9628291313957648,
      "grad_norm": 0.539155089767216,
      "learning_rate": 3.610393396551726e-08,
      "loss": 0.4624,
      "step": 13123
    },
    {
      "epoch": 0.9629025009859039,
      "grad_norm": 0.5632910649593876,
      "learning_rate": 3.5961542819180626e-08,
      "loss": 0.4705,
      "step": 13124
    },
    {
      "epoch": 0.962975870576043,
      "grad_norm": 1.6396464569048965,
      "learning_rate": 3.5819432004389196e-08,
      "loss": 0.766,
      "step": 13125
    },
    {
      "epoch": 0.9630492401661821,
      "grad_norm": 1.65316269166121,
      "learning_rate": 3.567760152916766e-08,
      "loss": 0.7229,
      "step": 13126
    },
    {
      "epoch": 0.9631226097563212,
      "grad_norm": 1.7038661535355617,
      "learning_rate": 3.553605140152572e-08,
      "loss": 0.5867,
      "step": 13127
    },
    {
      "epoch": 0.9631959793464604,
      "grad_norm": 1.6835271622481534,
      "learning_rate": 3.539478162945753e-08,
      "loss": 0.7434,
      "step": 13128
    },
    {
      "epoch": 0.9631959793464604,
      "lm_loss": 2.3072,
      "step": 13128,
      "vm_loss": 0.1035
    },
    {
      "epoch": 0.9631959793464604,
      "lm_loss": 1.3245,
      "step": 13128,
      "vm_loss": 0.0553
    },
    {
      "epoch": 0.9631959793464604,
      "lm_loss": 0.2878,
      "step": 13128,
      "vm_loss": 0.1002
    },
    {
      "epoch": 0.9631959793464604,
      "lm_loss": 0.2975,
      "step": 13128,
      "vm_loss": 0.0911
    },
    {
      "epoch": 0.9631959793464604,
      "lm_loss": 0.9143,
      "step": 13128,
      "vm_loss": 0.1109
    },
    {
      "epoch": 0.9631959793464604,
      "lm_loss": 1.3177,
      "step": 13128,
      "vm_loss": 0.0731
    },
    {
      "epoch": 0.9631959793464604,
      "lm_loss": 0.2412,
      "step": 13128,
      "vm_loss": 0.0708
    },
    {
      "epoch": 0.9631959793464604,
      "lm_loss": 0.7419,
      "step": 13128,
      "vm_loss": 0.2157
    },
    {
      "epoch": 0.9632693489365995,
      "grad_norm": 1.5939406161964578,
      "learning_rate": 3.525379222094061e-08,
      "loss": 0.7197,
      "step": 13129
    },
    {
      "epoch": 0.9633427185267386,
      "grad_norm": 0.5338335676679226,
      "learning_rate": 3.511308318393691e-08,
      "loss": 0.4692,
      "step": 13130
    },
    {
      "epoch": 0.9634160881168777,
      "grad_norm": 0.5402923298813087,
      "learning_rate": 3.497265452639287e-08,
      "loss": 0.4739,
      "step": 13131
    },
    {
      "epoch": 0.9634894577070169,
      "grad_norm": 1.4331308836443066,
      "learning_rate": 3.4832506256238795e-08,
      "loss": 0.5957,
      "step": 13132
    },
    {
      "epoch": 0.963562827297156,
      "grad_norm": 1.513803960914314,
      "learning_rate": 3.469263838138892e-08,
      "loss": 0.5965,
      "step": 13133
    },
    {
      "epoch": 0.9636361968872952,
      "grad_norm": 0.5650447389835538,
      "learning_rate": 3.455305090974193e-08,
      "loss": 0.4813,
      "step": 13134
    },
    {
      "epoch": 0.9637095664774342,
      "grad_norm": 1.6686298417766645,
      "learning_rate": 3.44137438491815e-08,
      "loss": 0.7854,
      "step": 13135
    },
    {
      "epoch": 0.9637829360675734,
      "grad_norm": 1.4705731026345605,
      "learning_rate": 3.427471720757358e-08,
      "loss": 0.616,
      "step": 13136
    },
    {
      "epoch": 0.9637829360675734,
      "lm_loss": 0.0987,
      "step": 13136,
      "vm_loss": 0.0789
    },
    {
      "epoch": 0.9637829360675734,
      "lm_loss": 0.6472,
      "step": 13136,
      "vm_loss": 0.069
    },
    {
      "epoch": 0.9637829360675734,
      "lm_loss": 0.0192,
      "step": 13136,
      "vm_loss": 0.1188
    },
    {
      "epoch": 0.9637829360675734,
      "lm_loss": 1.2273,
      "step": 13136,
      "vm_loss": 0.1151
    },
    {
      "epoch": 0.9637829360675734,
      "lm_loss": 0.1493,
      "step": 13136,
      "vm_loss": 0.0862
    },
    {
      "epoch": 0.9637829360675734,
      "lm_loss": 0.0963,
      "step": 13136,
      "vm_loss": 0.0846
    },
    {
      "epoch": 0.9637829360675734,
      "lm_loss": 0.9047,
      "step": 13136,
      "vm_loss": 0.1201
    },
    {
      "epoch": 0.9637829360675734,
      "lm_loss": 0.2512,
      "step": 13136,
      "vm_loss": 0.1582
    },
    {
      "epoch": 0.9638563056577125,
      "grad_norm": 1.4464831926671426,
      "learning_rate": 3.4135970992769663e-08,
      "loss": 0.6941,
      "step": 13137
    },
    {
      "epoch": 0.9639296752478517,
      "grad_norm": 0.5390072708264999,
      "learning_rate": 3.3997505212605695e-08,
      "loss": 0.4733,
      "step": 13138
    },
    {
      "epoch": 0.9640030448379908,
      "grad_norm": 0.5533614689631833,
      "learning_rate": 3.385931987490043e-08,
      "loss": 0.469,
      "step": 13139
    },
    {
      "epoch": 0.9640764144281299,
      "grad_norm": 1.5410108689101523,
      "learning_rate": 3.3721414987457624e-08,
      "loss": 0.73,
      "step": 13140
    },
    {
      "epoch": 0.964149784018269,
      "grad_norm": 0.5449748569972854,
      "learning_rate": 3.358379055806549e-08,
      "loss": 0.477,
      "step": 13141
    },
    {
      "epoch": 0.9642231536084082,
      "grad_norm": 0.5610354847782082,
      "learning_rate": 3.344644659449614e-08,
      "loss": 0.4828,
      "step": 13142
    },
    {
      "epoch": 0.9642965231985473,
      "grad_norm": 1.6419624396862962,
      "learning_rate": 3.330938310450504e-08,
      "loss": 0.7086,
      "step": 13143
    },
    {
      "epoch": 0.9643698927886865,
      "grad_norm": 1.5174966101613998,
      "learning_rate": 3.317260009583267e-08,
      "loss": 0.6516,
      "step": 13144
    },
    {
      "epoch": 0.9643698927886865,
      "lm_loss": 0.5472,
      "step": 13144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9643698927886865,
      "lm_loss": 0.4832,
      "step": 13144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9643698927886865,
      "lm_loss": 0.4078,
      "step": 13144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9643698927886865,
      "lm_loss": 0.4248,
      "step": 13144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9643698927886865,
      "lm_loss": 0.3816,
      "step": 13144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9643698927886865,
      "lm_loss": 0.5893,
      "step": 13144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9643698927886865,
      "lm_loss": 0.4949,
      "step": 13144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9643698927886865,
      "lm_loss": 0.5879,
      "step": 13144,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9644432623788255,
      "grad_norm": 0.546226661974076,
      "learning_rate": 3.303609757620396e-08,
      "loss": 0.4715,
      "step": 13145
    },
    {
      "epoch": 0.9645166319689646,
      "grad_norm": 1.3642775593451768,
      "learning_rate": 3.289987555332774e-08,
      "loss": 0.6825,
      "step": 13146
    },
    {
      "epoch": 0.9645900015591038,
      "grad_norm": 1.5109518780826763,
      "learning_rate": 3.276393403489564e-08,
      "loss": 0.8191,
      "step": 13147
    },
    {
      "epoch": 0.9646633711492429,
      "grad_norm": 0.5359765979921981,
      "learning_rate": 3.2628273028585956e-08,
      "loss": 0.4574,
      "step": 13148
    },
    {
      "epoch": 0.9647367407393821,
      "grad_norm": 1.6471405996142725,
      "learning_rate": 3.249289254205867e-08,
      "loss": 0.7445,
      "step": 13149
    },
    {
      "epoch": 0.9648101103295211,
      "grad_norm": 1.525684394387894,
      "learning_rate": 3.23577925829599e-08,
      "loss": 0.6517,
      "step": 13150
    },
    {
      "epoch": 0.9648834799196603,
      "grad_norm": 1.5612921188169928,
      "learning_rate": 3.222297315891853e-08,
      "loss": 0.6973,
      "step": 13151
    },
    {
      "epoch": 0.9649568495097994,
      "grad_norm": 0.5293447293389022,
      "learning_rate": 3.2088434277548485e-08,
      "loss": 0.4609,
      "step": 13152
    },
    {
      "epoch": 0.9649568495097994,
      "lm_loss": 1.084,
      "step": 13152,
      "vm_loss": 0.0667
    },
    {
      "epoch": 0.9649568495097994,
      "lm_loss": 0.0531,
      "step": 13152,
      "vm_loss": 0.0516
    },
    {
      "epoch": 0.9649568495097994,
      "lm_loss": 0.1373,
      "step": 13152,
      "vm_loss": 0.1903
    },
    {
      "epoch": 0.9649568495097994,
      "lm_loss": 0.4439,
      "step": 13152,
      "vm_loss": 0.1324
    },
    {
      "epoch": 0.9649568495097994,
      "lm_loss": 0.7602,
      "step": 13152,
      "vm_loss": 0.089
    },
    {
      "epoch": 0.9649568495097994,
      "lm_loss": 0.5193,
      "step": 13152,
      "vm_loss": 0.0644
    },
    {
      "epoch": 0.9649568495097994,
      "lm_loss": 0.242,
      "step": 13152,
      "vm_loss": 0.0631
    },
    {
      "epoch": 0.9649568495097994,
      "lm_loss": 0.9693,
      "step": 13152,
      "vm_loss": 0.14
    },
    {
      "epoch": 0.9650302190999386,
      "grad_norm": 1.6090061569510963,
      "learning_rate": 3.1954175946447565e-08,
      "loss": 0.7133,
      "step": 13153
    },
    {
      "epoch": 0.9651035886900777,
      "grad_norm": 1.5074853955387906,
      "learning_rate": 3.182019817319692e-08,
      "loss": 0.6617,
      "step": 13154
    },
    {
      "epoch": 0.9651769582802168,
      "grad_norm": 1.6774442392479154,
      "learning_rate": 3.1686500965363295e-08,
      "loss": 0.8232,
      "step": 13155
    },
    {
      "epoch": 0.9652503278703559,
      "grad_norm": 1.6512812966773867,
      "learning_rate": 3.1553084330497306e-08,
      "loss": 0.7338,
      "step": 13156
    },
    {
      "epoch": 0.9653236974604951,
      "grad_norm": 1.4475615798226678,
      "learning_rate": 3.141994827613237e-08,
      "loss": 0.6801,
      "step": 13157
    },
    {
      "epoch": 0.9653970670506342,
      "grad_norm": 1.365022219187247,
      "learning_rate": 3.1287092809787476e-08,
      "loss": 0.657,
      "step": 13158
    },
    {
      "epoch": 0.9654704366407734,
      "grad_norm": 0.5370263957196647,
      "learning_rate": 3.1154517938965514e-08,
      "loss": 0.4593,
      "step": 13159
    },
    {
      "epoch": 0.9655438062309124,
      "grad_norm": 1.6128355142332773,
      "learning_rate": 3.102222367115326e-08,
      "loss": 0.7126,
      "step": 13160
    },
    {
      "epoch": 0.9655438062309124,
      "lm_loss": 0.6039,
      "step": 13160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9655438062309124,
      "lm_loss": 0.5921,
      "step": 13160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9655438062309124,
      "lm_loss": 0.5361,
      "step": 13160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9655438062309124,
      "lm_loss": 0.328,
      "step": 13160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9655438062309124,
      "lm_loss": 0.3449,
      "step": 13160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9655438062309124,
      "lm_loss": 0.4379,
      "step": 13160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9655438062309124,
      "lm_loss": 0.4381,
      "step": 13160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9655438062309124,
      "lm_loss": 0.448,
      "step": 13160,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9656171758210516,
      "grad_norm": 0.549678189356234,
      "learning_rate": 3.0890210013820866e-08,
      "loss": 0.4745,
      "step": 13161
    },
    {
      "epoch": 0.9656905454111907,
      "grad_norm": 1.299343392789533,
      "learning_rate": 3.075847697442458e-08,
      "loss": 0.6053,
      "step": 13162
    },
    {
      "epoch": 0.9657639150013299,
      "grad_norm": 1.50716979436742,
      "learning_rate": 3.0627024560402895e-08,
      "loss": 0.7123,
      "step": 13163
    },
    {
      "epoch": 0.965837284591469,
      "grad_norm": 1.4033661075681687,
      "learning_rate": 3.049585277917988e-08,
      "loss": 0.7056,
      "step": 13164
    },
    {
      "epoch": 0.965910654181608,
      "grad_norm": 0.5418396228000628,
      "learning_rate": 3.036496163816238e-08,
      "loss": 0.4788,
      "step": 13165
    },
    {
      "epoch": 0.9659840237717472,
      "grad_norm": 1.4883000151526247,
      "learning_rate": 3.0234351144742824e-08,
      "loss": 0.7798,
      "step": 13166
    },
    {
      "epoch": 0.9660573933618863,
      "grad_norm": 1.4871052071370894,
      "learning_rate": 3.010402130629697e-08,
      "loss": 0.6731,
      "step": 13167
    },
    {
      "epoch": 0.9661307629520255,
      "grad_norm": 0.5444385892825143,
      "learning_rate": 2.997397213018449e-08,
      "loss": 0.4888,
      "step": 13168
    },
    {
      "epoch": 0.9661307629520255,
      "lm_loss": 0.6226,
      "step": 13168,
      "vm_loss": 0.069
    },
    {
      "epoch": 0.9661307629520255,
      "lm_loss": 1.2932,
      "step": 13168,
      "vm_loss": 0.1007
    },
    {
      "epoch": 0.9661307629520255,
      "lm_loss": 0.0248,
      "step": 13168,
      "vm_loss": 0.1618
    },
    {
      "epoch": 0.9661307629520255,
      "lm_loss": 0.3283,
      "step": 13168,
      "vm_loss": 0.1276
    },
    {
      "epoch": 0.9661307629520255,
      "lm_loss": 0.7062,
      "step": 13168,
      "vm_loss": 0.1703
    },
    {
      "epoch": 0.9661307629520255,
      "lm_loss": 0.3741,
      "step": 13168,
      "vm_loss": 0.1837
    },
    {
      "epoch": 0.9661307629520255,
      "lm_loss": 0.4088,
      "step": 13168,
      "vm_loss": 0.0567
    },
    {
      "epoch": 0.9661307629520255,
      "lm_loss": 0.9499,
      "step": 13168,
      "vm_loss": 0.2598
    },
    {
      "epoch": 0.9662041325421646,
      "grad_norm": 1.3618724427550706,
      "learning_rate": 2.984420362375007e-08,
      "loss": 0.7365,
      "step": 13169
    },
    {
      "epoch": 0.9662775021323037,
      "grad_norm": 1.4650558135419944,
      "learning_rate": 2.971471579432228e-08,
      "loss": 0.6757,
      "step": 13170
    },
    {
      "epoch": 0.9663508717224428,
      "grad_norm": 0.5363238995830653,
      "learning_rate": 2.95855086492125e-08,
      "loss": 0.4599,
      "step": 13171
    },
    {
      "epoch": 0.966424241312582,
      "grad_norm": 1.5541928118292407,
      "learning_rate": 2.945658219571823e-08,
      "loss": 0.7807,
      "step": 13172
    },
    {
      "epoch": 0.9664976109027211,
      "grad_norm": 1.6329040901717926,
      "learning_rate": 2.932793644112031e-08,
      "loss": 0.7208,
      "step": 13173
    },
    {
      "epoch": 0.9665709804928603,
      "grad_norm": 1.533229113422601,
      "learning_rate": 2.9199571392683477e-08,
      "loss": 0.7655,
      "step": 13174
    },
    {
      "epoch": 0.9666443500829993,
      "grad_norm": 0.5552927147512529,
      "learning_rate": 2.9071487057656943e-08,
      "loss": 0.4873,
      "step": 13175
    },
    {
      "epoch": 0.9667177196731385,
      "grad_norm": 1.4755968586668815,
      "learning_rate": 2.8943683443273807e-08,
      "loss": 0.6813,
      "step": 13176
    },
    {
      "epoch": 0.9667177196731385,
      "lm_loss": 0.4253,
      "step": 13176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9667177196731385,
      "lm_loss": 0.5053,
      "step": 13176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9667177196731385,
      "lm_loss": 0.3477,
      "step": 13176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9667177196731385,
      "lm_loss": 0.4568,
      "step": 13176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9667177196731385,
      "lm_loss": 0.4134,
      "step": 13176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9667177196731385,
      "lm_loss": 0.4419,
      "step": 13176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9667177196731385,
      "lm_loss": 0.7086,
      "step": 13176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9667177196731385,
      "lm_loss": 0.3359,
      "step": 13176,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9667910892632776,
      "grad_norm": 0.5556783349921264,
      "learning_rate": 2.8816160556751073e-08,
      "loss": 0.4688,
      "step": 13177
    },
    {
      "epoch": 0.9668644588534168,
      "grad_norm": 1.4577776267821345,
      "learning_rate": 2.8688918405290757e-08,
      "loss": 0.8134,
      "step": 13178
    },
    {
      "epoch": 0.9669378284435558,
      "grad_norm": 1.459459271191988,
      "learning_rate": 2.8561956996078778e-08,
      "loss": 0.8133,
      "step": 13179
    },
    {
      "epoch": 0.967011198033695,
      "grad_norm": 1.4020005775548656,
      "learning_rate": 2.84352763362844e-08,
      "loss": 0.7041,
      "step": 13180
    },
    {
      "epoch": 0.9670845676238341,
      "grad_norm": 1.3953522905592137,
      "learning_rate": 2.8308876433061905e-08,
      "loss": 0.5915,
      "step": 13181
    },
    {
      "epoch": 0.9671579372139733,
      "grad_norm": 1.609611081015232,
      "learning_rate": 2.818275729354947e-08,
      "loss": 0.704,
      "step": 13182
    },
    {
      "epoch": 0.9672313068041124,
      "grad_norm": 1.458214150940703,
      "learning_rate": 2.805691892486917e-08,
      "loss": 0.7125,
      "step": 13183
    },
    {
      "epoch": 0.9673046763942514,
      "grad_norm": 1.4399608592541777,
      "learning_rate": 2.7931361334127548e-08,
      "loss": 0.7806,
      "step": 13184
    },
    {
      "epoch": 0.9673046763942514,
      "lm_loss": 1.1434,
      "step": 13184,
      "vm_loss": 0.0806
    },
    {
      "epoch": 0.9673046763942514,
      "lm_loss": 0.8193,
      "step": 13184,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.9673046763942514,
      "lm_loss": 0.1688,
      "step": 13184,
      "vm_loss": 0.1864
    },
    {
      "epoch": 0.9673046763942514,
      "lm_loss": 0.5018,
      "step": 13184,
      "vm_loss": 0.2403
    },
    {
      "epoch": 0.9673046763942514,
      "lm_loss": 0.6349,
      "step": 13184,
      "vm_loss": 0.1545
    },
    {
      "epoch": 0.9673046763942514,
      "lm_loss": 0.7497,
      "step": 13184,
      "vm_loss": 0.1121
    },
    {
      "epoch": 0.9673046763942514,
      "lm_loss": 0.0628,
      "step": 13184,
      "vm_loss": 0.0497
    },
    {
      "epoch": 0.9673046763942514,
      "lm_loss": 0.1632,
      "step": 13184,
      "vm_loss": 0.1488
    },
    {
      "epoch": 0.9673780459843906,
      "grad_norm": 1.574114401582448,
      "learning_rate": 2.780608452841449e-08,
      "loss": 0.6696,
      "step": 13185
    },
    {
      "epoch": 0.9674514155745297,
      "grad_norm": 1.5092583615826463,
      "learning_rate": 2.7681088514806e-08,
      "loss": 0.6889,
      "step": 13186
    },
    {
      "epoch": 0.9675247851646689,
      "grad_norm": 1.7053622874269792,
      "learning_rate": 2.755637330035976e-08,
      "loss": 0.711,
      "step": 13187
    },
    {
      "epoch": 0.967598154754808,
      "grad_norm": 0.5454329381395313,
      "learning_rate": 2.7431938892119592e-08,
      "loss": 0.4895,
      "step": 13188
    },
    {
      "epoch": 0.9676715243449471,
      "grad_norm": 1.6222392548797848,
      "learning_rate": 2.7307785297111533e-08,
      "loss": 0.7188,
      "step": 13189
    },
    {
      "epoch": 0.9677448939350862,
      "grad_norm": 1.4715530819663067,
      "learning_rate": 2.718391252234831e-08,
      "loss": 0.8376,
      "step": 13190
    },
    {
      "epoch": 0.9678182635252254,
      "grad_norm": 0.5489879299269856,
      "learning_rate": 2.7060320574823774e-08,
      "loss": 0.4707,
      "step": 13191
    },
    {
      "epoch": 0.9678916331153645,
      "grad_norm": 1.5992879275694427,
      "learning_rate": 2.6937009461519005e-08,
      "loss": 0.8438,
      "step": 13192
    },
    {
      "epoch": 0.9678916331153645,
      "lm_loss": 0.5466,
      "step": 13192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9678916331153645,
      "lm_loss": 0.4085,
      "step": 13192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9678916331153645,
      "lm_loss": 0.4929,
      "step": 13192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9678916331153645,
      "lm_loss": 0.4382,
      "step": 13192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9678916331153645,
      "lm_loss": 0.5905,
      "step": 13192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9678916331153645,
      "lm_loss": 0.4776,
      "step": 13192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9678916331153645,
      "lm_loss": 0.5074,
      "step": 13192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9678916331153645,
      "lm_loss": 0.4584,
      "step": 13192,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9679650027055037,
      "grad_norm": 0.5302732134689867,
      "learning_rate": 2.681397918939621e-08,
      "loss": 0.4808,
      "step": 13193
    },
    {
      "epoch": 0.9680383722956427,
      "grad_norm": 0.5496419777015373,
      "learning_rate": 2.6691229765403724e-08,
      "loss": 0.4618,
      "step": 13194
    },
    {
      "epoch": 0.9681117418857819,
      "grad_norm": 1.5231449173365346,
      "learning_rate": 2.6568761196474336e-08,
      "loss": 0.6734,
      "step": 13195
    },
    {
      "epoch": 0.968185111475921,
      "grad_norm": 1.8065927013634124,
      "learning_rate": 2.6446573489523063e-08,
      "loss": 0.8098,
      "step": 13196
    },
    {
      "epoch": 0.9682584810660602,
      "grad_norm": 0.5447493175872486,
      "learning_rate": 2.632466665145106e-08,
      "loss": 0.4865,
      "step": 13197
    },
    {
      "epoch": 0.9683318506561993,
      "grad_norm": 1.5376748349830864,
      "learning_rate": 2.62030406891417e-08,
      "loss": 0.7795,
      "step": 13198
    },
    {
      "epoch": 0.9684052202463383,
      "grad_norm": 1.4339595977133466,
      "learning_rate": 2.60816956094645e-08,
      "loss": 0.7104,
      "step": 13199
    },
    {
      "epoch": 0.9684785898364775,
      "grad_norm": 1.54029018328479,
      "learning_rate": 2.5960631419271186e-08,
      "loss": 0.662,
      "step": 13200
    },
    {
      "epoch": 0.9684785898364775,
      "lm_loss": 0.885,
      "step": 13200,
      "vm_loss": 0.0437
    },
    {
      "epoch": 0.9684785898364775,
      "lm_loss": 0.1934,
      "step": 13200,
      "vm_loss": 0.1382
    },
    {
      "epoch": 0.9684785898364775,
      "lm_loss": 1.0725,
      "step": 13200,
      "vm_loss": 0.1802
    },
    {
      "epoch": 0.9684785898364775,
      "lm_loss": 0.8748,
      "step": 13200,
      "vm_loss": 0.132
    },
    {
      "epoch": 0.9684785898364775,
      "lm_loss": 0.4383,
      "step": 13200,
      "vm_loss": 0.1678
    },
    {
      "epoch": 0.9684785898364775,
      "lm_loss": 0.4177,
      "step": 13200,
      "vm_loss": 0.0735
    },
    {
      "epoch": 0.9684785898364775,
      "lm_loss": 0.18,
      "step": 13200,
      "vm_loss": 0.0671
    },
    {
      "epoch": 0.9684785898364775,
      "lm_loss": 0.2023,
      "step": 13200,
      "vm_loss": 0.2698
    },
    {
      "epoch": 0.9685519594266166,
      "grad_norm": 1.4694078484746447,
      "learning_rate": 2.5839848125399634e-08,
      "loss": 0.6473,
      "step": 13201
    },
    {
      "epoch": 0.9686253290167558,
      "grad_norm": 1.4260858149280315,
      "learning_rate": 2.571934573466939e-08,
      "loss": 0.6503,
      "step": 13202
    },
    {
      "epoch": 0.968698698606895,
      "grad_norm": 1.5023894446594779,
      "learning_rate": 2.5599124253886664e-08,
      "loss": 0.7418,
      "step": 13203
    },
    {
      "epoch": 0.968772068197034,
      "grad_norm": 1.5091636495512066,
      "learning_rate": 2.5479183689840482e-08,
      "loss": 0.6751,
      "step": 13204
    },
    {
      "epoch": 0.9688454377871731,
      "grad_norm": 0.5316182286370121,
      "learning_rate": 2.5359524049303754e-08,
      "loss": 0.4636,
      "step": 13205
    },
    {
      "epoch": 0.9689188073773123,
      "grad_norm": 1.3695590531654014,
      "learning_rate": 2.5240145339034406e-08,
      "loss": 0.733,
      "step": 13206
    },
    {
      "epoch": 0.9689921769674514,
      "grad_norm": 1.6949466688867487,
      "learning_rate": 2.5121047565773714e-08,
      "loss": 0.7508,
      "step": 13207
    },
    {
      "epoch": 0.9690655465575906,
      "grad_norm": 1.418592801828061,
      "learning_rate": 2.5002230736247412e-08,
      "loss": 0.7561,
      "step": 13208
    },
    {
      "epoch": 0.9690655465575906,
      "lm_loss": 1.637,
      "step": 13208,
      "vm_loss": 0.0816
    },
    {
      "epoch": 0.9690655465575906,
      "lm_loss": 1.693,
      "step": 13208,
      "vm_loss": 0.1814
    },
    {
      "epoch": 0.9690655465575906,
      "lm_loss": 0.1364,
      "step": 13208,
      "vm_loss": 0.1177
    },
    {
      "epoch": 0.9690655465575906,
      "lm_loss": 0.642,
      "step": 13208,
      "vm_loss": 0.1102
    },
    {
      "epoch": 0.9690655465575906,
      "lm_loss": 1.0352,
      "step": 13208,
      "vm_loss": 0.1448
    },
    {
      "epoch": 0.9690655465575906,
      "lm_loss": 1.0501,
      "step": 13208,
      "vm_loss": 0.1161
    },
    {
      "epoch": 0.9690655465575906,
      "lm_loss": 0.6582,
      "step": 13208,
      "vm_loss": 0.117
    },
    {
      "epoch": 0.9690655465575906,
      "lm_loss": 0.8949,
      "step": 13208,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.9691389161477296,
      "grad_norm": 1.6420122651663633,
      "learning_rate": 2.488369485716513e-08,
      "loss": 0.7964,
      "step": 13209
    },
    {
      "epoch": 0.9692122857378688,
      "grad_norm": 1.647078667216966,
      "learning_rate": 2.476543993522207e-08,
      "loss": 0.7117,
      "step": 13210
    },
    {
      "epoch": 0.9692856553280079,
      "grad_norm": 1.6727993682186275,
      "learning_rate": 2.464746597709511e-08,
      "loss": 0.6733,
      "step": 13211
    },
    {
      "epoch": 0.9693590249181471,
      "grad_norm": 1.5996978925859573,
      "learning_rate": 2.4529772989447252e-08,
      "loss": 0.6938,
      "step": 13212
    },
    {
      "epoch": 0.9694323945082862,
      "grad_norm": 1.4424132669634613,
      "learning_rate": 2.4412360978924854e-08,
      "loss": 0.7053,
      "step": 13213
    },
    {
      "epoch": 0.9695057640984253,
      "grad_norm": 1.2822552824584703,
      "learning_rate": 2.4295229952157608e-08,
      "loss": 0.6846,
      "step": 13214
    },
    {
      "epoch": 0.9695791336885644,
      "grad_norm": 1.4966089302407495,
      "learning_rate": 2.417837991576133e-08,
      "loss": 0.6447,
      "step": 13215
    },
    {
      "epoch": 0.9696525032787036,
      "grad_norm": 1.4736855813705079,
      "learning_rate": 2.406181087633408e-08,
      "loss": 0.65,
      "step": 13216
    },
    {
      "epoch": 0.9696525032787036,
      "lm_loss": 1.0756,
      "step": 13216,
      "vm_loss": 0.1156
    },
    {
      "epoch": 0.9696525032787036,
      "lm_loss": 1.018,
      "step": 13216,
      "vm_loss": 0.0758
    },
    {
      "epoch": 0.9696525032787036,
      "lm_loss": 0.0085,
      "step": 13216,
      "vm_loss": 0.1013
    },
    {
      "epoch": 0.9696525032787036,
      "lm_loss": 0.2151,
      "step": 13216,
      "vm_loss": 0.1402
    },
    {
      "epoch": 0.9696525032787036,
      "lm_loss": 0.6336,
      "step": 13216,
      "vm_loss": 0.1378
    },
    {
      "epoch": 0.9696525032787036,
      "lm_loss": 0.4927,
      "step": 13216,
      "vm_loss": 0.0863
    },
    {
      "epoch": 0.9696525032787036,
      "lm_loss": 0.4809,
      "step": 13216,
      "vm_loss": 0.0806
    },
    {
      "epoch": 0.9696525032787036,
      "lm_loss": 0.908,
      "step": 13216,
      "vm_loss": 0.1829
    },
    {
      "epoch": 0.9697258728688427,
      "grad_norm": 1.3335025425291764,
      "learning_rate": 2.394552284045948e-08,
      "loss": 0.7592,
      "step": 13217
    },
    {
      "epoch": 0.9697992424589819,
      "grad_norm": 1.4403583416625017,
      "learning_rate": 2.382951581470394e-08,
      "loss": 0.7641,
      "step": 13218
    },
    {
      "epoch": 0.9698726120491209,
      "grad_norm": 1.4183511862583245,
      "learning_rate": 2.371378980561889e-08,
      "loss": 0.6503,
      "step": 13219
    },
    {
      "epoch": 0.96994598163926,
      "grad_norm": 1.584189256539153,
      "learning_rate": 2.359834481973966e-08,
      "loss": 0.7617,
      "step": 13220
    },
    {
      "epoch": 0.9700193512293992,
      "grad_norm": 1.3541916092565232,
      "learning_rate": 2.3483180863585474e-08,
      "loss": 0.7869,
      "step": 13221
    },
    {
      "epoch": 0.9700927208195383,
      "grad_norm": 1.6572548133514131,
      "learning_rate": 2.336829794366058e-08,
      "loss": 0.7187,
      "step": 13222
    },
    {
      "epoch": 0.9701660904096775,
      "grad_norm": 0.5587793334589989,
      "learning_rate": 2.3253696066452557e-08,
      "loss": 0.4788,
      "step": 13223
    },
    {
      "epoch": 0.9702394599998165,
      "grad_norm": 0.5527000432266109,
      "learning_rate": 2.313937523843235e-08,
      "loss": 0.4741,
      "step": 13224
    },
    {
      "epoch": 0.9702394599998165,
      "lm_loss": 0.2455,
      "step": 13224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9702394599998165,
      "lm_loss": 0.5571,
      "step": 13224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9702394599998165,
      "lm_loss": 0.4467,
      "step": 13224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9702394599998165,
      "lm_loss": 0.4826,
      "step": 13224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9702394599998165,
      "lm_loss": 0.4039,
      "step": 13224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9702394599998165,
      "lm_loss": 0.4907,
      "step": 13224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9702394599998165,
      "lm_loss": 0.4312,
      "step": 13224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9702394599998165,
      "lm_loss": 0.5087,
      "step": 13224,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9703128295899557,
      "grad_norm": 0.5310926442646338,
      "learning_rate": 2.302533546605701e-08,
      "loss": 0.4681,
      "step": 13225
    },
    {
      "epoch": 0.9703861991800948,
      "grad_norm": 1.4618202197851258,
      "learning_rate": 2.2911576755766385e-08,
      "loss": 0.8264,
      "step": 13226
    },
    {
      "epoch": 0.970459568770234,
      "grad_norm": 1.3830281911592246,
      "learning_rate": 2.279809911398423e-08,
      "loss": 0.6431,
      "step": 13227
    },
    {
      "epoch": 0.9705329383603731,
      "grad_norm": 1.5558668761342203,
      "learning_rate": 2.268490254711986e-08,
      "loss": 0.7833,
      "step": 13228
    },
    {
      "epoch": 0.9706063079505122,
      "grad_norm": 1.9170472413335204,
      "learning_rate": 2.2571987061564827e-08,
      "loss": 0.7386,
      "step": 13229
    },
    {
      "epoch": 0.9706796775406513,
      "grad_norm": 1.5882494533921612,
      "learning_rate": 2.24593526636957e-08,
      "loss": 0.7353,
      "step": 13230
    },
    {
      "epoch": 0.9707530471307905,
      "grad_norm": 1.4574548262462077,
      "learning_rate": 2.2346999359874055e-08,
      "loss": 0.7642,
      "step": 13231
    },
    {
      "epoch": 0.9708264167209296,
      "grad_norm": 0.548359502587185,
      "learning_rate": 2.223492715644371e-08,
      "loss": 0.4701,
      "step": 13232
    },
    {
      "epoch": 0.9708264167209296,
      "lm_loss": 0.5476,
      "step": 13232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9708264167209296,
      "lm_loss": 0.4025,
      "step": 13232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9708264167209296,
      "lm_loss": 0.4403,
      "step": 13232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9708264167209296,
      "lm_loss": 0.5255,
      "step": 13232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9708264167209296,
      "lm_loss": 0.3037,
      "step": 13232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9708264167209296,
      "lm_loss": 0.4611,
      "step": 13232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9708264167209296,
      "lm_loss": 0.474,
      "step": 13232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9708264167209296,
      "lm_loss": 0.4306,
      "step": 13232,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9708997863110688,
      "grad_norm": 0.5234773277975088,
      "learning_rate": 2.2123136059735152e-08,
      "loss": 0.467,
      "step": 13233
    },
    {
      "epoch": 0.9709731559012078,
      "grad_norm": 1.5829713226849491,
      "learning_rate": 2.201162607606e-08,
      "loss": 0.6929,
      "step": 13234
    },
    {
      "epoch": 0.971046525491347,
      "grad_norm": 1.4750280401146587,
      "learning_rate": 2.1900397211715996e-08,
      "loss": 0.6936,
      "step": 13235
    },
    {
      "epoch": 0.9711198950814861,
      "grad_norm": 1.4941192160439418,
      "learning_rate": 2.178944947298478e-08,
      "loss": 0.7796,
      "step": 13236
    },
    {
      "epoch": 0.9711932646716253,
      "grad_norm": 1.6430640898126763,
      "learning_rate": 2.1678782866131898e-08,
      "loss": 0.7,
      "step": 13237
    },
    {
      "epoch": 0.9712666342617644,
      "grad_norm": 0.5477335791940563,
      "learning_rate": 2.156839739740679e-08,
      "loss": 0.491,
      "step": 13238
    },
    {
      "epoch": 0.9713400038519034,
      "grad_norm": 1.5302971493913793,
      "learning_rate": 2.1458293073042814e-08,
      "loss": 0.768,
      "step": 13239
    },
    {
      "epoch": 0.9714133734420426,
      "grad_norm": 1.3692078515561217,
      "learning_rate": 2.1348469899258316e-08,
      "loss": 0.6815,
      "step": 13240
    },
    {
      "epoch": 0.9714133734420426,
      "lm_loss": 0.735,
      "step": 13240,
      "vm_loss": 0.0857
    },
    {
      "epoch": 0.9714133734420426,
      "lm_loss": 1.3605,
      "step": 13240,
      "vm_loss": 0.0566
    },
    {
      "epoch": 0.9714133734420426,
      "lm_loss": 0.8885,
      "step": 13240,
      "vm_loss": 0.1542
    },
    {
      "epoch": 0.9714133734420426,
      "lm_loss": 0.8324,
      "step": 13240,
      "vm_loss": 0.307
    },
    {
      "epoch": 0.9714133734420426,
      "lm_loss": 0.4174,
      "step": 13240,
      "vm_loss": 0.1772
    },
    {
      "epoch": 0.9714133734420426,
      "lm_loss": 0.6376,
      "step": 13240,
      "vm_loss": 0.0983
    },
    {
      "epoch": 0.9714133734420426,
      "lm_loss": 0.5586,
      "step": 13240,
      "vm_loss": 0.0825
    },
    {
      "epoch": 0.9714133734420426,
      "lm_loss": 0.4292,
      "step": 13240,
      "vm_loss": 0.1576
    },
    {
      "epoch": 0.9714867430321817,
      "grad_norm": 1.340032025690902,
      "learning_rate": 2.1238927882255567e-08,
      "loss": 0.7077,
      "step": 13241
    },
    {
      "epoch": 0.9715601126223209,
      "grad_norm": 1.4912877229196362,
      "learning_rate": 2.1129667028219615e-08,
      "loss": 0.7434,
      "step": 13242
    },
    {
      "epoch": 0.97163348221246,
      "grad_norm": 1.4224395567337866,
      "learning_rate": 2.1020687343321634e-08,
      "loss": 0.6807,
      "step": 13243
    },
    {
      "epoch": 0.9717068518025991,
      "grad_norm": 1.510859562239118,
      "learning_rate": 2.091198883371559e-08,
      "loss": 0.6282,
      "step": 13244
    },
    {
      "epoch": 0.9717802213927382,
      "grad_norm": 0.5604326190457416,
      "learning_rate": 2.080357150553991e-08,
      "loss": 0.4774,
      "step": 13245
    },
    {
      "epoch": 0.9718535909828774,
      "grad_norm": 0.5432902391390652,
      "learning_rate": 2.069543536491747e-08,
      "loss": 0.4673,
      "step": 13246
    },
    {
      "epoch": 0.9719269605730165,
      "grad_norm": 1.5328704102460562,
      "learning_rate": 2.0587580417955054e-08,
      "loss": 0.7383,
      "step": 13247
    },
    {
      "epoch": 0.9720003301631557,
      "grad_norm": 1.4390524070912485,
      "learning_rate": 2.048000667074279e-08,
      "loss": 0.6438,
      "step": 13248
    },
    {
      "epoch": 0.9720003301631557,
      "lm_loss": 1.416,
      "step": 13248,
      "vm_loss": 0.1824
    },
    {
      "epoch": 0.9720003301631557,
      "lm_loss": 1.3558,
      "step": 13248,
      "vm_loss": 0.2092
    },
    {
      "epoch": 0.9720003301631557,
      "lm_loss": 0.4023,
      "step": 13248,
      "vm_loss": 0.0679
    },
    {
      "epoch": 0.9720003301631557,
      "lm_loss": 1.5556,
      "step": 13248,
      "vm_loss": 0.1626
    },
    {
      "epoch": 0.9720003301631557,
      "lm_loss": 0.3359,
      "step": 13248,
      "vm_loss": 0.14
    },
    {
      "epoch": 0.9720003301631557,
      "lm_loss": 0.1234,
      "step": 13248,
      "vm_loss": 0.116
    },
    {
      "epoch": 0.9720003301631557,
      "lm_loss": 0.0247,
      "step": 13248,
      "vm_loss": 0.0537
    },
    {
      "epoch": 0.9720003301631557,
      "lm_loss": 1.1213,
      "step": 13248,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.9720736997532947,
      "grad_norm": 1.423432069709963,
      "learning_rate": 2.0372714129356375e-08,
      "loss": 0.6925,
      "step": 13249
    },
    {
      "epoch": 0.9721470693434339,
      "grad_norm": 0.5332577687404813,
      "learning_rate": 2.026570279985485e-08,
      "loss": 0.4707,
      "step": 13250
    },
    {
      "epoch": 0.972220438933573,
      "grad_norm": 1.5378448360583932,
      "learning_rate": 2.0158972688281152e-08,
      "loss": 0.6878,
      "step": 13251
    },
    {
      "epoch": 0.9722938085237122,
      "grad_norm": 1.4943385725768923,
      "learning_rate": 2.005252380066214e-08,
      "loss": 0.7071,
      "step": 13252
    },
    {
      "epoch": 0.9723671781138513,
      "grad_norm": 1.4512939964043106,
      "learning_rate": 1.9946356143010214e-08,
      "loss": 0.834,
      "step": 13253
    },
    {
      "epoch": 0.9724405477039904,
      "grad_norm": 0.5557588835646624,
      "learning_rate": 1.9840469721320587e-08,
      "loss": 0.4905,
      "step": 13254
    },
    {
      "epoch": 0.9725139172941295,
      "grad_norm": 1.2869450602273076,
      "learning_rate": 1.9734864541572363e-08,
      "loss": 0.6692,
      "step": 13255
    },
    {
      "epoch": 0.9725872868842687,
      "grad_norm": 1.579279931781651,
      "learning_rate": 1.9629540609730212e-08,
      "loss": 0.7755,
      "step": 13256
    },
    {
      "epoch": 0.9725872868842687,
      "lm_loss": 0.3645,
      "step": 13256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9725872868842687,
      "lm_loss": 0.5622,
      "step": 13256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9725872868842687,
      "lm_loss": 0.4748,
      "step": 13256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9725872868842687,
      "lm_loss": 0.4509,
      "step": 13256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9725872868842687,
      "lm_loss": 0.4017,
      "step": 13256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9725872868842687,
      "lm_loss": 0.5894,
      "step": 13256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9725872868842687,
      "lm_loss": 0.644,
      "step": 13256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9725872868842687,
      "lm_loss": 0.4085,
      "step": 13256,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9726606564744078,
      "grad_norm": 0.548126301699472,
      "learning_rate": 1.952449793174105e-08,
      "loss": 0.4797,
      "step": 13257
    },
    {
      "epoch": 0.9727340260645468,
      "grad_norm": 0.5379475440634023,
      "learning_rate": 1.94197365135379e-08,
      "loss": 0.4844,
      "step": 13258
    },
    {
      "epoch": 0.972807395654686,
      "grad_norm": 1.453238550517806,
      "learning_rate": 1.9315256361036038e-08,
      "loss": 0.8084,
      "step": 13259
    },
    {
      "epoch": 0.9728807652448251,
      "grad_norm": 0.5255074911686823,
      "learning_rate": 1.9211057480136296e-08,
      "loss": 0.46,
      "step": 13260
    },
    {
      "epoch": 0.9729541348349643,
      "grad_norm": 0.5504684889760436,
      "learning_rate": 1.9107139876722857e-08,
      "loss": 0.4611,
      "step": 13261
    },
    {
      "epoch": 0.9730275044251034,
      "grad_norm": 0.5353609209637252,
      "learning_rate": 1.9003503556663806e-08,
      "loss": 0.4668,
      "step": 13262
    },
    {
      "epoch": 0.9731008740152425,
      "grad_norm": 1.483718147478028,
      "learning_rate": 1.890014852581279e-08,
      "loss": 0.781,
      "step": 13263
    },
    {
      "epoch": 0.9731742436053816,
      "grad_norm": 1.3740156899065512,
      "learning_rate": 1.87970747900057e-08,
      "loss": 0.6079,
      "step": 13264
    },
    {
      "epoch": 0.9731742436053816,
      "lm_loss": 0.5791,
      "step": 13264,
      "vm_loss": 0.022
    },
    {
      "epoch": 0.9731742436053816,
      "lm_loss": 1.5973,
      "step": 13264,
      "vm_loss": 0.1826
    },
    {
      "epoch": 0.9731742436053816,
      "lm_loss": 0.5188,
      "step": 13264,
      "vm_loss": 0.0897
    },
    {
      "epoch": 0.9731742436053816,
      "lm_loss": 0.033,
      "step": 13264,
      "vm_loss": 0.0743
    },
    {
      "epoch": 0.9731742436053816,
      "lm_loss": 0.8072,
      "step": 13264,
      "vm_loss": 0.1436
    },
    {
      "epoch": 0.9731742436053816,
      "lm_loss": 0.2684,
      "step": 13264,
      "vm_loss": 0.0573
    },
    {
      "epoch": 0.9731742436053816,
      "lm_loss": 1.0147,
      "step": 13264,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.9731742436053816,
      "lm_loss": 0.0694,
      "step": 13264,
      "vm_loss": 0.0972
    },
    {
      "epoch": 0.9732476131955208,
      "grad_norm": 1.4735411269858114,
      "learning_rate": 1.8694282355063432e-08,
      "loss": 0.7728,
      "step": 13265
    },
    {
      "epoch": 0.9733209827856599,
      "grad_norm": 1.614633267559733,
      "learning_rate": 1.859177122679079e-08,
      "loss": 0.6878,
      "step": 13266
    },
    {
      "epoch": 0.9733943523757991,
      "grad_norm": 1.5972410328723063,
      "learning_rate": 1.8489541410977585e-08,
      "loss": 0.7707,
      "step": 13267
    },
    {
      "epoch": 0.9734677219659381,
      "grad_norm": 1.396875441068407,
      "learning_rate": 1.838759291339587e-08,
      "loss": 0.8339,
      "step": 13268
    },
    {
      "epoch": 0.9735410915560773,
      "grad_norm": 0.5430206198368548,
      "learning_rate": 1.8285925739803812e-08,
      "loss": 0.4695,
      "step": 13269
    },
    {
      "epoch": 0.9736144611462164,
      "grad_norm": 1.5178150572071414,
      "learning_rate": 1.818453989594293e-08,
      "loss": 0.6736,
      "step": 13270
    },
    {
      "epoch": 0.9736878307363556,
      "grad_norm": 0.5325337837166476,
      "learning_rate": 1.8083435387538094e-08,
      "loss": 0.4736,
      "step": 13271
    },
    {
      "epoch": 0.9737612003264947,
      "grad_norm": 1.4746594166648364,
      "learning_rate": 1.798261222029918e-08,
      "loss": 0.6399,
      "step": 13272
    },
    {
      "epoch": 0.9737612003264947,
      "lm_loss": 1.6113,
      "step": 13272,
      "vm_loss": 0.0162
    },
    {
      "epoch": 0.9737612003264947,
      "lm_loss": 0.4949,
      "step": 13272,
      "vm_loss": 0.1169
    },
    {
      "epoch": 0.9737612003264947,
      "lm_loss": 0.338,
      "step": 13272,
      "vm_loss": 0.0683
    },
    {
      "epoch": 0.9737612003264947,
      "lm_loss": 0.0196,
      "step": 13272,
      "vm_loss": 0.1777
    },
    {
      "epoch": 0.9737612003264947,
      "lm_loss": 0.8175,
      "step": 13272,
      "vm_loss": 0.1563
    },
    {
      "epoch": 0.9737612003264947,
      "lm_loss": 0.2854,
      "step": 13272,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.9737612003264947,
      "lm_loss": 0.6021,
      "step": 13272,
      "vm_loss": 0.1615
    },
    {
      "epoch": 0.9737612003264947,
      "lm_loss": 0.9006,
      "step": 13272,
      "vm_loss": 0.2284
    },
    {
      "epoch": 0.9738345699166338,
      "grad_norm": 1.3732441741556476,
      "learning_rate": 1.7882070399919403e-08,
      "loss": 0.6867,
      "step": 13273
    },
    {
      "epoch": 0.9739079395067729,
      "grad_norm": 1.4670795332685567,
      "learning_rate": 1.7781809932077565e-08,
      "loss": 0.6877,
      "step": 13274
    },
    {
      "epoch": 0.973981309096912,
      "grad_norm": 1.592118426919964,
      "learning_rate": 1.768183082243524e-08,
      "loss": 0.8164,
      "step": 13275
    },
    {
      "epoch": 0.9740546786870512,
      "grad_norm": 1.4367636958615604,
      "learning_rate": 1.7582133076638473e-08,
      "loss": 0.7154,
      "step": 13276
    },
    {
      "epoch": 0.9741280482771904,
      "grad_norm": 0.5484406192357082,
      "learning_rate": 1.7482716700317203e-08,
      "loss": 0.4655,
      "step": 13277
    },
    {
      "epoch": 0.9742014178673294,
      "grad_norm": 1.3592796337539472,
      "learning_rate": 1.738358169908583e-08,
      "loss": 0.6854,
      "step": 13278
    },
    {
      "epoch": 0.9742747874574685,
      "grad_norm": 1.5751087514630264,
      "learning_rate": 1.72847280785432e-08,
      "loss": 0.7071,
      "step": 13279
    },
    {
      "epoch": 0.9743481570476077,
      "grad_norm": 1.4317702269638093,
      "learning_rate": 1.7186155844270967e-08,
      "loss": 0.6662,
      "step": 13280
    },
    {
      "epoch": 0.9743481570476077,
      "lm_loss": 0.4758,
      "step": 13280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9743481570476077,
      "lm_loss": 0.3835,
      "step": 13280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9743481570476077,
      "lm_loss": 0.5095,
      "step": 13280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9743481570476077,
      "lm_loss": 0.3655,
      "step": 13280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9743481570476077,
      "lm_loss": 0.4507,
      "step": 13280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9743481570476077,
      "lm_loss": 0.4107,
      "step": 13280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9743481570476077,
      "lm_loss": 0.4171,
      "step": 13280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9743481570476077,
      "lm_loss": 0.5274,
      "step": 13280,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9744215266377468,
      "grad_norm": 0.5624920885204365,
      "learning_rate": 1.70878650018369e-08,
      "loss": 0.5,
      "step": 13281
    },
    {
      "epoch": 0.974494896227886,
      "grad_norm": 1.4859626504982453,
      "learning_rate": 1.6989855556790446e-08,
      "loss": 0.7069,
      "step": 13282
    },
    {
      "epoch": 0.974568265818025,
      "grad_norm": 1.4421510970693463,
      "learning_rate": 1.689212751466718e-08,
      "loss": 0.7378,
      "step": 13283
    },
    {
      "epoch": 0.9746416354081642,
      "grad_norm": 0.5409136449127065,
      "learning_rate": 1.6794680880986015e-08,
      "loss": 0.4714,
      "step": 13284
    },
    {
      "epoch": 0.9747150049983033,
      "grad_norm": 1.372511291061318,
      "learning_rate": 1.6697515661250328e-08,
      "loss": 0.5533,
      "step": 13285
    },
    {
      "epoch": 0.9747883745884425,
      "grad_norm": 0.541799329788417,
      "learning_rate": 1.6600631860946293e-08,
      "loss": 0.488,
      "step": 13286
    },
    {
      "epoch": 0.9748617441785816,
      "grad_norm": 1.4483814283630145,
      "learning_rate": 1.650402948554619e-08,
      "loss": 0.8025,
      "step": 13287
    },
    {
      "epoch": 0.9749351137687207,
      "grad_norm": 0.5420829309856827,
      "learning_rate": 1.640770854050455e-08,
      "loss": 0.482,
      "step": 13288
    },
    {
      "epoch": 0.9749351137687207,
      "lm_loss": 0.5156,
      "step": 13288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9749351137687207,
      "lm_loss": 0.4729,
      "step": 13288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9749351137687207,
      "lm_loss": 0.4312,
      "step": 13288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9749351137687207,
      "lm_loss": 0.5186,
      "step": 13288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9749351137687207,
      "lm_loss": 0.5447,
      "step": 13288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9749351137687207,
      "lm_loss": 0.3352,
      "step": 13288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9749351137687207,
      "lm_loss": 0.367,
      "step": 13288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9749351137687207,
      "lm_loss": 0.5777,
      "step": 13288,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9750084833588598,
      "grad_norm": 0.5362973879058164,
      "learning_rate": 1.631166903126147e-08,
      "loss": 0.472,
      "step": 13289
    },
    {
      "epoch": 0.975081852948999,
      "grad_norm": 0.5537529451847938,
      "learning_rate": 1.6215910963240377e-08,
      "loss": 0.4876,
      "step": 13290
    },
    {
      "epoch": 0.9751552225391381,
      "grad_norm": 1.4643323150621166,
      "learning_rate": 1.612043434184862e-08,
      "loss": 0.648,
      "step": 13291
    },
    {
      "epoch": 0.9752285921292773,
      "grad_norm": 1.586319086899204,
      "learning_rate": 1.6025239172478556e-08,
      "loss": 0.7395,
      "step": 13292
    },
    {
      "epoch": 0.9753019617194163,
      "grad_norm": 1.431932728307194,
      "learning_rate": 1.593032546050588e-08,
      "loss": 0.6489,
      "step": 13293
    },
    {
      "epoch": 0.9753753313095554,
      "grad_norm": 0.543323247152903,
      "learning_rate": 1.583569321129075e-08,
      "loss": 0.4926,
      "step": 13294
    },
    {
      "epoch": 0.9754487008996946,
      "grad_norm": 0.5267957141696129,
      "learning_rate": 1.5741342430176666e-08,
      "loss": 0.4621,
      "step": 13295
    },
    {
      "epoch": 0.9755220704898337,
      "grad_norm": 1.622770500296671,
      "learning_rate": 1.5647273122492702e-08,
      "loss": 0.6914,
      "step": 13296
    },
    {
      "epoch": 0.9755220704898337,
      "lm_loss": 1.3205,
      "step": 13296,
      "vm_loss": 0.0789
    },
    {
      "epoch": 0.9755220704898337,
      "lm_loss": 0.4947,
      "step": 13296,
      "vm_loss": 0.1155
    },
    {
      "epoch": 0.9755220704898337,
      "lm_loss": 1.0129,
      "step": 13296,
      "vm_loss": 0.0783
    },
    {
      "epoch": 0.9755220704898337,
      "lm_loss": 0.1394,
      "step": 13296,
      "vm_loss": 0.1048
    },
    {
      "epoch": 0.9755220704898337,
      "lm_loss": 1.2588,
      "step": 13296,
      "vm_loss": 0.2095
    },
    {
      "epoch": 0.9755220704898337,
      "lm_loss": 0.5688,
      "step": 13296,
      "vm_loss": 0.0763
    },
    {
      "epoch": 0.9755220704898337,
      "lm_loss": 0.7877,
      "step": 13296,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.9755220704898337,
      "lm_loss": 1.4429,
      "step": 13296,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.9755954400799729,
      "grad_norm": 1.5919776263009997,
      "learning_rate": 1.555348529355072e-08,
      "loss": 0.868,
      "step": 13297
    },
    {
      "epoch": 0.9756688096701119,
      "grad_norm": 0.5424575447985648,
      "learning_rate": 1.5459978948647037e-08,
      "loss": 0.4749,
      "step": 13298
    },
    {
      "epoch": 0.9757421792602511,
      "grad_norm": 1.7069696233832166,
      "learning_rate": 1.536675409306243e-08,
      "loss": 0.733,
      "step": 13299
    },
    {
      "epoch": 0.9758155488503902,
      "grad_norm": 1.3417620426182917,
      "learning_rate": 1.5273810732061023e-08,
      "loss": 0.7202,
      "step": 13300
    },
    {
      "epoch": 0.9758889184405294,
      "grad_norm": 1.445085447966436,
      "learning_rate": 1.5181148870892505e-08,
      "loss": 0.769,
      "step": 13301
    },
    {
      "epoch": 0.9759622880306685,
      "grad_norm": 1.5541165531989851,
      "learning_rate": 1.50887685147888e-08,
      "loss": 0.6948,
      "step": 13302
    },
    {
      "epoch": 0.9760356576208076,
      "grad_norm": 0.5499692479930457,
      "learning_rate": 1.49966696689674e-08,
      "loss": 0.4751,
      "step": 13303
    },
    {
      "epoch": 0.9761090272109467,
      "grad_norm": 1.4796541529266807,
      "learning_rate": 1.4904852338629706e-08,
      "loss": 0.7595,
      "step": 13304
    },
    {
      "epoch": 0.9761090272109467,
      "lm_loss": 1.2361,
      "step": 13304,
      "vm_loss": 0.0657
    },
    {
      "epoch": 0.9761090272109467,
      "lm_loss": 1.038,
      "step": 13304,
      "vm_loss": 0.197
    },
    {
      "epoch": 0.9761090272109467,
      "lm_loss": 0.34,
      "step": 13304,
      "vm_loss": 0.1501
    },
    {
      "epoch": 0.9761090272109467,
      "lm_loss": 0.6134,
      "step": 13304,
      "vm_loss": 0.168
    },
    {
      "epoch": 0.9761090272109467,
      "lm_loss": 0.6176,
      "step": 13304,
      "vm_loss": 0.1414
    },
    {
      "epoch": 0.9761090272109467,
      "lm_loss": 0.573,
      "step": 13304,
      "vm_loss": 0.2554
    },
    {
      "epoch": 0.9761090272109467,
      "lm_loss": 0.8192,
      "step": 13304,
      "vm_loss": 0.125
    },
    {
      "epoch": 0.9761090272109467,
      "lm_loss": 0.8317,
      "step": 13304,
      "vm_loss": 0.1101
    },
    {
      "epoch": 0.9761823968010859,
      "grad_norm": 1.5705220907718895,
      "learning_rate": 1.4813316528959898e-08,
      "loss": 0.7493,
      "step": 13305
    },
    {
      "epoch": 0.976255766391225,
      "grad_norm": 1.5308914281926862,
      "learning_rate": 1.4722062245127733e-08,
      "loss": 0.7166,
      "step": 13306
    },
    {
      "epoch": 0.9763291359813642,
      "grad_norm": 1.7143132062089188,
      "learning_rate": 1.4631089492286866e-08,
      "loss": 0.7825,
      "step": 13307
    },
    {
      "epoch": 0.9764025055715032,
      "grad_norm": 1.4592558909354247,
      "learning_rate": 1.4540398275573741e-08,
      "loss": 0.6843,
      "step": 13308
    },
    {
      "epoch": 0.9764758751616424,
      "grad_norm": 1.4716125168711705,
      "learning_rate": 1.4449988600111486e-08,
      "loss": 0.7592,
      "step": 13309
    },
    {
      "epoch": 0.9765492447517815,
      "grad_norm": 1.7497455129597523,
      "learning_rate": 1.435986047100435e-08,
      "loss": 0.6691,
      "step": 13310
    },
    {
      "epoch": 0.9766226143419207,
      "grad_norm": 1.6933784162222485,
      "learning_rate": 1.4270013893342704e-08,
      "loss": 0.8384,
      "step": 13311
    },
    {
      "epoch": 0.9766959839320598,
      "grad_norm": 1.3795246745521657,
      "learning_rate": 1.418044887220027e-08,
      "loss": 0.7156,
      "step": 13312
    },
    {
      "epoch": 0.9766959839320598,
      "lm_loss": 1.2177,
      "step": 13312,
      "vm_loss": 0.0601
    },
    {
      "epoch": 0.9766959839320598,
      "lm_loss": 1.0978,
      "step": 13312,
      "vm_loss": 0.1562
    },
    {
      "epoch": 0.9766959839320598,
      "lm_loss": 0.1092,
      "step": 13312,
      "vm_loss": 0.1757
    },
    {
      "epoch": 0.9766959839320598,
      "lm_loss": 0.5068,
      "step": 13312,
      "vm_loss": 0.2062
    },
    {
      "epoch": 0.9766959839320598,
      "lm_loss": 0.4598,
      "step": 13312,
      "vm_loss": 0.041
    },
    {
      "epoch": 0.9766959839320598,
      "lm_loss": 0.1677,
      "step": 13312,
      "vm_loss": 0.0753
    },
    {
      "epoch": 0.9766959839320598,
      "lm_loss": 0.5879,
      "step": 13312,
      "vm_loss": 0.2474
    },
    {
      "epoch": 0.9766959839320598,
      "lm_loss": 0.088,
      "step": 13312,
      "vm_loss": 0.0655
    },
    {
      "epoch": 0.9767693535221988,
      "grad_norm": 1.5007354433714475,
      "learning_rate": 1.4091165412635222e-08,
      "loss": 0.7274,
      "step": 13313
    },
    {
      "epoch": 0.976842723112338,
      "grad_norm": 1.5916763329113144,
      "learning_rate": 1.4002163519689083e-08,
      "loss": 0.6659,
      "step": 13314
    },
    {
      "epoch": 0.9769160927024771,
      "grad_norm": 1.746935220687597,
      "learning_rate": 1.3913443198388943e-08,
      "loss": 0.8202,
      "step": 13315
    },
    {
      "epoch": 0.9769894622926163,
      "grad_norm": 1.5596409573184917,
      "learning_rate": 1.3825004453744129e-08,
      "loss": 0.7681,
      "step": 13316
    },
    {
      "epoch": 0.9770628318827554,
      "grad_norm": 1.5750549507006633,
      "learning_rate": 1.3736847290749533e-08,
      "loss": 0.6745,
      "step": 13317
    },
    {
      "epoch": 0.9771362014728945,
      "grad_norm": 1.6415098200171787,
      "learning_rate": 1.3648971714383397e-08,
      "loss": 0.7439,
      "step": 13318
    },
    {
      "epoch": 0.9772095710630336,
      "grad_norm": 1.5150874755994852,
      "learning_rate": 1.3561377729608417e-08,
      "loss": 0.6935,
      "step": 13319
    },
    {
      "epoch": 0.9772829406531728,
      "grad_norm": 0.54273346992259,
      "learning_rate": 1.347406534137119e-08,
      "loss": 0.4624,
      "step": 13320
    },
    {
      "epoch": 0.9772829406531728,
      "lm_loss": 1.0447,
      "step": 13320,
      "vm_loss": 0.0557
    },
    {
      "epoch": 0.9772829406531728,
      "lm_loss": 1.1539,
      "step": 13320,
      "vm_loss": 0.067
    },
    {
      "epoch": 0.9772829406531728,
      "lm_loss": 0.7642,
      "step": 13320,
      "vm_loss": 0.1278
    },
    {
      "epoch": 0.9772829406531728,
      "lm_loss": 0.0025,
      "step": 13320,
      "vm_loss": 0.0908
    },
    {
      "epoch": 0.9772829406531728,
      "lm_loss": 0.7278,
      "step": 13320,
      "vm_loss": 0.1335
    },
    {
      "epoch": 0.9772829406531728,
      "lm_loss": 0.8498,
      "step": 13320,
      "vm_loss": 0.1156
    },
    {
      "epoch": 0.9772829406531728,
      "lm_loss": 0.7375,
      "step": 13320,
      "vm_loss": 0.0752
    },
    {
      "epoch": 0.9772829406531728,
      "lm_loss": 0.5499,
      "step": 13320,
      "vm_loss": 0.0694
    },
    {
      "epoch": 0.9773563102433119,
      "grad_norm": 1.5133021150117227,
      "learning_rate": 1.3387034554601663e-08,
      "loss": 0.7394,
      "step": 13321
    },
    {
      "epoch": 0.9774296798334511,
      "grad_norm": 1.400794189154413,
      "learning_rate": 1.330028537421646e-08,
      "loss": 0.6778,
      "step": 13322
    },
    {
      "epoch": 0.9775030494235901,
      "grad_norm": 1.767055256105749,
      "learning_rate": 1.3213817805112772e-08,
      "loss": 0.681,
      "step": 13323
    },
    {
      "epoch": 0.9775764190137293,
      "grad_norm": 1.4905975511577572,
      "learning_rate": 1.3127631852175026e-08,
      "loss": 0.7432,
      "step": 13324
    },
    {
      "epoch": 0.9776497886038684,
      "grad_norm": 0.5377594695722094,
      "learning_rate": 1.3041727520269332e-08,
      "loss": 0.4769,
      "step": 13325
    },
    {
      "epoch": 0.9777231581940076,
      "grad_norm": 0.542451747043003,
      "learning_rate": 1.2956104814247361e-08,
      "loss": 0.4879,
      "step": 13326
    },
    {
      "epoch": 0.9777965277841467,
      "grad_norm": 0.5519003561888921,
      "learning_rate": 1.2870763738944137e-08,
      "loss": 0.4811,
      "step": 13327
    },
    {
      "epoch": 0.9778698973742858,
      "grad_norm": 1.5170223259960134,
      "learning_rate": 1.2785704299179135e-08,
      "loss": 0.6968,
      "step": 13328
    },
    {
      "epoch": 0.9778698973742858,
      "lm_loss": 1.3163,
      "step": 13328,
      "vm_loss": 0.0799
    },
    {
      "epoch": 0.9778698973742858,
      "lm_loss": 1.3477,
      "step": 13328,
      "vm_loss": 0.1057
    },
    {
      "epoch": 0.9778698973742858,
      "lm_loss": 0.4355,
      "step": 13328,
      "vm_loss": 0.1386
    },
    {
      "epoch": 0.9778698973742858,
      "lm_loss": 0.3728,
      "step": 13328,
      "vm_loss": 0.1653
    },
    {
      "epoch": 0.9778698973742858,
      "lm_loss": 0.6127,
      "step": 13328,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.9778698973742858,
      "lm_loss": 0.6645,
      "step": 13328,
      "vm_loss": 0.0991
    },
    {
      "epoch": 0.9778698973742858,
      "lm_loss": 0.8187,
      "step": 13328,
      "vm_loss": 0.0804
    },
    {
      "epoch": 0.9778698973742858,
      "lm_loss": 1.3868,
      "step": 13328,
      "vm_loss": 0.1479
    },
    {
      "epoch": 0.9779432669644249,
      "grad_norm": 1.6029208109540136,
      "learning_rate": 1.2700926499756295e-08,
      "loss": 0.8141,
      "step": 13329
    },
    {
      "epoch": 0.9780166365545641,
      "grad_norm": 1.487011510472885,
      "learning_rate": 1.2616430345462894e-08,
      "loss": 0.9539,
      "step": 13330
    },
    {
      "epoch": 0.9780900061447032,
      "grad_norm": 1.5493561558750164,
      "learning_rate": 1.2532215841071227e-08,
      "loss": 0.803,
      "step": 13331
    },
    {
      "epoch": 0.9781633757348424,
      "grad_norm": 1.5947078723624317,
      "learning_rate": 1.2448282991335825e-08,
      "loss": 0.7038,
      "step": 13332
    },
    {
      "epoch": 0.9782367453249814,
      "grad_norm": 1.61756788294662,
      "learning_rate": 1.2364631800997894e-08,
      "loss": 0.7729,
      "step": 13333
    },
    {
      "epoch": 0.9783101149151205,
      "grad_norm": 1.6436000573751197,
      "learning_rate": 1.2281262274780881e-08,
      "loss": 0.6748,
      "step": 13334
    },
    {
      "epoch": 0.9783834845052597,
      "grad_norm": 0.5353512077440788,
      "learning_rate": 1.2198174417392684e-08,
      "loss": 0.4685,
      "step": 13335
    },
    {
      "epoch": 0.9784568540953988,
      "grad_norm": 1.4573638251009726,
      "learning_rate": 1.211536823352566e-08,
      "loss": 0.665,
      "step": 13336
    },
    {
      "epoch": 0.9784568540953988,
      "lm_loss": 1.4428,
      "step": 13336,
      "vm_loss": 0.0701
    },
    {
      "epoch": 0.9784568540953988,
      "lm_loss": 0.0623,
      "step": 13336,
      "vm_loss": 0.1344
    },
    {
      "epoch": 0.9784568540953988,
      "lm_loss": 0.2818,
      "step": 13336,
      "vm_loss": 0.1283
    },
    {
      "epoch": 0.9784568540953988,
      "lm_loss": 0.0893,
      "step": 13336,
      "vm_loss": 0.0932
    },
    {
      "epoch": 0.9784568540953988,
      "lm_loss": 0.6228,
      "step": 13336,
      "vm_loss": 0.1064
    },
    {
      "epoch": 0.9784568540953988,
      "lm_loss": 0.6444,
      "step": 13336,
      "vm_loss": 0.1343
    },
    {
      "epoch": 0.9784568540953988,
      "lm_loss": 0.7176,
      "step": 13336,
      "vm_loss": 0.1521
    },
    {
      "epoch": 0.9784568540953988,
      "lm_loss": 0.0025,
      "step": 13336,
      "vm_loss": 0.0885
    },
    {
      "epoch": 0.978530223685538,
      "grad_norm": 1.4280726914198552,
      "learning_rate": 1.2032843727856069e-08,
      "loss": 0.6575,
      "step": 13337
    },
    {
      "epoch": 0.978603593275677,
      "grad_norm": 1.5589731754873277,
      "learning_rate": 1.1950600905044074e-08,
      "loss": 0.7578,
      "step": 13338
    },
    {
      "epoch": 0.9786769628658162,
      "grad_norm": 1.552914270338762,
      "learning_rate": 1.1868639769734847e-08,
      "loss": 0.6816,
      "step": 13339
    },
    {
      "epoch": 0.9787503324559553,
      "grad_norm": 1.5884296512772118,
      "learning_rate": 1.1786960326555797e-08,
      "loss": 0.7078,
      "step": 13340
    },
    {
      "epoch": 0.9788237020460945,
      "grad_norm": 1.5199608259669946,
      "learning_rate": 1.1705562580120455e-08,
      "loss": 0.6864,
      "step": 13341
    },
    {
      "epoch": 0.9788970716362335,
      "grad_norm": 1.5013215822241572,
      "learning_rate": 1.1624446535025701e-08,
      "loss": 0.6611,
      "step": 13342
    },
    {
      "epoch": 0.9789704412263727,
      "grad_norm": 1.5125609542739407,
      "learning_rate": 1.1543612195851206e-08,
      "loss": 0.7743,
      "step": 13343
    },
    {
      "epoch": 0.9790438108165118,
      "grad_norm": 1.4772973891539085,
      "learning_rate": 1.1463059567163314e-08,
      "loss": 0.7506,
      "step": 13344
    },
    {
      "epoch": 0.9790438108165118,
      "lm_loss": 1.387,
      "step": 13344,
      "vm_loss": 0.0508
    },
    {
      "epoch": 0.9790438108165118,
      "lm_loss": 0.8834,
      "step": 13344,
      "vm_loss": 0.0957
    },
    {
      "epoch": 0.9790438108165118,
      "lm_loss": 0.6988,
      "step": 13344,
      "vm_loss": 0.0906
    },
    {
      "epoch": 0.9790438108165118,
      "lm_loss": 0.3095,
      "step": 13344,
      "vm_loss": 0.0905
    },
    {
      "epoch": 0.9790438108165118,
      "lm_loss": 0.814,
      "step": 13344,
      "vm_loss": 0.158
    },
    {
      "epoch": 0.9790438108165118,
      "lm_loss": 0.704,
      "step": 13344,
      "vm_loss": 0.155
    },
    {
      "epoch": 0.9790438108165118,
      "lm_loss": 0.7941,
      "step": 13344,
      "vm_loss": 0.0459
    },
    {
      "epoch": 0.9790438108165118,
      "lm_loss": 0.081,
      "step": 13344,
      "vm_loss": 0.0614
    },
    {
      "epoch": 0.979117180406651,
      "grad_norm": 1.6243117750497917,
      "learning_rate": 1.1382788653510058e-08,
      "loss": 0.7666,
      "step": 13345
    },
    {
      "epoch": 0.9791905499967901,
      "grad_norm": 0.5396889052911197,
      "learning_rate": 1.1302799459425028e-08,
      "loss": 0.4674,
      "step": 13346
    },
    {
      "epoch": 0.9792639195869292,
      "grad_norm": 1.4263487798797065,
      "learning_rate": 1.1223091989425173e-08,
      "loss": 0.7039,
      "step": 13347
    },
    {
      "epoch": 0.9793372891770683,
      "grad_norm": 1.4685317185707234,
      "learning_rate": 1.1143666248011331e-08,
      "loss": 0.7273,
      "step": 13348
    },
    {
      "epoch": 0.9794106587672075,
      "grad_norm": 0.5474094179243633,
      "learning_rate": 1.1064522239669916e-08,
      "loss": 0.4709,
      "step": 13349
    },
    {
      "epoch": 0.9794840283573466,
      "grad_norm": 1.4172673900556154,
      "learning_rate": 1.0985659968869578e-08,
      "loss": 0.5857,
      "step": 13350
    },
    {
      "epoch": 0.9795573979474858,
      "grad_norm": 0.5548998636832633,
      "learning_rate": 1.0907079440063972e-08,
      "loss": 0.4759,
      "step": 13351
    },
    {
      "epoch": 0.9796307675376248,
      "grad_norm": 1.6941529759369887,
      "learning_rate": 1.0828780657691217e-08,
      "loss": 0.7671,
      "step": 13352
    },
    {
      "epoch": 0.9796307675376248,
      "lm_loss": 0.1607,
      "step": 13352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9796307675376248,
      "lm_loss": 0.3975,
      "step": 13352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9796307675376248,
      "lm_loss": 0.5554,
      "step": 13352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9796307675376248,
      "lm_loss": 0.4516,
      "step": 13352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9796307675376248,
      "lm_loss": 0.5775,
      "step": 13352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9796307675376248,
      "lm_loss": 0.4951,
      "step": 13352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9796307675376248,
      "lm_loss": 0.4444,
      "step": 13352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9796307675376248,
      "lm_loss": 0.53,
      "step": 13352,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9797041371277639,
      "grad_norm": 0.5374779991477724,
      "learning_rate": 1.0750763626172778e-08,
      "loss": 0.4722,
      "step": 13353
    },
    {
      "epoch": 0.9797775067179031,
      "grad_norm": 1.314547078189994,
      "learning_rate": 1.0673028349914017e-08,
      "loss": 0.5775,
      "step": 13354
    },
    {
      "epoch": 0.9798508763080422,
      "grad_norm": 1.4093122052735758,
      "learning_rate": 1.059557483330531e-08,
      "loss": 0.6772,
      "step": 13355
    },
    {
      "epoch": 0.9799242458981814,
      "grad_norm": 1.587712416333789,
      "learning_rate": 1.0518403080720385e-08,
      "loss": 0.7936,
      "step": 13356
    },
    {
      "epoch": 0.9799976154883204,
      "grad_norm": 0.5305645054553286,
      "learning_rate": 1.0441513096517974e-08,
      "loss": 0.471,
      "step": 13357
    },
    {
      "epoch": 0.9800709850784596,
      "grad_norm": 1.5009968780303513,
      "learning_rate": 1.036490488503905e-08,
      "loss": 0.7207,
      "step": 13358
    },
    {
      "epoch": 0.9801443546685987,
      "grad_norm": 0.5474617705412667,
      "learning_rate": 1.0288578450611264e-08,
      "loss": 0.4776,
      "step": 13359
    },
    {
      "epoch": 0.9802177242587379,
      "grad_norm": 0.5354100295985184,
      "learning_rate": 1.0212533797543944e-08,
      "loss": 0.4754,
      "step": 13360
    },
    {
      "epoch": 0.9802177242587379,
      "lm_loss": 1.3996,
      "step": 13360,
      "vm_loss": 0.089
    },
    {
      "epoch": 0.9802177242587379,
      "lm_loss": 0.6156,
      "step": 13360,
      "vm_loss": 0.1351
    },
    {
      "epoch": 0.9802177242587379,
      "lm_loss": 0.5939,
      "step": 13360,
      "vm_loss": 0.1481
    },
    {
      "epoch": 0.9802177242587379,
      "lm_loss": 0.1072,
      "step": 13360,
      "vm_loss": 0.1678
    },
    {
      "epoch": 0.9802177242587379,
      "lm_loss": 0.6582,
      "step": 13360,
      "vm_loss": 0.1767
    },
    {
      "epoch": 0.9802177242587379,
      "lm_loss": 0.1575,
      "step": 13360,
      "vm_loss": 0.0803
    },
    {
      "epoch": 0.9802177242587379,
      "lm_loss": 0.2315,
      "step": 13360,
      "vm_loss": 0.0782
    },
    {
      "epoch": 0.9802177242587379,
      "lm_loss": 0.0382,
      "step": 13360,
      "vm_loss": 0.2069
    },
    {
      "epoch": 0.980291093848877,
      "grad_norm": 1.384288089889383,
      "learning_rate": 1.0136770930131435e-08,
      "loss": 0.6738,
      "step": 13361
    },
    {
      "epoch": 0.9803644634390161,
      "grad_norm": 1.3825288272949052,
      "learning_rate": 1.006128985265309e-08,
      "loss": 0.6902,
      "step": 13362
    },
    {
      "epoch": 0.9804378330291552,
      "grad_norm": 1.7291143961794757,
      "learning_rate": 9.986090569371054e-09,
      "loss": 0.7443,
      "step": 13363
    },
    {
      "epoch": 0.9805112026192944,
      "grad_norm": 1.4947610104713138,
      "learning_rate": 9.911173084532488e-09,
      "loss": 0.7512,
      "step": 13364
    },
    {
      "epoch": 0.9805845722094335,
      "grad_norm": 1.4420780797560944,
      "learning_rate": 9.836537402366786e-09,
      "loss": 0.745,
      "step": 13365
    },
    {
      "epoch": 0.9806579417995727,
      "grad_norm": 1.384324585758401,
      "learning_rate": 9.762183527090574e-09,
      "loss": 0.7258,
      "step": 13366
    },
    {
      "epoch": 0.9807313113897117,
      "grad_norm": 1.4862085493556219,
      "learning_rate": 9.688111462901606e-09,
      "loss": 0.6659,
      "step": 13367
    },
    {
      "epoch": 0.9808046809798509,
      "grad_norm": 0.5190304593993653,
      "learning_rate": 9.614321213983203e-09,
      "loss": 0.4664,
      "step": 13368
    },
    {
      "epoch": 0.9808046809798509,
      "lm_loss": 1.2246,
      "step": 13368,
      "vm_loss": 0.0601
    },
    {
      "epoch": 0.9808046809798509,
      "lm_loss": 0.0512,
      "step": 13368,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.9808046809798509,
      "lm_loss": 0.3826,
      "step": 13368,
      "vm_loss": 0.0912
    },
    {
      "epoch": 0.9808046809798509,
      "lm_loss": 1.0472,
      "step": 13368,
      "vm_loss": 0.0827
    },
    {
      "epoch": 0.9808046809798509,
      "lm_loss": 0.8629,
      "step": 13368,
      "vm_loss": 0.1617
    },
    {
      "epoch": 0.9808046809798509,
      "lm_loss": 0.4274,
      "step": 13368,
      "vm_loss": 0.1754
    },
    {
      "epoch": 0.9808046809798509,
      "lm_loss": 0.1518,
      "step": 13368,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.9808046809798509,
      "lm_loss": 0.4493,
      "step": 13368,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.98087805056999,
      "grad_norm": 1.4664562677606632,
      "learning_rate": 9.54081278450314e-09,
      "loss": 0.7555,
      "step": 13369
    },
    {
      "epoch": 0.9809514201601292,
      "grad_norm": 1.542117138574877,
      "learning_rate": 9.467586178611432e-09,
      "loss": 0.6684,
      "step": 13370
    },
    {
      "epoch": 0.9810247897502683,
      "grad_norm": 0.5430460008389629,
      "learning_rate": 9.394641400444215e-09,
      "loss": 0.4629,
      "step": 13371
    },
    {
      "epoch": 0.9810981593404073,
      "grad_norm": 1.6168175427776166,
      "learning_rate": 9.321978454120418e-09,
      "loss": 0.6979,
      "step": 13372
    },
    {
      "epoch": 0.9811715289305465,
      "grad_norm": 0.5371125591226035,
      "learning_rate": 9.249597343743977e-09,
      "loss": 0.455,
      "step": 13373
    },
    {
      "epoch": 0.9812448985206856,
      "grad_norm": 1.5704545899829845,
      "learning_rate": 9.177498073401625e-09,
      "loss": 0.6854,
      "step": 13374
    },
    {
      "epoch": 0.9813182681108248,
      "grad_norm": 1.364520331625234,
      "learning_rate": 9.105680647166215e-09,
      "loss": 0.6385,
      "step": 13375
    },
    {
      "epoch": 0.9813916377009639,
      "grad_norm": 0.5441316435340182,
      "learning_rate": 9.034145069092282e-09,
      "loss": 0.4869,
      "step": 13376
    },
    {
      "epoch": 0.9813916377009639,
      "lm_loss": 1.2999,
      "step": 13376,
      "vm_loss": 0.0865
    },
    {
      "epoch": 0.9813916377009639,
      "lm_loss": 1.246,
      "step": 13376,
      "vm_loss": 0.1385
    },
    {
      "epoch": 0.9813916377009639,
      "lm_loss": 0.4803,
      "step": 13376,
      "vm_loss": 0.0519
    },
    {
      "epoch": 0.9813916377009639,
      "lm_loss": 0.5932,
      "step": 13376,
      "vm_loss": 0.1022
    },
    {
      "epoch": 0.9813916377009639,
      "lm_loss": 0.5624,
      "step": 13376,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.9813916377009639,
      "lm_loss": 0.264,
      "step": 13376,
      "vm_loss": 0.0586
    },
    {
      "epoch": 0.9813916377009639,
      "lm_loss": 0.6612,
      "step": 13376,
      "vm_loss": 0.1303
    },
    {
      "epoch": 0.9813916377009639,
      "lm_loss": 0.2779,
      "step": 13376,
      "vm_loss": 0.0512
    },
    {
      "epoch": 0.981465007291103,
      "grad_norm": 1.4884695299395896,
      "learning_rate": 8.96289134322048e-09,
      "loss": 0.6812,
      "step": 13377
    },
    {
      "epoch": 0.9815383768812421,
      "grad_norm": 0.5524962842215544,
      "learning_rate": 8.891919473573707e-09,
      "loss": 0.483,
      "step": 13378
    },
    {
      "epoch": 0.9816117464713813,
      "grad_norm": 1.6115167947406772,
      "learning_rate": 8.821229464160974e-09,
      "loss": 0.6484,
      "step": 13379
    },
    {
      "epoch": 0.9816851160615204,
      "grad_norm": 1.5227491472369081,
      "learning_rate": 8.75082131897409e-09,
      "loss": 0.7278,
      "step": 13380
    },
    {
      "epoch": 0.9817584856516596,
      "grad_norm": 1.7624965397188552,
      "learning_rate": 8.680695041988208e-09,
      "loss": 0.7218,
      "step": 13381
    },
    {
      "epoch": 0.9818318552417986,
      "grad_norm": 1.5135306244425974,
      "learning_rate": 8.610850637164603e-09,
      "loss": 0.6961,
      "step": 13382
    },
    {
      "epoch": 0.9819052248319378,
      "grad_norm": 1.5280817786215362,
      "learning_rate": 8.541288108447343e-09,
      "loss": 0.7526,
      "step": 13383
    },
    {
      "epoch": 0.9819785944220769,
      "grad_norm": 1.644287834932947,
      "learning_rate": 8.472007459764953e-09,
      "loss": 0.7282,
      "step": 13384
    },
    {
      "epoch": 0.9819785944220769,
      "lm_loss": 1.5646,
      "step": 13384,
      "vm_loss": 0.0566
    },
    {
      "epoch": 0.9819785944220769,
      "lm_loss": 1.2833,
      "step": 13384,
      "vm_loss": 0.1235
    },
    {
      "epoch": 0.9819785944220769,
      "lm_loss": 0.549,
      "step": 13384,
      "vm_loss": 0.0813
    },
    {
      "epoch": 0.9819785944220769,
      "lm_loss": 0.6965,
      "step": 13384,
      "vm_loss": 0.2671
    },
    {
      "epoch": 0.9819785944220769,
      "lm_loss": 0.5463,
      "step": 13384,
      "vm_loss": 0.1243
    },
    {
      "epoch": 0.9819785944220769,
      "lm_loss": 0.8048,
      "step": 13384,
      "vm_loss": 0.1105
    },
    {
      "epoch": 0.9819785944220769,
      "lm_loss": 0.1394,
      "step": 13384,
      "vm_loss": 0.0741
    },
    {
      "epoch": 0.9819785944220769,
      "lm_loss": 0.8323,
      "step": 13384,
      "vm_loss": 0.1308
    },
    {
      "epoch": 0.9820519640122161,
      "grad_norm": 1.5240313975086304,
      "learning_rate": 8.403008695029302e-09,
      "loss": 0.6849,
      "step": 13385
    },
    {
      "epoch": 0.9821253336023552,
      "grad_norm": 1.448744996962838,
      "learning_rate": 8.334291818137274e-09,
      "loss": 0.6595,
      "step": 13386
    },
    {
      "epoch": 0.9821987031924942,
      "grad_norm": 1.6499691902397475,
      "learning_rate": 8.2658568329691e-09,
      "loss": 0.684,
      "step": 13387
    },
    {
      "epoch": 0.9822720727826334,
      "grad_norm": 0.5289651995195374,
      "learning_rate": 8.197703743390573e-09,
      "loss": 0.4744,
      "step": 13388
    },
    {
      "epoch": 0.9823454423727725,
      "grad_norm": 1.5741327072248448,
      "learning_rate": 8.129832553249173e-09,
      "loss": 0.6021,
      "step": 13389
    },
    {
      "epoch": 0.9824188119629117,
      "grad_norm": 1.433049383402745,
      "learning_rate": 8.0622432663785e-09,
      "loss": 0.7321,
      "step": 13390
    },
    {
      "epoch": 0.9824921815530508,
      "grad_norm": 0.5442803756378993,
      "learning_rate": 7.994935886594945e-09,
      "loss": 0.4791,
      "step": 13391
    },
    {
      "epoch": 0.9825655511431899,
      "grad_norm": 1.5925496108594408,
      "learning_rate": 7.927910417699913e-09,
      "loss": 0.6955,
      "step": 13392
    },
    {
      "epoch": 0.9825655511431899,
      "lm_loss": 1.3256,
      "step": 13392,
      "vm_loss": 0.1
    },
    {
      "epoch": 0.9825655511431899,
      "lm_loss": 0.5667,
      "step": 13392,
      "vm_loss": 0.0726
    },
    {
      "epoch": 0.9825655511431899,
      "lm_loss": 0.5961,
      "step": 13392,
      "vm_loss": 0.1477
    },
    {
      "epoch": 0.9825655511431899,
      "lm_loss": 0.1309,
      "step": 13392,
      "vm_loss": 0.0986
    },
    {
      "epoch": 0.9825655511431899,
      "lm_loss": 0.5212,
      "step": 13392,
      "vm_loss": 0.0864
    },
    {
      "epoch": 0.9825655511431899,
      "lm_loss": 0.4649,
      "step": 13392,
      "vm_loss": 0.0615
    },
    {
      "epoch": 0.9825655511431899,
      "lm_loss": 0.3893,
      "step": 13392,
      "vm_loss": 0.1004
    },
    {
      "epoch": 0.9825655511431899,
      "lm_loss": 1.2458,
      "step": 13392,
      "vm_loss": 0.1418
    },
    {
      "epoch": 0.982638920733329,
      "grad_norm": 1.6739043906751012,
      "learning_rate": 7.861166863478709e-09,
      "loss": 0.6529,
      "step": 13393
    },
    {
      "epoch": 0.9827122903234682,
      "grad_norm": 0.5481003755063812,
      "learning_rate": 7.794705227699984e-09,
      "loss": 0.485,
      "step": 13394
    },
    {
      "epoch": 0.9827856599136073,
      "grad_norm": 1.5612114346059582,
      "learning_rate": 7.728525514117402e-09,
      "loss": 0.7263,
      "step": 13395
    },
    {
      "epoch": 0.9828590295037465,
      "grad_norm": 1.663597348689005,
      "learning_rate": 7.662627726467976e-09,
      "loss": 0.7765,
      "step": 13396
    },
    {
      "epoch": 0.9829323990938855,
      "grad_norm": 1.5677576919105884,
      "learning_rate": 7.597011868473725e-09,
      "loss": 0.6034,
      "step": 13397
    },
    {
      "epoch": 0.9830057686840247,
      "grad_norm": 1.585336872564817,
      "learning_rate": 7.53167794383891e-09,
      "loss": 0.7649,
      "step": 13398
    },
    {
      "epoch": 0.9830791382741638,
      "grad_norm": 0.5465250377766863,
      "learning_rate": 7.46662595625447e-09,
      "loss": 0.4891,
      "step": 13399
    },
    {
      "epoch": 0.983152507864303,
      "grad_norm": 1.5929818908899955,
      "learning_rate": 7.401855909393574e-09,
      "loss": 0.7583,
      "step": 13400
    },
    {
      "epoch": 0.983152507864303,
      "lm_loss": 1.3476,
      "step": 13400,
      "vm_loss": 0.0886
    },
    {
      "epoch": 0.983152507864303,
      "lm_loss": 1.2293,
      "step": 13400,
      "vm_loss": 0.1362
    },
    {
      "epoch": 0.983152507864303,
      "lm_loss": 0.9729,
      "step": 13400,
      "vm_loss": 0.0871
    },
    {
      "epoch": 0.983152507864303,
      "lm_loss": 0.707,
      "step": 13400,
      "vm_loss": 0.185
    },
    {
      "epoch": 0.983152507864303,
      "lm_loss": 0.0746,
      "step": 13400,
      "vm_loss": 0.0795
    },
    {
      "epoch": 0.983152507864303,
      "lm_loss": 0.3032,
      "step": 13400,
      "vm_loss": 0.1154
    },
    {
      "epoch": 0.983152507864303,
      "lm_loss": 0.6886,
      "step": 13400,
      "vm_loss": 0.1516
    },
    {
      "epoch": 0.983152507864303,
      "lm_loss": 0.4681,
      "step": 13400,
      "vm_loss": 0.2371
    },
    {
      "epoch": 0.9832258774544421,
      "grad_norm": 1.453241513710178,
      "learning_rate": 7.337367806913298e-09,
      "loss": 0.8296,
      "step": 13401
    },
    {
      "epoch": 0.9832992470445812,
      "grad_norm": 0.5403634645178521,
      "learning_rate": 7.273161652456284e-09,
      "loss": 0.4839,
      "step": 13402
    },
    {
      "epoch": 0.9833726166347203,
      "grad_norm": 1.5490985411187301,
      "learning_rate": 7.209237449647966e-09,
      "loss": 0.6304,
      "step": 13403
    },
    {
      "epoch": 0.9834459862248595,
      "grad_norm": 0.539448598404615,
      "learning_rate": 7.145595202098787e-09,
      "loss": 0.4632,
      "step": 13404
    },
    {
      "epoch": 0.9835193558149986,
      "grad_norm": 1.4066976148833188,
      "learning_rate": 7.082234913402542e-09,
      "loss": 0.6989,
      "step": 13405
    },
    {
      "epoch": 0.9835927254051378,
      "grad_norm": 1.4445360777971448,
      "learning_rate": 7.019156587136921e-09,
      "loss": 0.7041,
      "step": 13406
    },
    {
      "epoch": 0.9836660949952768,
      "grad_norm": 1.7142224769868721,
      "learning_rate": 6.956360226864079e-09,
      "loss": 0.8557,
      "step": 13407
    },
    {
      "epoch": 0.983739464585416,
      "grad_norm": 1.5642553945558295,
      "learning_rate": 6.893845836131175e-09,
      "loss": 0.7746,
      "step": 13408
    },
    {
      "epoch": 0.983739464585416,
      "lm_loss": 1.3142,
      "step": 13408,
      "vm_loss": 0.0478
    },
    {
      "epoch": 0.983739464585416,
      "lm_loss": 0.4235,
      "step": 13408,
      "vm_loss": 0.0463
    },
    {
      "epoch": 0.983739464585416,
      "lm_loss": 0.0913,
      "step": 13408,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.983739464585416,
      "lm_loss": 0.7321,
      "step": 13408,
      "vm_loss": 0.1277
    },
    {
      "epoch": 0.983739464585416,
      "lm_loss": 0.7634,
      "step": 13408,
      "vm_loss": 0.1641
    },
    {
      "epoch": 0.983739464585416,
      "lm_loss": 0.2022,
      "step": 13408,
      "vm_loss": 0.0774
    },
    {
      "epoch": 0.983739464585416,
      "lm_loss": 0.1113,
      "step": 13408,
      "vm_loss": 0.172
    },
    {
      "epoch": 0.983739464585416,
      "lm_loss": 0.1357,
      "step": 13408,
      "vm_loss": 0.1088
    },
    {
      "epoch": 0.9838128341755551,
      "grad_norm": 1.5001702578903588,
      "learning_rate": 6.831613418468163e-09,
      "loss": 0.7698,
      "step": 13409
    },
    {
      "epoch": 0.9838862037656942,
      "grad_norm": 1.6206491764423876,
      "learning_rate": 6.769662977388902e-09,
      "loss": 0.7898,
      "step": 13410
    },
    {
      "epoch": 0.9839595733558334,
      "grad_norm": 1.4944292046861336,
      "learning_rate": 6.7079945163922575e-09,
      "loss": 0.7985,
      "step": 13411
    },
    {
      "epoch": 0.9840329429459724,
      "grad_norm": 1.5885285142235681,
      "learning_rate": 6.646608038961e-09,
      "loss": 0.7437,
      "step": 13412
    },
    {
      "epoch": 0.9841063125361116,
      "grad_norm": 1.536690472294344,
      "learning_rate": 6.5855035485618e-09,
      "loss": 0.6489,
      "step": 13413
    },
    {
      "epoch": 0.9841796821262507,
      "grad_norm": 1.5243791897897665,
      "learning_rate": 6.524681048644677e-09,
      "loss": 0.6783,
      "step": 13414
    },
    {
      "epoch": 0.9842530517163899,
      "grad_norm": 1.402848808521071,
      "learning_rate": 6.464140542645214e-09,
      "loss": 0.6734,
      "step": 13415
    },
    {
      "epoch": 0.984326421306529,
      "grad_norm": 1.6965006836757084,
      "learning_rate": 6.403882033982345e-09,
      "loss": 0.6929,
      "step": 13416
    },
    {
      "epoch": 0.984326421306529,
      "lm_loss": 0.3362,
      "step": 13416,
      "vm_loss": 0.1292
    },
    {
      "epoch": 0.984326421306529,
      "lm_loss": 0.0167,
      "step": 13416,
      "vm_loss": 0.0433
    },
    {
      "epoch": 0.984326421306529,
      "lm_loss": 1.089,
      "step": 13416,
      "vm_loss": 0.1469
    },
    {
      "epoch": 0.984326421306529,
      "lm_loss": 0.9415,
      "step": 13416,
      "vm_loss": 0.1721
    },
    {
      "epoch": 0.984326421306529,
      "lm_loss": 0.2298,
      "step": 13416,
      "vm_loss": 0.0926
    },
    {
      "epoch": 0.984326421306529,
      "lm_loss": 0.2446,
      "step": 13416,
      "vm_loss": 0.0789
    },
    {
      "epoch": 0.984326421306529,
      "lm_loss": 0.0777,
      "step": 13416,
      "vm_loss": 0.1261
    },
    {
      "epoch": 0.984326421306529,
      "lm_loss": 0.5151,
      "step": 13416,
      "vm_loss": 0.0715
    },
    {
      "epoch": 0.9843997908966681,
      "grad_norm": 1.5176280770724624,
      "learning_rate": 6.3439055260583475e-09,
      "loss": 0.7268,
      "step": 13417
    },
    {
      "epoch": 0.9844731604868072,
      "grad_norm": 0.5380194789891509,
      "learning_rate": 6.284211022260511e-09,
      "loss": 0.4655,
      "step": 13418
    },
    {
      "epoch": 0.9845465300769464,
      "grad_norm": 0.5408737892448294,
      "learning_rate": 6.224798525960029e-09,
      "loss": 0.4637,
      "step": 13419
    },
    {
      "epoch": 0.9846198996670855,
      "grad_norm": 0.5372814823118933,
      "learning_rate": 6.1656680405125515e-09,
      "loss": 0.4683,
      "step": 13420
    },
    {
      "epoch": 0.9846932692572246,
      "grad_norm": 0.5304147425673196,
      "learning_rate": 6.1068195692565166e-09,
      "loss": 0.4483,
      "step": 13421
    },
    {
      "epoch": 0.9847666388473637,
      "grad_norm": 1.601231451611326,
      "learning_rate": 6.048253115515379e-09,
      "loss": 0.718,
      "step": 13422
    },
    {
      "epoch": 0.9848400084375029,
      "grad_norm": 1.6326567319979093,
      "learning_rate": 5.989968682596492e-09,
      "loss": 0.6706,
      "step": 13423
    },
    {
      "epoch": 0.984913378027642,
      "grad_norm": 1.470406651034875,
      "learning_rate": 5.931966273792222e-09,
      "loss": 0.7024,
      "step": 13424
    },
    {
      "epoch": 0.984913378027642,
      "lm_loss": 1.1965,
      "step": 13424,
      "vm_loss": 0.0742
    },
    {
      "epoch": 0.984913378027642,
      "lm_loss": 0.4878,
      "step": 13424,
      "vm_loss": 0.1639
    },
    {
      "epoch": 0.984913378027642,
      "lm_loss": 0.6993,
      "step": 13424,
      "vm_loss": 0.1426
    },
    {
      "epoch": 0.984913378027642,
      "lm_loss": 0.5443,
      "step": 13424,
      "vm_loss": 0.1578
    },
    {
      "epoch": 0.984913378027642,
      "lm_loss": 0.472,
      "step": 13424,
      "vm_loss": 0.1643
    },
    {
      "epoch": 0.984913378027642,
      "lm_loss": 0.1958,
      "step": 13424,
      "vm_loss": 0.1586
    },
    {
      "epoch": 0.984913378027642,
      "lm_loss": 0.7336,
      "step": 13424,
      "vm_loss": 0.1991
    },
    {
      "epoch": 0.984913378027642,
      "lm_loss": 0.3651,
      "step": 13424,
      "vm_loss": 0.0727
    },
    {
      "epoch": 0.9849867476177812,
      "grad_norm": 1.5141577281549552,
      "learning_rate": 5.8742458923766174e-09,
      "loss": 0.7244,
      "step": 13425
    },
    {
      "epoch": 0.9850601172079202,
      "grad_norm": 1.434785276937755,
      "learning_rate": 5.816807541610958e-09,
      "loss": 0.647,
      "step": 13426
    },
    {
      "epoch": 0.9851334867980593,
      "grad_norm": 1.5398901953596966,
      "learning_rate": 5.75965122473765e-09,
      "loss": 0.7311,
      "step": 13427
    },
    {
      "epoch": 0.9852068563881985,
      "grad_norm": 1.3797950955347789,
      "learning_rate": 5.702776944984667e-09,
      "loss": 0.6521,
      "step": 13428
    },
    {
      "epoch": 0.9852802259783376,
      "grad_norm": 1.5096525531418243,
      "learning_rate": 5.646184705563884e-09,
      "loss": 0.8424,
      "step": 13429
    },
    {
      "epoch": 0.9853535955684768,
      "grad_norm": 0.5409828312774233,
      "learning_rate": 5.5898745096716336e-09,
      "loss": 0.4669,
      "step": 13430
    },
    {
      "epoch": 0.9854269651586158,
      "grad_norm": 0.5336912954522217,
      "learning_rate": 5.533846360488149e-09,
      "loss": 0.4676,
      "step": 13431
    },
    {
      "epoch": 0.985500334748755,
      "grad_norm": 1.5717461882818091,
      "learning_rate": 5.478100261176456e-09,
      "loss": 0.7864,
      "step": 13432
    },
    {
      "epoch": 0.985500334748755,
      "lm_loss": 0.5489,
      "step": 13432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.985500334748755,
      "lm_loss": 0.3921,
      "step": 13432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.985500334748755,
      "lm_loss": 0.4177,
      "step": 13432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.985500334748755,
      "lm_loss": 0.3106,
      "step": 13432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.985500334748755,
      "lm_loss": 0.6399,
      "step": 13432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.985500334748755,
      "lm_loss": 0.4315,
      "step": 13432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.985500334748755,
      "lm_loss": 0.5566,
      "step": 13432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.985500334748755,
      "lm_loss": 0.3806,
      "step": 13432,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9855737043388941,
      "grad_norm": 0.5636237829689777,
      "learning_rate": 5.422636214885146e-09,
      "loss": 0.4844,
      "step": 13433
    },
    {
      "epoch": 0.9856470739290333,
      "grad_norm": 1.3547054596888686,
      "learning_rate": 5.36745422474616e-09,
      "loss": 0.7173,
      "step": 13434
    },
    {
      "epoch": 0.9857204435191724,
      "grad_norm": 1.4992029983449582,
      "learning_rate": 5.3125542938764485e-09,
      "loss": 0.8241,
      "step": 13435
    },
    {
      "epoch": 0.9857938131093115,
      "grad_norm": 1.6390864501808455,
      "learning_rate": 5.257936425375754e-09,
      "loss": 0.7328,
      "step": 13436
    },
    {
      "epoch": 0.9858671826994506,
      "grad_norm": 1.6702014735926676,
      "learning_rate": 5.203600622328831e-09,
      "loss": 0.771,
      "step": 13437
    },
    {
      "epoch": 0.9859405522895898,
      "grad_norm": 0.5392020987370563,
      "learning_rate": 5.14954688780378e-09,
      "loss": 0.4759,
      "step": 13438
    },
    {
      "epoch": 0.9860139218797289,
      "grad_norm": 0.5379201450452912,
      "learning_rate": 5.095775224853161e-09,
      "loss": 0.4704,
      "step": 13439
    },
    {
      "epoch": 0.9860872914698681,
      "grad_norm": 0.5555050128877422,
      "learning_rate": 5.042285636514544e-09,
      "loss": 0.49,
      "step": 13440
    },
    {
      "epoch": 0.9860872914698681,
      "lm_loss": 0.3281,
      "step": 13440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9860872914698681,
      "lm_loss": 0.3717,
      "step": 13440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9860872914698681,
      "lm_loss": 0.4196,
      "step": 13440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9860872914698681,
      "lm_loss": 0.6234,
      "step": 13440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9860872914698681,
      "lm_loss": 0.4721,
      "step": 13440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9860872914698681,
      "lm_loss": 0.5784,
      "step": 13440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9860872914698681,
      "lm_loss": 0.61,
      "step": 13440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9860872914698681,
      "lm_loss": 0.47,
      "step": 13440,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9861606610600071,
      "grad_norm": 0.5399796042025788,
      "learning_rate": 4.989078125807178e-09,
      "loss": 0.4629,
      "step": 13441
    },
    {
      "epoch": 0.9862340306501463,
      "grad_norm": 1.4481785167722532,
      "learning_rate": 4.93615269573644e-09,
      "loss": 0.7105,
      "step": 13442
    },
    {
      "epoch": 0.9863074002402854,
      "grad_norm": 1.609238367409768,
      "learning_rate": 4.883509349291604e-09,
      "loss": 0.6968,
      "step": 13443
    },
    {
      "epoch": 0.9863807698304246,
      "grad_norm": 0.537004104511605,
      "learning_rate": 4.8311480894452925e-09,
      "loss": 0.473,
      "step": 13444
    },
    {
      "epoch": 0.9864541394205637,
      "grad_norm": 1.5584198337961475,
      "learning_rate": 4.779068919154029e-09,
      "loss": 0.6412,
      "step": 13445
    },
    {
      "epoch": 0.9865275090107027,
      "grad_norm": 1.5188320716057775,
      "learning_rate": 4.727271841359349e-09,
      "loss": 0.7056,
      "step": 13446
    },
    {
      "epoch": 0.9866008786008419,
      "grad_norm": 1.3768531363917365,
      "learning_rate": 4.6757568589861355e-09,
      "loss": 0.6592,
      "step": 13447
    },
    {
      "epoch": 0.986674248190981,
      "grad_norm": 1.381058027965786,
      "learning_rate": 4.624523974943729e-09,
      "loss": 0.6762,
      "step": 13448
    },
    {
      "epoch": 0.986674248190981,
      "lm_loss": 1.877,
      "step": 13448,
      "vm_loss": 0.0887
    },
    {
      "epoch": 0.986674248190981,
      "lm_loss": 0.3697,
      "step": 13448,
      "vm_loss": 0.1002
    },
    {
      "epoch": 0.986674248190981,
      "lm_loss": 0.3066,
      "step": 13448,
      "vm_loss": 0.0415
    },
    {
      "epoch": 0.986674248190981,
      "lm_loss": 0.1081,
      "step": 13448,
      "vm_loss": 0.1184
    },
    {
      "epoch": 0.986674248190981,
      "lm_loss": 0.7831,
      "step": 13448,
      "vm_loss": 0.169
    },
    {
      "epoch": 0.986674248190981,
      "lm_loss": 0.6651,
      "step": 13448,
      "vm_loss": 0.0837
    },
    {
      "epoch": 0.986674248190981,
      "lm_loss": 0.5676,
      "step": 13448,
      "vm_loss": 0.1776
    },
    {
      "epoch": 0.986674248190981,
      "lm_loss": 0.0547,
      "step": 13448,
      "vm_loss": 0.093
    },
    {
      "epoch": 0.9867476177811202,
      "grad_norm": 1.6589371207473043,
      "learning_rate": 4.573573192125369e-09,
      "loss": 0.7391,
      "step": 13449
    },
    {
      "epoch": 0.9868209873712593,
      "grad_norm": 1.5107487979245253,
      "learning_rate": 4.522904513408199e-09,
      "loss": 0.6865,
      "step": 13450
    },
    {
      "epoch": 0.9868943569613984,
      "grad_norm": 1.5013383701404543,
      "learning_rate": 4.47251794165382e-09,
      "loss": 0.7539,
      "step": 13451
    },
    {
      "epoch": 0.9869677265515375,
      "grad_norm": 1.5187697281943966,
      "learning_rate": 4.422413479707177e-09,
      "loss": 0.6948,
      "step": 13452
    },
    {
      "epoch": 0.9870410961416767,
      "grad_norm": 1.4594429018662076,
      "learning_rate": 4.372591130398784e-09,
      "loss": 0.6854,
      "step": 13453
    },
    {
      "epoch": 0.9871144657318158,
      "grad_norm": 0.5276822980678675,
      "learning_rate": 4.323050896541392e-09,
      "loss": 0.4769,
      "step": 13454
    },
    {
      "epoch": 0.987187835321955,
      "grad_norm": 0.5442147832621987,
      "learning_rate": 4.2737927809327615e-09,
      "loss": 0.4702,
      "step": 13455
    },
    {
      "epoch": 0.987261204912094,
      "grad_norm": 1.513774761940381,
      "learning_rate": 4.224816786354557e-09,
      "loss": 0.7732,
      "step": 13456
    },
    {
      "epoch": 0.987261204912094,
      "lm_loss": 1.4819,
      "step": 13456,
      "vm_loss": 0.06
    },
    {
      "epoch": 0.987261204912094,
      "lm_loss": 0.8108,
      "step": 13456,
      "vm_loss": 0.135
    },
    {
      "epoch": 0.987261204912094,
      "lm_loss": 0.4171,
      "step": 13456,
      "vm_loss": 0.1908
    },
    {
      "epoch": 0.987261204912094,
      "lm_loss": 0.9388,
      "step": 13456,
      "vm_loss": 0.1261
    },
    {
      "epoch": 0.987261204912094,
      "lm_loss": 0.4091,
      "step": 13456,
      "vm_loss": 0.0922
    },
    {
      "epoch": 0.987261204912094,
      "lm_loss": 0.8945,
      "step": 13456,
      "vm_loss": 0.1993
    },
    {
      "epoch": 0.987261204912094,
      "lm_loss": 0.9295,
      "step": 13456,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.987261204912094,
      "lm_loss": 0.2844,
      "step": 13456,
      "vm_loss": 0.0577
    },
    {
      "epoch": 0.9873345745022332,
      "grad_norm": 1.604093476570783,
      "learning_rate": 4.1761229155729e-09,
      "loss": 0.6749,
      "step": 13457
    },
    {
      "epoch": 0.9874079440923723,
      "grad_norm": 1.6370504339533427,
      "learning_rate": 4.127711171337812e-09,
      "loss": 0.77,
      "step": 13458
    },
    {
      "epoch": 0.9874813136825115,
      "grad_norm": 1.3629824394495211,
      "learning_rate": 4.079581556382661e-09,
      "loss": 0.5812,
      "step": 13459
    },
    {
      "epoch": 0.9875546832726506,
      "grad_norm": 1.480607768172869,
      "learning_rate": 4.031734073426386e-09,
      "loss": 0.7583,
      "step": 13460
    },
    {
      "epoch": 0.9876280528627897,
      "grad_norm": 1.669765691790608,
      "learning_rate": 3.984168725169602e-09,
      "loss": 0.8181,
      "step": 13461
    },
    {
      "epoch": 0.9877014224529288,
      "grad_norm": 0.5458657808733962,
      "learning_rate": 3.936885514299604e-09,
      "loss": 0.475,
      "step": 13462
    },
    {
      "epoch": 0.987774792043068,
      "grad_norm": 1.3925971575463123,
      "learning_rate": 3.889884443485925e-09,
      "loss": 0.7777,
      "step": 13463
    },
    {
      "epoch": 0.9878481616332071,
      "grad_norm": 0.5292005403704162,
      "learning_rate": 3.8431655153836626e-09,
      "loss": 0.451,
      "step": 13464
    },
    {
      "epoch": 0.9878481616332071,
      "lm_loss": 1.3214,
      "step": 13464,
      "vm_loss": 0.0728
    },
    {
      "epoch": 0.9878481616332071,
      "lm_loss": 0.7052,
      "step": 13464,
      "vm_loss": 0.142
    },
    {
      "epoch": 0.9878481616332071,
      "lm_loss": 0.5545,
      "step": 13464,
      "vm_loss": 0.0805
    },
    {
      "epoch": 0.9878481616332071,
      "lm_loss": 0.4105,
      "step": 13464,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.9878481616332071,
      "lm_loss": 0.5728,
      "step": 13464,
      "vm_loss": 0.0634
    },
    {
      "epoch": 0.9878481616332071,
      "lm_loss": 0.0493,
      "step": 13464,
      "vm_loss": 0.1191
    },
    {
      "epoch": 0.9878481616332071,
      "lm_loss": 0.8108,
      "step": 13464,
      "vm_loss": 0.1992
    },
    {
      "epoch": 0.9878481616332071,
      "lm_loss": 0.804,
      "step": 13464,
      "vm_loss": 0.2004
    },
    {
      "epoch": 0.9879215312233462,
      "grad_norm": 1.550622725764904,
      "learning_rate": 3.7967287326301505e-09,
      "loss": 0.7589,
      "step": 13465
    },
    {
      "epoch": 0.9879949008134853,
      "grad_norm": 1.497908987685733,
      "learning_rate": 3.750574097848292e-09,
      "loss": 0.763,
      "step": 13466
    },
    {
      "epoch": 0.9880682704036244,
      "grad_norm": 1.5957881289806395,
      "learning_rate": 3.7047016136443346e-09,
      "loss": 0.6745,
      "step": 13467
    },
    {
      "epoch": 0.9881416399937636,
      "grad_norm": 1.6553822055615406,
      "learning_rate": 3.65911128260954e-09,
      "loss": 0.7075,
      "step": 13468
    },
    {
      "epoch": 0.9882150095839027,
      "grad_norm": 1.3915943025322701,
      "learning_rate": 3.613803107317959e-09,
      "loss": 0.7442,
      "step": 13469
    },
    {
      "epoch": 0.9882883791740419,
      "grad_norm": 1.47544081953769,
      "learning_rate": 3.5687770903275464e-09,
      "loss": 0.7054,
      "step": 13470
    },
    {
      "epoch": 0.9883617487641809,
      "grad_norm": 1.4455269033596683,
      "learning_rate": 3.5240332341823778e-09,
      "loss": 0.6918,
      "step": 13471
    },
    {
      "epoch": 0.9884351183543201,
      "grad_norm": 0.5644897506354667,
      "learning_rate": 3.479571541408211e-09,
      "loss": 0.4625,
      "step": 13472
    },
    {
      "epoch": 0.9884351183543201,
      "lm_loss": 0.9518,
      "step": 13472,
      "vm_loss": 0.0875
    },
    {
      "epoch": 0.9884351183543201,
      "lm_loss": 1.4197,
      "step": 13472,
      "vm_loss": 0.136
    },
    {
      "epoch": 0.9884351183543201,
      "lm_loss": 0.6815,
      "step": 13472,
      "vm_loss": 0.1411
    },
    {
      "epoch": 0.9884351183543201,
      "lm_loss": 0.9095,
      "step": 13472,
      "vm_loss": 0.1697
    },
    {
      "epoch": 0.9884351183543201,
      "lm_loss": 0.6336,
      "step": 13472,
      "vm_loss": 0.0995
    },
    {
      "epoch": 0.9884351183543201,
      "lm_loss": 0.489,
      "step": 13472,
      "vm_loss": 0.1334
    },
    {
      "epoch": 0.9884351183543201,
      "lm_loss": 0.4041,
      "step": 13472,
      "vm_loss": 0.1474
    },
    {
      "epoch": 0.9884351183543201,
      "lm_loss": 0.9339,
      "step": 13472,
      "vm_loss": 0.192
    },
    {
      "epoch": 0.9885084879444592,
      "grad_norm": 1.7882990756959656,
      "learning_rate": 3.435392014516925e-09,
      "loss": 0.7954,
      "step": 13473
    },
    {
      "epoch": 0.9885818575345984,
      "grad_norm": 1.611819198994068,
      "learning_rate": 3.3914946560020813e-09,
      "loss": 0.7479,
      "step": 13474
    },
    {
      "epoch": 0.9886552271247375,
      "grad_norm": 1.4318109088728959,
      "learning_rate": 3.3478794683439176e-09,
      "loss": 0.7356,
      "step": 13475
    },
    {
      "epoch": 0.9887285967148766,
      "grad_norm": 0.5396233743196469,
      "learning_rate": 3.304546454004909e-09,
      "loss": 0.4541,
      "step": 13476
    },
    {
      "epoch": 0.9888019663050157,
      "grad_norm": 0.5307834223704855,
      "learning_rate": 3.2614956154319868e-09,
      "loss": 0.4788,
      "step": 13477
    },
    {
      "epoch": 0.9888753358951549,
      "grad_norm": 1.6815288799821175,
      "learning_rate": 3.2187269550570944e-09,
      "loss": 0.6939,
      "step": 13478
    },
    {
      "epoch": 0.988948705485294,
      "grad_norm": 0.5491316559563277,
      "learning_rate": 3.176240475294967e-09,
      "loss": 0.4752,
      "step": 13479
    },
    {
      "epoch": 0.9890220750754332,
      "grad_norm": 1.4986433337271203,
      "learning_rate": 3.134036178544797e-09,
      "loss": 0.6479,
      "step": 13480
    },
    {
      "epoch": 0.9890220750754332,
      "lm_loss": 1.0763,
      "step": 13480,
      "vm_loss": 0.0693
    },
    {
      "epoch": 0.9890220750754332,
      "lm_loss": 1.0874,
      "step": 13480,
      "vm_loss": 0.1139
    },
    {
      "epoch": 0.9890220750754332,
      "lm_loss": 0.6503,
      "step": 13480,
      "vm_loss": 0.1865
    },
    {
      "epoch": 0.9890220750754332,
      "lm_loss": 0.9562,
      "step": 13480,
      "vm_loss": 0.1467
    },
    {
      "epoch": 0.9890220750754332,
      "lm_loss": 0.2472,
      "step": 13480,
      "vm_loss": 0.0849
    },
    {
      "epoch": 0.9890220750754332,
      "lm_loss": 0.2999,
      "step": 13480,
      "vm_loss": 0.0746
    },
    {
      "epoch": 0.9890220750754332,
      "lm_loss": 0.8589,
      "step": 13480,
      "vm_loss": 0.0612
    },
    {
      "epoch": 0.9890220750754332,
      "lm_loss": 0.2257,
      "step": 13480,
      "vm_loss": 0.0814
    },
    {
      "epoch": 0.9890954446655722,
      "grad_norm": 1.3530936209989304,
      "learning_rate": 3.0921140671902326e-09,
      "loss": 0.6898,
      "step": 13481
    },
    {
      "epoch": 0.9891688142557113,
      "grad_norm": 0.536240872429566,
      "learning_rate": 3.05047414359827e-09,
      "loss": 0.4658,
      "step": 13482
    },
    {
      "epoch": 0.9892421838458505,
      "grad_norm": 1.4387765392465117,
      "learning_rate": 3.009116410120916e-09,
      "loss": 0.653,
      "step": 13483
    },
    {
      "epoch": 0.9893155534359896,
      "grad_norm": 1.506184963598324,
      "learning_rate": 2.9680408690935246e-09,
      "loss": 0.7008,
      "step": 13484
    },
    {
      "epoch": 0.9893889230261288,
      "grad_norm": 0.5175589212291496,
      "learning_rate": 2.9272475228359076e-09,
      "loss": 0.4531,
      "step": 13485
    },
    {
      "epoch": 0.9894622926162678,
      "grad_norm": 1.4946916003663822,
      "learning_rate": 2.8867363736517775e-09,
      "loss": 0.6733,
      "step": 13486
    },
    {
      "epoch": 0.989535662206407,
      "grad_norm": 1.5062701402059888,
      "learning_rate": 2.846507423828748e-09,
      "loss": 0.6473,
      "step": 13487
    },
    {
      "epoch": 0.9896090317965461,
      "grad_norm": 1.578100978294977,
      "learning_rate": 2.806560675638892e-09,
      "loss": 0.6744,
      "step": 13488
    },
    {
      "epoch": 0.9896090317965461,
      "lm_loss": 0.6502,
      "step": 13488,
      "vm_loss": 0.0594
    },
    {
      "epoch": 0.9896090317965461,
      "lm_loss": 0.9762,
      "step": 13488,
      "vm_loss": 0.0688
    },
    {
      "epoch": 0.9896090317965461,
      "lm_loss": 0.0965,
      "step": 13488,
      "vm_loss": 0.1666
    },
    {
      "epoch": 0.9896090317965461,
      "lm_loss": 0.939,
      "step": 13488,
      "vm_loss": 0.1931
    },
    {
      "epoch": 0.9896090317965461,
      "lm_loss": 0.5232,
      "step": 13488,
      "vm_loss": 0.182
    },
    {
      "epoch": 0.9896090317965461,
      "lm_loss": 0.7267,
      "step": 13488,
      "vm_loss": 0.0894
    },
    {
      "epoch": 0.9896090317965461,
      "lm_loss": 0.1717,
      "step": 13488,
      "vm_loss": 0.0794
    },
    {
      "epoch": 0.9896090317965461,
      "lm_loss": 0.0853,
      "step": 13488,
      "vm_loss": 0.0784
    },
    {
      "epoch": 0.9896824013866853,
      "grad_norm": 1.462326492377116,
      "learning_rate": 2.7668961313376263e-09,
      "loss": 0.7006,
      "step": 13489
    },
    {
      "epoch": 0.9897557709768244,
      "grad_norm": 1.4302666446562022,
      "learning_rate": 2.7275137931653815e-09,
      "loss": 0.7318,
      "step": 13490
    },
    {
      "epoch": 0.9898291405669635,
      "grad_norm": 1.5535717377669416,
      "learning_rate": 2.6884136633459344e-09,
      "loss": 0.8413,
      "step": 13491
    },
    {
      "epoch": 0.9899025101571026,
      "grad_norm": 1.4827170841670347,
      "learning_rate": 2.6495957440869634e-09,
      "loss": 0.7383,
      "step": 13492
    },
    {
      "epoch": 0.9899758797472418,
      "grad_norm": 1.4696897988853719,
      "learning_rate": 2.611060037581159e-09,
      "loss": 0.8014,
      "step": 13493
    },
    {
      "epoch": 0.9900492493373809,
      "grad_norm": 1.4215912029103273,
      "learning_rate": 2.572806546005113e-09,
      "loss": 0.7066,
      "step": 13494
    },
    {
      "epoch": 0.9901226189275201,
      "grad_norm": 0.5428943108957214,
      "learning_rate": 2.53483527151821e-09,
      "loss": 0.4838,
      "step": 13495
    },
    {
      "epoch": 0.9901959885176591,
      "grad_norm": 1.8395242156163742,
      "learning_rate": 2.4971462162654002e-09,
      "loss": 0.8377,
      "step": 13496
    },
    {
      "epoch": 0.9901959885176591,
      "lm_loss": 0.5119,
      "step": 13496,
      "vm_loss": 0.0504
    },
    {
      "epoch": 0.9901959885176591,
      "lm_loss": 1.4016,
      "step": 13496,
      "vm_loss": 0.0862
    },
    {
      "epoch": 0.9901959885176591,
      "lm_loss": 0.2696,
      "step": 13496,
      "vm_loss": 0.0942
    },
    {
      "epoch": 0.9901959885176591,
      "lm_loss": 0.3192,
      "step": 13496,
      "vm_loss": 0.0857
    },
    {
      "epoch": 0.9901959885176591,
      "lm_loss": 0.9691,
      "step": 13496,
      "vm_loss": 0.1944
    },
    {
      "epoch": 0.9901959885176591,
      "lm_loss": 0.2021,
      "step": 13496,
      "vm_loss": 0.1467
    },
    {
      "epoch": 0.9901959885176591,
      "lm_loss": 0.0943,
      "step": 13496,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.9901959885176591,
      "lm_loss": 0.7467,
      "step": 13496,
      "vm_loss": 0.0946
    },
    {
      "epoch": 0.9902693581077983,
      "grad_norm": 1.4674073586172904,
      "learning_rate": 2.4597393823749817e-09,
      "loss": 0.6805,
      "step": 13497
    },
    {
      "epoch": 0.9903427276979374,
      "grad_norm": 1.5455948112306692,
      "learning_rate": 2.422614771958598e-09,
      "loss": 0.7407,
      "step": 13498
    },
    {
      "epoch": 0.9904160972880766,
      "grad_norm": 0.5451399998737528,
      "learning_rate": 2.385772387114016e-09,
      "loss": 0.4754,
      "step": 13499
    },
    {
      "epoch": 0.9904894668782156,
      "grad_norm": 1.6421310742895505,
      "learning_rate": 2.349212229921238e-09,
      "loss": 0.7613,
      "step": 13500
    },
    {
      "epoch": 0.9905628364683547,
      "grad_norm": 1.5254936488325153,
      "learning_rate": 2.3129343024447248e-09,
      "loss": 0.7829,
      "step": 13501
    },
    {
      "epoch": 0.9906362060584939,
      "grad_norm": 0.5444760476044814,
      "learning_rate": 2.2769386067333922e-09,
      "loss": 0.4736,
      "step": 13502
    },
    {
      "epoch": 0.990709575648633,
      "grad_norm": 0.5418388453389974,
      "learning_rate": 2.2412251448200583e-09,
      "loss": 0.4635,
      "step": 13503
    },
    {
      "epoch": 0.9907829452387722,
      "grad_norm": 1.4775128652986669,
      "learning_rate": 2.205793918721444e-09,
      "loss": 0.7353,
      "step": 13504
    },
    {
      "epoch": 0.9907829452387722,
      "lm_loss": 1.281,
      "step": 13504,
      "vm_loss": 0.0649
    },
    {
      "epoch": 0.9907829452387722,
      "lm_loss": 1.3413,
      "step": 13504,
      "vm_loss": 0.0765
    },
    {
      "epoch": 0.9907829452387722,
      "lm_loss": 0.2423,
      "step": 13504,
      "vm_loss": 0.0815
    },
    {
      "epoch": 0.9907829452387722,
      "lm_loss": 0.6863,
      "step": 13504,
      "vm_loss": 0.0993
    },
    {
      "epoch": 0.9907829452387722,
      "lm_loss": 0.6751,
      "step": 13504,
      "vm_loss": 0.1655
    },
    {
      "epoch": 0.9907829452387722,
      "lm_loss": 0.2882,
      "step": 13504,
      "vm_loss": 0.1446
    },
    {
      "epoch": 0.9907829452387722,
      "lm_loss": 0.0008,
      "step": 13504,
      "vm_loss": 0.0863
    },
    {
      "epoch": 0.9907829452387722,
      "lm_loss": 0.9246,
      "step": 13504,
      "vm_loss": 0.1999
    },
    {
      "epoch": 0.9908563148289112,
      "grad_norm": 1.7845445525703871,
      "learning_rate": 2.1706449304387255e-09,
      "loss": 0.7036,
      "step": 13505
    },
    {
      "epoch": 0.9909296844190504,
      "grad_norm": 0.5437401836642112,
      "learning_rate": 2.1357781819564273e-09,
      "loss": 0.4614,
      "step": 13506
    },
    {
      "epoch": 0.9910030540091895,
      "grad_norm": 1.490221483047835,
      "learning_rate": 2.101193675243529e-09,
      "loss": 0.7941,
      "step": 13507
    },
    {
      "epoch": 0.9910764235993287,
      "grad_norm": 1.4245635299793495,
      "learning_rate": 2.066891412253469e-09,
      "loss": 0.706,
      "step": 13508
    },
    {
      "epoch": 0.9911497931894678,
      "grad_norm": 1.5308445551919092,
      "learning_rate": 2.0328713949230304e-09,
      "loss": 0.7216,
      "step": 13509
    },
    {
      "epoch": 0.9912231627796069,
      "grad_norm": 1.4941573878436862,
      "learning_rate": 1.99913362517401e-09,
      "loss": 0.7622,
      "step": 13510
    },
    {
      "epoch": 0.991296532369746,
      "grad_norm": 1.4804182462876483,
      "learning_rate": 1.9656781049109953e-09,
      "loss": 0.7328,
      "step": 13511
    },
    {
      "epoch": 0.9913699019598852,
      "grad_norm": 0.5471383211541619,
      "learning_rate": 1.9325048360235854e-09,
      "loss": 0.4822,
      "step": 13512
    },
    {
      "epoch": 0.9913699019598852,
      "lm_loss": 1.2144,
      "step": 13512,
      "vm_loss": 0.0632
    },
    {
      "epoch": 0.9913699019598852,
      "lm_loss": 0.7316,
      "step": 13512,
      "vm_loss": 0.084
    },
    {
      "epoch": 0.9913699019598852,
      "lm_loss": 0.621,
      "step": 13512,
      "vm_loss": 0.1746
    },
    {
      "epoch": 0.9913699019598852,
      "lm_loss": 0.2724,
      "step": 13512,
      "vm_loss": 0.0838
    },
    {
      "epoch": 0.9913699019598852,
      "lm_loss": 0.876,
      "step": 13512,
      "vm_loss": 0.1068
    },
    {
      "epoch": 0.9913699019598852,
      "lm_loss": 0.0595,
      "step": 13512,
      "vm_loss": 0.0769
    },
    {
      "epoch": 0.9913699019598852,
      "lm_loss": 0.3365,
      "step": 13512,
      "vm_loss": 0.1074
    },
    {
      "epoch": 0.9913699019598852,
      "lm_loss": 1.1707,
      "step": 13512,
      "vm_loss": 0.224
    },
    {
      "epoch": 0.9914432715500243,
      "grad_norm": 1.4864914347974012,
      "learning_rate": 1.899613820385282e-09,
      "loss": 0.6638,
      "step": 13513
    },
    {
      "epoch": 0.9915166411401635,
      "grad_norm": 0.5439936408514217,
      "learning_rate": 1.867005059852933e-09,
      "loss": 0.4687,
      "step": 13514
    },
    {
      "epoch": 0.9915900107303025,
      "grad_norm": 1.4665745514429909,
      "learning_rate": 1.834678556268954e-09,
      "loss": 0.7126,
      "step": 13515
    },
    {
      "epoch": 0.9916633803204417,
      "grad_norm": 1.505123089794676,
      "learning_rate": 1.8026343114579958e-09,
      "loss": 0.6443,
      "step": 13516
    },
    {
      "epoch": 0.9917367499105808,
      "grad_norm": 1.4583395945545752,
      "learning_rate": 1.7708723272302775e-09,
      "loss": 0.734,
      "step": 13517
    },
    {
      "epoch": 0.99181011950072,
      "grad_norm": 0.5494069896472695,
      "learning_rate": 1.7393926053799194e-09,
      "loss": 0.484,
      "step": 13518
    },
    {
      "epoch": 0.9918834890908591,
      "grad_norm": 1.4208728843890537,
      "learning_rate": 1.708195147683278e-09,
      "loss": 0.6855,
      "step": 13519
    },
    {
      "epoch": 0.9919568586809981,
      "grad_norm": 0.5625933024936843,
      "learning_rate": 1.677279955902833e-09,
      "loss": 0.4947,
      "step": 13520
    },
    {
      "epoch": 0.9919568586809981,
      "lm_loss": 1.0098,
      "step": 13520,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.9919568586809981,
      "lm_loss": 1.3773,
      "step": 13520,
      "vm_loss": 0.104
    },
    {
      "epoch": 0.9919568586809981,
      "lm_loss": 0.4769,
      "step": 13520,
      "vm_loss": 0.1995
    },
    {
      "epoch": 0.9919568586809981,
      "lm_loss": 0.6932,
      "step": 13520,
      "vm_loss": 0.0472
    },
    {
      "epoch": 0.9919568586809981,
      "lm_loss": 0.767,
      "step": 13520,
      "vm_loss": 0.1242
    },
    {
      "epoch": 0.9919568586809981,
      "lm_loss": 0.6934,
      "step": 13520,
      "vm_loss": 0.1731
    },
    {
      "epoch": 0.9919568586809981,
      "lm_loss": 0.5569,
      "step": 13520,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.9919568586809981,
      "lm_loss": 0.8668,
      "step": 13520,
      "vm_loss": 0.1676
    },
    {
      "epoch": 0.9920302282711373,
      "grad_norm": 1.6064888772170487,
      "learning_rate": 1.6466470317849647e-09,
      "loss": 0.7735,
      "step": 13521
    },
    {
      "epoch": 0.9921035978612764,
      "grad_norm": 1.440799835759608,
      "learning_rate": 1.616296377059401e-09,
      "loss": 0.7377,
      "step": 13522
    },
    {
      "epoch": 0.9921769674514156,
      "grad_norm": 1.7407311591089467,
      "learning_rate": 1.5862279934392155e-09,
      "loss": 0.6984,
      "step": 13523
    },
    {
      "epoch": 0.9922503370415547,
      "grad_norm": 0.5543159806810906,
      "learning_rate": 1.5564418826236049e-09,
      "loss": 0.4739,
      "step": 13524
    },
    {
      "epoch": 0.9923237066316938,
      "grad_norm": 1.693576632492285,
      "learning_rate": 1.526938046294002e-09,
      "loss": 0.8011,
      "step": 13525
    },
    {
      "epoch": 0.9923970762218329,
      "grad_norm": 1.5563174594740083,
      "learning_rate": 1.4977164861162962e-09,
      "loss": 0.7258,
      "step": 13526
    },
    {
      "epoch": 0.9924704458119721,
      "grad_norm": 1.501871035485066,
      "learning_rate": 1.4687772037413895e-09,
      "loss": 0.7835,
      "step": 13527
    },
    {
      "epoch": 0.9925438154021112,
      "grad_norm": 0.540161321330615,
      "learning_rate": 1.4401202008035297e-09,
      "loss": 0.4868,
      "step": 13528
    },
    {
      "epoch": 0.9925438154021112,
      "lm_loss": 1.1329,
      "step": 13528,
      "vm_loss": 0.0477
    },
    {
      "epoch": 0.9925438154021112,
      "lm_loss": 0.6911,
      "step": 13528,
      "vm_loss": 0.1103
    },
    {
      "epoch": 0.9925438154021112,
      "lm_loss": 1.1001,
      "step": 13528,
      "vm_loss": 0.0941
    },
    {
      "epoch": 0.9925438154021112,
      "lm_loss": 0.5125,
      "step": 13528,
      "vm_loss": 0.1178
    },
    {
      "epoch": 0.9925438154021112,
      "lm_loss": 0.5307,
      "step": 13528,
      "vm_loss": 0.1127
    },
    {
      "epoch": 0.9925438154021112,
      "lm_loss": 0.4415,
      "step": 13528,
      "vm_loss": 0.0733
    },
    {
      "epoch": 0.9925438154021112,
      "lm_loss": 0.0684,
      "step": 13528,
      "vm_loss": 0.1134
    },
    {
      "epoch": 0.9925438154021112,
      "lm_loss": 0.4573,
      "step": 13528,
      "vm_loss": 0.045
    },
    {
      "epoch": 0.9926171849922504,
      "grad_norm": 1.4430458970213842,
      "learning_rate": 1.4117454789208673e-09,
      "loss": 0.6662,
      "step": 13529
    },
    {
      "epoch": 0.9926905545823894,
      "grad_norm": 1.5914437330087943,
      "learning_rate": 1.383653039695454e-09,
      "loss": 0.7859,
      "step": 13530
    },
    {
      "epoch": 0.9927639241725286,
      "grad_norm": 1.4631057985972846,
      "learning_rate": 1.3558428847137983e-09,
      "loss": 0.7609,
      "step": 13531
    },
    {
      "epoch": 0.9928372937626677,
      "grad_norm": 0.5554790699302832,
      "learning_rate": 1.3283150155468661e-09,
      "loss": 0.4845,
      "step": 13532
    },
    {
      "epoch": 0.9929106633528069,
      "grad_norm": 0.5320994661673382,
      "learning_rate": 1.3010694337495244e-09,
      "loss": 0.4771,
      "step": 13533
    },
    {
      "epoch": 0.992984032942946,
      "grad_norm": 1.702244683057557,
      "learning_rate": 1.2741061408594324e-09,
      "loss": 0.754,
      "step": 13534
    },
    {
      "epoch": 0.993057402533085,
      "grad_norm": 1.466376422748083,
      "learning_rate": 1.247425138399261e-09,
      "loss": 0.6365,
      "step": 13535
    },
    {
      "epoch": 0.9931307721232242,
      "grad_norm": 1.3118533084975144,
      "learning_rate": 1.2210264278766926e-09,
      "loss": 0.6677,
      "step": 13536
    },
    {
      "epoch": 0.9931307721232242,
      "lm_loss": 1.4823,
      "step": 13536,
      "vm_loss": 0.0767
    },
    {
      "epoch": 0.9931307721232242,
      "lm_loss": 0.2521,
      "step": 13536,
      "vm_loss": 0.0775
    },
    {
      "epoch": 0.9931307721232242,
      "lm_loss": 0.7139,
      "step": 13536,
      "vm_loss": 0.1765
    },
    {
      "epoch": 0.9931307721232242,
      "lm_loss": 0.3036,
      "step": 13536,
      "vm_loss": 0.071
    },
    {
      "epoch": 0.9931307721232242,
      "lm_loss": 0.2052,
      "step": 13536,
      "vm_loss": 0.1108
    },
    {
      "epoch": 0.9931307721232242,
      "lm_loss": 0.4822,
      "step": 13536,
      "vm_loss": 0.1262
    },
    {
      "epoch": 0.9931307721232242,
      "lm_loss": 0.21,
      "step": 13536,
      "vm_loss": 0.145
    },
    {
      "epoch": 0.9931307721232242,
      "lm_loss": 0.6999,
      "step": 13536,
      "vm_loss": 0.0864
    },
    {
      "epoch": 0.9932041417133634,
      "grad_norm": 1.5415336115304359,
      "learning_rate": 1.194910010781647e-09,
      "loss": 0.7307,
      "step": 13537
    },
    {
      "epoch": 0.9932775113035025,
      "grad_norm": 1.61516761286585,
      "learning_rate": 1.1690758885896103e-09,
      "loss": 0.7851,
      "step": 13538
    },
    {
      "epoch": 0.9933508808936417,
      "grad_norm": 0.5355936571926724,
      "learning_rate": 1.1435240627594157e-09,
      "loss": 0.4646,
      "step": 13539
    },
    {
      "epoch": 0.9934242504837807,
      "grad_norm": 1.5801993020133804,
      "learning_rate": 1.1182545347337982e-09,
      "loss": 0.7877,
      "step": 13540
    },
    {
      "epoch": 0.9934976200739198,
      "grad_norm": 0.5393447799615986,
      "learning_rate": 1.093267305939949e-09,
      "loss": 0.4743,
      "step": 13541
    },
    {
      "epoch": 0.993570989664059,
      "grad_norm": 1.4325938086542895,
      "learning_rate": 1.0685623777884068e-09,
      "loss": 0.7183,
      "step": 13542
    },
    {
      "epoch": 0.9936443592541981,
      "grad_norm": 1.4388283107020832,
      "learning_rate": 1.0441397516752771e-09,
      "loss": 0.733,
      "step": 13543
    },
    {
      "epoch": 0.9937177288443373,
      "grad_norm": 1.3467128668207105,
      "learning_rate": 1.0199994289794568e-09,
      "loss": 0.6986,
      "step": 13544
    },
    {
      "epoch": 0.9937177288443373,
      "lm_loss": 1.4795,
      "step": 13544,
      "vm_loss": 0.1652
    },
    {
      "epoch": 0.9937177288443373,
      "lm_loss": 0.4762,
      "step": 13544,
      "vm_loss": 0.1788
    },
    {
      "epoch": 0.9937177288443373,
      "lm_loss": 0.0849,
      "step": 13544,
      "vm_loss": 0.1034
    },
    {
      "epoch": 0.9937177288443373,
      "lm_loss": 1.1067,
      "step": 13544,
      "vm_loss": 0.196
    },
    {
      "epoch": 0.9937177288443373,
      "lm_loss": 0.3892,
      "step": 13544,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.9937177288443373,
      "lm_loss": 0.4575,
      "step": 13544,
      "vm_loss": 0.1894
    },
    {
      "epoch": 0.9937177288443373,
      "lm_loss": 0.67,
      "step": 13544,
      "vm_loss": 0.0824
    },
    {
      "epoch": 0.9937177288443373,
      "lm_loss": 0.858,
      "step": 13544,
      "vm_loss": 0.0973
    },
    {
      "epoch": 0.9937910984344763,
      "grad_norm": 1.4088672397554562,
      "learning_rate": 9.961414110637447e-10,
      "loss": 0.6958,
      "step": 13545
    },
    {
      "epoch": 0.9938644680246155,
      "grad_norm": 1.6392009479477523,
      "learning_rate": 9.725656992753961e-10,
      "loss": 0.7204,
      "step": 13546
    },
    {
      "epoch": 0.9939378376147546,
      "grad_norm": 1.3969424707461728,
      "learning_rate": 9.492722949466793e-10,
      "loss": 0.7773,
      "step": 13547
    },
    {
      "epoch": 0.9940112072048938,
      "grad_norm": 0.5350569135219323,
      "learning_rate": 9.262611993920978e-10,
      "loss": 0.4904,
      "step": 13548
    },
    {
      "epoch": 0.9940845767950329,
      "grad_norm": 1.4474114218724798,
      "learning_rate": 9.03532413911723e-10,
      "loss": 0.7287,
      "step": 13549
    },
    {
      "epoch": 0.994157946385172,
      "grad_norm": 1.5879949399421613,
      "learning_rate": 8.810859397889726e-10,
      "loss": 0.7242,
      "step": 13550
    },
    {
      "epoch": 0.9942313159753111,
      "grad_norm": 0.5348855502882304,
      "learning_rate": 8.589217782911663e-10,
      "loss": 0.4751,
      "step": 13551
    },
    {
      "epoch": 0.9943046855654503,
      "grad_norm": 1.4915250146942765,
      "learning_rate": 8.370399306706356e-10,
      "loss": 0.7911,
      "step": 13552
    },
    {
      "epoch": 0.9943046855654503,
      "lm_loss": 0.3144,
      "step": 13552,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.9943046855654503,
      "lm_loss": 0.7341,
      "step": 13552,
      "vm_loss": 0.2152
    },
    {
      "epoch": 0.9943046855654503,
      "lm_loss": 0.1022,
      "step": 13552,
      "vm_loss": 0.1258
    },
    {
      "epoch": 0.9943046855654503,
      "lm_loss": 0.6871,
      "step": 13552,
      "vm_loss": 0.2594
    },
    {
      "epoch": 0.9943046855654503,
      "lm_loss": 0.073,
      "step": 13552,
      "vm_loss": 0.1048
    },
    {
      "epoch": 0.9943046855654503,
      "lm_loss": 0.6269,
      "step": 13552,
      "vm_loss": 0.0683
    },
    {
      "epoch": 0.9943046855654503,
      "lm_loss": 0.6421,
      "step": 13552,
      "vm_loss": 0.1627
    },
    {
      "epoch": 0.9943046855654503,
      "lm_loss": 1.213,
      "step": 13552,
      "vm_loss": 0.139
    },
    {
      "epoch": 0.9943780551555894,
      "grad_norm": 1.4239271544562675,
      "learning_rate": 8.154403981625036e-10,
      "loss": 0.7213,
      "step": 13553
    },
    {
      "epoch": 0.9944514247457286,
      "grad_norm": 1.446729293191747,
      "learning_rate": 7.941231819863504e-10,
      "loss": 0.7249,
      "step": 13554
    },
    {
      "epoch": 0.9945247943358676,
      "grad_norm": 1.5030255045937917,
      "learning_rate": 7.730882833467679e-10,
      "loss": 0.8626,
      "step": 13555
    },
    {
      "epoch": 0.9945981639260068,
      "grad_norm": 0.5315532582620435,
      "learning_rate": 7.523357034311396e-10,
      "loss": 0.4598,
      "step": 13556
    },
    {
      "epoch": 0.9946715335161459,
      "grad_norm": 1.658019138580148,
      "learning_rate": 7.31865443411861e-10,
      "loss": 0.7078,
      "step": 13557
    },
    {
      "epoch": 0.994744903106285,
      "grad_norm": 0.5223776846750753,
      "learning_rate": 7.116775044441193e-10,
      "loss": 0.448,
      "step": 13558
    },
    {
      "epoch": 0.9948182726964242,
      "grad_norm": 1.49159133958308,
      "learning_rate": 6.917718876686685e-10,
      "loss": 0.782,
      "step": 13559
    },
    {
      "epoch": 0.9948916422865632,
      "grad_norm": 0.5276495474541186,
      "learning_rate": 6.721485942096095e-10,
      "loss": 0.461,
      "step": 13560
    },
    {
      "epoch": 0.9948916422865632,
      "lm_loss": 1.0479,
      "step": 13560,
      "vm_loss": 0.0335
    },
    {
      "epoch": 0.9948916422865632,
      "lm_loss": 1.1857,
      "step": 13560,
      "vm_loss": 0.0853
    },
    {
      "epoch": 0.9948916422865632,
      "lm_loss": 0.8057,
      "step": 13560,
      "vm_loss": 0.1506
    },
    {
      "epoch": 0.9948916422865632,
      "lm_loss": 0.5798,
      "step": 13560,
      "vm_loss": 0.1902
    },
    {
      "epoch": 0.9948916422865632,
      "lm_loss": 0.8414,
      "step": 13560,
      "vm_loss": 0.1626
    },
    {
      "epoch": 0.9948916422865632,
      "lm_loss": 0.7873,
      "step": 13560,
      "vm_loss": 0.1801
    },
    {
      "epoch": 0.9948916422865632,
      "lm_loss": 0.4021,
      "step": 13560,
      "vm_loss": 0.1311
    },
    {
      "epoch": 0.9948916422865632,
      "lm_loss": 0.5517,
      "step": 13560,
      "vm_loss": 0.1301
    },
    {
      "epoch": 0.9949650118767024,
      "grad_norm": 1.5573185983884577,
      "learning_rate": 6.528076251749449e-10,
      "loss": 0.7851,
      "step": 13561
    },
    {
      "epoch": 0.9950383814668415,
      "grad_norm": 1.5460894838051167,
      "learning_rate": 6.337489816565789e-10,
      "loss": 0.7637,
      "step": 13562
    },
    {
      "epoch": 0.9951117510569807,
      "grad_norm": 0.5329393717042297,
      "learning_rate": 6.14972664731428e-10,
      "loss": 0.477,
      "step": 13563
    },
    {
      "epoch": 0.9951851206471198,
      "grad_norm": 0.5402541078717764,
      "learning_rate": 5.964786754597551e-10,
      "loss": 0.4722,
      "step": 13564
    },
    {
      "epoch": 0.9952584902372589,
      "grad_norm": 1.7062103354425024,
      "learning_rate": 5.782670148851699e-10,
      "loss": 0.7929,
      "step": 13565
    },
    {
      "epoch": 0.995331859827398,
      "grad_norm": 1.489000070560673,
      "learning_rate": 5.603376840368491e-10,
      "loss": 0.7186,
      "step": 13566
    },
    {
      "epoch": 0.9954052294175372,
      "grad_norm": 0.557182410057373,
      "learning_rate": 5.426906839273161e-10,
      "loss": 0.5083,
      "step": 13567
    },
    {
      "epoch": 0.9954785990076763,
      "grad_norm": 0.5337140174044858,
      "learning_rate": 5.25326015552996e-10,
      "loss": 0.4576,
      "step": 13568
    },
    {
      "epoch": 0.9954785990076763,
      "lm_loss": 1.1399,
      "step": 13568,
      "vm_loss": 0.0285
    },
    {
      "epoch": 0.9954785990076763,
      "lm_loss": 1.7199,
      "step": 13568,
      "vm_loss": 0.0881
    },
    {
      "epoch": 0.9954785990076763,
      "lm_loss": 0.0707,
      "step": 13568,
      "vm_loss": 0.2662
    },
    {
      "epoch": 0.9954785990076763,
      "lm_loss": 0.8045,
      "step": 13568,
      "vm_loss": 0.1471
    },
    {
      "epoch": 0.9954785990076763,
      "lm_loss": 0.6893,
      "step": 13568,
      "vm_loss": 0.2353
    },
    {
      "epoch": 0.9954785990076763,
      "lm_loss": 0.0518,
      "step": 13568,
      "vm_loss": 0.0902
    },
    {
      "epoch": 0.9954785990076763,
      "lm_loss": 0.6588,
      "step": 13568,
      "vm_loss": 0.1497
    },
    {
      "epoch": 0.9954785990076763,
      "lm_loss": 0.0377,
      "step": 13568,
      "vm_loss": 0.0837
    },
    {
      "epoch": 0.9955519685978155,
      "grad_norm": 1.5199371357198581,
      "learning_rate": 5.08243679894771e-10,
      "loss": 0.7956,
      "step": 13569
    },
    {
      "epoch": 0.9956253381879545,
      "grad_norm": 0.5422975464101729,
      "learning_rate": 4.914436779168697e-10,
      "loss": 0.4755,
      "step": 13570
    },
    {
      "epoch": 0.9956987077780937,
      "grad_norm": 0.5329043330599506,
      "learning_rate": 4.749260105679776e-10,
      "loss": 0.4751,
      "step": 13571
    },
    {
      "epoch": 0.9957720773682328,
      "grad_norm": 1.6192690889435382,
      "learning_rate": 4.586906787812373e-10,
      "loss": 0.6984,
      "step": 13572
    },
    {
      "epoch": 0.995845446958372,
      "grad_norm": 1.5460053108949683,
      "learning_rate": 4.4273768347313785e-10,
      "loss": 0.7627,
      "step": 13573
    },
    {
      "epoch": 0.9959188165485111,
      "grad_norm": 1.4775309364596372,
      "learning_rate": 4.270670255451803e-10,
      "loss": 0.7299,
      "step": 13574
    },
    {
      "epoch": 0.9959921861386501,
      "grad_norm": 1.6298392721144046,
      "learning_rate": 4.1167870588221246e-10,
      "loss": 0.6765,
      "step": 13575
    },
    {
      "epoch": 0.9960655557287893,
      "grad_norm": 0.5403389218196271,
      "learning_rate": 3.965727253524287e-10,
      "loss": 0.4789,
      "step": 13576
    },
    {
      "epoch": 0.9960655557287893,
      "lm_loss": 1.5252,
      "step": 13576,
      "vm_loss": 0.0647
    },
    {
      "epoch": 0.9960655557287893,
      "lm_loss": 0.7419,
      "step": 13576,
      "vm_loss": 0.127
    },
    {
      "epoch": 0.9960655557287893,
      "lm_loss": 1.2607,
      "step": 13576,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.9960655557287893,
      "lm_loss": 1.0065,
      "step": 13576,
      "vm_loss": 0.0713
    },
    {
      "epoch": 0.9960655557287893,
      "lm_loss": 0.1431,
      "step": 13576,
      "vm_loss": 0.1381
    },
    {
      "epoch": 0.9960655557287893,
      "lm_loss": 0.6421,
      "step": 13576,
      "vm_loss": 0.1722
    },
    {
      "epoch": 0.9960655557287893,
      "lm_loss": 1.0683,
      "step": 13576,
      "vm_loss": 0.0995
    },
    {
      "epoch": 0.9960655557287893,
      "lm_loss": 0.1408,
      "step": 13576,
      "vm_loss": 0.1366
    },
    {
      "epoch": 0.9961389253189284,
      "grad_norm": 1.7023208432230064,
      "learning_rate": 3.8174908480959063e-10,
      "loss": 0.7803,
      "step": 13577
    },
    {
      "epoch": 0.9962122949090676,
      "grad_norm": 0.5362775662510302,
      "learning_rate": 3.6720778509136135e-10,
      "loss": 0.4504,
      "step": 13578
    },
    {
      "epoch": 0.9962856644992067,
      "grad_norm": 1.528905153598782,
      "learning_rate": 3.5294882701764064e-10,
      "loss": 0.7103,
      "step": 13579
    },
    {
      "epoch": 0.9963590340893458,
      "grad_norm": 0.5343310872245478,
      "learning_rate": 3.3897221139445047e-10,
      "loss": 0.4585,
      "step": 13580
    },
    {
      "epoch": 0.9964324036794849,
      "grad_norm": 0.5439847745472327,
      "learning_rate": 3.2527793901115936e-10,
      "loss": 0.4756,
      "step": 13581
    },
    {
      "epoch": 0.9965057732696241,
      "grad_norm": 1.5742597798608924,
      "learning_rate": 3.1186601064048247e-10,
      "loss": 0.7515,
      "step": 13582
    },
    {
      "epoch": 0.9965791428597632,
      "grad_norm": 1.5118270435094716,
      "learning_rate": 2.987364270407023e-10,
      "loss": 0.7625,
      "step": 13583
    },
    {
      "epoch": 0.9966525124499023,
      "grad_norm": 1.5596317970146683,
      "learning_rate": 2.8588918895289253e-10,
      "loss": 0.7276,
      "step": 13584
    },
    {
      "epoch": 0.9966525124499023,
      "lm_loss": 2.3527,
      "step": 13584,
      "vm_loss": 0.1006
    },
    {
      "epoch": 0.9966525124499023,
      "lm_loss": 0.9455,
      "step": 13584,
      "vm_loss": 0.12
    },
    {
      "epoch": 0.9966525124499023,
      "lm_loss": 0.0169,
      "step": 13584,
      "vm_loss": 0.0898
    },
    {
      "epoch": 0.9966525124499023,
      "lm_loss": 0.2994,
      "step": 13584,
      "vm_loss": 0.1686
    },
    {
      "epoch": 0.9966525124499023,
      "lm_loss": 0.4196,
      "step": 13584,
      "vm_loss": 0.0933
    },
    {
      "epoch": 0.9966525124499023,
      "lm_loss": 0.0038,
      "step": 13584,
      "vm_loss": 0.0953
    },
    {
      "epoch": 0.9966525124499023,
      "lm_loss": 0.8428,
      "step": 13584,
      "vm_loss": 0.0963
    },
    {
      "epoch": 0.9966525124499023,
      "lm_loss": 0.5727,
      "step": 13584,
      "vm_loss": 0.1502
    },
    {
      "epoch": 0.9967258820400414,
      "grad_norm": 2.0505526729593684,
      "learning_rate": 2.733242971020289e-10,
      "loss": 0.7724,
      "step": 13585
    },
    {
      "epoch": 0.9967992516301806,
      "grad_norm": 0.5697370690886213,
      "learning_rate": 2.6104175219865413e-10,
      "loss": 0.4779,
      "step": 13586
    },
    {
      "epoch": 0.9968726212203197,
      "grad_norm": 1.5645645808795132,
      "learning_rate": 2.490415549355474e-10,
      "loss": 0.7117,
      "step": 13587
    },
    {
      "epoch": 0.9969459908104589,
      "grad_norm": 1.6509180386176225,
      "learning_rate": 2.3732370599105494e-10,
      "loss": 0.6977,
      "step": 13588
    },
    {
      "epoch": 0.9970193604005979,
      "grad_norm": 0.5628872045025505,
      "learning_rate": 2.2588820602631457e-10,
      "loss": 0.4774,
      "step": 13589
    },
    {
      "epoch": 0.9970927299907371,
      "grad_norm": 1.576312401673759,
      "learning_rate": 2.147350556874761e-10,
      "loss": 0.7209,
      "step": 13590
    },
    {
      "epoch": 0.9971660995808762,
      "grad_norm": 1.3953602127189557,
      "learning_rate": 2.03864255604036e-10,
      "loss": 0.6497,
      "step": 13591
    },
    {
      "epoch": 0.9972394691710154,
      "grad_norm": 1.3989023370267137,
      "learning_rate": 1.9327580639050269e-10,
      "loss": 0.8205,
      "step": 13592
    },
    {
      "epoch": 0.9972394691710154,
      "lm_loss": 0.5351,
      "step": 13592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9972394691710154,
      "lm_loss": 0.4238,
      "step": 13592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9972394691710154,
      "lm_loss": 0.504,
      "step": 13592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9972394691710154,
      "lm_loss": 0.4239,
      "step": 13592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9972394691710154,
      "lm_loss": 0.5248,
      "step": 13592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9972394691710154,
      "lm_loss": 0.4831,
      "step": 13592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9972394691710154,
      "lm_loss": 0.3816,
      "step": 13592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9972394691710154,
      "lm_loss": 0.4046,
      "step": 13592,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9973128387611545,
      "grad_norm": 0.535010964454687,
      "learning_rate": 1.8296970864417618e-10,
      "loss": 0.4607,
      "step": 13593
    },
    {
      "epoch": 0.9973862083512935,
      "grad_norm": 1.4534446812525994,
      "learning_rate": 1.7294596294792354e-10,
      "loss": 0.7306,
      "step": 13594
    },
    {
      "epoch": 0.9974595779414327,
      "grad_norm": 1.4688441840198874,
      "learning_rate": 1.632045698668483e-10,
      "loss": 0.7715,
      "step": 13595
    },
    {
      "epoch": 0.9975329475315718,
      "grad_norm": 0.5322067077612906,
      "learning_rate": 1.5374552995162105e-10,
      "loss": 0.4713,
      "step": 13596
    },
    {
      "epoch": 0.997606317121711,
      "grad_norm": 1.484538736198197,
      "learning_rate": 1.4456884373625913e-10,
      "loss": 0.7927,
      "step": 13597
    },
    {
      "epoch": 0.9976796867118501,
      "grad_norm": 1.4800845307016612,
      "learning_rate": 1.3567451173923661e-10,
      "loss": 0.6037,
      "step": 13598
    },
    {
      "epoch": 0.9977530563019892,
      "grad_norm": 1.6285376303960728,
      "learning_rate": 1.2706253446237437e-10,
      "loss": 0.7732,
      "step": 13599
    },
    {
      "epoch": 0.9978264258921283,
      "grad_norm": 1.9651600980353807,
      "learning_rate": 1.1873291239250518e-10,
      "loss": 0.7495,
      "step": 13600
    },
    {
      "epoch": 0.9978264258921283,
      "lm_loss": 0.6434,
      "step": 13600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9978264258921283,
      "lm_loss": 0.3939,
      "step": 13600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9978264258921283,
      "lm_loss": 0.4248,
      "step": 13600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9978264258921283,
      "lm_loss": 0.4603,
      "step": 13600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9978264258921283,
      "lm_loss": 0.3374,
      "step": 13600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9978264258921283,
      "lm_loss": 0.4133,
      "step": 13600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9978264258921283,
      "lm_loss": 0.3152,
      "step": 13600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9978264258921283,
      "lm_loss": 0.4911,
      "step": 13600,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9978997954822675,
      "grad_norm": 0.5518995921040681,
      "learning_rate": 1.1068564599980847e-10,
      "loss": 0.46,
      "step": 13601
    },
    {
      "epoch": 0.9979731650724066,
      "grad_norm": 1.4364205457415449,
      "learning_rate": 1.0292073573836548e-10,
      "loss": 0.7543,
      "step": 13602
    },
    {
      "epoch": 0.9980465346625458,
      "grad_norm": 1.5510316036712188,
      "learning_rate": 9.54381820472694e-11,
      "loss": 0.7094,
      "step": 13603
    },
    {
      "epoch": 0.9981199042526848,
      "grad_norm": 1.5827765691838838,
      "learning_rate": 8.8237985348405e-11,
      "loss": 0.7116,
      "step": 13604
    },
    {
      "epoch": 0.998193273842824,
      "grad_norm": 1.5115077156244112,
      "learning_rate": 8.132014604922411e-11,
      "loss": 0.7344,
      "step": 13605
    },
    {
      "epoch": 0.9982666434329631,
      "grad_norm": 1.509014817114492,
      "learning_rate": 7.468466453997014e-11,
      "loss": 0.7634,
      "step": 13606
    },
    {
      "epoch": 0.9983400130231023,
      "grad_norm": 1.5642087817852708,
      "learning_rate": 6.833154119534336e-11,
      "loss": 0.6526,
      "step": 13607
    },
    {
      "epoch": 0.9984133826132414,
      "grad_norm": 2.097567083662179,
      "learning_rate": 6.22607763745009e-11,
      "loss": 0.6924,
      "step": 13608
    },
    {
      "epoch": 0.9984133826132414,
      "lm_loss": 0.5333,
      "step": 13608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9984133826132414,
      "lm_loss": 0.3808,
      "step": 13608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9984133826132414,
      "lm_loss": 0.5979,
      "step": 13608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9984133826132414,
      "lm_loss": 0.5436,
      "step": 13608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9984133826132414,
      "lm_loss": 0.4502,
      "step": 13608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9984133826132414,
      "lm_loss": 0.3666,
      "step": 13608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9984133826132414,
      "lm_loss": 0.5509,
      "step": 13608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9984133826132414,
      "lm_loss": 0.4519,
      "step": 13608,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9984867522033805,
      "grad_norm": 0.5380437786995331,
      "learning_rate": 5.6472370419391464e-11,
      "loss": 0.4505,
      "step": 13609
    },
    {
      "epoch": 0.9985601217935196,
      "grad_norm": 1.5150124023534934,
      "learning_rate": 5.096632365753085e-11,
      "loss": 0.7562,
      "step": 13610
    },
    {
      "epoch": 0.9986334913836588,
      "grad_norm": 1.6132400238884073,
      "learning_rate": 4.5742636399781494e-11,
      "loss": 0.7124,
      "step": 13611
    },
    {
      "epoch": 0.9987068609737979,
      "grad_norm": 1.6064094632133867,
      "learning_rate": 4.0801308941462726e-11,
      "loss": 0.7562,
      "step": 13612
    },
    {
      "epoch": 0.998780230563937,
      "grad_norm": 1.483702640922607,
      "learning_rate": 3.6142341561240525e-11,
      "loss": 0.6957,
      "step": 13613
    },
    {
      "epoch": 0.9988536001540761,
      "grad_norm": 1.5850759190490582,
      "learning_rate": 3.1765734521682634e-11,
      "loss": 0.6337,
      "step": 13614
    },
    {
      "epoch": 0.9989269697442152,
      "grad_norm": 0.5773770025827901,
      "learning_rate": 2.7671488070923903e-11,
      "loss": 0.494,
      "step": 13615
    },
    {
      "epoch": 0.9990003393343544,
      "grad_norm": 0.5688713397466723,
      "learning_rate": 2.3859602439890716e-11,
      "loss": 0.4717,
      "step": 13616
    },
    {
      "epoch": 0.9990003393343544,
      "lm_loss": 0.3181,
      "step": 13616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9990003393343544,
      "lm_loss": 0.4133,
      "step": 13616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9990003393343544,
      "lm_loss": 0.4562,
      "step": 13616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9990003393343544,
      "lm_loss": 0.3643,
      "step": 13616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9990003393343544,
      "lm_loss": 0.5102,
      "step": 13616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9990003393343544,
      "lm_loss": 0.5026,
      "step": 13616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9990003393343544,
      "lm_loss": 0.586,
      "step": 13616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9990003393343544,
      "lm_loss": 0.6052,
      "step": 13616,
      "vm_loss": 0.0
    },
    {
      "epoch": 0.9990737089244935,
      "grad_norm": 0.5381208322996899,
      "learning_rate": 2.0330077843411232e-11,
      "loss": 0.4706,
      "step": 13617
    },
    {
      "epoch": 0.9991470785146327,
      "grad_norm": 1.4446404678285472,
      "learning_rate": 1.7082914481325596e-11,
      "loss": 0.6949,
      "step": 13618
    },
    {
      "epoch": 0.9992204481047717,
      "grad_norm": 1.4845432009987194,
      "learning_rate": 1.4118112536820605e-11,
      "loss": 0.6677,
      "step": 13619
    },
    {
      "epoch": 0.9992938176949109,
      "grad_norm": 1.6719389052257871,
      "learning_rate": 1.1435672177539936e-11,
      "loss": 0.7167,
      "step": 13620
    },
    {
      "epoch": 0.99936718728505,
      "grad_norm": 1.6481530686870907,
      "learning_rate": 9.03559355447392e-12,
      "loss": 0.6698,
      "step": 13621
    },
    {
      "epoch": 0.9994405568751892,
      "grad_norm": 1.5095677510740895,
      "learning_rate": 6.917876803624879e-12,
      "loss": 0.8373,
      "step": 13622
    },
    {
      "epoch": 0.9995139264653283,
      "grad_norm": 0.5419956015652062,
      "learning_rate": 5.082522044341787e-12,
      "loss": 0.4924,
      "step": 13623
    },
    {
      "epoch": 0.9995872960554674,
      "grad_norm": 1.3703678440316887,
      "learning_rate": 3.529529379875385e-12,
      "loss": 0.6471,
      "step": 13624
    },
    {
      "epoch": 0.9995872960554674,
      "lm_loss": 1.4283,
      "step": 13624,
      "vm_loss": 0.0245
    },
    {
      "epoch": 0.9995872960554674,
      "lm_loss": 1.0721,
      "step": 13624,
      "vm_loss": 0.0896
    },
    {
      "epoch": 0.9995872960554674,
      "lm_loss": 0.5663,
      "step": 13624,
      "vm_loss": 0.1909
    },
    {
      "epoch": 0.9995872960554674,
      "lm_loss": 0.0214,
      "step": 13624,
      "vm_loss": 0.1159
    },
    {
      "epoch": 0.9995872960554674,
      "lm_loss": 0.6223,
      "step": 13624,
      "vm_loss": 0.1153
    },
    {
      "epoch": 0.9995872960554674,
      "lm_loss": 0.1105,
      "step": 13624,
      "vm_loss": 0.077
    },
    {
      "epoch": 0.9995872960554674,
      "lm_loss": 0.3707,
      "step": 13624,
      "vm_loss": 0.0802
    },
    {
      "epoch": 0.9995872960554674,
      "lm_loss": 0.3849,
      "step": 13624,
      "vm_loss": 0.1255
    },
    {
      "epoch": 0.9996606656456065,
      "grad_norm": 1.6231006365099303,
      "learning_rate": 2.2588988990435156e-12,
      "loss": 0.7096,
      "step": 13625
    },
    {
      "epoch": 0.9997340352357457,
      "grad_norm": 1.512783794596358,
      "learning_rate": 1.270630672345341e-12,
      "loss": 0.7398,
      "step": 13626
    },
    {
      "epoch": 0.9998074048258848,
      "grad_norm": 0.5504413484435591,
      "learning_rate": 5.647247564022351e-13,
      "loss": 0.4831,
      "step": 13627
    },
    {
      "epoch": 0.999880774416024,
      "grad_norm": 1.5116325090481457,
      "learning_rate": 1.4118119118222695e-13,
      "loss": 0.6786,
      "step": 13628
    },
    {
      "epoch": 0.999954144006163,
      "grad_norm": 0.5674280051804622,
      "learning_rate": 0.0,
      "loss": 0.4786,
      "step": 13629
    },
    {
      "epoch": 0.999954144006163,
      "step": 13629,
      "total_flos": 9.745692120868454e+16,
      "train_loss": 0.7232364302086842,
      "train_runtime": 314600.026,
      "train_samples_per_second": 11.091,
      "train_steps_per_second": 0.043
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 13629,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 13628,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.745692120868454e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}