{
  "best_metric": 11.500072479248047,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.5961251862891207,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0029806259314456036,
      "grad_norm": 0.7637186050415039,
      "learning_rate": 1e-05,
      "loss": 11.9191,
      "step": 1
    },
    {
      "epoch": 0.0029806259314456036,
      "eval_loss": 11.92214298248291,
      "eval_runtime": 5.7791,
      "eval_samples_per_second": 97.765,
      "eval_steps_per_second": 24.571,
      "step": 1
    },
    {
      "epoch": 0.005961251862891207,
      "grad_norm": 0.6532636284828186,
      "learning_rate": 2e-05,
      "loss": 11.9169,
      "step": 2
    },
    {
      "epoch": 0.00894187779433681,
      "grad_norm": 0.6132882237434387,
      "learning_rate": 3e-05,
      "loss": 11.9157,
      "step": 3
    },
    {
      "epoch": 0.011922503725782414,
      "grad_norm": 0.7796972393989563,
      "learning_rate": 4e-05,
      "loss": 11.9163,
      "step": 4
    },
    {
      "epoch": 0.014903129657228018,
      "grad_norm": 0.6557272672653198,
      "learning_rate": 5e-05,
      "loss": 11.9202,
      "step": 5
    },
    {
      "epoch": 0.01788375558867362,
      "grad_norm": 0.6318836212158203,
      "learning_rate": 6e-05,
      "loss": 11.9138,
      "step": 6
    },
    {
      "epoch": 0.020864381520119227,
      "grad_norm": 0.6313462257385254,
      "learning_rate": 7e-05,
      "loss": 11.9177,
      "step": 7
    },
    {
      "epoch": 0.02384500745156483,
      "grad_norm": 0.6250380277633667,
      "learning_rate": 8e-05,
      "loss": 11.9123,
      "step": 8
    },
    {
      "epoch": 0.026825633383010434,
      "grad_norm": 0.6615362763404846,
      "learning_rate": 9e-05,
      "loss": 11.9051,
      "step": 9
    },
    {
      "epoch": 0.029806259314456036,
      "grad_norm": 0.7824218273162842,
      "learning_rate": 0.0001,
      "loss": 11.8986,
      "step": 10
    },
    {
      "epoch": 0.03278688524590164,
      "grad_norm": 0.7486230731010437,
      "learning_rate": 9.999316524962345e-05,
      "loss": 11.8925,
      "step": 11
    },
    {
      "epoch": 0.03576751117734724,
      "grad_norm": 0.7275559902191162,
      "learning_rate": 9.997266286704631e-05,
      "loss": 11.8877,
      "step": 12
    },
    {
      "epoch": 0.038748137108792845,
      "grad_norm": 0.7845178842544556,
      "learning_rate": 9.993849845741524e-05,
      "loss": 11.8765,
      "step": 13
    },
    {
      "epoch": 0.041728763040238454,
      "grad_norm": 0.7134093642234802,
      "learning_rate": 9.989068136093873e-05,
      "loss": 11.8714,
      "step": 14
    },
    {
      "epoch": 0.044709388971684055,
      "grad_norm": 0.7441971898078918,
      "learning_rate": 9.98292246503335e-05,
      "loss": 11.8617,
      "step": 15
    },
    {
      "epoch": 0.04769001490312966,
      "grad_norm": 0.6720972061157227,
      "learning_rate": 9.975414512725057e-05,
      "loss": 11.8598,
      "step": 16
    },
    {
      "epoch": 0.05067064083457526,
      "grad_norm": 0.6425601243972778,
      "learning_rate": 9.966546331768191e-05,
      "loss": 11.8499,
      "step": 17
    },
    {
      "epoch": 0.05365126676602087,
      "grad_norm": 0.6544840931892395,
      "learning_rate": 9.956320346634876e-05,
      "loss": 11.8418,
      "step": 18
    },
    {
      "epoch": 0.05663189269746647,
      "grad_norm": 0.5775944590568542,
      "learning_rate": 9.944739353007344e-05,
      "loss": 11.8356,
      "step": 19
    },
    {
      "epoch": 0.05961251862891207,
      "grad_norm": 0.5665213465690613,
      "learning_rate": 9.931806517013612e-05,
      "loss": 11.831,
      "step": 20
    },
    {
      "epoch": 0.06259314456035768,
      "grad_norm": 0.5018438696861267,
      "learning_rate": 9.917525374361912e-05,
      "loss": 11.8249,
      "step": 21
    },
    {
      "epoch": 0.06557377049180328,
      "grad_norm": 0.49436721205711365,
      "learning_rate": 9.901899829374047e-05,
      "loss": 11.8179,
      "step": 22
    },
    {
      "epoch": 0.06855439642324888,
      "grad_norm": 0.462222158908844,
      "learning_rate": 9.884934153917997e-05,
      "loss": 11.8142,
      "step": 23
    },
    {
      "epoch": 0.07153502235469449,
      "grad_norm": 0.46007296442985535,
      "learning_rate": 9.86663298624003e-05,
      "loss": 11.8061,
      "step": 24
    },
    {
      "epoch": 0.07451564828614009,
      "grad_norm": 0.4221821129322052,
      "learning_rate": 9.847001329696653e-05,
      "loss": 11.8107,
      "step": 25
    },
    {
      "epoch": 0.07749627421758569,
      "grad_norm": 0.3891777992248535,
      "learning_rate": 9.826044551386744e-05,
      "loss": 11.8014,
      "step": 26
    },
    {
      "epoch": 0.08047690014903129,
      "grad_norm": 0.37388280034065247,
      "learning_rate": 9.803768380684242e-05,
      "loss": 11.8019,
      "step": 27
    },
    {
      "epoch": 0.08345752608047691,
      "grad_norm": 0.38052019476890564,
      "learning_rate": 9.780178907671789e-05,
      "loss": 11.7944,
      "step": 28
    },
    {
      "epoch": 0.08643815201192251,
      "grad_norm": 0.37809059023857117,
      "learning_rate": 9.755282581475769e-05,
      "loss": 11.7918,
      "step": 29
    },
    {
      "epoch": 0.08941877794336811,
      "grad_norm": 0.3687851130962372,
      "learning_rate": 9.729086208503174e-05,
      "loss": 11.7889,
      "step": 30
    },
    {
      "epoch": 0.09239940387481371,
      "grad_norm": 0.32611191272735596,
      "learning_rate": 9.701596950580806e-05,
      "loss": 11.787,
      "step": 31
    },
    {
      "epoch": 0.09538002980625931,
      "grad_norm": 0.31833168864250183,
      "learning_rate": 9.672822322997305e-05,
      "loss": 11.778,
      "step": 32
    },
    {
      "epoch": 0.09836065573770492,
      "grad_norm": 0.3484291434288025,
      "learning_rate": 9.642770192448536e-05,
      "loss": 11.7858,
      "step": 33
    },
    {
      "epoch": 0.10134128166915052,
      "grad_norm": 0.32301127910614014,
      "learning_rate": 9.611448774886924e-05,
      "loss": 11.7825,
      "step": 34
    },
    {
      "epoch": 0.10432190760059612,
      "grad_norm": 0.34433430433273315,
      "learning_rate": 9.578866633275288e-05,
      "loss": 11.7792,
      "step": 35
    },
    {
      "epoch": 0.10730253353204174,
      "grad_norm": 0.36491772532463074,
      "learning_rate": 9.545032675245813e-05,
      "loss": 11.7766,
      "step": 36
    },
    {
      "epoch": 0.11028315946348734,
      "grad_norm": 0.3507254421710968,
      "learning_rate": 9.509956150664796e-05,
      "loss": 11.77,
      "step": 37
    },
    {
      "epoch": 0.11326378539493294,
      "grad_norm": 0.3505198061466217,
      "learning_rate": 9.473646649103818e-05,
      "loss": 11.7624,
      "step": 38
    },
    {
      "epoch": 0.11624441132637854,
      "grad_norm": 0.41520798206329346,
      "learning_rate": 9.43611409721806e-05,
      "loss": 11.7581,
      "step": 39
    },
    {
      "epoch": 0.11922503725782414,
      "grad_norm": 0.36798131465911865,
      "learning_rate": 9.397368756032445e-05,
      "loss": 11.7534,
      "step": 40
    },
    {
      "epoch": 0.12220566318926974,
      "grad_norm": 0.3230534791946411,
      "learning_rate": 9.357421218136386e-05,
      "loss": 11.7502,
      "step": 41
    },
    {
      "epoch": 0.12518628912071536,
      "grad_norm": 0.3596799373626709,
      "learning_rate": 9.316282404787871e-05,
      "loss": 11.7458,
      "step": 42
    },
    {
      "epoch": 0.12816691505216096,
      "grad_norm": 0.3787568509578705,
      "learning_rate": 9.273963562927695e-05,
      "loss": 11.7466,
      "step": 43
    },
    {
      "epoch": 0.13114754098360656,
      "grad_norm": 0.34623926877975464,
      "learning_rate": 9.230476262104677e-05,
      "loss": 11.738,
      "step": 44
    },
    {
      "epoch": 0.13412816691505217,
      "grad_norm": 0.3480023443698883,
      "learning_rate": 9.185832391312644e-05,
      "loss": 11.7292,
      "step": 45
    },
    {
      "epoch": 0.13710879284649777,
      "grad_norm": 0.34748515486717224,
      "learning_rate": 9.140044155740101e-05,
      "loss": 11.7259,
      "step": 46
    },
    {
      "epoch": 0.14008941877794337,
      "grad_norm": 0.36501240730285645,
      "learning_rate": 9.093124073433463e-05,
      "loss": 11.7246,
      "step": 47
    },
    {
      "epoch": 0.14307004470938897,
      "grad_norm": 0.3235962986946106,
      "learning_rate": 9.045084971874738e-05,
      "loss": 11.7132,
      "step": 48
    },
    {
      "epoch": 0.14605067064083457,
      "grad_norm": 0.29502999782562256,
      "learning_rate": 8.995939984474624e-05,
      "loss": 11.7135,
      "step": 49
    },
    {
      "epoch": 0.14903129657228018,
      "grad_norm": 0.5087769627571106,
      "learning_rate": 8.945702546981969e-05,
      "loss": 11.7054,
      "step": 50
    },
    {
      "epoch": 0.14903129657228018,
      "eval_loss": 11.707014083862305,
      "eval_runtime": 5.6054,
      "eval_samples_per_second": 100.796,
      "eval_steps_per_second": 25.333,
      "step": 50
    },
    {
      "epoch": 0.15201192250372578,
      "grad_norm": 0.6833786964416504,
      "learning_rate": 8.894386393810563e-05,
      "loss": 11.7033,
      "step": 51
    },
    {
      "epoch": 0.15499254843517138,
      "grad_norm": 0.3737204968929291,
      "learning_rate": 8.842005554284296e-05,
      "loss": 11.7182,
      "step": 52
    },
    {
      "epoch": 0.15797317436661698,
      "grad_norm": 0.2742517590522766,
      "learning_rate": 8.788574348801675e-05,
      "loss": 11.7064,
      "step": 53
    },
    {
      "epoch": 0.16095380029806258,
      "grad_norm": 0.38709139823913574,
      "learning_rate": 8.73410738492077e-05,
      "loss": 11.6838,
      "step": 54
    },
    {
      "epoch": 0.16393442622950818,
      "grad_norm": 0.2635756731033325,
      "learning_rate": 8.678619553365659e-05,
      "loss": 11.6956,
      "step": 55
    },
    {
      "epoch": 0.16691505216095381,
      "grad_norm": 0.256094366312027,
      "learning_rate": 8.622126023955446e-05,
      "loss": 11.7079,
      "step": 56
    },
    {
      "epoch": 0.16989567809239942,
      "grad_norm": 0.22819863259792328,
      "learning_rate": 8.564642241456986e-05,
      "loss": 11.6969,
      "step": 57
    },
    {
      "epoch": 0.17287630402384502,
      "grad_norm": 0.21696770191192627,
      "learning_rate": 8.506183921362443e-05,
      "loss": 11.685,
      "step": 58
    },
    {
      "epoch": 0.17585692995529062,
      "grad_norm": 0.23800955712795258,
      "learning_rate": 8.44676704559283e-05,
      "loss": 11.6767,
      "step": 59
    },
    {
      "epoch": 0.17883755588673622,
      "grad_norm": 0.2353067696094513,
      "learning_rate": 8.386407858128706e-05,
      "loss": 11.6679,
      "step": 60
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.2418794184923172,
      "learning_rate": 8.32512286056924e-05,
      "loss": 11.6745,
      "step": 61
    },
    {
      "epoch": 0.18479880774962743,
      "grad_norm": 0.23060879111289978,
      "learning_rate": 8.262928807620843e-05,
      "loss": 11.6568,
      "step": 62
    },
    {
      "epoch": 0.18777943368107303,
      "grad_norm": 0.19210241734981537,
      "learning_rate": 8.199842702516583e-05,
      "loss": 11.6549,
      "step": 63
    },
    {
      "epoch": 0.19076005961251863,
      "grad_norm": 0.29771679639816284,
      "learning_rate": 8.135881792367686e-05,
      "loss": 11.6719,
      "step": 64
    },
    {
      "epoch": 0.19374068554396423,
      "grad_norm": 0.21918320655822754,
      "learning_rate": 8.07106356344834e-05,
      "loss": 11.6534,
      "step": 65
    },
    {
      "epoch": 0.19672131147540983,
      "grad_norm": 0.20529407262802124,
      "learning_rate": 8.005405736415126e-05,
      "loss": 11.6535,
      "step": 66
    },
    {
      "epoch": 0.19970193740685543,
      "grad_norm": 0.2387327402830124,
      "learning_rate": 7.938926261462366e-05,
      "loss": 11.6392,
      "step": 67
    },
    {
      "epoch": 0.20268256333830104,
      "grad_norm": 0.1831187605857849,
      "learning_rate": 7.871643313414718e-05,
      "loss": 11.6419,
      "step": 68
    },
    {
      "epoch": 0.20566318926974664,
      "grad_norm": 0.20405712723731995,
      "learning_rate": 7.803575286758364e-05,
      "loss": 11.6266,
      "step": 69
    },
    {
      "epoch": 0.20864381520119224,
      "grad_norm": 0.20329934358596802,
      "learning_rate": 7.734740790612136e-05,
      "loss": 11.6344,
      "step": 70
    },
    {
      "epoch": 0.21162444113263784,
      "grad_norm": 0.18024663627147675,
      "learning_rate": 7.66515864363997e-05,
      "loss": 11.6267,
      "step": 71
    },
    {
      "epoch": 0.21460506706408347,
      "grad_norm": 0.20047903060913086,
      "learning_rate": 7.594847868906076e-05,
      "loss": 11.6106,
      "step": 72
    },
    {
      "epoch": 0.21758569299552907,
      "grad_norm": 0.22485877573490143,
      "learning_rate": 7.52382768867422e-05,
      "loss": 11.6094,
      "step": 73
    },
    {
      "epoch": 0.22056631892697467,
      "grad_norm": 0.2008683979511261,
      "learning_rate": 7.452117519152542e-05,
      "loss": 11.6076,
      "step": 74
    },
    {
      "epoch": 0.22354694485842028,
      "grad_norm": 0.17285442352294922,
      "learning_rate": 7.379736965185368e-05,
      "loss": 11.6061,
      "step": 75
    },
    {
      "epoch": 0.22652757078986588,
      "grad_norm": 0.23781253397464752,
      "learning_rate": 7.30670581489344e-05,
      "loss": 11.5934,
      "step": 76
    },
    {
      "epoch": 0.22950819672131148,
      "grad_norm": 0.1844850331544876,
      "learning_rate": 7.233044034264034e-05,
      "loss": 11.591,
      "step": 77
    },
    {
      "epoch": 0.23248882265275708,
      "grad_norm": 0.18350204825401306,
      "learning_rate": 7.158771761692464e-05,
      "loss": 11.5831,
      "step": 78
    },
    {
      "epoch": 0.23546944858420268,
      "grad_norm": 0.21184353530406952,
      "learning_rate": 7.083909302476453e-05,
      "loss": 11.586,
      "step": 79
    },
    {
      "epoch": 0.23845007451564829,
      "grad_norm": 0.2409575879573822,
      "learning_rate": 7.008477123264848e-05,
      "loss": 11.5685,
      "step": 80
    },
    {
      "epoch": 0.2414307004470939,
      "grad_norm": 0.20296820998191833,
      "learning_rate": 6.932495846462261e-05,
      "loss": 11.579,
      "step": 81
    },
    {
      "epoch": 0.2444113263785395,
      "grad_norm": 0.18394796550273895,
      "learning_rate": 6.855986244591104e-05,
      "loss": 11.5817,
      "step": 82
    },
    {
      "epoch": 0.2473919523099851,
      "grad_norm": 0.18706998229026794,
      "learning_rate": 6.778969234612584e-05,
      "loss": 11.5767,
      "step": 83
    },
    {
      "epoch": 0.2503725782414307,
      "grad_norm": 0.17890408635139465,
      "learning_rate": 6.701465872208216e-05,
      "loss": 11.5776,
      "step": 84
    },
    {
      "epoch": 0.2533532041728763,
      "grad_norm": 0.19777363538742065,
      "learning_rate": 6.623497346023418e-05,
      "loss": 11.5688,
      "step": 85
    },
    {
      "epoch": 0.2563338301043219,
      "grad_norm": 0.20829612016677856,
      "learning_rate": 6.545084971874738e-05,
      "loss": 11.578,
      "step": 86
    },
    {
      "epoch": 0.2593144560357675,
      "grad_norm": 0.16336099803447723,
      "learning_rate": 6.466250186922325e-05,
      "loss": 11.5655,
      "step": 87
    },
    {
      "epoch": 0.26229508196721313,
      "grad_norm": 0.2072550654411316,
      "learning_rate": 6.387014543809223e-05,
      "loss": 11.5685,
      "step": 88
    },
    {
      "epoch": 0.26527570789865873,
      "grad_norm": 0.17955659329891205,
      "learning_rate": 6.307399704769099e-05,
      "loss": 11.5658,
      "step": 89
    },
    {
      "epoch": 0.26825633383010433,
      "grad_norm": 0.15350516140460968,
      "learning_rate": 6.227427435703997e-05,
      "loss": 11.5537,
      "step": 90
    },
    {
      "epoch": 0.27123695976154993,
      "grad_norm": 0.18274275958538055,
      "learning_rate": 6.147119600233758e-05,
      "loss": 11.5443,
      "step": 91
    },
    {
      "epoch": 0.27421758569299554,
      "grad_norm": 0.15808114409446716,
      "learning_rate": 6.066498153718735e-05,
      "loss": 11.5532,
      "step": 92
    },
    {
      "epoch": 0.27719821162444114,
      "grad_norm": 0.22299650311470032,
      "learning_rate": 5.985585137257401e-05,
      "loss": 11.5499,
      "step": 93
    },
    {
      "epoch": 0.28017883755588674,
      "grad_norm": 0.21931232511997223,
      "learning_rate": 5.90440267166055e-05,
      "loss": 11.5491,
      "step": 94
    },
    {
      "epoch": 0.28315946348733234,
      "grad_norm": 0.3318580985069275,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 11.5497,
      "step": 95
    },
    {
      "epoch": 0.28614008941877794,
      "grad_norm": 0.22443974018096924,
      "learning_rate": 5.74131823855921e-05,
      "loss": 11.55,
      "step": 96
    },
    {
      "epoch": 0.28912071535022354,
      "grad_norm": 0.15305203199386597,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 11.5487,
      "step": 97
    },
    {
      "epoch": 0.29210134128166915,
      "grad_norm": 0.18104498088359833,
      "learning_rate": 5.577423184847932e-05,
      "loss": 11.5338,
      "step": 98
    },
    {
      "epoch": 0.29508196721311475,
      "grad_norm": 0.16398631036281586,
      "learning_rate": 5.495227651252315e-05,
      "loss": 11.5182,
      "step": 99
    },
    {
      "epoch": 0.29806259314456035,
      "grad_norm": 0.2035548985004425,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 11.5277,
      "step": 100
    },
    {
      "epoch": 0.29806259314456035,
      "eval_loss": 11.540384292602539,
      "eval_runtime": 5.6146,
      "eval_samples_per_second": 100.63,
      "eval_steps_per_second": 25.291,
      "step": 100
    },
    {
      "epoch": 0.30104321907600595,
      "grad_norm": 0.5054342150688171,
      "learning_rate": 5.330452921628497e-05,
      "loss": 11.5431,
      "step": 101
    },
    {
      "epoch": 0.30402384500745155,
      "grad_norm": 0.27865058183670044,
      "learning_rate": 5.247918773366112e-05,
      "loss": 11.5917,
      "step": 102
    },
    {
      "epoch": 0.30700447093889716,
      "grad_norm": 0.2700931429862976,
      "learning_rate": 5.165316846586541e-05,
      "loss": 11.5851,
      "step": 103
    },
    {
      "epoch": 0.30998509687034276,
      "grad_norm": 0.22124922275543213,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 11.5659,
      "step": 104
    },
    {
      "epoch": 0.31296572280178836,
      "grad_norm": 0.21236014366149902,
      "learning_rate": 5e-05,
      "loss": 11.529,
      "step": 105
    },
    {
      "epoch": 0.31594634873323396,
      "grad_norm": 0.1851605325937271,
      "learning_rate": 4.917330276168208e-05,
      "loss": 11.56,
      "step": 106
    },
    {
      "epoch": 0.31892697466467956,
      "grad_norm": 0.24979141354560852,
      "learning_rate": 4.834683153413459e-05,
      "loss": 11.565,
      "step": 107
    },
    {
      "epoch": 0.32190760059612517,
      "grad_norm": 0.1909526288509369,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 11.5583,
      "step": 108
    },
    {
      "epoch": 0.32488822652757077,
      "grad_norm": 0.19361820816993713,
      "learning_rate": 4.669547078371504e-05,
      "loss": 11.5451,
      "step": 109
    },
    {
      "epoch": 0.32786885245901637,
      "grad_norm": 0.21884016692638397,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 11.543,
      "step": 110
    },
    {
      "epoch": 0.33084947839046197,
      "grad_norm": 0.1892264038324356,
      "learning_rate": 4.504772348747687e-05,
      "loss": 11.5483,
      "step": 111
    },
    {
      "epoch": 0.33383010432190763,
      "grad_norm": 0.1816645711660385,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 11.5348,
      "step": 112
    },
    {
      "epoch": 0.33681073025335323,
      "grad_norm": 0.15595115721225739,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 11.5297,
      "step": 113
    },
    {
      "epoch": 0.33979135618479883,
      "grad_norm": 0.14673824608325958,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 11.5318,
      "step": 114
    },
    {
      "epoch": 0.34277198211624443,
      "grad_norm": 0.15732240676879883,
      "learning_rate": 4.17702704859633e-05,
      "loss": 11.5463,
      "step": 115
    },
    {
      "epoch": 0.34575260804769004,
      "grad_norm": 0.19519706070423126,
      "learning_rate": 4.095597328339452e-05,
      "loss": 11.5438,
      "step": 116
    },
    {
      "epoch": 0.34873323397913564,
      "grad_norm": 0.15034478902816772,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 11.5321,
      "step": 117
    },
    {
      "epoch": 0.35171385991058124,
      "grad_norm": 0.21822240948677063,
      "learning_rate": 3.933501846281267e-05,
      "loss": 11.5061,
      "step": 118
    },
    {
      "epoch": 0.35469448584202684,
      "grad_norm": 0.14199988543987274,
      "learning_rate": 3.852880399766243e-05,
      "loss": 11.5451,
      "step": 119
    },
    {
      "epoch": 0.35767511177347244,
      "grad_norm": 0.1756989061832428,
      "learning_rate": 3.772572564296005e-05,
      "loss": 11.522,
      "step": 120
    },
    {
      "epoch": 0.36065573770491804,
      "grad_norm": 0.17190110683441162,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 11.5216,
      "step": 121
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.14261436462402344,
      "learning_rate": 3.612985456190778e-05,
      "loss": 11.5245,
      "step": 122
    },
    {
      "epoch": 0.36661698956780925,
      "grad_norm": 0.1700989305973053,
      "learning_rate": 3.533749813077677e-05,
      "loss": 11.5033,
      "step": 123
    },
    {
      "epoch": 0.36959761549925485,
      "grad_norm": 0.14692789316177368,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 11.5126,
      "step": 124
    },
    {
      "epoch": 0.37257824143070045,
      "grad_norm": 0.14157594740390778,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 11.5113,
      "step": 125
    },
    {
      "epoch": 0.37555886736214605,
      "grad_norm": 0.14132097363471985,
      "learning_rate": 3.298534127791785e-05,
      "loss": 11.5042,
      "step": 126
    },
    {
      "epoch": 0.37853949329359166,
      "grad_norm": 0.1447451114654541,
      "learning_rate": 3.221030765387417e-05,
      "loss": 11.5097,
      "step": 127
    },
    {
      "epoch": 0.38152011922503726,
      "grad_norm": 0.13442525267601013,
      "learning_rate": 3.144013755408895e-05,
      "loss": 11.507,
      "step": 128
    },
    {
      "epoch": 0.38450074515648286,
      "grad_norm": 0.13315923511981964,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 11.4993,
      "step": 129
    },
    {
      "epoch": 0.38748137108792846,
      "grad_norm": 0.13803797960281372,
      "learning_rate": 2.991522876735154e-05,
      "loss": 11.4959,
      "step": 130
    },
    {
      "epoch": 0.39046199701937406,
      "grad_norm": 0.17156541347503662,
      "learning_rate": 2.916090697523549e-05,
      "loss": 11.5019,
      "step": 131
    },
    {
      "epoch": 0.39344262295081966,
      "grad_norm": 0.15189988911151886,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 11.5115,
      "step": 132
    },
    {
      "epoch": 0.39642324888226527,
      "grad_norm": 0.14400018751621246,
      "learning_rate": 2.766955965735968e-05,
      "loss": 11.5088,
      "step": 133
    },
    {
      "epoch": 0.39940387481371087,
      "grad_norm": 0.16279402375221252,
      "learning_rate": 2.693294185106562e-05,
      "loss": 11.5146,
      "step": 134
    },
    {
      "epoch": 0.40238450074515647,
      "grad_norm": 0.1359529346227646,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 11.4951,
      "step": 135
    },
    {
      "epoch": 0.40536512667660207,
      "grad_norm": 0.1630541831254959,
      "learning_rate": 2.547882480847461e-05,
      "loss": 11.4992,
      "step": 136
    },
    {
      "epoch": 0.4083457526080477,
      "grad_norm": 0.14012819528579712,
      "learning_rate": 2.476172311325783e-05,
      "loss": 11.5079,
      "step": 137
    },
    {
      "epoch": 0.4113263785394933,
      "grad_norm": 0.16439418494701385,
      "learning_rate": 2.405152131093926e-05,
      "loss": 11.496,
      "step": 138
    },
    {
      "epoch": 0.4143070044709389,
      "grad_norm": 0.13019785284996033,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 11.5034,
      "step": 139
    },
    {
      "epoch": 0.4172876304023845,
      "grad_norm": 0.1702735871076584,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 11.4945,
      "step": 140
    },
    {
      "epoch": 0.4202682563338301,
      "grad_norm": 0.16818252205848694,
      "learning_rate": 2.196424713241637e-05,
      "loss": 11.4981,
      "step": 141
    },
    {
      "epoch": 0.4232488822652757,
      "grad_norm": 0.13469691574573517,
      "learning_rate": 2.128356686585282e-05,
      "loss": 11.5004,
      "step": 142
    },
    {
      "epoch": 0.4262295081967213,
      "grad_norm": 0.15116199851036072,
      "learning_rate": 2.061073738537635e-05,
      "loss": 11.506,
      "step": 143
    },
    {
      "epoch": 0.42921013412816694,
      "grad_norm": 0.1493290662765503,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 11.5033,
      "step": 144
    },
    {
      "epoch": 0.43219076005961254,
      "grad_norm": 0.147053524851799,
      "learning_rate": 1.928936436551661e-05,
      "loss": 11.4964,
      "step": 145
    },
    {
      "epoch": 0.43517138599105815,
      "grad_norm": 0.19523854553699493,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 11.4983,
      "step": 146
    },
    {
      "epoch": 0.43815201192250375,
      "grad_norm": 0.18568909168243408,
      "learning_rate": 1.800157297483417e-05,
      "loss": 11.4926,
      "step": 147
    },
    {
      "epoch": 0.44113263785394935,
      "grad_norm": 0.16512249410152435,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 11.4944,
      "step": 148
    },
    {
      "epoch": 0.44411326378539495,
      "grad_norm": 0.1522430181503296,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 11.4959,
      "step": 149
    },
    {
      "epoch": 0.44709388971684055,
      "grad_norm": 0.1610044687986374,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 11.4846,
      "step": 150
    },
    {
      "epoch": 0.44709388971684055,
      "eval_loss": 11.503910064697266,
      "eval_runtime": 5.8673,
      "eval_samples_per_second": 96.296,
      "eval_steps_per_second": 24.202,
      "step": 150
    },
    {
      "epoch": 0.45007451564828616,
      "grad_norm": 0.42695456743240356,
      "learning_rate": 1.553232954407171e-05,
      "loss": 11.5182,
      "step": 151
    },
    {
      "epoch": 0.45305514157973176,
      "grad_norm": 0.2571699917316437,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 11.5436,
      "step": 152
    },
    {
      "epoch": 0.45603576751117736,
      "grad_norm": 0.24285241961479187,
      "learning_rate": 1.435357758543015e-05,
      "loss": 11.527,
      "step": 153
    },
    {
      "epoch": 0.45901639344262296,
      "grad_norm": 0.2131449580192566,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 11.5139,
      "step": 154
    },
    {
      "epoch": 0.46199701937406856,
      "grad_norm": 0.17757467925548553,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 11.5234,
      "step": 155
    },
    {
      "epoch": 0.46497764530551416,
      "grad_norm": 0.17251507937908173,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 11.5401,
      "step": 156
    },
    {
      "epoch": 0.46795827123695977,
      "grad_norm": 0.1669883131980896,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 11.5307,
      "step": 157
    },
    {
      "epoch": 0.47093889716840537,
      "grad_norm": 0.15614424645900726,
      "learning_rate": 1.157994445715706e-05,
      "loss": 11.5261,
      "step": 158
    },
    {
      "epoch": 0.47391952309985097,
      "grad_norm": 0.13682828843593597,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 11.5188,
      "step": 159
    },
    {
      "epoch": 0.47690014903129657,
      "grad_norm": 0.17259274423122406,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 11.5068,
      "step": 160
    },
    {
      "epoch": 0.4798807749627422,
      "grad_norm": 0.14623887836933136,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 11.5073,
      "step": 161
    },
    {
      "epoch": 0.4828614008941878,
      "grad_norm": 0.14129942655563354,
      "learning_rate": 9.549150281252633e-06,
      "loss": 11.5071,
      "step": 162
    },
    {
      "epoch": 0.4858420268256334,
      "grad_norm": 0.12252959609031677,
      "learning_rate": 9.068759265665384e-06,
      "loss": 11.5243,
      "step": 163
    },
    {
      "epoch": 0.488822652757079,
      "grad_norm": 0.13987858593463898,
      "learning_rate": 8.599558442598998e-06,
      "loss": 11.5232,
      "step": 164
    },
    {
      "epoch": 0.4918032786885246,
      "grad_norm": 0.15460610389709473,
      "learning_rate": 8.141676086873572e-06,
      "loss": 11.5081,
      "step": 165
    },
    {
      "epoch": 0.4947839046199702,
      "grad_norm": 0.13415540754795074,
      "learning_rate": 7.695237378953223e-06,
      "loss": 11.5063,
      "step": 166
    },
    {
      "epoch": 0.4977645305514158,
      "grad_norm": 0.1525568962097168,
      "learning_rate": 7.260364370723044e-06,
      "loss": 11.5237,
      "step": 167
    },
    {
      "epoch": 0.5007451564828614,
      "grad_norm": 0.11544854938983917,
      "learning_rate": 6.837175952121306e-06,
      "loss": 11.5101,
      "step": 168
    },
    {
      "epoch": 0.503725782414307,
      "grad_norm": 0.12548111379146576,
      "learning_rate": 6.425787818636131e-06,
      "loss": 11.4975,
      "step": 169
    },
    {
      "epoch": 0.5067064083457526,
      "grad_norm": 0.1126219853758812,
      "learning_rate": 6.026312439675552e-06,
      "loss": 11.5035,
      "step": 170
    },
    {
      "epoch": 0.5096870342771982,
      "grad_norm": 0.11540310084819794,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 11.4977,
      "step": 171
    },
    {
      "epoch": 0.5126676602086438,
      "grad_norm": 0.13346481323242188,
      "learning_rate": 5.263533508961827e-06,
      "loss": 11.4874,
      "step": 172
    },
    {
      "epoch": 0.5156482861400894,
      "grad_norm": 0.21232837438583374,
      "learning_rate": 4.900438493352055e-06,
      "loss": 11.5091,
      "step": 173
    },
    {
      "epoch": 0.518628912071535,
      "grad_norm": 0.1193055659532547,
      "learning_rate": 4.549673247541875e-06,
      "loss": 11.4966,
      "step": 174
    },
    {
      "epoch": 0.5216095380029806,
      "grad_norm": 0.13836126029491425,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 11.5067,
      "step": 175
    },
    {
      "epoch": 0.5245901639344263,
      "grad_norm": 0.1375090777873993,
      "learning_rate": 3.885512251130763e-06,
      "loss": 11.496,
      "step": 176
    },
    {
      "epoch": 0.5275707898658718,
      "grad_norm": 0.155953049659729,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 11.5012,
      "step": 177
    },
    {
      "epoch": 0.5305514157973175,
      "grad_norm": 0.13382163643836975,
      "learning_rate": 3.271776770026963e-06,
      "loss": 11.4917,
      "step": 178
    },
    {
      "epoch": 0.533532041728763,
      "grad_norm": 0.13938270509243011,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 11.4833,
      "step": 179
    },
    {
      "epoch": 0.5365126676602087,
      "grad_norm": 0.18127211928367615,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 11.4946,
      "step": 180
    },
    {
      "epoch": 0.5394932935916542,
      "grad_norm": 0.14780205488204956,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 11.4785,
      "step": 181
    },
    {
      "epoch": 0.5424739195230999,
      "grad_norm": 0.1618468016386032,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 11.4822,
      "step": 182
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.12401887774467468,
      "learning_rate": 1.962316193157593e-06,
      "loss": 11.4968,
      "step": 183
    },
    {
      "epoch": 0.5484351713859911,
      "grad_norm": 0.1382862776517868,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 11.4804,
      "step": 184
    },
    {
      "epoch": 0.5514157973174366,
      "grad_norm": 0.1842343509197235,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 11.5095,
      "step": 185
    },
    {
      "epoch": 0.5543964232488823,
      "grad_norm": 0.1770644634962082,
      "learning_rate": 1.333670137599713e-06,
      "loss": 11.5015,
      "step": 186
    },
    {
      "epoch": 0.5573770491803278,
      "grad_norm": 0.19411513209342957,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 11.4982,
      "step": 187
    },
    {
      "epoch": 0.5603576751117735,
      "grad_norm": 0.15455012023448944,
      "learning_rate": 9.810017062595322e-07,
      "loss": 11.4922,
      "step": 188
    },
    {
      "epoch": 0.563338301043219,
      "grad_norm": 0.20202553272247314,
      "learning_rate": 8.247462563808817e-07,
      "loss": 11.4819,
      "step": 189
    },
    {
      "epoch": 0.5663189269746647,
      "grad_norm": 0.21089744567871094,
      "learning_rate": 6.819348298638839e-07,
      "loss": 11.5076,
      "step": 190
    },
    {
      "epoch": 0.5692995529061102,
      "grad_norm": 0.1702708750963211,
      "learning_rate": 5.526064699265753e-07,
      "loss": 11.4968,
      "step": 191
    },
    {
      "epoch": 0.5722801788375559,
      "grad_norm": 0.17010365426540375,
      "learning_rate": 4.367965336512403e-07,
      "loss": 11.5067,
      "step": 192
    },
    {
      "epoch": 0.5752608047690015,
      "grad_norm": 0.14179681241512299,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 11.4925,
      "step": 193
    },
    {
      "epoch": 0.5782414307004471,
      "grad_norm": 0.14533187448978424,
      "learning_rate": 2.458548727494292e-07,
      "loss": 11.4847,
      "step": 194
    },
    {
      "epoch": 0.5812220566318927,
      "grad_norm": 0.1771630197763443,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 11.4908,
      "step": 195
    },
    {
      "epoch": 0.5842026825633383,
      "grad_norm": 0.17438878118991852,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 11.4905,
      "step": 196
    },
    {
      "epoch": 0.587183308494784,
      "grad_norm": 0.19026672840118408,
      "learning_rate": 6.150154258476315e-08,
      "loss": 11.4881,
      "step": 197
    },
    {
      "epoch": 0.5901639344262295,
      "grad_norm": 0.1565513163805008,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 11.4841,
      "step": 198
    },
    {
      "epoch": 0.5931445603576752,
      "grad_norm": 0.2124798595905304,
      "learning_rate": 6.834750376549792e-09,
      "loss": 11.4944,
      "step": 199
    },
    {
      "epoch": 0.5961251862891207,
      "grad_norm": 0.27714475989341736,
      "learning_rate": 0.0,
      "loss": 11.4892,
      "step": 200
    },
    {
      "epoch": 0.5961251862891207,
      "eval_loss": 11.500072479248047,
      "eval_runtime": 5.7035,
      "eval_samples_per_second": 99.061,
      "eval_steps_per_second": 24.897,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 220358246400000.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}