{
  "best_metric": NaN,
  "best_model_checkpoint": "miner_id_24/checkpoint-50",
  "epoch": 0.19895548371051977,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000994777418552599,
      "grad_norm": 0.04143854230642319,
      "learning_rate": 1e-05,
      "loss": 10.3813,
      "step": 1
    },
    {
      "epoch": 0.000994777418552599,
      "eval_loss": NaN,
      "eval_runtime": 3.754,
      "eval_samples_per_second": 450.983,
      "eval_steps_per_second": 112.946,
      "step": 1
    },
    {
      "epoch": 0.001989554837105198,
      "grad_norm": 0.042226918041706085,
      "learning_rate": 2e-05,
      "loss": 10.3807,
      "step": 2
    },
    {
      "epoch": 0.0029843322556577967,
      "grad_norm": 0.05301538482308388,
      "learning_rate": 3e-05,
      "loss": 10.381,
      "step": 3
    },
    {
      "epoch": 0.003979109674210396,
      "grad_norm": 0.04115965962409973,
      "learning_rate": 4e-05,
      "loss": 10.3781,
      "step": 4
    },
    {
      "epoch": 0.004973887092762994,
      "grad_norm": 0.0458495132625103,
      "learning_rate": 5e-05,
      "loss": 10.3812,
      "step": 5
    },
    {
      "epoch": 0.005968664511315593,
      "grad_norm": 0.043900974094867706,
      "learning_rate": 6e-05,
      "loss": 10.379,
      "step": 6
    },
    {
      "epoch": 0.006963441929868192,
      "grad_norm": 0.04340410977602005,
      "learning_rate": 7e-05,
      "loss": 10.3781,
      "step": 7
    },
    {
      "epoch": 0.007958219348420791,
      "grad_norm": 0.045883286744356155,
      "learning_rate": 8e-05,
      "loss": 10.3794,
      "step": 8
    },
    {
      "epoch": 0.008952996766973389,
      "grad_norm": 0.045194365084171295,
      "learning_rate": 9e-05,
      "loss": 10.3816,
      "step": 9
    },
    {
      "epoch": 0.009947774185525988,
      "grad_norm": 0.047754883766174316,
      "learning_rate": 0.0001,
      "loss": 10.3791,
      "step": 10
    },
    {
      "epoch": 0.010942551604078588,
      "grad_norm": 0.0554577112197876,
      "learning_rate": 9.999316524962345e-05,
      "loss": 10.3803,
      "step": 11
    },
    {
      "epoch": 0.011937329022631187,
      "grad_norm": 0.05751480907201767,
      "learning_rate": 9.997266286704631e-05,
      "loss": 10.3781,
      "step": 12
    },
    {
      "epoch": 0.012932106441183784,
      "grad_norm": 0.055258236825466156,
      "learning_rate": 9.993849845741524e-05,
      "loss": 10.377,
      "step": 13
    },
    {
      "epoch": 0.013926883859736384,
      "grad_norm": 0.05144908279180527,
      "learning_rate": 9.989068136093873e-05,
      "loss": 10.3758,
      "step": 14
    },
    {
      "epoch": 0.014921661278288983,
      "grad_norm": 0.0627208799123764,
      "learning_rate": 9.98292246503335e-05,
      "loss": 10.3784,
      "step": 15
    },
    {
      "epoch": 0.015916438696841582,
      "grad_norm": 0.06269817054271698,
      "learning_rate": 9.975414512725057e-05,
      "loss": 10.377,
      "step": 16
    },
    {
      "epoch": 0.01691121611539418,
      "grad_norm": 0.0643804520368576,
      "learning_rate": 9.966546331768191e-05,
      "loss": 10.376,
      "step": 17
    },
    {
      "epoch": 0.017905993533946778,
      "grad_norm": 0.07872988283634186,
      "learning_rate": 9.956320346634876e-05,
      "loss": 10.3741,
      "step": 18
    },
    {
      "epoch": 0.01890077095249938,
      "grad_norm": 0.09896330535411835,
      "learning_rate": 9.944739353007344e-05,
      "loss": 10.3723,
      "step": 19
    },
    {
      "epoch": 0.019895548371051976,
      "grad_norm": 0.09100668877363205,
      "learning_rate": 9.931806517013612e-05,
      "loss": 10.3702,
      "step": 20
    },
    {
      "epoch": 0.020890325789604577,
      "grad_norm": 0.11379483342170715,
      "learning_rate": 9.917525374361912e-05,
      "loss": 10.3709,
      "step": 21
    },
    {
      "epoch": 0.021885103208157175,
      "grad_norm": 0.0907195657491684,
      "learning_rate": 9.901899829374047e-05,
      "loss": 10.3721,
      "step": 22
    },
    {
      "epoch": 0.022879880626709773,
      "grad_norm": 0.10841047018766403,
      "learning_rate": 9.884934153917997e-05,
      "loss": 10.3696,
      "step": 23
    },
    {
      "epoch": 0.023874658045262374,
      "grad_norm": 0.13297222554683685,
      "learning_rate": 9.86663298624003e-05,
      "loss": 10.3672,
      "step": 24
    },
    {
      "epoch": 0.02486943546381497,
      "grad_norm": 0.11716096103191376,
      "learning_rate": 9.847001329696653e-05,
      "loss": 10.368,
      "step": 25
    },
    {
      "epoch": 0.02586421288236757,
      "grad_norm": 0.10968994349241257,
      "learning_rate": 9.826044551386744e-05,
      "loss": 10.3638,
      "step": 26
    },
    {
      "epoch": 0.02685899030092017,
      "grad_norm": 0.11097239702939987,
      "learning_rate": 9.803768380684242e-05,
      "loss": 10.3681,
      "step": 27
    },
    {
      "epoch": 0.027853767719472768,
      "grad_norm": 0.13693346083164215,
      "learning_rate": 9.780178907671789e-05,
      "loss": 10.3665,
      "step": 28
    },
    {
      "epoch": 0.028848545138025365,
      "grad_norm": 0.10598407685756683,
      "learning_rate": 9.755282581475769e-05,
      "loss": 10.3631,
      "step": 29
    },
    {
      "epoch": 0.029843322556577966,
      "grad_norm": 0.15644285082817078,
      "learning_rate": 9.729086208503174e-05,
      "loss": 10.3604,
      "step": 30
    },
    {
      "epoch": 0.030838099975130564,
      "grad_norm": 0.14097367227077484,
      "learning_rate": 9.701596950580806e-05,
      "loss": 10.3605,
      "step": 31
    },
    {
      "epoch": 0.031832877393683165,
      "grad_norm": 0.1672372817993164,
      "learning_rate": 9.672822322997305e-05,
      "loss": 10.3608,
      "step": 32
    },
    {
      "epoch": 0.03282765481223576,
      "grad_norm": 0.15272119641304016,
      "learning_rate": 9.642770192448536e-05,
      "loss": 10.3575,
      "step": 33
    },
    {
      "epoch": 0.03382243223078836,
      "grad_norm": 0.12398381531238556,
      "learning_rate": 9.611448774886924e-05,
      "loss": 10.3592,
      "step": 34
    },
    {
      "epoch": 0.03481720964934096,
      "grad_norm": 0.1535937786102295,
      "learning_rate": 9.578866633275288e-05,
      "loss": 10.3565,
      "step": 35
    },
    {
      "epoch": 0.035811987067893555,
      "grad_norm": 0.16945576667785645,
      "learning_rate": 9.545032675245813e-05,
      "loss": 10.3582,
      "step": 36
    },
    {
      "epoch": 0.03680676448644616,
      "grad_norm": 0.1461615264415741,
      "learning_rate": 9.509956150664796e-05,
      "loss": 10.3525,
      "step": 37
    },
    {
      "epoch": 0.03780154190499876,
      "grad_norm": 0.15130968391895294,
      "learning_rate": 9.473646649103818e-05,
      "loss": 10.3561,
      "step": 38
    },
    {
      "epoch": 0.038796319323551355,
      "grad_norm": 0.15193091332912445,
      "learning_rate": 9.43611409721806e-05,
      "loss": 10.3522,
      "step": 39
    },
    {
      "epoch": 0.03979109674210395,
      "grad_norm": 0.10653501003980637,
      "learning_rate": 9.397368756032445e-05,
      "loss": 10.3512,
      "step": 40
    },
    {
      "epoch": 0.04078587416065655,
      "grad_norm": 0.12497211247682571,
      "learning_rate": 9.357421218136386e-05,
      "loss": 10.3509,
      "step": 41
    },
    {
      "epoch": 0.041780651579209155,
      "grad_norm": 0.133211150765419,
      "learning_rate": 9.316282404787871e-05,
      "loss": 10.3494,
      "step": 42
    },
    {
      "epoch": 0.04277542899776175,
      "grad_norm": 0.17113818228244781,
      "learning_rate": 9.273963562927695e-05,
      "loss": 10.3492,
      "step": 43
    },
    {
      "epoch": 0.04377020641631435,
      "grad_norm": 0.15438786149024963,
      "learning_rate": 9.230476262104677e-05,
      "loss": 10.3528,
      "step": 44
    },
    {
      "epoch": 0.04476498383486695,
      "grad_norm": 0.14064650237560272,
      "learning_rate": 9.185832391312644e-05,
      "loss": 10.3518,
      "step": 45
    },
    {
      "epoch": 0.045759761253419545,
      "grad_norm": 0.17576594650745392,
      "learning_rate": 9.140044155740101e-05,
      "loss": 10.3512,
      "step": 46
    },
    {
      "epoch": 0.04675453867197214,
      "grad_norm": 0.16118112206459045,
      "learning_rate": 9.093124073433463e-05,
      "loss": 10.3441,
      "step": 47
    },
    {
      "epoch": 0.04774931609052475,
      "grad_norm": 0.22006335854530334,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.3546,
      "step": 48
    },
    {
      "epoch": 0.048744093509077345,
      "grad_norm": 0.2656434178352356,
      "learning_rate": 8.995939984474624e-05,
      "loss": 10.3503,
      "step": 49
    },
    {
      "epoch": 0.04973887092762994,
      "grad_norm": 0.3199167847633362,
      "learning_rate": 8.945702546981969e-05,
      "loss": 10.3521,
      "step": 50
    },
    {
      "epoch": 0.04973887092762994,
      "eval_loss": NaN,
      "eval_runtime": 3.739,
      "eval_samples_per_second": 452.8,
      "eval_steps_per_second": 113.4,
      "step": 50
    },
    {
      "epoch": 0.05073364834618254,
      "grad_norm": 0.14966224133968353,
      "learning_rate": 8.894386393810563e-05,
      "loss": 10.3632,
      "step": 51
    },
    {
      "epoch": 0.05172842576473514,
      "grad_norm": 0.14993417263031006,
      "learning_rate": 8.842005554284296e-05,
      "loss": 10.3607,
      "step": 52
    },
    {
      "epoch": 0.05272320318328774,
      "grad_norm": 0.173100084066391,
      "learning_rate": 8.788574348801675e-05,
      "loss": 10.3631,
      "step": 53
    },
    {
      "epoch": 0.05371798060184034,
      "grad_norm": 0.15190263092517853,
      "learning_rate": 8.73410738492077e-05,
      "loss": 10.361,
      "step": 54
    },
    {
      "epoch": 0.05471275802039294,
      "grad_norm": 0.13069787621498108,
      "learning_rate": 8.678619553365659e-05,
      "loss": 10.3601,
      "step": 55
    },
    {
      "epoch": 0.055707535438945535,
      "grad_norm": 0.1595054715871811,
      "learning_rate": 8.622126023955446e-05,
      "loss": 10.3589,
      "step": 56
    },
    {
      "epoch": 0.05670231285749813,
      "grad_norm": 0.13559356331825256,
      "learning_rate": 8.564642241456986e-05,
      "loss": 10.3597,
      "step": 57
    },
    {
      "epoch": 0.05769709027605073,
      "grad_norm": 0.14737024903297424,
      "learning_rate": 8.506183921362443e-05,
      "loss": 10.3564,
      "step": 58
    },
    {
      "epoch": 0.058691867694603335,
      "grad_norm": 0.14111341536045074,
      "learning_rate": 8.44676704559283e-05,
      "loss": 10.3579,
      "step": 59
    },
    {
      "epoch": 0.05968664511315593,
      "grad_norm": 0.15131875872612,
      "learning_rate": 8.386407858128706e-05,
      "loss": 10.3547,
      "step": 60
    },
    {
      "epoch": 0.06068142253170853,
      "grad_norm": 0.13234858214855194,
      "learning_rate": 8.32512286056924e-05,
      "loss": 10.3552,
      "step": 61
    },
    {
      "epoch": 0.06167619995026113,
      "grad_norm": 0.12116644531488419,
      "learning_rate": 8.262928807620843e-05,
      "loss": 10.3555,
      "step": 62
    },
    {
      "epoch": 0.06267097736881373,
      "grad_norm": 0.1282518059015274,
      "learning_rate": 8.199842702516583e-05,
      "loss": 10.3569,
      "step": 63
    },
    {
      "epoch": 0.06366575478736633,
      "grad_norm": 0.12213567644357681,
      "learning_rate": 8.135881792367686e-05,
      "loss": 10.3557,
      "step": 64
    },
    {
      "epoch": 0.06466053220591893,
      "grad_norm": 0.1226038858294487,
      "learning_rate": 8.07106356344834e-05,
      "loss": 10.3541,
      "step": 65
    },
    {
      "epoch": 0.06565530962447153,
      "grad_norm": 0.12056569010019302,
      "learning_rate": 8.005405736415126e-05,
      "loss": 10.3543,
      "step": 66
    },
    {
      "epoch": 0.06665008704302412,
      "grad_norm": 0.08386711031198502,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.3557,
      "step": 67
    },
    {
      "epoch": 0.06764486446157672,
      "grad_norm": 0.09664157778024673,
      "learning_rate": 7.871643313414718e-05,
      "loss": 10.3512,
      "step": 68
    },
    {
      "epoch": 0.06863964188012932,
      "grad_norm": 0.11908260732889175,
      "learning_rate": 7.803575286758364e-05,
      "loss": 10.3514,
      "step": 69
    },
    {
      "epoch": 0.06963441929868192,
      "grad_norm": 0.08827299624681473,
      "learning_rate": 7.734740790612136e-05,
      "loss": 10.3537,
      "step": 70
    },
    {
      "epoch": 0.07062919671723451,
      "grad_norm": 0.0886191725730896,
      "learning_rate": 7.66515864363997e-05,
      "loss": 10.3475,
      "step": 71
    },
    {
      "epoch": 0.07162397413578711,
      "grad_norm": 0.07376007735729218,
      "learning_rate": 7.594847868906076e-05,
      "loss": 10.3526,
      "step": 72
    },
    {
      "epoch": 0.07261875155433972,
      "grad_norm": 0.06961233913898468,
      "learning_rate": 7.52382768867422e-05,
      "loss": 10.3516,
      "step": 73
    },
    {
      "epoch": 0.07361352897289232,
      "grad_norm": 0.08884357661008835,
      "learning_rate": 7.452117519152542e-05,
      "loss": 10.3548,
      "step": 74
    },
    {
      "epoch": 0.07460830639144492,
      "grad_norm": 0.08113538473844528,
      "learning_rate": 7.379736965185368e-05,
      "loss": 10.3496,
      "step": 75
    },
    {
      "epoch": 0.07560308380999751,
      "grad_norm": 0.08134754747152328,
      "learning_rate": 7.30670581489344e-05,
      "loss": 10.3501,
      "step": 76
    },
    {
      "epoch": 0.07659786122855011,
      "grad_norm": 0.08616188913583755,
      "learning_rate": 7.233044034264034e-05,
      "loss": 10.3504,
      "step": 77
    },
    {
      "epoch": 0.07759263864710271,
      "grad_norm": 0.09940912574529648,
      "learning_rate": 7.158771761692464e-05,
      "loss": 10.3461,
      "step": 78
    },
    {
      "epoch": 0.07858741606565531,
      "grad_norm": 0.0934673622250557,
      "learning_rate": 7.083909302476453e-05,
      "loss": 10.3503,
      "step": 79
    },
    {
      "epoch": 0.0795821934842079,
      "grad_norm": 0.09255243092775345,
      "learning_rate": 7.008477123264848e-05,
      "loss": 10.3526,
      "step": 80
    },
    {
      "epoch": 0.0805769709027605,
      "grad_norm": 0.08574622124433517,
      "learning_rate": 6.932495846462261e-05,
      "loss": 10.3565,
      "step": 81
    },
    {
      "epoch": 0.0815717483213131,
      "grad_norm": 0.10347207635641098,
      "learning_rate": 6.855986244591104e-05,
      "loss": 10.3506,
      "step": 82
    },
    {
      "epoch": 0.0825665257398657,
      "grad_norm": 0.12859922647476196,
      "learning_rate": 6.778969234612584e-05,
      "loss": 10.3502,
      "step": 83
    },
    {
      "epoch": 0.08356130315841831,
      "grad_norm": 0.09400458633899689,
      "learning_rate": 6.701465872208216e-05,
      "loss": 10.3451,
      "step": 84
    },
    {
      "epoch": 0.08455608057697091,
      "grad_norm": 0.12339858710765839,
      "learning_rate": 6.623497346023418e-05,
      "loss": 10.3517,
      "step": 85
    },
    {
      "epoch": 0.0855508579955235,
      "grad_norm": 0.11353509873151779,
      "learning_rate": 6.545084971874738e-05,
      "loss": 10.3508,
      "step": 86
    },
    {
      "epoch": 0.0865456354140761,
      "grad_norm": 0.10968830436468124,
      "learning_rate": 6.466250186922325e-05,
      "loss": 10.3506,
      "step": 87
    },
    {
      "epoch": 0.0875404128326287,
      "grad_norm": 0.1172848492860794,
      "learning_rate": 6.387014543809223e-05,
      "loss": 10.348,
      "step": 88
    },
    {
      "epoch": 0.0885351902511813,
      "grad_norm": 0.12724760174751282,
      "learning_rate": 6.307399704769099e-05,
      "loss": 10.3467,
      "step": 89
    },
    {
      "epoch": 0.0895299676697339,
      "grad_norm": 0.13043548166751862,
      "learning_rate": 6.227427435703997e-05,
      "loss": 10.3479,
      "step": 90
    },
    {
      "epoch": 0.09052474508828649,
      "grad_norm": 0.14637860655784607,
      "learning_rate": 6.147119600233758e-05,
      "loss": 10.3465,
      "step": 91
    },
    {
      "epoch": 0.09151952250683909,
      "grad_norm": 0.12578149139881134,
      "learning_rate": 6.066498153718735e-05,
      "loss": 10.3433,
      "step": 92
    },
    {
      "epoch": 0.09251429992539169,
      "grad_norm": 0.12398236989974976,
      "learning_rate": 5.985585137257401e-05,
      "loss": 10.3484,
      "step": 93
    },
    {
      "epoch": 0.09350907734394429,
      "grad_norm": 0.17920762300491333,
      "learning_rate": 5.90440267166055e-05,
      "loss": 10.3467,
      "step": 94
    },
    {
      "epoch": 0.0945038547624969,
      "grad_norm": 0.12839432060718536,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 10.3512,
      "step": 95
    },
    {
      "epoch": 0.0954986321810495,
      "grad_norm": 0.14354673027992249,
      "learning_rate": 5.74131823855921e-05,
      "loss": 10.3419,
      "step": 96
    },
    {
      "epoch": 0.09649340959960209,
      "grad_norm": 0.15313208103179932,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 10.3446,
      "step": 97
    },
    {
      "epoch": 0.09748818701815469,
      "grad_norm": 0.17869329452514648,
      "learning_rate": 5.577423184847932e-05,
      "loss": 10.3434,
      "step": 98
    },
    {
      "epoch": 0.09848296443670729,
      "grad_norm": 0.19117513298988342,
      "learning_rate": 5.495227651252315e-05,
      "loss": 10.3381,
      "step": 99
    },
    {
      "epoch": 0.09947774185525989,
      "grad_norm": 0.3579142093658447,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 10.3342,
      "step": 100
    },
    {
      "epoch": 0.09947774185525989,
      "eval_loss": NaN,
      "eval_runtime": 3.7492,
      "eval_samples_per_second": 451.561,
      "eval_steps_per_second": 113.09,
      "step": 100
    },
    {
      "epoch": 0.10047251927381248,
      "grad_norm": 0.1144421398639679,
      "learning_rate": 5.330452921628497e-05,
      "loss": 10.3547,
      "step": 101
    },
    {
      "epoch": 0.10146729669236508,
      "grad_norm": 0.0968073308467865,
      "learning_rate": 5.247918773366112e-05,
      "loss": 10.3553,
      "step": 102
    },
    {
      "epoch": 0.10246207411091768,
      "grad_norm": 0.09298888593912125,
      "learning_rate": 5.165316846586541e-05,
      "loss": 10.3538,
      "step": 103
    },
    {
      "epoch": 0.10345685152947028,
      "grad_norm": 0.10922475904226303,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 10.3508,
      "step": 104
    },
    {
      "epoch": 0.10445162894802287,
      "grad_norm": 0.11373133212327957,
      "learning_rate": 5e-05,
      "loss": 10.3521,
      "step": 105
    },
    {
      "epoch": 0.10544640636657548,
      "grad_norm": 0.12363123893737793,
      "learning_rate": 4.917330276168208e-05,
      "loss": 10.3513,
      "step": 106
    },
    {
      "epoch": 0.10644118378512808,
      "grad_norm": 0.08842926472425461,
      "learning_rate": 4.834683153413459e-05,
      "loss": 10.352,
      "step": 107
    },
    {
      "epoch": 0.10743596120368068,
      "grad_norm": 0.1170431450009346,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 10.3539,
      "step": 108
    },
    {
      "epoch": 0.10843073862223328,
      "grad_norm": 0.10642478615045547,
      "learning_rate": 4.669547078371504e-05,
      "loss": 10.3508,
      "step": 109
    },
    {
      "epoch": 0.10942551604078588,
      "grad_norm": 0.10229886323213577,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 10.3516,
      "step": 110
    },
    {
      "epoch": 0.11042029345933847,
      "grad_norm": 0.10575897246599197,
      "learning_rate": 4.504772348747687e-05,
      "loss": 10.3498,
      "step": 111
    },
    {
      "epoch": 0.11141507087789107,
      "grad_norm": 0.08639760315418243,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 10.3516,
      "step": 112
    },
    {
      "epoch": 0.11240984829644367,
      "grad_norm": 0.08680632710456848,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 10.3496,
      "step": 113
    },
    {
      "epoch": 0.11340462571499627,
      "grad_norm": 0.09064137935638428,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 10.3516,
      "step": 114
    },
    {
      "epoch": 0.11439940313354886,
      "grad_norm": 0.09754966199398041,
      "learning_rate": 4.17702704859633e-05,
      "loss": 10.3511,
      "step": 115
    },
    {
      "epoch": 0.11539418055210146,
      "grad_norm": 0.08839617669582367,
      "learning_rate": 4.095597328339452e-05,
      "loss": 10.3517,
      "step": 116
    },
    {
      "epoch": 0.11638895797065407,
      "grad_norm": 0.08832965046167374,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 10.3492,
      "step": 117
    },
    {
      "epoch": 0.11738373538920667,
      "grad_norm": 0.07746206969022751,
      "learning_rate": 3.933501846281267e-05,
      "loss": 10.3472,
      "step": 118
    },
    {
      "epoch": 0.11837851280775927,
      "grad_norm": 0.08333753049373627,
      "learning_rate": 3.852880399766243e-05,
      "loss": 10.3497,
      "step": 119
    },
    {
      "epoch": 0.11937329022631187,
      "grad_norm": 0.0770106092095375,
      "learning_rate": 3.772572564296005e-05,
      "loss": 10.3504,
      "step": 120
    },
    {
      "epoch": 0.12036806764486446,
      "grad_norm": 0.060788776725530624,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 10.3496,
      "step": 121
    },
    {
      "epoch": 0.12136284506341706,
      "grad_norm": 0.0579555444419384,
      "learning_rate": 3.612985456190778e-05,
      "loss": 10.3485,
      "step": 122
    },
    {
      "epoch": 0.12235762248196966,
      "grad_norm": 0.06347642093896866,
      "learning_rate": 3.533749813077677e-05,
      "loss": 10.3513,
      "step": 123
    },
    {
      "epoch": 0.12335239990052226,
      "grad_norm": 0.07773666083812714,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.3451,
      "step": 124
    },
    {
      "epoch": 0.12434717731907485,
      "grad_norm": 0.06726665794849396,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 10.3485,
      "step": 125
    },
    {
      "epoch": 0.12534195473762746,
      "grad_norm": 0.09245743602514267,
      "learning_rate": 3.298534127791785e-05,
      "loss": 10.3442,
      "step": 126
    },
    {
      "epoch": 0.12633673215618005,
      "grad_norm": 0.09986552596092224,
      "learning_rate": 3.221030765387417e-05,
      "loss": 10.3488,
      "step": 127
    },
    {
      "epoch": 0.12733150957473266,
      "grad_norm": 0.09778209030628204,
      "learning_rate": 3.144013755408895e-05,
      "loss": 10.3424,
      "step": 128
    },
    {
      "epoch": 0.12832628699328524,
      "grad_norm": 0.08016423881053925,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 10.3512,
      "step": 129
    },
    {
      "epoch": 0.12932106441183785,
      "grad_norm": 0.07977841794490814,
      "learning_rate": 2.991522876735154e-05,
      "loss": 10.344,
      "step": 130
    },
    {
      "epoch": 0.13031584183039044,
      "grad_norm": 0.0810018852353096,
      "learning_rate": 2.916090697523549e-05,
      "loss": 10.3469,
      "step": 131
    },
    {
      "epoch": 0.13131061924894305,
      "grad_norm": 0.08203331381082535,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 10.3472,
      "step": 132
    },
    {
      "epoch": 0.13230539666749566,
      "grad_norm": 0.0677650049328804,
      "learning_rate": 2.766955965735968e-05,
      "loss": 10.3438,
      "step": 133
    },
    {
      "epoch": 0.13330017408604825,
      "grad_norm": 0.07421693950891495,
      "learning_rate": 2.693294185106562e-05,
      "loss": 10.3462,
      "step": 134
    },
    {
      "epoch": 0.13429495150460086,
      "grad_norm": 0.08530429750680923,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 10.3458,
      "step": 135
    },
    {
      "epoch": 0.13528972892315344,
      "grad_norm": 0.07112938165664673,
      "learning_rate": 2.547882480847461e-05,
      "loss": 10.3429,
      "step": 136
    },
    {
      "epoch": 0.13628450634170605,
      "grad_norm": 0.09052597731351852,
      "learning_rate": 2.476172311325783e-05,
      "loss": 10.3471,
      "step": 137
    },
    {
      "epoch": 0.13727928376025864,
      "grad_norm": 0.10110969841480255,
      "learning_rate": 2.405152131093926e-05,
      "loss": 10.3439,
      "step": 138
    },
    {
      "epoch": 0.13827406117881125,
      "grad_norm": 0.10673440247774124,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 10.3431,
      "step": 139
    },
    {
      "epoch": 0.13926883859736383,
      "grad_norm": 0.08431506901979446,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 10.3376,
      "step": 140
    },
    {
      "epoch": 0.14026361601591644,
      "grad_norm": 0.11080516874790192,
      "learning_rate": 2.196424713241637e-05,
      "loss": 10.3419,
      "step": 141
    },
    {
      "epoch": 0.14125839343446903,
      "grad_norm": 0.11397632956504822,
      "learning_rate": 2.128356686585282e-05,
      "loss": 10.3465,
      "step": 142
    },
    {
      "epoch": 0.14225317085302164,
      "grad_norm": 0.12718522548675537,
      "learning_rate": 2.061073738537635e-05,
      "loss": 10.3465,
      "step": 143
    },
    {
      "epoch": 0.14324794827157422,
      "grad_norm": 0.0826769545674324,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 10.3481,
      "step": 144
    },
    {
      "epoch": 0.14424272569012683,
      "grad_norm": 0.11485014110803604,
      "learning_rate": 1.928936436551661e-05,
      "loss": 10.3373,
      "step": 145
    },
    {
      "epoch": 0.14523750310867944,
      "grad_norm": 0.10438144207000732,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 10.3416,
      "step": 146
    },
    {
      "epoch": 0.14623228052723203,
      "grad_norm": 0.13776470720767975,
      "learning_rate": 1.800157297483417e-05,
      "loss": 10.3431,
      "step": 147
    },
    {
      "epoch": 0.14722705794578464,
      "grad_norm": 0.14553122222423553,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 10.3489,
      "step": 148
    },
    {
      "epoch": 0.14822183536433722,
      "grad_norm": 0.1501682698726654,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 10.3413,
      "step": 149
    },
    {
      "epoch": 0.14921661278288983,
      "grad_norm": 0.3195921778678894,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 10.327,
      "step": 150
    },
    {
      "epoch": 0.14921661278288983,
      "eval_loss": NaN,
      "eval_runtime": 3.7532,
      "eval_samples_per_second": 451.078,
      "eval_steps_per_second": 112.969,
      "step": 150
    },
    {
      "epoch": 0.15021139020144242,
      "grad_norm": 0.0833912044763565,
      "learning_rate": 1.553232954407171e-05,
      "loss": 10.3517,
      "step": 151
    },
    {
      "epoch": 0.15120616761999503,
      "grad_norm": 0.09314489364624023,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 10.3526,
      "step": 152
    },
    {
      "epoch": 0.1522009450385476,
      "grad_norm": 0.07763419300317764,
      "learning_rate": 1.435357758543015e-05,
      "loss": 10.35,
      "step": 153
    },
    {
      "epoch": 0.15319572245710023,
      "grad_norm": 0.0731983557343483,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 10.3514,
      "step": 154
    },
    {
      "epoch": 0.1541904998756528,
      "grad_norm": 0.08284006267786026,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 10.3514,
      "step": 155
    },
    {
      "epoch": 0.15518527729420542,
      "grad_norm": 0.08221081644296646,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 10.3507,
      "step": 156
    },
    {
      "epoch": 0.15618005471275803,
      "grad_norm": 0.06812987476587296,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 10.3506,
      "step": 157
    },
    {
      "epoch": 0.15717483213131062,
      "grad_norm": 0.09210630506277084,
      "learning_rate": 1.157994445715706e-05,
      "loss": 10.3525,
      "step": 158
    },
    {
      "epoch": 0.15816960954986323,
      "grad_norm": 0.07099930942058563,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 10.3488,
      "step": 159
    },
    {
      "epoch": 0.1591643869684158,
      "grad_norm": 0.06311217695474625,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 10.35,
      "step": 160
    },
    {
      "epoch": 0.16015916438696842,
      "grad_norm": 0.08011385798454285,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 10.3505,
      "step": 161
    },
    {
      "epoch": 0.161153941805521,
      "grad_norm": 0.07472092658281326,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.3497,
      "step": 162
    },
    {
      "epoch": 0.16214871922407362,
      "grad_norm": 0.07421920448541641,
      "learning_rate": 9.068759265665384e-06,
      "loss": 10.3506,
      "step": 163
    },
    {
      "epoch": 0.1631434966426262,
      "grad_norm": 0.07388707250356674,
      "learning_rate": 8.599558442598998e-06,
      "loss": 10.3491,
      "step": 164
    },
    {
      "epoch": 0.1641382740611788,
      "grad_norm": 0.05804355442523956,
      "learning_rate": 8.141676086873572e-06,
      "loss": 10.3494,
      "step": 165
    },
    {
      "epoch": 0.1651330514797314,
      "grad_norm": 0.07939363270998001,
      "learning_rate": 7.695237378953223e-06,
      "loss": 10.3503,
      "step": 166
    },
    {
      "epoch": 0.166127828898284,
      "grad_norm": 0.07092492282390594,
      "learning_rate": 7.260364370723044e-06,
      "loss": 10.3491,
      "step": 167
    },
    {
      "epoch": 0.16712260631683662,
      "grad_norm": 0.056517764925956726,
      "learning_rate": 6.837175952121306e-06,
      "loss": 10.3499,
      "step": 168
    },
    {
      "epoch": 0.1681173837353892,
      "grad_norm": 0.04937941953539848,
      "learning_rate": 6.425787818636131e-06,
      "loss": 10.3486,
      "step": 169
    },
    {
      "epoch": 0.16911216115394181,
      "grad_norm": 0.07365497946739197,
      "learning_rate": 6.026312439675552e-06,
      "loss": 10.3519,
      "step": 170
    },
    {
      "epoch": 0.1701069385724944,
      "grad_norm": 0.061077579855918884,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 10.3472,
      "step": 171
    },
    {
      "epoch": 0.171101715991047,
      "grad_norm": 0.052605871111154556,
      "learning_rate": 5.263533508961827e-06,
      "loss": 10.3506,
      "step": 172
    },
    {
      "epoch": 0.1720964934095996,
      "grad_norm": 0.07988005131483078,
      "learning_rate": 4.900438493352055e-06,
      "loss": 10.3532,
      "step": 173
    },
    {
      "epoch": 0.1730912708281522,
      "grad_norm": 0.06954702734947205,
      "learning_rate": 4.549673247541875e-06,
      "loss": 10.3458,
      "step": 174
    },
    {
      "epoch": 0.1740860482467048,
      "grad_norm": 0.0634763315320015,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 10.3462,
      "step": 175
    },
    {
      "epoch": 0.1750808256652574,
      "grad_norm": 0.06692229956388474,
      "learning_rate": 3.885512251130763e-06,
      "loss": 10.3442,
      "step": 176
    },
    {
      "epoch": 0.17607560308380998,
      "grad_norm": 0.07760122418403625,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 10.3461,
      "step": 177
    },
    {
      "epoch": 0.1770703805023626,
      "grad_norm": 0.09306254982948303,
      "learning_rate": 3.271776770026963e-06,
      "loss": 10.3459,
      "step": 178
    },
    {
      "epoch": 0.1780651579209152,
      "grad_norm": 0.06654608249664307,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 10.3486,
      "step": 179
    },
    {
      "epoch": 0.1790599353394678,
      "grad_norm": 0.07079225778579712,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 10.3408,
      "step": 180
    },
    {
      "epoch": 0.1800547127580204,
      "grad_norm": 0.08374956250190735,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.3416,
      "step": 181
    },
    {
      "epoch": 0.18104949017657299,
      "grad_norm": 0.09131907671689987,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 10.3394,
      "step": 182
    },
    {
      "epoch": 0.1820442675951256,
      "grad_norm": 0.06627543270587921,
      "learning_rate": 1.962316193157593e-06,
      "loss": 10.3446,
      "step": 183
    },
    {
      "epoch": 0.18303904501367818,
      "grad_norm": 0.07468573749065399,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 10.3436,
      "step": 184
    },
    {
      "epoch": 0.1840338224322308,
      "grad_norm": 0.08352243155241013,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 10.3431,
      "step": 185
    },
    {
      "epoch": 0.18502859985078338,
      "grad_norm": 0.10297807306051254,
      "learning_rate": 1.333670137599713e-06,
      "loss": 10.342,
      "step": 186
    },
    {
      "epoch": 0.186023377269336,
      "grad_norm": 0.07954021543264389,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 10.3454,
      "step": 187
    },
    {
      "epoch": 0.18701815468788857,
      "grad_norm": 0.08290816843509674,
      "learning_rate": 9.810017062595322e-07,
      "loss": 10.3426,
      "step": 188
    },
    {
      "epoch": 0.18801293210644118,
      "grad_norm": 0.08984582126140594,
      "learning_rate": 8.247462563808817e-07,
      "loss": 10.3392,
      "step": 189
    },
    {
      "epoch": 0.1890077095249938,
      "grad_norm": 0.08822706341743469,
      "learning_rate": 6.819348298638839e-07,
      "loss": 10.348,
      "step": 190
    },
    {
      "epoch": 0.19000248694354638,
      "grad_norm": 0.10973329097032547,
      "learning_rate": 5.526064699265753e-07,
      "loss": 10.3435,
      "step": 191
    },
    {
      "epoch": 0.190997264362099,
      "grad_norm": 0.09264547377824783,
      "learning_rate": 4.367965336512403e-07,
      "loss": 10.3459,
      "step": 192
    },
    {
      "epoch": 0.19199204178065157,
      "grad_norm": 0.12172121554613113,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 10.3422,
      "step": 193
    },
    {
      "epoch": 0.19298681919920418,
      "grad_norm": 0.09719396382570267,
      "learning_rate": 2.458548727494292e-07,
      "loss": 10.3409,
      "step": 194
    },
    {
      "epoch": 0.19398159661775677,
      "grad_norm": 0.1207161471247673,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.3384,
      "step": 195
    },
    {
      "epoch": 0.19497637403630938,
      "grad_norm": 0.10838349163532257,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 10.3453,
      "step": 196
    },
    {
      "epoch": 0.19597115145486196,
      "grad_norm": 0.1181037649512291,
      "learning_rate": 6.150154258476315e-08,
      "loss": 10.3443,
      "step": 197
    },
    {
      "epoch": 0.19696592887341458,
      "grad_norm": 0.11356630176305771,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 10.3357,
      "step": 198
    },
    {
      "epoch": 0.19796070629196716,
      "grad_norm": 0.16243548691272736,
      "learning_rate": 6.834750376549792e-09,
      "loss": 10.3378,
      "step": 199
    },
    {
      "epoch": 0.19895548371051977,
      "grad_norm": 0.24875065684318542,
      "learning_rate": 0.0,
      "loss": 10.3428,
      "step": 200
    },
    {
      "epoch": 0.19895548371051977,
      "eval_loss": NaN,
      "eval_runtime": 3.7481,
      "eval_samples_per_second": 451.701,
      "eval_steps_per_second": 113.125,
      "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": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 22310132121600.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}