{
  "best_metric": 2.3232147693634033,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.025809781907342882,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001290489095367144,
      "grad_norm": 1.8824232816696167,
      "learning_rate": 1e-05,
      "loss": 8.9881,
      "step": 1
    },
    {
      "epoch": 0.0001290489095367144,
      "eval_loss": 2.636259078979492,
      "eval_runtime": 1413.0548,
      "eval_samples_per_second": 9.236,
      "eval_steps_per_second": 2.309,
      "step": 1
    },
    {
      "epoch": 0.0002580978190734288,
      "grad_norm": 2.8287086486816406,
      "learning_rate": 2e-05,
      "loss": 9.2454,
      "step": 2
    },
    {
      "epoch": 0.00038714672861014324,
      "grad_norm": 2.5714402198791504,
      "learning_rate": 3e-05,
      "loss": 9.0849,
      "step": 3
    },
    {
      "epoch": 0.0005161956381468576,
      "grad_norm": 2.7806339263916016,
      "learning_rate": 4e-05,
      "loss": 9.4978,
      "step": 4
    },
    {
      "epoch": 0.0006452445476835721,
      "grad_norm": 2.4504265785217285,
      "learning_rate": 5e-05,
      "loss": 9.7492,
      "step": 5
    },
    {
      "epoch": 0.0007742934572202865,
      "grad_norm": 2.520981550216675,
      "learning_rate": 6e-05,
      "loss": 8.8263,
      "step": 6
    },
    {
      "epoch": 0.0009033423667570009,
      "grad_norm": 2.629502534866333,
      "learning_rate": 7e-05,
      "loss": 9.3517,
      "step": 7
    },
    {
      "epoch": 0.0010323912762937152,
      "grad_norm": 2.9892454147338867,
      "learning_rate": 8e-05,
      "loss": 9.6262,
      "step": 8
    },
    {
      "epoch": 0.0011614401858304297,
      "grad_norm": 2.9270966053009033,
      "learning_rate": 9e-05,
      "loss": 9.0262,
      "step": 9
    },
    {
      "epoch": 0.0012904890953671442,
      "grad_norm": 3.554283618927002,
      "learning_rate": 0.0001,
      "loss": 9.1454,
      "step": 10
    },
    {
      "epoch": 0.0014195380049038585,
      "grad_norm": 2.9322144985198975,
      "learning_rate": 9.999316524962345e-05,
      "loss": 9.189,
      "step": 11
    },
    {
      "epoch": 0.001548586914440573,
      "grad_norm": 2.8467869758605957,
      "learning_rate": 9.997266286704631e-05,
      "loss": 9.13,
      "step": 12
    },
    {
      "epoch": 0.0016776358239772875,
      "grad_norm": 2.985856294631958,
      "learning_rate": 9.993849845741524e-05,
      "loss": 9.3862,
      "step": 13
    },
    {
      "epoch": 0.0018066847335140017,
      "grad_norm": 2.986755132675171,
      "learning_rate": 9.989068136093873e-05,
      "loss": 9.3611,
      "step": 14
    },
    {
      "epoch": 0.0019357336430507162,
      "grad_norm": 2.891209840774536,
      "learning_rate": 9.98292246503335e-05,
      "loss": 8.9278,
      "step": 15
    },
    {
      "epoch": 0.0020647825525874305,
      "grad_norm": 3.200324535369873,
      "learning_rate": 9.975414512725057e-05,
      "loss": 9.3208,
      "step": 16
    },
    {
      "epoch": 0.002193831462124145,
      "grad_norm": 3.082493782043457,
      "learning_rate": 9.966546331768191e-05,
      "loss": 9.3286,
      "step": 17
    },
    {
      "epoch": 0.0023228803716608595,
      "grad_norm": 2.9336154460906982,
      "learning_rate": 9.956320346634876e-05,
      "loss": 9.1107,
      "step": 18
    },
    {
      "epoch": 0.0024519292811975737,
      "grad_norm": 2.870712995529175,
      "learning_rate": 9.944739353007344e-05,
      "loss": 9.2296,
      "step": 19
    },
    {
      "epoch": 0.0025809781907342884,
      "grad_norm": 3.229865789413452,
      "learning_rate": 9.931806517013612e-05,
      "loss": 9.0709,
      "step": 20
    },
    {
      "epoch": 0.0027100271002710027,
      "grad_norm": 2.955902576446533,
      "learning_rate": 9.917525374361912e-05,
      "loss": 9.1987,
      "step": 21
    },
    {
      "epoch": 0.002839076009807717,
      "grad_norm": 3.0389182567596436,
      "learning_rate": 9.901899829374047e-05,
      "loss": 8.9725,
      "step": 22
    },
    {
      "epoch": 0.0029681249193444317,
      "grad_norm": 4.6783766746521,
      "learning_rate": 9.884934153917997e-05,
      "loss": 9.3348,
      "step": 23
    },
    {
      "epoch": 0.003097173828881146,
      "grad_norm": 3.08036208152771,
      "learning_rate": 9.86663298624003e-05,
      "loss": 9.3108,
      "step": 24
    },
    {
      "epoch": 0.0032262227384178602,
      "grad_norm": 3.42242693901062,
      "learning_rate": 9.847001329696653e-05,
      "loss": 9.0909,
      "step": 25
    },
    {
      "epoch": 0.003355271647954575,
      "grad_norm": 3.5331575870513916,
      "learning_rate": 9.826044551386744e-05,
      "loss": 9.3532,
      "step": 26
    },
    {
      "epoch": 0.003484320557491289,
      "grad_norm": 7.4575581550598145,
      "learning_rate": 9.803768380684242e-05,
      "loss": 9.2112,
      "step": 27
    },
    {
      "epoch": 0.0036133694670280035,
      "grad_norm": 3.2820820808410645,
      "learning_rate": 9.780178907671789e-05,
      "loss": 9.4624,
      "step": 28
    },
    {
      "epoch": 0.003742418376564718,
      "grad_norm": 3.459367275238037,
      "learning_rate": 9.755282581475769e-05,
      "loss": 9.2565,
      "step": 29
    },
    {
      "epoch": 0.0038714672861014324,
      "grad_norm": 3.8929476737976074,
      "learning_rate": 9.729086208503174e-05,
      "loss": 9.669,
      "step": 30
    },
    {
      "epoch": 0.004000516195638147,
      "grad_norm": 3.531162738800049,
      "learning_rate": 9.701596950580806e-05,
      "loss": 9.2843,
      "step": 31
    },
    {
      "epoch": 0.004129565105174861,
      "grad_norm": 3.5504400730133057,
      "learning_rate": 9.672822322997305e-05,
      "loss": 9.5244,
      "step": 32
    },
    {
      "epoch": 0.004258614014711576,
      "grad_norm": 3.8479762077331543,
      "learning_rate": 9.642770192448536e-05,
      "loss": 9.287,
      "step": 33
    },
    {
      "epoch": 0.00438766292424829,
      "grad_norm": 3.660515308380127,
      "learning_rate": 9.611448774886924e-05,
      "loss": 9.1514,
      "step": 34
    },
    {
      "epoch": 0.004516711833785004,
      "grad_norm": 3.883052110671997,
      "learning_rate": 9.578866633275288e-05,
      "loss": 9.7452,
      "step": 35
    },
    {
      "epoch": 0.004645760743321719,
      "grad_norm": 3.631749153137207,
      "learning_rate": 9.545032675245813e-05,
      "loss": 8.7108,
      "step": 36
    },
    {
      "epoch": 0.004774809652858434,
      "grad_norm": 4.567019939422607,
      "learning_rate": 9.509956150664796e-05,
      "loss": 9.0823,
      "step": 37
    },
    {
      "epoch": 0.0049038585623951475,
      "grad_norm": 3.8664674758911133,
      "learning_rate": 9.473646649103818e-05,
      "loss": 9.0209,
      "step": 38
    },
    {
      "epoch": 0.005032907471931862,
      "grad_norm": 3.753689765930176,
      "learning_rate": 9.43611409721806e-05,
      "loss": 9.1999,
      "step": 39
    },
    {
      "epoch": 0.005161956381468577,
      "grad_norm": 4.090752124786377,
      "learning_rate": 9.397368756032445e-05,
      "loss": 9.5598,
      "step": 40
    },
    {
      "epoch": 0.005291005291005291,
      "grad_norm": 5.097810745239258,
      "learning_rate": 9.357421218136386e-05,
      "loss": 9.3718,
      "step": 41
    },
    {
      "epoch": 0.005420054200542005,
      "grad_norm": 4.1418232917785645,
      "learning_rate": 9.316282404787871e-05,
      "loss": 9.2399,
      "step": 42
    },
    {
      "epoch": 0.00554910311007872,
      "grad_norm": 4.791266918182373,
      "learning_rate": 9.273963562927695e-05,
      "loss": 9.6049,
      "step": 43
    },
    {
      "epoch": 0.005678152019615434,
      "grad_norm": 5.000885486602783,
      "learning_rate": 9.230476262104677e-05,
      "loss": 9.2656,
      "step": 44
    },
    {
      "epoch": 0.005807200929152149,
      "grad_norm": 4.691623210906982,
      "learning_rate": 9.185832391312644e-05,
      "loss": 9.0675,
      "step": 45
    },
    {
      "epoch": 0.005936249838688863,
      "grad_norm": 8.883625030517578,
      "learning_rate": 9.140044155740101e-05,
      "loss": 10.0689,
      "step": 46
    },
    {
      "epoch": 0.006065298748225577,
      "grad_norm": 7.350927829742432,
      "learning_rate": 9.093124073433463e-05,
      "loss": 9.5976,
      "step": 47
    },
    {
      "epoch": 0.006194347657762292,
      "grad_norm": 5.843919277191162,
      "learning_rate": 9.045084971874738e-05,
      "loss": 9.7173,
      "step": 48
    },
    {
      "epoch": 0.006323396567299007,
      "grad_norm": 5.78486967086792,
      "learning_rate": 8.995939984474624e-05,
      "loss": 9.5343,
      "step": 49
    },
    {
      "epoch": 0.0064524454768357204,
      "grad_norm": 7.93582820892334,
      "learning_rate": 8.945702546981969e-05,
      "loss": 9.6716,
      "step": 50
    },
    {
      "epoch": 0.0064524454768357204,
      "eval_loss": 2.4635510444641113,
      "eval_runtime": 1423.6372,
      "eval_samples_per_second": 9.167,
      "eval_steps_per_second": 2.292,
      "step": 50
    },
    {
      "epoch": 0.006581494386372435,
      "grad_norm": 5.295307159423828,
      "learning_rate": 8.894386393810563e-05,
      "loss": 8.74,
      "step": 51
    },
    {
      "epoch": 0.00671054329590915,
      "grad_norm": 4.496520042419434,
      "learning_rate": 8.842005554284296e-05,
      "loss": 9.5416,
      "step": 52
    },
    {
      "epoch": 0.006839592205445864,
      "grad_norm": 3.4309699535369873,
      "learning_rate": 8.788574348801675e-05,
      "loss": 9.5004,
      "step": 53
    },
    {
      "epoch": 0.006968641114982578,
      "grad_norm": 2.202669382095337,
      "learning_rate": 8.73410738492077e-05,
      "loss": 9.1945,
      "step": 54
    },
    {
      "epoch": 0.007097690024519293,
      "grad_norm": 1.8157504796981812,
      "learning_rate": 8.678619553365659e-05,
      "loss": 8.8847,
      "step": 55
    },
    {
      "epoch": 0.007226738934056007,
      "grad_norm": 1.9221110343933105,
      "learning_rate": 8.622126023955446e-05,
      "loss": 8.8982,
      "step": 56
    },
    {
      "epoch": 0.007355787843592722,
      "grad_norm": 2.0396029949188232,
      "learning_rate": 8.564642241456986e-05,
      "loss": 9.0756,
      "step": 57
    },
    {
      "epoch": 0.007484836753129436,
      "grad_norm": 2.0514283180236816,
      "learning_rate": 8.506183921362443e-05,
      "loss": 9.0645,
      "step": 58
    },
    {
      "epoch": 0.00761388566266615,
      "grad_norm": 2.847005844116211,
      "learning_rate": 8.44676704559283e-05,
      "loss": 8.8491,
      "step": 59
    },
    {
      "epoch": 0.007742934572202865,
      "grad_norm": 2.267468214035034,
      "learning_rate": 8.386407858128706e-05,
      "loss": 9.3444,
      "step": 60
    },
    {
      "epoch": 0.00787198348173958,
      "grad_norm": 2.3715107440948486,
      "learning_rate": 8.32512286056924e-05,
      "loss": 9.1397,
      "step": 61
    },
    {
      "epoch": 0.008001032391276294,
      "grad_norm": 2.2348217964172363,
      "learning_rate": 8.262928807620843e-05,
      "loss": 9.3976,
      "step": 62
    },
    {
      "epoch": 0.008130081300813009,
      "grad_norm": 2.58975887298584,
      "learning_rate": 8.199842702516583e-05,
      "loss": 9.1345,
      "step": 63
    },
    {
      "epoch": 0.008259130210349722,
      "grad_norm": 3.094756841659546,
      "learning_rate": 8.135881792367686e-05,
      "loss": 9.0197,
      "step": 64
    },
    {
      "epoch": 0.008388179119886437,
      "grad_norm": 2.9497580528259277,
      "learning_rate": 8.07106356344834e-05,
      "loss": 9.3347,
      "step": 65
    },
    {
      "epoch": 0.008517228029423151,
      "grad_norm": 2.713719129562378,
      "learning_rate": 8.005405736415126e-05,
      "loss": 9.2095,
      "step": 66
    },
    {
      "epoch": 0.008646276938959866,
      "grad_norm": 2.7610373497009277,
      "learning_rate": 7.938926261462366e-05,
      "loss": 9.2925,
      "step": 67
    },
    {
      "epoch": 0.00877532584849658,
      "grad_norm": 2.471407651901245,
      "learning_rate": 7.871643313414718e-05,
      "loss": 9.2705,
      "step": 68
    },
    {
      "epoch": 0.008904374758033295,
      "grad_norm": 2.827798366546631,
      "learning_rate": 7.803575286758364e-05,
      "loss": 9.7064,
      "step": 69
    },
    {
      "epoch": 0.009033423667570008,
      "grad_norm": 3.0900652408599854,
      "learning_rate": 7.734740790612136e-05,
      "loss": 8.8224,
      "step": 70
    },
    {
      "epoch": 0.009162472577106723,
      "grad_norm": 2.860865354537964,
      "learning_rate": 7.66515864363997e-05,
      "loss": 9.2759,
      "step": 71
    },
    {
      "epoch": 0.009291521486643438,
      "grad_norm": 3.471763849258423,
      "learning_rate": 7.594847868906076e-05,
      "loss": 9.1891,
      "step": 72
    },
    {
      "epoch": 0.009420570396180153,
      "grad_norm": 3.1697371006011963,
      "learning_rate": 7.52382768867422e-05,
      "loss": 8.8724,
      "step": 73
    },
    {
      "epoch": 0.009549619305716867,
      "grad_norm": 3.079185962677002,
      "learning_rate": 7.452117519152542e-05,
      "loss": 9.2962,
      "step": 74
    },
    {
      "epoch": 0.009678668215253582,
      "grad_norm": 2.8517227172851562,
      "learning_rate": 7.379736965185368e-05,
      "loss": 9.1186,
      "step": 75
    },
    {
      "epoch": 0.009807717124790295,
      "grad_norm": 3.036484956741333,
      "learning_rate": 7.30670581489344e-05,
      "loss": 9.8041,
      "step": 76
    },
    {
      "epoch": 0.00993676603432701,
      "grad_norm": 3.1947879791259766,
      "learning_rate": 7.233044034264034e-05,
      "loss": 9.1916,
      "step": 77
    },
    {
      "epoch": 0.010065814943863724,
      "grad_norm": 3.0642189979553223,
      "learning_rate": 7.158771761692464e-05,
      "loss": 8.8235,
      "step": 78
    },
    {
      "epoch": 0.010194863853400439,
      "grad_norm": 3.215501546859741,
      "learning_rate": 7.083909302476453e-05,
      "loss": 9.6942,
      "step": 79
    },
    {
      "epoch": 0.010323912762937154,
      "grad_norm": 3.0959300994873047,
      "learning_rate": 7.008477123264848e-05,
      "loss": 9.4002,
      "step": 80
    },
    {
      "epoch": 0.010452961672473868,
      "grad_norm": 3.484468698501587,
      "learning_rate": 6.932495846462261e-05,
      "loss": 9.736,
      "step": 81
    },
    {
      "epoch": 0.010582010582010581,
      "grad_norm": 3.5731234550476074,
      "learning_rate": 6.855986244591104e-05,
      "loss": 8.8871,
      "step": 82
    },
    {
      "epoch": 0.010711059491547296,
      "grad_norm": 3.703524589538574,
      "learning_rate": 6.778969234612584e-05,
      "loss": 9.2151,
      "step": 83
    },
    {
      "epoch": 0.01084010840108401,
      "grad_norm": 3.4570441246032715,
      "learning_rate": 6.701465872208216e-05,
      "loss": 9.4282,
      "step": 84
    },
    {
      "epoch": 0.010969157310620726,
      "grad_norm": 3.427992582321167,
      "learning_rate": 6.623497346023418e-05,
      "loss": 9.8639,
      "step": 85
    },
    {
      "epoch": 0.01109820622015744,
      "grad_norm": 3.327118396759033,
      "learning_rate": 6.545084971874738e-05,
      "loss": 9.2256,
      "step": 86
    },
    {
      "epoch": 0.011227255129694155,
      "grad_norm": 3.586247205734253,
      "learning_rate": 6.466250186922325e-05,
      "loss": 9.1051,
      "step": 87
    },
    {
      "epoch": 0.011356304039230868,
      "grad_norm": 3.6466622352600098,
      "learning_rate": 6.387014543809223e-05,
      "loss": 8.9656,
      "step": 88
    },
    {
      "epoch": 0.011485352948767583,
      "grad_norm": 3.89249324798584,
      "learning_rate": 6.307399704769099e-05,
      "loss": 9.6611,
      "step": 89
    },
    {
      "epoch": 0.011614401858304297,
      "grad_norm": 3.8333232402801514,
      "learning_rate": 6.227427435703997e-05,
      "loss": 9.2893,
      "step": 90
    },
    {
      "epoch": 0.011743450767841012,
      "grad_norm": 3.963622570037842,
      "learning_rate": 6.147119600233758e-05,
      "loss": 9.2436,
      "step": 91
    },
    {
      "epoch": 0.011872499677377727,
      "grad_norm": 4.480712890625,
      "learning_rate": 6.066498153718735e-05,
      "loss": 9.2674,
      "step": 92
    },
    {
      "epoch": 0.012001548586914441,
      "grad_norm": 4.302041053771973,
      "learning_rate": 5.985585137257401e-05,
      "loss": 9.2682,
      "step": 93
    },
    {
      "epoch": 0.012130597496451154,
      "grad_norm": 4.53174352645874,
      "learning_rate": 5.90440267166055e-05,
      "loss": 9.3679,
      "step": 94
    },
    {
      "epoch": 0.012259646405987869,
      "grad_norm": 12.704017639160156,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 8.7514,
      "step": 95
    },
    {
      "epoch": 0.012388695315524584,
      "grad_norm": 4.984100341796875,
      "learning_rate": 5.74131823855921e-05,
      "loss": 9.3493,
      "step": 96
    },
    {
      "epoch": 0.012517744225061299,
      "grad_norm": 5.241578102111816,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 9.672,
      "step": 97
    },
    {
      "epoch": 0.012646793134598013,
      "grad_norm": 5.066354751586914,
      "learning_rate": 5.577423184847932e-05,
      "loss": 9.5853,
      "step": 98
    },
    {
      "epoch": 0.012775842044134728,
      "grad_norm": 5.492655277252197,
      "learning_rate": 5.495227651252315e-05,
      "loss": 9.8459,
      "step": 99
    },
    {
      "epoch": 0.012904890953671441,
      "grad_norm": 6.962948799133301,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 10.5033,
      "step": 100
    },
    {
      "epoch": 0.012904890953671441,
      "eval_loss": 2.411465644836426,
      "eval_runtime": 1420.9881,
      "eval_samples_per_second": 9.184,
      "eval_steps_per_second": 2.296,
      "step": 100
    },
    {
      "epoch": 0.013033939863208156,
      "grad_norm": 6.244724750518799,
      "learning_rate": 5.330452921628497e-05,
      "loss": 8.9729,
      "step": 101
    },
    {
      "epoch": 0.01316298877274487,
      "grad_norm": 6.598541259765625,
      "learning_rate": 5.247918773366112e-05,
      "loss": 9.3254,
      "step": 102
    },
    {
      "epoch": 0.013292037682281585,
      "grad_norm": 5.932408332824707,
      "learning_rate": 5.165316846586541e-05,
      "loss": 9.2865,
      "step": 103
    },
    {
      "epoch": 0.0134210865918183,
      "grad_norm": 4.1782307624816895,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 8.8848,
      "step": 104
    },
    {
      "epoch": 0.013550135501355014,
      "grad_norm": 2.616091251373291,
      "learning_rate": 5e-05,
      "loss": 9.1734,
      "step": 105
    },
    {
      "epoch": 0.013679184410891727,
      "grad_norm": 2.0207128524780273,
      "learning_rate": 4.917330276168208e-05,
      "loss": 9.095,
      "step": 106
    },
    {
      "epoch": 0.013808233320428442,
      "grad_norm": 1.8105111122131348,
      "learning_rate": 4.834683153413459e-05,
      "loss": 8.6875,
      "step": 107
    },
    {
      "epoch": 0.013937282229965157,
      "grad_norm": 1.8800246715545654,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 9.3335,
      "step": 108
    },
    {
      "epoch": 0.014066331139501871,
      "grad_norm": 1.935013771057129,
      "learning_rate": 4.669547078371504e-05,
      "loss": 9.025,
      "step": 109
    },
    {
      "epoch": 0.014195380049038586,
      "grad_norm": 1.9459717273712158,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 9.1148,
      "step": 110
    },
    {
      "epoch": 0.014324428958575301,
      "grad_norm": 2.5711159706115723,
      "learning_rate": 4.504772348747687e-05,
      "loss": 9.5559,
      "step": 111
    },
    {
      "epoch": 0.014453477868112014,
      "grad_norm": 2.041684627532959,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 8.5859,
      "step": 112
    },
    {
      "epoch": 0.014582526777648729,
      "grad_norm": 2.1846988201141357,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 8.8542,
      "step": 113
    },
    {
      "epoch": 0.014711575687185443,
      "grad_norm": 2.409000873565674,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 9.3534,
      "step": 114
    },
    {
      "epoch": 0.014840624596722158,
      "grad_norm": 2.565107583999634,
      "learning_rate": 4.17702704859633e-05,
      "loss": 9.1375,
      "step": 115
    },
    {
      "epoch": 0.014969673506258873,
      "grad_norm": 2.407203435897827,
      "learning_rate": 4.095597328339452e-05,
      "loss": 8.972,
      "step": 116
    },
    {
      "epoch": 0.015098722415795587,
      "grad_norm": 2.518613815307617,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 9.1395,
      "step": 117
    },
    {
      "epoch": 0.0152277713253323,
      "grad_norm": 2.6896538734436035,
      "learning_rate": 3.933501846281267e-05,
      "loss": 9.6748,
      "step": 118
    },
    {
      "epoch": 0.015356820234869015,
      "grad_norm": 2.6174116134643555,
      "learning_rate": 3.852880399766243e-05,
      "loss": 9.1389,
      "step": 119
    },
    {
      "epoch": 0.01548586914440573,
      "grad_norm": 2.59897780418396,
      "learning_rate": 3.772572564296005e-05,
      "loss": 9.0712,
      "step": 120
    },
    {
      "epoch": 0.015614918053942444,
      "grad_norm": 2.4263978004455566,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 8.7425,
      "step": 121
    },
    {
      "epoch": 0.01574396696347916,
      "grad_norm": 2.7460546493530273,
      "learning_rate": 3.612985456190778e-05,
      "loss": 9.1534,
      "step": 122
    },
    {
      "epoch": 0.015873015873015872,
      "grad_norm": 2.5968704223632812,
      "learning_rate": 3.533749813077677e-05,
      "loss": 9.2163,
      "step": 123
    },
    {
      "epoch": 0.01600206478255259,
      "grad_norm": 2.787576198577881,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 8.5128,
      "step": 124
    },
    {
      "epoch": 0.0161311136920893,
      "grad_norm": 2.823814630508423,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 9.283,
      "step": 125
    },
    {
      "epoch": 0.016260162601626018,
      "grad_norm": 2.9372408390045166,
      "learning_rate": 3.298534127791785e-05,
      "loss": 9.8537,
      "step": 126
    },
    {
      "epoch": 0.01638921151116273,
      "grad_norm": 3.104567527770996,
      "learning_rate": 3.221030765387417e-05,
      "loss": 8.941,
      "step": 127
    },
    {
      "epoch": 0.016518260420699444,
      "grad_norm": 2.982133626937866,
      "learning_rate": 3.144013755408895e-05,
      "loss": 9.798,
      "step": 128
    },
    {
      "epoch": 0.01664730933023616,
      "grad_norm": 2.8185410499572754,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 8.771,
      "step": 129
    },
    {
      "epoch": 0.016776358239772873,
      "grad_norm": 3.1185972690582275,
      "learning_rate": 2.991522876735154e-05,
      "loss": 9.5178,
      "step": 130
    },
    {
      "epoch": 0.01690540714930959,
      "grad_norm": 2.9422974586486816,
      "learning_rate": 2.916090697523549e-05,
      "loss": 8.8508,
      "step": 131
    },
    {
      "epoch": 0.017034456058846303,
      "grad_norm": 3.1368095874786377,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 9.8087,
      "step": 132
    },
    {
      "epoch": 0.017163504968383016,
      "grad_norm": 3.2694547176361084,
      "learning_rate": 2.766955965735968e-05,
      "loss": 9.4227,
      "step": 133
    },
    {
      "epoch": 0.017292553877919732,
      "grad_norm": 3.2808034420013428,
      "learning_rate": 2.693294185106562e-05,
      "loss": 9.1016,
      "step": 134
    },
    {
      "epoch": 0.017421602787456445,
      "grad_norm": 3.4240145683288574,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 9.1178,
      "step": 135
    },
    {
      "epoch": 0.01755065169699316,
      "grad_norm": 3.7069358825683594,
      "learning_rate": 2.547882480847461e-05,
      "loss": 9.8316,
      "step": 136
    },
    {
      "epoch": 0.017679700606529875,
      "grad_norm": 3.5348432064056396,
      "learning_rate": 2.476172311325783e-05,
      "loss": 9.4054,
      "step": 137
    },
    {
      "epoch": 0.01780874951606659,
      "grad_norm": 3.763582468032837,
      "learning_rate": 2.405152131093926e-05,
      "loss": 8.8777,
      "step": 138
    },
    {
      "epoch": 0.017937798425603304,
      "grad_norm": 3.6810343265533447,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 9.0372,
      "step": 139
    },
    {
      "epoch": 0.018066847335140017,
      "grad_norm": 3.93106746673584,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 9.7601,
      "step": 140
    },
    {
      "epoch": 0.018195896244676733,
      "grad_norm": 4.078088760375977,
      "learning_rate": 2.196424713241637e-05,
      "loss": 9.2283,
      "step": 141
    },
    {
      "epoch": 0.018324945154213446,
      "grad_norm": 4.157682418823242,
      "learning_rate": 2.128356686585282e-05,
      "loss": 9.4837,
      "step": 142
    },
    {
      "epoch": 0.018453994063750163,
      "grad_norm": 4.0645751953125,
      "learning_rate": 2.061073738537635e-05,
      "loss": 9.1976,
      "step": 143
    },
    {
      "epoch": 0.018583042973286876,
      "grad_norm": 4.302129745483398,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 9.5948,
      "step": 144
    },
    {
      "epoch": 0.01871209188282359,
      "grad_norm": 5.034425735473633,
      "learning_rate": 1.928936436551661e-05,
      "loss": 9.9455,
      "step": 145
    },
    {
      "epoch": 0.018841140792360305,
      "grad_norm": 4.414882659912109,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 8.9202,
      "step": 146
    },
    {
      "epoch": 0.018970189701897018,
      "grad_norm": 4.810214519500732,
      "learning_rate": 1.800157297483417e-05,
      "loss": 10.1101,
      "step": 147
    },
    {
      "epoch": 0.019099238611433735,
      "grad_norm": 4.97913122177124,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 9.7973,
      "step": 148
    },
    {
      "epoch": 0.019228287520970447,
      "grad_norm": 6.313774108886719,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 9.4023,
      "step": 149
    },
    {
      "epoch": 0.019357336430507164,
      "grad_norm": 6.726715564727783,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 9.5693,
      "step": 150
    },
    {
      "epoch": 0.019357336430507164,
      "eval_loss": 2.3450348377227783,
      "eval_runtime": 1421.6073,
      "eval_samples_per_second": 9.18,
      "eval_steps_per_second": 2.295,
      "step": 150
    },
    {
      "epoch": 0.019486385340043877,
      "grad_norm": 3.5760293006896973,
      "learning_rate": 1.553232954407171e-05,
      "loss": 8.6555,
      "step": 151
    },
    {
      "epoch": 0.01961543424958059,
      "grad_norm": 3.826117992401123,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 8.569,
      "step": 152
    },
    {
      "epoch": 0.019744483159117306,
      "grad_norm": 4.005914211273193,
      "learning_rate": 1.435357758543015e-05,
      "loss": 9.1417,
      "step": 153
    },
    {
      "epoch": 0.01987353206865402,
      "grad_norm": 4.030627250671387,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 8.9886,
      "step": 154
    },
    {
      "epoch": 0.020002580978190736,
      "grad_norm": 3.994656801223755,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 9.0325,
      "step": 155
    },
    {
      "epoch": 0.02013162988772745,
      "grad_norm": 3.5603768825531006,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 8.9488,
      "step": 156
    },
    {
      "epoch": 0.02026067879726416,
      "grad_norm": 3.1016173362731934,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 8.8295,
      "step": 157
    },
    {
      "epoch": 0.020389727706800878,
      "grad_norm": 3.1204254627227783,
      "learning_rate": 1.157994445715706e-05,
      "loss": 9.4366,
      "step": 158
    },
    {
      "epoch": 0.02051877661633759,
      "grad_norm": 2.8520798683166504,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 9.1356,
      "step": 159
    },
    {
      "epoch": 0.020647825525874308,
      "grad_norm": 2.810343027114868,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 9.2959,
      "step": 160
    },
    {
      "epoch": 0.02077687443541102,
      "grad_norm": 2.5594828128814697,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 9.1708,
      "step": 161
    },
    {
      "epoch": 0.020905923344947737,
      "grad_norm": 2.9740025997161865,
      "learning_rate": 9.549150281252633e-06,
      "loss": 8.7406,
      "step": 162
    },
    {
      "epoch": 0.02103497225448445,
      "grad_norm": 2.275719165802002,
      "learning_rate": 9.068759265665384e-06,
      "loss": 9.0083,
      "step": 163
    },
    {
      "epoch": 0.021164021164021163,
      "grad_norm": 2.507056951522827,
      "learning_rate": 8.599558442598998e-06,
      "loss": 9.2756,
      "step": 164
    },
    {
      "epoch": 0.02129307007355788,
      "grad_norm": 2.2246103286743164,
      "learning_rate": 8.141676086873572e-06,
      "loss": 9.209,
      "step": 165
    },
    {
      "epoch": 0.021422118983094592,
      "grad_norm": 2.405937910079956,
      "learning_rate": 7.695237378953223e-06,
      "loss": 8.8886,
      "step": 166
    },
    {
      "epoch": 0.02155116789263131,
      "grad_norm": 2.2762365341186523,
      "learning_rate": 7.260364370723044e-06,
      "loss": 9.1101,
      "step": 167
    },
    {
      "epoch": 0.02168021680216802,
      "grad_norm": 2.4731991291046143,
      "learning_rate": 6.837175952121306e-06,
      "loss": 8.9141,
      "step": 168
    },
    {
      "epoch": 0.021809265711704735,
      "grad_norm": 2.4252381324768066,
      "learning_rate": 6.425787818636131e-06,
      "loss": 9.296,
      "step": 169
    },
    {
      "epoch": 0.02193831462124145,
      "grad_norm": 2.4549665451049805,
      "learning_rate": 6.026312439675552e-06,
      "loss": 9.4637,
      "step": 170
    },
    {
      "epoch": 0.022067363530778164,
      "grad_norm": 2.5885021686553955,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 9.2551,
      "step": 171
    },
    {
      "epoch": 0.02219641244031488,
      "grad_norm": 2.4765894412994385,
      "learning_rate": 5.263533508961827e-06,
      "loss": 9.549,
      "step": 172
    },
    {
      "epoch": 0.022325461349851593,
      "grad_norm": 2.590534210205078,
      "learning_rate": 4.900438493352055e-06,
      "loss": 9.5712,
      "step": 173
    },
    {
      "epoch": 0.02245451025938831,
      "grad_norm": 2.5699732303619385,
      "learning_rate": 4.549673247541875e-06,
      "loss": 8.7675,
      "step": 174
    },
    {
      "epoch": 0.022583559168925023,
      "grad_norm": 2.758275270462036,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 9.0924,
      "step": 175
    },
    {
      "epoch": 0.022712608078461736,
      "grad_norm": 2.9239282608032227,
      "learning_rate": 3.885512251130763e-06,
      "loss": 9.4239,
      "step": 176
    },
    {
      "epoch": 0.022841656987998452,
      "grad_norm": 2.9665844440460205,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 9.1295,
      "step": 177
    },
    {
      "epoch": 0.022970705897535165,
      "grad_norm": 2.919910192489624,
      "learning_rate": 3.271776770026963e-06,
      "loss": 9.5301,
      "step": 178
    },
    {
      "epoch": 0.02309975480707188,
      "grad_norm": 2.7769105434417725,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 9.2018,
      "step": 179
    },
    {
      "epoch": 0.023228803716608595,
      "grad_norm": 2.8075311183929443,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 9.2266,
      "step": 180
    },
    {
      "epoch": 0.023357852626145308,
      "grad_norm": 3.0806944370269775,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 9.5961,
      "step": 181
    },
    {
      "epoch": 0.023486901535682024,
      "grad_norm": 2.935178518295288,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 9.0992,
      "step": 182
    },
    {
      "epoch": 0.023615950445218737,
      "grad_norm": 3.109252691268921,
      "learning_rate": 1.962316193157593e-06,
      "loss": 9.5299,
      "step": 183
    },
    {
      "epoch": 0.023744999354755453,
      "grad_norm": 3.1644036769866943,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 8.9805,
      "step": 184
    },
    {
      "epoch": 0.023874048264292166,
      "grad_norm": 3.3138835430145264,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 9.2192,
      "step": 185
    },
    {
      "epoch": 0.024003097173828883,
      "grad_norm": 3.4772531986236572,
      "learning_rate": 1.333670137599713e-06,
      "loss": 9.9319,
      "step": 186
    },
    {
      "epoch": 0.024132146083365596,
      "grad_norm": 3.1823484897613525,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 8.83,
      "step": 187
    },
    {
      "epoch": 0.02426119499290231,
      "grad_norm": 3.5448269844055176,
      "learning_rate": 9.810017062595322e-07,
      "loss": 9.3681,
      "step": 188
    },
    {
      "epoch": 0.024390243902439025,
      "grad_norm": 3.576612710952759,
      "learning_rate": 8.247462563808817e-07,
      "loss": 9.113,
      "step": 189
    },
    {
      "epoch": 0.024519292811975738,
      "grad_norm": 4.022530555725098,
      "learning_rate": 6.819348298638839e-07,
      "loss": 9.6095,
      "step": 190
    },
    {
      "epoch": 0.024648341721512455,
      "grad_norm": 3.772294521331787,
      "learning_rate": 5.526064699265753e-07,
      "loss": 9.2514,
      "step": 191
    },
    {
      "epoch": 0.024777390631049168,
      "grad_norm": 3.9957714080810547,
      "learning_rate": 4.367965336512403e-07,
      "loss": 9.3345,
      "step": 192
    },
    {
      "epoch": 0.02490643954058588,
      "grad_norm": 4.381701946258545,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 10.194,
      "step": 193
    },
    {
      "epoch": 0.025035488450122597,
      "grad_norm": 4.343029022216797,
      "learning_rate": 2.458548727494292e-07,
      "loss": 9.8201,
      "step": 194
    },
    {
      "epoch": 0.02516453735965931,
      "grad_norm": 5.34747314453125,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.5577,
      "step": 195
    },
    {
      "epoch": 0.025293586269196026,
      "grad_norm": 5.110474586486816,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 9.82,
      "step": 196
    },
    {
      "epoch": 0.02542263517873274,
      "grad_norm": 5.267843246459961,
      "learning_rate": 6.150154258476315e-08,
      "loss": 9.6086,
      "step": 197
    },
    {
      "epoch": 0.025551684088269456,
      "grad_norm": 5.388737678527832,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 10.0599,
      "step": 198
    },
    {
      "epoch": 0.02568073299780617,
      "grad_norm": 6.308083534240723,
      "learning_rate": 6.834750376549792e-09,
      "loss": 10.3258,
      "step": 199
    },
    {
      "epoch": 0.025809781907342882,
      "grad_norm": 7.229769229888916,
      "learning_rate": 0.0,
      "loss": 9.957,
      "step": 200
    },
    {
      "epoch": 0.025809781907342882,
      "eval_loss": 2.3232147693634033,
      "eval_runtime": 1422.6911,
      "eval_samples_per_second": 9.173,
      "eval_steps_per_second": 2.294,
      "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": 4.665029231326003e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}