{
  "best_metric": 0.657630980014801,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 1.4678899082568808,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007339449541284404,
      "grad_norm": 1.2576438188552856,
      "learning_rate": 1e-05,
      "loss": 0.9112,
      "step": 1
    },
    {
      "epoch": 0.007339449541284404,
      "eval_loss": 1.086421012878418,
      "eval_runtime": 7.3885,
      "eval_samples_per_second": 31.129,
      "eval_steps_per_second": 7.85,
      "step": 1
    },
    {
      "epoch": 0.014678899082568808,
      "grad_norm": 1.320379614830017,
      "learning_rate": 2e-05,
      "loss": 0.9991,
      "step": 2
    },
    {
      "epoch": 0.022018348623853212,
      "grad_norm": 1.2803949117660522,
      "learning_rate": 3e-05,
      "loss": 1.0084,
      "step": 3
    },
    {
      "epoch": 0.029357798165137616,
      "grad_norm": 1.0986052751541138,
      "learning_rate": 4e-05,
      "loss": 0.8749,
      "step": 4
    },
    {
      "epoch": 0.03669724770642202,
      "grad_norm": 0.9205357432365417,
      "learning_rate": 5e-05,
      "loss": 0.9294,
      "step": 5
    },
    {
      "epoch": 0.044036697247706424,
      "grad_norm": 0.8968833684921265,
      "learning_rate": 6e-05,
      "loss": 0.6873,
      "step": 6
    },
    {
      "epoch": 0.05137614678899083,
      "grad_norm": 0.7835147976875305,
      "learning_rate": 7e-05,
      "loss": 0.8424,
      "step": 7
    },
    {
      "epoch": 0.05871559633027523,
      "grad_norm": 0.638231098651886,
      "learning_rate": 8e-05,
      "loss": 0.8373,
      "step": 8
    },
    {
      "epoch": 0.06605504587155964,
      "grad_norm": 0.8740332722663879,
      "learning_rate": 9e-05,
      "loss": 0.8565,
      "step": 9
    },
    {
      "epoch": 0.07339449541284404,
      "grad_norm": 0.7018423676490784,
      "learning_rate": 0.0001,
      "loss": 0.7369,
      "step": 10
    },
    {
      "epoch": 0.08073394495412844,
      "grad_norm": 0.7135981917381287,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.671,
      "step": 11
    },
    {
      "epoch": 0.08807339449541285,
      "grad_norm": 0.791584312915802,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.7039,
      "step": 12
    },
    {
      "epoch": 0.09541284403669725,
      "grad_norm": 0.6780750751495361,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.7024,
      "step": 13
    },
    {
      "epoch": 0.10275229357798166,
      "grad_norm": 0.6978002190589905,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.8127,
      "step": 14
    },
    {
      "epoch": 0.11009174311926606,
      "grad_norm": 0.6280326843261719,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.7367,
      "step": 15
    },
    {
      "epoch": 0.11743119266055047,
      "grad_norm": 0.5914128422737122,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.7386,
      "step": 16
    },
    {
      "epoch": 0.12477064220183487,
      "grad_norm": 0.5690916180610657,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.6967,
      "step": 17
    },
    {
      "epoch": 0.13211009174311927,
      "grad_norm": 0.6618253588676453,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.8074,
      "step": 18
    },
    {
      "epoch": 0.13944954128440368,
      "grad_norm": 0.5351158976554871,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.7263,
      "step": 19
    },
    {
      "epoch": 0.14678899082568808,
      "grad_norm": 0.5452873110771179,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.7654,
      "step": 20
    },
    {
      "epoch": 0.15412844036697249,
      "grad_norm": 0.5408823490142822,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.6872,
      "step": 21
    },
    {
      "epoch": 0.1614678899082569,
      "grad_norm": 0.5791258811950684,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.7021,
      "step": 22
    },
    {
      "epoch": 0.1688073394495413,
      "grad_norm": 0.6007850170135498,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.7315,
      "step": 23
    },
    {
      "epoch": 0.1761467889908257,
      "grad_norm": 0.5800233483314514,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.7132,
      "step": 24
    },
    {
      "epoch": 0.1834862385321101,
      "grad_norm": 0.6535601019859314,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.7046,
      "step": 25
    },
    {
      "epoch": 0.1908256880733945,
      "grad_norm": 0.7141433954238892,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.8598,
      "step": 26
    },
    {
      "epoch": 0.1981651376146789,
      "grad_norm": 0.6294669508934021,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.9351,
      "step": 27
    },
    {
      "epoch": 0.20550458715596331,
      "grad_norm": 0.6602486371994019,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.8266,
      "step": 28
    },
    {
      "epoch": 0.21284403669724772,
      "grad_norm": 0.6365261077880859,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.8377,
      "step": 29
    },
    {
      "epoch": 0.22018348623853212,
      "grad_norm": 0.6225994229316711,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.8686,
      "step": 30
    },
    {
      "epoch": 0.22752293577981653,
      "grad_norm": 0.6388337016105652,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.8258,
      "step": 31
    },
    {
      "epoch": 0.23486238532110093,
      "grad_norm": 0.6848137974739075,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.9546,
      "step": 32
    },
    {
      "epoch": 0.24220183486238533,
      "grad_norm": 0.760630190372467,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.9698,
      "step": 33
    },
    {
      "epoch": 0.24954128440366974,
      "grad_norm": 1.4575302600860596,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.289,
      "step": 34
    },
    {
      "epoch": 0.25688073394495414,
      "grad_norm": 0.4609055817127228,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.54,
      "step": 35
    },
    {
      "epoch": 0.26422018348623855,
      "grad_norm": 0.47219833731651306,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.5379,
      "step": 36
    },
    {
      "epoch": 0.27155963302752295,
      "grad_norm": 0.43857619166374207,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.5831,
      "step": 37
    },
    {
      "epoch": 0.27889908256880735,
      "grad_norm": 0.44134005904197693,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.6643,
      "step": 38
    },
    {
      "epoch": 0.28623853211009176,
      "grad_norm": 0.3944750726222992,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.5563,
      "step": 39
    },
    {
      "epoch": 0.29357798165137616,
      "grad_norm": 0.39067232608795166,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.593,
      "step": 40
    },
    {
      "epoch": 0.30091743119266057,
      "grad_norm": 0.5718076229095459,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.5941,
      "step": 41
    },
    {
      "epoch": 0.30825688073394497,
      "grad_norm": 0.3971261978149414,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.7517,
      "step": 42
    },
    {
      "epoch": 0.3155963302752294,
      "grad_norm": 0.3980463147163391,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.7895,
      "step": 43
    },
    {
      "epoch": 0.3229357798165138,
      "grad_norm": 0.3441670536994934,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.4756,
      "step": 44
    },
    {
      "epoch": 0.3302752293577982,
      "grad_norm": 0.3822227418422699,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.5701,
      "step": 45
    },
    {
      "epoch": 0.3376146788990826,
      "grad_norm": 0.40735456347465515,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.6265,
      "step": 46
    },
    {
      "epoch": 0.344954128440367,
      "grad_norm": 0.4540577232837677,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.699,
      "step": 47
    },
    {
      "epoch": 0.3522935779816514,
      "grad_norm": 0.4145893454551697,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.7058,
      "step": 48
    },
    {
      "epoch": 0.3596330275229358,
      "grad_norm": 0.4242461025714874,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.6845,
      "step": 49
    },
    {
      "epoch": 0.3669724770642202,
      "grad_norm": 0.4225294291973114,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.5969,
      "step": 50
    },
    {
      "epoch": 0.3669724770642202,
      "eval_loss": 0.7089372873306274,
      "eval_runtime": 7.3832,
      "eval_samples_per_second": 31.152,
      "eval_steps_per_second": 7.856,
      "step": 50
    },
    {
      "epoch": 0.3743119266055046,
      "grad_norm": 0.45440104603767395,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.6622,
      "step": 51
    },
    {
      "epoch": 0.381651376146789,
      "grad_norm": 0.4386112093925476,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.6675,
      "step": 52
    },
    {
      "epoch": 0.3889908256880734,
      "grad_norm": 0.43003344535827637,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.6793,
      "step": 53
    },
    {
      "epoch": 0.3963302752293578,
      "grad_norm": 0.5001123547554016,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.6817,
      "step": 54
    },
    {
      "epoch": 0.4036697247706422,
      "grad_norm": 0.5568297505378723,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.6538,
      "step": 55
    },
    {
      "epoch": 0.41100917431192663,
      "grad_norm": 0.44233083724975586,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.717,
      "step": 56
    },
    {
      "epoch": 0.41834862385321103,
      "grad_norm": 0.5203274488449097,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.6568,
      "step": 57
    },
    {
      "epoch": 0.42568807339449544,
      "grad_norm": 0.529590368270874,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.7246,
      "step": 58
    },
    {
      "epoch": 0.43302752293577984,
      "grad_norm": 0.47836318612098694,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.6922,
      "step": 59
    },
    {
      "epoch": 0.44036697247706424,
      "grad_norm": 0.5171796083450317,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.7058,
      "step": 60
    },
    {
      "epoch": 0.44770642201834865,
      "grad_norm": 0.49430549144744873,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.6869,
      "step": 61
    },
    {
      "epoch": 0.45504587155963305,
      "grad_norm": 0.5193171501159668,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.7945,
      "step": 62
    },
    {
      "epoch": 0.46238532110091746,
      "grad_norm": 0.6313068866729736,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.7861,
      "step": 63
    },
    {
      "epoch": 0.46972477064220186,
      "grad_norm": 0.6041170358657837,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.7491,
      "step": 64
    },
    {
      "epoch": 0.47706422018348627,
      "grad_norm": 0.6043881773948669,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.9072,
      "step": 65
    },
    {
      "epoch": 0.48440366972477067,
      "grad_norm": 0.6289061307907104,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.9472,
      "step": 66
    },
    {
      "epoch": 0.4917431192660551,
      "grad_norm": 0.7074716091156006,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.0706,
      "step": 67
    },
    {
      "epoch": 0.4990825688073395,
      "grad_norm": 1.0272139310836792,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.249,
      "step": 68
    },
    {
      "epoch": 0.5064220183486239,
      "grad_norm": 0.4037744998931885,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.5478,
      "step": 69
    },
    {
      "epoch": 0.5137614678899083,
      "grad_norm": 0.4181773364543915,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.6575,
      "step": 70
    },
    {
      "epoch": 0.5211009174311927,
      "grad_norm": 0.4181150496006012,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.5975,
      "step": 71
    },
    {
      "epoch": 0.5284403669724771,
      "grad_norm": 0.3705148994922638,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.5462,
      "step": 72
    },
    {
      "epoch": 0.5357798165137615,
      "grad_norm": 0.3600141704082489,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.5123,
      "step": 73
    },
    {
      "epoch": 0.5431192660550459,
      "grad_norm": 0.34120720624923706,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.5915,
      "step": 74
    },
    {
      "epoch": 0.5504587155963303,
      "grad_norm": 0.3528464436531067,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5901,
      "step": 75
    },
    {
      "epoch": 0.5577981651376147,
      "grad_norm": 0.3352064788341522,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.6071,
      "step": 76
    },
    {
      "epoch": 0.5651376146788991,
      "grad_norm": 0.32289859652519226,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.5741,
      "step": 77
    },
    {
      "epoch": 0.5724770642201835,
      "grad_norm": 0.32375475764274597,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.526,
      "step": 78
    },
    {
      "epoch": 0.5798165137614679,
      "grad_norm": 0.3686944544315338,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.6329,
      "step": 79
    },
    {
      "epoch": 0.5871559633027523,
      "grad_norm": 0.40908193588256836,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.5515,
      "step": 80
    },
    {
      "epoch": 0.5944954128440367,
      "grad_norm": 0.4456939697265625,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.6521,
      "step": 81
    },
    {
      "epoch": 0.6018348623853211,
      "grad_norm": 0.3769167959690094,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.5615,
      "step": 82
    },
    {
      "epoch": 0.6091743119266055,
      "grad_norm": 0.41746941208839417,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.6451,
      "step": 83
    },
    {
      "epoch": 0.6165137614678899,
      "grad_norm": 0.3857053816318512,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.6505,
      "step": 84
    },
    {
      "epoch": 0.6238532110091743,
      "grad_norm": 0.37709927558898926,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.6044,
      "step": 85
    },
    {
      "epoch": 0.6311926605504588,
      "grad_norm": 0.3770582675933838,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.5986,
      "step": 86
    },
    {
      "epoch": 0.6385321100917432,
      "grad_norm": 0.4132257103919983,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.6329,
      "step": 87
    },
    {
      "epoch": 0.6458715596330276,
      "grad_norm": 0.40162786841392517,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.6208,
      "step": 88
    },
    {
      "epoch": 0.653211009174312,
      "grad_norm": 0.428101122379303,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.6981,
      "step": 89
    },
    {
      "epoch": 0.6605504587155964,
      "grad_norm": 0.41267475485801697,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.6747,
      "step": 90
    },
    {
      "epoch": 0.6678899082568808,
      "grad_norm": 0.4253780245780945,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.6754,
      "step": 91
    },
    {
      "epoch": 0.6752293577981652,
      "grad_norm": 0.4702104330062866,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.6771,
      "step": 92
    },
    {
      "epoch": 0.6825688073394496,
      "grad_norm": 0.4644729495048523,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.7873,
      "step": 93
    },
    {
      "epoch": 0.689908256880734,
      "grad_norm": 0.5347480177879333,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.776,
      "step": 94
    },
    {
      "epoch": 0.6972477064220184,
      "grad_norm": 0.48614299297332764,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.6903,
      "step": 95
    },
    {
      "epoch": 0.7045871559633028,
      "grad_norm": 0.5975285768508911,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.7214,
      "step": 96
    },
    {
      "epoch": 0.7119266055045872,
      "grad_norm": 0.5876331329345703,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.6721,
      "step": 97
    },
    {
      "epoch": 0.7192660550458716,
      "grad_norm": 0.6095494031906128,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.8158,
      "step": 98
    },
    {
      "epoch": 0.726605504587156,
      "grad_norm": 0.5459724068641663,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.793,
      "step": 99
    },
    {
      "epoch": 0.7339449541284404,
      "grad_norm": 0.5794693827629089,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.8975,
      "step": 100
    },
    {
      "epoch": 0.7339449541284404,
      "eval_loss": 0.6698483228683472,
      "eval_runtime": 7.4072,
      "eval_samples_per_second": 31.051,
      "eval_steps_per_second": 7.83,
      "step": 100
    },
    {
      "epoch": 0.7412844036697248,
      "grad_norm": 0.7477515339851379,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.0842,
      "step": 101
    },
    {
      "epoch": 0.7486238532110092,
      "grad_norm": 0.94322669506073,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.0228,
      "step": 102
    },
    {
      "epoch": 0.7559633027522936,
      "grad_norm": 0.36179572343826294,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.5232,
      "step": 103
    },
    {
      "epoch": 0.763302752293578,
      "grad_norm": 0.3528522849082947,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.5148,
      "step": 104
    },
    {
      "epoch": 0.7706422018348624,
      "grad_norm": 0.3581383228302002,
      "learning_rate": 5e-05,
      "loss": 0.6041,
      "step": 105
    },
    {
      "epoch": 0.7779816513761468,
      "grad_norm": 0.3716886043548584,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.5869,
      "step": 106
    },
    {
      "epoch": 0.7853211009174312,
      "grad_norm": 0.34180161356925964,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.622,
      "step": 107
    },
    {
      "epoch": 0.7926605504587156,
      "grad_norm": 0.3448285758495331,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.6203,
      "step": 108
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.337371289730072,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.6097,
      "step": 109
    },
    {
      "epoch": 0.8073394495412844,
      "grad_norm": 0.3692116439342499,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.5712,
      "step": 110
    },
    {
      "epoch": 0.8146788990825689,
      "grad_norm": 0.3329354524612427,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.4834,
      "step": 111
    },
    {
      "epoch": 0.8220183486238533,
      "grad_norm": 0.2949000597000122,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.4317,
      "step": 112
    },
    {
      "epoch": 0.8293577981651377,
      "grad_norm": 0.3322964608669281,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.5143,
      "step": 113
    },
    {
      "epoch": 0.8366972477064221,
      "grad_norm": 0.3326318562030792,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.6001,
      "step": 114
    },
    {
      "epoch": 0.8440366972477065,
      "grad_norm": 0.3536280393600464,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.6035,
      "step": 115
    },
    {
      "epoch": 0.8513761467889909,
      "grad_norm": 0.3589414060115814,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.5224,
      "step": 116
    },
    {
      "epoch": 0.8587155963302753,
      "grad_norm": 0.4147566854953766,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.6198,
      "step": 117
    },
    {
      "epoch": 0.8660550458715597,
      "grad_norm": 0.3640088140964508,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.5701,
      "step": 118
    },
    {
      "epoch": 0.8733944954128441,
      "grad_norm": 0.39355286955833435,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.6648,
      "step": 119
    },
    {
      "epoch": 0.8807339449541285,
      "grad_norm": 0.42057371139526367,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.631,
      "step": 120
    },
    {
      "epoch": 0.8880733944954129,
      "grad_norm": 0.39816609025001526,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.5714,
      "step": 121
    },
    {
      "epoch": 0.8954128440366973,
      "grad_norm": 0.4015522599220276,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.6463,
      "step": 122
    },
    {
      "epoch": 0.9027522935779817,
      "grad_norm": 0.4073263704776764,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.6142,
      "step": 123
    },
    {
      "epoch": 0.9100917431192661,
      "grad_norm": 0.4220086336135864,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.607,
      "step": 124
    },
    {
      "epoch": 0.9174311926605505,
      "grad_norm": 0.46245715022087097,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.6343,
      "step": 125
    },
    {
      "epoch": 0.9247706422018349,
      "grad_norm": 0.4634447395801544,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.676,
      "step": 126
    },
    {
      "epoch": 0.9321100917431193,
      "grad_norm": 0.44423627853393555,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.7772,
      "step": 127
    },
    {
      "epoch": 0.9394495412844037,
      "grad_norm": 0.5185724496841431,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.7921,
      "step": 128
    },
    {
      "epoch": 0.9467889908256881,
      "grad_norm": 0.4980521500110626,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.7116,
      "step": 129
    },
    {
      "epoch": 0.9541284403669725,
      "grad_norm": 0.4910784959793091,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.6899,
      "step": 130
    },
    {
      "epoch": 0.9614678899082569,
      "grad_norm": 0.49023112654685974,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.7204,
      "step": 131
    },
    {
      "epoch": 0.9688073394495413,
      "grad_norm": 0.5643883347511292,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.9618,
      "step": 132
    },
    {
      "epoch": 0.9761467889908257,
      "grad_norm": 0.5237534046173096,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.7147,
      "step": 133
    },
    {
      "epoch": 0.9834862385321101,
      "grad_norm": 0.5938643217086792,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.8781,
      "step": 134
    },
    {
      "epoch": 0.9908256880733946,
      "grad_norm": 0.717574417591095,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.9911,
      "step": 135
    },
    {
      "epoch": 0.998165137614679,
      "grad_norm": 0.8267049789428711,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.0187,
      "step": 136
    },
    {
      "epoch": 1.0055045871559634,
      "grad_norm": 2.545386552810669,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.3663,
      "step": 137
    },
    {
      "epoch": 1.0128440366972478,
      "grad_norm": 0.30662229657173157,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.4236,
      "step": 138
    },
    {
      "epoch": 1.0201834862385322,
      "grad_norm": 0.32362738251686096,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.5572,
      "step": 139
    },
    {
      "epoch": 1.0275229357798166,
      "grad_norm": 0.3168981969356537,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.5269,
      "step": 140
    },
    {
      "epoch": 1.034862385321101,
      "grad_norm": 0.31193283200263977,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.572,
      "step": 141
    },
    {
      "epoch": 1.0422018348623854,
      "grad_norm": 0.32214057445526123,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.5702,
      "step": 142
    },
    {
      "epoch": 1.0495412844036698,
      "grad_norm": 0.31561291217803955,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.5514,
      "step": 143
    },
    {
      "epoch": 1.0568807339449542,
      "grad_norm": 0.3159513771533966,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.5772,
      "step": 144
    },
    {
      "epoch": 1.0642201834862386,
      "grad_norm": 0.3018777370452881,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.535,
      "step": 145
    },
    {
      "epoch": 1.071559633027523,
      "grad_norm": 0.3041372001171112,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.5148,
      "step": 146
    },
    {
      "epoch": 1.0788990825688074,
      "grad_norm": 0.3117336630821228,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.4676,
      "step": 147
    },
    {
      "epoch": 1.0862385321100918,
      "grad_norm": 0.312052458524704,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.4992,
      "step": 148
    },
    {
      "epoch": 1.0935779816513762,
      "grad_norm": 0.3615710437297821,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.5694,
      "step": 149
    },
    {
      "epoch": 1.1009174311926606,
      "grad_norm": 0.3654485046863556,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.6718,
      "step": 150
    },
    {
      "epoch": 1.1009174311926606,
      "eval_loss": 0.657630980014801,
      "eval_runtime": 7.3958,
      "eval_samples_per_second": 31.099,
      "eval_steps_per_second": 7.842,
      "step": 150
    },
    {
      "epoch": 1.108256880733945,
      "grad_norm": 0.35208287835121155,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.5099,
      "step": 151
    },
    {
      "epoch": 1.1155963302752294,
      "grad_norm": 0.3770894706249237,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.6071,
      "step": 152
    },
    {
      "epoch": 1.1229357798165138,
      "grad_norm": 0.3409969210624695,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.4753,
      "step": 153
    },
    {
      "epoch": 1.1302752293577982,
      "grad_norm": 0.39421525597572327,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.6188,
      "step": 154
    },
    {
      "epoch": 1.1376146788990826,
      "grad_norm": 0.38094234466552734,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.6225,
      "step": 155
    },
    {
      "epoch": 1.144954128440367,
      "grad_norm": 0.3630848824977875,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.5918,
      "step": 156
    },
    {
      "epoch": 1.1522935779816514,
      "grad_norm": 0.4022403061389923,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.637,
      "step": 157
    },
    {
      "epoch": 1.1596330275229358,
      "grad_norm": 0.3792831003665924,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.5433,
      "step": 158
    },
    {
      "epoch": 1.1669724770642202,
      "grad_norm": 0.39754539728164673,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.5978,
      "step": 159
    },
    {
      "epoch": 1.1743119266055047,
      "grad_norm": 0.39391881227493286,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.6028,
      "step": 160
    },
    {
      "epoch": 1.181651376146789,
      "grad_norm": 0.4423828125,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.5758,
      "step": 161
    },
    {
      "epoch": 1.1889908256880735,
      "grad_norm": 0.4398263692855835,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.7979,
      "step": 162
    },
    {
      "epoch": 1.1963302752293579,
      "grad_norm": 0.45425429940223694,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.7243,
      "step": 163
    },
    {
      "epoch": 1.2036697247706423,
      "grad_norm": 0.4375683665275574,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.6185,
      "step": 164
    },
    {
      "epoch": 1.2110091743119267,
      "grad_norm": 0.4564600884914398,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.676,
      "step": 165
    },
    {
      "epoch": 1.218348623853211,
      "grad_norm": 0.47862768173217773,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.7066,
      "step": 166
    },
    {
      "epoch": 1.2256880733944955,
      "grad_norm": 0.5144950747489929,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.7415,
      "step": 167
    },
    {
      "epoch": 1.2330275229357799,
      "grad_norm": 0.4956972301006317,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.6938,
      "step": 168
    },
    {
      "epoch": 1.2403669724770643,
      "grad_norm": 0.6631423234939575,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.8086,
      "step": 169
    },
    {
      "epoch": 1.2477064220183487,
      "grad_norm": 0.7612147927284241,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.9677,
      "step": 170
    },
    {
      "epoch": 1.255045871559633,
      "grad_norm": 0.4201236665248871,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.4762,
      "step": 171
    },
    {
      "epoch": 1.2623853211009175,
      "grad_norm": 0.3686876595020294,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.5071,
      "step": 172
    },
    {
      "epoch": 1.269724770642202,
      "grad_norm": 0.29836687445640564,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.4445,
      "step": 173
    },
    {
      "epoch": 1.2770642201834863,
      "grad_norm": 0.31192854046821594,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.5124,
      "step": 174
    },
    {
      "epoch": 1.2844036697247707,
      "grad_norm": 0.3132554888725281,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.5598,
      "step": 175
    },
    {
      "epoch": 1.2917431192660551,
      "grad_norm": 0.3496793806552887,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.5556,
      "step": 176
    },
    {
      "epoch": 1.2990825688073395,
      "grad_norm": 0.31457391381263733,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.4844,
      "step": 177
    },
    {
      "epoch": 1.306422018348624,
      "grad_norm": 0.3034902513027191,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.4528,
      "step": 178
    },
    {
      "epoch": 1.3137614678899083,
      "grad_norm": 0.3396747410297394,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.5139,
      "step": 179
    },
    {
      "epoch": 1.3211009174311927,
      "grad_norm": 0.3272293210029602,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.5646,
      "step": 180
    },
    {
      "epoch": 1.3284403669724771,
      "grad_norm": 0.35281357169151306,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.6153,
      "step": 181
    },
    {
      "epoch": 1.3357798165137615,
      "grad_norm": 0.35483184456825256,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.5263,
      "step": 182
    },
    {
      "epoch": 1.343119266055046,
      "grad_norm": 0.36175045371055603,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.5793,
      "step": 183
    },
    {
      "epoch": 1.3504587155963304,
      "grad_norm": 0.3474538326263428,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.5315,
      "step": 184
    },
    {
      "epoch": 1.3577981651376148,
      "grad_norm": 0.35895705223083496,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.51,
      "step": 185
    },
    {
      "epoch": 1.3651376146788992,
      "grad_norm": 0.37117087841033936,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.5293,
      "step": 186
    },
    {
      "epoch": 1.3724770642201836,
      "grad_norm": 0.3558691442012787,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.4994,
      "step": 187
    },
    {
      "epoch": 1.379816513761468,
      "grad_norm": 0.3714943528175354,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.6044,
      "step": 188
    },
    {
      "epoch": 1.3871559633027524,
      "grad_norm": 0.3848516345024109,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.5971,
      "step": 189
    },
    {
      "epoch": 1.3944954128440368,
      "grad_norm": 0.3662392199039459,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.5318,
      "step": 190
    },
    {
      "epoch": 1.4018348623853212,
      "grad_norm": 0.3739279508590698,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.5653,
      "step": 191
    },
    {
      "epoch": 1.4091743119266056,
      "grad_norm": 0.38193321228027344,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.5314,
      "step": 192
    },
    {
      "epoch": 1.41651376146789,
      "grad_norm": 0.3724275529384613,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.5957,
      "step": 193
    },
    {
      "epoch": 1.4238532110091744,
      "grad_norm": 0.4420280456542969,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.6163,
      "step": 194
    },
    {
      "epoch": 1.4311926605504588,
      "grad_norm": 0.40835294127464294,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.6157,
      "step": 195
    },
    {
      "epoch": 1.4385321100917432,
      "grad_norm": 0.4198923408985138,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.6766,
      "step": 196
    },
    {
      "epoch": 1.4458715596330276,
      "grad_norm": 0.43725141882896423,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.6375,
      "step": 197
    },
    {
      "epoch": 1.453211009174312,
      "grad_norm": 0.43340611457824707,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.6489,
      "step": 198
    },
    {
      "epoch": 1.4605504587155964,
      "grad_norm": 0.4456269443035126,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.597,
      "step": 199
    },
    {
      "epoch": 1.4678899082568808,
      "grad_norm": 0.4575132727622986,
      "learning_rate": 0.0,
      "loss": 0.6429,
      "step": 200
    },
    {
      "epoch": 1.4678899082568808,
      "eval_loss": 0.6582860350608826,
      "eval_runtime": 7.388,
      "eval_samples_per_second": 31.131,
      "eval_steps_per_second": 7.851,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.088494791806157e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}