{
  "best_metric": 0.1271982491016388,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 2.572347266881029,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.012861736334405145,
      "grad_norm": 4.657966136932373,
      "learning_rate": 1e-05,
      "loss": 0.8068,
      "step": 1
    },
    {
      "epoch": 0.012861736334405145,
      "eval_loss": 0.33266839385032654,
      "eval_runtime": 12.2161,
      "eval_samples_per_second": 10.724,
      "eval_steps_per_second": 2.701,
      "step": 1
    },
    {
      "epoch": 0.02572347266881029,
      "grad_norm": 5.956973552703857,
      "learning_rate": 2e-05,
      "loss": 0.9749,
      "step": 2
    },
    {
      "epoch": 0.03858520900321544,
      "grad_norm": 5.02043342590332,
      "learning_rate": 3e-05,
      "loss": 1.0221,
      "step": 3
    },
    {
      "epoch": 0.05144694533762058,
      "grad_norm": 4.180272102355957,
      "learning_rate": 4e-05,
      "loss": 0.8701,
      "step": 4
    },
    {
      "epoch": 0.06430868167202572,
      "grad_norm": 4.415040969848633,
      "learning_rate": 5e-05,
      "loss": 0.8123,
      "step": 5
    },
    {
      "epoch": 0.07717041800643087,
      "grad_norm": 6.058294296264648,
      "learning_rate": 6e-05,
      "loss": 0.8215,
      "step": 6
    },
    {
      "epoch": 0.09003215434083602,
      "grad_norm": 4.734410285949707,
      "learning_rate": 7e-05,
      "loss": 0.8118,
      "step": 7
    },
    {
      "epoch": 0.10289389067524116,
      "grad_norm": 3.4613609313964844,
      "learning_rate": 8e-05,
      "loss": 0.6817,
      "step": 8
    },
    {
      "epoch": 0.1157556270096463,
      "grad_norm": 3.305318832397461,
      "learning_rate": 9e-05,
      "loss": 0.7011,
      "step": 9
    },
    {
      "epoch": 0.12861736334405144,
      "grad_norm": 3.3293144702911377,
      "learning_rate": 0.0001,
      "loss": 0.6714,
      "step": 10
    },
    {
      "epoch": 0.1414790996784566,
      "grad_norm": 3.1625776290893555,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.681,
      "step": 11
    },
    {
      "epoch": 0.15434083601286175,
      "grad_norm": 2.8392910957336426,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.5941,
      "step": 12
    },
    {
      "epoch": 0.16720257234726688,
      "grad_norm": 3.101907730102539,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.611,
      "step": 13
    },
    {
      "epoch": 0.18006430868167203,
      "grad_norm": 3.5103650093078613,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.652,
      "step": 14
    },
    {
      "epoch": 0.19292604501607716,
      "grad_norm": 3.8682022094726562,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.6699,
      "step": 15
    },
    {
      "epoch": 0.2057877813504823,
      "grad_norm": 3.6441659927368164,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.6599,
      "step": 16
    },
    {
      "epoch": 0.21864951768488747,
      "grad_norm": 3.809445858001709,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.574,
      "step": 17
    },
    {
      "epoch": 0.2315112540192926,
      "grad_norm": 23.959035873413086,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.5928,
      "step": 18
    },
    {
      "epoch": 0.24437299035369775,
      "grad_norm": 15.54284954071045,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.8449,
      "step": 19
    },
    {
      "epoch": 0.2572347266881029,
      "grad_norm": 2.597175359725952,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.698,
      "step": 20
    },
    {
      "epoch": 0.27009646302250806,
      "grad_norm": 2.7330052852630615,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.6823,
      "step": 21
    },
    {
      "epoch": 0.2829581993569132,
      "grad_norm": 2.402292013168335,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.672,
      "step": 22
    },
    {
      "epoch": 0.2958199356913183,
      "grad_norm": 2.4805665016174316,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.7418,
      "step": 23
    },
    {
      "epoch": 0.3086816720257235,
      "grad_norm": 3.5304245948791504,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.6791,
      "step": 24
    },
    {
      "epoch": 0.3215434083601286,
      "grad_norm": 2.4033923149108887,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.5491,
      "step": 25
    },
    {
      "epoch": 0.33440514469453375,
      "grad_norm": 8.79857063293457,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.6595,
      "step": 26
    },
    {
      "epoch": 0.34726688102893893,
      "grad_norm": 2.726418972015381,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.6144,
      "step": 27
    },
    {
      "epoch": 0.36012861736334406,
      "grad_norm": 2.586483955383301,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.6029,
      "step": 28
    },
    {
      "epoch": 0.3729903536977492,
      "grad_norm": 2.677011013031006,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.6229,
      "step": 29
    },
    {
      "epoch": 0.3858520900321543,
      "grad_norm": 2.5549943447113037,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.5839,
      "step": 30
    },
    {
      "epoch": 0.3987138263665595,
      "grad_norm": 2.3411028385162354,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.5922,
      "step": 31
    },
    {
      "epoch": 0.4115755627009646,
      "grad_norm": 2.2514898777008057,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.5846,
      "step": 32
    },
    {
      "epoch": 0.42443729903536975,
      "grad_norm": 2.266422748565674,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.5518,
      "step": 33
    },
    {
      "epoch": 0.43729903536977494,
      "grad_norm": 2.5696330070495605,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.5678,
      "step": 34
    },
    {
      "epoch": 0.45016077170418006,
      "grad_norm": 2.581017017364502,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.5695,
      "step": 35
    },
    {
      "epoch": 0.4630225080385852,
      "grad_norm": 2.769589424133301,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.5163,
      "step": 36
    },
    {
      "epoch": 0.4758842443729904,
      "grad_norm": 3.0649688243865967,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.6094,
      "step": 37
    },
    {
      "epoch": 0.4887459807073955,
      "grad_norm": 3.4942679405212402,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.5095,
      "step": 38
    },
    {
      "epoch": 0.5016077170418006,
      "grad_norm": 2.097593069076538,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.556,
      "step": 39
    },
    {
      "epoch": 0.5144694533762058,
      "grad_norm": 2.3842356204986572,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.6736,
      "step": 40
    },
    {
      "epoch": 0.5273311897106109,
      "grad_norm": 1.9987016916275024,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.654,
      "step": 41
    },
    {
      "epoch": 0.5401929260450161,
      "grad_norm": 2.1255781650543213,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.6425,
      "step": 42
    },
    {
      "epoch": 0.5530546623794212,
      "grad_norm": 1.8313132524490356,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.5481,
      "step": 43
    },
    {
      "epoch": 0.5659163987138264,
      "grad_norm": 1.8115752935409546,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.4981,
      "step": 44
    },
    {
      "epoch": 0.5787781350482315,
      "grad_norm": 1.8278248310089111,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.5915,
      "step": 45
    },
    {
      "epoch": 0.5916398713826366,
      "grad_norm": 2.037808418273926,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.5385,
      "step": 46
    },
    {
      "epoch": 0.6045016077170418,
      "grad_norm": 1.866446614265442,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.4693,
      "step": 47
    },
    {
      "epoch": 0.617363344051447,
      "grad_norm": 2.0997281074523926,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.5544,
      "step": 48
    },
    {
      "epoch": 0.6302250803858521,
      "grad_norm": 2.162992000579834,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.4923,
      "step": 49
    },
    {
      "epoch": 0.6430868167202572,
      "grad_norm": 2.618288040161133,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.608,
      "step": 50
    },
    {
      "epoch": 0.6430868167202572,
      "eval_loss": 0.13918624818325043,
      "eval_runtime": 12.4602,
      "eval_samples_per_second": 10.513,
      "eval_steps_per_second": 2.648,
      "step": 50
    },
    {
      "epoch": 0.6559485530546624,
      "grad_norm": 2.701051950454712,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.6194,
      "step": 51
    },
    {
      "epoch": 0.6688102893890675,
      "grad_norm": 2.478145122528076,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.6276,
      "step": 52
    },
    {
      "epoch": 0.6816720257234726,
      "grad_norm": 2.3256773948669434,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.5805,
      "step": 53
    },
    {
      "epoch": 0.6945337620578779,
      "grad_norm": 2.4030654430389404,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.6028,
      "step": 54
    },
    {
      "epoch": 0.707395498392283,
      "grad_norm": 2.430856704711914,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.5498,
      "step": 55
    },
    {
      "epoch": 0.7202572347266881,
      "grad_norm": 2.3979735374450684,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.4958,
      "step": 56
    },
    {
      "epoch": 0.7331189710610932,
      "grad_norm": 2.6257119178771973,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.5262,
      "step": 57
    },
    {
      "epoch": 0.7459807073954984,
      "grad_norm": 2.0465402603149414,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.6362,
      "step": 58
    },
    {
      "epoch": 0.7588424437299035,
      "grad_norm": 2.1365973949432373,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.6563,
      "step": 59
    },
    {
      "epoch": 0.7717041800643086,
      "grad_norm": 2.0593950748443604,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.6288,
      "step": 60
    },
    {
      "epoch": 0.7845659163987139,
      "grad_norm": 1.8679747581481934,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.5811,
      "step": 61
    },
    {
      "epoch": 0.797427652733119,
      "grad_norm": 1.6986397504806519,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.5894,
      "step": 62
    },
    {
      "epoch": 0.8102893890675241,
      "grad_norm": 1.858195424079895,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.6084,
      "step": 63
    },
    {
      "epoch": 0.8231511254019293,
      "grad_norm": 1.9211241006851196,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.5519,
      "step": 64
    },
    {
      "epoch": 0.8360128617363344,
      "grad_norm": 2.0085325241088867,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.551,
      "step": 65
    },
    {
      "epoch": 0.8488745980707395,
      "grad_norm": 1.8719388246536255,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.5303,
      "step": 66
    },
    {
      "epoch": 0.8617363344051447,
      "grad_norm": 2.3183629512786865,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.6535,
      "step": 67
    },
    {
      "epoch": 0.8745980707395499,
      "grad_norm": 2.222489356994629,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.5604,
      "step": 68
    },
    {
      "epoch": 0.887459807073955,
      "grad_norm": 2.1058013439178467,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.5256,
      "step": 69
    },
    {
      "epoch": 0.9003215434083601,
      "grad_norm": 2.1135175228118896,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.4931,
      "step": 70
    },
    {
      "epoch": 0.9131832797427653,
      "grad_norm": 2.310624361038208,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.5908,
      "step": 71
    },
    {
      "epoch": 0.9260450160771704,
      "grad_norm": 2.490583896636963,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.541,
      "step": 72
    },
    {
      "epoch": 0.9389067524115756,
      "grad_norm": 2.5393335819244385,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.5692,
      "step": 73
    },
    {
      "epoch": 0.9517684887459807,
      "grad_norm": 2.5098774433135986,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.5421,
      "step": 74
    },
    {
      "epoch": 0.9646302250803859,
      "grad_norm": 2.646827220916748,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5843,
      "step": 75
    },
    {
      "epoch": 0.977491961414791,
      "grad_norm": 3.6071279048919678,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.5131,
      "step": 76
    },
    {
      "epoch": 0.9903536977491961,
      "grad_norm": 2.4894869327545166,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.628,
      "step": 77
    },
    {
      "epoch": 1.0032154340836013,
      "grad_norm": 2.9163973331451416,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.5614,
      "step": 78
    },
    {
      "epoch": 1.0160771704180065,
      "grad_norm": 1.5033864974975586,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.4138,
      "step": 79
    },
    {
      "epoch": 1.0289389067524115,
      "grad_norm": 1.304566502571106,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.3716,
      "step": 80
    },
    {
      "epoch": 1.0418006430868167,
      "grad_norm": 1.380507230758667,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.3569,
      "step": 81
    },
    {
      "epoch": 1.0546623794212218,
      "grad_norm": 1.2010259628295898,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.3096,
      "step": 82
    },
    {
      "epoch": 1.067524115755627,
      "grad_norm": 1.3432456254959106,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.2847,
      "step": 83
    },
    {
      "epoch": 1.0803858520900322,
      "grad_norm": 1.593232274055481,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.3568,
      "step": 84
    },
    {
      "epoch": 1.0932475884244373,
      "grad_norm": 1.7396436929702759,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.3227,
      "step": 85
    },
    {
      "epoch": 1.1061093247588425,
      "grad_norm": 1.4870123863220215,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.2605,
      "step": 86
    },
    {
      "epoch": 1.1189710610932475,
      "grad_norm": 1.8556181192398071,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.3089,
      "step": 87
    },
    {
      "epoch": 1.1318327974276527,
      "grad_norm": 1.6006104946136475,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.2573,
      "step": 88
    },
    {
      "epoch": 1.144694533762058,
      "grad_norm": 1.8679314851760864,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.2764,
      "step": 89
    },
    {
      "epoch": 1.157556270096463,
      "grad_norm": 2.121520519256592,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.31,
      "step": 90
    },
    {
      "epoch": 1.1704180064308682,
      "grad_norm": 1.9375683069229126,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.2847,
      "step": 91
    },
    {
      "epoch": 1.1832797427652733,
      "grad_norm": 2.5801634788513184,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.3366,
      "step": 92
    },
    {
      "epoch": 1.1961414790996785,
      "grad_norm": 2.4023470878601074,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.3209,
      "step": 93
    },
    {
      "epoch": 1.2090032154340835,
      "grad_norm": 1.9816339015960693,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3414,
      "step": 94
    },
    {
      "epoch": 1.2218649517684887,
      "grad_norm": 1.9100102186203003,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.2783,
      "step": 95
    },
    {
      "epoch": 1.234726688102894,
      "grad_norm": 1.908983588218689,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.2937,
      "step": 96
    },
    {
      "epoch": 1.247588424437299,
      "grad_norm": 2.0047850608825684,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.2735,
      "step": 97
    },
    {
      "epoch": 1.2604501607717042,
      "grad_norm": 1.884751319885254,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3963,
      "step": 98
    },
    {
      "epoch": 1.2733118971061093,
      "grad_norm": 2.369093418121338,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.4739,
      "step": 99
    },
    {
      "epoch": 1.2861736334405145,
      "grad_norm": 1.8225935697555542,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.3989,
      "step": 100
    },
    {
      "epoch": 1.2861736334405145,
      "eval_loss": 0.15167492628097534,
      "eval_runtime": 12.4383,
      "eval_samples_per_second": 10.532,
      "eval_steps_per_second": 2.653,
      "step": 100
    },
    {
      "epoch": 1.2990353697749195,
      "grad_norm": 1.3703364133834839,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.3045,
      "step": 101
    },
    {
      "epoch": 1.3118971061093248,
      "grad_norm": 1.5422170162200928,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.401,
      "step": 102
    },
    {
      "epoch": 1.32475884244373,
      "grad_norm": 1.8546526432037354,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.3142,
      "step": 103
    },
    {
      "epoch": 1.337620578778135,
      "grad_norm": 1.4363991022109985,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.3074,
      "step": 104
    },
    {
      "epoch": 1.3504823151125402,
      "grad_norm": 1.5756192207336426,
      "learning_rate": 5e-05,
      "loss": 0.3496,
      "step": 105
    },
    {
      "epoch": 1.3633440514469453,
      "grad_norm": 1.2166670560836792,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.2786,
      "step": 106
    },
    {
      "epoch": 1.3762057877813505,
      "grad_norm": 1.4348070621490479,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.2937,
      "step": 107
    },
    {
      "epoch": 1.3890675241157555,
      "grad_norm": 1.3649669885635376,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.2425,
      "step": 108
    },
    {
      "epoch": 1.4019292604501608,
      "grad_norm": 1.358501672744751,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.2527,
      "step": 109
    },
    {
      "epoch": 1.414790996784566,
      "grad_norm": 1.563353419303894,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.2571,
      "step": 110
    },
    {
      "epoch": 1.427652733118971,
      "grad_norm": 1.7664766311645508,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.3088,
      "step": 111
    },
    {
      "epoch": 1.4405144694533762,
      "grad_norm": 1.8658229112625122,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.3299,
      "step": 112
    },
    {
      "epoch": 1.4533762057877815,
      "grad_norm": 2.0176994800567627,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.3356,
      "step": 113
    },
    {
      "epoch": 1.4662379421221865,
      "grad_norm": 2.3260958194732666,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3097,
      "step": 114
    },
    {
      "epoch": 1.4790996784565915,
      "grad_norm": 2.7366232872009277,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.3387,
      "step": 115
    },
    {
      "epoch": 1.4919614147909968,
      "grad_norm": 2.857238531112671,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.4011,
      "step": 116
    },
    {
      "epoch": 1.504823151125402,
      "grad_norm": 1.866429328918457,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.3519,
      "step": 117
    },
    {
      "epoch": 1.517684887459807,
      "grad_norm": 2.011643648147583,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.3556,
      "step": 118
    },
    {
      "epoch": 1.5305466237942122,
      "grad_norm": 1.8237982988357544,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.328,
      "step": 119
    },
    {
      "epoch": 1.5434083601286175,
      "grad_norm": 1.8693746328353882,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.3467,
      "step": 120
    },
    {
      "epoch": 1.5562700964630225,
      "grad_norm": 1.7072639465332031,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.3681,
      "step": 121
    },
    {
      "epoch": 1.5691318327974275,
      "grad_norm": 1.6630723476409912,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.3607,
      "step": 122
    },
    {
      "epoch": 1.5819935691318328,
      "grad_norm": 1.6173055171966553,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.3427,
      "step": 123
    },
    {
      "epoch": 1.594855305466238,
      "grad_norm": 1.2936534881591797,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.256,
      "step": 124
    },
    {
      "epoch": 1.607717041800643,
      "grad_norm": 1.3732786178588867,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.309,
      "step": 125
    },
    {
      "epoch": 1.6205787781350482,
      "grad_norm": 1.4616271257400513,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.3308,
      "step": 126
    },
    {
      "epoch": 1.6334405144694535,
      "grad_norm": 1.388346791267395,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.3024,
      "step": 127
    },
    {
      "epoch": 1.6463022508038585,
      "grad_norm": 1.5876193046569824,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.3376,
      "step": 128
    },
    {
      "epoch": 1.6591639871382635,
      "grad_norm": 1.3425132036209106,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.3003,
      "step": 129
    },
    {
      "epoch": 1.6720257234726688,
      "grad_norm": 1.372531533241272,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.2788,
      "step": 130
    },
    {
      "epoch": 1.684887459807074,
      "grad_norm": 1.4891109466552734,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.2723,
      "step": 131
    },
    {
      "epoch": 1.697749196141479,
      "grad_norm": 1.8936517238616943,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.3107,
      "step": 132
    },
    {
      "epoch": 1.7106109324758842,
      "grad_norm": 1.7436033487319946,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.3117,
      "step": 133
    },
    {
      "epoch": 1.7234726688102895,
      "grad_norm": 1.913162350654602,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.3405,
      "step": 134
    },
    {
      "epoch": 1.7363344051446945,
      "grad_norm": 2.1614837646484375,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.3254,
      "step": 135
    },
    {
      "epoch": 1.7491961414790995,
      "grad_norm": 1.3987747430801392,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.2954,
      "step": 136
    },
    {
      "epoch": 1.762057877813505,
      "grad_norm": 1.6411230564117432,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.3113,
      "step": 137
    },
    {
      "epoch": 1.77491961414791,
      "grad_norm": 1.4385764598846436,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.3092,
      "step": 138
    },
    {
      "epoch": 1.787781350482315,
      "grad_norm": 1.6263717412948608,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.3417,
      "step": 139
    },
    {
      "epoch": 1.8006430868167203,
      "grad_norm": 1.6047823429107666,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.2946,
      "step": 140
    },
    {
      "epoch": 1.8135048231511255,
      "grad_norm": 1.7206212282180786,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.2867,
      "step": 141
    },
    {
      "epoch": 1.8263665594855305,
      "grad_norm": 1.6301636695861816,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.3079,
      "step": 142
    },
    {
      "epoch": 1.8392282958199357,
      "grad_norm": 1.7266063690185547,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.3515,
      "step": 143
    },
    {
      "epoch": 1.852090032154341,
      "grad_norm": 1.591522455215454,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.3002,
      "step": 144
    },
    {
      "epoch": 1.864951768488746,
      "grad_norm": 1.6167227029800415,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.2827,
      "step": 145
    },
    {
      "epoch": 1.877813504823151,
      "grad_norm": 1.6780692338943481,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.289,
      "step": 146
    },
    {
      "epoch": 1.8906752411575563,
      "grad_norm": 1.7239904403686523,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.3121,
      "step": 147
    },
    {
      "epoch": 1.9035369774919615,
      "grad_norm": 1.2845085859298706,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.2366,
      "step": 148
    },
    {
      "epoch": 1.9163987138263665,
      "grad_norm": 1.6193195581436157,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.3141,
      "step": 149
    },
    {
      "epoch": 1.9292604501607717,
      "grad_norm": 1.7007980346679688,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.2956,
      "step": 150
    },
    {
      "epoch": 1.9292604501607717,
      "eval_loss": 0.1271982491016388,
      "eval_runtime": 12.4446,
      "eval_samples_per_second": 10.527,
      "eval_steps_per_second": 2.652,
      "step": 150
    },
    {
      "epoch": 1.942122186495177,
      "grad_norm": 1.7696309089660645,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.2456,
      "step": 151
    },
    {
      "epoch": 1.954983922829582,
      "grad_norm": 1.8788748979568481,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.2717,
      "step": 152
    },
    {
      "epoch": 1.967845659163987,
      "grad_norm": 1.8966642618179321,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.2979,
      "step": 153
    },
    {
      "epoch": 1.9807073954983923,
      "grad_norm": 2.306748867034912,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.3188,
      "step": 154
    },
    {
      "epoch": 1.9935691318327975,
      "grad_norm": 1.178938627243042,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.2191,
      "step": 155
    },
    {
      "epoch": 2.0064308681672025,
      "grad_norm": 1.580810546875,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.2331,
      "step": 156
    },
    {
      "epoch": 2.0192926045016075,
      "grad_norm": 0.912643551826477,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.1715,
      "step": 157
    },
    {
      "epoch": 2.032154340836013,
      "grad_norm": 0.958441972732544,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.1678,
      "step": 158
    },
    {
      "epoch": 2.045016077170418,
      "grad_norm": 0.934238851070404,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.1705,
      "step": 159
    },
    {
      "epoch": 2.057877813504823,
      "grad_norm": 0.9232479929924011,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.1627,
      "step": 160
    },
    {
      "epoch": 2.0707395498392285,
      "grad_norm": 1.0311024188995361,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.1759,
      "step": 161
    },
    {
      "epoch": 2.0836012861736335,
      "grad_norm": 0.8430033922195435,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.1331,
      "step": 162
    },
    {
      "epoch": 2.0964630225080385,
      "grad_norm": 1.0932472944259644,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.1756,
      "step": 163
    },
    {
      "epoch": 2.1093247588424435,
      "grad_norm": 0.9494350552558899,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.1364,
      "step": 164
    },
    {
      "epoch": 2.122186495176849,
      "grad_norm": 0.9491251111030579,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.1626,
      "step": 165
    },
    {
      "epoch": 2.135048231511254,
      "grad_norm": 0.879464864730835,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.1417,
      "step": 166
    },
    {
      "epoch": 2.147909967845659,
      "grad_norm": 1.0843161344528198,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.1539,
      "step": 167
    },
    {
      "epoch": 2.1607717041800645,
      "grad_norm": 1.0936511754989624,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.1611,
      "step": 168
    },
    {
      "epoch": 2.1736334405144695,
      "grad_norm": 0.9530389904975891,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.1505,
      "step": 169
    },
    {
      "epoch": 2.1864951768488745,
      "grad_norm": 1.0707767009735107,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.1471,
      "step": 170
    },
    {
      "epoch": 2.19935691318328,
      "grad_norm": 1.1714147329330444,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.1413,
      "step": 171
    },
    {
      "epoch": 2.212218649517685,
      "grad_norm": 1.1402021646499634,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.1687,
      "step": 172
    },
    {
      "epoch": 2.22508038585209,
      "grad_norm": 1.1536344289779663,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.14,
      "step": 173
    },
    {
      "epoch": 2.237942122186495,
      "grad_norm": 1.3489817380905151,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.1838,
      "step": 174
    },
    {
      "epoch": 2.2508038585209005,
      "grad_norm": 1.4208256006240845,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.1827,
      "step": 175
    },
    {
      "epoch": 2.2636655948553055,
      "grad_norm": 1.0411112308502197,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.1455,
      "step": 176
    },
    {
      "epoch": 2.2765273311897105,
      "grad_norm": 1.1936169862747192,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.1843,
      "step": 177
    },
    {
      "epoch": 2.289389067524116,
      "grad_norm": 1.1074681282043457,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.1674,
      "step": 178
    },
    {
      "epoch": 2.302250803858521,
      "grad_norm": 1.1440948247909546,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.1676,
      "step": 179
    },
    {
      "epoch": 2.315112540192926,
      "grad_norm": 1.013526439666748,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.1472,
      "step": 180
    },
    {
      "epoch": 2.327974276527331,
      "grad_norm": 0.9656915664672852,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.1295,
      "step": 181
    },
    {
      "epoch": 2.3408360128617365,
      "grad_norm": 1.0586071014404297,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.1265,
      "step": 182
    },
    {
      "epoch": 2.3536977491961415,
      "grad_norm": 1.0153192281723022,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.1337,
      "step": 183
    },
    {
      "epoch": 2.3665594855305465,
      "grad_norm": 1.029646873474121,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.133,
      "step": 184
    },
    {
      "epoch": 2.379421221864952,
      "grad_norm": 1.0104255676269531,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.1267,
      "step": 185
    },
    {
      "epoch": 2.392282958199357,
      "grad_norm": 1.163865327835083,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.1445,
      "step": 186
    },
    {
      "epoch": 2.405144694533762,
      "grad_norm": 1.071718692779541,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.1529,
      "step": 187
    },
    {
      "epoch": 2.418006430868167,
      "grad_norm": 1.1187633275985718,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.1585,
      "step": 188
    },
    {
      "epoch": 2.4308681672025725,
      "grad_norm": 1.1866687536239624,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.1506,
      "step": 189
    },
    {
      "epoch": 2.4437299035369775,
      "grad_norm": 1.1966333389282227,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.1513,
      "step": 190
    },
    {
      "epoch": 2.4565916398713825,
      "grad_norm": 1.1807458400726318,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.1376,
      "step": 191
    },
    {
      "epoch": 2.469453376205788,
      "grad_norm": 1.3857436180114746,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.1529,
      "step": 192
    },
    {
      "epoch": 2.482315112540193,
      "grad_norm": 1.5167961120605469,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.1393,
      "step": 193
    },
    {
      "epoch": 2.495176848874598,
      "grad_norm": 2.958528995513916,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.2454,
      "step": 194
    },
    {
      "epoch": 2.508038585209003,
      "grad_norm": 1.5139999389648438,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.1615,
      "step": 195
    },
    {
      "epoch": 2.5209003215434085,
      "grad_norm": 1.3296087980270386,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.1762,
      "step": 196
    },
    {
      "epoch": 2.5337620578778135,
      "grad_norm": 1.0163224935531616,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.1209,
      "step": 197
    },
    {
      "epoch": 2.5466237942122185,
      "grad_norm": 1.1030423641204834,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.1341,
      "step": 198
    },
    {
      "epoch": 2.559485530546624,
      "grad_norm": 1.1762861013412476,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.1531,
      "step": 199
    },
    {
      "epoch": 2.572347266881029,
      "grad_norm": 1.088255524635315,
      "learning_rate": 0.0,
      "loss": 0.1467,
      "step": 200
    },
    {
      "epoch": 2.572347266881029,
      "eval_loss": 0.13435882329940796,
      "eval_runtime": 12.4529,
      "eval_samples_per_second": 10.52,
      "eval_steps_per_second": 2.65,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 2.894193473177518e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}