{
  "best_metric": 2.2457525730133057,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.012011470954761797,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.005735477380899e-05,
      "grad_norm": 0.7277234792709351,
      "learning_rate": 1e-05,
      "loss": 1.3234,
      "step": 1
    },
    {
      "epoch": 6.005735477380899e-05,
      "eval_loss": 3.1906468868255615,
      "eval_runtime": 2134.556,
      "eval_samples_per_second": 13.138,
      "eval_steps_per_second": 3.285,
      "step": 1
    },
    {
      "epoch": 0.00012011470954761798,
      "grad_norm": 0.8249174952507019,
      "learning_rate": 2e-05,
      "loss": 1.4146,
      "step": 2
    },
    {
      "epoch": 0.00018017206432142696,
      "grad_norm": 0.8658198118209839,
      "learning_rate": 3e-05,
      "loss": 1.5297,
      "step": 3
    },
    {
      "epoch": 0.00024022941909523595,
      "grad_norm": 0.9398988485336304,
      "learning_rate": 4e-05,
      "loss": 1.6596,
      "step": 4
    },
    {
      "epoch": 0.00030028677386904494,
      "grad_norm": 0.823463499546051,
      "learning_rate": 5e-05,
      "loss": 1.6762,
      "step": 5
    },
    {
      "epoch": 0.00036034412864285393,
      "grad_norm": 0.7399727702140808,
      "learning_rate": 6e-05,
      "loss": 1.6846,
      "step": 6
    },
    {
      "epoch": 0.0004204014834166629,
      "grad_norm": 0.9010226726531982,
      "learning_rate": 7e-05,
      "loss": 1.644,
      "step": 7
    },
    {
      "epoch": 0.0004804588381904719,
      "grad_norm": 0.9326359629631042,
      "learning_rate": 8e-05,
      "loss": 1.5877,
      "step": 8
    },
    {
      "epoch": 0.0005405161929642808,
      "grad_norm": 0.977419912815094,
      "learning_rate": 9e-05,
      "loss": 1.7177,
      "step": 9
    },
    {
      "epoch": 0.0006005735477380899,
      "grad_norm": 0.6996713876724243,
      "learning_rate": 0.0001,
      "loss": 1.4415,
      "step": 10
    },
    {
      "epoch": 0.0006606309025118988,
      "grad_norm": 0.7411949634552002,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5325,
      "step": 11
    },
    {
      "epoch": 0.0007206882572857079,
      "grad_norm": 0.8528963327407837,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.4692,
      "step": 12
    },
    {
      "epoch": 0.0007807456120595168,
      "grad_norm": 0.8318995237350464,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.6418,
      "step": 13
    },
    {
      "epoch": 0.0008408029668333258,
      "grad_norm": 0.7479487061500549,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.7038,
      "step": 14
    },
    {
      "epoch": 0.0009008603216071348,
      "grad_norm": 0.7285621166229248,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4805,
      "step": 15
    },
    {
      "epoch": 0.0009609176763809438,
      "grad_norm": 0.7259240746498108,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.6445,
      "step": 16
    },
    {
      "epoch": 0.0010209750311547527,
      "grad_norm": 0.7934219837188721,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.367,
      "step": 17
    },
    {
      "epoch": 0.0010810323859285617,
      "grad_norm": 0.8407491445541382,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.4958,
      "step": 18
    },
    {
      "epoch": 0.0011410897407023708,
      "grad_norm": 0.9245749115943909,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6083,
      "step": 19
    },
    {
      "epoch": 0.0012011470954761798,
      "grad_norm": 0.8109903931617737,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.5333,
      "step": 20
    },
    {
      "epoch": 0.0012612044502499887,
      "grad_norm": 1.2378709316253662,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6405,
      "step": 21
    },
    {
      "epoch": 0.0013212618050237976,
      "grad_norm": 1.558487057685852,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.5916,
      "step": 22
    },
    {
      "epoch": 0.0013813191597976068,
      "grad_norm": 0.9527750015258789,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.593,
      "step": 23
    },
    {
      "epoch": 0.0014413765145714157,
      "grad_norm": 1.060105562210083,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.7384,
      "step": 24
    },
    {
      "epoch": 0.0015014338693452246,
      "grad_norm": 1.1204156875610352,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.9215,
      "step": 25
    },
    {
      "epoch": 0.0015614912241190336,
      "grad_norm": 1.1645809412002563,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.8122,
      "step": 26
    },
    {
      "epoch": 0.0016215485788928427,
      "grad_norm": 1.0866835117340088,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.8427,
      "step": 27
    },
    {
      "epoch": 0.0016816059336666517,
      "grad_norm": 1.4639962911605835,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.7834,
      "step": 28
    },
    {
      "epoch": 0.0017416632884404606,
      "grad_norm": 1.2034151554107666,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.7954,
      "step": 29
    },
    {
      "epoch": 0.0018017206432142695,
      "grad_norm": 1.3376408815383911,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.8065,
      "step": 30
    },
    {
      "epoch": 0.0018617779979880787,
      "grad_norm": 1.7506548166275024,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.0368,
      "step": 31
    },
    {
      "epoch": 0.0019218353527618876,
      "grad_norm": 1.909898042678833,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.766,
      "step": 32
    },
    {
      "epoch": 0.0019818927075356966,
      "grad_norm": 1.9195573329925537,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.0777,
      "step": 33
    },
    {
      "epoch": 0.0020419500623095055,
      "grad_norm": 1.7940828800201416,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.2604,
      "step": 34
    },
    {
      "epoch": 0.0021020074170833144,
      "grad_norm": 2.1280953884124756,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.3557,
      "step": 35
    },
    {
      "epoch": 0.0021620647718571234,
      "grad_norm": 2.3466970920562744,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.6686,
      "step": 36
    },
    {
      "epoch": 0.0022221221266309327,
      "grad_norm": 2.250682830810547,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.7733,
      "step": 37
    },
    {
      "epoch": 0.0022821794814047417,
      "grad_norm": 2.6434786319732666,
      "learning_rate": 9.473646649103818e-05,
      "loss": 3.0848,
      "step": 38
    },
    {
      "epoch": 0.0023422368361785506,
      "grad_norm": 4.73193359375,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.0783,
      "step": 39
    },
    {
      "epoch": 0.0024022941909523595,
      "grad_norm": 3.889577865600586,
      "learning_rate": 9.397368756032445e-05,
      "loss": 3.4497,
      "step": 40
    },
    {
      "epoch": 0.0024623515457261685,
      "grad_norm": 4.585144996643066,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.6834,
      "step": 41
    },
    {
      "epoch": 0.0025224089004999774,
      "grad_norm": 5.334516525268555,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.5614,
      "step": 42
    },
    {
      "epoch": 0.0025824662552737863,
      "grad_norm": 6.08420467376709,
      "learning_rate": 9.273963562927695e-05,
      "loss": 4.0365,
      "step": 43
    },
    {
      "epoch": 0.0026425236100475953,
      "grad_norm": 6.18115234375,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.5424,
      "step": 44
    },
    {
      "epoch": 0.0027025809648214046,
      "grad_norm": 9.245948791503906,
      "learning_rate": 9.185832391312644e-05,
      "loss": 4.1178,
      "step": 45
    },
    {
      "epoch": 0.0027626383195952136,
      "grad_norm": 5.46101713180542,
      "learning_rate": 9.140044155740101e-05,
      "loss": 3.632,
      "step": 46
    },
    {
      "epoch": 0.0028226956743690225,
      "grad_norm": 10.615842819213867,
      "learning_rate": 9.093124073433463e-05,
      "loss": 4.2407,
      "step": 47
    },
    {
      "epoch": 0.0028827530291428314,
      "grad_norm": 22.699947357177734,
      "learning_rate": 9.045084971874738e-05,
      "loss": 4.6777,
      "step": 48
    },
    {
      "epoch": 0.0029428103839166404,
      "grad_norm": 25.06355094909668,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.5389,
      "step": 49
    },
    {
      "epoch": 0.0030028677386904493,
      "grad_norm": 17.44358253479004,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.6876,
      "step": 50
    },
    {
      "epoch": 0.0030028677386904493,
      "eval_loss": 2.82340669631958,
      "eval_runtime": 2140.029,
      "eval_samples_per_second": 13.104,
      "eval_steps_per_second": 3.276,
      "step": 50
    },
    {
      "epoch": 0.0030629250934642582,
      "grad_norm": 4.516435623168945,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.8638,
      "step": 51
    },
    {
      "epoch": 0.003122982448238067,
      "grad_norm": 2.2428054809570312,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.7122,
      "step": 52
    },
    {
      "epoch": 0.0031830398030118765,
      "grad_norm": 1.1682690382003784,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5114,
      "step": 53
    },
    {
      "epoch": 0.0032430971577856855,
      "grad_norm": 1.2726598978042603,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.4385,
      "step": 54
    },
    {
      "epoch": 0.0033031545125594944,
      "grad_norm": 0.6764863729476929,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.3759,
      "step": 55
    },
    {
      "epoch": 0.0033632118673333033,
      "grad_norm": 0.6642225384712219,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4671,
      "step": 56
    },
    {
      "epoch": 0.0034232692221071123,
      "grad_norm": 0.7281270027160645,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.5034,
      "step": 57
    },
    {
      "epoch": 0.003483326576880921,
      "grad_norm": 0.5902082920074463,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4262,
      "step": 58
    },
    {
      "epoch": 0.00354338393165473,
      "grad_norm": 0.6462448239326477,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.5712,
      "step": 59
    },
    {
      "epoch": 0.003603441286428539,
      "grad_norm": 0.6664770245552063,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.4912,
      "step": 60
    },
    {
      "epoch": 0.0036634986412023484,
      "grad_norm": 0.7605299949645996,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.6861,
      "step": 61
    },
    {
      "epoch": 0.0037235559959761574,
      "grad_norm": 0.6704346537590027,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.538,
      "step": 62
    },
    {
      "epoch": 0.0037836133507499663,
      "grad_norm": 0.6854274272918701,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4547,
      "step": 63
    },
    {
      "epoch": 0.0038436707055237752,
      "grad_norm": 0.6620988845825195,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.475,
      "step": 64
    },
    {
      "epoch": 0.003903728060297584,
      "grad_norm": 0.7582455277442932,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.7002,
      "step": 65
    },
    {
      "epoch": 0.003963785415071393,
      "grad_norm": 0.6943215131759644,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.5913,
      "step": 66
    },
    {
      "epoch": 0.004023842769845202,
      "grad_norm": 1.0203649997711182,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5227,
      "step": 67
    },
    {
      "epoch": 0.004083900124619011,
      "grad_norm": 0.8940017223358154,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.6607,
      "step": 68
    },
    {
      "epoch": 0.00414395747939282,
      "grad_norm": 0.9065792560577393,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.6817,
      "step": 69
    },
    {
      "epoch": 0.004204014834166629,
      "grad_norm": 0.8339809775352478,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.5294,
      "step": 70
    },
    {
      "epoch": 0.004264072188940438,
      "grad_norm": 1.0335615873336792,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.8012,
      "step": 71
    },
    {
      "epoch": 0.004324129543714247,
      "grad_norm": 0.8233577609062195,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6343,
      "step": 72
    },
    {
      "epoch": 0.0043841868984880565,
      "grad_norm": 0.8685855269432068,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.8167,
      "step": 73
    },
    {
      "epoch": 0.0044442442532618654,
      "grad_norm": 0.9024196863174438,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.7911,
      "step": 74
    },
    {
      "epoch": 0.004504301608035674,
      "grad_norm": 1.0157138109207153,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.6895,
      "step": 75
    },
    {
      "epoch": 0.004564358962809483,
      "grad_norm": 0.9315282702445984,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.6535,
      "step": 76
    },
    {
      "epoch": 0.004624416317583292,
      "grad_norm": 1.0565733909606934,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.792,
      "step": 77
    },
    {
      "epoch": 0.004684473672357101,
      "grad_norm": 1.0977641344070435,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.7772,
      "step": 78
    },
    {
      "epoch": 0.00474453102713091,
      "grad_norm": 1.0870914459228516,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.8946,
      "step": 79
    },
    {
      "epoch": 0.004804588381904719,
      "grad_norm": 1.1871083974838257,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.7883,
      "step": 80
    },
    {
      "epoch": 0.004864645736678528,
      "grad_norm": 1.1695376634597778,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.9889,
      "step": 81
    },
    {
      "epoch": 0.004924703091452337,
      "grad_norm": 1.3979439735412598,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.9121,
      "step": 82
    },
    {
      "epoch": 0.004984760446226146,
      "grad_norm": 1.7781084775924683,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.3537,
      "step": 83
    },
    {
      "epoch": 0.005044817800999955,
      "grad_norm": 2.4839773178100586,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.2461,
      "step": 84
    },
    {
      "epoch": 0.005104875155773764,
      "grad_norm": 2.7168943881988525,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.4214,
      "step": 85
    },
    {
      "epoch": 0.005164932510547573,
      "grad_norm": 2.0098538398742676,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.5479,
      "step": 86
    },
    {
      "epoch": 0.005224989865321382,
      "grad_norm": 2.1378700733184814,
      "learning_rate": 6.466250186922325e-05,
      "loss": 3.0993,
      "step": 87
    },
    {
      "epoch": 0.0052850472200951905,
      "grad_norm": 2.6309289932250977,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.916,
      "step": 88
    },
    {
      "epoch": 0.005345104574869,
      "grad_norm": 3.0686724185943604,
      "learning_rate": 6.307399704769099e-05,
      "loss": 3.217,
      "step": 89
    },
    {
      "epoch": 0.005405161929642809,
      "grad_norm": 3.5666792392730713,
      "learning_rate": 6.227427435703997e-05,
      "loss": 3.6197,
      "step": 90
    },
    {
      "epoch": 0.005465219284416618,
      "grad_norm": 4.670599937438965,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.8187,
      "step": 91
    },
    {
      "epoch": 0.005525276639190427,
      "grad_norm": 5.340340614318848,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.9059,
      "step": 92
    },
    {
      "epoch": 0.005585333993964236,
      "grad_norm": 5.360260963439941,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.7257,
      "step": 93
    },
    {
      "epoch": 0.005645391348738045,
      "grad_norm": 4.779595375061035,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.976,
      "step": 94
    },
    {
      "epoch": 0.005705448703511854,
      "grad_norm": 4.226884365081787,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 3.8745,
      "step": 95
    },
    {
      "epoch": 0.005765506058285663,
      "grad_norm": 6.577067852020264,
      "learning_rate": 5.74131823855921e-05,
      "loss": 3.9981,
      "step": 96
    },
    {
      "epoch": 0.005825563413059472,
      "grad_norm": 6.186820030212402,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 4.2383,
      "step": 97
    },
    {
      "epoch": 0.005885620767833281,
      "grad_norm": 9.95272445678711,
      "learning_rate": 5.577423184847932e-05,
      "loss": 4.388,
      "step": 98
    },
    {
      "epoch": 0.00594567812260709,
      "grad_norm": 11.007781982421875,
      "learning_rate": 5.495227651252315e-05,
      "loss": 4.35,
      "step": 99
    },
    {
      "epoch": 0.006005735477380899,
      "grad_norm": 13.059420585632324,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 5.1192,
      "step": 100
    },
    {
      "epoch": 0.006005735477380899,
      "eval_loss": 2.867065191268921,
      "eval_runtime": 2140.465,
      "eval_samples_per_second": 13.102,
      "eval_steps_per_second": 3.275,
      "step": 100
    },
    {
      "epoch": 0.0060657928321547075,
      "grad_norm": 5.179087162017822,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.9935,
      "step": 101
    },
    {
      "epoch": 0.0061258501869285165,
      "grad_norm": 4.025508880615234,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.8939,
      "step": 102
    },
    {
      "epoch": 0.006185907541702325,
      "grad_norm": 2.6050190925598145,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.7391,
      "step": 103
    },
    {
      "epoch": 0.006245964896476134,
      "grad_norm": 1.267030119895935,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.5506,
      "step": 104
    },
    {
      "epoch": 0.006306022251249943,
      "grad_norm": 1.056959867477417,
      "learning_rate": 5e-05,
      "loss": 1.725,
      "step": 105
    },
    {
      "epoch": 0.006366079606023753,
      "grad_norm": 0.6427236199378967,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.5067,
      "step": 106
    },
    {
      "epoch": 0.006426136960797562,
      "grad_norm": 0.7136508226394653,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.5774,
      "step": 107
    },
    {
      "epoch": 0.006486194315571371,
      "grad_norm": 0.6463544368743896,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.4658,
      "step": 108
    },
    {
      "epoch": 0.00654625167034518,
      "grad_norm": 0.6630975604057312,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4437,
      "step": 109
    },
    {
      "epoch": 0.006606309025118989,
      "grad_norm": 0.606503427028656,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5019,
      "step": 110
    },
    {
      "epoch": 0.006666366379892798,
      "grad_norm": 0.7609089016914368,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.7205,
      "step": 111
    },
    {
      "epoch": 0.006726423734666607,
      "grad_norm": 0.7030594348907471,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.4174,
      "step": 112
    },
    {
      "epoch": 0.006786481089440416,
      "grad_norm": 0.6835500597953796,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.4883,
      "step": 113
    },
    {
      "epoch": 0.0068465384442142245,
      "grad_norm": 0.822965145111084,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.6526,
      "step": 114
    },
    {
      "epoch": 0.0069065957989880335,
      "grad_norm": 0.736895740032196,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.5497,
      "step": 115
    },
    {
      "epoch": 0.006966653153761842,
      "grad_norm": 0.7960342764854431,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.5518,
      "step": 116
    },
    {
      "epoch": 0.007026710508535651,
      "grad_norm": 0.7455801367759705,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.4373,
      "step": 117
    },
    {
      "epoch": 0.00708676786330946,
      "grad_norm": 0.74431973695755,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.6076,
      "step": 118
    },
    {
      "epoch": 0.007146825218083269,
      "grad_norm": 0.8275505304336548,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.7249,
      "step": 119
    },
    {
      "epoch": 0.007206882572857078,
      "grad_norm": 0.9723110795021057,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.6594,
      "step": 120
    },
    {
      "epoch": 0.007266939927630887,
      "grad_norm": 0.7745776176452637,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.5868,
      "step": 121
    },
    {
      "epoch": 0.007326997282404697,
      "grad_norm": 1.0984992980957031,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.7533,
      "step": 122
    },
    {
      "epoch": 0.007387054637178506,
      "grad_norm": 0.8350873589515686,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.5889,
      "step": 123
    },
    {
      "epoch": 0.007447111991952315,
      "grad_norm": 0.8925673365592957,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.8666,
      "step": 124
    },
    {
      "epoch": 0.007507169346726124,
      "grad_norm": 0.9474840760231018,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.8208,
      "step": 125
    },
    {
      "epoch": 0.007567226701499933,
      "grad_norm": 0.9493285417556763,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.755,
      "step": 126
    },
    {
      "epoch": 0.0076272840562737415,
      "grad_norm": 1.3120079040527344,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.6636,
      "step": 127
    },
    {
      "epoch": 0.0076873414110475505,
      "grad_norm": 0.9975591897964478,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.8085,
      "step": 128
    },
    {
      "epoch": 0.007747398765821359,
      "grad_norm": 1.1535462141036987,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.8213,
      "step": 129
    },
    {
      "epoch": 0.007807456120595168,
      "grad_norm": 1.311443567276001,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.8648,
      "step": 130
    },
    {
      "epoch": 0.007867513475368978,
      "grad_norm": 1.0834039449691772,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.895,
      "step": 131
    },
    {
      "epoch": 0.007927570830142786,
      "grad_norm": 1.0714950561523438,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.0328,
      "step": 132
    },
    {
      "epoch": 0.007987628184916596,
      "grad_norm": 1.6697895526885986,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.4142,
      "step": 133
    },
    {
      "epoch": 0.008047685539690404,
      "grad_norm": 1.4947165250778198,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.2941,
      "step": 134
    },
    {
      "epoch": 0.008107742894464214,
      "grad_norm": 1.6730867624282837,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.6055,
      "step": 135
    },
    {
      "epoch": 0.008167800249238022,
      "grad_norm": 2.1663410663604736,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.9374,
      "step": 136
    },
    {
      "epoch": 0.008227857604011832,
      "grad_norm": 2.37280011177063,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.8922,
      "step": 137
    },
    {
      "epoch": 0.00828791495878564,
      "grad_norm": 2.307690143585205,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.1167,
      "step": 138
    },
    {
      "epoch": 0.00834797231355945,
      "grad_norm": 2.4942281246185303,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 3.4561,
      "step": 139
    },
    {
      "epoch": 0.008408029668333258,
      "grad_norm": 3.406865119934082,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 3.7567,
      "step": 140
    },
    {
      "epoch": 0.008468087023107067,
      "grad_norm": 4.235611915588379,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.9085,
      "step": 141
    },
    {
      "epoch": 0.008528144377880876,
      "grad_norm": 3.753329038619995,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.9523,
      "step": 142
    },
    {
      "epoch": 0.008588201732654685,
      "grad_norm": 4.188745975494385,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.0361,
      "step": 143
    },
    {
      "epoch": 0.008648259087428493,
      "grad_norm": 4.07323694229126,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 4.0014,
      "step": 144
    },
    {
      "epoch": 0.008708316442202303,
      "grad_norm": 4.6655683517456055,
      "learning_rate": 1.928936436551661e-05,
      "loss": 3.9193,
      "step": 145
    },
    {
      "epoch": 0.008768373796976113,
      "grad_norm": 5.27960205078125,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 3.9531,
      "step": 146
    },
    {
      "epoch": 0.008828431151749921,
      "grad_norm": 6.609133243560791,
      "learning_rate": 1.800157297483417e-05,
      "loss": 4.1015,
      "step": 147
    },
    {
      "epoch": 0.008888488506523731,
      "grad_norm": 9.882762908935547,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 4.6635,
      "step": 148
    },
    {
      "epoch": 0.008948545861297539,
      "grad_norm": 7.035732269287109,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 3.9155,
      "step": 149
    },
    {
      "epoch": 0.009008603216071349,
      "grad_norm": 16.392091751098633,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.8104,
      "step": 150
    },
    {
      "epoch": 0.009008603216071349,
      "eval_loss": 2.292562961578369,
      "eval_runtime": 2141.0151,
      "eval_samples_per_second": 13.098,
      "eval_steps_per_second": 3.275,
      "step": 150
    },
    {
      "epoch": 0.009068660570845157,
      "grad_norm": 0.48265546560287476,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.26,
      "step": 151
    },
    {
      "epoch": 0.009128717925618967,
      "grad_norm": 0.5253238677978516,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.2976,
      "step": 152
    },
    {
      "epoch": 0.009188775280392775,
      "grad_norm": 0.5875234603881836,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3534,
      "step": 153
    },
    {
      "epoch": 0.009248832635166584,
      "grad_norm": 0.5230312943458557,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.4221,
      "step": 154
    },
    {
      "epoch": 0.009308889989940393,
      "grad_norm": 0.5650403499603271,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4197,
      "step": 155
    },
    {
      "epoch": 0.009368947344714202,
      "grad_norm": 0.6009732484817505,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.5457,
      "step": 156
    },
    {
      "epoch": 0.00942900469948801,
      "grad_norm": 0.6655670404434204,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.3704,
      "step": 157
    },
    {
      "epoch": 0.00948906205426182,
      "grad_norm": 0.6313867568969727,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.6597,
      "step": 158
    },
    {
      "epoch": 0.009549119409035628,
      "grad_norm": 0.6830577850341797,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.5639,
      "step": 159
    },
    {
      "epoch": 0.009609176763809438,
      "grad_norm": 0.8179656267166138,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.3819,
      "step": 160
    },
    {
      "epoch": 0.009669234118583246,
      "grad_norm": 0.6520982980728149,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.5832,
      "step": 161
    },
    {
      "epoch": 0.009729291473357056,
      "grad_norm": 0.6750402450561523,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4861,
      "step": 162
    },
    {
      "epoch": 0.009789348828130866,
      "grad_norm": 0.7585417032241821,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.5695,
      "step": 163
    },
    {
      "epoch": 0.009849406182904674,
      "grad_norm": 0.6254276633262634,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.508,
      "step": 164
    },
    {
      "epoch": 0.009909463537678484,
      "grad_norm": 0.7563104629516602,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.6226,
      "step": 165
    },
    {
      "epoch": 0.009969520892452292,
      "grad_norm": 0.7516692280769348,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.615,
      "step": 166
    },
    {
      "epoch": 0.010029578247226102,
      "grad_norm": 0.6805993914604187,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.4733,
      "step": 167
    },
    {
      "epoch": 0.01008963560199991,
      "grad_norm": 0.7976547479629517,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.5344,
      "step": 168
    },
    {
      "epoch": 0.01014969295677372,
      "grad_norm": 0.8522383570671082,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.5984,
      "step": 169
    },
    {
      "epoch": 0.010209750311547527,
      "grad_norm": 0.8134978413581848,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.5362,
      "step": 170
    },
    {
      "epoch": 0.010269807666321337,
      "grad_norm": 0.8034011125564575,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.6478,
      "step": 171
    },
    {
      "epoch": 0.010329865021095145,
      "grad_norm": 0.9884563088417053,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.7466,
      "step": 172
    },
    {
      "epoch": 0.010389922375868955,
      "grad_norm": 0.7845810055732727,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.6212,
      "step": 173
    },
    {
      "epoch": 0.010449979730642763,
      "grad_norm": 0.7723093628883362,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.6498,
      "step": 174
    },
    {
      "epoch": 0.010510037085416573,
      "grad_norm": 1.0075610876083374,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.6671,
      "step": 175
    },
    {
      "epoch": 0.010570094440190381,
      "grad_norm": 0.9355666637420654,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.7672,
      "step": 176
    },
    {
      "epoch": 0.01063015179496419,
      "grad_norm": 0.9750188589096069,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.7507,
      "step": 177
    },
    {
      "epoch": 0.010690209149738,
      "grad_norm": 0.9941683411598206,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.7685,
      "step": 178
    },
    {
      "epoch": 0.010750266504511809,
      "grad_norm": 1.2474644184112549,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.0103,
      "step": 179
    },
    {
      "epoch": 0.010810323859285619,
      "grad_norm": 1.0567797422409058,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.8378,
      "step": 180
    },
    {
      "epoch": 0.010870381214059427,
      "grad_norm": 1.2146373987197876,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.9224,
      "step": 181
    },
    {
      "epoch": 0.010930438568833236,
      "grad_norm": 1.2096244096755981,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.088,
      "step": 182
    },
    {
      "epoch": 0.010990495923607044,
      "grad_norm": 1.2452001571655273,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.0728,
      "step": 183
    },
    {
      "epoch": 0.011050553278380854,
      "grad_norm": 1.471971869468689,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.9715,
      "step": 184
    },
    {
      "epoch": 0.011110610633154662,
      "grad_norm": 1.5556915998458862,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.577,
      "step": 185
    },
    {
      "epoch": 0.011170667987928472,
      "grad_norm": 1.8677924871444702,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.7585,
      "step": 186
    },
    {
      "epoch": 0.01123072534270228,
      "grad_norm": 1.878254771232605,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.8857,
      "step": 187
    },
    {
      "epoch": 0.01129078269747609,
      "grad_norm": 2.483644962310791,
      "learning_rate": 9.810017062595322e-07,
      "loss": 3.1029,
      "step": 188
    },
    {
      "epoch": 0.011350840052249898,
      "grad_norm": 2.463496685028076,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.9694,
      "step": 189
    },
    {
      "epoch": 0.011410897407023708,
      "grad_norm": 2.559252977371216,
      "learning_rate": 6.819348298638839e-07,
      "loss": 3.5838,
      "step": 190
    },
    {
      "epoch": 0.011470954761797516,
      "grad_norm": 3.3962340354919434,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.5984,
      "step": 191
    },
    {
      "epoch": 0.011531012116571326,
      "grad_norm": 3.19893217086792,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.8059,
      "step": 192
    },
    {
      "epoch": 0.011591069471345134,
      "grad_norm": 4.175753116607666,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 3.9952,
      "step": 193
    },
    {
      "epoch": 0.011651126826118944,
      "grad_norm": 4.384350299835205,
      "learning_rate": 2.458548727494292e-07,
      "loss": 3.7542,
      "step": 194
    },
    {
      "epoch": 0.011711184180892753,
      "grad_norm": 4.514298439025879,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 3.8213,
      "step": 195
    },
    {
      "epoch": 0.011771241535666561,
      "grad_norm": 4.254011154174805,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 3.9888,
      "step": 196
    },
    {
      "epoch": 0.011831298890440371,
      "grad_norm": 6.089274883270264,
      "learning_rate": 6.150154258476315e-08,
      "loss": 3.9635,
      "step": 197
    },
    {
      "epoch": 0.01189135624521418,
      "grad_norm": 5.965423107147217,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 4.2933,
      "step": 198
    },
    {
      "epoch": 0.011951413599987989,
      "grad_norm": 5.93259859085083,
      "learning_rate": 6.834750376549792e-09,
      "loss": 4.6476,
      "step": 199
    },
    {
      "epoch": 0.012011470954761797,
      "grad_norm": 10.676911354064941,
      "learning_rate": 0.0,
      "loss": 4.8538,
      "step": 200
    },
    {
      "epoch": 0.012011470954761797,
      "eval_loss": 2.2457525730133057,
      "eval_runtime": 2143.2019,
      "eval_samples_per_second": 13.085,
      "eval_steps_per_second": 3.271,
      "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": 3.0321122932791706e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}