{
  "best_metric": 3.217210531234741,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3198720511795282,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001599360255897641,
      "grad_norm": 7.648873329162598,
      "learning_rate": 1e-05,
      "loss": 6.1486,
      "step": 1
    },
    {
      "epoch": 0.001599360255897641,
      "eval_loss": 7.470527648925781,
      "eval_runtime": 30.6658,
      "eval_samples_per_second": 34.338,
      "eval_steps_per_second": 8.609,
      "step": 1
    },
    {
      "epoch": 0.003198720511795282,
      "grad_norm": 7.270749092102051,
      "learning_rate": 2e-05,
      "loss": 6.436,
      "step": 2
    },
    {
      "epoch": 0.004798080767692923,
      "grad_norm": 6.999588489532471,
      "learning_rate": 3e-05,
      "loss": 6.2846,
      "step": 3
    },
    {
      "epoch": 0.006397441023590564,
      "grad_norm": 7.06572961807251,
      "learning_rate": 4e-05,
      "loss": 6.2746,
      "step": 4
    },
    {
      "epoch": 0.007996801279488205,
      "grad_norm": 5.899146556854248,
      "learning_rate": 5e-05,
      "loss": 6.0562,
      "step": 5
    },
    {
      "epoch": 0.009596161535385846,
      "grad_norm": 6.20655632019043,
      "learning_rate": 6e-05,
      "loss": 5.8245,
      "step": 6
    },
    {
      "epoch": 0.011195521791283487,
      "grad_norm": 5.553204536437988,
      "learning_rate": 7e-05,
      "loss": 5.5328,
      "step": 7
    },
    {
      "epoch": 0.012794882047181128,
      "grad_norm": 4.747158527374268,
      "learning_rate": 8e-05,
      "loss": 5.3374,
      "step": 8
    },
    {
      "epoch": 0.014394242303078768,
      "grad_norm": 5.379507064819336,
      "learning_rate": 9e-05,
      "loss": 4.8272,
      "step": 9
    },
    {
      "epoch": 0.01599360255897641,
      "grad_norm": 4.724349498748779,
      "learning_rate": 0.0001,
      "loss": 4.7132,
      "step": 10
    },
    {
      "epoch": 0.01759296281487405,
      "grad_norm": 4.929715156555176,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.3929,
      "step": 11
    },
    {
      "epoch": 0.01919232307077169,
      "grad_norm": 5.725976467132568,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.3365,
      "step": 12
    },
    {
      "epoch": 0.020791683326669332,
      "grad_norm": 8.988175392150879,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.1484,
      "step": 13
    },
    {
      "epoch": 0.022391043582566973,
      "grad_norm": 6.153397560119629,
      "learning_rate": 9.989068136093873e-05,
      "loss": 4.2506,
      "step": 14
    },
    {
      "epoch": 0.023990403838464614,
      "grad_norm": 4.292699337005615,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.1935,
      "step": 15
    },
    {
      "epoch": 0.025589764094362255,
      "grad_norm": 4.61167049407959,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.1873,
      "step": 16
    },
    {
      "epoch": 0.027189124350259896,
      "grad_norm": 4.89919376373291,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.9584,
      "step": 17
    },
    {
      "epoch": 0.028788484606157537,
      "grad_norm": 4.3355712890625,
      "learning_rate": 9.956320346634876e-05,
      "loss": 4.2508,
      "step": 18
    },
    {
      "epoch": 0.030387844862055178,
      "grad_norm": 4.595262050628662,
      "learning_rate": 9.944739353007344e-05,
      "loss": 4.3774,
      "step": 19
    },
    {
      "epoch": 0.03198720511795282,
      "grad_norm": 3.7050914764404297,
      "learning_rate": 9.931806517013612e-05,
      "loss": 3.914,
      "step": 20
    },
    {
      "epoch": 0.03358656537385046,
      "grad_norm": 3.3008880615234375,
      "learning_rate": 9.917525374361912e-05,
      "loss": 3.8408,
      "step": 21
    },
    {
      "epoch": 0.0351859256297481,
      "grad_norm": 3.080190658569336,
      "learning_rate": 9.901899829374047e-05,
      "loss": 3.6535,
      "step": 22
    },
    {
      "epoch": 0.036785285885645745,
      "grad_norm": 3.556636095046997,
      "learning_rate": 9.884934153917997e-05,
      "loss": 3.5662,
      "step": 23
    },
    {
      "epoch": 0.03838464614154338,
      "grad_norm": 3.1297287940979004,
      "learning_rate": 9.86663298624003e-05,
      "loss": 3.5994,
      "step": 24
    },
    {
      "epoch": 0.03998400639744103,
      "grad_norm": 3.9206912517547607,
      "learning_rate": 9.847001329696653e-05,
      "loss": 3.9169,
      "step": 25
    },
    {
      "epoch": 0.041583366653338664,
      "grad_norm": 3.640852212905884,
      "learning_rate": 9.826044551386744e-05,
      "loss": 3.838,
      "step": 26
    },
    {
      "epoch": 0.04318272690923631,
      "grad_norm": 3.509394645690918,
      "learning_rate": 9.803768380684242e-05,
      "loss": 3.6196,
      "step": 27
    },
    {
      "epoch": 0.044782087165133946,
      "grad_norm": 4.766598701477051,
      "learning_rate": 9.780178907671789e-05,
      "loss": 3.8965,
      "step": 28
    },
    {
      "epoch": 0.04638144742103159,
      "grad_norm": 3.7237844467163086,
      "learning_rate": 9.755282581475769e-05,
      "loss": 3.9251,
      "step": 29
    },
    {
      "epoch": 0.04798080767692923,
      "grad_norm": 3.9853508472442627,
      "learning_rate": 9.729086208503174e-05,
      "loss": 3.9083,
      "step": 30
    },
    {
      "epoch": 0.04958016793282687,
      "grad_norm": 4.01790189743042,
      "learning_rate": 9.701596950580806e-05,
      "loss": 3.7745,
      "step": 31
    },
    {
      "epoch": 0.05117952818872451,
      "grad_norm": 3.656719923019409,
      "learning_rate": 9.672822322997305e-05,
      "loss": 3.7139,
      "step": 32
    },
    {
      "epoch": 0.052778888444622155,
      "grad_norm": 3.675185203552246,
      "learning_rate": 9.642770192448536e-05,
      "loss": 3.5013,
      "step": 33
    },
    {
      "epoch": 0.05437824870051979,
      "grad_norm": 3.8905913829803467,
      "learning_rate": 9.611448774886924e-05,
      "loss": 3.6408,
      "step": 34
    },
    {
      "epoch": 0.055977608956417436,
      "grad_norm": 3.9639244079589844,
      "learning_rate": 9.578866633275288e-05,
      "loss": 3.5655,
      "step": 35
    },
    {
      "epoch": 0.057576969212315074,
      "grad_norm": 3.7178659439086914,
      "learning_rate": 9.545032675245813e-05,
      "loss": 3.6803,
      "step": 36
    },
    {
      "epoch": 0.05917632946821272,
      "grad_norm": 3.434163808822632,
      "learning_rate": 9.509956150664796e-05,
      "loss": 3.5271,
      "step": 37
    },
    {
      "epoch": 0.060775689724110356,
      "grad_norm": 3.590907096862793,
      "learning_rate": 9.473646649103818e-05,
      "loss": 3.6892,
      "step": 38
    },
    {
      "epoch": 0.062375049980008,
      "grad_norm": 4.963756084442139,
      "learning_rate": 9.43611409721806e-05,
      "loss": 3.4258,
      "step": 39
    },
    {
      "epoch": 0.06397441023590564,
      "grad_norm": 5.007536888122559,
      "learning_rate": 9.397368756032445e-05,
      "loss": 3.7067,
      "step": 40
    },
    {
      "epoch": 0.06557377049180328,
      "grad_norm": 3.881434679031372,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.6699,
      "step": 41
    },
    {
      "epoch": 0.06717313074770093,
      "grad_norm": 4.175216197967529,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.1801,
      "step": 42
    },
    {
      "epoch": 0.06877249100359856,
      "grad_norm": 3.6534883975982666,
      "learning_rate": 9.273963562927695e-05,
      "loss": 3.3731,
      "step": 43
    },
    {
      "epoch": 0.0703718512594962,
      "grad_norm": 4.464999198913574,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.4255,
      "step": 44
    },
    {
      "epoch": 0.07197121151539385,
      "grad_norm": 3.687138319015503,
      "learning_rate": 9.185832391312644e-05,
      "loss": 3.5646,
      "step": 45
    },
    {
      "epoch": 0.07357057177129149,
      "grad_norm": 4.271640777587891,
      "learning_rate": 9.140044155740101e-05,
      "loss": 3.4668,
      "step": 46
    },
    {
      "epoch": 0.07516993202718912,
      "grad_norm": 4.933566093444824,
      "learning_rate": 9.093124073433463e-05,
      "loss": 3.3196,
      "step": 47
    },
    {
      "epoch": 0.07676929228308677,
      "grad_norm": 4.293997287750244,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.7283,
      "step": 48
    },
    {
      "epoch": 0.07836865253898441,
      "grad_norm": 5.004490375518799,
      "learning_rate": 8.995939984474624e-05,
      "loss": 3.6459,
      "step": 49
    },
    {
      "epoch": 0.07996801279488205,
      "grad_norm": 16.031347274780273,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.1382,
      "step": 50
    },
    {
      "epoch": 0.07996801279488205,
      "eval_loss": 5.342373371124268,
      "eval_runtime": 30.8242,
      "eval_samples_per_second": 34.161,
      "eval_steps_per_second": 8.565,
      "step": 50
    },
    {
      "epoch": 0.08156737305077968,
      "grad_norm": 32.6708984375,
      "learning_rate": 8.894386393810563e-05,
      "loss": 6.4842,
      "step": 51
    },
    {
      "epoch": 0.08316673330667733,
      "grad_norm": 24.282751083374023,
      "learning_rate": 8.842005554284296e-05,
      "loss": 5.1974,
      "step": 52
    },
    {
      "epoch": 0.08476609356257497,
      "grad_norm": 11.314471244812012,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.3299,
      "step": 53
    },
    {
      "epoch": 0.08636545381847262,
      "grad_norm": 4.212861061096191,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.0722,
      "step": 54
    },
    {
      "epoch": 0.08796481407437025,
      "grad_norm": 3.833353281021118,
      "learning_rate": 8.678619553365659e-05,
      "loss": 4.1604,
      "step": 55
    },
    {
      "epoch": 0.08956417433026789,
      "grad_norm": 3.640949249267578,
      "learning_rate": 8.622126023955446e-05,
      "loss": 4.1033,
      "step": 56
    },
    {
      "epoch": 0.09116353458616554,
      "grad_norm": 3.6259710788726807,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.8915,
      "step": 57
    },
    {
      "epoch": 0.09276289484206318,
      "grad_norm": 3.829282760620117,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.8561,
      "step": 58
    },
    {
      "epoch": 0.09436225509796081,
      "grad_norm": 3.1499533653259277,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.7496,
      "step": 59
    },
    {
      "epoch": 0.09596161535385846,
      "grad_norm": 2.857248067855835,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.6361,
      "step": 60
    },
    {
      "epoch": 0.0975609756097561,
      "grad_norm": 2.9103546142578125,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.6957,
      "step": 61
    },
    {
      "epoch": 0.09916033586565375,
      "grad_norm": 3.3865253925323486,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.7705,
      "step": 62
    },
    {
      "epoch": 0.10075969612155138,
      "grad_norm": 3.29915714263916,
      "learning_rate": 8.199842702516583e-05,
      "loss": 3.6468,
      "step": 63
    },
    {
      "epoch": 0.10235905637744902,
      "grad_norm": 3.5676019191741943,
      "learning_rate": 8.135881792367686e-05,
      "loss": 3.725,
      "step": 64
    },
    {
      "epoch": 0.10395841663334666,
      "grad_norm": 3.2117583751678467,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.731,
      "step": 65
    },
    {
      "epoch": 0.10555777688924431,
      "grad_norm": 3.1759660243988037,
      "learning_rate": 8.005405736415126e-05,
      "loss": 3.6828,
      "step": 66
    },
    {
      "epoch": 0.10715713714514194,
      "grad_norm": 3.3611044883728027,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.4644,
      "step": 67
    },
    {
      "epoch": 0.10875649740103958,
      "grad_norm": 2.910616636276245,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.5939,
      "step": 68
    },
    {
      "epoch": 0.11035585765693723,
      "grad_norm": 3.056044578552246,
      "learning_rate": 7.803575286758364e-05,
      "loss": 3.7692,
      "step": 69
    },
    {
      "epoch": 0.11195521791283487,
      "grad_norm": 3.103546619415283,
      "learning_rate": 7.734740790612136e-05,
      "loss": 3.7308,
      "step": 70
    },
    {
      "epoch": 0.1135545781687325,
      "grad_norm": 2.6254987716674805,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.1862,
      "step": 71
    },
    {
      "epoch": 0.11515393842463015,
      "grad_norm": 3.5531318187713623,
      "learning_rate": 7.594847868906076e-05,
      "loss": 3.919,
      "step": 72
    },
    {
      "epoch": 0.11675329868052779,
      "grad_norm": 3.053609848022461,
      "learning_rate": 7.52382768867422e-05,
      "loss": 3.2853,
      "step": 73
    },
    {
      "epoch": 0.11835265893642544,
      "grad_norm": 2.562059164047241,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.5101,
      "step": 74
    },
    {
      "epoch": 0.11995201919232307,
      "grad_norm": 3.014320135116577,
      "learning_rate": 7.379736965185368e-05,
      "loss": 3.4666,
      "step": 75
    },
    {
      "epoch": 0.12155137944822071,
      "grad_norm": 3.250333786010742,
      "learning_rate": 7.30670581489344e-05,
      "loss": 3.2705,
      "step": 76
    },
    {
      "epoch": 0.12315073970411836,
      "grad_norm": 3.2853028774261475,
      "learning_rate": 7.233044034264034e-05,
      "loss": 3.6906,
      "step": 77
    },
    {
      "epoch": 0.124750099960016,
      "grad_norm": 3.3443684577941895,
      "learning_rate": 7.158771761692464e-05,
      "loss": 3.3191,
      "step": 78
    },
    {
      "epoch": 0.12634946021591364,
      "grad_norm": 3.1872684955596924,
      "learning_rate": 7.083909302476453e-05,
      "loss": 3.6239,
      "step": 79
    },
    {
      "epoch": 0.12794882047181128,
      "grad_norm": 2.8262345790863037,
      "learning_rate": 7.008477123264848e-05,
      "loss": 3.234,
      "step": 80
    },
    {
      "epoch": 0.1295481807277089,
      "grad_norm": 3.2739531993865967,
      "learning_rate": 6.932495846462261e-05,
      "loss": 3.6397,
      "step": 81
    },
    {
      "epoch": 0.13114754098360656,
      "grad_norm": 3.359330892562866,
      "learning_rate": 6.855986244591104e-05,
      "loss": 3.4603,
      "step": 82
    },
    {
      "epoch": 0.1327469012395042,
      "grad_norm": 3.23645281791687,
      "learning_rate": 6.778969234612584e-05,
      "loss": 3.2836,
      "step": 83
    },
    {
      "epoch": 0.13434626149540185,
      "grad_norm": 3.0637645721435547,
      "learning_rate": 6.701465872208216e-05,
      "loss": 3.4558,
      "step": 84
    },
    {
      "epoch": 0.13594562175129948,
      "grad_norm": 5.187008857727051,
      "learning_rate": 6.623497346023418e-05,
      "loss": 3.548,
      "step": 85
    },
    {
      "epoch": 0.13754498200719711,
      "grad_norm": 3.6012625694274902,
      "learning_rate": 6.545084971874738e-05,
      "loss": 3.4211,
      "step": 86
    },
    {
      "epoch": 0.13914434226309477,
      "grad_norm": 3.0747921466827393,
      "learning_rate": 6.466250186922325e-05,
      "loss": 3.2654,
      "step": 87
    },
    {
      "epoch": 0.1407437025189924,
      "grad_norm": 3.4943957328796387,
      "learning_rate": 6.387014543809223e-05,
      "loss": 3.3613,
      "step": 88
    },
    {
      "epoch": 0.14234306277489003,
      "grad_norm": 4.116997241973877,
      "learning_rate": 6.307399704769099e-05,
      "loss": 3.6086,
      "step": 89
    },
    {
      "epoch": 0.1439424230307877,
      "grad_norm": 3.429699182510376,
      "learning_rate": 6.227427435703997e-05,
      "loss": 3.1947,
      "step": 90
    },
    {
      "epoch": 0.14554178328668532,
      "grad_norm": 3.231152296066284,
      "learning_rate": 6.147119600233758e-05,
      "loss": 3.3577,
      "step": 91
    },
    {
      "epoch": 0.14714114354258298,
      "grad_norm": 3.69215726852417,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.564,
      "step": 92
    },
    {
      "epoch": 0.1487405037984806,
      "grad_norm": 3.653225898742676,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.2543,
      "step": 93
    },
    {
      "epoch": 0.15033986405437824,
      "grad_norm": 3.6382060050964355,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.2178,
      "step": 94
    },
    {
      "epoch": 0.1519392243102759,
      "grad_norm": 5.147819995880127,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 3.3943,
      "step": 95
    },
    {
      "epoch": 0.15353858456617353,
      "grad_norm": 3.4799234867095947,
      "learning_rate": 5.74131823855921e-05,
      "loss": 3.4956,
      "step": 96
    },
    {
      "epoch": 0.15513794482207116,
      "grad_norm": 4.478823661804199,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 3.5668,
      "step": 97
    },
    {
      "epoch": 0.15673730507796882,
      "grad_norm": 3.7702443599700928,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.138,
      "step": 98
    },
    {
      "epoch": 0.15833666533386645,
      "grad_norm": 4.2597737312316895,
      "learning_rate": 5.495227651252315e-05,
      "loss": 3.3189,
      "step": 99
    },
    {
      "epoch": 0.1599360255897641,
      "grad_norm": 10.079608917236328,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 3.8804,
      "step": 100
    },
    {
      "epoch": 0.1599360255897641,
      "eval_loss": 4.284668445587158,
      "eval_runtime": 30.757,
      "eval_samples_per_second": 34.236,
      "eval_steps_per_second": 8.583,
      "step": 100
    },
    {
      "epoch": 0.16153538584566174,
      "grad_norm": 22.57893943786621,
      "learning_rate": 5.330452921628497e-05,
      "loss": 5.5792,
      "step": 101
    },
    {
      "epoch": 0.16313474610155937,
      "grad_norm": 21.179407119750977,
      "learning_rate": 5.247918773366112e-05,
      "loss": 5.247,
      "step": 102
    },
    {
      "epoch": 0.16473410635745703,
      "grad_norm": 17.917861938476562,
      "learning_rate": 5.165316846586541e-05,
      "loss": 4.7219,
      "step": 103
    },
    {
      "epoch": 0.16633346661335466,
      "grad_norm": 13.462945938110352,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 4.2791,
      "step": 104
    },
    {
      "epoch": 0.1679328268692523,
      "grad_norm": 7.720991611480713,
      "learning_rate": 5e-05,
      "loss": 3.7668,
      "step": 105
    },
    {
      "epoch": 0.16953218712514995,
      "grad_norm": 4.960745811462402,
      "learning_rate": 4.917330276168208e-05,
      "loss": 3.7819,
      "step": 106
    },
    {
      "epoch": 0.17113154738104758,
      "grad_norm": 3.39565372467041,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.4241,
      "step": 107
    },
    {
      "epoch": 0.17273090763694524,
      "grad_norm": 3.3428215980529785,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.3786,
      "step": 108
    },
    {
      "epoch": 0.17433026789284287,
      "grad_norm": 3.824470281600952,
      "learning_rate": 4.669547078371504e-05,
      "loss": 3.5418,
      "step": 109
    },
    {
      "epoch": 0.1759296281487405,
      "grad_norm": 3.2780776023864746,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 3.3343,
      "step": 110
    },
    {
      "epoch": 0.17752898840463815,
      "grad_norm": 3.5912094116210938,
      "learning_rate": 4.504772348747687e-05,
      "loss": 3.4995,
      "step": 111
    },
    {
      "epoch": 0.17912834866053579,
      "grad_norm": 3.4991390705108643,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.4045,
      "step": 112
    },
    {
      "epoch": 0.18072770891643342,
      "grad_norm": 3.4763553142547607,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 3.3832,
      "step": 113
    },
    {
      "epoch": 0.18232706917233107,
      "grad_norm": 3.423131227493286,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.3545,
      "step": 114
    },
    {
      "epoch": 0.1839264294282287,
      "grad_norm": 3.057525396347046,
      "learning_rate": 4.17702704859633e-05,
      "loss": 3.6266,
      "step": 115
    },
    {
      "epoch": 0.18552578968412636,
      "grad_norm": 3.1434223651885986,
      "learning_rate": 4.095597328339452e-05,
      "loss": 3.6233,
      "step": 116
    },
    {
      "epoch": 0.187125149940024,
      "grad_norm": 2.9135935306549072,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 3.4099,
      "step": 117
    },
    {
      "epoch": 0.18872451019592162,
      "grad_norm": 2.9700891971588135,
      "learning_rate": 3.933501846281267e-05,
      "loss": 3.2518,
      "step": 118
    },
    {
      "epoch": 0.19032387045181928,
      "grad_norm": 2.8020801544189453,
      "learning_rate": 3.852880399766243e-05,
      "loss": 3.3184,
      "step": 119
    },
    {
      "epoch": 0.1919232307077169,
      "grad_norm": 2.6947579383850098,
      "learning_rate": 3.772572564296005e-05,
      "loss": 3.401,
      "step": 120
    },
    {
      "epoch": 0.19352259096361454,
      "grad_norm": 2.535811185836792,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 3.1938,
      "step": 121
    },
    {
      "epoch": 0.1951219512195122,
      "grad_norm": 3.152883529663086,
      "learning_rate": 3.612985456190778e-05,
      "loss": 3.2137,
      "step": 122
    },
    {
      "epoch": 0.19672131147540983,
      "grad_norm": 2.810652732849121,
      "learning_rate": 3.533749813077677e-05,
      "loss": 3.2135,
      "step": 123
    },
    {
      "epoch": 0.1983206717313075,
      "grad_norm": 2.7589709758758545,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 3.0276,
      "step": 124
    },
    {
      "epoch": 0.19992003198720512,
      "grad_norm": 4.1250128746032715,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 3.2273,
      "step": 125
    },
    {
      "epoch": 0.20151939224310275,
      "grad_norm": 2.6969239711761475,
      "learning_rate": 3.298534127791785e-05,
      "loss": 3.3883,
      "step": 126
    },
    {
      "epoch": 0.2031187524990004,
      "grad_norm": 2.96826434135437,
      "learning_rate": 3.221030765387417e-05,
      "loss": 3.5533,
      "step": 127
    },
    {
      "epoch": 0.20471811275489804,
      "grad_norm": 3.048295259475708,
      "learning_rate": 3.144013755408895e-05,
      "loss": 3.3319,
      "step": 128
    },
    {
      "epoch": 0.20631747301079567,
      "grad_norm": 3.083566188812256,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 3.1547,
      "step": 129
    },
    {
      "epoch": 0.20791683326669333,
      "grad_norm": 3.1777796745300293,
      "learning_rate": 2.991522876735154e-05,
      "loss": 3.3645,
      "step": 130
    },
    {
      "epoch": 0.20951619352259096,
      "grad_norm": 3.0185019969940186,
      "learning_rate": 2.916090697523549e-05,
      "loss": 3.0949,
      "step": 131
    },
    {
      "epoch": 0.21111555377848862,
      "grad_norm": 3.4715216159820557,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 3.0134,
      "step": 132
    },
    {
      "epoch": 0.21271491403438625,
      "grad_norm": 3.3303847312927246,
      "learning_rate": 2.766955965735968e-05,
      "loss": 3.1923,
      "step": 133
    },
    {
      "epoch": 0.21431427429028388,
      "grad_norm": 3.7659823894500732,
      "learning_rate": 2.693294185106562e-05,
      "loss": 3.1971,
      "step": 134
    },
    {
      "epoch": 0.21591363454618154,
      "grad_norm": 3.3014018535614014,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 3.3184,
      "step": 135
    },
    {
      "epoch": 0.21751299480207917,
      "grad_norm": 3.0923290252685547,
      "learning_rate": 2.547882480847461e-05,
      "loss": 3.2694,
      "step": 136
    },
    {
      "epoch": 0.2191123550579768,
      "grad_norm": 2.9805493354797363,
      "learning_rate": 2.476172311325783e-05,
      "loss": 3.3121,
      "step": 137
    },
    {
      "epoch": 0.22071171531387446,
      "grad_norm": 3.2085442543029785,
      "learning_rate": 2.405152131093926e-05,
      "loss": 3.0345,
      "step": 138
    },
    {
      "epoch": 0.2223110755697721,
      "grad_norm": 4.016242027282715,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 3.0142,
      "step": 139
    },
    {
      "epoch": 0.22391043582566975,
      "grad_norm": 3.0115888118743896,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 3.1706,
      "step": 140
    },
    {
      "epoch": 0.22550979608156738,
      "grad_norm": 3.3110365867614746,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.0881,
      "step": 141
    },
    {
      "epoch": 0.227109156337465,
      "grad_norm": 3.372243881225586,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.275,
      "step": 142
    },
    {
      "epoch": 0.22870851659336267,
      "grad_norm": 4.122211456298828,
      "learning_rate": 2.061073738537635e-05,
      "loss": 3.1405,
      "step": 143
    },
    {
      "epoch": 0.2303078768492603,
      "grad_norm": 3.8763859272003174,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 3.4546,
      "step": 144
    },
    {
      "epoch": 0.23190723710515793,
      "grad_norm": 3.9019641876220703,
      "learning_rate": 1.928936436551661e-05,
      "loss": 3.1657,
      "step": 145
    },
    {
      "epoch": 0.23350659736105558,
      "grad_norm": 3.8251686096191406,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 3.3089,
      "step": 146
    },
    {
      "epoch": 0.23510595761695322,
      "grad_norm": 4.333549499511719,
      "learning_rate": 1.800157297483417e-05,
      "loss": 3.0533,
      "step": 147
    },
    {
      "epoch": 0.23670531787285087,
      "grad_norm": 4.1478166580200195,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 3.1415,
      "step": 148
    },
    {
      "epoch": 0.2383046781287485,
      "grad_norm": 6.864630222320557,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 3.0762,
      "step": 149
    },
    {
      "epoch": 0.23990403838464613,
      "grad_norm": 7.250833988189697,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 3.1724,
      "step": 150
    },
    {
      "epoch": 0.23990403838464613,
      "eval_loss": 3.5659871101379395,
      "eval_runtime": 30.7558,
      "eval_samples_per_second": 34.237,
      "eval_steps_per_second": 8.584,
      "step": 150
    },
    {
      "epoch": 0.2415033986405438,
      "grad_norm": 11.09546184539795,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.902,
      "step": 151
    },
    {
      "epoch": 0.24310275889644142,
      "grad_norm": 10.348603248596191,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 4.0178,
      "step": 152
    },
    {
      "epoch": 0.24470211915233905,
      "grad_norm": 10.795031547546387,
      "learning_rate": 1.435357758543015e-05,
      "loss": 4.2387,
      "step": 153
    },
    {
      "epoch": 0.2463014794082367,
      "grad_norm": 10.855401992797852,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 4.0424,
      "step": 154
    },
    {
      "epoch": 0.24790083966413434,
      "grad_norm": 9.387147903442383,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 3.7014,
      "step": 155
    },
    {
      "epoch": 0.249500199920032,
      "grad_norm": 8.649288177490234,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 3.8136,
      "step": 156
    },
    {
      "epoch": 0.2510995601759296,
      "grad_norm": 8.260371208190918,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 3.647,
      "step": 157
    },
    {
      "epoch": 0.2526989204318273,
      "grad_norm": 7.8920183181762695,
      "learning_rate": 1.157994445715706e-05,
      "loss": 3.5079,
      "step": 158
    },
    {
      "epoch": 0.2542982806877249,
      "grad_norm": 7.546154022216797,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.7488,
      "step": 159
    },
    {
      "epoch": 0.25589764094362255,
      "grad_norm": 5.844323635101318,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 3.3266,
      "step": 160
    },
    {
      "epoch": 0.2574970011995202,
      "grad_norm": 5.708782196044922,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 3.2808,
      "step": 161
    },
    {
      "epoch": 0.2590963614554178,
      "grad_norm": 4.468139171600342,
      "learning_rate": 9.549150281252633e-06,
      "loss": 3.5266,
      "step": 162
    },
    {
      "epoch": 0.2606957217113155,
      "grad_norm": 4.116734504699707,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.1623,
      "step": 163
    },
    {
      "epoch": 0.26229508196721313,
      "grad_norm": 3.7286252975463867,
      "learning_rate": 8.599558442598998e-06,
      "loss": 3.3157,
      "step": 164
    },
    {
      "epoch": 0.26389444222311076,
      "grad_norm": 3.5930635929107666,
      "learning_rate": 8.141676086873572e-06,
      "loss": 3.2194,
      "step": 165
    },
    {
      "epoch": 0.2654938024790084,
      "grad_norm": 3.347017288208008,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.1885,
      "step": 166
    },
    {
      "epoch": 0.267093162734906,
      "grad_norm": 2.9623448848724365,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.3232,
      "step": 167
    },
    {
      "epoch": 0.2686925229908037,
      "grad_norm": 3.4055099487304688,
      "learning_rate": 6.837175952121306e-06,
      "loss": 3.1354,
      "step": 168
    },
    {
      "epoch": 0.27029188324670134,
      "grad_norm": 3.100329637527466,
      "learning_rate": 6.425787818636131e-06,
      "loss": 3.2555,
      "step": 169
    },
    {
      "epoch": 0.27189124350259897,
      "grad_norm": 2.8353939056396484,
      "learning_rate": 6.026312439675552e-06,
      "loss": 3.2862,
      "step": 170
    },
    {
      "epoch": 0.2734906037584966,
      "grad_norm": 3.1711535453796387,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 3.2783,
      "step": 171
    },
    {
      "epoch": 0.27508996401439423,
      "grad_norm": 3.049731969833374,
      "learning_rate": 5.263533508961827e-06,
      "loss": 3.3446,
      "step": 172
    },
    {
      "epoch": 0.27668932427029186,
      "grad_norm": 2.9547441005706787,
      "learning_rate": 4.900438493352055e-06,
      "loss": 3.3257,
      "step": 173
    },
    {
      "epoch": 0.27828868452618954,
      "grad_norm": 2.9613542556762695,
      "learning_rate": 4.549673247541875e-06,
      "loss": 3.2181,
      "step": 174
    },
    {
      "epoch": 0.2798880447820872,
      "grad_norm": 2.891442060470581,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 3.2939,
      "step": 175
    },
    {
      "epoch": 0.2814874050379848,
      "grad_norm": 3.8245656490325928,
      "learning_rate": 3.885512251130763e-06,
      "loss": 3.1729,
      "step": 176
    },
    {
      "epoch": 0.28308676529388244,
      "grad_norm": 3.1802756786346436,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 3.3786,
      "step": 177
    },
    {
      "epoch": 0.28468612554978007,
      "grad_norm": 3.655912399291992,
      "learning_rate": 3.271776770026963e-06,
      "loss": 3.2414,
      "step": 178
    },
    {
      "epoch": 0.28628548580567775,
      "grad_norm": 3.212887763977051,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 3.3478,
      "step": 179
    },
    {
      "epoch": 0.2878848460615754,
      "grad_norm": 3.5430848598480225,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 3.2744,
      "step": 180
    },
    {
      "epoch": 0.289484206317473,
      "grad_norm": 3.1065115928649902,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 3.1332,
      "step": 181
    },
    {
      "epoch": 0.29108356657337064,
      "grad_norm": 3.7347214221954346,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 3.3164,
      "step": 182
    },
    {
      "epoch": 0.2926829268292683,
      "grad_norm": 3.184579849243164,
      "learning_rate": 1.962316193157593e-06,
      "loss": 3.2521,
      "step": 183
    },
    {
      "epoch": 0.29428228708516596,
      "grad_norm": 3.4126813411712646,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 3.2861,
      "step": 184
    },
    {
      "epoch": 0.2958816473410636,
      "grad_norm": 3.7646429538726807,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 3.1318,
      "step": 185
    },
    {
      "epoch": 0.2974810075969612,
      "grad_norm": 3.781168222427368,
      "learning_rate": 1.333670137599713e-06,
      "loss": 3.0655,
      "step": 186
    },
    {
      "epoch": 0.29908036785285885,
      "grad_norm": 3.8673880100250244,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 3.2974,
      "step": 187
    },
    {
      "epoch": 0.3006797281087565,
      "grad_norm": 3.7562146186828613,
      "learning_rate": 9.810017062595322e-07,
      "loss": 3.1772,
      "step": 188
    },
    {
      "epoch": 0.3022790883646541,
      "grad_norm": 4.822183609008789,
      "learning_rate": 8.247462563808817e-07,
      "loss": 3.3681,
      "step": 189
    },
    {
      "epoch": 0.3038784486205518,
      "grad_norm": 4.63447904586792,
      "learning_rate": 6.819348298638839e-07,
      "loss": 3.4163,
      "step": 190
    },
    {
      "epoch": 0.30547780887644943,
      "grad_norm": 4.988310813903809,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.5577,
      "step": 191
    },
    {
      "epoch": 0.30707716913234706,
      "grad_norm": 4.339115619659424,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.121,
      "step": 192
    },
    {
      "epoch": 0.3086765293882447,
      "grad_norm": 5.104634761810303,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 3.3514,
      "step": 193
    },
    {
      "epoch": 0.3102758896441423,
      "grad_norm": 4.544820308685303,
      "learning_rate": 2.458548727494292e-07,
      "loss": 3.0763,
      "step": 194
    },
    {
      "epoch": 0.31187524990004,
      "grad_norm": 4.733354091644287,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 3.1469,
      "step": 195
    },
    {
      "epoch": 0.31347461015593764,
      "grad_norm": 5.4134907722473145,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 3.2098,
      "step": 196
    },
    {
      "epoch": 0.31507397041183527,
      "grad_norm": 6.04467248916626,
      "learning_rate": 6.150154258476315e-08,
      "loss": 3.2694,
      "step": 197
    },
    {
      "epoch": 0.3166733306677329,
      "grad_norm": 6.402224540710449,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 3.2773,
      "step": 198
    },
    {
      "epoch": 0.31827269092363053,
      "grad_norm": 8.071269035339355,
      "learning_rate": 6.834750376549792e-09,
      "loss": 3.4497,
      "step": 199
    },
    {
      "epoch": 0.3198720511795282,
      "grad_norm": 11.67451000213623,
      "learning_rate": 0.0,
      "loss": 4.2834,
      "step": 200
    },
    {
      "epoch": 0.3198720511795282,
      "eval_loss": 3.217210531234741,
      "eval_runtime": 30.7922,
      "eval_samples_per_second": 34.197,
      "eval_steps_per_second": 8.574,
      "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": 5.4428946137088e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}