{
  "best_metric": 0.27505218982696533,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.6739679865206403,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003369839932603201,
      "grad_norm": 7.51057243347168,
      "learning_rate": 1e-05,
      "loss": 2.1494,
      "step": 1
    },
    {
      "epoch": 0.003369839932603201,
      "eval_loss": 0.8092716336250305,
      "eval_runtime": 120.282,
      "eval_samples_per_second": 4.157,
      "eval_steps_per_second": 1.039,
      "step": 1
    },
    {
      "epoch": 0.006739679865206402,
      "grad_norm": 9.15483570098877,
      "learning_rate": 2e-05,
      "loss": 2.7371,
      "step": 2
    },
    {
      "epoch": 0.010109519797809604,
      "grad_norm": 7.816402912139893,
      "learning_rate": 3e-05,
      "loss": 2.5099,
      "step": 3
    },
    {
      "epoch": 0.013479359730412805,
      "grad_norm": 5.4665608406066895,
      "learning_rate": 4e-05,
      "loss": 2.3463,
      "step": 4
    },
    {
      "epoch": 0.016849199663016005,
      "grad_norm": 5.093127727508545,
      "learning_rate": 5e-05,
      "loss": 2.1346,
      "step": 5
    },
    {
      "epoch": 0.020219039595619208,
      "grad_norm": 5.4731855392456055,
      "learning_rate": 6e-05,
      "loss": 2.2381,
      "step": 6
    },
    {
      "epoch": 0.02358887952822241,
      "grad_norm": 4.477018356323242,
      "learning_rate": 7e-05,
      "loss": 1.927,
      "step": 7
    },
    {
      "epoch": 0.02695871946082561,
      "grad_norm": 4.402273178100586,
      "learning_rate": 8e-05,
      "loss": 2.0187,
      "step": 8
    },
    {
      "epoch": 0.030328559393428812,
      "grad_norm": 3.9021198749542236,
      "learning_rate": 9e-05,
      "loss": 1.752,
      "step": 9
    },
    {
      "epoch": 0.03369839932603201,
      "grad_norm": 4.180506229400635,
      "learning_rate": 0.0001,
      "loss": 1.8939,
      "step": 10
    },
    {
      "epoch": 0.03706823925863521,
      "grad_norm": 3.6741297245025635,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.7022,
      "step": 11
    },
    {
      "epoch": 0.040438079191238416,
      "grad_norm": 3.5289134979248047,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.8531,
      "step": 12
    },
    {
      "epoch": 0.04380791912384162,
      "grad_norm": 4.152334690093994,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.85,
      "step": 13
    },
    {
      "epoch": 0.04717775905644482,
      "grad_norm": 3.1703765392303467,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.6875,
      "step": 14
    },
    {
      "epoch": 0.05054759898904802,
      "grad_norm": 3.482504367828369,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.8679,
      "step": 15
    },
    {
      "epoch": 0.05391743892165122,
      "grad_norm": 2.8432414531707764,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.777,
      "step": 16
    },
    {
      "epoch": 0.05728727885425442,
      "grad_norm": 3.414099931716919,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.6603,
      "step": 17
    },
    {
      "epoch": 0.060657118786857624,
      "grad_norm": 3.249868869781494,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.7211,
      "step": 18
    },
    {
      "epoch": 0.06402695871946082,
      "grad_norm": 2.949674606323242,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.6861,
      "step": 19
    },
    {
      "epoch": 0.06739679865206402,
      "grad_norm": 2.7226510047912598,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.6053,
      "step": 20
    },
    {
      "epoch": 0.07076663858466722,
      "grad_norm": 3.109335422515869,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.6572,
      "step": 21
    },
    {
      "epoch": 0.07413647851727043,
      "grad_norm": 3.3253257274627686,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.5954,
      "step": 22
    },
    {
      "epoch": 0.07750631844987363,
      "grad_norm": 3.090282917022705,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.6515,
      "step": 23
    },
    {
      "epoch": 0.08087615838247683,
      "grad_norm": 2.9994139671325684,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.7576,
      "step": 24
    },
    {
      "epoch": 0.08424599831508003,
      "grad_norm": 3.127659320831299,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.6768,
      "step": 25
    },
    {
      "epoch": 0.08761583824768324,
      "grad_norm": 3.3083384037017822,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.568,
      "step": 26
    },
    {
      "epoch": 0.09098567818028644,
      "grad_norm": 3.118006706237793,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.5629,
      "step": 27
    },
    {
      "epoch": 0.09435551811288964,
      "grad_norm": 3.579113245010376,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.5379,
      "step": 28
    },
    {
      "epoch": 0.09772535804549284,
      "grad_norm": 3.026153802871704,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.4824,
      "step": 29
    },
    {
      "epoch": 0.10109519797809605,
      "grad_norm": 2.802640914916992,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.2536,
      "step": 30
    },
    {
      "epoch": 0.10446503791069923,
      "grad_norm": 3.0235989093780518,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.3792,
      "step": 31
    },
    {
      "epoch": 0.10783487784330244,
      "grad_norm": 3.0789220333099365,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.2263,
      "step": 32
    },
    {
      "epoch": 0.11120471777590564,
      "grad_norm": 2.703223943710327,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.1993,
      "step": 33
    },
    {
      "epoch": 0.11457455770850884,
      "grad_norm": 2.7497401237487793,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.4319,
      "step": 34
    },
    {
      "epoch": 0.11794439764111204,
      "grad_norm": 2.8858532905578613,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.549,
      "step": 35
    },
    {
      "epoch": 0.12131423757371525,
      "grad_norm": 2.815218925476074,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5933,
      "step": 36
    },
    {
      "epoch": 0.12468407750631845,
      "grad_norm": 2.8283677101135254,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.599,
      "step": 37
    },
    {
      "epoch": 0.12805391743892164,
      "grad_norm": 2.8242433071136475,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.559,
      "step": 38
    },
    {
      "epoch": 0.13142375737152484,
      "grad_norm": 2.870215654373169,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.6841,
      "step": 39
    },
    {
      "epoch": 0.13479359730412804,
      "grad_norm": 2.8067023754119873,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.6646,
      "step": 40
    },
    {
      "epoch": 0.13816343723673125,
      "grad_norm": 2.9085981845855713,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.7161,
      "step": 41
    },
    {
      "epoch": 0.14153327716933445,
      "grad_norm": 2.7535951137542725,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.5564,
      "step": 42
    },
    {
      "epoch": 0.14490311710193765,
      "grad_norm": 2.7437710762023926,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.5449,
      "step": 43
    },
    {
      "epoch": 0.14827295703454085,
      "grad_norm": 2.6580138206481934,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.4529,
      "step": 44
    },
    {
      "epoch": 0.15164279696714406,
      "grad_norm": 3.217501163482666,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.4437,
      "step": 45
    },
    {
      "epoch": 0.15501263689974726,
      "grad_norm": 4.156802177429199,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5593,
      "step": 46
    },
    {
      "epoch": 0.15838247683235046,
      "grad_norm": 2.9678525924682617,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.4153,
      "step": 47
    },
    {
      "epoch": 0.16175231676495366,
      "grad_norm": 2.6032907962799072,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3261,
      "step": 48
    },
    {
      "epoch": 0.16512215669755687,
      "grad_norm": 2.3966689109802246,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3488,
      "step": 49
    },
    {
      "epoch": 0.16849199663016007,
      "grad_norm": 3.1099071502685547,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.5242,
      "step": 50
    },
    {
      "epoch": 0.16849199663016007,
      "eval_loss": 0.3645605444908142,
      "eval_runtime": 121.6366,
      "eval_samples_per_second": 4.111,
      "eval_steps_per_second": 1.028,
      "step": 50
    },
    {
      "epoch": 0.17186183656276327,
      "grad_norm": 2.536522388458252,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.5471,
      "step": 51
    },
    {
      "epoch": 0.17523167649536647,
      "grad_norm": 2.1130993366241455,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.5713,
      "step": 52
    },
    {
      "epoch": 0.17860151642796968,
      "grad_norm": 1.899036169052124,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.5713,
      "step": 53
    },
    {
      "epoch": 0.18197135636057288,
      "grad_norm": 2.3696439266204834,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.583,
      "step": 54
    },
    {
      "epoch": 0.18534119629317608,
      "grad_norm": 2.156263589859009,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.4893,
      "step": 55
    },
    {
      "epoch": 0.18871103622577928,
      "grad_norm": 1.9296045303344727,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4985,
      "step": 56
    },
    {
      "epoch": 0.19208087615838249,
      "grad_norm": 1.9608573913574219,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4801,
      "step": 57
    },
    {
      "epoch": 0.1954507160909857,
      "grad_norm": 2.0413198471069336,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4525,
      "step": 58
    },
    {
      "epoch": 0.1988205560235889,
      "grad_norm": 1.96634840965271,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.4184,
      "step": 59
    },
    {
      "epoch": 0.2021903959561921,
      "grad_norm": 1.993112325668335,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.342,
      "step": 60
    },
    {
      "epoch": 0.2055602358887953,
      "grad_norm": 1.9955614805221558,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.2742,
      "step": 61
    },
    {
      "epoch": 0.20893007582139847,
      "grad_norm": 1.9142816066741943,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.114,
      "step": 62
    },
    {
      "epoch": 0.21229991575400167,
      "grad_norm": 2.074475049972534,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.3422,
      "step": 63
    },
    {
      "epoch": 0.21566975568660487,
      "grad_norm": 2.186030626296997,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.3134,
      "step": 64
    },
    {
      "epoch": 0.21903959561920808,
      "grad_norm": 2.226123571395874,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.6073,
      "step": 65
    },
    {
      "epoch": 0.22240943555181128,
      "grad_norm": 2.3604118824005127,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.5057,
      "step": 66
    },
    {
      "epoch": 0.22577927548441448,
      "grad_norm": 2.158315896987915,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5419,
      "step": 67
    },
    {
      "epoch": 0.22914911541701768,
      "grad_norm": 2.1268155574798584,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.4379,
      "step": 68
    },
    {
      "epoch": 0.2325189553496209,
      "grad_norm": 2.069324254989624,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.3387,
      "step": 69
    },
    {
      "epoch": 0.2358887952822241,
      "grad_norm": 2.181563377380371,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.6706,
      "step": 70
    },
    {
      "epoch": 0.2392586352148273,
      "grad_norm": 2.1285486221313477,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.2435,
      "step": 71
    },
    {
      "epoch": 0.2426284751474305,
      "grad_norm": 2.0617074966430664,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.2847,
      "step": 72
    },
    {
      "epoch": 0.2459983150800337,
      "grad_norm": 2.0489954948425293,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.3647,
      "step": 73
    },
    {
      "epoch": 0.2493681550126369,
      "grad_norm": 2.035676956176758,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.1035,
      "step": 74
    },
    {
      "epoch": 0.2527379949452401,
      "grad_norm": 1.9856700897216797,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.0442,
      "step": 75
    },
    {
      "epoch": 0.2561078348778433,
      "grad_norm": 2.228813886642456,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.2683,
      "step": 76
    },
    {
      "epoch": 0.2594776748104465,
      "grad_norm": 2.338102340698242,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.2413,
      "step": 77
    },
    {
      "epoch": 0.2628475147430497,
      "grad_norm": 2.1351075172424316,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.0997,
      "step": 78
    },
    {
      "epoch": 0.2662173546756529,
      "grad_norm": 2.4263010025024414,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1819,
      "step": 79
    },
    {
      "epoch": 0.2695871946082561,
      "grad_norm": 2.098903179168701,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.0219,
      "step": 80
    },
    {
      "epoch": 0.2729570345408593,
      "grad_norm": 2.0994834899902344,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.9817,
      "step": 81
    },
    {
      "epoch": 0.2763268744734625,
      "grad_norm": 2.0242087841033936,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.8622,
      "step": 82
    },
    {
      "epoch": 0.2796967144060657,
      "grad_norm": 1.9342875480651855,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.9491,
      "step": 83
    },
    {
      "epoch": 0.2830665543386689,
      "grad_norm": 2.0760669708251953,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.1232,
      "step": 84
    },
    {
      "epoch": 0.2864363942712721,
      "grad_norm": 2.1343932151794434,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.1115,
      "step": 85
    },
    {
      "epoch": 0.2898062342038753,
      "grad_norm": 2.311663866043091,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.325,
      "step": 86
    },
    {
      "epoch": 0.2931760741364785,
      "grad_norm": 2.1452784538269043,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.301,
      "step": 87
    },
    {
      "epoch": 0.2965459140690817,
      "grad_norm": 2.288419246673584,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.2846,
      "step": 88
    },
    {
      "epoch": 0.2999157540016849,
      "grad_norm": 2.253816604614258,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.2931,
      "step": 89
    },
    {
      "epoch": 0.3032855939342881,
      "grad_norm": 2.3119921684265137,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.3536,
      "step": 90
    },
    {
      "epoch": 0.3066554338668913,
      "grad_norm": 2.3031837940216064,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.4333,
      "step": 91
    },
    {
      "epoch": 0.3100252737994945,
      "grad_norm": 2.3394315242767334,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.3935,
      "step": 92
    },
    {
      "epoch": 0.3133951137320977,
      "grad_norm": 2.75341534614563,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.351,
      "step": 93
    },
    {
      "epoch": 0.3167649536647009,
      "grad_norm": 2.6161062717437744,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.3347,
      "step": 94
    },
    {
      "epoch": 0.3201347935973041,
      "grad_norm": 2.364361524581909,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.9233,
      "step": 95
    },
    {
      "epoch": 0.3235046335299073,
      "grad_norm": 2.500354290008545,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.339,
      "step": 96
    },
    {
      "epoch": 0.32687447346251053,
      "grad_norm": 2.3215596675872803,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.2591,
      "step": 97
    },
    {
      "epoch": 0.33024431339511373,
      "grad_norm": 2.3461391925811768,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.2868,
      "step": 98
    },
    {
      "epoch": 0.33361415332771693,
      "grad_norm": 2.204437494277954,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.0907,
      "step": 99
    },
    {
      "epoch": 0.33698399326032014,
      "grad_norm": 2.5249979496002197,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.2808,
      "step": 100
    },
    {
      "epoch": 0.33698399326032014,
      "eval_loss": 0.314230740070343,
      "eval_runtime": 121.707,
      "eval_samples_per_second": 4.108,
      "eval_steps_per_second": 1.027,
      "step": 100
    },
    {
      "epoch": 0.34035383319292334,
      "grad_norm": 2.035517692565918,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4091,
      "step": 101
    },
    {
      "epoch": 0.34372367312552654,
      "grad_norm": 1.6589703559875488,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2788,
      "step": 102
    },
    {
      "epoch": 0.34709351305812974,
      "grad_norm": 1.6942716836929321,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.4311,
      "step": 103
    },
    {
      "epoch": 0.35046335299073295,
      "grad_norm": 1.6722877025604248,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.4587,
      "step": 104
    },
    {
      "epoch": 0.35383319292333615,
      "grad_norm": 1.6814851760864258,
      "learning_rate": 5e-05,
      "loss": 1.4119,
      "step": 105
    },
    {
      "epoch": 0.35720303285593935,
      "grad_norm": 1.7735084295272827,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.3435,
      "step": 106
    },
    {
      "epoch": 0.36057287278854255,
      "grad_norm": 1.7777315378189087,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2643,
      "step": 107
    },
    {
      "epoch": 0.36394271272114576,
      "grad_norm": 1.7229983806610107,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.4055,
      "step": 108
    },
    {
      "epoch": 0.36731255265374896,
      "grad_norm": 1.9317232370376587,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4089,
      "step": 109
    },
    {
      "epoch": 0.37068239258635216,
      "grad_norm": 1.8933583498001099,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.5829,
      "step": 110
    },
    {
      "epoch": 0.37405223251895536,
      "grad_norm": 1.8013765811920166,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.2522,
      "step": 111
    },
    {
      "epoch": 0.37742207245155857,
      "grad_norm": 1.92638099193573,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.2129,
      "step": 112
    },
    {
      "epoch": 0.38079191238416177,
      "grad_norm": 1.7256003618240356,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.1744,
      "step": 113
    },
    {
      "epoch": 0.38416175231676497,
      "grad_norm": 1.9628052711486816,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.3731,
      "step": 114
    },
    {
      "epoch": 0.3875315922493682,
      "grad_norm": 1.7589800357818604,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1395,
      "step": 115
    },
    {
      "epoch": 0.3909014321819714,
      "grad_norm": 1.9581555128097534,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.3804,
      "step": 116
    },
    {
      "epoch": 0.3942712721145746,
      "grad_norm": 1.763586163520813,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.2846,
      "step": 117
    },
    {
      "epoch": 0.3976411120471778,
      "grad_norm": 1.9399962425231934,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.3103,
      "step": 118
    },
    {
      "epoch": 0.401010951979781,
      "grad_norm": 2.0409631729125977,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.3013,
      "step": 119
    },
    {
      "epoch": 0.4043807919123842,
      "grad_norm": 1.9230983257293701,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.3784,
      "step": 120
    },
    {
      "epoch": 0.4077506318449874,
      "grad_norm": 1.9140021800994873,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.3676,
      "step": 121
    },
    {
      "epoch": 0.4111204717775906,
      "grad_norm": 1.8744556903839111,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2209,
      "step": 122
    },
    {
      "epoch": 0.4144903117101938,
      "grad_norm": 1.9762202501296997,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2086,
      "step": 123
    },
    {
      "epoch": 0.41786015164279694,
      "grad_norm": 1.9981001615524292,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.1732,
      "step": 124
    },
    {
      "epoch": 0.42122999157540014,
      "grad_norm": 2.0948734283447266,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.3828,
      "step": 125
    },
    {
      "epoch": 0.42459983150800334,
      "grad_norm": 1.8933742046356201,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.8188,
      "step": 126
    },
    {
      "epoch": 0.42796967144060655,
      "grad_norm": 1.9235304594039917,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.1265,
      "step": 127
    },
    {
      "epoch": 0.43133951137320975,
      "grad_norm": 2.0142617225646973,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.0303,
      "step": 128
    },
    {
      "epoch": 0.43470935130581295,
      "grad_norm": 1.917381763458252,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.8551,
      "step": 129
    },
    {
      "epoch": 0.43807919123841615,
      "grad_norm": 1.9005807638168335,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.8717,
      "step": 130
    },
    {
      "epoch": 0.44144903117101936,
      "grad_norm": 2.029496192932129,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.1523,
      "step": 131
    },
    {
      "epoch": 0.44481887110362256,
      "grad_norm": 2.0164740085601807,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.1836,
      "step": 132
    },
    {
      "epoch": 0.44818871103622576,
      "grad_norm": 1.8247592449188232,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.9052,
      "step": 133
    },
    {
      "epoch": 0.45155855096882896,
      "grad_norm": 2.021507978439331,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.0794,
      "step": 134
    },
    {
      "epoch": 0.45492839090143217,
      "grad_norm": 1.948276400566101,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.8556,
      "step": 135
    },
    {
      "epoch": 0.45829823083403537,
      "grad_norm": 2.1332807540893555,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.4236,
      "step": 136
    },
    {
      "epoch": 0.46166807076663857,
      "grad_norm": 2.0226590633392334,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.1958,
      "step": 137
    },
    {
      "epoch": 0.4650379106992418,
      "grad_norm": 2.013789176940918,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.1031,
      "step": 138
    },
    {
      "epoch": 0.468407750631845,
      "grad_norm": 2.3079583644866943,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.32,
      "step": 139
    },
    {
      "epoch": 0.4717775905644482,
      "grad_norm": 2.1277973651885986,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.1596,
      "step": 140
    },
    {
      "epoch": 0.4751474304970514,
      "grad_norm": 2.163935661315918,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2472,
      "step": 141
    },
    {
      "epoch": 0.4785172704296546,
      "grad_norm": 2.427151918411255,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.2116,
      "step": 142
    },
    {
      "epoch": 0.4818871103622578,
      "grad_norm": 2.2841601371765137,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1287,
      "step": 143
    },
    {
      "epoch": 0.485256950294861,
      "grad_norm": 2.5958619117736816,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.1741,
      "step": 144
    },
    {
      "epoch": 0.4886267902274642,
      "grad_norm": 2.314521312713623,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.0629,
      "step": 145
    },
    {
      "epoch": 0.4919966301600674,
      "grad_norm": 2.4749999046325684,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.0139,
      "step": 146
    },
    {
      "epoch": 0.4953664700926706,
      "grad_norm": 2.472071409225464,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.1252,
      "step": 147
    },
    {
      "epoch": 0.4987363100252738,
      "grad_norm": 2.380768299102783,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1219,
      "step": 148
    },
    {
      "epoch": 0.502106149957877,
      "grad_norm": 2.444328546524048,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.1542,
      "step": 149
    },
    {
      "epoch": 0.5054759898904801,
      "grad_norm": 2.586287260055542,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.066,
      "step": 150
    },
    {
      "epoch": 0.5054759898904801,
      "eval_loss": 0.287589430809021,
      "eval_runtime": 121.6025,
      "eval_samples_per_second": 4.112,
      "eval_steps_per_second": 1.028,
      "step": 150
    },
    {
      "epoch": 0.5088458298230834,
      "grad_norm": 2.356450080871582,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.4228,
      "step": 151
    },
    {
      "epoch": 0.5122156697556866,
      "grad_norm": 2.0100934505462646,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.217,
      "step": 152
    },
    {
      "epoch": 0.5155855096882898,
      "grad_norm": 1.7375375032424927,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.2351,
      "step": 153
    },
    {
      "epoch": 0.518955349620893,
      "grad_norm": 1.5888546705245972,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.0818,
      "step": 154
    },
    {
      "epoch": 0.5223251895534962,
      "grad_norm": 1.5656553506851196,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.1639,
      "step": 155
    },
    {
      "epoch": 0.5256950294860994,
      "grad_norm": 1.5815587043762207,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.1753,
      "step": 156
    },
    {
      "epoch": 0.5290648694187026,
      "grad_norm": 1.674997329711914,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.2684,
      "step": 157
    },
    {
      "epoch": 0.5324347093513058,
      "grad_norm": 1.6887853145599365,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.1651,
      "step": 158
    },
    {
      "epoch": 0.535804549283909,
      "grad_norm": 1.752082347869873,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.3349,
      "step": 159
    },
    {
      "epoch": 0.5391743892165122,
      "grad_norm": 1.9607328176498413,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.418,
      "step": 160
    },
    {
      "epoch": 0.5425442291491154,
      "grad_norm": 1.7222490310668945,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.1937,
      "step": 161
    },
    {
      "epoch": 0.5459140690817186,
      "grad_norm": 1.6560667753219604,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0829,
      "step": 162
    },
    {
      "epoch": 0.5492839090143218,
      "grad_norm": 1.6664199829101562,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.165,
      "step": 163
    },
    {
      "epoch": 0.552653748946925,
      "grad_norm": 1.543486475944519,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.0827,
      "step": 164
    },
    {
      "epoch": 0.5560235888795282,
      "grad_norm": 1.8627314567565918,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.2044,
      "step": 165
    },
    {
      "epoch": 0.5593934288121314,
      "grad_norm": 1.6388341188430786,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.9012,
      "step": 166
    },
    {
      "epoch": 0.5627632687447346,
      "grad_norm": 1.6322858333587646,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.8965,
      "step": 167
    },
    {
      "epoch": 0.5661331086773378,
      "grad_norm": 1.947608232498169,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.2191,
      "step": 168
    },
    {
      "epoch": 0.569502948609941,
      "grad_norm": 1.8948485851287842,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.323,
      "step": 169
    },
    {
      "epoch": 0.5728727885425442,
      "grad_norm": 2.0614988803863525,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.3929,
      "step": 170
    },
    {
      "epoch": 0.5762426284751474,
      "grad_norm": 1.884502649307251,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.1308,
      "step": 171
    },
    {
      "epoch": 0.5796124684077506,
      "grad_norm": 1.893863320350647,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.2453,
      "step": 172
    },
    {
      "epoch": 0.5829823083403538,
      "grad_norm": 1.896811842918396,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.2367,
      "step": 173
    },
    {
      "epoch": 0.586352148272957,
      "grad_norm": 1.8429944515228271,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0618,
      "step": 174
    },
    {
      "epoch": 0.5897219882055602,
      "grad_norm": 2.052502393722534,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.0696,
      "step": 175
    },
    {
      "epoch": 0.5930918281381634,
      "grad_norm": 1.6713838577270508,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.8484,
      "step": 176
    },
    {
      "epoch": 0.5964616680707666,
      "grad_norm": 1.938137173652649,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.0222,
      "step": 177
    },
    {
      "epoch": 0.5998315080033698,
      "grad_norm": 2.0545225143432617,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.2909,
      "step": 178
    },
    {
      "epoch": 0.603201347935973,
      "grad_norm": 1.8221138715744019,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.827,
      "step": 179
    },
    {
      "epoch": 0.6065711878685762,
      "grad_norm": 1.816070795059204,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.8751,
      "step": 180
    },
    {
      "epoch": 0.6099410278011794,
      "grad_norm": 1.9226304292678833,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.083,
      "step": 181
    },
    {
      "epoch": 0.6133108677337826,
      "grad_norm": 1.904410719871521,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.7344,
      "step": 182
    },
    {
      "epoch": 0.6166807076663858,
      "grad_norm": 2.142829418182373,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.1868,
      "step": 183
    },
    {
      "epoch": 0.620050547598989,
      "grad_norm": 1.7813894748687744,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.8987,
      "step": 184
    },
    {
      "epoch": 0.6234203875315922,
      "grad_norm": 2.0589540004730225,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.2215,
      "step": 185
    },
    {
      "epoch": 0.6267902274641954,
      "grad_norm": 2.139035940170288,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1805,
      "step": 186
    },
    {
      "epoch": 0.6301600673967986,
      "grad_norm": 1.9569933414459229,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.1102,
      "step": 187
    },
    {
      "epoch": 0.6335299073294018,
      "grad_norm": 2.062237501144409,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.9966,
      "step": 188
    },
    {
      "epoch": 0.636899747262005,
      "grad_norm": 2.160637617111206,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.2641,
      "step": 189
    },
    {
      "epoch": 0.6402695871946082,
      "grad_norm": 2.121643304824829,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.219,
      "step": 190
    },
    {
      "epoch": 0.6436394271272114,
      "grad_norm": 2.1960558891296387,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.2312,
      "step": 191
    },
    {
      "epoch": 0.6470092670598147,
      "grad_norm": 2.3870770931243896,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.3038,
      "step": 192
    },
    {
      "epoch": 0.6503791069924179,
      "grad_norm": 2.4442238807678223,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1942,
      "step": 193
    },
    {
      "epoch": 0.6537489469250211,
      "grad_norm": 2.187699556350708,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.0049,
      "step": 194
    },
    {
      "epoch": 0.6571187868576243,
      "grad_norm": 2.6958065032958984,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.8884,
      "step": 195
    },
    {
      "epoch": 0.6604886267902275,
      "grad_norm": 2.5944459438323975,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2507,
      "step": 196
    },
    {
      "epoch": 0.6638584667228307,
      "grad_norm": 2.352285385131836,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.101,
      "step": 197
    },
    {
      "epoch": 0.6672283066554339,
      "grad_norm": 2.5292348861694336,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.1908,
      "step": 198
    },
    {
      "epoch": 0.6705981465880371,
      "grad_norm": 2.3451972007751465,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.9805,
      "step": 199
    },
    {
      "epoch": 0.6739679865206403,
      "grad_norm": 2.6243679523468018,
      "learning_rate": 0.0,
      "loss": 1.0318,
      "step": 200
    },
    {
      "epoch": 0.6739679865206403,
      "eval_loss": 0.27505218982696533,
      "eval_runtime": 121.997,
      "eval_samples_per_second": 4.098,
      "eval_steps_per_second": 1.025,
      "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": 5.673907105636024e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}