{
  "best_metric": 0.14367078244686127,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3401360544217687,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0017006802721088435,
      "grad_norm": 0.5296435952186584,
      "learning_rate": 1e-05,
      "loss": 0.1901,
      "step": 1
    },
    {
      "epoch": 0.0017006802721088435,
      "eval_loss": 0.5936321020126343,
      "eval_runtime": 99.2183,
      "eval_samples_per_second": 9.978,
      "eval_steps_per_second": 2.5,
      "step": 1
    },
    {
      "epoch": 0.003401360544217687,
      "grad_norm": 0.46403637528419495,
      "learning_rate": 2e-05,
      "loss": 0.1959,
      "step": 2
    },
    {
      "epoch": 0.00510204081632653,
      "grad_norm": 0.5744128823280334,
      "learning_rate": 3e-05,
      "loss": 0.2195,
      "step": 3
    },
    {
      "epoch": 0.006802721088435374,
      "grad_norm": 0.5589671730995178,
      "learning_rate": 4e-05,
      "loss": 0.2317,
      "step": 4
    },
    {
      "epoch": 0.008503401360544218,
      "grad_norm": 0.5408557653427124,
      "learning_rate": 5e-05,
      "loss": 0.2098,
      "step": 5
    },
    {
      "epoch": 0.01020408163265306,
      "grad_norm": 0.514587938785553,
      "learning_rate": 6e-05,
      "loss": 0.2101,
      "step": 6
    },
    {
      "epoch": 0.011904761904761904,
      "grad_norm": 0.46519431471824646,
      "learning_rate": 7e-05,
      "loss": 0.1949,
      "step": 7
    },
    {
      "epoch": 0.013605442176870748,
      "grad_norm": 0.48039916157722473,
      "learning_rate": 8e-05,
      "loss": 0.2285,
      "step": 8
    },
    {
      "epoch": 0.015306122448979591,
      "grad_norm": 0.3992774188518524,
      "learning_rate": 9e-05,
      "loss": 0.1829,
      "step": 9
    },
    {
      "epoch": 0.017006802721088437,
      "grad_norm": 0.37706080079078674,
      "learning_rate": 0.0001,
      "loss": 0.1575,
      "step": 10
    },
    {
      "epoch": 0.01870748299319728,
      "grad_norm": 0.33965542912483215,
      "learning_rate": 9.999316524962345e-05,
      "loss": 0.184,
      "step": 11
    },
    {
      "epoch": 0.02040816326530612,
      "grad_norm": 0.3842385709285736,
      "learning_rate": 9.997266286704631e-05,
      "loss": 0.1588,
      "step": 12
    },
    {
      "epoch": 0.022108843537414966,
      "grad_norm": 0.5256059169769287,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.2427,
      "step": 13
    },
    {
      "epoch": 0.023809523809523808,
      "grad_norm": 0.5123369693756104,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.1808,
      "step": 14
    },
    {
      "epoch": 0.025510204081632654,
      "grad_norm": 0.42671579122543335,
      "learning_rate": 9.98292246503335e-05,
      "loss": 0.1534,
      "step": 15
    },
    {
      "epoch": 0.027210884353741496,
      "grad_norm": 0.362161785364151,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.1293,
      "step": 16
    },
    {
      "epoch": 0.02891156462585034,
      "grad_norm": 0.4058263301849365,
      "learning_rate": 9.966546331768191e-05,
      "loss": 0.1596,
      "step": 17
    },
    {
      "epoch": 0.030612244897959183,
      "grad_norm": 0.36066436767578125,
      "learning_rate": 9.956320346634876e-05,
      "loss": 0.1386,
      "step": 18
    },
    {
      "epoch": 0.03231292517006803,
      "grad_norm": 0.34079501032829285,
      "learning_rate": 9.944739353007344e-05,
      "loss": 0.1577,
      "step": 19
    },
    {
      "epoch": 0.034013605442176874,
      "grad_norm": 0.34605899453163147,
      "learning_rate": 9.931806517013612e-05,
      "loss": 0.1488,
      "step": 20
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 0.29584625363349915,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.1254,
      "step": 21
    },
    {
      "epoch": 0.03741496598639456,
      "grad_norm": 0.3019483983516693,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.1597,
      "step": 22
    },
    {
      "epoch": 0.0391156462585034,
      "grad_norm": 0.347933828830719,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.1477,
      "step": 23
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 0.35964134335517883,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.162,
      "step": 24
    },
    {
      "epoch": 0.04251700680272109,
      "grad_norm": 0.3269605338573456,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.1221,
      "step": 25
    },
    {
      "epoch": 0.04421768707482993,
      "grad_norm": 0.39577123522758484,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.16,
      "step": 26
    },
    {
      "epoch": 0.04591836734693878,
      "grad_norm": 1.1477911472320557,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.2382,
      "step": 27
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 0.40509581565856934,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.1499,
      "step": 28
    },
    {
      "epoch": 0.04931972789115646,
      "grad_norm": 0.3921804428100586,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.1597,
      "step": 29
    },
    {
      "epoch": 0.05102040816326531,
      "grad_norm": 0.3694240152835846,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.1531,
      "step": 30
    },
    {
      "epoch": 0.05272108843537415,
      "grad_norm": 0.48857390880584717,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.1802,
      "step": 31
    },
    {
      "epoch": 0.05442176870748299,
      "grad_norm": 0.518921434879303,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.2015,
      "step": 32
    },
    {
      "epoch": 0.05612244897959184,
      "grad_norm": 0.45379164814949036,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.1666,
      "step": 33
    },
    {
      "epoch": 0.05782312925170068,
      "grad_norm": 0.6328680515289307,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.314,
      "step": 34
    },
    {
      "epoch": 0.05952380952380952,
      "grad_norm": 0.41080063581466675,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.1411,
      "step": 35
    },
    {
      "epoch": 0.061224489795918366,
      "grad_norm": 0.47048354148864746,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.1458,
      "step": 36
    },
    {
      "epoch": 0.06292517006802721,
      "grad_norm": 0.4776947796344757,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.2109,
      "step": 37
    },
    {
      "epoch": 0.06462585034013606,
      "grad_norm": 0.5354244709014893,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.1569,
      "step": 38
    },
    {
      "epoch": 0.0663265306122449,
      "grad_norm": 0.4933754801750183,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.1632,
      "step": 39
    },
    {
      "epoch": 0.06802721088435375,
      "grad_norm": 0.526614248752594,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.2226,
      "step": 40
    },
    {
      "epoch": 0.06972789115646258,
      "grad_norm": 0.6038437485694885,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.2194,
      "step": 41
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 0.5840771198272705,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.2245,
      "step": 42
    },
    {
      "epoch": 0.07312925170068027,
      "grad_norm": 0.5161733031272888,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.1846,
      "step": 43
    },
    {
      "epoch": 0.07482993197278912,
      "grad_norm": 0.5809618830680847,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.2079,
      "step": 44
    },
    {
      "epoch": 0.07653061224489796,
      "grad_norm": 0.5962471961975098,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.2366,
      "step": 45
    },
    {
      "epoch": 0.0782312925170068,
      "grad_norm": 0.6618085503578186,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.2596,
      "step": 46
    },
    {
      "epoch": 0.07993197278911565,
      "grad_norm": 0.5380584597587585,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.2301,
      "step": 47
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 0.631349503993988,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.2292,
      "step": 48
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 1.3007142543792725,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.2722,
      "step": 49
    },
    {
      "epoch": 0.08503401360544217,
      "grad_norm": 2.0485315322875977,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.3315,
      "step": 50
    },
    {
      "epoch": 0.08503401360544217,
      "eval_loss": 0.17823144793510437,
      "eval_runtime": 100.0032,
      "eval_samples_per_second": 9.9,
      "eval_steps_per_second": 2.48,
      "step": 50
    },
    {
      "epoch": 0.08673469387755102,
      "grad_norm": 0.2999962270259857,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.1368,
      "step": 51
    },
    {
      "epoch": 0.08843537414965986,
      "grad_norm": 0.30443039536476135,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.0848,
      "step": 52
    },
    {
      "epoch": 0.09013605442176871,
      "grad_norm": 0.25375548005104065,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.1295,
      "step": 53
    },
    {
      "epoch": 0.09183673469387756,
      "grad_norm": 0.17752642929553986,
      "learning_rate": 8.73410738492077e-05,
      "loss": 0.0672,
      "step": 54
    },
    {
      "epoch": 0.0935374149659864,
      "grad_norm": 0.18084105849266052,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.1213,
      "step": 55
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.26786619424819946,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.1161,
      "step": 56
    },
    {
      "epoch": 0.09693877551020408,
      "grad_norm": 0.22582994401454926,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.1128,
      "step": 57
    },
    {
      "epoch": 0.09863945578231292,
      "grad_norm": 0.2891770899295807,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.139,
      "step": 58
    },
    {
      "epoch": 0.10034013605442177,
      "grad_norm": 0.2367025464773178,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.1246,
      "step": 59
    },
    {
      "epoch": 0.10204081632653061,
      "grad_norm": 0.18042106926441193,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.1039,
      "step": 60
    },
    {
      "epoch": 0.10374149659863946,
      "grad_norm": 0.19133102893829346,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.1255,
      "step": 61
    },
    {
      "epoch": 0.1054421768707483,
      "grad_norm": 0.2340659350156784,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.1529,
      "step": 62
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 0.2856714725494385,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.1337,
      "step": 63
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 0.24162790179252625,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.119,
      "step": 64
    },
    {
      "epoch": 0.11054421768707483,
      "grad_norm": 0.19808998703956604,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.1318,
      "step": 65
    },
    {
      "epoch": 0.11224489795918367,
      "grad_norm": 0.2100956290960312,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.1239,
      "step": 66
    },
    {
      "epoch": 0.11394557823129252,
      "grad_norm": 0.20403148233890533,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.1008,
      "step": 67
    },
    {
      "epoch": 0.11564625850340136,
      "grad_norm": 0.2087327241897583,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.1288,
      "step": 68
    },
    {
      "epoch": 0.11734693877551021,
      "grad_norm": 0.20853887498378754,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.1189,
      "step": 69
    },
    {
      "epoch": 0.11904761904761904,
      "grad_norm": 0.261677086353302,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.1297,
      "step": 70
    },
    {
      "epoch": 0.12074829931972789,
      "grad_norm": 0.25025635957717896,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.1426,
      "step": 71
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 0.2275075763463974,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.1177,
      "step": 72
    },
    {
      "epoch": 0.12414965986394558,
      "grad_norm": 0.19386127591133118,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.1212,
      "step": 73
    },
    {
      "epoch": 0.12585034013605442,
      "grad_norm": 0.2676469683647156,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.1361,
      "step": 74
    },
    {
      "epoch": 0.12755102040816327,
      "grad_norm": 0.4179306924343109,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.1405,
      "step": 75
    },
    {
      "epoch": 0.1292517006802721,
      "grad_norm": 0.285391241312027,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.1426,
      "step": 76
    },
    {
      "epoch": 0.13095238095238096,
      "grad_norm": 0.37067097425460815,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.125,
      "step": 77
    },
    {
      "epoch": 0.1326530612244898,
      "grad_norm": 0.27059873938560486,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.1446,
      "step": 78
    },
    {
      "epoch": 0.13435374149659865,
      "grad_norm": 0.32261890172958374,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.1377,
      "step": 79
    },
    {
      "epoch": 0.1360544217687075,
      "grad_norm": 0.4681248068809509,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.2576,
      "step": 80
    },
    {
      "epoch": 0.1377551020408163,
      "grad_norm": 0.39090844988822937,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.2131,
      "step": 81
    },
    {
      "epoch": 0.13945578231292516,
      "grad_norm": 0.2745368778705597,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.1218,
      "step": 82
    },
    {
      "epoch": 0.141156462585034,
      "grad_norm": 0.4222804009914398,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.2313,
      "step": 83
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.26606062054634094,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.1231,
      "step": 84
    },
    {
      "epoch": 0.1445578231292517,
      "grad_norm": 0.39861971139907837,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.1935,
      "step": 85
    },
    {
      "epoch": 0.14625850340136054,
      "grad_norm": 0.34635743498802185,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.1142,
      "step": 86
    },
    {
      "epoch": 0.14795918367346939,
      "grad_norm": 0.4157678186893463,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.188,
      "step": 87
    },
    {
      "epoch": 0.14965986394557823,
      "grad_norm": 0.3788713216781616,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.1562,
      "step": 88
    },
    {
      "epoch": 0.15136054421768708,
      "grad_norm": 0.3447698950767517,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.1649,
      "step": 89
    },
    {
      "epoch": 0.15306122448979592,
      "grad_norm": 0.40397417545318604,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.1885,
      "step": 90
    },
    {
      "epoch": 0.15476190476190477,
      "grad_norm": 0.44152534008026123,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.2558,
      "step": 91
    },
    {
      "epoch": 0.1564625850340136,
      "grad_norm": 0.35763365030288696,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.1404,
      "step": 92
    },
    {
      "epoch": 0.15816326530612246,
      "grad_norm": 0.39760518074035645,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.1549,
      "step": 93
    },
    {
      "epoch": 0.1598639455782313,
      "grad_norm": 0.5707307457923889,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.3187,
      "step": 94
    },
    {
      "epoch": 0.16156462585034015,
      "grad_norm": 0.5781694054603577,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.208,
      "step": 95
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 0.44152382016181946,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.174,
      "step": 96
    },
    {
      "epoch": 0.1649659863945578,
      "grad_norm": 0.5395876169204712,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.1957,
      "step": 97
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.6618689894676208,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.2403,
      "step": 98
    },
    {
      "epoch": 0.1683673469387755,
      "grad_norm": 0.7297075390815735,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.2676,
      "step": 99
    },
    {
      "epoch": 0.17006802721088435,
      "grad_norm": 1.3715710639953613,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.2514,
      "step": 100
    },
    {
      "epoch": 0.17006802721088435,
      "eval_loss": 0.15225209295749664,
      "eval_runtime": 100.0235,
      "eval_samples_per_second": 9.898,
      "eval_steps_per_second": 2.479,
      "step": 100
    },
    {
      "epoch": 0.1717687074829932,
      "grad_norm": 0.13049182295799255,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.1053,
      "step": 101
    },
    {
      "epoch": 0.17346938775510204,
      "grad_norm": 0.1473456174135208,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.1182,
      "step": 102
    },
    {
      "epoch": 0.17517006802721088,
      "grad_norm": 0.1675950139760971,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.1675,
      "step": 103
    },
    {
      "epoch": 0.17687074829931973,
      "grad_norm": 0.14025190472602844,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.0941,
      "step": 104
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 0.15604698657989502,
      "learning_rate": 5e-05,
      "loss": 0.1037,
      "step": 105
    },
    {
      "epoch": 0.18027210884353742,
      "grad_norm": 0.15615573525428772,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.1098,
      "step": 106
    },
    {
      "epoch": 0.18197278911564627,
      "grad_norm": 0.13778981566429138,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.0907,
      "step": 107
    },
    {
      "epoch": 0.1836734693877551,
      "grad_norm": 0.13400906324386597,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.0833,
      "step": 108
    },
    {
      "epoch": 0.18537414965986396,
      "grad_norm": 0.17634348571300507,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.1152,
      "step": 109
    },
    {
      "epoch": 0.1870748299319728,
      "grad_norm": 0.13390770554542542,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.092,
      "step": 110
    },
    {
      "epoch": 0.18877551020408162,
      "grad_norm": 0.1868814378976822,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.1241,
      "step": 111
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.18534260988235474,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.1185,
      "step": 112
    },
    {
      "epoch": 0.1921768707482993,
      "grad_norm": 0.21472658216953278,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.1177,
      "step": 113
    },
    {
      "epoch": 0.19387755102040816,
      "grad_norm": 0.23310808837413788,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.1606,
      "step": 114
    },
    {
      "epoch": 0.195578231292517,
      "grad_norm": 0.17960843443870544,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.1139,
      "step": 115
    },
    {
      "epoch": 0.19727891156462585,
      "grad_norm": 0.2276669293642044,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.1171,
      "step": 116
    },
    {
      "epoch": 0.1989795918367347,
      "grad_norm": 0.19684050977230072,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.1292,
      "step": 117
    },
    {
      "epoch": 0.20068027210884354,
      "grad_norm": 0.19612234830856323,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.1155,
      "step": 118
    },
    {
      "epoch": 0.20238095238095238,
      "grad_norm": 0.21086162328720093,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.1058,
      "step": 119
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 0.23637105524539948,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.1429,
      "step": 120
    },
    {
      "epoch": 0.20578231292517007,
      "grad_norm": 0.28456252813339233,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.1784,
      "step": 121
    },
    {
      "epoch": 0.20748299319727892,
      "grad_norm": 0.2100301831960678,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.1132,
      "step": 122
    },
    {
      "epoch": 0.20918367346938777,
      "grad_norm": 0.25739890336990356,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.1655,
      "step": 123
    },
    {
      "epoch": 0.2108843537414966,
      "grad_norm": 0.2325243353843689,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.1201,
      "step": 124
    },
    {
      "epoch": 0.21258503401360543,
      "grad_norm": 0.31694191694259644,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.2064,
      "step": 125
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.2747313380241394,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.1295,
      "step": 126
    },
    {
      "epoch": 0.21598639455782312,
      "grad_norm": 0.3434375524520874,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.2123,
      "step": 127
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 0.24401719868183136,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.1366,
      "step": 128
    },
    {
      "epoch": 0.2193877551020408,
      "grad_norm": 0.3210381865501404,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.1794,
      "step": 129
    },
    {
      "epoch": 0.22108843537414966,
      "grad_norm": 0.28831300139427185,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.1572,
      "step": 130
    },
    {
      "epoch": 0.2227891156462585,
      "grad_norm": 0.4137888252735138,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.2545,
      "step": 131
    },
    {
      "epoch": 0.22448979591836735,
      "grad_norm": 0.28466734290122986,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.1587,
      "step": 132
    },
    {
      "epoch": 0.2261904761904762,
      "grad_norm": 0.2973605692386627,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.1551,
      "step": 133
    },
    {
      "epoch": 0.22789115646258504,
      "grad_norm": 0.2863537073135376,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.1244,
      "step": 134
    },
    {
      "epoch": 0.22959183673469388,
      "grad_norm": 0.3383525609970093,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.1681,
      "step": 135
    },
    {
      "epoch": 0.23129251700680273,
      "grad_norm": 0.31313425302505493,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.1551,
      "step": 136
    },
    {
      "epoch": 0.23299319727891157,
      "grad_norm": 0.3133527338504791,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.1276,
      "step": 137
    },
    {
      "epoch": 0.23469387755102042,
      "grad_norm": 0.3251171112060547,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.1457,
      "step": 138
    },
    {
      "epoch": 0.23639455782312926,
      "grad_norm": 0.32641586661338806,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.1519,
      "step": 139
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.4531185030937195,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.2294,
      "step": 140
    },
    {
      "epoch": 0.23979591836734693,
      "grad_norm": 0.3772891163825989,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.1405,
      "step": 141
    },
    {
      "epoch": 0.24149659863945577,
      "grad_norm": 0.37547656893730164,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.1491,
      "step": 142
    },
    {
      "epoch": 0.24319727891156462,
      "grad_norm": 0.47029197216033936,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.2363,
      "step": 143
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 0.5126447677612305,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.3044,
      "step": 144
    },
    {
      "epoch": 0.2465986394557823,
      "grad_norm": 0.44240790605545044,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.1866,
      "step": 145
    },
    {
      "epoch": 0.24829931972789115,
      "grad_norm": 0.5072203278541565,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.1994,
      "step": 146
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5118880867958069,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.2285,
      "step": 147
    },
    {
      "epoch": 0.25170068027210885,
      "grad_norm": 0.5532385110855103,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.1671,
      "step": 148
    },
    {
      "epoch": 0.2534013605442177,
      "grad_norm": 0.8175846338272095,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.1871,
      "step": 149
    },
    {
      "epoch": 0.25510204081632654,
      "grad_norm": 1.3380439281463623,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.2504,
      "step": 150
    },
    {
      "epoch": 0.25510204081632654,
      "eval_loss": 0.1447044014930725,
      "eval_runtime": 99.6548,
      "eval_samples_per_second": 9.934,
      "eval_steps_per_second": 2.489,
      "step": 150
    },
    {
      "epoch": 0.2568027210884354,
      "grad_norm": 0.12096025794744492,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.0891,
      "step": 151
    },
    {
      "epoch": 0.2585034013605442,
      "grad_norm": 0.13122454285621643,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.1108,
      "step": 152
    },
    {
      "epoch": 0.2602040816326531,
      "grad_norm": 0.14695559442043304,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.1426,
      "step": 153
    },
    {
      "epoch": 0.2619047619047619,
      "grad_norm": 0.12018829584121704,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.0709,
      "step": 154
    },
    {
      "epoch": 0.26360544217687076,
      "grad_norm": 0.13240677118301392,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.0956,
      "step": 155
    },
    {
      "epoch": 0.2653061224489796,
      "grad_norm": 0.1313542127609253,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.0723,
      "step": 156
    },
    {
      "epoch": 0.26700680272108845,
      "grad_norm": 0.1426219940185547,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.096,
      "step": 157
    },
    {
      "epoch": 0.2687074829931973,
      "grad_norm": 0.14496323466300964,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.1016,
      "step": 158
    },
    {
      "epoch": 0.27040816326530615,
      "grad_norm": 0.1569526195526123,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.1185,
      "step": 159
    },
    {
      "epoch": 0.272108843537415,
      "grad_norm": 0.15756379067897797,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.1106,
      "step": 160
    },
    {
      "epoch": 0.27380952380952384,
      "grad_norm": 0.1848234236240387,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.1734,
      "step": 161
    },
    {
      "epoch": 0.2755102040816326,
      "grad_norm": 0.17146940529346466,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.1256,
      "step": 162
    },
    {
      "epoch": 0.27721088435374147,
      "grad_norm": 0.18103595077991486,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.128,
      "step": 163
    },
    {
      "epoch": 0.2789115646258503,
      "grad_norm": 0.16318972408771515,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.0987,
      "step": 164
    },
    {
      "epoch": 0.28061224489795916,
      "grad_norm": 0.15253207087516785,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.1011,
      "step": 165
    },
    {
      "epoch": 0.282312925170068,
      "grad_norm": 0.1747894287109375,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.107,
      "step": 166
    },
    {
      "epoch": 0.28401360544217685,
      "grad_norm": 0.18291305005550385,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.1206,
      "step": 167
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.21150052547454834,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.1405,
      "step": 168
    },
    {
      "epoch": 0.28741496598639454,
      "grad_norm": 0.1849411427974701,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.1197,
      "step": 169
    },
    {
      "epoch": 0.2891156462585034,
      "grad_norm": 0.1709270477294922,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.1049,
      "step": 170
    },
    {
      "epoch": 0.29081632653061223,
      "grad_norm": 0.2308603972196579,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.1681,
      "step": 171
    },
    {
      "epoch": 0.2925170068027211,
      "grad_norm": 0.17844510078430176,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.1144,
      "step": 172
    },
    {
      "epoch": 0.2942176870748299,
      "grad_norm": 0.20301955938339233,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.1168,
      "step": 173
    },
    {
      "epoch": 0.29591836734693877,
      "grad_norm": 0.22025449573993683,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.1379,
      "step": 174
    },
    {
      "epoch": 0.2976190476190476,
      "grad_norm": 0.22694650292396545,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.117,
      "step": 175
    },
    {
      "epoch": 0.29931972789115646,
      "grad_norm": 0.2415693700313568,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.1377,
      "step": 176
    },
    {
      "epoch": 0.3010204081632653,
      "grad_norm": 0.26638922095298767,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.1569,
      "step": 177
    },
    {
      "epoch": 0.30272108843537415,
      "grad_norm": 0.2395794540643692,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.1355,
      "step": 178
    },
    {
      "epoch": 0.304421768707483,
      "grad_norm": 0.30046385526657104,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.1803,
      "step": 179
    },
    {
      "epoch": 0.30612244897959184,
      "grad_norm": 0.29656267166137695,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.1796,
      "step": 180
    },
    {
      "epoch": 0.3078231292517007,
      "grad_norm": 0.2734433710575104,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.1635,
      "step": 181
    },
    {
      "epoch": 0.30952380952380953,
      "grad_norm": 0.31922054290771484,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.206,
      "step": 182
    },
    {
      "epoch": 0.3112244897959184,
      "grad_norm": 0.3922654092311859,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.2243,
      "step": 183
    },
    {
      "epoch": 0.3129251700680272,
      "grad_norm": 0.3273572623729706,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.1757,
      "step": 184
    },
    {
      "epoch": 0.31462585034013607,
      "grad_norm": 0.4013892412185669,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.1965,
      "step": 185
    },
    {
      "epoch": 0.3163265306122449,
      "grad_norm": 0.3162621557712555,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.1546,
      "step": 186
    },
    {
      "epoch": 0.31802721088435376,
      "grad_norm": 0.3780801594257355,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.2012,
      "step": 187
    },
    {
      "epoch": 0.3197278911564626,
      "grad_norm": 0.32307010889053345,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.152,
      "step": 188
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 0.3758290708065033,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.1879,
      "step": 189
    },
    {
      "epoch": 0.3231292517006803,
      "grad_norm": 0.36550796031951904,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.1701,
      "step": 190
    },
    {
      "epoch": 0.32482993197278914,
      "grad_norm": 0.41870972514152527,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2054,
      "step": 191
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 0.3498254120349884,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.1232,
      "step": 192
    },
    {
      "epoch": 0.3282312925170068,
      "grad_norm": 0.45467662811279297,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.2094,
      "step": 193
    },
    {
      "epoch": 0.3299319727891156,
      "grad_norm": 0.5509414672851562,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.2777,
      "step": 194
    },
    {
      "epoch": 0.33163265306122447,
      "grad_norm": 0.4783499538898468,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.1986,
      "step": 195
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.5205704569816589,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.233,
      "step": 196
    },
    {
      "epoch": 0.33503401360544216,
      "grad_norm": 0.7216671109199524,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.3241,
      "step": 197
    },
    {
      "epoch": 0.336734693877551,
      "grad_norm": 0.6502580642700195,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.2387,
      "step": 198
    },
    {
      "epoch": 0.33843537414965985,
      "grad_norm": 0.9961100816726685,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.2379,
      "step": 199
    },
    {
      "epoch": 0.3401360544217687,
      "grad_norm": 1.435985803604126,
      "learning_rate": 0.0,
      "loss": 0.2531,
      "step": 200
    },
    {
      "epoch": 0.3401360544217687,
      "eval_loss": 0.14367078244686127,
      "eval_runtime": 99.9926,
      "eval_samples_per_second": 9.901,
      "eval_steps_per_second": 2.48,
      "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": 3.683300708985078e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}