{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 29.948867786705623,
  "global_step": 164000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.18,
      "learning_rate": 5e-05,
      "loss": 7.0207,
      "step": 1000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.96937775600196e-05,
      "loss": 5.1703,
      "step": 2000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.93875551200392e-05,
      "loss": 4.7048,
      "step": 3000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.9081332680058794e-05,
      "loss": 4.3894,
      "step": 4000
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.87751102400784e-05,
      "loss": 4.1348,
      "step": 5000
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.8468887800097995e-05,
      "loss": 3.9447,
      "step": 6000
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.816266536011759e-05,
      "loss": 3.8286,
      "step": 7000
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.785644292013719e-05,
      "loss": 3.7462,
      "step": 8000
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.7550220480156786e-05,
      "loss": 3.693,
      "step": 9000
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.7243998040176384e-05,
      "loss": 3.6311,
      "step": 10000
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.693777560019598e-05,
      "loss": 3.5843,
      "step": 11000
    },
    {
      "epoch": 2.19,
      "learning_rate": 4.6631553160215585e-05,
      "loss": 3.4624,
      "step": 12000
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.632533072023518e-05,
      "loss": 3.4432,
      "step": 13000
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.601910828025478e-05,
      "loss": 3.423,
      "step": 14000
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.5712885840274376e-05,
      "loss": 3.4042,
      "step": 15000
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.540666340029397e-05,
      "loss": 3.3902,
      "step": 16000
    },
    {
      "epoch": 3.1,
      "learning_rate": 4.510044096031357e-05,
      "loss": 3.2973,
      "step": 17000
    },
    {
      "epoch": 3.29,
      "learning_rate": 4.479421852033317e-05,
      "loss": 3.2482,
      "step": 18000
    },
    {
      "epoch": 3.47,
      "learning_rate": 4.448799608035277e-05,
      "loss": 3.2427,
      "step": 19000
    },
    {
      "epoch": 3.65,
      "learning_rate": 4.418177364037237e-05,
      "loss": 3.2309,
      "step": 20000
    },
    {
      "epoch": 3.83,
      "learning_rate": 4.3875551200391966e-05,
      "loss": 3.2193,
      "step": 21000
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.356932876041156e-05,
      "loss": 3.2065,
      "step": 22000
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.326310632043116e-05,
      "loss": 3.0733,
      "step": 23000
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.295688388045076e-05,
      "loss": 3.0868,
      "step": 24000
    },
    {
      "epoch": 4.57,
      "learning_rate": 4.2650661440470355e-05,
      "loss": 3.0884,
      "step": 25000
    },
    {
      "epoch": 4.75,
      "learning_rate": 4.234443900048996e-05,
      "loss": 3.0843,
      "step": 26000
    },
    {
      "epoch": 4.93,
      "learning_rate": 4.2038216560509556e-05,
      "loss": 3.0768,
      "step": 27000
    },
    {
      "epoch": 5.11,
      "learning_rate": 4.173199412052915e-05,
      "loss": 2.9834,
      "step": 28000
    },
    {
      "epoch": 5.3,
      "learning_rate": 4.142577168054875e-05,
      "loss": 2.9267,
      "step": 29000
    },
    {
      "epoch": 5.48,
      "learning_rate": 4.111954924056835e-05,
      "loss": 2.9463,
      "step": 30000
    },
    {
      "epoch": 5.66,
      "learning_rate": 4.081332680058795e-05,
      "loss": 2.9523,
      "step": 31000
    },
    {
      "epoch": 5.84,
      "learning_rate": 4.050710436060755e-05,
      "loss": 2.9561,
      "step": 32000
    },
    {
      "epoch": 6.03,
      "learning_rate": 4.0200881920627145e-05,
      "loss": 2.9237,
      "step": 33000
    },
    {
      "epoch": 6.21,
      "learning_rate": 3.989465948064674e-05,
      "loss": 2.7803,
      "step": 34000
    },
    {
      "epoch": 6.39,
      "learning_rate": 3.958843704066634e-05,
      "loss": 2.8068,
      "step": 35000
    },
    {
      "epoch": 6.57,
      "learning_rate": 3.9282214600685944e-05,
      "loss": 2.8223,
      "step": 36000
    },
    {
      "epoch": 6.76,
      "learning_rate": 3.897599216070554e-05,
      "loss": 2.8238,
      "step": 37000
    },
    {
      "epoch": 6.94,
      "learning_rate": 3.866976972072514e-05,
      "loss": 2.8341,
      "step": 38000
    },
    {
      "epoch": 7.12,
      "learning_rate": 3.8363547280744735e-05,
      "loss": 2.6976,
      "step": 39000
    },
    {
      "epoch": 7.3,
      "learning_rate": 3.805732484076434e-05,
      "loss": 2.6679,
      "step": 40000
    },
    {
      "epoch": 7.49,
      "learning_rate": 3.7751102400783936e-05,
      "loss": 2.691,
      "step": 41000
    },
    {
      "epoch": 7.67,
      "learning_rate": 3.744487996080353e-05,
      "loss": 2.7001,
      "step": 42000
    },
    {
      "epoch": 7.85,
      "learning_rate": 3.713865752082313e-05,
      "loss": 2.7189,
      "step": 43000
    },
    {
      "epoch": 8.04,
      "learning_rate": 3.683243508084273e-05,
      "loss": 2.6829,
      "step": 44000
    },
    {
      "epoch": 8.22,
      "learning_rate": 3.6526212640862325e-05,
      "loss": 2.5262,
      "step": 45000
    },
    {
      "epoch": 8.4,
      "learning_rate": 3.621999020088192e-05,
      "loss": 2.5547,
      "step": 46000
    },
    {
      "epoch": 8.58,
      "learning_rate": 3.5913767760901526e-05,
      "loss": 2.5926,
      "step": 47000
    },
    {
      "epoch": 8.77,
      "learning_rate": 3.560754532092112e-05,
      "loss": 2.5971,
      "step": 48000
    },
    {
      "epoch": 8.95,
      "learning_rate": 3.530132288094072e-05,
      "loss": 2.6053,
      "step": 49000
    },
    {
      "epoch": 9.13,
      "learning_rate": 3.499510044096032e-05,
      "loss": 2.4615,
      "step": 50000
    },
    {
      "epoch": 9.31,
      "learning_rate": 3.4688878000979914e-05,
      "loss": 2.4412,
      "step": 51000
    },
    {
      "epoch": 9.5,
      "learning_rate": 3.438265556099951e-05,
      "loss": 2.4633,
      "step": 52000
    },
    {
      "epoch": 9.68,
      "learning_rate": 3.407643312101911e-05,
      "loss": 2.469,
      "step": 53000
    },
    {
      "epoch": 9.86,
      "learning_rate": 3.377021068103871e-05,
      "loss": 2.5006,
      "step": 54000
    },
    {
      "epoch": 10.04,
      "learning_rate": 3.346398824105831e-05,
      "loss": 2.4556,
      "step": 55000
    },
    {
      "epoch": 10.23,
      "learning_rate": 3.315776580107791e-05,
      "loss": 2.3134,
      "step": 56000
    },
    {
      "epoch": 10.41,
      "learning_rate": 3.2851543361097504e-05,
      "loss": 2.3445,
      "step": 57000
    },
    {
      "epoch": 10.59,
      "learning_rate": 3.25453209211171e-05,
      "loss": 2.3709,
      "step": 58000
    },
    {
      "epoch": 10.77,
      "learning_rate": 3.22390984811367e-05,
      "loss": 2.3831,
      "step": 59000
    },
    {
      "epoch": 10.96,
      "learning_rate": 3.1932876041156296e-05,
      "loss": 2.4089,
      "step": 60000
    },
    {
      "epoch": 11.14,
      "learning_rate": 3.162665360117589e-05,
      "loss": 2.248,
      "step": 61000
    },
    {
      "epoch": 11.32,
      "learning_rate": 3.13204311611955e-05,
      "loss": 2.2344,
      "step": 62000
    },
    {
      "epoch": 11.5,
      "learning_rate": 3.1014208721215094e-05,
      "loss": 2.2553,
      "step": 63000
    },
    {
      "epoch": 11.69,
      "learning_rate": 3.070798628123469e-05,
      "loss": 2.2816,
      "step": 64000
    },
    {
      "epoch": 11.87,
      "learning_rate": 3.0401763841254288e-05,
      "loss": 2.3067,
      "step": 65000
    },
    {
      "epoch": 12.05,
      "learning_rate": 3.0095541401273885e-05,
      "loss": 2.248,
      "step": 66000
    },
    {
      "epoch": 12.24,
      "learning_rate": 2.9789318961293483e-05,
      "loss": 2.1257,
      "step": 67000
    },
    {
      "epoch": 12.42,
      "learning_rate": 2.948309652131308e-05,
      "loss": 2.1559,
      "step": 68000
    },
    {
      "epoch": 12.6,
      "learning_rate": 2.9176874081332684e-05,
      "loss": 2.1743,
      "step": 69000
    },
    {
      "epoch": 12.78,
      "learning_rate": 2.887065164135228e-05,
      "loss": 2.1997,
      "step": 70000
    },
    {
      "epoch": 12.97,
      "learning_rate": 2.8564429201371878e-05,
      "loss": 2.2321,
      "step": 71000
    },
    {
      "epoch": 13.15,
      "learning_rate": 2.8258206761391475e-05,
      "loss": 2.0603,
      "step": 72000
    },
    {
      "epoch": 13.33,
      "learning_rate": 2.7951984321411072e-05,
      "loss": 2.0491,
      "step": 73000
    },
    {
      "epoch": 13.51,
      "learning_rate": 2.764576188143067e-05,
      "loss": 2.079,
      "step": 74000
    },
    {
      "epoch": 13.7,
      "learning_rate": 2.7339539441450267e-05,
      "loss": 2.1119,
      "step": 75000
    },
    {
      "epoch": 13.88,
      "learning_rate": 2.703331700146987e-05,
      "loss": 2.1355,
      "step": 76000
    },
    {
      "epoch": 14.06,
      "learning_rate": 2.6727094561489468e-05,
      "loss": 2.068,
      "step": 77000
    },
    {
      "epoch": 14.24,
      "learning_rate": 2.6420872121509065e-05,
      "loss": 1.96,
      "step": 78000
    },
    {
      "epoch": 14.43,
      "learning_rate": 2.6114649681528662e-05,
      "loss": 1.9827,
      "step": 79000
    },
    {
      "epoch": 14.61,
      "learning_rate": 2.5808427241548263e-05,
      "loss": 2.0163,
      "step": 80000
    },
    {
      "epoch": 14.79,
      "learning_rate": 2.550220480156786e-05,
      "loss": 2.0496,
      "step": 81000
    },
    {
      "epoch": 14.97,
      "learning_rate": 2.5195982361587457e-05,
      "loss": 2.0579,
      "step": 82000
    },
    {
      "epoch": 15.16,
      "learning_rate": 2.4889759921607057e-05,
      "loss": 1.8884,
      "step": 83000
    },
    {
      "epoch": 15.34,
      "learning_rate": 2.4583537481626655e-05,
      "loss": 1.8992,
      "step": 84000
    },
    {
      "epoch": 15.52,
      "learning_rate": 2.4277315041646255e-05,
      "loss": 1.9383,
      "step": 85000
    },
    {
      "epoch": 15.7,
      "learning_rate": 2.3971092601665852e-05,
      "loss": 1.9512,
      "step": 86000
    },
    {
      "epoch": 15.89,
      "learning_rate": 2.366487016168545e-05,
      "loss": 1.9814,
      "step": 87000
    },
    {
      "epoch": 16.07,
      "learning_rate": 2.335864772170505e-05,
      "loss": 1.9075,
      "step": 88000
    },
    {
      "epoch": 16.25,
      "learning_rate": 2.3052425281724647e-05,
      "loss": 1.8141,
      "step": 89000
    },
    {
      "epoch": 16.44,
      "learning_rate": 2.2746202841744244e-05,
      "loss": 1.8431,
      "step": 90000
    },
    {
      "epoch": 16.62,
      "learning_rate": 2.243998040176384e-05,
      "loss": 1.8723,
      "step": 91000
    },
    {
      "epoch": 16.8,
      "learning_rate": 2.2133757961783442e-05,
      "loss": 1.9034,
      "step": 92000
    },
    {
      "epoch": 16.98,
      "learning_rate": 2.182753552180304e-05,
      "loss": 1.9112,
      "step": 93000
    },
    {
      "epoch": 17.17,
      "learning_rate": 2.1521313081822636e-05,
      "loss": 1.7565,
      "step": 94000
    },
    {
      "epoch": 17.35,
      "learning_rate": 2.1215090641842237e-05,
      "loss": 1.7656,
      "step": 95000
    },
    {
      "epoch": 17.53,
      "learning_rate": 2.0908868201861834e-05,
      "loss": 1.7856,
      "step": 96000
    },
    {
      "epoch": 17.71,
      "learning_rate": 2.060264576188143e-05,
      "loss": 1.8185,
      "step": 97000
    },
    {
      "epoch": 17.9,
      "learning_rate": 2.029642332190103e-05,
      "loss": 1.8397,
      "step": 98000
    },
    {
      "epoch": 18.08,
      "learning_rate": 1.999020088192063e-05,
      "loss": 1.7726,
      "step": 99000
    },
    {
      "epoch": 18.26,
      "learning_rate": 1.9683978441940226e-05,
      "loss": 1.6883,
      "step": 100000
    },
    {
      "epoch": 18.44,
      "learning_rate": 1.9377756001959823e-05,
      "loss": 1.724,
      "step": 101000
    },
    {
      "epoch": 18.63,
      "learning_rate": 1.9071533561979424e-05,
      "loss": 1.751,
      "step": 102000
    },
    {
      "epoch": 18.81,
      "learning_rate": 1.876531112199902e-05,
      "loss": 1.7654,
      "step": 103000
    },
    {
      "epoch": 18.99,
      "learning_rate": 1.8459088682018618e-05,
      "loss": 1.7807,
      "step": 104000
    },
    {
      "epoch": 19.17,
      "learning_rate": 1.8152866242038215e-05,
      "loss": 1.6287,
      "step": 105000
    },
    {
      "epoch": 19.36,
      "learning_rate": 1.7846643802057816e-05,
      "loss": 1.6529,
      "step": 106000
    },
    {
      "epoch": 19.54,
      "learning_rate": 1.7540421362077413e-05,
      "loss": 1.6782,
      "step": 107000
    },
    {
      "epoch": 19.72,
      "learning_rate": 1.723419892209701e-05,
      "loss": 1.6989,
      "step": 108000
    },
    {
      "epoch": 19.91,
      "learning_rate": 1.692797648211661e-05,
      "loss": 1.7147,
      "step": 109000
    },
    {
      "epoch": 20.09,
      "learning_rate": 1.6621754042136208e-05,
      "loss": 1.6426,
      "step": 110000
    },
    {
      "epoch": 20.27,
      "learning_rate": 1.6315531602155805e-05,
      "loss": 1.5852,
      "step": 111000
    },
    {
      "epoch": 20.45,
      "learning_rate": 1.6009309162175405e-05,
      "loss": 1.6156,
      "step": 112000
    },
    {
      "epoch": 20.64,
      "learning_rate": 1.5703086722195003e-05,
      "loss": 1.6307,
      "step": 113000
    },
    {
      "epoch": 20.82,
      "learning_rate": 1.53968642822146e-05,
      "loss": 1.6493,
      "step": 114000
    },
    {
      "epoch": 21.0,
      "learning_rate": 1.5090641842234199e-05,
      "loss": 1.6623,
      "step": 115000
    },
    {
      "epoch": 21.18,
      "learning_rate": 1.47844194022538e-05,
      "loss": 1.5248,
      "step": 116000
    },
    {
      "epoch": 21.37,
      "learning_rate": 1.4478196962273396e-05,
      "loss": 1.551,
      "step": 117000
    },
    {
      "epoch": 21.55,
      "learning_rate": 1.4171974522292993e-05,
      "loss": 1.5738,
      "step": 118000
    },
    {
      "epoch": 21.73,
      "learning_rate": 1.386575208231259e-05,
      "loss": 1.5854,
      "step": 119000
    },
    {
      "epoch": 21.91,
      "learning_rate": 1.3559529642332191e-05,
      "loss": 1.599,
      "step": 120000
    },
    {
      "epoch": 22.1,
      "learning_rate": 1.3253307202351788e-05,
      "loss": 1.5402,
      "step": 121000
    },
    {
      "epoch": 22.28,
      "learning_rate": 1.2947084762371387e-05,
      "loss": 1.486,
      "step": 122000
    },
    {
      "epoch": 22.46,
      "learning_rate": 1.2640862322390986e-05,
      "loss": 1.5188,
      "step": 123000
    },
    {
      "epoch": 22.64,
      "learning_rate": 1.2334639882410585e-05,
      "loss": 1.5442,
      "step": 124000
    },
    {
      "epoch": 22.83,
      "learning_rate": 1.2028417442430182e-05,
      "loss": 1.5483,
      "step": 125000
    },
    {
      "epoch": 23.01,
      "learning_rate": 1.1722195002449781e-05,
      "loss": 1.5501,
      "step": 126000
    },
    {
      "epoch": 23.19,
      "learning_rate": 1.1415972562469378e-05,
      "loss": 1.4471,
      "step": 127000
    },
    {
      "epoch": 23.37,
      "learning_rate": 1.1109750122488977e-05,
      "loss": 1.4578,
      "step": 128000
    },
    {
      "epoch": 23.56,
      "learning_rate": 1.0803527682508574e-05,
      "loss": 1.4712,
      "step": 129000
    },
    {
      "epoch": 23.74,
      "learning_rate": 1.0497305242528173e-05,
      "loss": 1.4918,
      "step": 130000
    },
    {
      "epoch": 23.92,
      "learning_rate": 1.0191082802547772e-05,
      "loss": 1.5129,
      "step": 131000
    },
    {
      "epoch": 24.11,
      "learning_rate": 9.884860362567369e-06,
      "loss": 1.4455,
      "step": 132000
    },
    {
      "epoch": 24.29,
      "learning_rate": 9.578637922586968e-06,
      "loss": 1.4141,
      "step": 133000
    },
    {
      "epoch": 24.47,
      "learning_rate": 9.272415482606565e-06,
      "loss": 1.4301,
      "step": 134000
    },
    {
      "epoch": 24.65,
      "learning_rate": 8.966193042626164e-06,
      "loss": 1.4454,
      "step": 135000
    },
    {
      "epoch": 24.84,
      "learning_rate": 8.659970602645761e-06,
      "loss": 1.4599,
      "step": 136000
    },
    {
      "epoch": 25.02,
      "learning_rate": 8.35374816266536e-06,
      "loss": 1.4573,
      "step": 137000
    },
    {
      "epoch": 25.2,
      "learning_rate": 8.047525722684959e-06,
      "loss": 1.3704,
      "step": 138000
    },
    {
      "epoch": 25.38,
      "learning_rate": 7.741303282704558e-06,
      "loss": 1.3845,
      "step": 139000
    },
    {
      "epoch": 25.57,
      "learning_rate": 7.4350808427241555e-06,
      "loss": 1.4078,
      "step": 140000
    },
    {
      "epoch": 25.75,
      "learning_rate": 7.128858402743753e-06,
      "loss": 1.4147,
      "step": 141000
    },
    {
      "epoch": 25.93,
      "learning_rate": 6.8226359627633515e-06,
      "loss": 1.4223,
      "step": 142000
    },
    {
      "epoch": 26.11,
      "learning_rate": 6.5164135227829495e-06,
      "loss": 1.3655,
      "step": 143000
    },
    {
      "epoch": 26.3,
      "learning_rate": 6.210191082802548e-06,
      "loss": 1.3528,
      "step": 144000
    },
    {
      "epoch": 26.48,
      "learning_rate": 5.903968642822146e-06,
      "loss": 1.3626,
      "step": 145000
    },
    {
      "epoch": 26.66,
      "learning_rate": 5.597746202841744e-06,
      "loss": 1.3719,
      "step": 146000
    },
    {
      "epoch": 26.84,
      "learning_rate": 5.291523762861342e-06,
      "loss": 1.3808,
      "step": 147000
    },
    {
      "epoch": 27.03,
      "learning_rate": 4.985301322880941e-06,
      "loss": 1.3752,
      "step": 148000
    },
    {
      "epoch": 27.21,
      "learning_rate": 4.679078882900539e-06,
      "loss": 1.3171,
      "step": 149000
    },
    {
      "epoch": 27.39,
      "learning_rate": 4.372856442920137e-06,
      "loss": 1.3321,
      "step": 150000
    },
    {
      "epoch": 27.57,
      "learning_rate": 4.066634002939736e-06,
      "loss": 1.3365,
      "step": 151000
    },
    {
      "epoch": 27.76,
      "learning_rate": 3.7604115629593337e-06,
      "loss": 1.3424,
      "step": 152000
    },
    {
      "epoch": 27.94,
      "learning_rate": 3.454189122978932e-06,
      "loss": 1.3471,
      "step": 153000
    },
    {
      "epoch": 28.12,
      "learning_rate": 3.14796668299853e-06,
      "loss": 1.3095,
      "step": 154000
    },
    {
      "epoch": 28.31,
      "learning_rate": 2.8417442430181285e-06,
      "loss": 1.3113,
      "step": 155000
    },
    {
      "epoch": 28.49,
      "learning_rate": 2.5355218030377265e-06,
      "loss": 1.3017,
      "step": 156000
    },
    {
      "epoch": 28.67,
      "learning_rate": 2.229299363057325e-06,
      "loss": 1.3147,
      "step": 157000
    },
    {
      "epoch": 28.85,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 1.3135,
      "step": 158000
    },
    {
      "epoch": 29.04,
      "learning_rate": 1.6168544830965214e-06,
      "loss": 1.309,
      "step": 159000
    },
    {
      "epoch": 29.22,
      "learning_rate": 1.3106320431161196e-06,
      "loss": 1.2784,
      "step": 160000
    },
    {
      "epoch": 29.4,
      "learning_rate": 1.0044096031357178e-06,
      "loss": 1.2903,
      "step": 161000
    },
    {
      "epoch": 29.58,
      "learning_rate": 6.98187163155316e-07,
      "loss": 1.2919,
      "step": 162000
    },
    {
      "epoch": 29.77,
      "learning_rate": 3.9196472317491427e-07,
      "loss": 1.2802,
      "step": 163000
    },
    {
      "epoch": 29.95,
      "learning_rate": 8.57422831945125e-08,
      "loss": 1.2904,
      "step": 164000
    }
  ],
  "max_steps": 164280,
  "num_train_epochs": 30,
  "total_flos": 3.45466660974336e+17,
  "trial_name": null,
  "trial_params": null
}