{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 68268,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.963379621491768e-05,
      "loss": 1.7826,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.926759242983536e-05,
      "loss": 1.5222,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.890138864475303e-05,
      "loss": 1.4992,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.853518485967071e-05,
      "loss": 1.4969,
      "step": 2000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.816898107458839e-05,
      "loss": 1.4997,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.780277728950607e-05,
      "loss": 1.4956,
      "step": 3000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7436573504423745e-05,
      "loss": 1.4934,
      "step": 3500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.707036971934142e-05,
      "loss": 1.4995,
      "step": 4000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6704165934259096e-05,
      "loss": 1.4932,
      "step": 4500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.633796214917678e-05,
      "loss": 1.4914,
      "step": 5000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.5971758364094454e-05,
      "loss": 1.4894,
      "step": 5500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5605554579012136e-05,
      "loss": 1.4955,
      "step": 6000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.5239350793929805e-05,
      "loss": 1.4833,
      "step": 6500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.487314700884749e-05,
      "loss": 1.4888,
      "step": 7000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.450694322376516e-05,
      "loss": 1.4858,
      "step": 7500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.414073943868284e-05,
      "loss": 1.4881,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.377453565360052e-05,
      "loss": 1.4882,
      "step": 8500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.3408331868518196e-05,
      "loss": 1.4854,
      "step": 9000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.304212808343587e-05,
      "loss": 1.4933,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.267592429835355e-05,
      "loss": 1.4912,
      "step": 10000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.230972051327123e-05,
      "loss": 1.4782,
      "step": 10500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.1943516728188905e-05,
      "loss": 1.4891,
      "step": 11000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.157731294310658e-05,
      "loss": 1.4848,
      "step": 11500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.1211109158024256e-05,
      "loss": 1.4839,
      "step": 12000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.084490537294194e-05,
      "loss": 1.4822,
      "step": 12500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.0478701587859614e-05,
      "loss": 1.4835,
      "step": 13000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.0112497802777296e-05,
      "loss": 1.4849,
      "step": 13500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.9746294017694965e-05,
      "loss": 1.4839,
      "step": 14000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.938009023261265e-05,
      "loss": 1.4838,
      "step": 14500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.901388644753032e-05,
      "loss": 1.4827,
      "step": 15000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.8647682662448e-05,
      "loss": 1.4803,
      "step": 15500
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.828147887736568e-05,
      "loss": 1.4756,
      "step": 16000
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.7915275092283356e-05,
      "loss": 1.4815,
      "step": 16500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.754907130720103e-05,
      "loss": 1.4737,
      "step": 17000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.718286752211871e-05,
      "loss": 1.4749,
      "step": 17500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.681666373703639e-05,
      "loss": 1.4952,
      "step": 18000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.6450459951954065e-05,
      "loss": 1.4818,
      "step": 18500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.608425616687174e-05,
      "loss": 1.4868,
      "step": 19000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.5718052381789416e-05,
      "loss": 1.4774,
      "step": 19500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.53518485967071e-05,
      "loss": 1.4758,
      "step": 20000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.4985644811624774e-05,
      "loss": 1.4752,
      "step": 20500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4619441026542456e-05,
      "loss": 1.4782,
      "step": 21000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.425323724146013e-05,
      "loss": 1.4807,
      "step": 21500
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.388703345637781e-05,
      "loss": 1.4754,
      "step": 22000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.352082967129548e-05,
      "loss": 1.4764,
      "step": 22500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3154625886213165e-05,
      "loss": 1.4096,
      "step": 23000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.278842210113084e-05,
      "loss": 1.3221,
      "step": 23500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.2422218316048517e-05,
      "loss": 1.3234,
      "step": 24000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.205601453096619e-05,
      "loss": 1.3303,
      "step": 24500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.168981074588387e-05,
      "loss": 1.3323,
      "step": 25000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.132360696080155e-05,
      "loss": 1.3368,
      "step": 25500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.0957403175719225e-05,
      "loss": 1.3301,
      "step": 26000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.05911993906369e-05,
      "loss": 1.3436,
      "step": 26500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.022499560555458e-05,
      "loss": 1.3404,
      "step": 27000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.985879182047226e-05,
      "loss": 1.33,
      "step": 27500
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.9492588035389934e-05,
      "loss": 1.3362,
      "step": 28000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9126384250307613e-05,
      "loss": 1.3349,
      "step": 28500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.876018046522529e-05,
      "loss": 1.3422,
      "step": 29000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.839397668014297e-05,
      "loss": 1.3331,
      "step": 29500
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.8027772895060643e-05,
      "loss": 1.3389,
      "step": 30000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7661569109978326e-05,
      "loss": 1.3308,
      "step": 30500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7295365324896e-05,
      "loss": 1.3319,
      "step": 31000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6929161539813673e-05,
      "loss": 1.3305,
      "step": 31500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6562957754731356e-05,
      "loss": 1.3361,
      "step": 32000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6196753969649028e-05,
      "loss": 1.3386,
      "step": 32500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.583055018456671e-05,
      "loss": 1.3307,
      "step": 33000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5464346399484386e-05,
      "loss": 1.3332,
      "step": 33500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5098142614402065e-05,
      "loss": 1.3437,
      "step": 34000
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4731938829319744e-05,
      "loss": 1.341,
      "step": 34500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4365735044237416e-05,
      "loss": 1.3401,
      "step": 35000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.3999531259155095e-05,
      "loss": 1.3393,
      "step": 35500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3633327474072774e-05,
      "loss": 1.3426,
      "step": 36000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.326712368899045e-05,
      "loss": 1.3376,
      "step": 36500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2900919903908128e-05,
      "loss": 1.334,
      "step": 37000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2534716118825804e-05,
      "loss": 1.3362,
      "step": 37500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2168512333743483e-05,
      "loss": 1.3303,
      "step": 38000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.180230854866116e-05,
      "loss": 1.3385,
      "step": 38500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1436104763578837e-05,
      "loss": 1.3364,
      "step": 39000
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1069900978496516e-05,
      "loss": 1.3394,
      "step": 39500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.070369719341419e-05,
      "loss": 1.3315,
      "step": 40000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.033749340833187e-05,
      "loss": 1.3435,
      "step": 40500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.997128962324955e-05,
      "loss": 1.3421,
      "step": 41000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9605085838167225e-05,
      "loss": 1.3384,
      "step": 41500
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9238882053084904e-05,
      "loss": 1.3388,
      "step": 42000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.887267826800258e-05,
      "loss": 1.3366,
      "step": 42500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8506474482920255e-05,
      "loss": 1.347,
      "step": 43000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8140270697837934e-05,
      "loss": 1.336,
      "step": 43500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.777406691275561e-05,
      "loss": 1.337,
      "step": 44000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7407863127673288e-05,
      "loss": 1.3428,
      "step": 44500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.7041659342590964e-05,
      "loss": 1.3451,
      "step": 45000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6675455557508643e-05,
      "loss": 1.3463,
      "step": 45500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.630925177242632e-05,
      "loss": 1.2271,
      "step": 46000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.5943047987343997e-05,
      "loss": 1.2264,
      "step": 46500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5576844202261676e-05,
      "loss": 1.2323,
      "step": 47000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5210640417179353e-05,
      "loss": 1.2248,
      "step": 47500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.484443663209703e-05,
      "loss": 1.2215,
      "step": 48000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4478232847014708e-05,
      "loss": 1.2293,
      "step": 48500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4112029061932385e-05,
      "loss": 1.2319,
      "step": 49000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3745825276850064e-05,
      "loss": 1.2262,
      "step": 49500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3379621491767741e-05,
      "loss": 1.2289,
      "step": 50000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.3013417706685415e-05,
      "loss": 1.2317,
      "step": 50500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2647213921603094e-05,
      "loss": 1.2269,
      "step": 51000
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2281010136520773e-05,
      "loss": 1.2277,
      "step": 51500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1914806351438449e-05,
      "loss": 1.2306,
      "step": 52000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1548602566356126e-05,
      "loss": 1.2358,
      "step": 52500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1182398781273803e-05,
      "loss": 1.2277,
      "step": 53000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.081619499619148e-05,
      "loss": 1.2271,
      "step": 53500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0449991211109159e-05,
      "loss": 1.2384,
      "step": 54000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0083787426026836e-05,
      "loss": 1.2308,
      "step": 54500
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.717583640944514e-06,
      "loss": 1.223,
      "step": 55000
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.35137985586219e-06,
      "loss": 1.2304,
      "step": 55500
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.985176070779868e-06,
      "loss": 1.2296,
      "step": 56000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.618972285697545e-06,
      "loss": 1.2297,
      "step": 56500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.252768500615223e-06,
      "loss": 1.2309,
      "step": 57000
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.8865647155329e-06,
      "loss": 1.2306,
      "step": 57500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.520360930450578e-06,
      "loss": 1.228,
      "step": 58000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.154157145368255e-06,
      "loss": 1.2297,
      "step": 58500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.787953360285932e-06,
      "loss": 1.2363,
      "step": 59000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.4217495752036104e-06,
      "loss": 1.2327,
      "step": 59500
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.055545790121287e-06,
      "loss": 1.2316,
      "step": 60000
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.689342005038965e-06,
      "loss": 1.23,
      "step": 60500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.323138219956641e-06,
      "loss": 1.2277,
      "step": 61000
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.956934434874319e-06,
      "loss": 1.2272,
      "step": 61500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.590730649791997e-06,
      "loss": 1.2312,
      "step": 62000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.224526864709674e-06,
      "loss": 1.2337,
      "step": 62500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.858323079627351e-06,
      "loss": 1.228,
      "step": 63000
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.4921192945450287e-06,
      "loss": 1.2276,
      "step": 63500
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.125915509462706e-06,
      "loss": 1.2256,
      "step": 64000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.7597117243803832e-06,
      "loss": 1.2339,
      "step": 64500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3935079392980604e-06,
      "loss": 1.2302,
      "step": 65000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.027304154215738e-06,
      "loss": 1.2282,
      "step": 65500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6611003691334153e-06,
      "loss": 1.2263,
      "step": 66000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.2948965840510928e-06,
      "loss": 1.2292,
      "step": 66500
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.286927989687702e-07,
      "loss": 1.2356,
      "step": 67000
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.624890138864476e-07,
      "loss": 1.2314,
      "step": 67500
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.962852288041249e-07,
      "loss": 1.2327,
      "step": 68000
    },
    {
      "epoch": 3.0,
      "step": 68268,
      "total_flos": 2748096779848450560,
      "train_runtime": 192268.3522,
      "train_samples_per_second": 0.355
    }
  ],
  "max_steps": 68268,
  "num_train_epochs": 3,
  "total_flos": 2748096779848450560,
  "trial_name": null,
  "trial_params": null
}