{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9996687966007222,
  "global_step": 247500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.9959609341551486e-05,
      "loss": 1.1369,
      "step": 500
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9919218683102975e-05,
      "loss": 1.1452,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.987882802465446e-05,
      "loss": 1.0938,
      "step": 1500
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9838437366205944e-05,
      "loss": 1.0933,
      "step": 2000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9798046707757432e-05,
      "loss": 1.1217,
      "step": 2500
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9757656049308917e-05,
      "loss": 1.0401,
      "step": 3000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9717265390860402e-05,
      "loss": 1.0169,
      "step": 3500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.967687473241189e-05,
      "loss": 1.1245,
      "step": 4000
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9636484073963375e-05,
      "loss": 1.0174,
      "step": 4500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.959609341551486e-05,
      "loss": 1.024,
      "step": 5000
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9555702757066348e-05,
      "loss": 1.0263,
      "step": 5500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9515312098617832e-05,
      "loss": 1.042,
      "step": 6000
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9474921440169317e-05,
      "loss": 1.0445,
      "step": 6500
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9434530781720805e-05,
      "loss": 1.0388,
      "step": 7000
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.939414012327229e-05,
      "loss": 1.0556,
      "step": 7500
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9353749464823778e-05,
      "loss": 1.0525,
      "step": 8000
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9313358806375263e-05,
      "loss": 1.09,
      "step": 8500
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9272968147926748e-05,
      "loss": 1.0217,
      "step": 9000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9232577489478236e-05,
      "loss": 1.0252,
      "step": 9500
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.919218683102972e-05,
      "loss": 1.0263,
      "step": 10000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.915179617258121e-05,
      "loss": 1.041,
      "step": 10500
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9111405514132694e-05,
      "loss": 1.0094,
      "step": 11000
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.907101485568418e-05,
      "loss": 1.0667,
      "step": 11500
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9030624197235667e-05,
      "loss": 1.0105,
      "step": 12000
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.899023353878715e-05,
      "loss": 1.0098,
      "step": 12500
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.8949842880338636e-05,
      "loss": 1.0656,
      "step": 13000
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.8909452221890124e-05,
      "loss": 1.0973,
      "step": 13500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.886906156344161e-05,
      "loss": 0.9736,
      "step": 14000
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.8828670904993094e-05,
      "loss": 1.0193,
      "step": 14500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.8788280246544582e-05,
      "loss": 0.9827,
      "step": 15000
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.8747889588096067e-05,
      "loss": 1.0219,
      "step": 15500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.870749892964755e-05,
      "loss": 1.0261,
      "step": 16000
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.866710827119904e-05,
      "loss": 0.993,
      "step": 16500
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8626717612750524e-05,
      "loss": 1.0186,
      "step": 17000
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.858632695430201e-05,
      "loss": 0.9749,
      "step": 17500
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8545936295853497e-05,
      "loss": 1.0013,
      "step": 18000
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.8505545637404982e-05,
      "loss": 0.9743,
      "step": 18500
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8465154978956467e-05,
      "loss": 0.9796,
      "step": 19000
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8424764320507955e-05,
      "loss": 1.0136,
      "step": 19500
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.838437366205944e-05,
      "loss": 1.0118,
      "step": 20000
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8343983003610925e-05,
      "loss": 1.0037,
      "step": 20500
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8303592345162413e-05,
      "loss": 0.9659,
      "step": 21000
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.8263201686713897e-05,
      "loss": 1.0248,
      "step": 21500
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.8222811028265382e-05,
      "loss": 1.0028,
      "step": 22000
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.818242036981687e-05,
      "loss": 1.0514,
      "step": 22500
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.8142029711368355e-05,
      "loss": 1.0166,
      "step": 23000
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.810163905291984e-05,
      "loss": 0.9976,
      "step": 23500
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.8061248394471328e-05,
      "loss": 1.0427,
      "step": 24000
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.8020857736022813e-05,
      "loss": 1.0069,
      "step": 24500
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.79804670775743e-05,
      "loss": 1.035,
      "step": 25000
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.7940076419125786e-05,
      "loss": 0.9844,
      "step": 25500
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7899685760677274e-05,
      "loss": 0.9708,
      "step": 26000
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.785929510222876e-05,
      "loss": 1.0022,
      "step": 26500
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7818904443780243e-05,
      "loss": 1.037,
      "step": 27000
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.777851378533173e-05,
      "loss": 0.9876,
      "step": 27500
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7738123126883216e-05,
      "loss": 0.9781,
      "step": 28000
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.7697732468434705e-05,
      "loss": 0.9913,
      "step": 28500
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.765734180998619e-05,
      "loss": 1.0041,
      "step": 29000
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.7616951151537674e-05,
      "loss": 1.0104,
      "step": 29500
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.757656049308916e-05,
      "loss": 0.9863,
      "step": 30000
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.7536169834640647e-05,
      "loss": 1.0076,
      "step": 30500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.7495779176192132e-05,
      "loss": 0.9887,
      "step": 31000
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.7455388517743617e-05,
      "loss": 0.9595,
      "step": 31500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.7414997859295105e-05,
      "loss": 0.9512,
      "step": 32000
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.737460720084659e-05,
      "loss": 0.9956,
      "step": 32500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.7334216542398074e-05,
      "loss": 0.9986,
      "step": 33000
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.7293825883949562e-05,
      "loss": 0.9447,
      "step": 33500
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.7253435225501047e-05,
      "loss": 0.9576,
      "step": 34000
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.7213044567052532e-05,
      "loss": 0.9977,
      "step": 34500
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.717265390860402e-05,
      "loss": 0.9411,
      "step": 35000
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.7132263250155505e-05,
      "loss": 1.0055,
      "step": 35500
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.709187259170699e-05,
      "loss": 1.0319,
      "step": 36000
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.7051481933258478e-05,
      "loss": 0.9554,
      "step": 36500
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.7011091274809963e-05,
      "loss": 0.9689,
      "step": 37000
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.6970700616361447e-05,
      "loss": 0.9857,
      "step": 37500
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.6930309957912935e-05,
      "loss": 0.9764,
      "step": 38000
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.688991929946442e-05,
      "loss": 0.9616,
      "step": 38500
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.6849528641015905e-05,
      "loss": 0.9978,
      "step": 39000
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.6809137982567393e-05,
      "loss": 0.9387,
      "step": 39500
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.6768747324118878e-05,
      "loss": 0.9403,
      "step": 40000
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.6728356665670366e-05,
      "loss": 0.9615,
      "step": 40500
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.668796600722185e-05,
      "loss": 0.9558,
      "step": 41000
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.6647575348773336e-05,
      "loss": 0.9855,
      "step": 41500
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.6607184690324824e-05,
      "loss": 1.0088,
      "step": 42000
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.656679403187631e-05,
      "loss": 0.9751,
      "step": 42500
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.6526403373427797e-05,
      "loss": 0.9547,
      "step": 43000
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.648601271497928e-05,
      "loss": 0.9572,
      "step": 43500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6445622056530766e-05,
      "loss": 0.9317,
      "step": 44000
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6405231398082254e-05,
      "loss": 0.9836,
      "step": 44500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.636484073963374e-05,
      "loss": 0.9795,
      "step": 45000
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6324450081185227e-05,
      "loss": 0.9789,
      "step": 45500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.6284059422736712e-05,
      "loss": 0.9628,
      "step": 46000
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.6243668764288197e-05,
      "loss": 0.9666,
      "step": 46500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.6203278105839685e-05,
      "loss": 0.9821,
      "step": 47000
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.616288744739117e-05,
      "loss": 0.9409,
      "step": 47500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.6122496788942655e-05,
      "loss": 1.0201,
      "step": 48000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.6082106130494143e-05,
      "loss": 0.9693,
      "step": 48500
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.6041715472045627e-05,
      "loss": 0.9974,
      "step": 49000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.6001324813597112e-05,
      "loss": 0.9703,
      "step": 49500
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.5960934155148597e-05,
      "loss": 0.9504,
      "step": 50000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.5920543496700085e-05,
      "loss": 0.9595,
      "step": 50500
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.588015283825157e-05,
      "loss": 0.9734,
      "step": 51000
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.5839762179803055e-05,
      "loss": 0.945,
      "step": 51500
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.5799371521354543e-05,
      "loss": 0.9776,
      "step": 52000
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.5758980862906028e-05,
      "loss": 0.9735,
      "step": 52500
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.5718590204457512e-05,
      "loss": 1.007,
      "step": 53000
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5678199546009e-05,
      "loss": 0.9473,
      "step": 53500
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5637808887560485e-05,
      "loss": 0.9741,
      "step": 54000
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.559741822911197e-05,
      "loss": 0.9218,
      "step": 54500
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5557027570663458e-05,
      "loss": 0.9297,
      "step": 55000
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.5516636912214943e-05,
      "loss": 0.9443,
      "step": 55500
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.5476246253766428e-05,
      "loss": 0.9553,
      "step": 56000
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.5435855595317916e-05,
      "loss": 1.0098,
      "step": 56500
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.53954649368694e-05,
      "loss": 0.9684,
      "step": 57000
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.535507427842089e-05,
      "loss": 0.9874,
      "step": 57500
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.5314683619972374e-05,
      "loss": 0.9602,
      "step": 58000
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.527429296152386e-05,
      "loss": 0.9409,
      "step": 58500
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.5233902303075345e-05,
      "loss": 0.9379,
      "step": 59000
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.5193511644626833e-05,
      "loss": 0.9913,
      "step": 59500
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.5153120986178318e-05,
      "loss": 0.9938,
      "step": 60000
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.5112730327729802e-05,
      "loss": 0.9803,
      "step": 60500
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.507233966928129e-05,
      "loss": 0.9494,
      "step": 61000
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.5031949010832775e-05,
      "loss": 0.9691,
      "step": 61500
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.4991558352384262e-05,
      "loss": 0.9712,
      "step": 62000
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.4951167693935748e-05,
      "loss": 0.9205,
      "step": 62500
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.4910777035487233e-05,
      "loss": 0.9914,
      "step": 63000
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.487038637703872e-05,
      "loss": 0.9812,
      "step": 63500
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.4829995718590206e-05,
      "loss": 0.9621,
      "step": 64000
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.4789605060141692e-05,
      "loss": 0.964,
      "step": 64500
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.4749214401693177e-05,
      "loss": 0.9682,
      "step": 65000
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.4708823743244664e-05,
      "loss": 0.983,
      "step": 65500
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.466843308479615e-05,
      "loss": 0.9768,
      "step": 66000
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.4628042426347635e-05,
      "loss": 0.962,
      "step": 66500
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.4587651767899123e-05,
      "loss": 1.0047,
      "step": 67000
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.4547261109450608e-05,
      "loss": 0.9667,
      "step": 67500
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.4506870451002093e-05,
      "loss": 1.0206,
      "step": 68000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.446647979255358e-05,
      "loss": 0.937,
      "step": 68500
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.4426089134105066e-05,
      "loss": 0.9796,
      "step": 69000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.438569847565655e-05,
      "loss": 0.9328,
      "step": 69500
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.4345307817208037e-05,
      "loss": 0.9244,
      "step": 70000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.4304917158759523e-05,
      "loss": 0.9437,
      "step": 70500
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.426452650031101e-05,
      "loss": 0.9466,
      "step": 71000
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4224135841862494e-05,
      "loss": 0.9172,
      "step": 71500
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4183745183413981e-05,
      "loss": 0.9681,
      "step": 72000
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4143354524965467e-05,
      "loss": 0.9339,
      "step": 72500
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.4102963866516952e-05,
      "loss": 0.9663,
      "step": 73000
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.406257320806844e-05,
      "loss": 0.9324,
      "step": 73500
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.4022182549619925e-05,
      "loss": 0.9784,
      "step": 74000
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.398179189117141e-05,
      "loss": 0.9536,
      "step": 74500
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.3941401232722898e-05,
      "loss": 0.9418,
      "step": 75000
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.3901010574274383e-05,
      "loss": 0.9225,
      "step": 75500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.3860619915825868e-05,
      "loss": 0.9436,
      "step": 76000
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.3820229257377356e-05,
      "loss": 0.9656,
      "step": 76500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.377983859892884e-05,
      "loss": 0.9451,
      "step": 77000
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.3739447940480325e-05,
      "loss": 0.9152,
      "step": 77500
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.3699057282031813e-05,
      "loss": 0.9567,
      "step": 78000
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.3658666623583298e-05,
      "loss": 0.9142,
      "step": 78500
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.3618275965134785e-05,
      "loss": 0.9347,
      "step": 79000
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.3577885306686271e-05,
      "loss": 0.9835,
      "step": 79500
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.3537494648237758e-05,
      "loss": 0.9171,
      "step": 80000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.3497103989789242e-05,
      "loss": 0.9485,
      "step": 80500
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.3456713331340729e-05,
      "loss": 0.9879,
      "step": 81000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.3416322672892215e-05,
      "loss": 0.9326,
      "step": 81500
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.33759320144437e-05,
      "loss": 0.9512,
      "step": 82000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.3335541355995188e-05,
      "loss": 0.9233,
      "step": 82500
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.3295150697546673e-05,
      "loss": 0.9232,
      "step": 83000
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.3254760039098158e-05,
      "loss": 0.9353,
      "step": 83500
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.3214369380649646e-05,
      "loss": 0.9536,
      "step": 84000
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.317397872220113e-05,
      "loss": 0.9547,
      "step": 84500
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.3133588063752615e-05,
      "loss": 0.9323,
      "step": 85000
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.3093197405304104e-05,
      "loss": 0.9414,
      "step": 85500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.3052806746855588e-05,
      "loss": 0.9694,
      "step": 86000
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.3012416088407073e-05,
      "loss": 0.931,
      "step": 86500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.2972025429958561e-05,
      "loss": 0.949,
      "step": 87000
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.2931634771510046e-05,
      "loss": 0.9511,
      "step": 87500
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.2891244113061532e-05,
      "loss": 0.9378,
      "step": 88000
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.2850853454613019e-05,
      "loss": 0.9475,
      "step": 88500
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.2810462796164504e-05,
      "loss": 0.9869,
      "step": 89000
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.277007213771599e-05,
      "loss": 0.9298,
      "step": 89500
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.2729681479267475e-05,
      "loss": 0.9133,
      "step": 90000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.2689290820818963e-05,
      "loss": 0.9378,
      "step": 90500
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.2648900162370448e-05,
      "loss": 0.9598,
      "step": 91000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.2608509503921933e-05,
      "loss": 0.9438,
      "step": 91500
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.256811884547342e-05,
      "loss": 0.9571,
      "step": 92000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.2527728187024906e-05,
      "loss": 0.9755,
      "step": 92500
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.248733752857639e-05,
      "loss": 0.9469,
      "step": 93000
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.2446946870127878e-05,
      "loss": 0.9402,
      "step": 93500
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.2406556211679363e-05,
      "loss": 0.9425,
      "step": 94000
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.236616555323085e-05,
      "loss": 0.9816,
      "step": 94500
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.2325774894782336e-05,
      "loss": 0.9738,
      "step": 95000
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.2285384236333821e-05,
      "loss": 0.9631,
      "step": 95500
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.2244993577885307e-05,
      "loss": 0.9766,
      "step": 96000
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.2204602919436794e-05,
      "loss": 0.9972,
      "step": 96500
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.216421226098828e-05,
      "loss": 0.9506,
      "step": 97000
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.2123821602539765e-05,
      "loss": 0.9384,
      "step": 97500
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.2083430944091252e-05,
      "loss": 0.9416,
      "step": 98000
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.2043040285642738e-05,
      "loss": 0.9529,
      "step": 98500
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.2002649627194223e-05,
      "loss": 0.9469,
      "step": 99000
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.1962258968745711e-05,
      "loss": 0.8995,
      "step": 99500
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.1921868310297196e-05,
      "loss": 0.9206,
      "step": 100000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.188147765184868e-05,
      "loss": 0.9306,
      "step": 100500
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.1841086993400169e-05,
      "loss": 0.9977,
      "step": 101000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.1800696334951653e-05,
      "loss": 0.9301,
      "step": 101500
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.1760305676503138e-05,
      "loss": 0.9425,
      "step": 102000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.1719915018054626e-05,
      "loss": 0.8925,
      "step": 102500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.1679524359606111e-05,
      "loss": 0.9098,
      "step": 103000
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.1639133701157598e-05,
      "loss": 0.9129,
      "step": 103500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.1598743042709084e-05,
      "loss": 0.9247,
      "step": 104000
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.1558352384260569e-05,
      "loss": 0.9092,
      "step": 104500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.1517961725812055e-05,
      "loss": 0.9261,
      "step": 105000
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1477571067363542e-05,
      "loss": 0.9342,
      "step": 105500
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1437180408915028e-05,
      "loss": 0.9485,
      "step": 106000
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1396789750466513e-05,
      "loss": 0.8406,
      "step": 106500
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1356399092018e-05,
      "loss": 0.9847,
      "step": 107000
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.1316008433569486e-05,
      "loss": 0.9163,
      "step": 107500
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.127561777512097e-05,
      "loss": 0.9396,
      "step": 108000
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1235227116672459e-05,
      "loss": 0.928,
      "step": 108500
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1194836458223944e-05,
      "loss": 0.9116,
      "step": 109000
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1154445799775428e-05,
      "loss": 0.9316,
      "step": 109500
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.1114055141326913e-05,
      "loss": 0.915,
      "step": 110000
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1073664482878401e-05,
      "loss": 0.9313,
      "step": 110500
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1033273824429886e-05,
      "loss": 0.9005,
      "step": 111000
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.0992883165981372e-05,
      "loss": 0.9014,
      "step": 111500
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.0952492507532859e-05,
      "loss": 0.923,
      "step": 112000
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.0912101849084344e-05,
      "loss": 0.9115,
      "step": 112500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.087171119063583e-05,
      "loss": 0.9482,
      "step": 113000
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0831320532187317e-05,
      "loss": 0.9236,
      "step": 113500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0790929873738803e-05,
      "loss": 0.9172,
      "step": 114000
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0750539215290288e-05,
      "loss": 0.9374,
      "step": 114500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.0710148556841776e-05,
      "loss": 0.9542,
      "step": 115000
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.066975789839326e-05,
      "loss": 0.9249,
      "step": 115500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0629367239944745e-05,
      "loss": 0.93,
      "step": 116000
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0588976581496234e-05,
      "loss": 0.9265,
      "step": 116500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0548585923047718e-05,
      "loss": 0.8939,
      "step": 117000
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.0508195264599203e-05,
      "loss": 0.9213,
      "step": 117500
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0467804606150691e-05,
      "loss": 0.9554,
      "step": 118000
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0427413947702176e-05,
      "loss": 0.9253,
      "step": 118500
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0387023289253661e-05,
      "loss": 0.9161,
      "step": 119000
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0346632630805149e-05,
      "loss": 0.8829,
      "step": 119500
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0306241972356634e-05,
      "loss": 0.9496,
      "step": 120000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.026585131390812e-05,
      "loss": 0.8863,
      "step": 120500
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0225460655459607e-05,
      "loss": 0.911,
      "step": 121000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0185069997011091e-05,
      "loss": 0.936,
      "step": 121500
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0144679338562578e-05,
      "loss": 0.9418,
      "step": 122000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0104288680114064e-05,
      "loss": 0.9327,
      "step": 122500
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0063898021665551e-05,
      "loss": 0.9365,
      "step": 123000
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0023507363217036e-05,
      "loss": 0.9386,
      "step": 123500
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.983116704768522e-06,
      "loss": 0.9122,
      "step": 124000
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.942726046320009e-06,
      "loss": 0.9047,
      "step": 124500
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.902335387871493e-06,
      "loss": 0.9056,
      "step": 125000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.86194472942298e-06,
      "loss": 0.9117,
      "step": 125500
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.821554070974466e-06,
      "loss": 0.9186,
      "step": 126000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.781163412525951e-06,
      "loss": 0.9437,
      "step": 126500
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.740772754077437e-06,
      "loss": 0.8771,
      "step": 127000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.700382095628924e-06,
      "loss": 0.9197,
      "step": 127500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.659991437180409e-06,
      "loss": 0.9057,
      "step": 128000
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.619600778731895e-06,
      "loss": 0.946,
      "step": 128500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.579210120283382e-06,
      "loss": 0.884,
      "step": 129000
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.538819461834868e-06,
      "loss": 0.9384,
      "step": 129500
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.498428803386353e-06,
      "loss": 0.913,
      "step": 130000
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.45803814493784e-06,
      "loss": 0.8897,
      "step": 130500
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.417647486489326e-06,
      "loss": 0.9152,
      "step": 131000
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.377256828040812e-06,
      "loss": 0.8772,
      "step": 131500
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.336866169592299e-06,
      "loss": 0.9231,
      "step": 132000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.296475511143783e-06,
      "loss": 0.9081,
      "step": 132500
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.25608485269527e-06,
      "loss": 0.9135,
      "step": 133000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.215694194246755e-06,
      "loss": 0.9331,
      "step": 133500
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.175303535798241e-06,
      "loss": 0.95,
      "step": 134000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.134912877349728e-06,
      "loss": 0.9205,
      "step": 134500
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.094522218901212e-06,
      "loss": 0.8951,
      "step": 135000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.054131560452699e-06,
      "loss": 0.9264,
      "step": 135500
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.013740902004185e-06,
      "loss": 0.879,
      "step": 136000
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.97335024355567e-06,
      "loss": 0.9176,
      "step": 136500
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.932959585107157e-06,
      "loss": 0.905,
      "step": 137000
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.892568926658643e-06,
      "loss": 0.9876,
      "step": 137500
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.85217826821013e-06,
      "loss": 0.958,
      "step": 138000
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.811787609761614e-06,
      "loss": 0.8806,
      "step": 138500
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.7713969513131e-06,
      "loss": 0.9072,
      "step": 139000
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.731006292864587e-06,
      "loss": 0.91,
      "step": 139500
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.690615634416074e-06,
      "loss": 0.8987,
      "step": 140000
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.65022497596756e-06,
      "loss": 0.921,
      "step": 140500
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.609834317519045e-06,
      "loss": 0.955,
      "step": 141000
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.569443659070531e-06,
      "loss": 0.9451,
      "step": 141500
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.529053000622018e-06,
      "loss": 0.8754,
      "step": 142000
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.488662342173503e-06,
      "loss": 0.9248,
      "step": 142500
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.448271683724989e-06,
      "loss": 0.9433,
      "step": 143000
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.407881025276474e-06,
      "loss": 0.9401,
      "step": 143500
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.36749036682796e-06,
      "loss": 0.926,
      "step": 144000
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.327099708379447e-06,
      "loss": 0.9239,
      "step": 144500
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.286709049930931e-06,
      "loss": 0.9226,
      "step": 145000
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.246318391482418e-06,
      "loss": 0.8886,
      "step": 145500
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.205927733033904e-06,
      "loss": 0.9008,
      "step": 146000
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.16553707458539e-06,
      "loss": 0.8982,
      "step": 146500
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.125146416136877e-06,
      "loss": 0.9281,
      "step": 147000
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.084755757688362e-06,
      "loss": 0.8951,
      "step": 147500
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.044365099239849e-06,
      "loss": 0.8947,
      "step": 148000
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.003974440791335e-06,
      "loss": 0.903,
      "step": 148500
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.963583782342821e-06,
      "loss": 0.8872,
      "step": 149000
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.923193123894306e-06,
      "loss": 0.923,
      "step": 149500
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.882802465445793e-06,
      "loss": 0.8929,
      "step": 150000
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.842411806997279e-06,
      "loss": 0.9127,
      "step": 150500
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.802021148548764e-06,
      "loss": 0.8948,
      "step": 151000
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.76163049010025e-06,
      "loss": 0.9375,
      "step": 151500
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.721239831651737e-06,
      "loss": 0.9496,
      "step": 152000
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.680849173203222e-06,
      "loss": 0.8727,
      "step": 152500
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.640458514754708e-06,
      "loss": 0.8689,
      "step": 153000
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.600067856306194e-06,
      "loss": 0.8999,
      "step": 153500
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.55967719785768e-06,
      "loss": 0.9765,
      "step": 154000
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.519286539409167e-06,
      "loss": 0.9241,
      "step": 154500
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.478895880960651e-06,
      "loss": 0.8522,
      "step": 155000
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.438505222512138e-06,
      "loss": 0.919,
      "step": 155500
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.398114564063624e-06,
      "loss": 0.9277,
      "step": 156000
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.35772390561511e-06,
      "loss": 0.9202,
      "step": 156500
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.3173332471665955e-06,
      "loss": 0.8872,
      "step": 157000
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.276942588718082e-06,
      "loss": 0.9106,
      "step": 157500
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.236551930269568e-06,
      "loss": 0.9147,
      "step": 158000
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.196161271821054e-06,
      "loss": 0.8906,
      "step": 158500
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.1557706133725405e-06,
      "loss": 0.8656,
      "step": 159000
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.115379954924025e-06,
      "loss": 0.8898,
      "step": 159500
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.074989296475512e-06,
      "loss": 0.8818,
      "step": 160000
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.034598638026998e-06,
      "loss": 0.9191,
      "step": 160500
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.994207979578484e-06,
      "loss": 0.9219,
      "step": 161000
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.9538173211299694e-06,
      "loss": 0.9275,
      "step": 161500
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.913426662681456e-06,
      "loss": 0.9306,
      "step": 162000
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.8730360042329415e-06,
      "loss": 0.903,
      "step": 162500
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.832645345784428e-06,
      "loss": 0.914,
      "step": 163000
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.792254687335913e-06,
      "loss": 0.9402,
      "step": 163500
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.751864028887399e-06,
      "loss": 0.9246,
      "step": 164000
    },
    {
      "epoch": 0.66,
      "learning_rate": 6.711473370438886e-06,
      "loss": 0.8842,
      "step": 164500
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.671082711990371e-06,
      "loss": 0.8796,
      "step": 165000
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.630692053541857e-06,
      "loss": 0.9058,
      "step": 165500
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.590301395093343e-06,
      "loss": 0.8838,
      "step": 166000
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.549910736644829e-06,
      "loss": 0.9267,
      "step": 166500
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.5095200781963154e-06,
      "loss": 0.8483,
      "step": 167000
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.469129419747802e-06,
      "loss": 0.9276,
      "step": 167500
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.428738761299287e-06,
      "loss": 0.9106,
      "step": 168000
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.388348102850773e-06,
      "loss": 0.8863,
      "step": 168500
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.34795744440226e-06,
      "loss": 0.9168,
      "step": 169000
    },
    {
      "epoch": 0.68,
      "learning_rate": 6.307566785953745e-06,
      "loss": 0.8849,
      "step": 169500
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.267176127505231e-06,
      "loss": 0.8818,
      "step": 170000
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.226785469056717e-06,
      "loss": 0.8809,
      "step": 170500
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.186394810608203e-06,
      "loss": 0.9211,
      "step": 171000
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.146004152159689e-06,
      "loss": 0.9225,
      "step": 171500
    },
    {
      "epoch": 0.69,
      "learning_rate": 6.105613493711176e-06,
      "loss": 0.9148,
      "step": 172000
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.065222835262661e-06,
      "loss": 0.9156,
      "step": 172500
    },
    {
      "epoch": 0.7,
      "learning_rate": 6.024832176814147e-06,
      "loss": 0.9236,
      "step": 173000
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.984441518365633e-06,
      "loss": 0.8661,
      "step": 173500
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.944050859917119e-06,
      "loss": 0.9123,
      "step": 174000
    },
    {
      "epoch": 0.7,
      "learning_rate": 5.903660201468605e-06,
      "loss": 0.894,
      "step": 174500
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.86326954302009e-06,
      "loss": 0.9081,
      "step": 175000
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.822878884571577e-06,
      "loss": 0.8975,
      "step": 175500
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.782488226123063e-06,
      "loss": 0.8747,
      "step": 176000
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.742097567674548e-06,
      "loss": 0.9273,
      "step": 176500
    },
    {
      "epoch": 0.71,
      "learning_rate": 5.7017069092260345e-06,
      "loss": 0.8736,
      "step": 177000
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.661316250777521e-06,
      "loss": 0.8794,
      "step": 177500
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.6209255923290066e-06,
      "loss": 0.9084,
      "step": 178000
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.580534933880492e-06,
      "loss": 0.8854,
      "step": 178500
    },
    {
      "epoch": 0.72,
      "learning_rate": 5.540144275431979e-06,
      "loss": 0.9172,
      "step": 179000
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.499753616983464e-06,
      "loss": 0.9057,
      "step": 179500
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.459362958534951e-06,
      "loss": 0.9074,
      "step": 180000
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.418972300086437e-06,
      "loss": 0.9053,
      "step": 180500
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.378581641637922e-06,
      "loss": 0.893,
      "step": 181000
    },
    {
      "epoch": 0.73,
      "learning_rate": 5.338190983189408e-06,
      "loss": 0.8762,
      "step": 181500
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.297800324740895e-06,
      "loss": 0.8405,
      "step": 182000
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.2574096662923805e-06,
      "loss": 0.8939,
      "step": 182500
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.217019007843866e-06,
      "loss": 0.8593,
      "step": 183000
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.176628349395352e-06,
      "loss": 0.9004,
      "step": 183500
    },
    {
      "epoch": 0.74,
      "learning_rate": 5.136237690946838e-06,
      "loss": 0.903,
      "step": 184000
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.095847032498325e-06,
      "loss": 0.8813,
      "step": 184500
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.055456374049809e-06,
      "loss": 0.8457,
      "step": 185000
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.015065715601296e-06,
      "loss": 0.8782,
      "step": 185500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.9746750571527815e-06,
      "loss": 0.8841,
      "step": 186000
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.934284398704268e-06,
      "loss": 0.8989,
      "step": 186500
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.893893740255754e-06,
      "loss": 0.8968,
      "step": 187000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.85350308180724e-06,
      "loss": 0.845,
      "step": 187500
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.8131124233587265e-06,
      "loss": 0.85,
      "step": 188000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.772721764910212e-06,
      "loss": 0.8625,
      "step": 188500
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.732331106461698e-06,
      "loss": 0.8833,
      "step": 189000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.691940448013184e-06,
      "loss": 0.9184,
      "step": 189500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.65154978956467e-06,
      "loss": 0.889,
      "step": 190000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.611159131116155e-06,
      "loss": 0.8676,
      "step": 190500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.570768472667642e-06,
      "loss": 0.8962,
      "step": 191000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.530377814219128e-06,
      "loss": 0.8647,
      "step": 191500
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.489987155770614e-06,
      "loss": 0.8796,
      "step": 192000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.4495964973220996e-06,
      "loss": 0.9004,
      "step": 192500
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.409205838873586e-06,
      "loss": 0.8612,
      "step": 193000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.368815180425072e-06,
      "loss": 0.8612,
      "step": 193500
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.328424521976557e-06,
      "loss": 0.9331,
      "step": 194000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.288033863528044e-06,
      "loss": 0.8905,
      "step": 194500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.247643205079529e-06,
      "loss": 0.9151,
      "step": 195000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.207252546631016e-06,
      "loss": 0.8583,
      "step": 195500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.166861888182501e-06,
      "loss": 0.9202,
      "step": 196000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.126471229733988e-06,
      "loss": 0.8658,
      "step": 196500
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.0860805712854735e-06,
      "loss": 0.9142,
      "step": 197000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.045689912836959e-06,
      "loss": 0.8564,
      "step": 197500
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.0052992543884456e-06,
      "loss": 0.8799,
      "step": 198000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.964908595939931e-06,
      "loss": 0.8971,
      "step": 198500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.924517937491417e-06,
      "loss": 0.8952,
      "step": 199000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.884127279042903e-06,
      "loss": 0.8958,
      "step": 199500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.84373662059439e-06,
      "loss": 0.8655,
      "step": 200000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.8033459621458753e-06,
      "loss": 0.9043,
      "step": 200500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.762955303697361e-06,
      "loss": 0.8537,
      "step": 201000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.7225646452488474e-06,
      "loss": 0.8716,
      "step": 201500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.682173986800333e-06,
      "loss": 0.9348,
      "step": 202000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.641783328351819e-06,
      "loss": 0.9048,
      "step": 202500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.601392669903305e-06,
      "loss": 0.8867,
      "step": 203000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.561002011454791e-06,
      "loss": 0.88,
      "step": 203500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.5206113530062767e-06,
      "loss": 0.8901,
      "step": 204000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.480220694557763e-06,
      "loss": 0.8875,
      "step": 204500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.4398300361092492e-06,
      "loss": 0.8641,
      "step": 205000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.399439377660735e-06,
      "loss": 0.8864,
      "step": 205500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.3590487192122205e-06,
      "loss": 0.8575,
      "step": 206000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.318658060763707e-06,
      "loss": 0.8565,
      "step": 206500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.278267402315193e-06,
      "loss": 0.8848,
      "step": 207000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.2378767438666786e-06,
      "loss": 0.879,
      "step": 207500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.197486085418165e-06,
      "loss": 0.8805,
      "step": 208000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.1570954269696507e-06,
      "loss": 0.8843,
      "step": 208500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.1167047685211367e-06,
      "loss": 0.8928,
      "step": 209000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.0763141100726227e-06,
      "loss": 0.8742,
      "step": 209500
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.0359234516241088e-06,
      "loss": 0.8339,
      "step": 210000
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.9955327931755944e-06,
      "loss": 0.8919,
      "step": 210500
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.9551421347270804e-06,
      "loss": 0.8697,
      "step": 211000
    },
    {
      "epoch": 0.85,
      "learning_rate": 2.914751476278567e-06,
      "loss": 0.8697,
      "step": 211500
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.8743608178300525e-06,
      "loss": 0.8874,
      "step": 212000
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.833970159381538e-06,
      "loss": 0.8891,
      "step": 212500
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.7935795009330246e-06,
      "loss": 0.9008,
      "step": 213000
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.7531888424845106e-06,
      "loss": 0.9161,
      "step": 213500
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.7127981840359962e-06,
      "loss": 0.9063,
      "step": 214000
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.6724075255874827e-06,
      "loss": 0.8291,
      "step": 214500
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.6320168671389683e-06,
      "loss": 0.8717,
      "step": 215000
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.5916262086904543e-06,
      "loss": 0.8955,
      "step": 215500
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.55123555024194e-06,
      "loss": 0.8767,
      "step": 216000
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.5108448917934264e-06,
      "loss": 0.8314,
      "step": 216500
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.470454233344912e-06,
      "loss": 0.8677,
      "step": 217000
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.430063574896398e-06,
      "loss": 0.8947,
      "step": 217500
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.389672916447884e-06,
      "loss": 0.8415,
      "step": 218000
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.34928225799937e-06,
      "loss": 0.9029,
      "step": 218500
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.3088915995508558e-06,
      "loss": 0.8588,
      "step": 219000
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.2685009411023422e-06,
      "loss": 0.8956,
      "step": 219500
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.2281102826538283e-06,
      "loss": 0.8821,
      "step": 220000
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.187719624205314e-06,
      "loss": 0.8565,
      "step": 220500
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.1473289657568e-06,
      "loss": 0.8276,
      "step": 221000
    },
    {
      "epoch": 0.89,
      "learning_rate": 2.106938307308286e-06,
      "loss": 0.8415,
      "step": 221500
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.066547648859772e-06,
      "loss": 0.8428,
      "step": 222000
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.026156990411258e-06,
      "loss": 0.8904,
      "step": 222500
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9857663319627436e-06,
      "loss": 0.8784,
      "step": 223000
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9453756735142297e-06,
      "loss": 0.8684,
      "step": 223500
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9049850150657157e-06,
      "loss": 0.9086,
      "step": 224000
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.8645943566172017e-06,
      "loss": 0.8657,
      "step": 224500
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.8242036981686878e-06,
      "loss": 0.8467,
      "step": 225000
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7838130397201736e-06,
      "loss": 0.8263,
      "step": 225500
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7434223812716597e-06,
      "loss": 0.8868,
      "step": 226000
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.7030317228231455e-06,
      "loss": 0.8744,
      "step": 226500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6626410643746315e-06,
      "loss": 0.8546,
      "step": 227000
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6222504059261176e-06,
      "loss": 0.8566,
      "step": 227500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.5818597474776034e-06,
      "loss": 0.9289,
      "step": 228000
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.5414690890290896e-06,
      "loss": 0.8764,
      "step": 228500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.5010784305805752e-06,
      "loss": 0.8833,
      "step": 229000
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.4606877721320615e-06,
      "loss": 0.8887,
      "step": 229500
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.4202971136835475e-06,
      "loss": 0.8791,
      "step": 230000
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.3799064552350334e-06,
      "loss": 0.8815,
      "step": 230500
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.3395157967865194e-06,
      "loss": 0.8936,
      "step": 231000
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.2991251383380052e-06,
      "loss": 0.8494,
      "step": 231500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.2587344798894913e-06,
      "loss": 0.8743,
      "step": 232000
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.2183438214409773e-06,
      "loss": 0.888,
      "step": 232500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.1779531629924631e-06,
      "loss": 0.8898,
      "step": 233000
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.1375625045439492e-06,
      "loss": 0.8638,
      "step": 233500
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0971718460954352e-06,
      "loss": 0.8727,
      "step": 234000
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.056781187646921e-06,
      "loss": 0.853,
      "step": 234500
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.016390529198407e-06,
      "loss": 0.8674,
      "step": 235000
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.75999870749893e-07,
      "loss": 0.8506,
      "step": 235500
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.35609212301379e-07,
      "loss": 0.8516,
      "step": 236000
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.952185538528651e-07,
      "loss": 0.8636,
      "step": 236500
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.54827895404351e-07,
      "loss": 0.8434,
      "step": 237000
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.144372369558369e-07,
      "loss": 0.884,
      "step": 237500
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.740465785073229e-07,
      "loss": 0.89,
      "step": 238000
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.336559200588089e-07,
      "loss": 0.8915,
      "step": 238500
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.932652616102948e-07,
      "loss": 0.8749,
      "step": 239000
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.528746031617808e-07,
      "loss": 0.8963,
      "step": 239500
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.124839447132667e-07,
      "loss": 0.8905,
      "step": 240000
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.720932862647527e-07,
      "loss": 0.8566,
      "step": 240500
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.317026278162387e-07,
      "loss": 0.8718,
      "step": 241000
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.913119693677246e-07,
      "loss": 0.8383,
      "step": 241500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.5092131091921064e-07,
      "loss": 0.8678,
      "step": 242000
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.1053065247069657e-07,
      "loss": 0.811,
      "step": 242500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.701399940221826e-07,
      "loss": 0.8551,
      "step": 243000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.2974933557366854e-07,
      "loss": 0.9153,
      "step": 243500
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.8935867712515453e-07,
      "loss": 0.8721,
      "step": 244000
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.489680186766405e-07,
      "loss": 0.8805,
      "step": 244500
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.0857736022812644e-07,
      "loss": 0.8698,
      "step": 245000
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.6818670177961243e-07,
      "loss": 0.8793,
      "step": 245500
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.277960433310984e-07,
      "loss": 0.8455,
      "step": 246000
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.740538488258437e-08,
      "loss": 0.884,
      "step": 246500
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.7014726434070325e-08,
      "loss": 0.8582,
      "step": 247000
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.624067985556301e-09,
      "loss": 0.8516,
      "step": 247500
    }
  ],
  "max_steps": 247582,
  "num_train_epochs": 1,
  "total_flos": 9502800674291712.0,
  "trial_name": null,
  "trial_params": null
}