{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.2713042787334632,
  "eval_steps": 500,
  "global_step": 96000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.996689311774132e-05,
      "loss": 0.5436,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9933786235482635e-05,
      "loss": 0.5506,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.990067935322395e-05,
      "loss": 0.5253,
      "step": 1500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.986757247096527e-05,
      "loss": 0.5099,
      "step": 2000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.98345318024711e-05,
      "loss": 0.5289,
      "step": 2500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980142492021241e-05,
      "loss": 0.5223,
      "step": 3000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976831803795373e-05,
      "loss": 0.5273,
      "step": 3500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973521115569505e-05,
      "loss": 0.5321,
      "step": 4000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.970210427343636e-05,
      "loss": 0.4965,
      "step": 4500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.966899739117768e-05,
      "loss": 0.5026,
      "step": 5000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9635890508919e-05,
      "loss": 0.5123,
      "step": 5500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9602783626660317e-05,
      "loss": 0.5154,
      "step": 6000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9569676744401626e-05,
      "loss": 0.5258,
      "step": 6500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.953656986214294e-05,
      "loss": 0.5225,
      "step": 7000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.950352919364878e-05,
      "loss": 0.5051,
      "step": 7500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9470422311390094e-05,
      "loss": 0.5136,
      "step": 8000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.943731542913141e-05,
      "loss": 0.5306,
      "step": 8500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.940420854687273e-05,
      "loss": 0.4884,
      "step": 9000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.937110166461404e-05,
      "loss": 0.4916,
      "step": 9500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.933799478235536e-05,
      "loss": 0.5096,
      "step": 10000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.930495411386119e-05,
      "loss": 0.5053,
      "step": 10500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.927184723160251e-05,
      "loss": 0.4981,
      "step": 11000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.923874034934383e-05,
      "loss": 0.4947,
      "step": 11500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.920563346708514e-05,
      "loss": 0.4974,
      "step": 12000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.917252658482645e-05,
      "loss": 0.5037,
      "step": 12500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9139419702567776e-05,
      "loss": 0.4952,
      "step": 13000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.9106379034073605e-05,
      "loss": 0.5019,
      "step": 13500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.907327215181492e-05,
      "loss": 0.5044,
      "step": 14000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.904016526955624e-05,
      "loss": 0.4936,
      "step": 14500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9007058387297554e-05,
      "loss": 0.4975,
      "step": 15000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.897395150503887e-05,
      "loss": 0.4725,
      "step": 15500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.8940844622780186e-05,
      "loss": 0.4622,
      "step": 16000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.89077377405215e-05,
      "loss": 0.4824,
      "step": 16500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.887469707202734e-05,
      "loss": 0.5536,
      "step": 17000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.884159018976865e-05,
      "loss": 0.4653,
      "step": 17500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.8808549521274483e-05,
      "loss": 0.5132,
      "step": 18000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.87754426390158e-05,
      "loss": 0.5161,
      "step": 18500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.8742335756757116e-05,
      "loss": 0.4661,
      "step": 19000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.870922887449843e-05,
      "loss": 0.4929,
      "step": 19500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.867618820600426e-05,
      "loss": 0.4834,
      "step": 20000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.8643081323745584e-05,
      "loss": 0.4791,
      "step": 20500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.86099744414869e-05,
      "loss": 0.5235,
      "step": 21000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8576867559228217e-05,
      "loss": 0.4893,
      "step": 21500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.854376067696953e-05,
      "loss": 0.4616,
      "step": 22000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.851065379471085e-05,
      "loss": 0.4951,
      "step": 22500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8477546912452165e-05,
      "loss": 0.4882,
      "step": 23000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8444440030193475e-05,
      "loss": 0.5253,
      "step": 23500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.84113331479348e-05,
      "loss": 0.4815,
      "step": 24000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8378226265676114e-05,
      "loss": 0.4928,
      "step": 24500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.834518559718195e-05,
      "loss": 0.503,
      "step": 25000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.831207871492326e-05,
      "loss": 0.481,
      "step": 25500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.8278971832664575e-05,
      "loss": 0.4613,
      "step": 26000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.824586495040589e-05,
      "loss": 0.4727,
      "step": 26500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.821282428191173e-05,
      "loss": 0.4639,
      "step": 27000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.8179717399653044e-05,
      "loss": 0.4989,
      "step": 27500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.814661051739436e-05,
      "loss": 0.4595,
      "step": 28000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8113503635135676e-05,
      "loss": 0.4803,
      "step": 28500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8080396752876985e-05,
      "loss": 0.4768,
      "step": 29000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.804735608438282e-05,
      "loss": 0.4642,
      "step": 29500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.801424920212414e-05,
      "loss": 0.4939,
      "step": 30000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.798114231986546e-05,
      "loss": 0.4717,
      "step": 30500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.794803543760677e-05,
      "loss": 0.4689,
      "step": 31000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.7914994769112606e-05,
      "loss": 0.4973,
      "step": 31500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.788188788685392e-05,
      "loss": 0.4369,
      "step": 32000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.784878100459524e-05,
      "loss": 0.476,
      "step": 32500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.7815674122336554e-05,
      "loss": 0.4594,
      "step": 33000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.778256724007787e-05,
      "loss": 0.4863,
      "step": 33500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.77495265715837e-05,
      "loss": 0.4514,
      "step": 34000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.771641968932502e-05,
      "loss": 0.4843,
      "step": 34500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.768331280706633e-05,
      "loss": 0.4473,
      "step": 35000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.765020592480765e-05,
      "loss": 0.4865,
      "step": 35500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.761709904254897e-05,
      "loss": 0.4509,
      "step": 36000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.75840583740548e-05,
      "loss": 0.4599,
      "step": 36500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.7550951491796117e-05,
      "loss": 0.4589,
      "step": 37000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.751784460953743e-05,
      "loss": 0.4565,
      "step": 37500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.748473772727875e-05,
      "loss": 0.4489,
      "step": 38000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.7451697058784585e-05,
      "loss": 0.4637,
      "step": 38500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.74185901765259e-05,
      "loss": 0.4596,
      "step": 39000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.738548329426721e-05,
      "loss": 0.4647,
      "step": 39500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.7352376412008533e-05,
      "loss": 0.4587,
      "step": 40000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.731926952974985e-05,
      "loss": 0.4866,
      "step": 40500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.728616264749116e-05,
      "loss": 0.4546,
      "step": 41000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.7253121978996995e-05,
      "loss": 0.4345,
      "step": 41500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.722008131050283e-05,
      "loss": 0.4867,
      "step": 42000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.718697442824415e-05,
      "loss": 0.4494,
      "step": 42500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.715386754598546e-05,
      "loss": 0.4447,
      "step": 43000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.712076066372678e-05,
      "loss": 0.4674,
      "step": 43500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.7087653781468096e-05,
      "loss": 0.4343,
      "step": 44000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.705454689920941e-05,
      "loss": 0.4304,
      "step": 44500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.702144001695072e-05,
      "loss": 0.4324,
      "step": 45000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.6988333134692044e-05,
      "loss": 0.4772,
      "step": 45500
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.695522625243336e-05,
      "loss": 0.4649,
      "step": 46000
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.6922185583939196e-05,
      "loss": 0.4497,
      "step": 46500
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.6889078701680506e-05,
      "loss": 0.4224,
      "step": 47000
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.685597181942182e-05,
      "loss": 0.4572,
      "step": 47500
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.682286493716314e-05,
      "loss": 0.4737,
      "step": 48000
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.6789824268668974e-05,
      "loss": 0.4895,
      "step": 48500
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.675671738641029e-05,
      "loss": 0.4577,
      "step": 49000
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.6723610504151606e-05,
      "loss": 0.4541,
      "step": 49500
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.669050362189292e-05,
      "loss": 0.4484,
      "step": 50000
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.665739673963423e-05,
      "loss": 0.4472,
      "step": 50500
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.662435607114007e-05,
      "loss": 0.4398,
      "step": 51000
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.6591249188881384e-05,
      "loss": 0.4781,
      "step": 51500
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.655814230662271e-05,
      "loss": 0.4201,
      "step": 52000
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.6525035424364016e-05,
      "loss": 0.4385,
      "step": 52500
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.649192854210533e-05,
      "loss": 0.4245,
      "step": 53000
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.645882165984665e-05,
      "loss": 0.4525,
      "step": 53500
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.6425780991352485e-05,
      "loss": 0.4499,
      "step": 54000
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.63926741090938e-05,
      "loss": 0.4356,
      "step": 54500
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.635956722683512e-05,
      "loss": 0.4598,
      "step": 55000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.632646034457643e-05,
      "loss": 0.4508,
      "step": 55500
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.629341967608227e-05,
      "loss": 0.4536,
      "step": 56000
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.6260312793823585e-05,
      "loss": 0.4486,
      "step": 56500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.6227205911564895e-05,
      "loss": 0.4463,
      "step": 57000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.619409902930622e-05,
      "loss": 0.4503,
      "step": 57500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.6160992147047534e-05,
      "loss": 0.4612,
      "step": 58000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.612795147855336e-05,
      "loss": 0.4388,
      "step": 58500
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.609484459629468e-05,
      "loss": 0.4547,
      "step": 59000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.6061737714035996e-05,
      "loss": 0.4624,
      "step": 59500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.602863083177731e-05,
      "loss": 0.4479,
      "step": 60000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.599552394951863e-05,
      "loss": 0.4512,
      "step": 60500
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.5962417067259944e-05,
      "loss": 0.4451,
      "step": 61000
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.592931018500126e-05,
      "loss": 0.4318,
      "step": 61500
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.5896269516507096e-05,
      "loss": 0.4188,
      "step": 62000
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.5863162634248406e-05,
      "loss": 0.4371,
      "step": 62500
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.583005575198973e-05,
      "loss": 0.4519,
      "step": 63000
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.5796948869731045e-05,
      "loss": 0.4491,
      "step": 63500
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.5763841987472354e-05,
      "loss": 0.4397,
      "step": 64000
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.573073510521367e-05,
      "loss": 0.4308,
      "step": 64500
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.5697628222954994e-05,
      "loss": 0.4286,
      "step": 65000
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.566458755446082e-05,
      "loss": 0.4587,
      "step": 65500
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.563148067220214e-05,
      "loss": 0.4436,
      "step": 66000
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.5598373789943455e-05,
      "loss": 0.4672,
      "step": 66500
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.556526690768477e-05,
      "loss": 0.4294,
      "step": 67000
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.553216002542609e-05,
      "loss": 0.4454,
      "step": 67500
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.5499053143167404e-05,
      "loss": 0.4225,
      "step": 68000
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.546594626090872e-05,
      "loss": 0.4256,
      "step": 68500
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.5432905592414556e-05,
      "loss": 0.4179,
      "step": 69000
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.539979871015587e-05,
      "loss": 0.4521,
      "step": 69500
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.536669182789718e-05,
      "loss": 0.4415,
      "step": 70000
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.5333584945638504e-05,
      "loss": 0.4102,
      "step": 70500
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.530054427714433e-05,
      "loss": 0.3905,
      "step": 71000
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.5267437394885656e-05,
      "loss": 0.4331,
      "step": 71500
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.5234330512626966e-05,
      "loss": 0.4289,
      "step": 72000
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.520122363036828e-05,
      "loss": 0.4246,
      "step": 72500
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.51681167481096e-05,
      "loss": 0.4141,
      "step": 73000
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.5135076079615434e-05,
      "loss": 0.4365,
      "step": 73500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.510196919735675e-05,
      "loss": 0.4117,
      "step": 74000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.5068862315098066e-05,
      "loss": 0.4113,
      "step": 74500
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.503575543283938e-05,
      "loss": 0.4297,
      "step": 75000
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.500271476434522e-05,
      "loss": 0.4252,
      "step": 75500
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.496960788208653e-05,
      "loss": 0.4208,
      "step": 76000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.4936500999827844e-05,
      "loss": 0.43,
      "step": 76500
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.490339411756917e-05,
      "loss": 0.3911,
      "step": 77000
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.4870353449074996e-05,
      "loss": 0.4056,
      "step": 77500
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.483724656681631e-05,
      "loss": 0.4315,
      "step": 78000
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.480413968455763e-05,
      "loss": 0.4423,
      "step": 78500
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.4771032802298945e-05,
      "loss": 0.4177,
      "step": 79000
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.473792592004026e-05,
      "loss": 0.3952,
      "step": 79500
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.470481903778158e-05,
      "loss": 0.3925,
      "step": 80000
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.4671778369287406e-05,
      "loss": 0.4147,
      "step": 80500
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.463867148702873e-05,
      "loss": 0.4091,
      "step": 81000
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.460556460477004e-05,
      "loss": 0.41,
      "step": 81500
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.4572457722511355e-05,
      "loss": 0.3953,
      "step": 82000
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.453935084025268e-05,
      "loss": 0.4294,
      "step": 82500
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.450631017175851e-05,
      "loss": 0.3982,
      "step": 83000
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.447320328949982e-05,
      "loss": 0.4094,
      "step": 83500
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.444009640724114e-05,
      "loss": 0.3892,
      "step": 84000
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.4406989524982456e-05,
      "loss": 0.3843,
      "step": 84500
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.437394885648829e-05,
      "loss": 0.4414,
      "step": 85000
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.434084197422961e-05,
      "loss": 0.4325,
      "step": 85500
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.430773509197092e-05,
      "loss": 0.4319,
      "step": 86000
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.427462820971224e-05,
      "loss": 0.4216,
      "step": 86500
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.4241521327453556e-05,
      "loss": 0.4205,
      "step": 87000
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.4208414445194866e-05,
      "loss": 0.4056,
      "step": 87500
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.41753737767007e-05,
      "loss": 0.4287,
      "step": 88000
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.414226689444202e-05,
      "loss": 0.4034,
      "step": 88500
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.4109160012183334e-05,
      "loss": 0.3886,
      "step": 89000
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.407605312992465e-05,
      "loss": 0.4426,
      "step": 89500
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.4042946247665966e-05,
      "loss": 0.4298,
      "step": 90000
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.40099055791718e-05,
      "loss": 0.4066,
      "step": 90500
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.397679869691312e-05,
      "loss": 0.4017,
      "step": 91000
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.394369181465443e-05,
      "loss": 0.4057,
      "step": 91500
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.391058493239575e-05,
      "loss": 0.3921,
      "step": 92000
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.387747805013707e-05,
      "loss": 0.3989,
      "step": 92500
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.3844371167878377e-05,
      "loss": 0.3896,
      "step": 93000
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.381133049938421e-05,
      "loss": 0.3896,
      "step": 93500
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.377822361712553e-05,
      "loss": 0.407,
      "step": 94000
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.3745116734866845e-05,
      "loss": 0.4147,
      "step": 94500
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.371200985260816e-05,
      "loss": 0.3921,
      "step": 95000
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.367890297034948e-05,
      "loss": 0.3924,
      "step": 95500
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.3645796088090793e-05,
      "loss": 0.3874,
      "step": 96000
    }
  ],
  "logging_steps": 500,
  "max_steps": 755130,
  "num_train_epochs": 10,
  "save_steps": 2000,
  "total_flos": 3.56621066698752e+17,
  "trial_name": null,
  "trial_params": null
}