{
  "best_metric": 1.2051972150802612,
  "best_model_checkpoint": "gator/output_mlm/checkpoint-225000",
  "epoch": 3.0,
  "global_step": 233133,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9892765074013544e-05,
      "loss": 1.6961,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.978553014802709e-05,
      "loss": 1.6895,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9678295222040635e-05,
      "loss": 1.6703,
      "step": 1500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9571060296054184e-05,
      "loss": 1.6708,
      "step": 2000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.946382537006773e-05,
      "loss": 1.6626,
      "step": 2500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.935659044408128e-05,
      "loss": 1.659,
      "step": 3000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9249355518094824e-05,
      "loss": 1.652,
      "step": 3500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.914212059210837e-05,
      "loss": 1.6472,
      "step": 4000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9034885666121915e-05,
      "loss": 1.6317,
      "step": 4500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.8927650740135464e-05,
      "loss": 1.6339,
      "step": 5000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8820415814149006e-05,
      "loss": 1.6211,
      "step": 5500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.8713180888162555e-05,
      "loss": 1.632,
      "step": 6000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.86059459621761e-05,
      "loss": 1.624,
      "step": 6500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8498711036189646e-05,
      "loss": 1.6209,
      "step": 7000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8391476110203195e-05,
      "loss": 1.6198,
      "step": 7500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.828424118421674e-05,
      "loss": 1.6184,
      "step": 8000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8177006258230286e-05,
      "loss": 1.6141,
      "step": 8500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.806977133224383e-05,
      "loss": 1.6055,
      "step": 9000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.7962536406257376e-05,
      "loss": 1.6018,
      "step": 9500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.785530148027092e-05,
      "loss": 1.6092,
      "step": 10000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.774806655428447e-05,
      "loss": 1.5921,
      "step": 10500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.764083162829801e-05,
      "loss": 1.5968,
      "step": 11000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.753359670231156e-05,
      "loss": 1.6,
      "step": 11500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.742636177632511e-05,
      "loss": 1.5957,
      "step": 12000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.731912685033865e-05,
      "loss": 1.5944,
      "step": 12500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.72118919243522e-05,
      "loss": 1.5849,
      "step": 13000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.710465699836574e-05,
      "loss": 1.5823,
      "step": 13500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.699742207237929e-05,
      "loss": 1.5947,
      "step": 14000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.689018714639283e-05,
      "loss": 1.5874,
      "step": 14500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.678295222040638e-05,
      "loss": 1.5832,
      "step": 15000
    },
    {
      "epoch": 0.19,
      "eval_loss": 1.4992027282714844,
      "eval_runtime": 3385.2625,
      "eval_samples_per_second": 39.094,
      "eval_steps_per_second": 4.887,
      "step": 15000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.667571729441992e-05,
      "loss": 1.5678,
      "step": 15500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.656848236843347e-05,
      "loss": 1.5862,
      "step": 16000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.646124744244702e-05,
      "loss": 1.5828,
      "step": 16500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.635401251646057e-05,
      "loss": 1.5777,
      "step": 17000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.624677759047411e-05,
      "loss": 1.5722,
      "step": 17500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.613954266448765e-05,
      "loss": 1.5723,
      "step": 18000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.60323077385012e-05,
      "loss": 1.5827,
      "step": 18500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.5925072812514744e-05,
      "loss": 1.572,
      "step": 19000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.581783788652829e-05,
      "loss": 1.565,
      "step": 19500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5710602960541835e-05,
      "loss": 1.5628,
      "step": 20000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5603368034555384e-05,
      "loss": 1.5598,
      "step": 20500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5496133108568926e-05,
      "loss": 1.5571,
      "step": 21000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.538889818258248e-05,
      "loss": 1.5546,
      "step": 21500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.5281663256596023e-05,
      "loss": 1.5578,
      "step": 22000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.517442833060957e-05,
      "loss": 1.554,
      "step": 22500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.5067193404623114e-05,
      "loss": 1.5488,
      "step": 23000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.495995847863666e-05,
      "loss": 1.5512,
      "step": 23500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4852723552650205e-05,
      "loss": 1.5636,
      "step": 24000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4745488626663754e-05,
      "loss": 1.5437,
      "step": 24500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.4638253700677296e-05,
      "loss": 1.5515,
      "step": 25000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.453101877469084e-05,
      "loss": 1.5587,
      "step": 25500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.4423783848704394e-05,
      "loss": 1.5413,
      "step": 26000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.4316548922717936e-05,
      "loss": 1.5381,
      "step": 26500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.4209313996731485e-05,
      "loss": 1.5483,
      "step": 27000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.410207907074503e-05,
      "loss": 1.5436,
      "step": 27500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.3994844144758576e-05,
      "loss": 1.5457,
      "step": 28000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.388760921877212e-05,
      "loss": 1.5367,
      "step": 28500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.378037429278567e-05,
      "loss": 1.5404,
      "step": 29000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.367313936679921e-05,
      "loss": 1.5414,
      "step": 29500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.356590444081276e-05,
      "loss": 1.5325,
      "step": 30000
    },
    {
      "epoch": 0.39,
      "eval_loss": 1.4652618169784546,
      "eval_runtime": 3382.9946,
      "eval_samples_per_second": 39.12,
      "eval_steps_per_second": 4.89,
      "step": 30000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3458669514826307e-05,
      "loss": 1.5386,
      "step": 30500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.335143458883985e-05,
      "loss": 1.5513,
      "step": 31000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.32441996628534e-05,
      "loss": 1.5384,
      "step": 31500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.313696473686694e-05,
      "loss": 1.5381,
      "step": 32000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.302972981088049e-05,
      "loss": 1.5429,
      "step": 32500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.292249488489403e-05,
      "loss": 1.5325,
      "step": 33000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.281525995890758e-05,
      "loss": 1.5388,
      "step": 33500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.270802503292112e-05,
      "loss": 1.5353,
      "step": 34000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.260079010693467e-05,
      "loss": 1.5311,
      "step": 34500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.249355518094821e-05,
      "loss": 1.5356,
      "step": 35000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.238632025496176e-05,
      "loss": 1.5347,
      "step": 35500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.227908532897531e-05,
      "loss": 1.5198,
      "step": 36000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.217185040298885e-05,
      "loss": 1.5196,
      "step": 36500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.20646154770024e-05,
      "loss": 1.5173,
      "step": 37000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.195738055101594e-05,
      "loss": 1.52,
      "step": 37500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.185014562502949e-05,
      "loss": 1.5155,
      "step": 38000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1742910699043034e-05,
      "loss": 1.512,
      "step": 38500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.163567577305658e-05,
      "loss": 1.5146,
      "step": 39000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.1528440847070125e-05,
      "loss": 1.5138,
      "step": 39500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.142120592108368e-05,
      "loss": 1.5101,
      "step": 40000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.131397099509722e-05,
      "loss": 1.4977,
      "step": 40500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.120673606911077e-05,
      "loss": 1.5153,
      "step": 41000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.1099501143124314e-05,
      "loss": 1.5053,
      "step": 41500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.0992266217137856e-05,
      "loss": 1.5088,
      "step": 42000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.0885031291151405e-05,
      "loss": 1.5111,
      "step": 42500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.077779636516495e-05,
      "loss": 1.5069,
      "step": 43000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.0670561439178496e-05,
      "loss": 1.5085,
      "step": 43500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.056332651319204e-05,
      "loss": 1.5064,
      "step": 44000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.0456091587205586e-05,
      "loss": 1.5004,
      "step": 44500
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.0348856661219135e-05,
      "loss": 1.4979,
      "step": 45000
    },
    {
      "epoch": 0.58,
      "eval_loss": 1.4358571767807007,
      "eval_runtime": 3383.7351,
      "eval_samples_per_second": 39.112,
      "eval_steps_per_second": 4.889,
      "step": 45000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.0241621735232684e-05,
      "loss": 1.5037,
      "step": 45500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.0134386809246226e-05,
      "loss": 1.4942,
      "step": 46000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.0027151883259775e-05,
      "loss": 1.4907,
      "step": 46500
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.991991695727332e-05,
      "loss": 1.4985,
      "step": 47000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9812682031286866e-05,
      "loss": 1.4996,
      "step": 47500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.970544710530041e-05,
      "loss": 1.4959,
      "step": 48000
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.959821217931396e-05,
      "loss": 1.4876,
      "step": 48500
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.94909772533275e-05,
      "loss": 1.5026,
      "step": 49000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.938374232734105e-05,
      "loss": 1.4985,
      "step": 49500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.92765074013546e-05,
      "loss": 1.4923,
      "step": 50000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.916927247536814e-05,
      "loss": 1.5005,
      "step": 50500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.906203754938169e-05,
      "loss": 1.5018,
      "step": 51000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.895480262339523e-05,
      "loss": 1.4829,
      "step": 51500
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.884756769740878e-05,
      "loss": 1.4858,
      "step": 52000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.874033277142232e-05,
      "loss": 1.497,
      "step": 52500
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.863309784543587e-05,
      "loss": 1.4849,
      "step": 53000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.852586291944941e-05,
      "loss": 1.4851,
      "step": 53500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.841862799346296e-05,
      "loss": 1.4893,
      "step": 54000
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.831139306747651e-05,
      "loss": 1.4887,
      "step": 54500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.820415814149005e-05,
      "loss": 1.4805,
      "step": 55000
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.80969232155036e-05,
      "loss": 1.4815,
      "step": 55500
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.798968828951714e-05,
      "loss": 1.477,
      "step": 56000
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.788245336353069e-05,
      "loss": 1.4791,
      "step": 56500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.777521843754423e-05,
      "loss": 1.4915,
      "step": 57000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.766798351155778e-05,
      "loss": 1.4935,
      "step": 57500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.7560748585571324e-05,
      "loss": 1.4711,
      "step": 58000
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.745351365958487e-05,
      "loss": 1.4789,
      "step": 58500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.734627873359842e-05,
      "loss": 1.471,
      "step": 59000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.7239043807611964e-05,
      "loss": 1.4733,
      "step": 59500
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.713180888162551e-05,
      "loss": 1.4715,
      "step": 60000
    },
    {
      "epoch": 0.77,
      "eval_loss": 1.4038593769073486,
      "eval_runtime": 3383.4345,
      "eval_samples_per_second": 39.115,
      "eval_steps_per_second": 4.889,
      "step": 60000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.7024573955639055e-05,
      "loss": 1.4746,
      "step": 60500
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.6917339029652604e-05,
      "loss": 1.4745,
      "step": 61000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.6810104103666146e-05,
      "loss": 1.4699,
      "step": 61500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.6702869177679695e-05,
      "loss": 1.4672,
      "step": 62000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.659563425169324e-05,
      "loss": 1.4654,
      "step": 62500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.6488399325706786e-05,
      "loss": 1.4628,
      "step": 63000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6381164399720335e-05,
      "loss": 1.4698,
      "step": 63500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.6273929473733884e-05,
      "loss": 1.4667,
      "step": 64000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.6166694547747426e-05,
      "loss": 1.4591,
      "step": 64500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.6059459621760974e-05,
      "loss": 1.4674,
      "step": 65000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.5952224695774517e-05,
      "loss": 1.467,
      "step": 65500
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.5844989769788065e-05,
      "loss": 1.4652,
      "step": 66000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.573775484380161e-05,
      "loss": 1.4609,
      "step": 66500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.563051991781515e-05,
      "loss": 1.4657,
      "step": 67000
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.55232849918287e-05,
      "loss": 1.4614,
      "step": 67500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.541605006584225e-05,
      "loss": 1.4527,
      "step": 68000
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.5308815139855796e-05,
      "loss": 1.452,
      "step": 68500
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.520158021386934e-05,
      "loss": 1.4666,
      "step": 69000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.509434528788289e-05,
      "loss": 1.4544,
      "step": 69500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.498711036189643e-05,
      "loss": 1.448,
      "step": 70000
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.487987543590998e-05,
      "loss": 1.4554,
      "step": 70500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.477264050992352e-05,
      "loss": 1.4451,
      "step": 71000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.466540558393707e-05,
      "loss": 1.4546,
      "step": 71500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.455817065795061e-05,
      "loss": 1.4573,
      "step": 72000
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.445093573196416e-05,
      "loss": 1.4486,
      "step": 72500
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.434370080597771e-05,
      "loss": 1.4474,
      "step": 73000
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.423646587999125e-05,
      "loss": 1.456,
      "step": 73500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.41292309540048e-05,
      "loss": 1.4537,
      "step": 74000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.402199602801834e-05,
      "loss": 1.4532,
      "step": 74500
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.391476110203189e-05,
      "loss": 1.4448,
      "step": 75000
    },
    {
      "epoch": 0.97,
      "eval_loss": 1.3876514434814453,
      "eval_runtime": 3381.619,
      "eval_samples_per_second": 39.136,
      "eval_steps_per_second": 4.892,
      "step": 75000
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.380752617604543e-05,
      "loss": 1.4447,
      "step": 75500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.370029125005898e-05,
      "loss": 1.4423,
      "step": 76000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.3593056324072524e-05,
      "loss": 1.4477,
      "step": 76500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.348582139808607e-05,
      "loss": 1.4408,
      "step": 77000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.337858647209962e-05,
      "loss": 1.45,
      "step": 77500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3271351546113163e-05,
      "loss": 1.4316,
      "step": 78000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.316411662012671e-05,
      "loss": 1.4364,
      "step": 78500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.3056881694140254e-05,
      "loss": 1.4296,
      "step": 79000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.29496467681538e-05,
      "loss": 1.426,
      "step": 79500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.2842411842167345e-05,
      "loss": 1.4301,
      "step": 80000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.2735176916180894e-05,
      "loss": 1.436,
      "step": 80500
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.2627941990194436e-05,
      "loss": 1.4253,
      "step": 81000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.2520707064207985e-05,
      "loss": 1.4303,
      "step": 81500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.2413472138221534e-05,
      "loss": 1.4286,
      "step": 82000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.230623721223508e-05,
      "loss": 1.4331,
      "step": 82500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.2199002286248625e-05,
      "loss": 1.4307,
      "step": 83000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.2091767360262174e-05,
      "loss": 1.419,
      "step": 83500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.1984532434275716e-05,
      "loss": 1.422,
      "step": 84000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.187729750828926e-05,
      "loss": 1.4204,
      "step": 84500
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.177006258230281e-05,
      "loss": 1.4301,
      "step": 85000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.166282765631635e-05,
      "loss": 1.4126,
      "step": 85500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.15555927303299e-05,
      "loss": 1.4248,
      "step": 86000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.144835780434344e-05,
      "loss": 1.423,
      "step": 86500
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.1341122878356995e-05,
      "loss": 1.4192,
      "step": 87000
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.123388795237054e-05,
      "loss": 1.418,
      "step": 87500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.1126653026384086e-05,
      "loss": 1.4194,
      "step": 88000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.101941810039763e-05,
      "loss": 1.4,
      "step": 88500
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.091218317441118e-05,
      "loss": 1.4023,
      "step": 89000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.080494824842472e-05,
      "loss": 1.4258,
      "step": 89500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.069771332243827e-05,
      "loss": 1.4191,
      "step": 90000
    },
    {
      "epoch": 1.16,
      "eval_loss": 1.3602887392044067,
      "eval_runtime": 3380.9917,
      "eval_samples_per_second": 39.144,
      "eval_steps_per_second": 4.893,
      "step": 90000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.059047839645181e-05,
      "loss": 1.413,
      "step": 90500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0483243470465356e-05,
      "loss": 1.4107,
      "step": 91000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0376008544478908e-05,
      "loss": 1.4151,
      "step": 91500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.0268773618492454e-05,
      "loss": 1.4119,
      "step": 92000
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0161538692505996e-05,
      "loss": 1.4049,
      "step": 92500
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.005430376651954e-05,
      "loss": 1.4124,
      "step": 93000
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.9947068840533087e-05,
      "loss": 1.4025,
      "step": 93500
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9839833914546632e-05,
      "loss": 1.4116,
      "step": 94000
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9732598988560177e-05,
      "loss": 1.4148,
      "step": 94500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9625364062573723e-05,
      "loss": 1.4042,
      "step": 95000
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.951812913658727e-05,
      "loss": 1.4009,
      "step": 95500
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9410894210600814e-05,
      "loss": 1.3971,
      "step": 96000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9303659284614366e-05,
      "loss": 1.4069,
      "step": 96500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.919642435862791e-05,
      "loss": 1.4172,
      "step": 97000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9089189432641457e-05,
      "loss": 1.3947,
      "step": 97500
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.8981954506655003e-05,
      "loss": 1.3998,
      "step": 98000
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.8874719580668548e-05,
      "loss": 1.4033,
      "step": 98500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.8767484654682093e-05,
      "loss": 1.3905,
      "step": 99000
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.866024972869564e-05,
      "loss": 1.3934,
      "step": 99500
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.855301480270918e-05,
      "loss": 1.3938,
      "step": 100000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8445779876722727e-05,
      "loss": 1.4021,
      "step": 100500
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.833854495073628e-05,
      "loss": 1.3977,
      "step": 101000
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.8231310024749824e-05,
      "loss": 1.4034,
      "step": 101500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.812407509876337e-05,
      "loss": 1.3978,
      "step": 102000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.8016840172776915e-05,
      "loss": 1.3985,
      "step": 102500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.790960524679046e-05,
      "loss": 1.3881,
      "step": 103000
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7802370320804006e-05,
      "loss": 1.3916,
      "step": 103500
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.769513539481755e-05,
      "loss": 1.3981,
      "step": 104000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7587900468831097e-05,
      "loss": 1.392,
      "step": 104500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7480665542844643e-05,
      "loss": 1.3988,
      "step": 105000
    },
    {
      "epoch": 1.35,
      "eval_loss": 1.3425246477127075,
      "eval_runtime": 3384.4417,
      "eval_samples_per_second": 39.104,
      "eval_steps_per_second": 4.888,
      "step": 105000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.737343061685819e-05,
      "loss": 1.3916,
      "step": 105500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7266195690871737e-05,
      "loss": 1.3971,
      "step": 106000
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7158960764885282e-05,
      "loss": 1.3993,
      "step": 106500
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.7051725838898828e-05,
      "loss": 1.3899,
      "step": 107000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6944490912912373e-05,
      "loss": 1.3771,
      "step": 107500
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.683725598692592e-05,
      "loss": 1.3988,
      "step": 108000
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6730021060939464e-05,
      "loss": 1.3913,
      "step": 108500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.662278613495301e-05,
      "loss": 1.3929,
      "step": 109000
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6515551208966555e-05,
      "loss": 1.3879,
      "step": 109500
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.64083162829801e-05,
      "loss": 1.385,
      "step": 110000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.630108135699365e-05,
      "loss": 1.3798,
      "step": 110500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6193846431007195e-05,
      "loss": 1.3844,
      "step": 111000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.608661150502074e-05,
      "loss": 1.3789,
      "step": 111500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.5979376579034286e-05,
      "loss": 1.3734,
      "step": 112000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.587214165304783e-05,
      "loss": 1.3851,
      "step": 112500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5764906727061377e-05,
      "loss": 1.3802,
      "step": 113000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5657671801074922e-05,
      "loss": 1.3762,
      "step": 113500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5550436875088468e-05,
      "loss": 1.3746,
      "step": 114000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5443201949102013e-05,
      "loss": 1.38,
      "step": 114500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5335967023115565e-05,
      "loss": 1.3842,
      "step": 115000
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.522873209712911e-05,
      "loss": 1.3718,
      "step": 115500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5121497171142656e-05,
      "loss": 1.3839,
      "step": 116000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.5014262245156202e-05,
      "loss": 1.375,
      "step": 116500
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4907027319169744e-05,
      "loss": 1.3647,
      "step": 117000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.479979239318329e-05,
      "loss": 1.3787,
      "step": 117500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4692557467196835e-05,
      "loss": 1.3692,
      "step": 118000
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4585322541210384e-05,
      "loss": 1.371,
      "step": 118500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.447808761522393e-05,
      "loss": 1.3635,
      "step": 119000
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4370852689237475e-05,
      "loss": 1.368,
      "step": 119500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.426361776325102e-05,
      "loss": 1.3699,
      "step": 120000
    },
    {
      "epoch": 1.54,
      "eval_loss": 1.323049545288086,
      "eval_runtime": 3382.1272,
      "eval_samples_per_second": 39.13,
      "eval_steps_per_second": 4.891,
      "step": 120000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4156382837264566e-05,
      "loss": 1.3738,
      "step": 120500
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.4049147911278114e-05,
      "loss": 1.3661,
      "step": 121000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.394191298529166e-05,
      "loss": 1.3682,
      "step": 121500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3834678059305205e-05,
      "loss": 1.3731,
      "step": 122000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.372744313331875e-05,
      "loss": 1.3668,
      "step": 122500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3620208207332296e-05,
      "loss": 1.3665,
      "step": 123000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3512973281345842e-05,
      "loss": 1.3606,
      "step": 123500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3405738355359387e-05,
      "loss": 1.3559,
      "step": 124000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.3298503429372933e-05,
      "loss": 1.3663,
      "step": 124500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3191268503386478e-05,
      "loss": 1.3656,
      "step": 125000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3084033577400027e-05,
      "loss": 1.3765,
      "step": 125500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.2976798651413573e-05,
      "loss": 1.3688,
      "step": 126000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2869563725427118e-05,
      "loss": 1.3631,
      "step": 126500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2762328799440664e-05,
      "loss": 1.3493,
      "step": 127000
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.265509387345421e-05,
      "loss": 1.36,
      "step": 127500
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2547858947467758e-05,
      "loss": 1.3563,
      "step": 128000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2440624021481303e-05,
      "loss": 1.3518,
      "step": 128500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.233338909549485e-05,
      "loss": 1.3553,
      "step": 129000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.222615416950839e-05,
      "loss": 1.3541,
      "step": 129500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.211891924352194e-05,
      "loss": 1.3482,
      "step": 130000
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.2011684317535485e-05,
      "loss": 1.3502,
      "step": 130500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.190444939154903e-05,
      "loss": 1.352,
      "step": 131000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1797214465562576e-05,
      "loss": 1.3555,
      "step": 131500
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.168997953957612e-05,
      "loss": 1.3587,
      "step": 132000
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.158274461358967e-05,
      "loss": 1.3456,
      "step": 132500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1475509687603216e-05,
      "loss": 1.3486,
      "step": 133000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.136827476161676e-05,
      "loss": 1.3491,
      "step": 133500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1261039835630307e-05,
      "loss": 1.3479,
      "step": 134000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1153804909643852e-05,
      "loss": 1.3405,
      "step": 134500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1046569983657398e-05,
      "loss": 1.3493,
      "step": 135000
    },
    {
      "epoch": 1.74,
      "eval_loss": 1.301226258277893,
      "eval_runtime": 3380.5336,
      "eval_samples_per_second": 39.149,
      "eval_steps_per_second": 4.894,
      "step": 135000
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.0939335057670943e-05,
      "loss": 1.3478,
      "step": 135500
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.083210013168449e-05,
      "loss": 1.3441,
      "step": 136000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0724865205698034e-05,
      "loss": 1.3483,
      "step": 136500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.061763027971158e-05,
      "loss": 1.3367,
      "step": 137000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.051039535372513e-05,
      "loss": 1.3468,
      "step": 137500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0403160427738674e-05,
      "loss": 1.3434,
      "step": 138000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.029592550175222e-05,
      "loss": 1.3377,
      "step": 138500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0188690575765765e-05,
      "loss": 1.3383,
      "step": 139000
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0081455649779314e-05,
      "loss": 1.3427,
      "step": 139500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.997422072379286e-05,
      "loss": 1.3477,
      "step": 140000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9866985797806405e-05,
      "loss": 1.3364,
      "step": 140500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.975975087181995e-05,
      "loss": 1.3447,
      "step": 141000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9652515945833492e-05,
      "loss": 1.3365,
      "step": 141500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.954528101984704e-05,
      "loss": 1.3231,
      "step": 142000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9438046093860587e-05,
      "loss": 1.3362,
      "step": 142500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9330811167874132e-05,
      "loss": 1.3415,
      "step": 143000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9223576241887678e-05,
      "loss": 1.3323,
      "step": 143500
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9116341315901223e-05,
      "loss": 1.3316,
      "step": 144000
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.9009106389914772e-05,
      "loss": 1.3304,
      "step": 144500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8901871463928317e-05,
      "loss": 1.329,
      "step": 145000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8794636537941863e-05,
      "loss": 1.3268,
      "step": 145500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8687401611955408e-05,
      "loss": 1.337,
      "step": 146000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8580166685968957e-05,
      "loss": 1.3277,
      "step": 146500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.84729317599825e-05,
      "loss": 1.323,
      "step": 147000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8365696833996045e-05,
      "loss": 1.3318,
      "step": 147500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.825846190800959e-05,
      "loss": 1.3227,
      "step": 148000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8151226982023136e-05,
      "loss": 1.3334,
      "step": 148500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8043992056036685e-05,
      "loss": 1.3255,
      "step": 149000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.793675713005023e-05,
      "loss": 1.3244,
      "step": 149500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7829522204063775e-05,
      "loss": 1.3201,
      "step": 150000
    },
    {
      "epoch": 1.93,
      "eval_loss": 1.2773252725601196,
      "eval_runtime": 3341.8511,
      "eval_samples_per_second": 39.602,
      "eval_steps_per_second": 4.95,
      "step": 150000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.772228727807732e-05,
      "loss": 1.3188,
      "step": 150500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7615052352090866e-05,
      "loss": 1.3203,
      "step": 151000
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7507817426104415e-05,
      "loss": 1.3123,
      "step": 151500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.740058250011796e-05,
      "loss": 1.319,
      "step": 152000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7293347574131506e-05,
      "loss": 1.3225,
      "step": 152500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.718611264814505e-05,
      "loss": 1.3231,
      "step": 153000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.7078877722158597e-05,
      "loss": 1.3213,
      "step": 153500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.6971642796172143e-05,
      "loss": 1.315,
      "step": 154000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6864407870185688e-05,
      "loss": 1.3203,
      "step": 154500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6757172944199234e-05,
      "loss": 1.3208,
      "step": 155000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.664993801821278e-05,
      "loss": 1.3155,
      "step": 155500
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6542703092226328e-05,
      "loss": 1.3038,
      "step": 156000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6435468166239873e-05,
      "loss": 1.2996,
      "step": 156500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.632823324025342e-05,
      "loss": 1.3023,
      "step": 157000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6220998314266964e-05,
      "loss": 1.3112,
      "step": 157500
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.611376338828051e-05,
      "loss": 1.3002,
      "step": 158000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.600652846229406e-05,
      "loss": 1.2967,
      "step": 158500
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.58992935363076e-05,
      "loss": 1.2996,
      "step": 159000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5792058610321146e-05,
      "loss": 1.3006,
      "step": 159500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.568482368433469e-05,
      "loss": 1.2903,
      "step": 160000
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.557758875834824e-05,
      "loss": 1.3077,
      "step": 160500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5470353832361786e-05,
      "loss": 1.2952,
      "step": 161000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.536311890637533e-05,
      "loss": 1.2984,
      "step": 161500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5255883980388877e-05,
      "loss": 1.3062,
      "step": 162000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5148649054402422e-05,
      "loss": 1.3005,
      "step": 162500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.5041414128415971e-05,
      "loss": 1.2961,
      "step": 163000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.4934179202429515e-05,
      "loss": 1.299,
      "step": 163500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.482694427644306e-05,
      "loss": 1.2931,
      "step": 164000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4719709350456606e-05,
      "loss": 1.2962,
      "step": 164500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4612474424470151e-05,
      "loss": 1.2993,
      "step": 165000
    },
    {
      "epoch": 2.12,
      "eval_loss": 1.261716604232788,
      "eval_runtime": 3339.8508,
      "eval_samples_per_second": 39.626,
      "eval_steps_per_second": 4.953,
      "step": 165000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.45052394984837e-05,
      "loss": 1.2979,
      "step": 165500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4398004572497246e-05,
      "loss": 1.2992,
      "step": 166000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4290769646510791e-05,
      "loss": 1.3039,
      "step": 166500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4183534720524335e-05,
      "loss": 1.2953,
      "step": 167000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4076299794537884e-05,
      "loss": 1.2837,
      "step": 167500
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.396906486855143e-05,
      "loss": 1.2971,
      "step": 168000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3861829942564975e-05,
      "loss": 1.2963,
      "step": 168500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.375459501657852e-05,
      "loss": 1.288,
      "step": 169000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3647360090592066e-05,
      "loss": 1.2953,
      "step": 169500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3540125164605613e-05,
      "loss": 1.2892,
      "step": 170000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3432890238619158e-05,
      "loss": 1.2931,
      "step": 170500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3325655312632704e-05,
      "loss": 1.2865,
      "step": 171000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.321842038664625e-05,
      "loss": 1.2925,
      "step": 171500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3111185460659795e-05,
      "loss": 1.286,
      "step": 172000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.3003950534673342e-05,
      "loss": 1.2949,
      "step": 172500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2896715608686887e-05,
      "loss": 1.2841,
      "step": 173000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2789480682700433e-05,
      "loss": 1.2869,
      "step": 173500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2682245756713978e-05,
      "loss": 1.2791,
      "step": 174000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2575010830727527e-05,
      "loss": 1.2822,
      "step": 174500
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2467775904741071e-05,
      "loss": 1.2857,
      "step": 175000
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2360540978754616e-05,
      "loss": 1.2913,
      "step": 175500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2253306052768162e-05,
      "loss": 1.273,
      "step": 176000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2146071126781709e-05,
      "loss": 1.2783,
      "step": 176500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.2038836200795255e-05,
      "loss": 1.2707,
      "step": 177000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1931601274808802e-05,
      "loss": 1.2747,
      "step": 177500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1824366348822347e-05,
      "loss": 1.2814,
      "step": 178000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1717131422835893e-05,
      "loss": 1.2862,
      "step": 178500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1609896496849438e-05,
      "loss": 1.2773,
      "step": 179000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1502661570862984e-05,
      "loss": 1.2783,
      "step": 179500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.139542664487653e-05,
      "loss": 1.2745,
      "step": 180000
    },
    {
      "epoch": 2.32,
      "eval_loss": 1.2489670515060425,
      "eval_runtime": 3344.7969,
      "eval_samples_per_second": 39.567,
      "eval_steps_per_second": 4.946,
      "step": 180000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1288191718890076e-05,
      "loss": 1.2783,
      "step": 180500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1180956792903623e-05,
      "loss": 1.2782,
      "step": 181000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1073721866917169e-05,
      "loss": 1.2727,
      "step": 181500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0966486940930713e-05,
      "loss": 1.2769,
      "step": 182000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.085925201494426e-05,
      "loss": 1.2743,
      "step": 182500
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0752017088957805e-05,
      "loss": 1.271,
      "step": 183000
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0644782162971352e-05,
      "loss": 1.2747,
      "step": 183500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0537547236984898e-05,
      "loss": 1.2733,
      "step": 184000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0430312310998443e-05,
      "loss": 1.2719,
      "step": 184500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0323077385011989e-05,
      "loss": 1.2724,
      "step": 185000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0215842459025534e-05,
      "loss": 1.2699,
      "step": 185500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0108607533039081e-05,
      "loss": 1.2707,
      "step": 186000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0001372607052627e-05,
      "loss": 1.2805,
      "step": 186500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.894137681066174e-06,
      "loss": 1.2695,
      "step": 187000
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.78690275507972e-06,
      "loss": 1.2618,
      "step": 187500
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.679667829093265e-06,
      "loss": 1.2636,
      "step": 188000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.57243290310681e-06,
      "loss": 1.2738,
      "step": 188500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.465197977120356e-06,
      "loss": 1.266,
      "step": 189000
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.357963051133903e-06,
      "loss": 1.2702,
      "step": 189500
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.250728125147449e-06,
      "loss": 1.2629,
      "step": 190000
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.143493199160994e-06,
      "loss": 1.2653,
      "step": 190500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.03625827317454e-06,
      "loss": 1.2544,
      "step": 191000
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.929023347188087e-06,
      "loss": 1.2661,
      "step": 191500
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.821788421201632e-06,
      "loss": 1.2697,
      "step": 192000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.714553495215178e-06,
      "loss": 1.262,
      "step": 192500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.607318569228725e-06,
      "loss": 1.2724,
      "step": 193000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.50008364324227e-06,
      "loss": 1.2617,
      "step": 193500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.392848717255816e-06,
      "loss": 1.263,
      "step": 194000
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.285613791269361e-06,
      "loss": 1.2629,
      "step": 194500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.178378865282908e-06,
      "loss": 1.2614,
      "step": 195000
    },
    {
      "epoch": 2.51,
      "eval_loss": 1.2283380031585693,
      "eval_runtime": 3383.2987,
      "eval_samples_per_second": 39.117,
      "eval_steps_per_second": 4.89,
      "step": 195000
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.071143939296454e-06,
      "loss": 1.2589,
      "step": 195500
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.96390901331e-06,
      "loss": 1.2497,
      "step": 196000
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.856674087323545e-06,
      "loss": 1.256,
      "step": 196500
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.74943916133709e-06,
      "loss": 1.2563,
      "step": 197000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.642204235350637e-06,
      "loss": 1.2531,
      "step": 197500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.534969309364183e-06,
      "loss": 1.2586,
      "step": 198000
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.427734383377728e-06,
      "loss": 1.2589,
      "step": 198500
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.320499457391275e-06,
      "loss": 1.2623,
      "step": 199000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.21326453140482e-06,
      "loss": 1.2573,
      "step": 199500
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.106029605418367e-06,
      "loss": 1.2487,
      "step": 200000
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.998794679431912e-06,
      "loss": 1.2522,
      "step": 200500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.891559753445459e-06,
      "loss": 1.2486,
      "step": 201000
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.7843248274590046e-06,
      "loss": 1.253,
      "step": 201500
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.67708990147255e-06,
      "loss": 1.2453,
      "step": 202000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.569854975486096e-06,
      "loss": 1.2463,
      "step": 202500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.462620049499642e-06,
      "loss": 1.2485,
      "step": 203000
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.355385123513188e-06,
      "loss": 1.2521,
      "step": 203500
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.248150197526734e-06,
      "loss": 1.2441,
      "step": 204000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.14091527154028e-06,
      "loss": 1.2513,
      "step": 204500
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.033680345553825e-06,
      "loss": 1.2594,
      "step": 205000
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.926445419567372e-06,
      "loss": 1.2539,
      "step": 205500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.819210493580918e-06,
      "loss": 1.2503,
      "step": 206000
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.7119755675944635e-06,
      "loss": 1.2463,
      "step": 206500
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.60474064160801e-06,
      "loss": 1.2378,
      "step": 207000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.497505715621555e-06,
      "loss": 1.2447,
      "step": 207500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.390270789635101e-06,
      "loss": 1.2434,
      "step": 208000
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.283035863648647e-06,
      "loss": 1.2469,
      "step": 208500
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.175800937662193e-06,
      "loss": 1.2505,
      "step": 209000
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.068566011675739e-06,
      "loss": 1.2485,
      "step": 209500
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.961331085689285e-06,
      "loss": 1.2424,
      "step": 210000
    },
    {
      "epoch": 2.7,
      "eval_loss": 1.2152148485183716,
      "eval_runtime": 3385.5322,
      "eval_samples_per_second": 39.091,
      "eval_steps_per_second": 4.886,
      "step": 210000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.854096159702831e-06,
      "loss": 1.2362,
      "step": 210500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.746861233716376e-06,
      "loss": 1.2378,
      "step": 211000
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.6396263077299224e-06,
      "loss": 1.2462,
      "step": 211500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.532391381743469e-06,
      "loss": 1.2415,
      "step": 212000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.425156455757014e-06,
      "loss": 1.248,
      "step": 212500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.3179215297705605e-06,
      "loss": 1.2307,
      "step": 213000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.210686603784107e-06,
      "loss": 1.2438,
      "step": 213500
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.1034516777976515e-06,
      "loss": 1.2451,
      "step": 214000
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.996216751811198e-06,
      "loss": 1.2388,
      "step": 214500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.888981825824744e-06,
      "loss": 1.2413,
      "step": 215000
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.78174689983829e-06,
      "loss": 1.241,
      "step": 215500
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.674511973851836e-06,
      "loss": 1.2288,
      "step": 216000
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.5672770478653818e-06,
      "loss": 1.2341,
      "step": 216500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.460042121878928e-06,
      "loss": 1.2368,
      "step": 217000
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.352807195892473e-06,
      "loss": 1.2298,
      "step": 217500
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.2455722699060195e-06,
      "loss": 1.2289,
      "step": 218000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.1383373439195654e-06,
      "loss": 1.2313,
      "step": 218500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.0311024179331113e-06,
      "loss": 1.2341,
      "step": 219000
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.923867491946657e-06,
      "loss": 1.2347,
      "step": 219500
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.816632565960203e-06,
      "loss": 1.2415,
      "step": 220000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.709397639973749e-06,
      "loss": 1.2351,
      "step": 220500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.602162713987295e-06,
      "loss": 1.2316,
      "step": 221000
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.494927788000841e-06,
      "loss": 1.229,
      "step": 221500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3876928620143866e-06,
      "loss": 1.2256,
      "step": 222000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.2804579360279325e-06,
      "loss": 1.2342,
      "step": 222500
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.173223010041479e-06,
      "loss": 1.2367,
      "step": 223000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.0659880840550243e-06,
      "loss": 1.2343,
      "step": 223500
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.95875315806857e-06,
      "loss": 1.2325,
      "step": 224000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.8515182320821163e-06,
      "loss": 1.2304,
      "step": 224500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7442833060956624e-06,
      "loss": 1.2296,
      "step": 225000
    },
    {
      "epoch": 2.9,
      "eval_loss": 1.2051972150802612,
      "eval_runtime": 3383.7385,
      "eval_samples_per_second": 39.112,
      "eval_steps_per_second": 4.889,
      "step": 225000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.637048380109208e-06,
      "loss": 1.2192,
      "step": 225500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.529813454122754e-06,
      "loss": 1.2305,
      "step": 226000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4225785281362999e-06,
      "loss": 1.2292,
      "step": 226500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.315343602149846e-06,
      "loss": 1.2292,
      "step": 227000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.2081086761633916e-06,
      "loss": 1.2286,
      "step": 227500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1008737501769378e-06,
      "loss": 1.2283,
      "step": 228000
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.936388241904836e-07,
      "loss": 1.2272,
      "step": 228500
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.864038982040294e-07,
      "loss": 1.2292,
      "step": 229000
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.791689722175754e-07,
      "loss": 1.2343,
      "step": 229500
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.719340462311213e-07,
      "loss": 1.2179,
      "step": 230000
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.646991202446672e-07,
      "loss": 1.2239,
      "step": 230500
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.5746419425821316e-07,
      "loss": 1.2263,
      "step": 231000
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.5022926827175906e-07,
      "loss": 1.2263,
      "step": 231500
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.4299434228530495e-07,
      "loss": 1.2282,
      "step": 232000
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.3575941629885087e-07,
      "loss": 1.2334,
      "step": 232500
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.8524490312396787e-08,
      "loss": 1.224,
      "step": 233000
    },
    {
      "epoch": 3.0,
      "step": 233133,
      "total_flos": 4.975552182291128e+18,
      "train_loss": 1.3875968830903724,
      "train_runtime": 695161.7821,
      "train_samples_per_second": 10.732,
      "train_steps_per_second": 0.335
    }
  ],
  "max_steps": 233133,
  "num_train_epochs": 3,
  "total_flos": 4.975552182291128e+18,
  "trial_name": null,
  "trial_params": null
}