{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 625000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.9996000000000003e-05,
      "loss": 4.9384,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9992e-05,
      "loss": 3.2955,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9988000000000002e-05,
      "loss": 2.879,
      "step": 1500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9984e-05,
      "loss": 2.6655,
      "step": 2000
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9980000000000002e-05,
      "loss": 2.5291,
      "step": 2500
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9976000000000003e-05,
      "loss": 2.4219,
      "step": 3000
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9972e-05,
      "loss": 2.3595,
      "step": 3500
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9968e-05,
      "loss": 2.2788,
      "step": 4000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9964e-05,
      "loss": 2.239,
      "step": 4500
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9960000000000002e-05,
      "loss": 2.1891,
      "step": 5000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9956000000000003e-05,
      "loss": 2.1492,
      "step": 5500
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9952e-05,
      "loss": 2.1163,
      "step": 6000
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9948e-05,
      "loss": 2.086,
      "step": 6500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9944e-05,
      "loss": 2.0627,
      "step": 7000
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9940000000000002e-05,
      "loss": 2.0396,
      "step": 7500
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9936000000000004e-05,
      "loss": 2.0061,
      "step": 8000
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9932e-05,
      "loss": 1.9973,
      "step": 8500
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9928e-05,
      "loss": 1.9651,
      "step": 9000
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9924e-05,
      "loss": 1.9509,
      "step": 9500
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9920000000000002e-05,
      "loss": 1.937,
      "step": 10000
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9916e-05,
      "loss": 1.9226,
      "step": 10500
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9912000000000002e-05,
      "loss": 1.9087,
      "step": 11000
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9908e-05,
      "loss": 1.8888,
      "step": 11500
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9904e-05,
      "loss": 1.879,
      "step": 12000
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9900000000000003e-05,
      "loss": 1.8612,
      "step": 12500
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9896e-05,
      "loss": 1.8466,
      "step": 13000
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9892000000000002e-05,
      "loss": 1.8376,
      "step": 13500
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9888e-05,
      "loss": 1.8353,
      "step": 14000
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9884e-05,
      "loss": 1.8239,
      "step": 14500
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9880000000000003e-05,
      "loss": 1.8111,
      "step": 15000
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9876e-05,
      "loss": 1.8014,
      "step": 15500
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9872000000000002e-05,
      "loss": 1.7967,
      "step": 16000
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9868e-05,
      "loss": 1.78,
      "step": 16500
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9864e-05,
      "loss": 1.7791,
      "step": 17000
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9860000000000003e-05,
      "loss": 1.7657,
      "step": 17500
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9856e-05,
      "loss": 1.7641,
      "step": 18000
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9852000000000002e-05,
      "loss": 1.769,
      "step": 18500
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9848e-05,
      "loss": 1.7516,
      "step": 19000
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9844000000000002e-05,
      "loss": 1.7521,
      "step": 19500
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9840000000000003e-05,
      "loss": 1.738,
      "step": 20000
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9836e-05,
      "loss": 1.7292,
      "step": 20500
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9832000000000003e-05,
      "loss": 1.7123,
      "step": 21000
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9828e-05,
      "loss": 1.7123,
      "step": 21500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9824000000000002e-05,
      "loss": 1.7095,
      "step": 22000
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.982e-05,
      "loss": 1.71,
      "step": 22500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9816e-05,
      "loss": 1.7009,
      "step": 23000
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9812000000000003e-05,
      "loss": 1.6947,
      "step": 23500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9808e-05,
      "loss": 1.6951,
      "step": 24000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9804000000000002e-05,
      "loss": 1.6828,
      "step": 24500
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.98e-05,
      "loss": 1.676,
      "step": 25000
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9796e-05,
      "loss": 1.6734,
      "step": 25500
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9792000000000003e-05,
      "loss": 1.6642,
      "step": 26000
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9788e-05,
      "loss": 1.6649,
      "step": 26500
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9784000000000002e-05,
      "loss": 1.667,
      "step": 27000
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.978e-05,
      "loss": 1.6556,
      "step": 27500
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9776000000000002e-05,
      "loss": 1.6532,
      "step": 28000
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9772000000000003e-05,
      "loss": 1.6545,
      "step": 28500
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9768e-05,
      "loss": 1.6481,
      "step": 29000
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9764000000000003e-05,
      "loss": 1.647,
      "step": 29500
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.976e-05,
      "loss": 1.6385,
      "step": 30000
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9756000000000002e-05,
      "loss": 1.6337,
      "step": 30500
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9752000000000003e-05,
      "loss": 1.6296,
      "step": 31000
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9748e-05,
      "loss": 1.6339,
      "step": 31500
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9744e-05,
      "loss": 1.63,
      "step": 32000
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.974e-05,
      "loss": 1.6224,
      "step": 32500
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9736000000000002e-05,
      "loss": 1.6207,
      "step": 33000
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9732000000000004e-05,
      "loss": 1.6101,
      "step": 33500
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9728e-05,
      "loss": 1.6095,
      "step": 34000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9724e-05,
      "loss": 1.6029,
      "step": 34500
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.972e-05,
      "loss": 1.6028,
      "step": 35000
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9716000000000002e-05,
      "loss": 1.5977,
      "step": 35500
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9712000000000004e-05,
      "loss": 1.5929,
      "step": 36000
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9708000000000002e-05,
      "loss": 1.5942,
      "step": 36500
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9704e-05,
      "loss": 1.5908,
      "step": 37000
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.97e-05,
      "loss": 1.5926,
      "step": 37500
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9696000000000003e-05,
      "loss": 1.5793,
      "step": 38000
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9692000000000004e-05,
      "loss": 1.5769,
      "step": 38500
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9688000000000002e-05,
      "loss": 1.5778,
      "step": 39000
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9684e-05,
      "loss": 1.58,
      "step": 39500
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.968e-05,
      "loss": 1.569,
      "step": 40000
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9676000000000003e-05,
      "loss": 1.5757,
      "step": 40500
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9672e-05,
      "loss": 1.5706,
      "step": 41000
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9668000000000002e-05,
      "loss": 1.5658,
      "step": 41500
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9664e-05,
      "loss": 1.5606,
      "step": 42000
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.966e-05,
      "loss": 1.5525,
      "step": 42500
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9656000000000003e-05,
      "loss": 1.5628,
      "step": 43000
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9652e-05,
      "loss": 1.5532,
      "step": 43500
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9648000000000002e-05,
      "loss": 1.5476,
      "step": 44000
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9644e-05,
      "loss": 1.5552,
      "step": 44500
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9640000000000002e-05,
      "loss": 1.5474,
      "step": 45000
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9636000000000003e-05,
      "loss": 1.544,
      "step": 45500
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9632e-05,
      "loss": 1.5491,
      "step": 46000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9628000000000002e-05,
      "loss": 1.5405,
      "step": 46500
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9624e-05,
      "loss": 1.5407,
      "step": 47000
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9620000000000002e-05,
      "loss": 1.539,
      "step": 47500
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9616000000000003e-05,
      "loss": 1.5269,
      "step": 48000
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9612e-05,
      "loss": 1.5229,
      "step": 48500
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9608000000000003e-05,
      "loss": 1.5364,
      "step": 49000
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9604e-05,
      "loss": 1.5312,
      "step": 49500
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 1.535,
      "step": 50000
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9596e-05,
      "loss": 1.5234,
      "step": 50500
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9592e-05,
      "loss": 1.5171,
      "step": 51000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9588000000000003e-05,
      "loss": 1.5166,
      "step": 51500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9584e-05,
      "loss": 1.5168,
      "step": 52000
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9580000000000002e-05,
      "loss": 1.5112,
      "step": 52500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9576e-05,
      "loss": 1.5218,
      "step": 53000
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9572e-05,
      "loss": 1.51,
      "step": 53500
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9568000000000003e-05,
      "loss": 1.5093,
      "step": 54000
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9564e-05,
      "loss": 1.5038,
      "step": 54500
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9560000000000002e-05,
      "loss": 1.5069,
      "step": 55000
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9556e-05,
      "loss": 1.5045,
      "step": 55500
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9552000000000002e-05,
      "loss": 1.4999,
      "step": 56000
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9548000000000003e-05,
      "loss": 1.5019,
      "step": 56500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9544e-05,
      "loss": 1.5001,
      "step": 57000
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9540000000000003e-05,
      "loss": 1.5014,
      "step": 57500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9536e-05,
      "loss": 1.4977,
      "step": 58000
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9532000000000002e-05,
      "loss": 1.4924,
      "step": 58500
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9528000000000003e-05,
      "loss": 1.4853,
      "step": 59000
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9524e-05,
      "loss": 1.4857,
      "step": 59500
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9520000000000003e-05,
      "loss": 1.4909,
      "step": 60000
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9516e-05,
      "loss": 1.4927,
      "step": 60500
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9512000000000002e-05,
      "loss": 1.4781,
      "step": 61000
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9508000000000004e-05,
      "loss": 1.4841,
      "step": 61500
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9504e-05,
      "loss": 1.48,
      "step": 62000
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.95e-05,
      "loss": 1.4897,
      "step": 62500
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9496e-05,
      "loss": 1.4823,
      "step": 63000
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9492000000000002e-05,
      "loss": 1.4801,
      "step": 63500
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9488000000000004e-05,
      "loss": 1.4817,
      "step": 64000
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9484000000000002e-05,
      "loss": 1.4841,
      "step": 64500
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.948e-05,
      "loss": 1.4823,
      "step": 65000
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9476e-05,
      "loss": 1.4758,
      "step": 65500
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9472000000000003e-05,
      "loss": 1.4731,
      "step": 66000
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9468000000000004e-05,
      "loss": 1.4695,
      "step": 66500
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.9464000000000002e-05,
      "loss": 1.4712,
      "step": 67000
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.946e-05,
      "loss": 1.4635,
      "step": 67500
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.9456e-05,
      "loss": 1.4711,
      "step": 68000
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.9452000000000003e-05,
      "loss": 1.472,
      "step": 68500
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.9448e-05,
      "loss": 1.4643,
      "step": 69000
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9444000000000002e-05,
      "loss": 1.4622,
      "step": 69500
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.944e-05,
      "loss": 1.4658,
      "step": 70000
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9436e-05,
      "loss": 1.4688,
      "step": 70500
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.9432000000000003e-05,
      "loss": 1.463,
      "step": 71000
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.9428e-05,
      "loss": 1.4603,
      "step": 71500
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.9424e-05,
      "loss": 1.4553,
      "step": 72000
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.942e-05,
      "loss": 1.4544,
      "step": 72500
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.9416000000000002e-05,
      "loss": 1.4559,
      "step": 73000
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.9412000000000003e-05,
      "loss": 1.4488,
      "step": 73500
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.9408e-05,
      "loss": 1.4493,
      "step": 74000
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.9404e-05,
      "loss": 1.4501,
      "step": 74500
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.94e-05,
      "loss": 1.4479,
      "step": 75000
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.9396000000000002e-05,
      "loss": 1.4461,
      "step": 75500
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.9392000000000003e-05,
      "loss": 1.4441,
      "step": 76000
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.9388e-05,
      "loss": 1.4549,
      "step": 76500
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.9384e-05,
      "loss": 1.4426,
      "step": 77000
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.938e-05,
      "loss": 1.4469,
      "step": 77500
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.9376000000000002e-05,
      "loss": 1.431,
      "step": 78000
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.9372000000000004e-05,
      "loss": 1.4444,
      "step": 78500
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.9368e-05,
      "loss": 1.4401,
      "step": 79000
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.9364e-05,
      "loss": 1.4343,
      "step": 79500
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.936e-05,
      "loss": 1.4358,
      "step": 80000
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.9356000000000002e-05,
      "loss": 1.4324,
      "step": 80500
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.9352e-05,
      "loss": 1.4299,
      "step": 81000
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.9348000000000002e-05,
      "loss": 1.4306,
      "step": 81500
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.9344e-05,
      "loss": 1.4315,
      "step": 82000
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.934e-05,
      "loss": 1.4348,
      "step": 82500
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.9336000000000003e-05,
      "loss": 1.4284,
      "step": 83000
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.9332e-05,
      "loss": 1.4277,
      "step": 83500
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.9328000000000002e-05,
      "loss": 1.4362,
      "step": 84000
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9324e-05,
      "loss": 1.4311,
      "step": 84500
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.932e-05,
      "loss": 1.4264,
      "step": 85000
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9316000000000003e-05,
      "loss": 1.4241,
      "step": 85500
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.9312e-05,
      "loss": 1.4169,
      "step": 86000
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.9308000000000002e-05,
      "loss": 1.4226,
      "step": 86500
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.9304e-05,
      "loss": 1.4244,
      "step": 87000
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.93e-05,
      "loss": 1.4194,
      "step": 87500
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.9296000000000003e-05,
      "loss": 1.4198,
      "step": 88000
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.9292e-05,
      "loss": 1.4175,
      "step": 88500
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.9288000000000002e-05,
      "loss": 1.4189,
      "step": 89000
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.9284e-05,
      "loss": 1.4248,
      "step": 89500
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.9280000000000002e-05,
      "loss": 1.4194,
      "step": 90000
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.9276e-05,
      "loss": 1.4118,
      "step": 90500
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.9272e-05,
      "loss": 1.4141,
      "step": 91000
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.9268000000000003e-05,
      "loss": 1.4151,
      "step": 91500
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.9264e-05,
      "loss": 1.4161,
      "step": 92000
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.9260000000000002e-05,
      "loss": 1.412,
      "step": 92500
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.9256e-05,
      "loss": 1.4086,
      "step": 93000
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.9252e-05,
      "loss": 1.4062,
      "step": 93500
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.9248000000000003e-05,
      "loss": 1.4112,
      "step": 94000
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.9244000000000004e-05,
      "loss": 1.399,
      "step": 94500
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.9240000000000002e-05,
      "loss": 1.4006,
      "step": 95000
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.9236e-05,
      "loss": 1.4046,
      "step": 95500
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.9232e-05,
      "loss": 1.4005,
      "step": 96000
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.9228000000000003e-05,
      "loss": 1.4049,
      "step": 96500
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.9224000000000004e-05,
      "loss": 1.4011,
      "step": 97000
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.9220000000000002e-05,
      "loss": 1.3991,
      "step": 97500
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.9216e-05,
      "loss": 1.3992,
      "step": 98000
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.9212000000000002e-05,
      "loss": 1.405,
      "step": 98500
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.9208000000000003e-05,
      "loss": 1.4007,
      "step": 99000
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9204e-05,
      "loss": 1.3973,
      "step": 99500
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 1.3959,
      "step": 100000
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9196e-05,
      "loss": 1.4014,
      "step": 100500
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9192000000000002e-05,
      "loss": 1.3928,
      "step": 101000
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9188000000000003e-05,
      "loss": 1.3946,
      "step": 101500
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.9184e-05,
      "loss": 1.3904,
      "step": 102000
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.918e-05,
      "loss": 1.3904,
      "step": 102500
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.9176e-05,
      "loss": 1.3949,
      "step": 103000
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.9172000000000002e-05,
      "loss": 1.3948,
      "step": 103500
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.9168000000000004e-05,
      "loss": 1.3859,
      "step": 104000
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.9164e-05,
      "loss": 1.3803,
      "step": 104500
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.916e-05,
      "loss": 1.3885,
      "step": 105000
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.9156e-05,
      "loss": 1.3773,
      "step": 105500
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.9152000000000002e-05,
      "loss": 1.3812,
      "step": 106000
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.9148000000000004e-05,
      "loss": 1.3844,
      "step": 106500
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.9144000000000002e-05,
      "loss": 1.386,
      "step": 107000
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.914e-05,
      "loss": 1.381,
      "step": 107500
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.9136e-05,
      "loss": 1.388,
      "step": 108000
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.9132000000000002e-05,
      "loss": 1.3822,
      "step": 108500
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.9128e-05,
      "loss": 1.3875,
      "step": 109000
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.9124000000000002e-05,
      "loss": 1.3869,
      "step": 109500
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.912e-05,
      "loss": 1.3877,
      "step": 110000
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.9116e-05,
      "loss": 1.3814,
      "step": 110500
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.9112000000000003e-05,
      "loss": 1.3852,
      "step": 111000
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.9108e-05,
      "loss": 1.3797,
      "step": 111500
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9104000000000002e-05,
      "loss": 1.3692,
      "step": 112000
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.91e-05,
      "loss": 1.3773,
      "step": 112500
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.9096e-05,
      "loss": 1.3822,
      "step": 113000
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.9092000000000003e-05,
      "loss": 1.3831,
      "step": 113500
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.9088e-05,
      "loss": 1.3746,
      "step": 114000
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.9084000000000002e-05,
      "loss": 1.3697,
      "step": 114500
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.908e-05,
      "loss": 1.3753,
      "step": 115000
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.9076e-05,
      "loss": 1.3764,
      "step": 115500
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.9072000000000003e-05,
      "loss": 1.3688,
      "step": 116000
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.9068e-05,
      "loss": 1.3679,
      "step": 116500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.9064000000000002e-05,
      "loss": 1.3676,
      "step": 117000
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.906e-05,
      "loss": 1.3733,
      "step": 117500
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.9056000000000002e-05,
      "loss": 1.3679,
      "step": 118000
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.9052000000000003e-05,
      "loss": 1.3595,
      "step": 118500
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.9048e-05,
      "loss": 1.3723,
      "step": 119000
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.9044000000000003e-05,
      "loss": 1.3695,
      "step": 119500
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.904e-05,
      "loss": 1.368,
      "step": 120000
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.9036000000000002e-05,
      "loss": 1.3656,
      "step": 120500
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.9032e-05,
      "loss": 1.364,
      "step": 121000
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.9028e-05,
      "loss": 1.357,
      "step": 121500
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.9024000000000003e-05,
      "loss": 1.3588,
      "step": 122000
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.902e-05,
      "loss": 1.361,
      "step": 122500
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.9016000000000002e-05,
      "loss": 1.3631,
      "step": 123000
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.9012e-05,
      "loss": 1.3566,
      "step": 123500
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.9008e-05,
      "loss": 1.3568,
      "step": 124000
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.9004000000000003e-05,
      "loss": 1.3615,
      "step": 124500
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.9e-05,
      "loss": 1.3577,
      "step": 125000
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.2994823455810547,
      "eval_runtime": 80.8005,
      "eval_samples_per_second": 166.002,
      "eval_steps_per_second": 2.599,
      "step": 125000
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8996000000000002e-05,
      "loss": 1.3472,
      "step": 125500
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8992e-05,
      "loss": 1.3557,
      "step": 126000
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8988000000000002e-05,
      "loss": 1.3538,
      "step": 126500
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.8984000000000003e-05,
      "loss": 1.3584,
      "step": 127000
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.898e-05,
      "loss": 1.3484,
      "step": 127500
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.8976000000000003e-05,
      "loss": 1.351,
      "step": 128000
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.8972e-05,
      "loss": 1.3485,
      "step": 128500
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.8968000000000002e-05,
      "loss": 1.3497,
      "step": 129000
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.8964000000000003e-05,
      "loss": 1.3455,
      "step": 129500
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.896e-05,
      "loss": 1.3437,
      "step": 130000
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.8956e-05,
      "loss": 1.3544,
      "step": 130500
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.8952e-05,
      "loss": 1.3606,
      "step": 131000
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.8948000000000002e-05,
      "loss": 1.3514,
      "step": 131500
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.8944000000000004e-05,
      "loss": 1.3377,
      "step": 132000
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.894e-05,
      "loss": 1.3421,
      "step": 132500
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.8936e-05,
      "loss": 1.3454,
      "step": 133000
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.8932e-05,
      "loss": 1.3386,
      "step": 133500
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.8928000000000002e-05,
      "loss": 1.3444,
      "step": 134000
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.8924000000000004e-05,
      "loss": 1.3426,
      "step": 134500
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.8920000000000002e-05,
      "loss": 1.344,
      "step": 135000
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.8916e-05,
      "loss": 1.3443,
      "step": 135500
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.8912e-05,
      "loss": 1.3431,
      "step": 136000
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.8908000000000003e-05,
      "loss": 1.341,
      "step": 136500
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.8904000000000004e-05,
      "loss": 1.3528,
      "step": 137000
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 1.3424,
      "step": 137500
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.8896e-05,
      "loss": 1.3506,
      "step": 138000
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.8892e-05,
      "loss": 1.3451,
      "step": 138500
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.8888000000000003e-05,
      "loss": 1.3451,
      "step": 139000
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.8884e-05,
      "loss": 1.3366,
      "step": 139500
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.8880000000000002e-05,
      "loss": 1.34,
      "step": 140000
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.8876e-05,
      "loss": 1.3465,
      "step": 140500
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.8872e-05,
      "loss": 1.3462,
      "step": 141000
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.8868000000000003e-05,
      "loss": 1.3359,
      "step": 141500
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.8864e-05,
      "loss": 1.3431,
      "step": 142000
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.886e-05,
      "loss": 1.337,
      "step": 142500
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.8856e-05,
      "loss": 1.3355,
      "step": 143000
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.8852000000000002e-05,
      "loss": 1.3374,
      "step": 143500
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.8848000000000003e-05,
      "loss": 1.3362,
      "step": 144000
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.8844e-05,
      "loss": 1.334,
      "step": 144500
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.884e-05,
      "loss": 1.3311,
      "step": 145000
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.8836e-05,
      "loss": 1.3281,
      "step": 145500
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.8832000000000002e-05,
      "loss": 1.338,
      "step": 146000
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.8828000000000003e-05,
      "loss": 1.3314,
      "step": 146500
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.8824e-05,
      "loss": 1.3319,
      "step": 147000
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.882e-05,
      "loss": 1.3316,
      "step": 147500
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.8816e-05,
      "loss": 1.329,
      "step": 148000
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.8812000000000002e-05,
      "loss": 1.3322,
      "step": 148500
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.8808e-05,
      "loss": 1.3236,
      "step": 149000
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.8804e-05,
      "loss": 1.3237,
      "step": 149500
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.88e-05,
      "loss": 1.3316,
      "step": 150000
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.8796e-05,
      "loss": 1.3354,
      "step": 150500
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.8792000000000002e-05,
      "loss": 1.3304,
      "step": 151000
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.8788e-05,
      "loss": 1.3295,
      "step": 151500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.8784000000000002e-05,
      "loss": 1.3275,
      "step": 152000
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.878e-05,
      "loss": 1.3288,
      "step": 152500
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.8776e-05,
      "loss": 1.3341,
      "step": 153000
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.8772000000000003e-05,
      "loss": 1.3203,
      "step": 153500
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.8768e-05,
      "loss": 1.3242,
      "step": 154000
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.8764000000000002e-05,
      "loss": 1.3234,
      "step": 154500
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.876e-05,
      "loss": 1.3264,
      "step": 155000
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.8756e-05,
      "loss": 1.3247,
      "step": 155500
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.8752000000000003e-05,
      "loss": 1.3219,
      "step": 156000
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.8748e-05,
      "loss": 1.3211,
      "step": 156500
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.8744000000000002e-05,
      "loss": 1.3198,
      "step": 157000
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.8740000000000004e-05,
      "loss": 1.3225,
      "step": 157500
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.8736e-05,
      "loss": 1.3175,
      "step": 158000
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.8732000000000003e-05,
      "loss": 1.3248,
      "step": 158500
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.8728e-05,
      "loss": 1.316,
      "step": 159000
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.8724000000000002e-05,
      "loss": 1.3208,
      "step": 159500
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.8720000000000004e-05,
      "loss": 1.3132,
      "step": 160000
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.8716000000000002e-05,
      "loss": 1.3097,
      "step": 160500
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.8712e-05,
      "loss": 1.3181,
      "step": 161000
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.8708e-05,
      "loss": 1.3199,
      "step": 161500
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.8704000000000003e-05,
      "loss": 1.3191,
      "step": 162000
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.8700000000000004e-05,
      "loss": 1.3189,
      "step": 162500
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.8696000000000002e-05,
      "loss": 1.319,
      "step": 163000
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.8692e-05,
      "loss": 1.3203,
      "step": 163500
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.8688e-05,
      "loss": 1.3155,
      "step": 164000
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.8684000000000003e-05,
      "loss": 1.3148,
      "step": 164500
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.8680000000000004e-05,
      "loss": 1.319,
      "step": 165000
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.8676000000000002e-05,
      "loss": 1.3105,
      "step": 165500
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.8672e-05,
      "loss": 1.3117,
      "step": 166000
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.8668e-05,
      "loss": 1.3119,
      "step": 166500
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.8664000000000003e-05,
      "loss": 1.3159,
      "step": 167000
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.866e-05,
      "loss": 1.307,
      "step": 167500
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.8656000000000002e-05,
      "loss": 1.3063,
      "step": 168000
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.8652e-05,
      "loss": 1.3102,
      "step": 168500
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.8648000000000002e-05,
      "loss": 1.3134,
      "step": 169000
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.8644000000000003e-05,
      "loss": 1.314,
      "step": 169500
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.864e-05,
      "loss": 1.3111,
      "step": 170000
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.8636e-05,
      "loss": 1.3047,
      "step": 170500
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.8632e-05,
      "loss": 1.3095,
      "step": 171000
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.8628000000000002e-05,
      "loss": 1.3066,
      "step": 171500
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.8624000000000003e-05,
      "loss": 1.3071,
      "step": 172000
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.862e-05,
      "loss": 1.2985,
      "step": 172500
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.8616e-05,
      "loss": 1.3103,
      "step": 173000
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.8612e-05,
      "loss": 1.3106,
      "step": 173500
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.8608000000000002e-05,
      "loss": 1.3055,
      "step": 174000
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.8604000000000003e-05,
      "loss": 1.3074,
      "step": 174500
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.86e-05,
      "loss": 1.3028,
      "step": 175000
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.8596e-05,
      "loss": 1.3027,
      "step": 175500
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.8592e-05,
      "loss": 1.2988,
      "step": 176000
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.8588000000000002e-05,
      "loss": 1.303,
      "step": 176500
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.8584000000000004e-05,
      "loss": 1.3009,
      "step": 177000
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.858e-05,
      "loss": 1.3023,
      "step": 177500
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.8576e-05,
      "loss": 1.3051,
      "step": 178000
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8572e-05,
      "loss": 1.3032,
      "step": 178500
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.8568000000000002e-05,
      "loss": 1.3051,
      "step": 179000
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8564e-05,
      "loss": 1.303,
      "step": 179500
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8560000000000002e-05,
      "loss": 1.2975,
      "step": 180000
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8556e-05,
      "loss": 1.2985,
      "step": 180500
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.8552e-05,
      "loss": 1.3044,
      "step": 181000
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.8548000000000003e-05,
      "loss": 1.2983,
      "step": 181500
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.8544e-05,
      "loss": 1.2999,
      "step": 182000
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.8540000000000002e-05,
      "loss": 1.3027,
      "step": 182500
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.8536e-05,
      "loss": 1.3021,
      "step": 183000
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.8532e-05,
      "loss": 1.2962,
      "step": 183500
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.8528000000000003e-05,
      "loss": 1.2929,
      "step": 184000
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.8524e-05,
      "loss": 1.3006,
      "step": 184500
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.8520000000000002e-05,
      "loss": 1.2888,
      "step": 185000
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.8516e-05,
      "loss": 1.2982,
      "step": 185500
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.8512e-05,
      "loss": 1.2911,
      "step": 186000
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.8508000000000003e-05,
      "loss": 1.2942,
      "step": 186500
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.8504e-05,
      "loss": 1.3,
      "step": 187000
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.8500000000000002e-05,
      "loss": 1.2996,
      "step": 187500
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.8496e-05,
      "loss": 1.2964,
      "step": 188000
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.8492000000000002e-05,
      "loss": 1.2862,
      "step": 188500
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.8488e-05,
      "loss": 1.2911,
      "step": 189000
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.8484e-05,
      "loss": 1.295,
      "step": 189500
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.8480000000000003e-05,
      "loss": 1.2934,
      "step": 190000
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.8476e-05,
      "loss": 1.2915,
      "step": 190500
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.8472000000000002e-05,
      "loss": 1.289,
      "step": 191000
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.8468e-05,
      "loss": 1.2858,
      "step": 191500
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.8464e-05,
      "loss": 1.298,
      "step": 192000
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.8460000000000003e-05,
      "loss": 1.2918,
      "step": 192500
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.8456e-05,
      "loss": 1.2922,
      "step": 193000
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.8452000000000002e-05,
      "loss": 1.2901,
      "step": 193500
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.8448e-05,
      "loss": 1.2854,
      "step": 194000
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.8444e-05,
      "loss": 1.2853,
      "step": 194500
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.8440000000000003e-05,
      "loss": 1.2896,
      "step": 195000
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.8436e-05,
      "loss": 1.2873,
      "step": 195500
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.8432000000000002e-05,
      "loss": 1.2856,
      "step": 196000
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.8428e-05,
      "loss": 1.2851,
      "step": 196500
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.8424000000000002e-05,
      "loss": 1.2843,
      "step": 197000
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.8420000000000003e-05,
      "loss": 1.2896,
      "step": 197500
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.8416e-05,
      "loss": 1.2893,
      "step": 198000
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.8412000000000003e-05,
      "loss": 1.2837,
      "step": 198500
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.8408e-05,
      "loss": 1.2922,
      "step": 199000
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.8404000000000002e-05,
      "loss": 1.289,
      "step": 199500
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 1.2923,
      "step": 200000
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.8396e-05,
      "loss": 1.2859,
      "step": 200500
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.8392e-05,
      "loss": 1.2819,
      "step": 201000
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.8388e-05,
      "loss": 1.2864,
      "step": 201500
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.8384000000000002e-05,
      "loss": 1.2828,
      "step": 202000
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.8380000000000004e-05,
      "loss": 1.2883,
      "step": 202500
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.8376e-05,
      "loss": 1.2833,
      "step": 203000
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.8372e-05,
      "loss": 1.284,
      "step": 203500
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.8368e-05,
      "loss": 1.2811,
      "step": 204000
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.8364000000000002e-05,
      "loss": 1.2818,
      "step": 204500
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.8360000000000004e-05,
      "loss": 1.279,
      "step": 205000
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.8356000000000002e-05,
      "loss": 1.2823,
      "step": 205500
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.8352e-05,
      "loss": 1.2842,
      "step": 206000
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.8348e-05,
      "loss": 1.2854,
      "step": 206500
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.8344000000000003e-05,
      "loss": 1.2787,
      "step": 207000
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.834e-05,
      "loss": 1.285,
      "step": 207500
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.8336000000000002e-05,
      "loss": 1.2851,
      "step": 208000
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.8332e-05,
      "loss": 1.277,
      "step": 208500
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.8328e-05,
      "loss": 1.2828,
      "step": 209000
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.8324000000000003e-05,
      "loss": 1.2741,
      "step": 209500
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.832e-05,
      "loss": 1.2744,
      "step": 210000
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.8316e-05,
      "loss": 1.2743,
      "step": 210500
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.8312e-05,
      "loss": 1.2722,
      "step": 211000
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.8308e-05,
      "loss": 1.2759,
      "step": 211500
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.8304000000000003e-05,
      "loss": 1.2757,
      "step": 212000
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.83e-05,
      "loss": 1.2806,
      "step": 212500
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.8296e-05,
      "loss": 1.276,
      "step": 213000
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.8292e-05,
      "loss": 1.2771,
      "step": 213500
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.8288000000000002e-05,
      "loss": 1.2759,
      "step": 214000
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.8284000000000003e-05,
      "loss": 1.2676,
      "step": 214500
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.828e-05,
      "loss": 1.2758,
      "step": 215000
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.8276e-05,
      "loss": 1.2667,
      "step": 215500
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.8272e-05,
      "loss": 1.2738,
      "step": 216000
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.8268000000000002e-05,
      "loss": 1.2748,
      "step": 216500
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.8264000000000003e-05,
      "loss": 1.2762,
      "step": 217000
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.826e-05,
      "loss": 1.2707,
      "step": 217500
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.8256e-05,
      "loss": 1.2667,
      "step": 218000
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.8252e-05,
      "loss": 1.2708,
      "step": 218500
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.8248000000000002e-05,
      "loss": 1.2674,
      "step": 219000
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.8244e-05,
      "loss": 1.2753,
      "step": 219500
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.824e-05,
      "loss": 1.2724,
      "step": 220000
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.8236000000000003e-05,
      "loss": 1.274,
      "step": 220500
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.8232e-05,
      "loss": 1.2764,
      "step": 221000
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.8228000000000002e-05,
      "loss": 1.2681,
      "step": 221500
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.8224e-05,
      "loss": 1.2742,
      "step": 222000
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.8220000000000002e-05,
      "loss": 1.2712,
      "step": 222500
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.8216000000000003e-05,
      "loss": 1.2666,
      "step": 223000
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.8212e-05,
      "loss": 1.2695,
      "step": 223500
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.8208000000000003e-05,
      "loss": 1.2698,
      "step": 224000
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.8204e-05,
      "loss": 1.2734,
      "step": 224500
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 1.2682,
      "step": 225000
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.8196000000000003e-05,
      "loss": 1.2715,
      "step": 225500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.8192e-05,
      "loss": 1.2699,
      "step": 226000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.8188000000000003e-05,
      "loss": 1.2718,
      "step": 226500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.8184e-05,
      "loss": 1.2621,
      "step": 227000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.8180000000000002e-05,
      "loss": 1.2642,
      "step": 227500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.8176000000000004e-05,
      "loss": 1.2685,
      "step": 228000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.8172e-05,
      "loss": 1.2645,
      "step": 228500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.8168e-05,
      "loss": 1.2697,
      "step": 229000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.8164e-05,
      "loss": 1.2671,
      "step": 229500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.8160000000000002e-05,
      "loss": 1.2652,
      "step": 230000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.8156000000000004e-05,
      "loss": 1.2592,
      "step": 230500
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.8152000000000002e-05,
      "loss": 1.2699,
      "step": 231000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.8148e-05,
      "loss": 1.2673,
      "step": 231500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8144e-05,
      "loss": 1.27,
      "step": 232000
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8140000000000003e-05,
      "loss": 1.2673,
      "step": 232500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8136000000000004e-05,
      "loss": 1.266,
      "step": 233000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8132000000000002e-05,
      "loss": 1.2697,
      "step": 233500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8128e-05,
      "loss": 1.2592,
      "step": 234000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8124e-05,
      "loss": 1.2657,
      "step": 234500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8120000000000003e-05,
      "loss": 1.2608,
      "step": 235000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8116000000000004e-05,
      "loss": 1.2617,
      "step": 235500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8112000000000002e-05,
      "loss": 1.2647,
      "step": 236000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8108e-05,
      "loss": 1.2623,
      "step": 236500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8104e-05,
      "loss": 1.2593,
      "step": 237000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8100000000000003e-05,
      "loss": 1.258,
      "step": 237500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8096e-05,
      "loss": 1.2627,
      "step": 238000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8092000000000002e-05,
      "loss": 1.2666,
      "step": 238500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8088e-05,
      "loss": 1.2642,
      "step": 239000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8084e-05,
      "loss": 1.2583,
      "step": 239500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8080000000000003e-05,
      "loss": 1.2588,
      "step": 240000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8076e-05,
      "loss": 1.2676,
      "step": 240500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.8072e-05,
      "loss": 1.2564,
      "step": 241000
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.8068e-05,
      "loss": 1.2554,
      "step": 241500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.8064000000000002e-05,
      "loss": 1.2543,
      "step": 242000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.8060000000000003e-05,
      "loss": 1.265,
      "step": 242500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.8056e-05,
      "loss": 1.2575,
      "step": 243000
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.8052e-05,
      "loss": 1.2511,
      "step": 243500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.8048e-05,
      "loss": 1.2581,
      "step": 244000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.8044000000000002e-05,
      "loss": 1.257,
      "step": 244500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.8040000000000003e-05,
      "loss": 1.2639,
      "step": 245000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.8036e-05,
      "loss": 1.2565,
      "step": 245500
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.8032e-05,
      "loss": 1.2567,
      "step": 246000
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.8028e-05,
      "loss": 1.2469,
      "step": 246500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.8024000000000002e-05,
      "loss": 1.2606,
      "step": 247000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.802e-05,
      "loss": 1.2514,
      "step": 247500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.8016e-05,
      "loss": 1.2505,
      "step": 248000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.8012e-05,
      "loss": 1.2537,
      "step": 248500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.8008e-05,
      "loss": 1.252,
      "step": 249000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.8004000000000002e-05,
      "loss": 1.2528,
      "step": 249500
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.8e-05,
      "loss": 1.256,
      "step": 250000
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.2035317420959473,
      "eval_runtime": 81.1073,
      "eval_samples_per_second": 165.373,
      "eval_steps_per_second": 2.589,
      "step": 250000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.7996000000000002e-05,
      "loss": 1.246,
      "step": 250500
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.7992e-05,
      "loss": 1.2571,
      "step": 251000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.7988e-05,
      "loss": 1.2529,
      "step": 251500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.7984000000000003e-05,
      "loss": 1.2474,
      "step": 252000
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.798e-05,
      "loss": 1.2493,
      "step": 252500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.7976000000000002e-05,
      "loss": 1.2492,
      "step": 253000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.7972e-05,
      "loss": 1.2522,
      "step": 253500
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.7968e-05,
      "loss": 1.2494,
      "step": 254000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.7964000000000003e-05,
      "loss": 1.2426,
      "step": 254500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.796e-05,
      "loss": 1.2522,
      "step": 255000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.7956000000000002e-05,
      "loss": 1.253,
      "step": 255500
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.7952e-05,
      "loss": 1.2517,
      "step": 256000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.7948e-05,
      "loss": 1.2489,
      "step": 256500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.7944000000000003e-05,
      "loss": 1.2497,
      "step": 257000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.794e-05,
      "loss": 1.2469,
      "step": 257500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.7936000000000002e-05,
      "loss": 1.2405,
      "step": 258000
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.7932e-05,
      "loss": 1.2477,
      "step": 258500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.7928000000000002e-05,
      "loss": 1.2427,
      "step": 259000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.7924e-05,
      "loss": 1.2446,
      "step": 259500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.792e-05,
      "loss": 1.2422,
      "step": 260000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.7916000000000003e-05,
      "loss": 1.2452,
      "step": 260500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.7912e-05,
      "loss": 1.2532,
      "step": 261000
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.7908000000000002e-05,
      "loss": 1.2431,
      "step": 261500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.7904e-05,
      "loss": 1.2471,
      "step": 262000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.79e-05,
      "loss": 1.2446,
      "step": 262500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.7896000000000003e-05,
      "loss": 1.2495,
      "step": 263000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.7892e-05,
      "loss": 1.2468,
      "step": 263500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.7888000000000002e-05,
      "loss": 1.2495,
      "step": 264000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.7884e-05,
      "loss": 1.2519,
      "step": 264500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.788e-05,
      "loss": 1.2434,
      "step": 265000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.7876000000000003e-05,
      "loss": 1.2411,
      "step": 265500
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.7872e-05,
      "loss": 1.235,
      "step": 266000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.7868000000000002e-05,
      "loss": 1.2424,
      "step": 266500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.7864e-05,
      "loss": 1.2456,
      "step": 267000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.7860000000000002e-05,
      "loss": 1.2521,
      "step": 267500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.7856000000000003e-05,
      "loss": 1.2368,
      "step": 268000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.7852e-05,
      "loss": 1.2382,
      "step": 268500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.7848e-05,
      "loss": 1.2447,
      "step": 269000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.7844e-05,
      "loss": 1.2445,
      "step": 269500
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.7840000000000002e-05,
      "loss": 1.2429,
      "step": 270000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.7836000000000003e-05,
      "loss": 1.2416,
      "step": 270500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7832e-05,
      "loss": 1.2376,
      "step": 271000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7828e-05,
      "loss": 1.2317,
      "step": 271500
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7824e-05,
      "loss": 1.2303,
      "step": 272000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7820000000000002e-05,
      "loss": 1.24,
      "step": 272500
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7816000000000004e-05,
      "loss": 1.2421,
      "step": 273000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7812e-05,
      "loss": 1.2416,
      "step": 273500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7808e-05,
      "loss": 1.2398,
      "step": 274000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.7804e-05,
      "loss": 1.2393,
      "step": 274500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 1.2455,
      "step": 275000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.7796000000000004e-05,
      "loss": 1.2321,
      "step": 275500
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.7792000000000002e-05,
      "loss": 1.2326,
      "step": 276000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.7788e-05,
      "loss": 1.2381,
      "step": 276500
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.7784e-05,
      "loss": 1.2398,
      "step": 277000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.7780000000000003e-05,
      "loss": 1.2367,
      "step": 277500
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.7776e-05,
      "loss": 1.2357,
      "step": 278000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.7772000000000002e-05,
      "loss": 1.2369,
      "step": 278500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.7768e-05,
      "loss": 1.2395,
      "step": 279000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.7764e-05,
      "loss": 1.2347,
      "step": 279500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.7760000000000003e-05,
      "loss": 1.2322,
      "step": 280000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.7756e-05,
      "loss": 1.2393,
      "step": 280500
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.7752e-05,
      "loss": 1.2349,
      "step": 281000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.7748e-05,
      "loss": 1.2363,
      "step": 281500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.7744e-05,
      "loss": 1.2316,
      "step": 282000
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.7740000000000003e-05,
      "loss": 1.237,
      "step": 282500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.7736e-05,
      "loss": 1.2345,
      "step": 283000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.7732000000000002e-05,
      "loss": 1.2369,
      "step": 283500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.7728e-05,
      "loss": 1.2333,
      "step": 284000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.7724000000000002e-05,
      "loss": 1.2326,
      "step": 284500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.7720000000000003e-05,
      "loss": 1.2256,
      "step": 285000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.7716e-05,
      "loss": 1.2268,
      "step": 285500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.7712000000000003e-05,
      "loss": 1.2354,
      "step": 286000
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.7708e-05,
      "loss": 1.2298,
      "step": 286500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.7704000000000002e-05,
      "loss": 1.2292,
      "step": 287000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.77e-05,
      "loss": 1.2328,
      "step": 287500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.7696e-05,
      "loss": 1.2332,
      "step": 288000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.7692000000000003e-05,
      "loss": 1.2337,
      "step": 288500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.7688e-05,
      "loss": 1.2337,
      "step": 289000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.7684000000000002e-05,
      "loss": 1.2338,
      "step": 289500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.768e-05,
      "loss": 1.2373,
      "step": 290000
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.7676e-05,
      "loss": 1.2335,
      "step": 290500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.7672000000000003e-05,
      "loss": 1.2338,
      "step": 291000
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.7668e-05,
      "loss": 1.2246,
      "step": 291500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.7664000000000002e-05,
      "loss": 1.2349,
      "step": 292000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.766e-05,
      "loss": 1.2303,
      "step": 292500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.7656000000000002e-05,
      "loss": 1.227,
      "step": 293000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.7652000000000003e-05,
      "loss": 1.2304,
      "step": 293500
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.7648e-05,
      "loss": 1.2341,
      "step": 294000
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.7644000000000003e-05,
      "loss": 1.2397,
      "step": 294500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.764e-05,
      "loss": 1.2389,
      "step": 295000
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.7636000000000002e-05,
      "loss": 1.2336,
      "step": 295500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.7632000000000003e-05,
      "loss": 1.2246,
      "step": 296000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.7628e-05,
      "loss": 1.2282,
      "step": 296500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.7624000000000003e-05,
      "loss": 1.222,
      "step": 297000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.762e-05,
      "loss": 1.2253,
      "step": 297500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.7616000000000002e-05,
      "loss": 1.234,
      "step": 298000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.7612000000000003e-05,
      "loss": 1.2259,
      "step": 298500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.7608e-05,
      "loss": 1.2297,
      "step": 299000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.7604e-05,
      "loss": 1.226,
      "step": 299500
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.76e-05,
      "loss": 1.2278,
      "step": 300000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.7596000000000002e-05,
      "loss": 1.2291,
      "step": 300500
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.7592000000000004e-05,
      "loss": 1.2291,
      "step": 301000
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.7588e-05,
      "loss": 1.2228,
      "step": 301500
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.7584e-05,
      "loss": 1.2252,
      "step": 302000
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.758e-05,
      "loss": 1.2322,
      "step": 302500
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.7576000000000002e-05,
      "loss": 1.2248,
      "step": 303000
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.7572000000000004e-05,
      "loss": 1.2278,
      "step": 303500
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.7568000000000002e-05,
      "loss": 1.2142,
      "step": 304000
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7564e-05,
      "loss": 1.2294,
      "step": 304500
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.756e-05,
      "loss": 1.2247,
      "step": 305000
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7556000000000003e-05,
      "loss": 1.2266,
      "step": 305500
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7552e-05,
      "loss": 1.2299,
      "step": 306000
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7548000000000002e-05,
      "loss": 1.2284,
      "step": 306500
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.7544e-05,
      "loss": 1.2205,
      "step": 307000
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.754e-05,
      "loss": 1.2305,
      "step": 307500
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.7536000000000003e-05,
      "loss": 1.2268,
      "step": 308000
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.7532e-05,
      "loss": 1.226,
      "step": 308500
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.7528e-05,
      "loss": 1.2249,
      "step": 309000
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.7524e-05,
      "loss": 1.2264,
      "step": 309500
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.752e-05,
      "loss": 1.2265,
      "step": 310000
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.7516000000000003e-05,
      "loss": 1.2197,
      "step": 310500
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.7512e-05,
      "loss": 1.2224,
      "step": 311000
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.7508e-05,
      "loss": 1.2167,
      "step": 311500
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7504e-05,
      "loss": 1.2284,
      "step": 312000
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 1.2235,
      "step": 312500
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.7496000000000003e-05,
      "loss": 1.2218,
      "step": 313000
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7492e-05,
      "loss": 1.215,
      "step": 313500
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.7488e-05,
      "loss": 1.2217,
      "step": 314000
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7484e-05,
      "loss": 1.2207,
      "step": 314500
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7480000000000002e-05,
      "loss": 1.219,
      "step": 315000
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.7476000000000003e-05,
      "loss": 1.225,
      "step": 315500
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7472e-05,
      "loss": 1.2214,
      "step": 316000
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.7468e-05,
      "loss": 1.218,
      "step": 316500
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.7464e-05,
      "loss": 1.2182,
      "step": 317000
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.7460000000000002e-05,
      "loss": 1.2143,
      "step": 317500
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.7456e-05,
      "loss": 1.219,
      "step": 318000
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.7452e-05,
      "loss": 1.2204,
      "step": 318500
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.7448e-05,
      "loss": 1.2143,
      "step": 319000
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.7444e-05,
      "loss": 1.2188,
      "step": 319500
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.7440000000000002e-05,
      "loss": 1.226,
      "step": 320000
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.7436e-05,
      "loss": 1.2146,
      "step": 320500
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.7432000000000002e-05,
      "loss": 1.2195,
      "step": 321000
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.7428e-05,
      "loss": 1.2196,
      "step": 321500
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.7424e-05,
      "loss": 1.2138,
      "step": 322000
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.7420000000000003e-05,
      "loss": 1.2272,
      "step": 322500
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.7416e-05,
      "loss": 1.221,
      "step": 323000
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.7412000000000002e-05,
      "loss": 1.2135,
      "step": 323500
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.7408e-05,
      "loss": 1.2085,
      "step": 324000
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.7404e-05,
      "loss": 1.2114,
      "step": 324500
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 1.21,
      "step": 325000
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.7396e-05,
      "loss": 1.213,
      "step": 325500
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.7392000000000002e-05,
      "loss": 1.2166,
      "step": 326000
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.7388e-05,
      "loss": 1.2193,
      "step": 326500
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.7384e-05,
      "loss": 1.2197,
      "step": 327000
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.7380000000000003e-05,
      "loss": 1.2207,
      "step": 327500
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.7376e-05,
      "loss": 1.2155,
      "step": 328000
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.7372000000000002e-05,
      "loss": 1.2246,
      "step": 328500
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.7368e-05,
      "loss": 1.2119,
      "step": 329000
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.7364000000000002e-05,
      "loss": 1.2119,
      "step": 329500
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.736e-05,
      "loss": 1.2175,
      "step": 330000
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.7356e-05,
      "loss": 1.2194,
      "step": 330500
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.7352000000000003e-05,
      "loss": 1.2144,
      "step": 331000
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.7348e-05,
      "loss": 1.2142,
      "step": 331500
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.7344000000000002e-05,
      "loss": 1.2139,
      "step": 332000
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.734e-05,
      "loss": 1.2169,
      "step": 332500
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.7336e-05,
      "loss": 1.2131,
      "step": 333000
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.7332000000000003e-05,
      "loss": 1.2145,
      "step": 333500
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.7328e-05,
      "loss": 1.2125,
      "step": 334000
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.7324000000000002e-05,
      "loss": 1.218,
      "step": 334500
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.732e-05,
      "loss": 1.2163,
      "step": 335000
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.7316e-05,
      "loss": 1.2136,
      "step": 335500
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.7312000000000003e-05,
      "loss": 1.2045,
      "step": 336000
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.7308e-05,
      "loss": 1.2145,
      "step": 336500
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.7304000000000002e-05,
      "loss": 1.2123,
      "step": 337000
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.73e-05,
      "loss": 1.2135,
      "step": 337500
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.7296000000000002e-05,
      "loss": 1.2174,
      "step": 338000
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.7292000000000003e-05,
      "loss": 1.2096,
      "step": 338500
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.7288e-05,
      "loss": 1.2167,
      "step": 339000
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.7284e-05,
      "loss": 1.2181,
      "step": 339500
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.728e-05,
      "loss": 1.2163,
      "step": 340000
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.7276000000000002e-05,
      "loss": 1.2114,
      "step": 340500
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.7272000000000003e-05,
      "loss": 1.2157,
      "step": 341000
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.7268e-05,
      "loss": 1.2043,
      "step": 341500
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.7264e-05,
      "loss": 1.2088,
      "step": 342000
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.726e-05,
      "loss": 1.2137,
      "step": 342500
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.7256000000000002e-05,
      "loss": 1.2033,
      "step": 343000
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.7252000000000004e-05,
      "loss": 1.2114,
      "step": 343500
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.7248e-05,
      "loss": 1.206,
      "step": 344000
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.7244e-05,
      "loss": 1.2084,
      "step": 344500
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.724e-05,
      "loss": 1.2118,
      "step": 345000
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.7236000000000002e-05,
      "loss": 1.2053,
      "step": 345500
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.7232000000000004e-05,
      "loss": 1.2034,
      "step": 346000
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.7228000000000002e-05,
      "loss": 1.2034,
      "step": 346500
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.7224e-05,
      "loss": 1.2055,
      "step": 347000
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.722e-05,
      "loss": 1.2068,
      "step": 347500
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.7216000000000003e-05,
      "loss": 1.2056,
      "step": 348000
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.7212e-05,
      "loss": 1.2027,
      "step": 348500
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.7208000000000002e-05,
      "loss": 1.2045,
      "step": 349000
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.7204e-05,
      "loss": 1.2044,
      "step": 349500
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.72e-05,
      "loss": 1.2073,
      "step": 350000
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.7196000000000003e-05,
      "loss": 1.2113,
      "step": 350500
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.7192e-05,
      "loss": 1.2081,
      "step": 351000
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.7188000000000002e-05,
      "loss": 1.2076,
      "step": 351500
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.7184e-05,
      "loss": 1.2116,
      "step": 352000
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.718e-05,
      "loss": 1.209,
      "step": 352500
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.7176000000000003e-05,
      "loss": 1.2089,
      "step": 353000
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7172e-05,
      "loss": 1.2073,
      "step": 353500
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7168000000000002e-05,
      "loss": 1.2057,
      "step": 354000
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.7164e-05,
      "loss": 1.2046,
      "step": 354500
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.7160000000000002e-05,
      "loss": 1.2061,
      "step": 355000
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.7156000000000003e-05,
      "loss": 1.2079,
      "step": 355500
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.7152e-05,
      "loss": 1.2075,
      "step": 356000
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.7148000000000003e-05,
      "loss": 1.2085,
      "step": 356500
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.7144e-05,
      "loss": 1.2058,
      "step": 357000
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.7140000000000002e-05,
      "loss": 1.2032,
      "step": 357500
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.7136e-05,
      "loss": 1.197,
      "step": 358000
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.7132e-05,
      "loss": 1.2035,
      "step": 358500
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.7128000000000003e-05,
      "loss": 1.2008,
      "step": 359000
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.7124e-05,
      "loss": 1.2024,
      "step": 359500
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.7120000000000002e-05,
      "loss": 1.1994,
      "step": 360000
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.7116e-05,
      "loss": 1.2064,
      "step": 360500
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.7112e-05,
      "loss": 1.2014,
      "step": 361000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.7108000000000003e-05,
      "loss": 1.2051,
      "step": 361500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7104e-05,
      "loss": 1.2027,
      "step": 362000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7100000000000002e-05,
      "loss": 1.2051,
      "step": 362500
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.7096e-05,
      "loss": 1.2101,
      "step": 363000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.7092000000000002e-05,
      "loss": 1.2065,
      "step": 363500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.7088000000000003e-05,
      "loss": 1.2112,
      "step": 364000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.7084e-05,
      "loss": 1.2,
      "step": 364500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.7080000000000002e-05,
      "loss": 1.1968,
      "step": 365000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.7076e-05,
      "loss": 1.2012,
      "step": 365500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.7072000000000002e-05,
      "loss": 1.2013,
      "step": 366000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.7068000000000003e-05,
      "loss": 1.2086,
      "step": 366500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.7064e-05,
      "loss": 1.2017,
      "step": 367000
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.7060000000000003e-05,
      "loss": 1.1961,
      "step": 367500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.7056e-05,
      "loss": 1.2033,
      "step": 368000
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.7052000000000002e-05,
      "loss": 1.205,
      "step": 368500
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.7048000000000003e-05,
      "loss": 1.1949,
      "step": 369000
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.7044e-05,
      "loss": 1.2033,
      "step": 369500
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.704e-05,
      "loss": 1.1956,
      "step": 370000
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.7036e-05,
      "loss": 1.1983,
      "step": 370500
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.7032000000000002e-05,
      "loss": 1.1971,
      "step": 371000
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.7028000000000004e-05,
      "loss": 1.194,
      "step": 371500
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.7024e-05,
      "loss": 1.2097,
      "step": 372000
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.702e-05,
      "loss": 1.2002,
      "step": 372500
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.7016e-05,
      "loss": 1.1966,
      "step": 373000
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.7012000000000002e-05,
      "loss": 1.1937,
      "step": 373500
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.7008000000000004e-05,
      "loss": 1.1968,
      "step": 374000
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.7004000000000002e-05,
      "loss": 1.195,
      "step": 374500
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.7e-05,
      "loss": 1.1981,
      "step": 375000
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.144347906112671,
      "eval_runtime": 81.5337,
      "eval_samples_per_second": 164.509,
      "eval_steps_per_second": 2.576,
      "step": 375000
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.6996e-05,
      "loss": 1.1964,
      "step": 375500
    },
    {
      "epoch": 3.01,
      "learning_rate": 1.6992000000000003e-05,
      "loss": 1.1956,
      "step": 376000
    },
    {
      "epoch": 3.01,
      "learning_rate": 1.6988e-05,
      "loss": 1.197,
      "step": 376500
    },
    {
      "epoch": 3.02,
      "learning_rate": 1.6984000000000002e-05,
      "loss": 1.1958,
      "step": 377000
    },
    {
      "epoch": 3.02,
      "learning_rate": 1.698e-05,
      "loss": 1.1972,
      "step": 377500
    },
    {
      "epoch": 3.02,
      "learning_rate": 1.6976e-05,
      "loss": 1.1957,
      "step": 378000
    },
    {
      "epoch": 3.03,
      "learning_rate": 1.6972000000000003e-05,
      "loss": 1.1937,
      "step": 378500
    },
    {
      "epoch": 3.03,
      "learning_rate": 1.6968e-05,
      "loss": 1.1905,
      "step": 379000
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.6964e-05,
      "loss": 1.1894,
      "step": 379500
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.696e-05,
      "loss": 1.2001,
      "step": 380000
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.6956e-05,
      "loss": 1.1898,
      "step": 380500
    },
    {
      "epoch": 3.05,
      "learning_rate": 1.6952000000000003e-05,
      "loss": 1.1942,
      "step": 381000
    },
    {
      "epoch": 3.05,
      "learning_rate": 1.6948e-05,
      "loss": 1.1924,
      "step": 381500
    },
    {
      "epoch": 3.06,
      "learning_rate": 1.6944e-05,
      "loss": 1.1963,
      "step": 382000
    },
    {
      "epoch": 3.06,
      "learning_rate": 1.694e-05,
      "loss": 1.1896,
      "step": 382500
    },
    {
      "epoch": 3.06,
      "learning_rate": 1.6936000000000002e-05,
      "loss": 1.2003,
      "step": 383000
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.6932000000000003e-05,
      "loss": 1.1931,
      "step": 383500
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.6928e-05,
      "loss": 1.1962,
      "step": 384000
    },
    {
      "epoch": 3.08,
      "learning_rate": 1.6924e-05,
      "loss": 1.1963,
      "step": 384500
    },
    {
      "epoch": 3.08,
      "learning_rate": 1.692e-05,
      "loss": 1.1862,
      "step": 385000
    },
    {
      "epoch": 3.08,
      "learning_rate": 1.6916000000000002e-05,
      "loss": 1.1885,
      "step": 385500
    },
    {
      "epoch": 3.09,
      "learning_rate": 1.6912000000000003e-05,
      "loss": 1.1898,
      "step": 386000
    },
    {
      "epoch": 3.09,
      "learning_rate": 1.6908e-05,
      "loss": 1.1947,
      "step": 386500
    },
    {
      "epoch": 3.1,
      "learning_rate": 1.6904e-05,
      "loss": 1.191,
      "step": 387000
    },
    {
      "epoch": 3.1,
      "learning_rate": 1.69e-05,
      "loss": 1.1958,
      "step": 387500
    },
    {
      "epoch": 3.1,
      "learning_rate": 1.6896000000000002e-05,
      "loss": 1.1936,
      "step": 388000
    },
    {
      "epoch": 3.11,
      "learning_rate": 1.6892e-05,
      "loss": 1.1912,
      "step": 388500
    },
    {
      "epoch": 3.11,
      "learning_rate": 1.6888e-05,
      "loss": 1.1888,
      "step": 389000
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.6884e-05,
      "loss": 1.201,
      "step": 389500
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.688e-05,
      "loss": 1.1902,
      "step": 390000
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.6876000000000002e-05,
      "loss": 1.1854,
      "step": 390500
    },
    {
      "epoch": 3.13,
      "learning_rate": 1.6872e-05,
      "loss": 1.1942,
      "step": 391000
    },
    {
      "epoch": 3.13,
      "learning_rate": 1.6868000000000002e-05,
      "loss": 1.1923,
      "step": 391500
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.6864e-05,
      "loss": 1.1982,
      "step": 392000
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.686e-05,
      "loss": 1.1952,
      "step": 392500
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.6856000000000003e-05,
      "loss": 1.1924,
      "step": 393000
    },
    {
      "epoch": 3.15,
      "learning_rate": 1.6852e-05,
      "loss": 1.1905,
      "step": 393500
    },
    {
      "epoch": 3.15,
      "learning_rate": 1.6848000000000002e-05,
      "loss": 1.1935,
      "step": 394000
    },
    {
      "epoch": 3.16,
      "learning_rate": 1.6844e-05,
      "loss": 1.1952,
      "step": 394500
    },
    {
      "epoch": 3.16,
      "learning_rate": 1.684e-05,
      "loss": 1.192,
      "step": 395000
    },
    {
      "epoch": 3.16,
      "learning_rate": 1.6836000000000003e-05,
      "loss": 1.1836,
      "step": 395500
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.6832e-05,
      "loss": 1.1882,
      "step": 396000
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.6828000000000002e-05,
      "loss": 1.1887,
      "step": 396500
    },
    {
      "epoch": 3.18,
      "learning_rate": 1.6824e-05,
      "loss": 1.1919,
      "step": 397000
    },
    {
      "epoch": 3.18,
      "learning_rate": 1.682e-05,
      "loss": 1.187,
      "step": 397500
    },
    {
      "epoch": 3.18,
      "learning_rate": 1.6816e-05,
      "loss": 1.1905,
      "step": 398000
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.6812e-05,
      "loss": 1.1838,
      "step": 398500
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.6808000000000002e-05,
      "loss": 1.1948,
      "step": 399000
    },
    {
      "epoch": 3.2,
      "learning_rate": 1.6804e-05,
      "loss": 1.1851,
      "step": 399500
    },
    {
      "epoch": 3.2,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 1.1852,
      "step": 400000
    },
    {
      "epoch": 3.2,
      "learning_rate": 1.6796e-05,
      "loss": 1.1913,
      "step": 400500
    },
    {
      "epoch": 3.21,
      "learning_rate": 1.6792e-05,
      "loss": 1.1823,
      "step": 401000
    },
    {
      "epoch": 3.21,
      "learning_rate": 1.6788000000000003e-05,
      "loss": 1.1902,
      "step": 401500
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.6784e-05,
      "loss": 1.1918,
      "step": 402000
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.6780000000000002e-05,
      "loss": 1.1869,
      "step": 402500
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.6776e-05,
      "loss": 1.1833,
      "step": 403000
    },
    {
      "epoch": 3.23,
      "learning_rate": 1.6772e-05,
      "loss": 1.1936,
      "step": 403500
    },
    {
      "epoch": 3.23,
      "learning_rate": 1.6768000000000003e-05,
      "loss": 1.1898,
      "step": 404000
    },
    {
      "epoch": 3.24,
      "learning_rate": 1.6764e-05,
      "loss": 1.1863,
      "step": 404500
    },
    {
      "epoch": 3.24,
      "learning_rate": 1.6760000000000002e-05,
      "loss": 1.1815,
      "step": 405000
    },
    {
      "epoch": 3.24,
      "learning_rate": 1.6756e-05,
      "loss": 1.1915,
      "step": 405500
    },
    {
      "epoch": 3.25,
      "learning_rate": 1.6752e-05,
      "loss": 1.1861,
      "step": 406000
    },
    {
      "epoch": 3.25,
      "learning_rate": 1.6748000000000003e-05,
      "loss": 1.187,
      "step": 406500
    },
    {
      "epoch": 3.26,
      "learning_rate": 1.6744e-05,
      "loss": 1.1894,
      "step": 407000
    },
    {
      "epoch": 3.26,
      "learning_rate": 1.6740000000000002e-05,
      "loss": 1.191,
      "step": 407500
    },
    {
      "epoch": 3.26,
      "learning_rate": 1.6736e-05,
      "loss": 1.1859,
      "step": 408000
    },
    {
      "epoch": 3.27,
      "learning_rate": 1.6732000000000002e-05,
      "loss": 1.1908,
      "step": 408500
    },
    {
      "epoch": 3.27,
      "learning_rate": 1.6728000000000003e-05,
      "loss": 1.1944,
      "step": 409000
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.6724e-05,
      "loss": 1.1877,
      "step": 409500
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.672e-05,
      "loss": 1.1828,
      "step": 410000
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.6716e-05,
      "loss": 1.1844,
      "step": 410500
    },
    {
      "epoch": 3.29,
      "learning_rate": 1.6712000000000002e-05,
      "loss": 1.1919,
      "step": 411000
    },
    {
      "epoch": 3.29,
      "learning_rate": 1.6708000000000003e-05,
      "loss": 1.1876,
      "step": 411500
    },
    {
      "epoch": 3.3,
      "learning_rate": 1.6704e-05,
      "loss": 1.1812,
      "step": 412000
    },
    {
      "epoch": 3.3,
      "learning_rate": 1.67e-05,
      "loss": 1.1826,
      "step": 412500
    },
    {
      "epoch": 3.3,
      "learning_rate": 1.6696e-05,
      "loss": 1.1894,
      "step": 413000
    },
    {
      "epoch": 3.31,
      "learning_rate": 1.6692000000000002e-05,
      "loss": 1.1832,
      "step": 413500
    },
    {
      "epoch": 3.31,
      "learning_rate": 1.6688000000000004e-05,
      "loss": 1.1883,
      "step": 414000
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.6684e-05,
      "loss": 1.1826,
      "step": 414500
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.668e-05,
      "loss": 1.1855,
      "step": 415000
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.6676e-05,
      "loss": 1.187,
      "step": 415500
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.6672000000000002e-05,
      "loss": 1.1851,
      "step": 416000
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.6668e-05,
      "loss": 1.1814,
      "step": 416500
    },
    {
      "epoch": 3.34,
      "learning_rate": 1.6664000000000002e-05,
      "loss": 1.1778,
      "step": 417000
    },
    {
      "epoch": 3.34,
      "learning_rate": 1.666e-05,
      "loss": 1.1759,
      "step": 417500
    },
    {
      "epoch": 3.34,
      "learning_rate": 1.6656e-05,
      "loss": 1.1782,
      "step": 418000
    },
    {
      "epoch": 3.35,
      "learning_rate": 1.6652000000000003e-05,
      "loss": 1.1881,
      "step": 418500
    },
    {
      "epoch": 3.35,
      "learning_rate": 1.6648e-05,
      "loss": 1.1909,
      "step": 419000
    },
    {
      "epoch": 3.36,
      "learning_rate": 1.6644000000000002e-05,
      "loss": 1.1796,
      "step": 419500
    },
    {
      "epoch": 3.36,
      "learning_rate": 1.664e-05,
      "loss": 1.1754,
      "step": 420000
    },
    {
      "epoch": 3.36,
      "learning_rate": 1.6636e-05,
      "loss": 1.1875,
      "step": 420500
    },
    {
      "epoch": 3.37,
      "learning_rate": 1.6632000000000003e-05,
      "loss": 1.1817,
      "step": 421000
    },
    {
      "epoch": 3.37,
      "learning_rate": 1.6628e-05,
      "loss": 1.1857,
      "step": 421500
    },
    {
      "epoch": 3.38,
      "learning_rate": 1.6624000000000002e-05,
      "loss": 1.1819,
      "step": 422000
    },
    {
      "epoch": 3.38,
      "learning_rate": 1.662e-05,
      "loss": 1.1793,
      "step": 422500
    },
    {
      "epoch": 3.38,
      "learning_rate": 1.6616e-05,
      "loss": 1.1847,
      "step": 423000
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.6612000000000003e-05,
      "loss": 1.185,
      "step": 423500
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.6608e-05,
      "loss": 1.1784,
      "step": 424000
    },
    {
      "epoch": 3.4,
      "learning_rate": 1.6604000000000002e-05,
      "loss": 1.1818,
      "step": 424500
    },
    {
      "epoch": 3.4,
      "learning_rate": 1.66e-05,
      "loss": 1.1802,
      "step": 425000
    },
    {
      "epoch": 3.4,
      "learning_rate": 1.6596000000000002e-05,
      "loss": 1.177,
      "step": 425500
    },
    {
      "epoch": 3.41,
      "learning_rate": 1.6592000000000003e-05,
      "loss": 1.1916,
      "step": 426000
    },
    {
      "epoch": 3.41,
      "learning_rate": 1.6588e-05,
      "loss": 1.1828,
      "step": 426500
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.6584000000000002e-05,
      "loss": 1.1867,
      "step": 427000
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.658e-05,
      "loss": 1.1824,
      "step": 427500
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.6576000000000002e-05,
      "loss": 1.1842,
      "step": 428000
    },
    {
      "epoch": 3.43,
      "learning_rate": 1.6572e-05,
      "loss": 1.1784,
      "step": 428500
    },
    {
      "epoch": 3.43,
      "learning_rate": 1.6568e-05,
      "loss": 1.1784,
      "step": 429000
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.6564000000000003e-05,
      "loss": 1.1832,
      "step": 429500
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.656e-05,
      "loss": 1.1866,
      "step": 430000
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.6556000000000002e-05,
      "loss": 1.1874,
      "step": 430500
    },
    {
      "epoch": 3.45,
      "learning_rate": 1.6552e-05,
      "loss": 1.1833,
      "step": 431000
    },
    {
      "epoch": 3.45,
      "learning_rate": 1.6548e-05,
      "loss": 1.177,
      "step": 431500
    },
    {
      "epoch": 3.46,
      "learning_rate": 1.6544000000000003e-05,
      "loss": 1.186,
      "step": 432000
    },
    {
      "epoch": 3.46,
      "learning_rate": 1.654e-05,
      "loss": 1.1775,
      "step": 432500
    },
    {
      "epoch": 3.46,
      "learning_rate": 1.6536000000000002e-05,
      "loss": 1.1769,
      "step": 433000
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.6532e-05,
      "loss": 1.1782,
      "step": 433500
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.6528e-05,
      "loss": 1.176,
      "step": 434000
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.6524000000000003e-05,
      "loss": 1.1798,
      "step": 434500
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.652e-05,
      "loss": 1.1817,
      "step": 435000
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.6516000000000002e-05,
      "loss": 1.1744,
      "step": 435500
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.6512e-05,
      "loss": 1.1763,
      "step": 436000
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.6508000000000002e-05,
      "loss": 1.1829,
      "step": 436500
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.6504000000000003e-05,
      "loss": 1.1812,
      "step": 437000
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.65e-05,
      "loss": 1.1802,
      "step": 437500
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.6496e-05,
      "loss": 1.1792,
      "step": 438000
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.6492e-05,
      "loss": 1.1719,
      "step": 438500
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.6488000000000002e-05,
      "loss": 1.1821,
      "step": 439000
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6484000000000003e-05,
      "loss": 1.1748,
      "step": 439500
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.648e-05,
      "loss": 1.1721,
      "step": 440000
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6476e-05,
      "loss": 1.1778,
      "step": 440500
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.6472e-05,
      "loss": 1.1827,
      "step": 441000
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.6468000000000002e-05,
      "loss": 1.1802,
      "step": 441500
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.6464000000000004e-05,
      "loss": 1.1715,
      "step": 442000
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.646e-05,
      "loss": 1.1792,
      "step": 442500
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.6456e-05,
      "loss": 1.1726,
      "step": 443000
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.6452e-05,
      "loss": 1.1764,
      "step": 443500
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.6448000000000002e-05,
      "loss": 1.1704,
      "step": 444000
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.6444000000000004e-05,
      "loss": 1.1757,
      "step": 444500
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.6440000000000002e-05,
      "loss": 1.1678,
      "step": 445000
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.6436e-05,
      "loss": 1.1785,
      "step": 445500
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.6432e-05,
      "loss": 1.1728,
      "step": 446000
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.6428000000000003e-05,
      "loss": 1.1744,
      "step": 446500
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.6424e-05,
      "loss": 1.1834,
      "step": 447000
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.6420000000000002e-05,
      "loss": 1.1728,
      "step": 447500
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.6416e-05,
      "loss": 1.1737,
      "step": 448000
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.6412e-05,
      "loss": 1.1758,
      "step": 448500
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.6408000000000003e-05,
      "loss": 1.1765,
      "step": 449000
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.6404e-05,
      "loss": 1.1766,
      "step": 449500
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.64e-05,
      "loss": 1.1749,
      "step": 450000
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.6396e-05,
      "loss": 1.1668,
      "step": 450500
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.6392e-05,
      "loss": 1.1831,
      "step": 451000
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.6388000000000003e-05,
      "loss": 1.1829,
      "step": 451500
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.6384e-05,
      "loss": 1.1731,
      "step": 452000
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.638e-05,
      "loss": 1.1816,
      "step": 452500
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.6376e-05,
      "loss": 1.1754,
      "step": 453000
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.6372000000000002e-05,
      "loss": 1.1797,
      "step": 453500
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.6368000000000003e-05,
      "loss": 1.1721,
      "step": 454000
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.6364e-05,
      "loss": 1.1729,
      "step": 454500
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.636e-05,
      "loss": 1.1752,
      "step": 455000
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.6356e-05,
      "loss": 1.1668,
      "step": 455500
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.6352000000000002e-05,
      "loss": 1.17,
      "step": 456000
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.6348e-05,
      "loss": 1.1812,
      "step": 456500
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.6344e-05,
      "loss": 1.1788,
      "step": 457000
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.634e-05,
      "loss": 1.1743,
      "step": 457500
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.6336e-05,
      "loss": 1.1749,
      "step": 458000
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.6332000000000002e-05,
      "loss": 1.1717,
      "step": 458500
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.6328e-05,
      "loss": 1.1752,
      "step": 459000
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.6324e-05,
      "loss": 1.1732,
      "step": 459500
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.632e-05,
      "loss": 1.1763,
      "step": 460000
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.6316e-05,
      "loss": 1.1745,
      "step": 460500
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.6312000000000002e-05,
      "loss": 1.1701,
      "step": 461000
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.6308e-05,
      "loss": 1.1702,
      "step": 461500
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.6304000000000002e-05,
      "loss": 1.1753,
      "step": 462000
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.63e-05,
      "loss": 1.1708,
      "step": 462500
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.6296e-05,
      "loss": 1.1715,
      "step": 463000
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.6292000000000003e-05,
      "loss": 1.1687,
      "step": 463500
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.6288e-05,
      "loss": 1.1749,
      "step": 464000
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.6284000000000002e-05,
      "loss": 1.1726,
      "step": 464500
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.628e-05,
      "loss": 1.1722,
      "step": 465000
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.6276e-05,
      "loss": 1.1736,
      "step": 465500
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.6272000000000003e-05,
      "loss": 1.17,
      "step": 466000
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.6268e-05,
      "loss": 1.1682,
      "step": 466500
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.6264000000000002e-05,
      "loss": 1.1698,
      "step": 467000
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.626e-05,
      "loss": 1.1782,
      "step": 467500
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.6256e-05,
      "loss": 1.1645,
      "step": 468000
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.6252e-05,
      "loss": 1.1628,
      "step": 468500
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.6248e-05,
      "loss": 1.1681,
      "step": 469000
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.6244000000000002e-05,
      "loss": 1.1764,
      "step": 469500
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.6240000000000004e-05,
      "loss": 1.1692,
      "step": 470000
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.6236000000000002e-05,
      "loss": 1.1682,
      "step": 470500
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.6232e-05,
      "loss": 1.1692,
      "step": 471000
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.6228e-05,
      "loss": 1.1715,
      "step": 471500
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.6224000000000003e-05,
      "loss": 1.1729,
      "step": 472000
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.6220000000000004e-05,
      "loss": 1.1708,
      "step": 472500
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.6216000000000002e-05,
      "loss": 1.1658,
      "step": 473000
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.6212e-05,
      "loss": 1.1732,
      "step": 473500
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.6208e-05,
      "loss": 1.1708,
      "step": 474000
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.6204000000000003e-05,
      "loss": 1.1728,
      "step": 474500
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.62e-05,
      "loss": 1.1706,
      "step": 475000
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.6196000000000002e-05,
      "loss": 1.1685,
      "step": 475500
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.6192e-05,
      "loss": 1.1747,
      "step": 476000
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.6188e-05,
      "loss": 1.1659,
      "step": 476500
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.6184000000000003e-05,
      "loss": 1.1724,
      "step": 477000
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.618e-05,
      "loss": 1.1669,
      "step": 477500
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.6176e-05,
      "loss": 1.1697,
      "step": 478000
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.6172e-05,
      "loss": 1.1685,
      "step": 478500
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.6168000000000002e-05,
      "loss": 1.1733,
      "step": 479000
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.6164000000000003e-05,
      "loss": 1.1688,
      "step": 479500
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.616e-05,
      "loss": 1.1703,
      "step": 480000
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.6156e-05,
      "loss": 1.1612,
      "step": 480500
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.6152e-05,
      "loss": 1.1667,
      "step": 481000
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.6148000000000002e-05,
      "loss": 1.1614,
      "step": 481500
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.6144000000000003e-05,
      "loss": 1.1668,
      "step": 482000
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.614e-05,
      "loss": 1.1684,
      "step": 482500
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.6136e-05,
      "loss": 1.1666,
      "step": 483000
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.6132e-05,
      "loss": 1.1691,
      "step": 483500
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.6128000000000002e-05,
      "loss": 1.1692,
      "step": 484000
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.6124000000000004e-05,
      "loss": 1.1654,
      "step": 484500
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.612e-05,
      "loss": 1.1644,
      "step": 485000
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.6116e-05,
      "loss": 1.1645,
      "step": 485500
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.6112e-05,
      "loss": 1.1735,
      "step": 486000
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.6108000000000002e-05,
      "loss": 1.1704,
      "step": 486500
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.6104e-05,
      "loss": 1.164,
      "step": 487000
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.6100000000000002e-05,
      "loss": 1.1728,
      "step": 487500
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.6096e-05,
      "loss": 1.1637,
      "step": 488000
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.6092e-05,
      "loss": 1.1682,
      "step": 488500
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.6088000000000002e-05,
      "loss": 1.167,
      "step": 489000
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.6084e-05,
      "loss": 1.167,
      "step": 489500
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.6080000000000002e-05,
      "loss": 1.1655,
      "step": 490000
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.6076e-05,
      "loss": 1.1643,
      "step": 490500
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.6072e-05,
      "loss": 1.1687,
      "step": 491000
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.6068000000000003e-05,
      "loss": 1.1636,
      "step": 491500
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.6064e-05,
      "loss": 1.1609,
      "step": 492000
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.6060000000000002e-05,
      "loss": 1.1678,
      "step": 492500
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.6056e-05,
      "loss": 1.1704,
      "step": 493000
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.6052e-05,
      "loss": 1.1666,
      "step": 493500
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.6048000000000003e-05,
      "loss": 1.1636,
      "step": 494000
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.6044e-05,
      "loss": 1.1686,
      "step": 494500
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.6040000000000002e-05,
      "loss": 1.1635,
      "step": 495000
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.6036e-05,
      "loss": 1.1607,
      "step": 495500
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.6032e-05,
      "loss": 1.1592,
      "step": 496000
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.6028e-05,
      "loss": 1.1602,
      "step": 496500
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.6024e-05,
      "loss": 1.1608,
      "step": 497000
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.6020000000000002e-05,
      "loss": 1.1578,
      "step": 497500
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.6016e-05,
      "loss": 1.1653,
      "step": 498000
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.6012000000000002e-05,
      "loss": 1.1625,
      "step": 498500
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.6008e-05,
      "loss": 1.1701,
      "step": 499000
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.6004e-05,
      "loss": 1.1644,
      "step": 499500
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.1652,
      "step": 500000
    },
    {
      "epoch": 4.0,
      "eval_loss": 1.1185648441314697,
      "eval_runtime": 90.4279,
      "eval_samples_per_second": 148.328,
      "eval_steps_per_second": 2.322,
      "step": 500000
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.5996e-05,
      "loss": 1.1599,
      "step": 500500
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.5992000000000002e-05,
      "loss": 1.1548,
      "step": 501000
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.5988e-05,
      "loss": 1.1586,
      "step": 501500
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.5984e-05,
      "loss": 1.1593,
      "step": 502000
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.5980000000000003e-05,
      "loss": 1.1563,
      "step": 502500
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.5976e-05,
      "loss": 1.1611,
      "step": 503000
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.5972000000000002e-05,
      "loss": 1.1596,
      "step": 503500
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.5968e-05,
      "loss": 1.167,
      "step": 504000
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.5964e-05,
      "loss": 1.1596,
      "step": 504500
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.5960000000000003e-05,
      "loss": 1.1587,
      "step": 505000
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.5956e-05,
      "loss": 1.1529,
      "step": 505500
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.5952000000000002e-05,
      "loss": 1.1622,
      "step": 506000
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.5948e-05,
      "loss": 1.1545,
      "step": 506500
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.5944000000000002e-05,
      "loss": 1.1604,
      "step": 507000
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.5940000000000003e-05,
      "loss": 1.1553,
      "step": 507500
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.5936e-05,
      "loss": 1.1561,
      "step": 508000
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.5932e-05,
      "loss": 1.161,
      "step": 508500
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.5928e-05,
      "loss": 1.1563,
      "step": 509000
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.5924000000000002e-05,
      "loss": 1.1626,
      "step": 509500
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.5920000000000003e-05,
      "loss": 1.1587,
      "step": 510000
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.5916e-05,
      "loss": 1.1594,
      "step": 510500
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.5912e-05,
      "loss": 1.1605,
      "step": 511000
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.5908e-05,
      "loss": 1.1605,
      "step": 511500
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.5904000000000002e-05,
      "loss": 1.1573,
      "step": 512000
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.5900000000000004e-05,
      "loss": 1.1549,
      "step": 512500
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.5896e-05,
      "loss": 1.162,
      "step": 513000
    },
    {
      "epoch": 4.11,
      "learning_rate": 1.5892e-05,
      "loss": 1.1561,
      "step": 513500
    },
    {
      "epoch": 4.11,
      "learning_rate": 1.5888e-05,
      "loss": 1.1526,
      "step": 514000
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.5884000000000002e-05,
      "loss": 1.1637,
      "step": 514500
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.588e-05,
      "loss": 1.1591,
      "step": 515000
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.5876000000000002e-05,
      "loss": 1.155,
      "step": 515500
    },
    {
      "epoch": 4.13,
      "learning_rate": 1.5872e-05,
      "loss": 1.1536,
      "step": 516000
    },
    {
      "epoch": 4.13,
      "learning_rate": 1.5868e-05,
      "loss": 1.1491,
      "step": 516500
    },
    {
      "epoch": 4.14,
      "learning_rate": 1.5864000000000003e-05,
      "loss": 1.1498,
      "step": 517000
    },
    {
      "epoch": 4.14,
      "learning_rate": 1.586e-05,
      "loss": 1.1559,
      "step": 517500
    },
    {
      "epoch": 4.14,
      "learning_rate": 1.5856e-05,
      "loss": 1.159,
      "step": 518000
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.5852e-05,
      "loss": 1.1526,
      "step": 518500
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.5848e-05,
      "loss": 1.1584,
      "step": 519000
    },
    {
      "epoch": 4.16,
      "learning_rate": 1.5844000000000003e-05,
      "loss": 1.1558,
      "step": 519500
    },
    {
      "epoch": 4.16,
      "learning_rate": 1.584e-05,
      "loss": 1.1523,
      "step": 520000
    },
    {
      "epoch": 4.16,
      "learning_rate": 1.5836e-05,
      "loss": 1.1543,
      "step": 520500
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.5832e-05,
      "loss": 1.1553,
      "step": 521000
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.5828e-05,
      "loss": 1.1503,
      "step": 521500
    },
    {
      "epoch": 4.18,
      "learning_rate": 1.5824000000000003e-05,
      "loss": 1.1507,
      "step": 522000
    },
    {
      "epoch": 4.18,
      "learning_rate": 1.582e-05,
      "loss": 1.1615,
      "step": 522500
    },
    {
      "epoch": 4.18,
      "learning_rate": 1.5816e-05,
      "loss": 1.1574,
      "step": 523000
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.5812e-05,
      "loss": 1.1551,
      "step": 523500
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.5808000000000002e-05,
      "loss": 1.1556,
      "step": 524000
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.5804000000000003e-05,
      "loss": 1.1576,
      "step": 524500
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.58e-05,
      "loss": 1.1539,
      "step": 525000
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.5796e-05,
      "loss": 1.1565,
      "step": 525500
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.5792e-05,
      "loss": 1.1587,
      "step": 526000
    },
    {
      "epoch": 4.21,
      "learning_rate": 1.5788000000000002e-05,
      "loss": 1.1537,
      "step": 526500
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.5784e-05,
      "loss": 1.1569,
      "step": 527000
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.578e-05,
      "loss": 1.1536,
      "step": 527500
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.5776e-05,
      "loss": 1.1576,
      "step": 528000
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.5772e-05,
      "loss": 1.1559,
      "step": 528500
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.5768000000000002e-05,
      "loss": 1.1601,
      "step": 529000
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.5764e-05,
      "loss": 1.1544,
      "step": 529500
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.576e-05,
      "loss": 1.1554,
      "step": 530000
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.5756e-05,
      "loss": 1.155,
      "step": 530500
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.5752e-05,
      "loss": 1.159,
      "step": 531000
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.5748000000000002e-05,
      "loss": 1.1549,
      "step": 531500
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.5744e-05,
      "loss": 1.1496,
      "step": 532000
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.5740000000000002e-05,
      "loss": 1.1525,
      "step": 532500
    },
    {
      "epoch": 4.26,
      "learning_rate": 1.5736000000000003e-05,
      "loss": 1.154,
      "step": 533000
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.5732e-05,
      "loss": 1.157,
      "step": 533500
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.5728000000000003e-05,
      "loss": 1.1542,
      "step": 534000
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.5724e-05,
      "loss": 1.1529,
      "step": 534500
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.5720000000000002e-05,
      "loss": 1.1508,
      "step": 535000
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.5716000000000003e-05,
      "loss": 1.1584,
      "step": 535500
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.5712e-05,
      "loss": 1.1536,
      "step": 536000
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.5708e-05,
      "loss": 1.1577,
      "step": 536500
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.5704e-05,
      "loss": 1.1519,
      "step": 537000
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.5700000000000002e-05,
      "loss": 1.1511,
      "step": 537500
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.5696000000000004e-05,
      "loss": 1.1562,
      "step": 538000
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.5692e-05,
      "loss": 1.1563,
      "step": 538500
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.5688e-05,
      "loss": 1.1557,
      "step": 539000
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.5684e-05,
      "loss": 1.1491,
      "step": 539500
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.5680000000000002e-05,
      "loss": 1.1512,
      "step": 540000
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.5676000000000004e-05,
      "loss": 1.1544,
      "step": 540500
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.5672000000000002e-05,
      "loss": 1.1547,
      "step": 541000
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.5668e-05,
      "loss": 1.1499,
      "step": 541500
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.5664e-05,
      "loss": 1.1503,
      "step": 542000
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.5660000000000003e-05,
      "loss": 1.1546,
      "step": 542500
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.5656000000000004e-05,
      "loss": 1.1511,
      "step": 543000
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.5652000000000002e-05,
      "loss": 1.1508,
      "step": 543500
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.5648e-05,
      "loss": 1.15,
      "step": 544000
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.5644e-05,
      "loss": 1.15,
      "step": 544500
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.5640000000000003e-05,
      "loss": 1.1526,
      "step": 545000
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.5636e-05,
      "loss": 1.1565,
      "step": 545500
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.5632000000000002e-05,
      "loss": 1.149,
      "step": 546000
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.5628e-05,
      "loss": 1.1605,
      "step": 546500
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.5624e-05,
      "loss": 1.1538,
      "step": 547000
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.5620000000000003e-05,
      "loss": 1.148,
      "step": 547500
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.5616e-05,
      "loss": 1.1501,
      "step": 548000
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.5612e-05,
      "loss": 1.1526,
      "step": 548500
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.5608e-05,
      "loss": 1.155,
      "step": 549000
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.5604000000000002e-05,
      "loss": 1.1529,
      "step": 549500
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 1.1506,
      "step": 550000
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.5596e-05,
      "loss": 1.1557,
      "step": 550500
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.5592e-05,
      "loss": 1.1498,
      "step": 551000
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.5588e-05,
      "loss": 1.153,
      "step": 551500
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.5584000000000002e-05,
      "loss": 1.1534,
      "step": 552000
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.5580000000000003e-05,
      "loss": 1.1564,
      "step": 552500
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.5576e-05,
      "loss": 1.1463,
      "step": 553000
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.5572e-05,
      "loss": 1.1479,
      "step": 553500
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.5568e-05,
      "loss": 1.15,
      "step": 554000
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.5564000000000002e-05,
      "loss": 1.1564,
      "step": 554500
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.556e-05,
      "loss": 1.146,
      "step": 555000
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.5556e-05,
      "loss": 1.1495,
      "step": 555500
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.5552e-05,
      "loss": 1.1479,
      "step": 556000
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.5548e-05,
      "loss": 1.1542,
      "step": 556500
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.5544000000000002e-05,
      "loss": 1.1566,
      "step": 557000
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.554e-05,
      "loss": 1.1505,
      "step": 557500
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.5536e-05,
      "loss": 1.149,
      "step": 558000
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.5532e-05,
      "loss": 1.1447,
      "step": 558500
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.5528e-05,
      "loss": 1.1472,
      "step": 559000
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.5524000000000002e-05,
      "loss": 1.1476,
      "step": 559500
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.552e-05,
      "loss": 1.1506,
      "step": 560000
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.5516000000000002e-05,
      "loss": 1.1446,
      "step": 560500
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.5512e-05,
      "loss": 1.1546,
      "step": 561000
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.5508e-05,
      "loss": 1.1502,
      "step": 561500
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.5504000000000003e-05,
      "loss": 1.148,
      "step": 562000
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.55e-05,
      "loss": 1.1537,
      "step": 562500
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.5496000000000002e-05,
      "loss": 1.1527,
      "step": 563000
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.5492e-05,
      "loss": 1.1474,
      "step": 563500
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.5488e-05,
      "loss": 1.1475,
      "step": 564000
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.5484000000000003e-05,
      "loss": 1.1504,
      "step": 564500
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.548e-05,
      "loss": 1.1429,
      "step": 565000
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.5476000000000002e-05,
      "loss": 1.1501,
      "step": 565500
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.5472e-05,
      "loss": 1.1445,
      "step": 566000
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.5468e-05,
      "loss": 1.1389,
      "step": 566500
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.5464e-05,
      "loss": 1.1516,
      "step": 567000
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.546e-05,
      "loss": 1.1445,
      "step": 567500
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.5456000000000002e-05,
      "loss": 1.1443,
      "step": 568000
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.5452e-05,
      "loss": 1.1487,
      "step": 568500
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.5448000000000002e-05,
      "loss": 1.1455,
      "step": 569000
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.5444e-05,
      "loss": 1.1459,
      "step": 569500
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.544e-05,
      "loss": 1.1431,
      "step": 570000
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.5436000000000003e-05,
      "loss": 1.1438,
      "step": 570500
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.5432e-05,
      "loss": 1.1398,
      "step": 571000
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.5428000000000002e-05,
      "loss": 1.1496,
      "step": 571500
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.5424e-05,
      "loss": 1.1434,
      "step": 572000
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.542e-05,
      "loss": 1.1453,
      "step": 572500
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.5416000000000003e-05,
      "loss": 1.1447,
      "step": 573000
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.5412e-05,
      "loss": 1.1413,
      "step": 573500
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.5408000000000002e-05,
      "loss": 1.1439,
      "step": 574000
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.5404e-05,
      "loss": 1.142,
      "step": 574500
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.54e-05,
      "loss": 1.1379,
      "step": 575000
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.5396000000000003e-05,
      "loss": 1.1464,
      "step": 575500
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.5392e-05,
      "loss": 1.1433,
      "step": 576000
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.5388e-05,
      "loss": 1.1473,
      "step": 576500
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.5384e-05,
      "loss": 1.1487,
      "step": 577000
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.5380000000000002e-05,
      "loss": 1.1462,
      "step": 577500
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.5376000000000003e-05,
      "loss": 1.1419,
      "step": 578000
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.5372e-05,
      "loss": 1.143,
      "step": 578500
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.5368e-05,
      "loss": 1.1429,
      "step": 579000
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.5364e-05,
      "loss": 1.1469,
      "step": 579500
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.5360000000000002e-05,
      "loss": 1.1418,
      "step": 580000
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.5356000000000003e-05,
      "loss": 1.1427,
      "step": 580500
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.5352e-05,
      "loss": 1.1446,
      "step": 581000
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.5348e-05,
      "loss": 1.1447,
      "step": 581500
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.5344e-05,
      "loss": 1.1427,
      "step": 582000
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.5340000000000002e-05,
      "loss": 1.1451,
      "step": 582500
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.5336000000000004e-05,
      "loss": 1.1531,
      "step": 583000
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.5332e-05,
      "loss": 1.1482,
      "step": 583500
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.5328e-05,
      "loss": 1.1402,
      "step": 584000
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.5324e-05,
      "loss": 1.1497,
      "step": 584500
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.5320000000000002e-05,
      "loss": 1.1405,
      "step": 585000
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.5316e-05,
      "loss": 1.1484,
      "step": 585500
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.5312000000000002e-05,
      "loss": 1.1422,
      "step": 586000
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.5308e-05,
      "loss": 1.1462,
      "step": 586500
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.5304e-05,
      "loss": 1.1398,
      "step": 587000
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.5300000000000003e-05,
      "loss": 1.1404,
      "step": 587500
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.5296e-05,
      "loss": 1.1431,
      "step": 588000
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.5292e-05,
      "loss": 1.1457,
      "step": 588500
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.5288e-05,
      "loss": 1.1436,
      "step": 589000
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.5284e-05,
      "loss": 1.1406,
      "step": 589500
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.5280000000000003e-05,
      "loss": 1.1449,
      "step": 590000
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.5276e-05,
      "loss": 1.1451,
      "step": 590500
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.5272e-05,
      "loss": 1.1442,
      "step": 591000
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.5268e-05,
      "loss": 1.1447,
      "step": 591500
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.5264e-05,
      "loss": 1.1363,
      "step": 592000
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.5260000000000003e-05,
      "loss": 1.1394,
      "step": 592500
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.5256000000000003e-05,
      "loss": 1.1427,
      "step": 593000
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.5252e-05,
      "loss": 1.1433,
      "step": 593500
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.5248e-05,
      "loss": 1.1434,
      "step": 594000
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.5244000000000002e-05,
      "loss": 1.1395,
      "step": 594500
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.5240000000000001e-05,
      "loss": 1.1394,
      "step": 595000
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.5236000000000001e-05,
      "loss": 1.1442,
      "step": 595500
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.5232000000000003e-05,
      "loss": 1.1438,
      "step": 596000
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.5228e-05,
      "loss": 1.1381,
      "step": 596500
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5224e-05,
      "loss": 1.1445,
      "step": 597000
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5220000000000002e-05,
      "loss": 1.1399,
      "step": 597500
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.5216000000000001e-05,
      "loss": 1.1416,
      "step": 598000
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5212000000000003e-05,
      "loss": 1.1322,
      "step": 598500
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.5208e-05,
      "loss": 1.1481,
      "step": 599000
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.5204e-05,
      "loss": 1.1438,
      "step": 599500
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 1.1459,
      "step": 600000
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.5196000000000002e-05,
      "loss": 1.1441,
      "step": 600500
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.5192000000000003e-05,
      "loss": 1.1437,
      "step": 601000
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.5188000000000001e-05,
      "loss": 1.1386,
      "step": 601500
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.5184e-05,
      "loss": 1.1394,
      "step": 602000
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.5180000000000002e-05,
      "loss": 1.1445,
      "step": 602500
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.5176000000000002e-05,
      "loss": 1.1377,
      "step": 603000
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.5172000000000001e-05,
      "loss": 1.1405,
      "step": 603500
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.5168000000000001e-05,
      "loss": 1.1404,
      "step": 604000
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5164e-05,
      "loss": 1.1472,
      "step": 604500
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.516e-05,
      "loss": 1.1384,
      "step": 605000
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5156000000000002e-05,
      "loss": 1.1396,
      "step": 605500
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5152000000000002e-05,
      "loss": 1.1337,
      "step": 606000
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5148e-05,
      "loss": 1.1399,
      "step": 606500
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5144000000000001e-05,
      "loss": 1.1407,
      "step": 607000
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.514e-05,
      "loss": 1.1419,
      "step": 607500
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.5136000000000002e-05,
      "loss": 1.1408,
      "step": 608000
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5132000000000002e-05,
      "loss": 1.1376,
      "step": 608500
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5128e-05,
      "loss": 1.1439,
      "step": 609000
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5124000000000001e-05,
      "loss": 1.1368,
      "step": 609500
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5120000000000001e-05,
      "loss": 1.1404,
      "step": 610000
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.5116000000000002e-05,
      "loss": 1.138,
      "step": 610500
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5112000000000002e-05,
      "loss": 1.1406,
      "step": 611000
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.5108e-05,
      "loss": 1.1373,
      "step": 611500
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.5104000000000001e-05,
      "loss": 1.1421,
      "step": 612000
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.5100000000000001e-05,
      "loss": 1.1362,
      "step": 612500
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.5096000000000003e-05,
      "loss": 1.1476,
      "step": 613000
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.5092000000000002e-05,
      "loss": 1.1333,
      "step": 613500
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.5088e-05,
      "loss": 1.1306,
      "step": 614000
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.5084000000000002e-05,
      "loss": 1.1361,
      "step": 614500
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.5080000000000001e-05,
      "loss": 1.1351,
      "step": 615000
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.5076000000000001e-05,
      "loss": 1.1363,
      "step": 615500
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.5072000000000002e-05,
      "loss": 1.1414,
      "step": 616000
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.5068e-05,
      "loss": 1.1449,
      "step": 616500
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.5064e-05,
      "loss": 1.137,
      "step": 617000
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.5060000000000001e-05,
      "loss": 1.1387,
      "step": 617500
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.5056000000000001e-05,
      "loss": 1.1336,
      "step": 618000
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.5052000000000003e-05,
      "loss": 1.1365,
      "step": 618500
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.5048e-05,
      "loss": 1.1387,
      "step": 619000
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.5044e-05,
      "loss": 1.1352,
      "step": 619500
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.5040000000000002e-05,
      "loss": 1.1321,
      "step": 620000
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.5036000000000001e-05,
      "loss": 1.1341,
      "step": 620500
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.5032000000000003e-05,
      "loss": 1.1283,
      "step": 621000
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.5028e-05,
      "loss": 1.135,
      "step": 621500
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.5024e-05,
      "loss": 1.1357,
      "step": 622000
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.5020000000000002e-05,
      "loss": 1.1351,
      "step": 622500
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.5016000000000002e-05,
      "loss": 1.1347,
      "step": 623000
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.5012000000000001e-05,
      "loss": 1.1358,
      "step": 623500
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.5008000000000001e-05,
      "loss": 1.1409,
      "step": 624000
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.5004e-05,
      "loss": 1.138,
      "step": 624500
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.135,
      "step": 625000
    },
    {
      "epoch": 5.0,
      "eval_loss": 1.0943782329559326,
      "eval_runtime": 105.3282,
      "eval_samples_per_second": 127.345,
      "eval_steps_per_second": 1.994,
      "step": 625000
    }
  ],
  "max_steps": 2500000,
  "num_train_epochs": 20,
  "total_flos": 7.883799881847431e+18,
  "trial_name": null,
  "trial_params": null
}