{
  "best_metric": 13.157993518599618,
  "best_model_checkpoint": "./checkpoint-5000",
  "epoch": 2.0911752404851525,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0875,
      "step": 25
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0896,
      "step": 50
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.5e-06,
      "loss": 0.0854,
      "step": 75
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0822,
      "step": 100
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.5e-06,
      "loss": 0.0888,
      "step": 125
    },
    {
      "epoch": 0.06,
      "learning_rate": 3e-06,
      "loss": 0.0932,
      "step": 150
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.48e-06,
      "loss": 0.0985,
      "step": 175
    },
    {
      "epoch": 0.08,
      "learning_rate": 3.980000000000001e-06,
      "loss": 0.0774,
      "step": 200
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.48e-06,
      "loss": 0.0943,
      "step": 225
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.980000000000001e-06,
      "loss": 0.0862,
      "step": 250
    },
    {
      "epoch": 0.12,
      "learning_rate": 5.480000000000001e-06,
      "loss": 0.0713,
      "step": 275
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.98e-06,
      "loss": 0.0788,
      "step": 300
    },
    {
      "epoch": 0.14,
      "learning_rate": 6.480000000000001e-06,
      "loss": 0.0698,
      "step": 325
    },
    {
      "epoch": 0.15,
      "learning_rate": 6.98e-06,
      "loss": 0.0847,
      "step": 350
    },
    {
      "epoch": 0.16,
      "learning_rate": 7.48e-06,
      "loss": 0.087,
      "step": 375
    },
    {
      "epoch": 0.17,
      "learning_rate": 7.980000000000002e-06,
      "loss": 0.0899,
      "step": 400
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.48e-06,
      "loss": 0.0825,
      "step": 425
    },
    {
      "epoch": 0.19,
      "learning_rate": 8.98e-06,
      "loss": 0.0725,
      "step": 450
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.48e-06,
      "loss": 0.0944,
      "step": 475
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.980000000000001e-06,
      "loss": 0.0983,
      "step": 500
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.946666666666667e-06,
      "loss": 0.0789,
      "step": 525
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.891111111111113e-06,
      "loss": 0.0909,
      "step": 550
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.835555555555556e-06,
      "loss": 0.0876,
      "step": 575
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.780000000000001e-06,
      "loss": 0.0826,
      "step": 600
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.724444444444445e-06,
      "loss": 0.0782,
      "step": 625
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.66888888888889e-06,
      "loss": 0.0826,
      "step": 650
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.613333333333335e-06,
      "loss": 0.0794,
      "step": 675
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.557777777777777e-06,
      "loss": 0.0808,
      "step": 700
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.502222222222223e-06,
      "loss": 0.0792,
      "step": 725
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.446666666666667e-06,
      "loss": 0.0775,
      "step": 750
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.391111111111111e-06,
      "loss": 0.1031,
      "step": 775
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.335555555555557e-06,
      "loss": 0.0897,
      "step": 800
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.280000000000001e-06,
      "loss": 0.0908,
      "step": 825
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.224444444444445e-06,
      "loss": 0.0853,
      "step": 850
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.168888888888889e-06,
      "loss": 0.0965,
      "step": 875
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.113333333333335e-06,
      "loss": 0.087,
      "step": 900
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.057777777777779e-06,
      "loss": 0.0797,
      "step": 925
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.002222222222223e-06,
      "loss": 0.091,
      "step": 950
    },
    {
      "epoch": 0.41,
      "learning_rate": 8.946666666666669e-06,
      "loss": 0.0891,
      "step": 975
    },
    {
      "epoch": 0.42,
      "learning_rate": 8.891111111111111e-06,
      "loss": 0.103,
      "step": 1000
    },
    {
      "epoch": 0.43,
      "learning_rate": 8.835555555555557e-06,
      "loss": 0.1141,
      "step": 1025
    },
    {
      "epoch": 0.44,
      "learning_rate": 8.78e-06,
      "loss": 0.0979,
      "step": 1050
    },
    {
      "epoch": 0.45,
      "learning_rate": 8.724444444444445e-06,
      "loss": 0.0801,
      "step": 1075
    },
    {
      "epoch": 0.46,
      "learning_rate": 8.66888888888889e-06,
      "loss": 0.0868,
      "step": 1100
    },
    {
      "epoch": 0.47,
      "learning_rate": 8.613333333333333e-06,
      "loss": 0.094,
      "step": 1125
    },
    {
      "epoch": 0.48,
      "learning_rate": 8.557777777777778e-06,
      "loss": 0.092,
      "step": 1150
    },
    {
      "epoch": 0.49,
      "learning_rate": 8.502222222222223e-06,
      "loss": 0.0843,
      "step": 1175
    },
    {
      "epoch": 0.5,
      "learning_rate": 8.446666666666668e-06,
      "loss": 0.0784,
      "step": 1200
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.391111111111112e-06,
      "loss": 0.081,
      "step": 1225
    },
    {
      "epoch": 0.52,
      "learning_rate": 8.335555555555556e-06,
      "loss": 0.0772,
      "step": 1250
    },
    {
      "epoch": 0.53,
      "learning_rate": 8.28e-06,
      "loss": 0.0914,
      "step": 1275
    },
    {
      "epoch": 0.54,
      "learning_rate": 8.224444444444444e-06,
      "loss": 0.0981,
      "step": 1300
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.16888888888889e-06,
      "loss": 0.0859,
      "step": 1325
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.113333333333334e-06,
      "loss": 0.1,
      "step": 1350
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.057777777777778e-06,
      "loss": 0.0998,
      "step": 1375
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.002222222222222e-06,
      "loss": 0.086,
      "step": 1400
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.946666666666666e-06,
      "loss": 0.0879,
      "step": 1425
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.891111111111112e-06,
      "loss": 0.0754,
      "step": 1450
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.835555555555556e-06,
      "loss": 0.092,
      "step": 1475
    },
    {
      "epoch": 0.63,
      "learning_rate": 7.78e-06,
      "loss": 0.0921,
      "step": 1500
    },
    {
      "epoch": 0.64,
      "learning_rate": 7.724444444444446e-06,
      "loss": 0.0995,
      "step": 1525
    },
    {
      "epoch": 0.65,
      "learning_rate": 7.66888888888889e-06,
      "loss": 0.0746,
      "step": 1550
    },
    {
      "epoch": 0.66,
      "learning_rate": 7.613333333333334e-06,
      "loss": 0.0779,
      "step": 1575
    },
    {
      "epoch": 0.67,
      "learning_rate": 7.557777777777779e-06,
      "loss": 0.0981,
      "step": 1600
    },
    {
      "epoch": 0.68,
      "learning_rate": 7.502222222222223e-06,
      "loss": 0.0775,
      "step": 1625
    },
    {
      "epoch": 0.69,
      "learning_rate": 7.446666666666668e-06,
      "loss": 0.096,
      "step": 1650
    },
    {
      "epoch": 0.7,
      "learning_rate": 7.3911111111111125e-06,
      "loss": 0.0901,
      "step": 1675
    },
    {
      "epoch": 0.71,
      "learning_rate": 7.335555555555556e-06,
      "loss": 0.0879,
      "step": 1700
    },
    {
      "epoch": 0.72,
      "learning_rate": 7.280000000000001e-06,
      "loss": 0.0805,
      "step": 1725
    },
    {
      "epoch": 0.73,
      "learning_rate": 7.224444444444445e-06,
      "loss": 0.0935,
      "step": 1750
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.1688888888888895e-06,
      "loss": 0.091,
      "step": 1775
    },
    {
      "epoch": 0.75,
      "learning_rate": 7.113333333333334e-06,
      "loss": 0.1043,
      "step": 1800
    },
    {
      "epoch": 0.76,
      "learning_rate": 7.057777777777778e-06,
      "loss": 0.0865,
      "step": 1825
    },
    {
      "epoch": 0.77,
      "learning_rate": 7.0022222222222225e-06,
      "loss": 0.1032,
      "step": 1850
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.946666666666667e-06,
      "loss": 0.0983,
      "step": 1875
    },
    {
      "epoch": 0.79,
      "learning_rate": 6.891111111111111e-06,
      "loss": 0.0874,
      "step": 1900
    },
    {
      "epoch": 0.81,
      "learning_rate": 6.835555555555556e-06,
      "loss": 0.0937,
      "step": 1925
    },
    {
      "epoch": 0.82,
      "learning_rate": 6.780000000000001e-06,
      "loss": 0.0836,
      "step": 1950
    },
    {
      "epoch": 0.83,
      "learning_rate": 6.724444444444444e-06,
      "loss": 0.0708,
      "step": 1975
    },
    {
      "epoch": 0.84,
      "learning_rate": 6.668888888888889e-06,
      "loss": 0.0839,
      "step": 2000
    },
    {
      "epoch": 0.85,
      "learning_rate": 6.613333333333334e-06,
      "loss": 0.092,
      "step": 2025
    },
    {
      "epoch": 0.86,
      "learning_rate": 6.557777777777778e-06,
      "loss": 0.0798,
      "step": 2050
    },
    {
      "epoch": 0.87,
      "learning_rate": 6.502222222222223e-06,
      "loss": 0.0634,
      "step": 2075
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.446666666666668e-06,
      "loss": 0.1018,
      "step": 2100
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.391111111111111e-06,
      "loss": 0.0864,
      "step": 2125
    },
    {
      "epoch": 0.9,
      "learning_rate": 6.335555555555556e-06,
      "loss": 0.0818,
      "step": 2150
    },
    {
      "epoch": 0.91,
      "learning_rate": 6.280000000000001e-06,
      "loss": 0.0789,
      "step": 2175
    },
    {
      "epoch": 0.92,
      "learning_rate": 6.224444444444445e-06,
      "loss": 0.089,
      "step": 2200
    },
    {
      "epoch": 0.93,
      "learning_rate": 6.16888888888889e-06,
      "loss": 0.0968,
      "step": 2225
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.1155555555555555e-06,
      "loss": 0.0935,
      "step": 2250
    },
    {
      "epoch": 0.95,
      "learning_rate": 6.0600000000000004e-06,
      "loss": 0.0703,
      "step": 2275
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.004444444444445e-06,
      "loss": 0.1025,
      "step": 2300
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.948888888888889e-06,
      "loss": 0.0912,
      "step": 2325
    },
    {
      "epoch": 0.98,
      "learning_rate": 5.893333333333334e-06,
      "loss": 0.0849,
      "step": 2350
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.837777777777777e-06,
      "loss": 0.0972,
      "step": 2375
    },
    {
      "epoch": 1.0,
      "learning_rate": 5.782222222222222e-06,
      "loss": 0.0777,
      "step": 2400
    },
    {
      "epoch": 1.01,
      "learning_rate": 5.726666666666667e-06,
      "loss": 0.0338,
      "step": 2425
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.671111111111112e-06,
      "loss": 0.0412,
      "step": 2450
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.615555555555556e-06,
      "loss": 0.0341,
      "step": 2475
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.560000000000001e-06,
      "loss": 0.0344,
      "step": 2500
    },
    {
      "epoch": 1.06,
      "learning_rate": 5.504444444444444e-06,
      "loss": 0.0435,
      "step": 2525
    },
    {
      "epoch": 1.07,
      "learning_rate": 5.448888888888889e-06,
      "loss": 0.0425,
      "step": 2550
    },
    {
      "epoch": 1.08,
      "learning_rate": 5.393333333333334e-06,
      "loss": 0.0281,
      "step": 2575
    },
    {
      "epoch": 1.09,
      "learning_rate": 5.337777777777779e-06,
      "loss": 0.0372,
      "step": 2600
    },
    {
      "epoch": 1.1,
      "learning_rate": 5.282222222222223e-06,
      "loss": 0.0435,
      "step": 2625
    },
    {
      "epoch": 1.11,
      "learning_rate": 5.226666666666667e-06,
      "loss": 0.043,
      "step": 2650
    },
    {
      "epoch": 1.12,
      "learning_rate": 5.171111111111111e-06,
      "loss": 0.0374,
      "step": 2675
    },
    {
      "epoch": 1.13,
      "learning_rate": 5.115555555555556e-06,
      "loss": 0.0432,
      "step": 2700
    },
    {
      "epoch": 1.14,
      "learning_rate": 5.060000000000001e-06,
      "loss": 0.0483,
      "step": 2725
    },
    {
      "epoch": 1.15,
      "learning_rate": 5.004444444444445e-06,
      "loss": 0.0413,
      "step": 2750
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.94888888888889e-06,
      "loss": 0.0357,
      "step": 2775
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.893333333333334e-06,
      "loss": 0.0312,
      "step": 2800
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.837777777777778e-06,
      "loss": 0.0371,
      "step": 2825
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.7822222222222226e-06,
      "loss": 0.0466,
      "step": 2850
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.7266666666666674e-06,
      "loss": 0.0473,
      "step": 2875
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.6711111111111115e-06,
      "loss": 0.0392,
      "step": 2900
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.6155555555555555e-06,
      "loss": 0.0395,
      "step": 2925
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.56e-06,
      "loss": 0.0307,
      "step": 2950
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.504444444444444e-06,
      "loss": 0.0474,
      "step": 2975
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.448888888888889e-06,
      "loss": 0.0405,
      "step": 3000
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.393333333333334e-06,
      "loss": 0.0343,
      "step": 3025
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.337777777777778e-06,
      "loss": 0.0406,
      "step": 3050
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.282222222222222e-06,
      "loss": 0.0454,
      "step": 3075
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.226666666666667e-06,
      "loss": 0.0478,
      "step": 3100
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.171111111111111e-06,
      "loss": 0.0364,
      "step": 3125
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.115555555555556e-06,
      "loss": 0.0385,
      "step": 3150
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.060000000000001e-06,
      "loss": 0.0448,
      "step": 3175
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.004444444444445e-06,
      "loss": 0.0395,
      "step": 3200
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.948888888888889e-06,
      "loss": 0.0439,
      "step": 3225
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.893333333333333e-06,
      "loss": 0.0438,
      "step": 3250
    },
    {
      "epoch": 1.37,
      "learning_rate": 3.837777777777778e-06,
      "loss": 0.048,
      "step": 3275
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.782222222222223e-06,
      "loss": 0.0403,
      "step": 3300
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.726666666666667e-06,
      "loss": 0.0347,
      "step": 3325
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.6711111111111113e-06,
      "loss": 0.0482,
      "step": 3350
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.615555555555556e-06,
      "loss": 0.043,
      "step": 3375
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.5600000000000002e-06,
      "loss": 0.0412,
      "step": 3400
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.5044444444444447e-06,
      "loss": 0.036,
      "step": 3425
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.4488888888888896e-06,
      "loss": 0.0444,
      "step": 3450
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.3933333333333336e-06,
      "loss": 0.0388,
      "step": 3475
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.337777777777778e-06,
      "loss": 0.0377,
      "step": 3500
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.282222222222223e-06,
      "loss": 0.0388,
      "step": 3525
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.226666666666667e-06,
      "loss": 0.0408,
      "step": 3550
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.1711111111111114e-06,
      "loss": 0.0365,
      "step": 3575
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.1155555555555555e-06,
      "loss": 0.0482,
      "step": 3600
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.0600000000000003e-06,
      "loss": 0.0433,
      "step": 3625
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.004444444444445e-06,
      "loss": 0.0373,
      "step": 3650
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.948888888888889e-06,
      "loss": 0.0412,
      "step": 3675
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.8933333333333337e-06,
      "loss": 0.0386,
      "step": 3700
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.837777777777778e-06,
      "loss": 0.0411,
      "step": 3725
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.7822222222222222e-06,
      "loss": 0.0496,
      "step": 3750
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.726666666666667e-06,
      "loss": 0.0418,
      "step": 3775
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.6711111111111116e-06,
      "loss": 0.0434,
      "step": 3800
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.6155555555555556e-06,
      "loss": 0.0358,
      "step": 3825
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.56e-06,
      "loss": 0.0487,
      "step": 3850
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.504444444444445e-06,
      "loss": 0.0377,
      "step": 3875
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.448888888888889e-06,
      "loss": 0.0437,
      "step": 3900
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.3933333333333334e-06,
      "loss": 0.0455,
      "step": 3925
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.337777777777778e-06,
      "loss": 0.0372,
      "step": 3950
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2822222222222223e-06,
      "loss": 0.033,
      "step": 3975
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.226666666666667e-06,
      "loss": 0.0468,
      "step": 4000
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.1711111111111113e-06,
      "loss": 0.0425,
      "step": 4025
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1155555555555557e-06,
      "loss": 0.0372,
      "step": 4050
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.06e-06,
      "loss": 0.0342,
      "step": 4075
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.0044444444444446e-06,
      "loss": 0.044,
      "step": 4100
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.948888888888889e-06,
      "loss": 0.042,
      "step": 4125
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.8933333333333333e-06,
      "loss": 0.0442,
      "step": 4150
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.837777777777778e-06,
      "loss": 0.0415,
      "step": 4175
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.7822222222222225e-06,
      "loss": 0.0458,
      "step": 4200
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.7266666666666667e-06,
      "loss": 0.0503,
      "step": 4225
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.6711111111111112e-06,
      "loss": 0.0527,
      "step": 4250
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.6155555555555559e-06,
      "loss": 0.042,
      "step": 4275
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.56e-06,
      "loss": 0.0497,
      "step": 4300
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.5044444444444446e-06,
      "loss": 0.0501,
      "step": 4325
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.4488888888888892e-06,
      "loss": 0.0451,
      "step": 4350
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.3933333333333335e-06,
      "loss": 0.0397,
      "step": 4375
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.337777777777778e-06,
      "loss": 0.0547,
      "step": 4400
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.2822222222222222e-06,
      "loss": 0.0376,
      "step": 4425
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.228888888888889e-06,
      "loss": 0.0468,
      "step": 4450
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.1733333333333335e-06,
      "loss": 0.0429,
      "step": 4475
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.117777777777778e-06,
      "loss": 0.0478,
      "step": 4500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.0622222222222222e-06,
      "loss": 0.0429,
      "step": 4525
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.0066666666666668e-06,
      "loss": 0.0477,
      "step": 4550
    },
    {
      "epoch": 1.91,
      "learning_rate": 9.511111111111111e-07,
      "loss": 0.0443,
      "step": 4575
    },
    {
      "epoch": 1.92,
      "learning_rate": 8.955555555555557e-07,
      "loss": 0.0484,
      "step": 4600
    },
    {
      "epoch": 1.93,
      "learning_rate": 8.400000000000001e-07,
      "loss": 0.0494,
      "step": 4625
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.844444444444445e-07,
      "loss": 0.0365,
      "step": 4650
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.28888888888889e-07,
      "loss": 0.0517,
      "step": 4675
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.733333333333334e-07,
      "loss": 0.0415,
      "step": 4700
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.177777777777778e-07,
      "loss": 0.0442,
      "step": 4725
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.622222222222223e-07,
      "loss": 0.0516,
      "step": 4750
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.066666666666667e-07,
      "loss": 0.048,
      "step": 4775
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.511111111111111e-07,
      "loss": 0.0256,
      "step": 4800
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.9555555555555557e-07,
      "loss": 0.0213,
      "step": 4825
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.4000000000000003e-07,
      "loss": 0.017,
      "step": 4850
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.844444444444445e-07,
      "loss": 0.0132,
      "step": 4875
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.2888888888888892e-07,
      "loss": 0.0219,
      "step": 4900
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.7333333333333335e-07,
      "loss": 0.0174,
      "step": 4925
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.1777777777777778e-07,
      "loss": 0.0177,
      "step": 4950
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.222222222222223e-08,
      "loss": 0.02,
      "step": 4975
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.666666666666667e-09,
      "loss": 0.0167,
      "step": 5000
    },
    {
      "epoch": 2.09,
      "eval_loss": 0.16790781915187836,
      "eval_runtime": 10745.9827,
      "eval_samples_per_second": 1.017,
      "eval_steps_per_second": 0.032,
      "eval_wer": 13.157993518599618,
      "step": 5000
    },
    {
      "epoch": 2.09,
      "step": 5000,
      "total_flos": 1.6325800221671424e+20,
      "train_loss": 0.06287762756943703,
      "train_runtime": 28287.0338,
      "train_samples_per_second": 5.656,
      "train_steps_per_second": 0.177
    }
  ],
  "max_steps": 5000,
  "num_train_epochs": 3,
  "total_flos": 1.6325800221671424e+20,
  "trial_name": null,
  "trial_params": null
}