cv9-special-batch8-small-concat2 / trainer_state.json
TheRains's picture
End of training
a790775
raw
history blame
No virus
25.8 kB
{
"best_metric": 12.900851161720727,
"best_model_checkpoint": "./output/small/cv9-special-batch8-small-concat2/checkpoint-5000",
"epoch": 1.0577533319229955,
"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": 4.4e-07,
"loss": 1.765,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.400000000000001e-07,
"loss": 1.5922,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 1.44e-06,
"loss": 1.4918,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.94e-06,
"loss": 1.2202,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.9272,
"step": 125
},
{
"epoch": 0.03,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.7901,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 3.44e-06,
"loss": 0.7731,
"step": 175
},
{
"epoch": 0.04,
"learning_rate": 3.94e-06,
"loss": 0.6941,
"step": 200
},
{
"epoch": 0.05,
"learning_rate": 4.440000000000001e-06,
"loss": 0.6778,
"step": 225
},
{
"epoch": 0.05,
"learning_rate": 4.94e-06,
"loss": 0.5758,
"step": 250
},
{
"epoch": 0.06,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.6661,
"step": 275
},
{
"epoch": 0.06,
"learning_rate": 5.94e-06,
"loss": 0.6216,
"step": 300
},
{
"epoch": 0.07,
"learning_rate": 6.440000000000001e-06,
"loss": 0.7256,
"step": 325
},
{
"epoch": 0.07,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.7756,
"step": 350
},
{
"epoch": 0.08,
"learning_rate": 7.440000000000001e-06,
"loss": 0.6896,
"step": 375
},
{
"epoch": 0.08,
"learning_rate": 7.94e-06,
"loss": 0.7236,
"step": 400
},
{
"epoch": 0.09,
"learning_rate": 8.44e-06,
"loss": 0.7168,
"step": 425
},
{
"epoch": 0.1,
"learning_rate": 8.94e-06,
"loss": 0.5927,
"step": 450
},
{
"epoch": 0.1,
"learning_rate": 9.440000000000001e-06,
"loss": 0.6174,
"step": 475
},
{
"epoch": 0.11,
"learning_rate": 9.940000000000001e-06,
"loss": 0.6707,
"step": 500
},
{
"epoch": 0.11,
"learning_rate": 9.951111111111111e-06,
"loss": 0.5178,
"step": 525
},
{
"epoch": 0.12,
"learning_rate": 9.895555555555557e-06,
"loss": 0.6022,
"step": 550
},
{
"epoch": 0.12,
"learning_rate": 9.84e-06,
"loss": 0.7153,
"step": 575
},
{
"epoch": 0.13,
"learning_rate": 9.784444444444445e-06,
"loss": 0.6291,
"step": 600
},
{
"epoch": 0.13,
"learning_rate": 9.72888888888889e-06,
"loss": 0.6074,
"step": 625
},
{
"epoch": 0.14,
"learning_rate": 9.673333333333334e-06,
"loss": 0.5406,
"step": 650
},
{
"epoch": 0.14,
"learning_rate": 9.617777777777778e-06,
"loss": 0.6762,
"step": 675
},
{
"epoch": 0.15,
"learning_rate": 9.562222222222223e-06,
"loss": 0.605,
"step": 700
},
{
"epoch": 0.15,
"learning_rate": 9.506666666666667e-06,
"loss": 0.537,
"step": 725
},
{
"epoch": 0.16,
"learning_rate": 9.451111111111112e-06,
"loss": 0.6735,
"step": 750
},
{
"epoch": 0.16,
"learning_rate": 9.395555555555556e-06,
"loss": 0.5854,
"step": 775
},
{
"epoch": 0.17,
"learning_rate": 9.340000000000002e-06,
"loss": 0.6176,
"step": 800
},
{
"epoch": 0.17,
"learning_rate": 9.284444444444444e-06,
"loss": 0.6349,
"step": 825
},
{
"epoch": 0.18,
"learning_rate": 9.22888888888889e-06,
"loss": 0.5164,
"step": 850
},
{
"epoch": 0.19,
"learning_rate": 9.173333333333334e-06,
"loss": 0.5469,
"step": 875
},
{
"epoch": 0.19,
"learning_rate": 9.117777777777778e-06,
"loss": 0.5752,
"step": 900
},
{
"epoch": 0.2,
"learning_rate": 9.062222222222224e-06,
"loss": 0.6189,
"step": 925
},
{
"epoch": 0.2,
"learning_rate": 9.006666666666666e-06,
"loss": 0.5858,
"step": 950
},
{
"epoch": 0.21,
"learning_rate": 8.951111111111112e-06,
"loss": 0.5193,
"step": 975
},
{
"epoch": 0.21,
"learning_rate": 8.895555555555556e-06,
"loss": 0.6518,
"step": 1000
},
{
"epoch": 0.21,
"eval_loss": 0.30868610739707947,
"eval_runtime": 603.5293,
"eval_samples_per_second": 6.001,
"eval_steps_per_second": 1.501,
"eval_wer": 18.951000690131124,
"step": 1000
},
{
"epoch": 0.22,
"learning_rate": 8.84e-06,
"loss": 0.5307,
"step": 1025
},
{
"epoch": 0.22,
"learning_rate": 8.784444444444446e-06,
"loss": 0.6095,
"step": 1050
},
{
"epoch": 0.23,
"learning_rate": 8.72888888888889e-06,
"loss": 0.6181,
"step": 1075
},
{
"epoch": 0.23,
"learning_rate": 8.673333333333334e-06,
"loss": 0.5303,
"step": 1100
},
{
"epoch": 0.24,
"learning_rate": 8.617777777777778e-06,
"loss": 0.5081,
"step": 1125
},
{
"epoch": 0.24,
"learning_rate": 8.562222222222224e-06,
"loss": 0.5449,
"step": 1150
},
{
"epoch": 0.25,
"learning_rate": 8.506666666666668e-06,
"loss": 0.5585,
"step": 1175
},
{
"epoch": 0.25,
"learning_rate": 8.451111111111112e-06,
"loss": 0.6223,
"step": 1200
},
{
"epoch": 0.26,
"learning_rate": 8.395555555555557e-06,
"loss": 0.5928,
"step": 1225
},
{
"epoch": 0.26,
"learning_rate": 8.34e-06,
"loss": 0.598,
"step": 1250
},
{
"epoch": 0.27,
"learning_rate": 8.284444444444446e-06,
"loss": 0.4904,
"step": 1275
},
{
"epoch": 0.28,
"learning_rate": 8.22888888888889e-06,
"loss": 0.55,
"step": 1300
},
{
"epoch": 0.28,
"learning_rate": 8.173333333333334e-06,
"loss": 0.5779,
"step": 1325
},
{
"epoch": 0.29,
"learning_rate": 8.11777777777778e-06,
"loss": 0.5066,
"step": 1350
},
{
"epoch": 0.29,
"learning_rate": 8.062222222222222e-06,
"loss": 0.567,
"step": 1375
},
{
"epoch": 0.3,
"learning_rate": 8.006666666666667e-06,
"loss": 0.5326,
"step": 1400
},
{
"epoch": 0.3,
"learning_rate": 7.951111111111111e-06,
"loss": 0.5796,
"step": 1425
},
{
"epoch": 0.31,
"learning_rate": 7.895555555555557e-06,
"loss": 0.4663,
"step": 1450
},
{
"epoch": 0.31,
"learning_rate": 7.840000000000001e-06,
"loss": 0.5558,
"step": 1475
},
{
"epoch": 0.32,
"learning_rate": 7.784444444444445e-06,
"loss": 0.5576,
"step": 1500
},
{
"epoch": 0.32,
"learning_rate": 7.72888888888889e-06,
"loss": 0.5029,
"step": 1525
},
{
"epoch": 0.33,
"learning_rate": 7.673333333333333e-06,
"loss": 0.5007,
"step": 1550
},
{
"epoch": 0.33,
"learning_rate": 7.617777777777778e-06,
"loss": 0.5262,
"step": 1575
},
{
"epoch": 0.34,
"learning_rate": 7.562222222222223e-06,
"loss": 0.4636,
"step": 1600
},
{
"epoch": 0.34,
"learning_rate": 7.506666666666668e-06,
"loss": 0.5441,
"step": 1625
},
{
"epoch": 0.35,
"learning_rate": 7.451111111111111e-06,
"loss": 0.4662,
"step": 1650
},
{
"epoch": 0.35,
"learning_rate": 7.395555555555556e-06,
"loss": 0.5125,
"step": 1675
},
{
"epoch": 0.36,
"learning_rate": 7.340000000000001e-06,
"loss": 0.5338,
"step": 1700
},
{
"epoch": 0.36,
"learning_rate": 7.284444444444445e-06,
"loss": 0.5841,
"step": 1725
},
{
"epoch": 0.37,
"learning_rate": 7.22888888888889e-06,
"loss": 0.5753,
"step": 1750
},
{
"epoch": 0.38,
"learning_rate": 7.173333333333335e-06,
"loss": 0.5169,
"step": 1775
},
{
"epoch": 0.38,
"learning_rate": 7.117777777777778e-06,
"loss": 0.5042,
"step": 1800
},
{
"epoch": 0.39,
"learning_rate": 7.062222222222223e-06,
"loss": 0.5476,
"step": 1825
},
{
"epoch": 0.39,
"learning_rate": 7.006666666666667e-06,
"loss": 0.5712,
"step": 1850
},
{
"epoch": 0.4,
"learning_rate": 6.951111111111112e-06,
"loss": 0.4997,
"step": 1875
},
{
"epoch": 0.4,
"learning_rate": 6.8955555555555565e-06,
"loss": 0.5533,
"step": 1900
},
{
"epoch": 0.41,
"learning_rate": 6.8400000000000014e-06,
"loss": 0.6333,
"step": 1925
},
{
"epoch": 0.41,
"learning_rate": 6.784444444444445e-06,
"loss": 0.4897,
"step": 1950
},
{
"epoch": 0.42,
"learning_rate": 6.7288888888888895e-06,
"loss": 0.5785,
"step": 1975
},
{
"epoch": 0.42,
"learning_rate": 6.6733333333333335e-06,
"loss": 0.542,
"step": 2000
},
{
"epoch": 0.42,
"eval_loss": 0.27949848771095276,
"eval_runtime": 599.3935,
"eval_samples_per_second": 6.043,
"eval_steps_per_second": 1.512,
"eval_wer": 16.696572348746262,
"step": 2000
},
{
"epoch": 0.43,
"learning_rate": 6.617777777777778e-06,
"loss": 0.5706,
"step": 2025
},
{
"epoch": 0.43,
"learning_rate": 6.562222222222223e-06,
"loss": 0.5552,
"step": 2050
},
{
"epoch": 0.44,
"learning_rate": 6.5066666666666665e-06,
"loss": 0.4484,
"step": 2075
},
{
"epoch": 0.44,
"learning_rate": 6.451111111111111e-06,
"loss": 0.473,
"step": 2100
},
{
"epoch": 0.45,
"learning_rate": 6.395555555555556e-06,
"loss": 0.5224,
"step": 2125
},
{
"epoch": 0.45,
"learning_rate": 6.34e-06,
"loss": 0.5241,
"step": 2150
},
{
"epoch": 0.46,
"learning_rate": 6.284444444444445e-06,
"loss": 0.4629,
"step": 2175
},
{
"epoch": 0.47,
"learning_rate": 6.22888888888889e-06,
"loss": 0.5274,
"step": 2200
},
{
"epoch": 0.47,
"learning_rate": 6.173333333333333e-06,
"loss": 0.4933,
"step": 2225
},
{
"epoch": 0.48,
"learning_rate": 6.117777777777778e-06,
"loss": 0.5455,
"step": 2250
},
{
"epoch": 0.48,
"learning_rate": 6.062222222222223e-06,
"loss": 0.5981,
"step": 2275
},
{
"epoch": 0.49,
"learning_rate": 6.006666666666667e-06,
"loss": 0.5124,
"step": 2300
},
{
"epoch": 0.49,
"learning_rate": 5.951111111111112e-06,
"loss": 0.5057,
"step": 2325
},
{
"epoch": 0.5,
"learning_rate": 5.895555555555557e-06,
"loss": 0.5179,
"step": 2350
},
{
"epoch": 0.5,
"learning_rate": 5.84e-06,
"loss": 0.5481,
"step": 2375
},
{
"epoch": 0.51,
"learning_rate": 5.784444444444445e-06,
"loss": 0.5023,
"step": 2400
},
{
"epoch": 0.51,
"learning_rate": 5.72888888888889e-06,
"loss": 0.5238,
"step": 2425
},
{
"epoch": 0.52,
"learning_rate": 5.673333333333334e-06,
"loss": 0.4803,
"step": 2450
},
{
"epoch": 0.52,
"learning_rate": 5.617777777777779e-06,
"loss": 0.4933,
"step": 2475
},
{
"epoch": 0.53,
"learning_rate": 5.562222222222222e-06,
"loss": 0.404,
"step": 2500
},
{
"epoch": 0.53,
"learning_rate": 5.506666666666667e-06,
"loss": 0.5172,
"step": 2525
},
{
"epoch": 0.54,
"learning_rate": 5.451111111111112e-06,
"loss": 0.5105,
"step": 2550
},
{
"epoch": 0.54,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.4696,
"step": 2575
},
{
"epoch": 0.55,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.5238,
"step": 2600
},
{
"epoch": 0.56,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.5245,
"step": 2625
},
{
"epoch": 0.56,
"learning_rate": 5.228888888888889e-06,
"loss": 0.5032,
"step": 2650
},
{
"epoch": 0.57,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.5344,
"step": 2675
},
{
"epoch": 0.57,
"learning_rate": 5.117777777777778e-06,
"loss": 0.4965,
"step": 2700
},
{
"epoch": 0.58,
"learning_rate": 5.062222222222222e-06,
"loss": 0.5335,
"step": 2725
},
{
"epoch": 0.58,
"learning_rate": 5.006666666666667e-06,
"loss": 0.4586,
"step": 2750
},
{
"epoch": 0.59,
"learning_rate": 4.951111111111111e-06,
"loss": 0.4969,
"step": 2775
},
{
"epoch": 0.59,
"learning_rate": 4.895555555555556e-06,
"loss": 0.4828,
"step": 2800
},
{
"epoch": 0.6,
"learning_rate": 4.84e-06,
"loss": 0.482,
"step": 2825
},
{
"epoch": 0.6,
"learning_rate": 4.784444444444445e-06,
"loss": 0.5194,
"step": 2850
},
{
"epoch": 0.61,
"learning_rate": 4.728888888888889e-06,
"loss": 0.4705,
"step": 2875
},
{
"epoch": 0.61,
"learning_rate": 4.673333333333333e-06,
"loss": 0.4917,
"step": 2900
},
{
"epoch": 0.62,
"learning_rate": 4.617777777777778e-06,
"loss": 0.5806,
"step": 2925
},
{
"epoch": 0.62,
"learning_rate": 4.564444444444445e-06,
"loss": 0.4658,
"step": 2950
},
{
"epoch": 0.63,
"learning_rate": 4.50888888888889e-06,
"loss": 0.5393,
"step": 2975
},
{
"epoch": 0.63,
"learning_rate": 4.453333333333334e-06,
"loss": 0.4933,
"step": 3000
},
{
"epoch": 0.63,
"eval_loss": 0.2543008327484131,
"eval_runtime": 601.9725,
"eval_samples_per_second": 6.017,
"eval_steps_per_second": 1.505,
"eval_wer": 14.304117782378652,
"step": 3000
},
{
"epoch": 0.64,
"learning_rate": 4.397777777777778e-06,
"loss": 0.4951,
"step": 3025
},
{
"epoch": 0.65,
"learning_rate": 4.3422222222222225e-06,
"loss": 0.4836,
"step": 3050
},
{
"epoch": 0.65,
"learning_rate": 4.2866666666666666e-06,
"loss": 0.4067,
"step": 3075
},
{
"epoch": 0.66,
"learning_rate": 4.2311111111111114e-06,
"loss": 0.5096,
"step": 3100
},
{
"epoch": 0.66,
"learning_rate": 4.175555555555556e-06,
"loss": 0.4604,
"step": 3125
},
{
"epoch": 0.67,
"learning_rate": 4.12e-06,
"loss": 0.4998,
"step": 3150
},
{
"epoch": 0.67,
"learning_rate": 4.064444444444444e-06,
"loss": 0.5341,
"step": 3175
},
{
"epoch": 0.68,
"learning_rate": 4.008888888888889e-06,
"loss": 0.4687,
"step": 3200
},
{
"epoch": 0.68,
"learning_rate": 3.953333333333333e-06,
"loss": 0.5172,
"step": 3225
},
{
"epoch": 0.69,
"learning_rate": 3.897777777777778e-06,
"loss": 0.4309,
"step": 3250
},
{
"epoch": 0.69,
"learning_rate": 3.842222222222223e-06,
"loss": 0.4654,
"step": 3275
},
{
"epoch": 0.7,
"learning_rate": 3.7866666666666667e-06,
"loss": 0.4889,
"step": 3300
},
{
"epoch": 0.7,
"learning_rate": 3.7311111111111116e-06,
"loss": 0.4463,
"step": 3325
},
{
"epoch": 0.71,
"learning_rate": 3.675555555555556e-06,
"loss": 0.4392,
"step": 3350
},
{
"epoch": 0.71,
"learning_rate": 3.62e-06,
"loss": 0.4444,
"step": 3375
},
{
"epoch": 0.72,
"learning_rate": 3.564444444444445e-06,
"loss": 0.5219,
"step": 3400
},
{
"epoch": 0.72,
"learning_rate": 3.508888888888889e-06,
"loss": 0.4283,
"step": 3425
},
{
"epoch": 0.73,
"learning_rate": 3.4533333333333334e-06,
"loss": 0.481,
"step": 3450
},
{
"epoch": 0.74,
"learning_rate": 3.3977777777777783e-06,
"loss": 0.4612,
"step": 3475
},
{
"epoch": 0.74,
"learning_rate": 3.3422222222222224e-06,
"loss": 0.4557,
"step": 3500
},
{
"epoch": 0.75,
"learning_rate": 3.286666666666667e-06,
"loss": 0.4588,
"step": 3525
},
{
"epoch": 0.75,
"learning_rate": 3.2311111111111117e-06,
"loss": 0.4832,
"step": 3550
},
{
"epoch": 0.76,
"learning_rate": 3.1755555555555557e-06,
"loss": 0.5374,
"step": 3575
},
{
"epoch": 0.76,
"learning_rate": 3.12e-06,
"loss": 0.4595,
"step": 3600
},
{
"epoch": 0.77,
"learning_rate": 3.064444444444445e-06,
"loss": 0.4913,
"step": 3625
},
{
"epoch": 0.77,
"learning_rate": 3.008888888888889e-06,
"loss": 0.4691,
"step": 3650
},
{
"epoch": 0.78,
"learning_rate": 2.9533333333333336e-06,
"loss": 0.5271,
"step": 3675
},
{
"epoch": 0.78,
"learning_rate": 2.8977777777777785e-06,
"loss": 0.4035,
"step": 3700
},
{
"epoch": 0.79,
"learning_rate": 2.8422222222222225e-06,
"loss": 0.4724,
"step": 3725
},
{
"epoch": 0.79,
"learning_rate": 2.786666666666667e-06,
"loss": 0.4687,
"step": 3750
},
{
"epoch": 0.8,
"learning_rate": 2.7311111111111114e-06,
"loss": 0.4633,
"step": 3775
},
{
"epoch": 0.8,
"learning_rate": 2.675555555555556e-06,
"loss": 0.403,
"step": 3800
},
{
"epoch": 0.81,
"learning_rate": 2.6200000000000003e-06,
"loss": 0.4574,
"step": 3825
},
{
"epoch": 0.81,
"learning_rate": 2.5644444444444444e-06,
"loss": 0.4457,
"step": 3850
},
{
"epoch": 0.82,
"learning_rate": 2.5088888888888892e-06,
"loss": 0.4732,
"step": 3875
},
{
"epoch": 0.83,
"learning_rate": 2.4533333333333333e-06,
"loss": 0.4697,
"step": 3900
},
{
"epoch": 0.83,
"learning_rate": 2.397777777777778e-06,
"loss": 0.4558,
"step": 3925
},
{
"epoch": 0.84,
"learning_rate": 2.342222222222222e-06,
"loss": 0.5054,
"step": 3950
},
{
"epoch": 0.84,
"learning_rate": 2.2866666666666667e-06,
"loss": 0.4104,
"step": 3975
},
{
"epoch": 0.85,
"learning_rate": 2.2311111111111115e-06,
"loss": 0.4943,
"step": 4000
},
{
"epoch": 0.85,
"eval_loss": 0.24349194765090942,
"eval_runtime": 591.1948,
"eval_samples_per_second": 6.127,
"eval_steps_per_second": 1.532,
"eval_wer": 13.50356567747872,
"step": 4000
},
{
"epoch": 0.85,
"learning_rate": 2.1755555555555556e-06,
"loss": 0.3879,
"step": 4025
},
{
"epoch": 0.86,
"learning_rate": 2.12e-06,
"loss": 0.4024,
"step": 4050
},
{
"epoch": 0.86,
"learning_rate": 2.064444444444445e-06,
"loss": 0.4621,
"step": 4075
},
{
"epoch": 0.87,
"learning_rate": 2.008888888888889e-06,
"loss": 0.4002,
"step": 4100
},
{
"epoch": 0.87,
"learning_rate": 1.9533333333333334e-06,
"loss": 0.4211,
"step": 4125
},
{
"epoch": 0.88,
"learning_rate": 1.8977777777777779e-06,
"loss": 0.538,
"step": 4150
},
{
"epoch": 0.88,
"learning_rate": 1.8422222222222225e-06,
"loss": 0.4723,
"step": 4175
},
{
"epoch": 0.89,
"learning_rate": 1.7866666666666668e-06,
"loss": 0.4394,
"step": 4200
},
{
"epoch": 0.89,
"learning_rate": 1.7311111111111112e-06,
"loss": 0.4386,
"step": 4225
},
{
"epoch": 0.9,
"learning_rate": 1.675555555555556e-06,
"loss": 0.4327,
"step": 4250
},
{
"epoch": 0.9,
"learning_rate": 1.6200000000000002e-06,
"loss": 0.4054,
"step": 4275
},
{
"epoch": 0.91,
"learning_rate": 1.5644444444444446e-06,
"loss": 0.4158,
"step": 4300
},
{
"epoch": 0.91,
"learning_rate": 1.5088888888888889e-06,
"loss": 0.4526,
"step": 4325
},
{
"epoch": 0.92,
"learning_rate": 1.4533333333333335e-06,
"loss": 0.4507,
"step": 4350
},
{
"epoch": 0.93,
"learning_rate": 1.397777777777778e-06,
"loss": 0.4554,
"step": 4375
},
{
"epoch": 0.93,
"learning_rate": 1.3422222222222222e-06,
"loss": 0.3666,
"step": 4400
},
{
"epoch": 0.94,
"learning_rate": 1.286666666666667e-06,
"loss": 0.46,
"step": 4425
},
{
"epoch": 0.94,
"learning_rate": 1.2311111111111112e-06,
"loss": 0.4839,
"step": 4450
},
{
"epoch": 0.95,
"learning_rate": 1.1755555555555556e-06,
"loss": 0.4566,
"step": 4475
},
{
"epoch": 0.95,
"learning_rate": 1.12e-06,
"loss": 0.4083,
"step": 4500
},
{
"epoch": 0.96,
"learning_rate": 1.0644444444444445e-06,
"loss": 0.4123,
"step": 4525
},
{
"epoch": 0.96,
"learning_rate": 1.008888888888889e-06,
"loss": 0.4659,
"step": 4550
},
{
"epoch": 0.97,
"learning_rate": 9.533333333333335e-07,
"loss": 0.4195,
"step": 4575
},
{
"epoch": 0.97,
"learning_rate": 8.977777777777778e-07,
"loss": 0.4304,
"step": 4600
},
{
"epoch": 0.98,
"learning_rate": 8.422222222222224e-07,
"loss": 0.4558,
"step": 4625
},
{
"epoch": 0.98,
"learning_rate": 7.866666666666667e-07,
"loss": 0.4637,
"step": 4650
},
{
"epoch": 0.99,
"learning_rate": 7.311111111111112e-07,
"loss": 0.4263,
"step": 4675
},
{
"epoch": 0.99,
"learning_rate": 6.755555555555555e-07,
"loss": 0.3846,
"step": 4700
},
{
"epoch": 1.0,
"learning_rate": 6.200000000000001e-07,
"loss": 0.4278,
"step": 4725
},
{
"epoch": 1.0,
"learning_rate": 5.644444444444445e-07,
"loss": 0.3001,
"step": 4750
},
{
"epoch": 1.01,
"learning_rate": 5.088888888888889e-07,
"loss": 0.288,
"step": 4775
},
{
"epoch": 1.02,
"learning_rate": 4.533333333333334e-07,
"loss": 0.2635,
"step": 4800
},
{
"epoch": 1.02,
"learning_rate": 3.9777777777777783e-07,
"loss": 0.2257,
"step": 4825
},
{
"epoch": 1.03,
"learning_rate": 3.422222222222223e-07,
"loss": 0.2932,
"step": 4850
},
{
"epoch": 1.03,
"learning_rate": 2.866666666666667e-07,
"loss": 0.2626,
"step": 4875
},
{
"epoch": 1.04,
"learning_rate": 2.3111111111111112e-07,
"loss": 0.3071,
"step": 4900
},
{
"epoch": 1.04,
"learning_rate": 1.7555555555555558e-07,
"loss": 0.2844,
"step": 4925
},
{
"epoch": 1.05,
"learning_rate": 1.2000000000000002e-07,
"loss": 0.2685,
"step": 4950
},
{
"epoch": 1.05,
"learning_rate": 6.444444444444445e-08,
"loss": 0.2903,
"step": 4975
},
{
"epoch": 1.06,
"learning_rate": 8.88888888888889e-09,
"loss": 0.2716,
"step": 5000
},
{
"epoch": 1.06,
"eval_loss": 0.23201139271259308,
"eval_runtime": 593.0687,
"eval_samples_per_second": 6.107,
"eval_steps_per_second": 1.528,
"eval_wer": 12.900851161720727,
"step": 5000
},
{
"epoch": 1.06,
"step": 5000,
"total_flos": 1.154168450039808e+19,
"train_loss": 0.5289400071144104,
"train_runtime": 8043.0939,
"train_samples_per_second": 4.973,
"train_steps_per_second": 0.622
}
],
"max_steps": 5000,
"num_train_epochs": 2,
"total_flos": 1.154168450039808e+19,
"trial_name": null,
"trial_params": null
}