whisper-small-eu / trainer_state.json
xezpeleta's picture
End of training
53e38ec
raw
history blame
No virus
25.8 kB
{
"best_metric": 18.775568066750374,
"best_model_checkpoint": "./checkpoint-4000",
"epoch": 12.0112,
"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": 1.6639,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.400000000000001e-07,
"loss": 1.4728,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.44e-06,
"loss": 1.1563,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.94e-06,
"loss": 0.9368,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.8889,
"step": 125
},
{
"epoch": 0.03,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.8114,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 3.44e-06,
"loss": 0.6924,
"step": 175
},
{
"epoch": 0.04,
"learning_rate": 3.94e-06,
"loss": 0.661,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 4.440000000000001e-06,
"loss": 0.6427,
"step": 225
},
{
"epoch": 0.05,
"learning_rate": 4.94e-06,
"loss": 0.5217,
"step": 250
},
{
"epoch": 0.06,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.4991,
"step": 275
},
{
"epoch": 0.06,
"learning_rate": 5.94e-06,
"loss": 0.4768,
"step": 300
},
{
"epoch": 0.07,
"learning_rate": 6.440000000000001e-06,
"loss": 0.4632,
"step": 325
},
{
"epoch": 0.07,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.4289,
"step": 350
},
{
"epoch": 0.07,
"learning_rate": 7.440000000000001e-06,
"loss": 0.404,
"step": 375
},
{
"epoch": 0.08,
"learning_rate": 7.94e-06,
"loss": 0.3833,
"step": 400
},
{
"epoch": 1.0,
"learning_rate": 8.44e-06,
"loss": 0.4201,
"step": 425
},
{
"epoch": 1.01,
"learning_rate": 8.94e-06,
"loss": 0.3989,
"step": 450
},
{
"epoch": 1.01,
"learning_rate": 9.440000000000001e-06,
"loss": 0.3453,
"step": 475
},
{
"epoch": 1.02,
"learning_rate": 9.940000000000001e-06,
"loss": 0.3559,
"step": 500
},
{
"epoch": 1.02,
"learning_rate": 9.951111111111111e-06,
"loss": 0.3429,
"step": 525
},
{
"epoch": 1.03,
"learning_rate": 9.895555555555557e-06,
"loss": 0.3701,
"step": 550
},
{
"epoch": 1.03,
"learning_rate": 9.84e-06,
"loss": 0.3084,
"step": 575
},
{
"epoch": 1.04,
"learning_rate": 9.784444444444445e-06,
"loss": 0.2956,
"step": 600
},
{
"epoch": 1.04,
"learning_rate": 9.72888888888889e-06,
"loss": 0.2923,
"step": 625
},
{
"epoch": 1.05,
"learning_rate": 9.673333333333334e-06,
"loss": 0.2786,
"step": 650
},
{
"epoch": 1.05,
"learning_rate": 9.617777777777778e-06,
"loss": 0.2427,
"step": 675
},
{
"epoch": 1.06,
"learning_rate": 9.562222222222223e-06,
"loss": 0.2387,
"step": 700
},
{
"epoch": 1.06,
"learning_rate": 9.506666666666667e-06,
"loss": 0.2371,
"step": 725
},
{
"epoch": 1.07,
"learning_rate": 9.451111111111112e-06,
"loss": 0.2114,
"step": 750
},
{
"epoch": 1.07,
"learning_rate": 9.395555555555556e-06,
"loss": 0.1894,
"step": 775
},
{
"epoch": 1.08,
"learning_rate": 9.340000000000002e-06,
"loss": 0.1867,
"step": 800
},
{
"epoch": 2.0,
"learning_rate": 9.284444444444444e-06,
"loss": 0.1927,
"step": 825
},
{
"epoch": 2.01,
"learning_rate": 9.22888888888889e-06,
"loss": 0.1951,
"step": 850
},
{
"epoch": 2.01,
"learning_rate": 9.173333333333334e-06,
"loss": 0.1775,
"step": 875
},
{
"epoch": 2.02,
"learning_rate": 9.117777777777778e-06,
"loss": 0.1672,
"step": 900
},
{
"epoch": 2.02,
"learning_rate": 9.062222222222224e-06,
"loss": 0.1591,
"step": 925
},
{
"epoch": 2.03,
"learning_rate": 9.006666666666666e-06,
"loss": 0.1703,
"step": 950
},
{
"epoch": 2.03,
"learning_rate": 8.951111111111112e-06,
"loss": 0.1636,
"step": 975
},
{
"epoch": 2.04,
"learning_rate": 8.895555555555556e-06,
"loss": 0.1413,
"step": 1000
},
{
"epoch": 2.04,
"eval_loss": 0.31775403022766113,
"eval_runtime": 963.5028,
"eval_samples_per_second": 6.841,
"eval_steps_per_second": 0.428,
"eval_wer": 22.013852322896838,
"step": 1000
},
{
"epoch": 2.04,
"learning_rate": 8.84e-06,
"loss": 0.1371,
"step": 1025
},
{
"epoch": 2.05,
"learning_rate": 8.784444444444446e-06,
"loss": 0.1494,
"step": 1050
},
{
"epoch": 2.05,
"learning_rate": 8.72888888888889e-06,
"loss": 0.1196,
"step": 1075
},
{
"epoch": 2.06,
"learning_rate": 8.673333333333334e-06,
"loss": 0.1152,
"step": 1100
},
{
"epoch": 2.06,
"learning_rate": 8.617777777777778e-06,
"loss": 0.1158,
"step": 1125
},
{
"epoch": 2.07,
"learning_rate": 8.562222222222224e-06,
"loss": 0.1054,
"step": 1150
},
{
"epoch": 2.07,
"learning_rate": 8.506666666666668e-06,
"loss": 0.0912,
"step": 1175
},
{
"epoch": 2.08,
"learning_rate": 8.451111111111112e-06,
"loss": 0.0915,
"step": 1200
},
{
"epoch": 2.08,
"learning_rate": 8.395555555555557e-06,
"loss": 0.0901,
"step": 1225
},
{
"epoch": 3.0,
"learning_rate": 8.34e-06,
"loss": 0.0953,
"step": 1250
},
{
"epoch": 3.01,
"learning_rate": 8.284444444444446e-06,
"loss": 0.0923,
"step": 1275
},
{
"epoch": 3.01,
"learning_rate": 8.22888888888889e-06,
"loss": 0.0749,
"step": 1300
},
{
"epoch": 3.02,
"learning_rate": 8.173333333333334e-06,
"loss": 0.0796,
"step": 1325
},
{
"epoch": 3.02,
"learning_rate": 8.11777777777778e-06,
"loss": 0.0786,
"step": 1350
},
{
"epoch": 3.03,
"learning_rate": 8.062222222222222e-06,
"loss": 0.0819,
"step": 1375
},
{
"epoch": 3.03,
"learning_rate": 8.006666666666667e-06,
"loss": 0.0747,
"step": 1400
},
{
"epoch": 3.04,
"learning_rate": 7.951111111111111e-06,
"loss": 0.0646,
"step": 1425
},
{
"epoch": 3.04,
"learning_rate": 7.895555555555557e-06,
"loss": 0.0707,
"step": 1450
},
{
"epoch": 3.05,
"learning_rate": 7.840000000000001e-06,
"loss": 0.0639,
"step": 1475
},
{
"epoch": 3.05,
"learning_rate": 7.784444444444445e-06,
"loss": 0.0544,
"step": 1500
},
{
"epoch": 3.06,
"learning_rate": 7.72888888888889e-06,
"loss": 0.0531,
"step": 1525
},
{
"epoch": 3.06,
"learning_rate": 7.673333333333333e-06,
"loss": 0.0518,
"step": 1550
},
{
"epoch": 3.07,
"learning_rate": 7.617777777777778e-06,
"loss": 0.0461,
"step": 1575
},
{
"epoch": 3.07,
"learning_rate": 7.562222222222223e-06,
"loss": 0.0424,
"step": 1600
},
{
"epoch": 3.08,
"learning_rate": 7.506666666666668e-06,
"loss": 0.0392,
"step": 1625
},
{
"epoch": 4.0,
"learning_rate": 7.451111111111111e-06,
"loss": 0.0376,
"step": 1650
},
{
"epoch": 4.01,
"learning_rate": 7.395555555555556e-06,
"loss": 0.0447,
"step": 1675
},
{
"epoch": 4.01,
"learning_rate": 7.340000000000001e-06,
"loss": 0.0362,
"step": 1700
},
{
"epoch": 4.02,
"learning_rate": 7.284444444444445e-06,
"loss": 0.0368,
"step": 1725
},
{
"epoch": 4.02,
"learning_rate": 7.22888888888889e-06,
"loss": 0.0367,
"step": 1750
},
{
"epoch": 4.03,
"learning_rate": 7.173333333333335e-06,
"loss": 0.0391,
"step": 1775
},
{
"epoch": 4.03,
"learning_rate": 7.117777777777778e-06,
"loss": 0.0331,
"step": 1800
},
{
"epoch": 4.04,
"learning_rate": 7.062222222222223e-06,
"loss": 0.028,
"step": 1825
},
{
"epoch": 4.04,
"learning_rate": 7.006666666666667e-06,
"loss": 0.0315,
"step": 1850
},
{
"epoch": 4.05,
"learning_rate": 6.951111111111112e-06,
"loss": 0.0308,
"step": 1875
},
{
"epoch": 4.05,
"learning_rate": 6.8955555555555565e-06,
"loss": 0.0248,
"step": 1900
},
{
"epoch": 4.06,
"learning_rate": 6.8400000000000014e-06,
"loss": 0.023,
"step": 1925
},
{
"epoch": 4.06,
"learning_rate": 6.784444444444445e-06,
"loss": 0.0233,
"step": 1950
},
{
"epoch": 4.07,
"learning_rate": 6.7288888888888895e-06,
"loss": 0.0236,
"step": 1975
},
{
"epoch": 4.07,
"learning_rate": 6.6733333333333335e-06,
"loss": 0.0181,
"step": 2000
},
{
"epoch": 4.07,
"eval_loss": 0.33755382895469666,
"eval_runtime": 952.5181,
"eval_samples_per_second": 6.92,
"eval_steps_per_second": 0.433,
"eval_wer": 20.286362347604197,
"step": 2000
},
{
"epoch": 4.08,
"learning_rate": 6.617777777777778e-06,
"loss": 0.0176,
"step": 2025
},
{
"epoch": 4.08,
"learning_rate": 6.562222222222223e-06,
"loss": 0.0177,
"step": 2050
},
{
"epoch": 5.0,
"learning_rate": 6.5066666666666665e-06,
"loss": 0.0198,
"step": 2075
},
{
"epoch": 5.01,
"learning_rate": 6.451111111111111e-06,
"loss": 0.0188,
"step": 2100
},
{
"epoch": 5.01,
"learning_rate": 6.395555555555556e-06,
"loss": 0.0153,
"step": 2125
},
{
"epoch": 5.02,
"learning_rate": 6.34e-06,
"loss": 0.0157,
"step": 2150
},
{
"epoch": 5.02,
"learning_rate": 6.284444444444445e-06,
"loss": 0.0196,
"step": 2175
},
{
"epoch": 5.03,
"learning_rate": 6.22888888888889e-06,
"loss": 0.0153,
"step": 2200
},
{
"epoch": 5.03,
"learning_rate": 6.173333333333333e-06,
"loss": 0.0132,
"step": 2225
},
{
"epoch": 5.04,
"learning_rate": 6.117777777777778e-06,
"loss": 0.0144,
"step": 2250
},
{
"epoch": 5.04,
"learning_rate": 6.062222222222223e-06,
"loss": 0.0145,
"step": 2275
},
{
"epoch": 5.05,
"learning_rate": 6.006666666666667e-06,
"loss": 0.0131,
"step": 2300
},
{
"epoch": 5.05,
"learning_rate": 5.951111111111112e-06,
"loss": 0.0107,
"step": 2325
},
{
"epoch": 5.06,
"learning_rate": 5.895555555555557e-06,
"loss": 0.0122,
"step": 2350
},
{
"epoch": 5.06,
"learning_rate": 5.84e-06,
"loss": 0.0121,
"step": 2375
},
{
"epoch": 5.07,
"learning_rate": 5.784444444444445e-06,
"loss": 0.0103,
"step": 2400
},
{
"epoch": 5.07,
"learning_rate": 5.72888888888889e-06,
"loss": 0.0089,
"step": 2425
},
{
"epoch": 5.08,
"learning_rate": 5.673333333333334e-06,
"loss": 0.0087,
"step": 2450
},
{
"epoch": 6.0,
"learning_rate": 5.617777777777779e-06,
"loss": 0.0087,
"step": 2475
},
{
"epoch": 6.01,
"learning_rate": 5.562222222222222e-06,
"loss": 0.0106,
"step": 2500
},
{
"epoch": 6.01,
"learning_rate": 5.506666666666667e-06,
"loss": 0.0094,
"step": 2525
},
{
"epoch": 6.02,
"learning_rate": 5.451111111111112e-06,
"loss": 0.0082,
"step": 2550
},
{
"epoch": 6.02,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.0079,
"step": 2575
},
{
"epoch": 6.03,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.0084,
"step": 2600
},
{
"epoch": 6.03,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.0076,
"step": 2625
},
{
"epoch": 6.04,
"learning_rate": 5.228888888888889e-06,
"loss": 0.0069,
"step": 2650
},
{
"epoch": 6.04,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.0067,
"step": 2675
},
{
"epoch": 6.05,
"learning_rate": 5.117777777777778e-06,
"loss": 0.0073,
"step": 2700
},
{
"epoch": 6.05,
"learning_rate": 5.062222222222222e-06,
"loss": 0.0062,
"step": 2725
},
{
"epoch": 6.06,
"learning_rate": 5.006666666666667e-06,
"loss": 0.0057,
"step": 2750
},
{
"epoch": 6.06,
"learning_rate": 4.951111111111111e-06,
"loss": 0.0064,
"step": 2775
},
{
"epoch": 6.07,
"learning_rate": 4.895555555555556e-06,
"loss": 0.0053,
"step": 2800
},
{
"epoch": 6.07,
"learning_rate": 4.84e-06,
"loss": 0.0047,
"step": 2825
},
{
"epoch": 6.08,
"learning_rate": 4.784444444444445e-06,
"loss": 0.005,
"step": 2850
},
{
"epoch": 6.08,
"learning_rate": 4.728888888888889e-06,
"loss": 0.0047,
"step": 2875
},
{
"epoch": 7.0,
"learning_rate": 4.673333333333333e-06,
"loss": 0.0052,
"step": 2900
},
{
"epoch": 7.01,
"learning_rate": 4.617777777777778e-06,
"loss": 0.005,
"step": 2925
},
{
"epoch": 7.01,
"learning_rate": 4.562222222222222e-06,
"loss": 0.0041,
"step": 2950
},
{
"epoch": 7.02,
"learning_rate": 4.506666666666667e-06,
"loss": 0.0045,
"step": 2975
},
{
"epoch": 7.02,
"learning_rate": 4.451111111111112e-06,
"loss": 0.0044,
"step": 3000
},
{
"epoch": 7.02,
"eval_loss": 0.3602650463581085,
"eval_runtime": 960.4855,
"eval_samples_per_second": 6.862,
"eval_steps_per_second": 0.429,
"eval_wer": 18.876827737048888,
"step": 3000
},
{
"epoch": 7.03,
"learning_rate": 4.395555555555556e-06,
"loss": 0.0043,
"step": 3025
},
{
"epoch": 7.03,
"learning_rate": 4.34e-06,
"loss": 0.0041,
"step": 3050
},
{
"epoch": 7.04,
"learning_rate": 4.284444444444445e-06,
"loss": 0.0035,
"step": 3075
},
{
"epoch": 7.04,
"learning_rate": 4.228888888888889e-06,
"loss": 0.0044,
"step": 3100
},
{
"epoch": 7.05,
"learning_rate": 4.173333333333334e-06,
"loss": 0.0039,
"step": 3125
},
{
"epoch": 7.05,
"learning_rate": 4.117777777777779e-06,
"loss": 0.0031,
"step": 3150
},
{
"epoch": 7.06,
"learning_rate": 4.062222222222223e-06,
"loss": 0.0032,
"step": 3175
},
{
"epoch": 7.06,
"learning_rate": 4.006666666666667e-06,
"loss": 0.0034,
"step": 3200
},
{
"epoch": 7.07,
"learning_rate": 3.951111111111112e-06,
"loss": 0.0028,
"step": 3225
},
{
"epoch": 7.07,
"learning_rate": 3.895555555555556e-06,
"loss": 0.0032,
"step": 3250
},
{
"epoch": 7.08,
"learning_rate": 3.8400000000000005e-06,
"loss": 0.003,
"step": 3275
},
{
"epoch": 8.0,
"learning_rate": 3.784444444444445e-06,
"loss": 0.0034,
"step": 3300
},
{
"epoch": 8.01,
"learning_rate": 3.728888888888889e-06,
"loss": 0.003,
"step": 3325
},
{
"epoch": 8.01,
"learning_rate": 3.673333333333334e-06,
"loss": 0.0027,
"step": 3350
},
{
"epoch": 8.02,
"learning_rate": 3.617777777777778e-06,
"loss": 0.0026,
"step": 3375
},
{
"epoch": 8.02,
"learning_rate": 3.5622222222222224e-06,
"loss": 0.0028,
"step": 3400
},
{
"epoch": 8.03,
"learning_rate": 3.5066666666666673e-06,
"loss": 0.0029,
"step": 3425
},
{
"epoch": 8.03,
"learning_rate": 3.4511111111111113e-06,
"loss": 0.0027,
"step": 3450
},
{
"epoch": 8.04,
"learning_rate": 3.3955555555555558e-06,
"loss": 0.0022,
"step": 3475
},
{
"epoch": 8.04,
"learning_rate": 3.3400000000000006e-06,
"loss": 0.0024,
"step": 3500
},
{
"epoch": 8.05,
"learning_rate": 3.2844444444444447e-06,
"loss": 0.0028,
"step": 3525
},
{
"epoch": 8.05,
"learning_rate": 3.228888888888889e-06,
"loss": 0.0025,
"step": 3550
},
{
"epoch": 8.06,
"learning_rate": 3.173333333333334e-06,
"loss": 0.0022,
"step": 3575
},
{
"epoch": 8.06,
"learning_rate": 3.117777777777778e-06,
"loss": 0.0021,
"step": 3600
},
{
"epoch": 8.07,
"learning_rate": 3.0622222222222225e-06,
"loss": 0.002,
"step": 3625
},
{
"epoch": 8.07,
"learning_rate": 3.0066666666666674e-06,
"loss": 0.0019,
"step": 3650
},
{
"epoch": 8.08,
"learning_rate": 2.9511111111111114e-06,
"loss": 0.0019,
"step": 3675
},
{
"epoch": 8.08,
"learning_rate": 2.895555555555556e-06,
"loss": 0.0019,
"step": 3700
},
{
"epoch": 9.0,
"learning_rate": 2.84e-06,
"loss": 0.0021,
"step": 3725
},
{
"epoch": 9.01,
"learning_rate": 2.784444444444445e-06,
"loss": 0.0021,
"step": 3750
},
{
"epoch": 9.01,
"learning_rate": 2.7288888888888893e-06,
"loss": 0.0018,
"step": 3775
},
{
"epoch": 9.02,
"learning_rate": 2.6733333333333333e-06,
"loss": 0.0019,
"step": 3800
},
{
"epoch": 9.02,
"learning_rate": 2.617777777777778e-06,
"loss": 0.0018,
"step": 3825
},
{
"epoch": 9.03,
"learning_rate": 2.5622222222222226e-06,
"loss": 0.0021,
"step": 3850
},
{
"epoch": 9.03,
"learning_rate": 2.5066666666666667e-06,
"loss": 0.0017,
"step": 3875
},
{
"epoch": 9.04,
"learning_rate": 2.451111111111111e-06,
"loss": 0.0018,
"step": 3900
},
{
"epoch": 9.04,
"learning_rate": 2.3955555555555556e-06,
"loss": 0.0017,
"step": 3925
},
{
"epoch": 9.05,
"learning_rate": 2.3400000000000005e-06,
"loss": 0.002,
"step": 3950
},
{
"epoch": 9.05,
"learning_rate": 2.2844444444444445e-06,
"loss": 0.0017,
"step": 3975
},
{
"epoch": 9.06,
"learning_rate": 2.228888888888889e-06,
"loss": 0.0016,
"step": 4000
},
{
"epoch": 9.06,
"eval_loss": 0.3812163174152374,
"eval_runtime": 968.0397,
"eval_samples_per_second": 6.809,
"eval_steps_per_second": 0.426,
"eval_wer": 18.775568066750374,
"step": 4000
},
{
"epoch": 9.06,
"learning_rate": 2.1733333333333334e-06,
"loss": 0.0016,
"step": 4025
},
{
"epoch": 9.07,
"learning_rate": 2.117777777777778e-06,
"loss": 0.0015,
"step": 4050
},
{
"epoch": 9.07,
"learning_rate": 2.0622222222222223e-06,
"loss": 0.0015,
"step": 4075
},
{
"epoch": 9.08,
"learning_rate": 2.006666666666667e-06,
"loss": 0.0015,
"step": 4100
},
{
"epoch": 10.0,
"learning_rate": 1.9511111111111113e-06,
"loss": 0.0015,
"step": 4125
},
{
"epoch": 10.01,
"learning_rate": 1.8955555555555557e-06,
"loss": 0.0016,
"step": 4150
},
{
"epoch": 10.01,
"learning_rate": 1.8400000000000002e-06,
"loss": 0.0015,
"step": 4175
},
{
"epoch": 10.02,
"learning_rate": 1.7844444444444444e-06,
"loss": 0.0015,
"step": 4200
},
{
"epoch": 10.02,
"learning_rate": 1.728888888888889e-06,
"loss": 0.0014,
"step": 4225
},
{
"epoch": 10.03,
"learning_rate": 1.6733333333333335e-06,
"loss": 0.0016,
"step": 4250
},
{
"epoch": 10.03,
"learning_rate": 1.6177777777777778e-06,
"loss": 0.0015,
"step": 4275
},
{
"epoch": 10.04,
"learning_rate": 1.5622222222222225e-06,
"loss": 0.0015,
"step": 4300
},
{
"epoch": 10.04,
"learning_rate": 1.506666666666667e-06,
"loss": 0.0016,
"step": 4325
},
{
"epoch": 10.05,
"learning_rate": 1.4511111111111112e-06,
"loss": 0.0015,
"step": 4350
},
{
"epoch": 10.05,
"learning_rate": 1.3955555555555556e-06,
"loss": 0.0014,
"step": 4375
},
{
"epoch": 10.06,
"learning_rate": 1.34e-06,
"loss": 0.0014,
"step": 4400
},
{
"epoch": 10.06,
"learning_rate": 1.2844444444444445e-06,
"loss": 0.0013,
"step": 4425
},
{
"epoch": 10.07,
"learning_rate": 1.228888888888889e-06,
"loss": 0.0013,
"step": 4450
},
{
"epoch": 10.07,
"learning_rate": 1.1733333333333335e-06,
"loss": 0.0013,
"step": 4475
},
{
"epoch": 10.08,
"learning_rate": 1.117777777777778e-06,
"loss": 0.0013,
"step": 4500
},
{
"epoch": 10.08,
"learning_rate": 1.0622222222222222e-06,
"loss": 0.0013,
"step": 4525
},
{
"epoch": 11.0,
"learning_rate": 1.0066666666666668e-06,
"loss": 0.0014,
"step": 4550
},
{
"epoch": 11.01,
"learning_rate": 9.511111111111111e-07,
"loss": 0.0014,
"step": 4575
},
{
"epoch": 11.01,
"learning_rate": 8.955555555555557e-07,
"loss": 0.0013,
"step": 4600
},
{
"epoch": 11.02,
"learning_rate": 8.400000000000001e-07,
"loss": 0.0013,
"step": 4625
},
{
"epoch": 11.02,
"learning_rate": 7.844444444444445e-07,
"loss": 0.0014,
"step": 4650
},
{
"epoch": 11.03,
"learning_rate": 7.28888888888889e-07,
"loss": 0.0015,
"step": 4675
},
{
"epoch": 11.03,
"learning_rate": 6.733333333333334e-07,
"loss": 0.0013,
"step": 4700
},
{
"epoch": 11.04,
"learning_rate": 6.177777777777778e-07,
"loss": 0.0013,
"step": 4725
},
{
"epoch": 11.04,
"learning_rate": 5.622222222222223e-07,
"loss": 0.0013,
"step": 4750
},
{
"epoch": 11.05,
"learning_rate": 5.066666666666667e-07,
"loss": 0.0013,
"step": 4775
},
{
"epoch": 11.05,
"learning_rate": 4.511111111111111e-07,
"loss": 0.0013,
"step": 4800
},
{
"epoch": 11.06,
"learning_rate": 3.9555555555555557e-07,
"loss": 0.0013,
"step": 4825
},
{
"epoch": 11.06,
"learning_rate": 3.4000000000000003e-07,
"loss": 0.0013,
"step": 4850
},
{
"epoch": 11.07,
"learning_rate": 2.844444444444445e-07,
"loss": 0.0013,
"step": 4875
},
{
"epoch": 11.07,
"learning_rate": 2.2888888888888892e-07,
"loss": 0.0012,
"step": 4900
},
{
"epoch": 11.08,
"learning_rate": 1.7333333333333335e-07,
"loss": 0.0012,
"step": 4925
},
{
"epoch": 12.0,
"learning_rate": 1.1777777777777778e-07,
"loss": 0.0013,
"step": 4950
},
{
"epoch": 12.01,
"learning_rate": 6.222222222222223e-08,
"loss": 0.0013,
"step": 4975
},
{
"epoch": 12.01,
"learning_rate": 6.666666666666667e-09,
"loss": 0.0012,
"step": 5000
},
{
"epoch": 12.01,
"eval_loss": 0.3913772702217102,
"eval_runtime": 973.2149,
"eval_samples_per_second": 6.772,
"eval_steps_per_second": 0.423,
"eval_wer": 18.83024828871157,
"step": 5000
},
{
"epoch": 12.01,
"step": 5000,
"total_flos": 4.616673800159232e+19,
"train_loss": 0.106446673027426,
"train_runtime": 27653.1068,
"train_samples_per_second": 5.786,
"train_steps_per_second": 0.181
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 4.616673800159232e+19,
"trial_name": null,
"trial_params": null
}