whisper-th-medium-combined / trainer_state.json
tensorops's picture
Revert "update model"
22bed6d
{
"best_metric": 17.947723376834624,
"best_model_checkpoint": "./checkpoint-5000",
"epoch": 41.666666666666664,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.21,
"learning_rate": 4.898977360288234e-06,
"loss": 0.546,
"step": 25
},
{
"epoch": 0.42,
"learning_rate": 6.160712527409633e-06,
"loss": 0.4065,
"step": 50
},
{
"epoch": 0.62,
"learning_rate": 6.85912902234906e-06,
"loss": 0.3709,
"step": 75
},
{
"epoch": 0.83,
"learning_rate": 7.344547104469332e-06,
"loss": 0.3362,
"step": 100
},
{
"epoch": 1.04,
"learning_rate": 7.716963756434345e-06,
"loss": 0.3056,
"step": 125
},
{
"epoch": 1.25,
"learning_rate": 8.019180844200955e-06,
"loss": 0.2381,
"step": 150
},
{
"epoch": 1.46,
"learning_rate": 8.27351214279797e-06,
"loss": 0.232,
"step": 175
},
{
"epoch": 1.67,
"learning_rate": 8.49307723936858e-06,
"loss": 0.2129,
"step": 200
},
{
"epoch": 1.88,
"learning_rate": 8.686247975778677e-06,
"loss": 0.214,
"step": 225
},
{
"epoch": 2.08,
"learning_rate": 8.858694625217149e-06,
"loss": 0.1779,
"step": 250
},
{
"epoch": 2.29,
"learning_rate": 9.014436199608479e-06,
"loss": 0.1383,
"step": 275
},
{
"epoch": 2.5,
"learning_rate": 9.156425255148058e-06,
"loss": 0.1322,
"step": 300
},
{
"epoch": 2.71,
"learning_rate": 9.28689473531776e-06,
"loss": 0.1354,
"step": 325
},
{
"epoch": 2.92,
"learning_rate": 9.407574351377137e-06,
"loss": 0.1304,
"step": 350
},
{
"epoch": 3.12,
"learning_rate": 9.519831289296397e-06,
"loss": 0.1044,
"step": 375
},
{
"epoch": 3.33,
"learning_rate": 9.624764935335318e-06,
"loss": 0.076,
"step": 400
},
{
"epoch": 3.54,
"learning_rate": 9.723272550712454e-06,
"loss": 0.079,
"step": 425
},
{
"epoch": 3.75,
"learning_rate": 9.816095971633122e-06,
"loss": 0.0791,
"step": 450
},
{
"epoch": 3.96,
"learning_rate": 9.90385555539545e-06,
"loss": 0.0794,
"step": 475
},
{
"epoch": 4.17,
"learning_rate": 9.987075336738768e-06,
"loss": 0.0534,
"step": 500
},
{
"epoch": 4.38,
"learning_rate": 9.955555555555556e-06,
"loss": 0.0446,
"step": 525
},
{
"epoch": 4.58,
"learning_rate": 9.9e-06,
"loss": 0.0456,
"step": 550
},
{
"epoch": 4.79,
"learning_rate": 9.844444444444446e-06,
"loss": 0.0457,
"step": 575
},
{
"epoch": 5.0,
"learning_rate": 9.78888888888889e-06,
"loss": 0.044,
"step": 600
},
{
"epoch": 5.21,
"learning_rate": 9.733333333333334e-06,
"loss": 0.0244,
"step": 625
},
{
"epoch": 5.42,
"learning_rate": 9.677777777777778e-06,
"loss": 0.0264,
"step": 650
},
{
"epoch": 5.62,
"learning_rate": 9.622222222222222e-06,
"loss": 0.0266,
"step": 675
},
{
"epoch": 5.83,
"learning_rate": 9.566666666666668e-06,
"loss": 0.0253,
"step": 700
},
{
"epoch": 6.04,
"learning_rate": 9.511111111111112e-06,
"loss": 0.024,
"step": 725
},
{
"epoch": 6.25,
"learning_rate": 9.455555555555557e-06,
"loss": 0.0136,
"step": 750
},
{
"epoch": 6.46,
"learning_rate": 9.4e-06,
"loss": 0.013,
"step": 775
},
{
"epoch": 6.67,
"learning_rate": 9.344444444444446e-06,
"loss": 0.0136,
"step": 800
},
{
"epoch": 6.88,
"learning_rate": 9.28888888888889e-06,
"loss": 0.0142,
"step": 825
},
{
"epoch": 7.08,
"learning_rate": 9.233333333333334e-06,
"loss": 0.0131,
"step": 850
},
{
"epoch": 7.29,
"learning_rate": 9.17777777777778e-06,
"loss": 0.0081,
"step": 875
},
{
"epoch": 7.5,
"learning_rate": 9.122222222222223e-06,
"loss": 0.0075,
"step": 900
},
{
"epoch": 7.71,
"learning_rate": 9.066666666666667e-06,
"loss": 0.0085,
"step": 925
},
{
"epoch": 7.92,
"learning_rate": 9.011111111111111e-06,
"loss": 0.0072,
"step": 950
},
{
"epoch": 8.12,
"learning_rate": 8.955555555555555e-06,
"loss": 0.0052,
"step": 975
},
{
"epoch": 8.33,
"learning_rate": 8.900000000000001e-06,
"loss": 0.004,
"step": 1000
},
{
"epoch": 8.54,
"learning_rate": 8.844444444444445e-06,
"loss": 0.0042,
"step": 1025
},
{
"epoch": 8.75,
"learning_rate": 8.788888888888891e-06,
"loss": 0.0045,
"step": 1050
},
{
"epoch": 8.96,
"learning_rate": 8.733333333333333e-06,
"loss": 0.0044,
"step": 1075
},
{
"epoch": 9.17,
"learning_rate": 8.677777777777779e-06,
"loss": 0.0035,
"step": 1100
},
{
"epoch": 9.38,
"learning_rate": 8.622222222222223e-06,
"loss": 0.0029,
"step": 1125
},
{
"epoch": 9.58,
"learning_rate": 8.566666666666667e-06,
"loss": 0.0024,
"step": 1150
},
{
"epoch": 9.79,
"learning_rate": 8.511111111111113e-06,
"loss": 0.0032,
"step": 1175
},
{
"epoch": 10.0,
"learning_rate": 8.455555555555555e-06,
"loss": 0.003,
"step": 1200
},
{
"epoch": 10.21,
"learning_rate": 8.400000000000001e-06,
"loss": 0.0021,
"step": 1225
},
{
"epoch": 10.42,
"learning_rate": 8.344444444444445e-06,
"loss": 0.0024,
"step": 1250
},
{
"epoch": 10.62,
"learning_rate": 8.288888888888889e-06,
"loss": 0.0023,
"step": 1275
},
{
"epoch": 10.83,
"learning_rate": 8.233333333333335e-06,
"loss": 0.0024,
"step": 1300
},
{
"epoch": 11.04,
"learning_rate": 8.177777777777779e-06,
"loss": 0.0019,
"step": 1325
},
{
"epoch": 11.25,
"learning_rate": 8.122222222222223e-06,
"loss": 0.0014,
"step": 1350
},
{
"epoch": 11.46,
"learning_rate": 8.066666666666667e-06,
"loss": 0.0015,
"step": 1375
},
{
"epoch": 11.67,
"learning_rate": 8.011111111111113e-06,
"loss": 0.0018,
"step": 1400
},
{
"epoch": 11.88,
"learning_rate": 7.955555555555557e-06,
"loss": 0.0015,
"step": 1425
},
{
"epoch": 12.08,
"learning_rate": 7.9e-06,
"loss": 0.0016,
"step": 1450
},
{
"epoch": 12.29,
"learning_rate": 7.844444444444446e-06,
"loss": 0.0015,
"step": 1475
},
{
"epoch": 12.5,
"learning_rate": 7.788888888888889e-06,
"loss": 0.0017,
"step": 1500
},
{
"epoch": 12.71,
"learning_rate": 7.733333333333334e-06,
"loss": 0.0016,
"step": 1525
},
{
"epoch": 12.92,
"learning_rate": 7.677777777777778e-06,
"loss": 0.002,
"step": 1550
},
{
"epoch": 13.12,
"learning_rate": 7.622222222222223e-06,
"loss": 0.0015,
"step": 1575
},
{
"epoch": 13.33,
"learning_rate": 7.566666666666667e-06,
"loss": 0.0015,
"step": 1600
},
{
"epoch": 13.54,
"learning_rate": 7.511111111111111e-06,
"loss": 0.0015,
"step": 1625
},
{
"epoch": 13.75,
"learning_rate": 7.455555555555556e-06,
"loss": 0.0014,
"step": 1650
},
{
"epoch": 13.96,
"learning_rate": 7.4e-06,
"loss": 0.0017,
"step": 1675
},
{
"epoch": 14.17,
"learning_rate": 7.344444444444445e-06,
"loss": 0.0014,
"step": 1700
},
{
"epoch": 14.38,
"learning_rate": 7.28888888888889e-06,
"loss": 0.0013,
"step": 1725
},
{
"epoch": 14.58,
"learning_rate": 7.233333333333334e-06,
"loss": 0.0013,
"step": 1750
},
{
"epoch": 14.79,
"learning_rate": 7.177777777777778e-06,
"loss": 0.0018,
"step": 1775
},
{
"epoch": 15.0,
"learning_rate": 7.122222222222222e-06,
"loss": 0.0015,
"step": 1800
},
{
"epoch": 15.21,
"learning_rate": 7.066666666666667e-06,
"loss": 0.0017,
"step": 1825
},
{
"epoch": 15.42,
"learning_rate": 7.011111111111112e-06,
"loss": 0.0014,
"step": 1850
},
{
"epoch": 15.62,
"learning_rate": 6.955555555555557e-06,
"loss": 0.0018,
"step": 1875
},
{
"epoch": 15.83,
"learning_rate": 6.9e-06,
"loss": 0.0018,
"step": 1900
},
{
"epoch": 16.04,
"learning_rate": 6.844444444444445e-06,
"loss": 0.0021,
"step": 1925
},
{
"epoch": 16.25,
"learning_rate": 6.788888888888889e-06,
"loss": 0.0024,
"step": 1950
},
{
"epoch": 16.46,
"learning_rate": 6.733333333333334e-06,
"loss": 0.0019,
"step": 1975
},
{
"epoch": 16.67,
"learning_rate": 6.677777777777779e-06,
"loss": 0.0018,
"step": 2000
},
{
"epoch": 16.88,
"learning_rate": 6.6222222222222236e-06,
"loss": 0.0021,
"step": 2025
},
{
"epoch": 17.08,
"learning_rate": 6.566666666666667e-06,
"loss": 0.0019,
"step": 2050
},
{
"epoch": 17.29,
"learning_rate": 6.511111111111112e-06,
"loss": 0.0017,
"step": 2075
},
{
"epoch": 17.5,
"learning_rate": 6.455555555555556e-06,
"loss": 0.0013,
"step": 2100
},
{
"epoch": 17.71,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.0021,
"step": 2125
},
{
"epoch": 17.92,
"learning_rate": 6.3444444444444454e-06,
"loss": 0.0015,
"step": 2150
},
{
"epoch": 18.12,
"learning_rate": 6.28888888888889e-06,
"loss": 0.0015,
"step": 2175
},
{
"epoch": 18.33,
"learning_rate": 6.2333333333333335e-06,
"loss": 0.0016,
"step": 2200
},
{
"epoch": 18.54,
"learning_rate": 6.177777777777778e-06,
"loss": 0.0014,
"step": 2225
},
{
"epoch": 18.75,
"learning_rate": 6.1222222222222224e-06,
"loss": 0.0017,
"step": 2250
},
{
"epoch": 18.96,
"learning_rate": 6.066666666666667e-06,
"loss": 0.0015,
"step": 2275
},
{
"epoch": 19.17,
"learning_rate": 6.011111111111112e-06,
"loss": 0.0016,
"step": 2300
},
{
"epoch": 19.38,
"learning_rate": 5.955555555555555e-06,
"loss": 0.0014,
"step": 2325
},
{
"epoch": 19.58,
"learning_rate": 5.9e-06,
"loss": 0.0013,
"step": 2350
},
{
"epoch": 19.79,
"learning_rate": 5.844444444444445e-06,
"loss": 0.0013,
"step": 2375
},
{
"epoch": 20.0,
"learning_rate": 5.788888888888889e-06,
"loss": 0.0014,
"step": 2400
},
{
"epoch": 20.21,
"learning_rate": 5.733333333333334e-06,
"loss": 0.0009,
"step": 2425
},
{
"epoch": 20.42,
"learning_rate": 5.677777777777779e-06,
"loss": 0.0012,
"step": 2450
},
{
"epoch": 20.62,
"learning_rate": 5.622222222222222e-06,
"loss": 0.0014,
"step": 2475
},
{
"epoch": 20.83,
"learning_rate": 5.566666666666667e-06,
"loss": 0.0011,
"step": 2500
},
{
"epoch": 21.04,
"learning_rate": 5.511111111111112e-06,
"loss": 0.0011,
"step": 2525
},
{
"epoch": 21.25,
"learning_rate": 5.455555555555556e-06,
"loss": 0.001,
"step": 2550
},
{
"epoch": 21.46,
"learning_rate": 5.400000000000001e-06,
"loss": 0.0011,
"step": 2575
},
{
"epoch": 21.67,
"learning_rate": 5.344444444444446e-06,
"loss": 0.001,
"step": 2600
},
{
"epoch": 21.88,
"learning_rate": 5.288888888888889e-06,
"loss": 0.0012,
"step": 2625
},
{
"epoch": 22.08,
"learning_rate": 5.233333333333334e-06,
"loss": 0.0008,
"step": 2650
},
{
"epoch": 22.29,
"learning_rate": 5.177777777777779e-06,
"loss": 0.0007,
"step": 2675
},
{
"epoch": 22.5,
"learning_rate": 5.122222222222223e-06,
"loss": 0.0007,
"step": 2700
},
{
"epoch": 22.71,
"learning_rate": 5.0666666666666676e-06,
"loss": 0.0009,
"step": 2725
},
{
"epoch": 22.92,
"learning_rate": 5.011111111111111e-06,
"loss": 0.0006,
"step": 2750
},
{
"epoch": 23.12,
"learning_rate": 4.9555555555555565e-06,
"loss": 0.0006,
"step": 2775
},
{
"epoch": 23.33,
"learning_rate": 4.9000000000000005e-06,
"loss": 0.0003,
"step": 2800
},
{
"epoch": 23.54,
"learning_rate": 4.8444444444444446e-06,
"loss": 0.0007,
"step": 2825
},
{
"epoch": 23.75,
"learning_rate": 4.7888888888888894e-06,
"loss": 0.0004,
"step": 2850
},
{
"epoch": 23.96,
"learning_rate": 4.7333333333333335e-06,
"loss": 0.0004,
"step": 2875
},
{
"epoch": 24.17,
"learning_rate": 4.677777777777778e-06,
"loss": 0.0003,
"step": 2900
},
{
"epoch": 24.38,
"learning_rate": 4.622222222222222e-06,
"loss": 0.0003,
"step": 2925
},
{
"epoch": 24.58,
"learning_rate": 4.566666666666667e-06,
"loss": 0.0003,
"step": 2950
},
{
"epoch": 24.79,
"learning_rate": 4.511111111111111e-06,
"loss": 0.0003,
"step": 2975
},
{
"epoch": 25.0,
"learning_rate": 4.455555555555555e-06,
"loss": 0.0002,
"step": 3000
},
{
"epoch": 25.21,
"learning_rate": 4.404444444444445e-06,
"loss": 0.0002,
"step": 3025
},
{
"epoch": 25.42,
"learning_rate": 4.348888888888889e-06,
"loss": 0.0002,
"step": 3050
},
{
"epoch": 25.62,
"learning_rate": 4.2933333333333334e-06,
"loss": 0.0002,
"step": 3075
},
{
"epoch": 25.83,
"learning_rate": 4.2377777777777775e-06,
"loss": 0.0002,
"step": 3100
},
{
"epoch": 26.04,
"learning_rate": 4.182222222222222e-06,
"loss": 0.0002,
"step": 3125
},
{
"epoch": 26.25,
"learning_rate": 4.126666666666667e-06,
"loss": 0.0003,
"step": 3150
},
{
"epoch": 26.46,
"learning_rate": 4.071111111111111e-06,
"loss": 0.0002,
"step": 3175
},
{
"epoch": 26.67,
"learning_rate": 4.015555555555556e-06,
"loss": 0.0003,
"step": 3200
},
{
"epoch": 26.88,
"learning_rate": 3.96e-06,
"loss": 0.0002,
"step": 3225
},
{
"epoch": 27.08,
"learning_rate": 3.904444444444444e-06,
"loss": 0.0001,
"step": 3250
},
{
"epoch": 27.29,
"learning_rate": 3.848888888888889e-06,
"loss": 0.0002,
"step": 3275
},
{
"epoch": 27.5,
"learning_rate": 3.793333333333334e-06,
"loss": 0.0002,
"step": 3300
},
{
"epoch": 27.71,
"learning_rate": 3.737777777777778e-06,
"loss": 0.0002,
"step": 3325
},
{
"epoch": 27.92,
"learning_rate": 3.6822222222222225e-06,
"loss": 0.0001,
"step": 3350
},
{
"epoch": 28.12,
"learning_rate": 3.6266666666666674e-06,
"loss": 0.0001,
"step": 3375
},
{
"epoch": 28.33,
"learning_rate": 3.5711111111111114e-06,
"loss": 0.0001,
"step": 3400
},
{
"epoch": 28.54,
"learning_rate": 3.515555555555556e-06,
"loss": 0.0002,
"step": 3425
},
{
"epoch": 28.75,
"learning_rate": 3.46e-06,
"loss": 0.0001,
"step": 3450
},
{
"epoch": 28.96,
"learning_rate": 3.404444444444445e-06,
"loss": 0.0001,
"step": 3475
},
{
"epoch": 29.17,
"learning_rate": 3.3488888888888892e-06,
"loss": 0.0001,
"step": 3500
},
{
"epoch": 29.38,
"learning_rate": 3.2933333333333333e-06,
"loss": 0.0001,
"step": 3525
},
{
"epoch": 29.58,
"learning_rate": 3.237777777777778e-06,
"loss": 0.0001,
"step": 3550
},
{
"epoch": 29.79,
"learning_rate": 3.1822222222222226e-06,
"loss": 0.0001,
"step": 3575
},
{
"epoch": 30.0,
"learning_rate": 3.1266666666666667e-06,
"loss": 0.0001,
"step": 3600
},
{
"epoch": 30.21,
"learning_rate": 3.0711111111111115e-06,
"loss": 0.0001,
"step": 3625
},
{
"epoch": 30.42,
"learning_rate": 3.015555555555556e-06,
"loss": 0.0001,
"step": 3650
},
{
"epoch": 30.62,
"learning_rate": 2.96e-06,
"loss": 0.0001,
"step": 3675
},
{
"epoch": 30.83,
"learning_rate": 2.904444444444445e-06,
"loss": 0.0001,
"step": 3700
},
{
"epoch": 31.04,
"learning_rate": 2.8488888888888894e-06,
"loss": 0.0001,
"step": 3725
},
{
"epoch": 31.25,
"learning_rate": 2.7933333333333334e-06,
"loss": 0.0001,
"step": 3750
},
{
"epoch": 31.46,
"learning_rate": 2.7377777777777783e-06,
"loss": 0.0001,
"step": 3775
},
{
"epoch": 31.67,
"learning_rate": 2.6822222222222223e-06,
"loss": 0.0001,
"step": 3800
},
{
"epoch": 31.88,
"learning_rate": 2.6266666666666668e-06,
"loss": 0.0001,
"step": 3825
},
{
"epoch": 32.08,
"learning_rate": 2.5711111111111112e-06,
"loss": 0.0001,
"step": 3850
},
{
"epoch": 32.29,
"learning_rate": 2.5155555555555557e-06,
"loss": 0.0001,
"step": 3875
},
{
"epoch": 32.5,
"learning_rate": 2.46e-06,
"loss": 0.0001,
"step": 3900
},
{
"epoch": 32.71,
"learning_rate": 2.4044444444444446e-06,
"loss": 0.0001,
"step": 3925
},
{
"epoch": 32.92,
"learning_rate": 2.348888888888889e-06,
"loss": 0.0001,
"step": 3950
},
{
"epoch": 33.12,
"learning_rate": 2.2933333333333335e-06,
"loss": 0.0001,
"step": 3975
},
{
"epoch": 33.33,
"learning_rate": 2.237777777777778e-06,
"loss": 0.0001,
"step": 4000
},
{
"epoch": 33.54,
"learning_rate": 2.1866666666666668e-06,
"loss": 0.0001,
"step": 4025
},
{
"epoch": 33.75,
"learning_rate": 2.1311111111111112e-06,
"loss": 0.0001,
"step": 4050
},
{
"epoch": 33.96,
"learning_rate": 2.0755555555555557e-06,
"loss": 0.0001,
"step": 4075
},
{
"epoch": 34.17,
"learning_rate": 2.02e-06,
"loss": 0.0001,
"step": 4100
},
{
"epoch": 34.38,
"learning_rate": 1.9644444444444446e-06,
"loss": 0.0001,
"step": 4125
},
{
"epoch": 34.58,
"learning_rate": 1.908888888888889e-06,
"loss": 0.0001,
"step": 4150
},
{
"epoch": 34.79,
"learning_rate": 1.8533333333333333e-06,
"loss": 0.0001,
"step": 4175
},
{
"epoch": 35.0,
"learning_rate": 1.797777777777778e-06,
"loss": 0.0001,
"step": 4200
},
{
"epoch": 35.21,
"learning_rate": 1.7422222222222224e-06,
"loss": 0.0001,
"step": 4225
},
{
"epoch": 35.42,
"learning_rate": 1.6866666666666667e-06,
"loss": 0.0001,
"step": 4250
},
{
"epoch": 35.62,
"learning_rate": 1.6311111111111114e-06,
"loss": 0.0001,
"step": 4275
},
{
"epoch": 35.83,
"learning_rate": 1.5755555555555558e-06,
"loss": 0.0001,
"step": 4300
},
{
"epoch": 36.04,
"learning_rate": 1.52e-06,
"loss": 0.0001,
"step": 4325
},
{
"epoch": 36.25,
"learning_rate": 1.4644444444444445e-06,
"loss": 0.0001,
"step": 4350
},
{
"epoch": 36.46,
"learning_rate": 1.4088888888888892e-06,
"loss": 0.0001,
"step": 4375
},
{
"epoch": 36.67,
"learning_rate": 1.3533333333333334e-06,
"loss": 0.0001,
"step": 4400
},
{
"epoch": 36.88,
"learning_rate": 1.2977777777777779e-06,
"loss": 0.0001,
"step": 4425
},
{
"epoch": 37.08,
"learning_rate": 1.2422222222222224e-06,
"loss": 0.0001,
"step": 4450
},
{
"epoch": 37.29,
"learning_rate": 1.1866666666666668e-06,
"loss": 0.0001,
"step": 4475
},
{
"epoch": 37.5,
"learning_rate": 1.131111111111111e-06,
"loss": 0.0001,
"step": 4500
},
{
"epoch": 37.71,
"learning_rate": 1.0755555555555557e-06,
"loss": 0.0001,
"step": 4525
},
{
"epoch": 37.92,
"learning_rate": 1.02e-06,
"loss": 0.0001,
"step": 4550
},
{
"epoch": 38.12,
"learning_rate": 9.644444444444444e-07,
"loss": 0.0001,
"step": 4575
},
{
"epoch": 38.33,
"learning_rate": 9.08888888888889e-07,
"loss": 0.0001,
"step": 4600
},
{
"epoch": 38.54,
"learning_rate": 8.533333333333334e-07,
"loss": 0.0001,
"step": 4625
},
{
"epoch": 38.75,
"learning_rate": 7.977777777777779e-07,
"loss": 0.0001,
"step": 4650
},
{
"epoch": 38.96,
"learning_rate": 7.422222222222223e-07,
"loss": 0.0001,
"step": 4675
},
{
"epoch": 39.17,
"learning_rate": 6.866666666666667e-07,
"loss": 0.0001,
"step": 4700
},
{
"epoch": 39.38,
"learning_rate": 6.311111111111112e-07,
"loss": 0.0001,
"step": 4725
},
{
"epoch": 39.58,
"learning_rate": 5.755555555555555e-07,
"loss": 0.0001,
"step": 4750
},
{
"epoch": 39.79,
"learning_rate": 5.2e-07,
"loss": 0.0001,
"step": 4775
},
{
"epoch": 40.0,
"learning_rate": 4.6444444444444446e-07,
"loss": 0.0001,
"step": 4800
},
{
"epoch": 40.21,
"learning_rate": 4.0888888888888897e-07,
"loss": 0.0001,
"step": 4825
},
{
"epoch": 40.42,
"learning_rate": 3.533333333333334e-07,
"loss": 0.0001,
"step": 4850
},
{
"epoch": 40.62,
"learning_rate": 2.977777777777778e-07,
"loss": 0.0001,
"step": 4875
},
{
"epoch": 40.83,
"learning_rate": 2.4222222222222224e-07,
"loss": 0.0001,
"step": 4900
},
{
"epoch": 41.04,
"learning_rate": 1.866666666666667e-07,
"loss": 0.0001,
"step": 4925
},
{
"epoch": 41.25,
"learning_rate": 1.3111111111111113e-07,
"loss": 0.0001,
"step": 4950
},
{
"epoch": 41.46,
"learning_rate": 7.555555555555556e-08,
"loss": 0.0001,
"step": 4975
},
{
"epoch": 41.67,
"learning_rate": 2e-08,
"loss": 0.0001,
"step": 5000
},
{
"epoch": 41.67,
"eval_loss": 0.345458984375,
"eval_runtime": 344.0011,
"eval_samples_per_second": 5.901,
"eval_steps_per_second": 0.186,
"eval_wer": 17.947723376834624,
"step": 5000
},
{
"epoch": 41.67,
"step": 5000,
"total_flos": 1.6225066659249494e+20,
"train_loss": 0.02298767093420029,
"train_runtime": 31651.1636,
"train_samples_per_second": 5.055,
"train_steps_per_second": 0.158
}
],
"max_steps": 5000,
"num_train_epochs": 42,
"total_flos": 1.6225066659249494e+20,
"trial_name": null,
"trial_params": null
}