whisper-small-mn / trainer_state.json
Dorjzodovsuren's picture
Upload 12 files
3f23b04
{
"best_metric": 52.89689034369886,
"best_model_checkpoint": "./whisper-small_v2_mn/checkpoint-2000",
"epoch": 31.746031746031747,
"eval_steps": 1000,
"global_step": 4000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.2,
"learning_rate": 4.2000000000000006e-07,
"loss": 5.8081,
"step": 25
},
{
"epoch": 0.4,
"learning_rate": 9.200000000000001e-07,
"loss": 4.3345,
"step": 50
},
{
"epoch": 0.6,
"learning_rate": 1.42e-06,
"loss": 3.1424,
"step": 75
},
{
"epoch": 0.79,
"learning_rate": 1.9200000000000003e-06,
"loss": 2.3848,
"step": 100
},
{
"epoch": 0.99,
"learning_rate": 2.42e-06,
"loss": 1.8362,
"step": 125
},
{
"epoch": 1.19,
"learning_rate": 2.92e-06,
"loss": 1.4852,
"step": 150
},
{
"epoch": 1.39,
"learning_rate": 3.4200000000000007e-06,
"loss": 1.2819,
"step": 175
},
{
"epoch": 1.59,
"learning_rate": 3.920000000000001e-06,
"loss": 1.1295,
"step": 200
},
{
"epoch": 1.79,
"learning_rate": 4.42e-06,
"loss": 0.9959,
"step": 225
},
{
"epoch": 1.98,
"learning_rate": 4.92e-06,
"loss": 0.9184,
"step": 250
},
{
"epoch": 2.18,
"learning_rate": 5.420000000000001e-06,
"loss": 0.7924,
"step": 275
},
{
"epoch": 2.38,
"learning_rate": 5.92e-06,
"loss": 0.6941,
"step": 300
},
{
"epoch": 2.58,
"learning_rate": 6.42e-06,
"loss": 0.6417,
"step": 325
},
{
"epoch": 2.78,
"learning_rate": 6.92e-06,
"loss": 0.5299,
"step": 350
},
{
"epoch": 2.98,
"learning_rate": 7.420000000000001e-06,
"loss": 0.4726,
"step": 375
},
{
"epoch": 3.17,
"learning_rate": 7.92e-06,
"loss": 0.3759,
"step": 400
},
{
"epoch": 3.37,
"learning_rate": 8.42e-06,
"loss": 0.3508,
"step": 425
},
{
"epoch": 3.57,
"learning_rate": 8.920000000000001e-06,
"loss": 0.3384,
"step": 450
},
{
"epoch": 3.77,
"learning_rate": 9.42e-06,
"loss": 0.3153,
"step": 475
},
{
"epoch": 3.97,
"learning_rate": 9.920000000000002e-06,
"loss": 0.3337,
"step": 500
},
{
"epoch": 4.17,
"learning_rate": 9.940000000000001e-06,
"loss": 0.2261,
"step": 525
},
{
"epoch": 4.37,
"learning_rate": 9.86857142857143e-06,
"loss": 0.1987,
"step": 550
},
{
"epoch": 4.56,
"learning_rate": 9.797142857142858e-06,
"loss": 0.1959,
"step": 575
},
{
"epoch": 4.76,
"learning_rate": 9.725714285714287e-06,
"loss": 0.1938,
"step": 600
},
{
"epoch": 4.96,
"learning_rate": 9.654285714285716e-06,
"loss": 0.2021,
"step": 625
},
{
"epoch": 5.16,
"learning_rate": 9.582857142857143e-06,
"loss": 0.1245,
"step": 650
},
{
"epoch": 5.36,
"learning_rate": 9.511428571428572e-06,
"loss": 0.1071,
"step": 675
},
{
"epoch": 5.56,
"learning_rate": 9.440000000000001e-06,
"loss": 0.1053,
"step": 700
},
{
"epoch": 5.75,
"learning_rate": 9.368571428571428e-06,
"loss": 0.1179,
"step": 725
},
{
"epoch": 5.95,
"learning_rate": 9.297142857142857e-06,
"loss": 0.1051,
"step": 750
},
{
"epoch": 6.15,
"learning_rate": 9.225714285714286e-06,
"loss": 0.0661,
"step": 775
},
{
"epoch": 6.35,
"learning_rate": 9.154285714285715e-06,
"loss": 0.0528,
"step": 800
},
{
"epoch": 6.55,
"learning_rate": 9.082857142857143e-06,
"loss": 0.0578,
"step": 825
},
{
"epoch": 6.75,
"learning_rate": 9.011428571428572e-06,
"loss": 0.0566,
"step": 850
},
{
"epoch": 6.94,
"learning_rate": 8.94e-06,
"loss": 0.0568,
"step": 875
},
{
"epoch": 7.14,
"learning_rate": 8.86857142857143e-06,
"loss": 0.0395,
"step": 900
},
{
"epoch": 7.34,
"learning_rate": 8.797142857142857e-06,
"loss": 0.0313,
"step": 925
},
{
"epoch": 7.54,
"learning_rate": 8.725714285714286e-06,
"loss": 0.0302,
"step": 950
},
{
"epoch": 7.74,
"learning_rate": 8.654285714285715e-06,
"loss": 0.0309,
"step": 975
},
{
"epoch": 7.94,
"learning_rate": 8.582857142857144e-06,
"loss": 0.0312,
"step": 1000
},
{
"epoch": 7.94,
"eval_loss": 0.640009880065918,
"eval_runtime": 477.2063,
"eval_samples_per_second": 3.952,
"eval_steps_per_second": 0.124,
"eval_wer": 54.06983087834152,
"step": 1000
},
{
"epoch": 8.13,
"learning_rate": 8.511428571428571e-06,
"loss": 0.0234,
"step": 1025
},
{
"epoch": 8.33,
"learning_rate": 8.44e-06,
"loss": 0.0184,
"step": 1050
},
{
"epoch": 8.53,
"learning_rate": 8.36857142857143e-06,
"loss": 0.0178,
"step": 1075
},
{
"epoch": 8.73,
"learning_rate": 8.297142857142859e-06,
"loss": 0.0199,
"step": 1100
},
{
"epoch": 8.93,
"learning_rate": 8.225714285714288e-06,
"loss": 0.0173,
"step": 1125
},
{
"epoch": 9.13,
"learning_rate": 8.154285714285715e-06,
"loss": 0.0131,
"step": 1150
},
{
"epoch": 9.33,
"learning_rate": 8.082857142857144e-06,
"loss": 0.0111,
"step": 1175
},
{
"epoch": 9.52,
"learning_rate": 8.011428571428573e-06,
"loss": 0.0115,
"step": 1200
},
{
"epoch": 9.72,
"learning_rate": 7.94e-06,
"loss": 0.01,
"step": 1225
},
{
"epoch": 9.92,
"learning_rate": 7.86857142857143e-06,
"loss": 0.013,
"step": 1250
},
{
"epoch": 10.12,
"learning_rate": 7.797142857142858e-06,
"loss": 0.0097,
"step": 1275
},
{
"epoch": 10.32,
"learning_rate": 7.725714285714286e-06,
"loss": 0.0074,
"step": 1300
},
{
"epoch": 10.52,
"learning_rate": 7.654285714285715e-06,
"loss": 0.0075,
"step": 1325
},
{
"epoch": 10.71,
"learning_rate": 7.5828571428571444e-06,
"loss": 0.0094,
"step": 1350
},
{
"epoch": 10.91,
"learning_rate": 7.511428571428572e-06,
"loss": 0.0107,
"step": 1375
},
{
"epoch": 11.11,
"learning_rate": 7.440000000000001e-06,
"loss": 0.0067,
"step": 1400
},
{
"epoch": 11.31,
"learning_rate": 7.36857142857143e-06,
"loss": 0.0058,
"step": 1425
},
{
"epoch": 11.51,
"learning_rate": 7.297142857142858e-06,
"loss": 0.0065,
"step": 1450
},
{
"epoch": 11.71,
"learning_rate": 7.225714285714286e-06,
"loss": 0.0069,
"step": 1475
},
{
"epoch": 11.9,
"learning_rate": 7.154285714285715e-06,
"loss": 0.0057,
"step": 1500
},
{
"epoch": 12.1,
"learning_rate": 7.082857142857143e-06,
"loss": 0.0049,
"step": 1525
},
{
"epoch": 12.3,
"learning_rate": 7.011428571428572e-06,
"loss": 0.0039,
"step": 1550
},
{
"epoch": 12.5,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.0044,
"step": 1575
},
{
"epoch": 12.7,
"learning_rate": 6.868571428571429e-06,
"loss": 0.0043,
"step": 1600
},
{
"epoch": 12.9,
"learning_rate": 6.797142857142858e-06,
"loss": 0.0041,
"step": 1625
},
{
"epoch": 13.1,
"learning_rate": 6.725714285714287e-06,
"loss": 0.0038,
"step": 1650
},
{
"epoch": 13.29,
"learning_rate": 6.654285714285716e-06,
"loss": 0.0027,
"step": 1675
},
{
"epoch": 13.49,
"learning_rate": 6.582857142857143e-06,
"loss": 0.0028,
"step": 1700
},
{
"epoch": 13.69,
"learning_rate": 6.511428571428572e-06,
"loss": 0.0027,
"step": 1725
},
{
"epoch": 13.89,
"learning_rate": 6.440000000000001e-06,
"loss": 0.0027,
"step": 1750
},
{
"epoch": 14.09,
"learning_rate": 6.368571428571429e-06,
"loss": 0.0021,
"step": 1775
},
{
"epoch": 14.29,
"learning_rate": 6.297142857142857e-06,
"loss": 0.0022,
"step": 1800
},
{
"epoch": 14.48,
"learning_rate": 6.225714285714286e-06,
"loss": 0.0033,
"step": 1825
},
{
"epoch": 14.68,
"learning_rate": 6.1542857142857145e-06,
"loss": 0.002,
"step": 1850
},
{
"epoch": 14.88,
"learning_rate": 6.0828571428571435e-06,
"loss": 0.0024,
"step": 1875
},
{
"epoch": 15.08,
"learning_rate": 6.011428571428572e-06,
"loss": 0.0021,
"step": 1900
},
{
"epoch": 15.28,
"learning_rate": 5.94e-06,
"loss": 0.0012,
"step": 1925
},
{
"epoch": 15.48,
"learning_rate": 5.868571428571429e-06,
"loss": 0.0017,
"step": 1950
},
{
"epoch": 15.67,
"learning_rate": 5.797142857142858e-06,
"loss": 0.0016,
"step": 1975
},
{
"epoch": 15.87,
"learning_rate": 5.725714285714287e-06,
"loss": 0.0021,
"step": 2000
},
{
"epoch": 15.87,
"eval_loss": 0.7654294371604919,
"eval_runtime": 475.0372,
"eval_samples_per_second": 3.97,
"eval_steps_per_second": 0.124,
"eval_wer": 52.89689034369886,
"step": 2000
},
{
"epoch": 16.07,
"learning_rate": 5.654285714285714e-06,
"loss": 0.0017,
"step": 2025
},
{
"epoch": 16.27,
"learning_rate": 5.582857142857143e-06,
"loss": 0.0017,
"step": 2050
},
{
"epoch": 16.47,
"learning_rate": 5.511428571428572e-06,
"loss": 0.0013,
"step": 2075
},
{
"epoch": 16.67,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.0012,
"step": 2100
},
{
"epoch": 16.87,
"learning_rate": 5.368571428571429e-06,
"loss": 0.0019,
"step": 2125
},
{
"epoch": 17.06,
"learning_rate": 5.297142857142858e-06,
"loss": 0.0015,
"step": 2150
},
{
"epoch": 17.26,
"learning_rate": 5.225714285714286e-06,
"loss": 0.0016,
"step": 2175
},
{
"epoch": 17.46,
"learning_rate": 5.154285714285715e-06,
"loss": 0.0013,
"step": 2200
},
{
"epoch": 17.66,
"learning_rate": 5.082857142857144e-06,
"loss": 0.0019,
"step": 2225
},
{
"epoch": 17.86,
"learning_rate": 5.011428571428571e-06,
"loss": 0.001,
"step": 2250
},
{
"epoch": 18.06,
"learning_rate": 4.94e-06,
"loss": 0.001,
"step": 2275
},
{
"epoch": 18.25,
"learning_rate": 4.868571428571429e-06,
"loss": 0.0008,
"step": 2300
},
{
"epoch": 18.45,
"learning_rate": 4.797142857142857e-06,
"loss": 0.0009,
"step": 2325
},
{
"epoch": 18.65,
"learning_rate": 4.725714285714286e-06,
"loss": 0.0007,
"step": 2350
},
{
"epoch": 18.85,
"learning_rate": 4.6542857142857145e-06,
"loss": 0.0008,
"step": 2375
},
{
"epoch": 19.05,
"learning_rate": 4.5828571428571435e-06,
"loss": 0.0007,
"step": 2400
},
{
"epoch": 19.25,
"learning_rate": 4.511428571428572e-06,
"loss": 0.0006,
"step": 2425
},
{
"epoch": 19.44,
"learning_rate": 4.440000000000001e-06,
"loss": 0.0006,
"step": 2450
},
{
"epoch": 19.64,
"learning_rate": 4.368571428571429e-06,
"loss": 0.0011,
"step": 2475
},
{
"epoch": 19.84,
"learning_rate": 4.297142857142858e-06,
"loss": 0.0006,
"step": 2500
},
{
"epoch": 20.04,
"learning_rate": 4.225714285714286e-06,
"loss": 0.0006,
"step": 2525
},
{
"epoch": 20.24,
"learning_rate": 4.154285714285714e-06,
"loss": 0.0005,
"step": 2550
},
{
"epoch": 20.44,
"learning_rate": 4.082857142857143e-06,
"loss": 0.0005,
"step": 2575
},
{
"epoch": 20.63,
"learning_rate": 4.011428571428571e-06,
"loss": 0.0005,
"step": 2600
},
{
"epoch": 20.83,
"learning_rate": 3.94e-06,
"loss": 0.0005,
"step": 2625
},
{
"epoch": 21.03,
"learning_rate": 3.8685714285714286e-06,
"loss": 0.0005,
"step": 2650
},
{
"epoch": 21.23,
"learning_rate": 3.7971428571428576e-06,
"loss": 0.0005,
"step": 2675
},
{
"epoch": 21.43,
"learning_rate": 3.7257142857142857e-06,
"loss": 0.0004,
"step": 2700
},
{
"epoch": 21.63,
"learning_rate": 3.6542857142857148e-06,
"loss": 0.0004,
"step": 2725
},
{
"epoch": 21.83,
"learning_rate": 3.582857142857143e-06,
"loss": 0.0004,
"step": 2750
},
{
"epoch": 22.02,
"learning_rate": 3.511428571428572e-06,
"loss": 0.0005,
"step": 2775
},
{
"epoch": 22.22,
"learning_rate": 3.44e-06,
"loss": 0.0004,
"step": 2800
},
{
"epoch": 22.42,
"learning_rate": 3.3685714285714287e-06,
"loss": 0.0004,
"step": 2825
},
{
"epoch": 22.62,
"learning_rate": 3.2971428571428577e-06,
"loss": 0.0004,
"step": 2850
},
{
"epoch": 22.82,
"learning_rate": 3.225714285714286e-06,
"loss": 0.0004,
"step": 2875
},
{
"epoch": 23.02,
"learning_rate": 3.154285714285715e-06,
"loss": 0.0004,
"step": 2900
},
{
"epoch": 23.21,
"learning_rate": 3.082857142857143e-06,
"loss": 0.0004,
"step": 2925
},
{
"epoch": 23.41,
"learning_rate": 3.0114285714285716e-06,
"loss": 0.0004,
"step": 2950
},
{
"epoch": 23.61,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.0004,
"step": 2975
},
{
"epoch": 23.81,
"learning_rate": 2.868571428571429e-06,
"loss": 0.0004,
"step": 3000
},
{
"epoch": 23.81,
"eval_loss": 0.8085867762565613,
"eval_runtime": 479.2865,
"eval_samples_per_second": 3.935,
"eval_steps_per_second": 0.123,
"eval_wer": 53.10420076377523,
"step": 3000
},
{
"epoch": 24.01,
"learning_rate": 2.797142857142857e-06,
"loss": 0.0004,
"step": 3025
},
{
"epoch": 24.21,
"learning_rate": 2.725714285714286e-06,
"loss": 0.0004,
"step": 3050
},
{
"epoch": 24.4,
"learning_rate": 2.654285714285714e-06,
"loss": 0.0004,
"step": 3075
},
{
"epoch": 24.6,
"learning_rate": 2.582857142857143e-06,
"loss": 0.0004,
"step": 3100
},
{
"epoch": 24.8,
"learning_rate": 2.5114285714285718e-06,
"loss": 0.0004,
"step": 3125
},
{
"epoch": 25.0,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.0004,
"step": 3150
},
{
"epoch": 25.2,
"learning_rate": 2.3685714285714285e-06,
"loss": 0.0004,
"step": 3175
},
{
"epoch": 25.4,
"learning_rate": 2.297142857142857e-06,
"loss": 0.0003,
"step": 3200
},
{
"epoch": 25.6,
"learning_rate": 2.2257142857142857e-06,
"loss": 0.0003,
"step": 3225
},
{
"epoch": 25.79,
"learning_rate": 2.1542857142857147e-06,
"loss": 0.0003,
"step": 3250
},
{
"epoch": 25.99,
"learning_rate": 2.0828571428571433e-06,
"loss": 0.0003,
"step": 3275
},
{
"epoch": 26.19,
"learning_rate": 2.0114285714285715e-06,
"loss": 0.0003,
"step": 3300
},
{
"epoch": 26.39,
"learning_rate": 1.94e-06,
"loss": 0.0003,
"step": 3325
},
{
"epoch": 26.59,
"learning_rate": 1.8685714285714289e-06,
"loss": 0.0003,
"step": 3350
},
{
"epoch": 26.79,
"learning_rate": 1.7971428571428572e-06,
"loss": 0.0003,
"step": 3375
},
{
"epoch": 26.98,
"learning_rate": 1.7257142857142858e-06,
"loss": 0.0003,
"step": 3400
},
{
"epoch": 27.18,
"learning_rate": 1.6542857142857144e-06,
"loss": 0.0003,
"step": 3425
},
{
"epoch": 27.38,
"learning_rate": 1.582857142857143e-06,
"loss": 0.0003,
"step": 3450
},
{
"epoch": 27.58,
"learning_rate": 1.5114285714285714e-06,
"loss": 0.0003,
"step": 3475
},
{
"epoch": 27.78,
"learning_rate": 1.44e-06,
"loss": 0.0003,
"step": 3500
},
{
"epoch": 27.98,
"learning_rate": 1.3685714285714286e-06,
"loss": 0.0003,
"step": 3525
},
{
"epoch": 28.17,
"learning_rate": 1.2971428571428574e-06,
"loss": 0.0003,
"step": 3550
},
{
"epoch": 28.37,
"learning_rate": 1.2257142857142857e-06,
"loss": 0.0003,
"step": 3575
},
{
"epoch": 28.57,
"learning_rate": 1.1542857142857143e-06,
"loss": 0.0003,
"step": 3600
},
{
"epoch": 28.77,
"learning_rate": 1.082857142857143e-06,
"loss": 0.0003,
"step": 3625
},
{
"epoch": 28.97,
"learning_rate": 1.0114285714285715e-06,
"loss": 0.0003,
"step": 3650
},
{
"epoch": 29.17,
"learning_rate": 9.400000000000001e-07,
"loss": 0.0003,
"step": 3675
},
{
"epoch": 29.37,
"learning_rate": 8.685714285714286e-07,
"loss": 0.0003,
"step": 3700
},
{
"epoch": 29.56,
"learning_rate": 7.971428571428572e-07,
"loss": 0.0003,
"step": 3725
},
{
"epoch": 29.76,
"learning_rate": 7.257142857142857e-07,
"loss": 0.0003,
"step": 3750
},
{
"epoch": 29.96,
"learning_rate": 6.542857142857144e-07,
"loss": 0.0003,
"step": 3775
},
{
"epoch": 30.16,
"learning_rate": 5.82857142857143e-07,
"loss": 0.0003,
"step": 3800
},
{
"epoch": 30.36,
"learning_rate": 5.114285714285714e-07,
"loss": 0.0003,
"step": 3825
},
{
"epoch": 30.56,
"learning_rate": 4.4e-07,
"loss": 0.0003,
"step": 3850
},
{
"epoch": 30.75,
"learning_rate": 3.685714285714286e-07,
"loss": 0.0003,
"step": 3875
},
{
"epoch": 30.95,
"learning_rate": 2.9714285714285715e-07,
"loss": 0.0003,
"step": 3900
},
{
"epoch": 31.15,
"learning_rate": 2.2571428571428574e-07,
"loss": 0.0003,
"step": 3925
},
{
"epoch": 31.35,
"learning_rate": 1.542857142857143e-07,
"loss": 0.0003,
"step": 3950
},
{
"epoch": 31.55,
"learning_rate": 8.285714285714285e-08,
"loss": 0.0003,
"step": 3975
},
{
"epoch": 31.75,
"learning_rate": 1.142857142857143e-08,
"loss": 0.0003,
"step": 4000
},
{
"epoch": 31.75,
"eval_loss": 0.8255746960639954,
"eval_runtime": 479.9003,
"eval_samples_per_second": 3.93,
"eval_steps_per_second": 0.123,
"eval_wer": 53.136933987997814,
"step": 4000
}
],
"logging_steps": 25,
"max_steps": 4000,
"num_input_tokens_seen": 0,
"num_train_epochs": 32,
"save_steps": 1000,
"total_flos": 3.69389315127509e+19,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}