whisper-uzbek / trainer_state.json
Makhmud's picture
Upload trainer_state.json
41d135f verified
{
"best_metric": 34.928488014365186,
"best_model_checkpoint": "./whisper-small-uz/checkpoint-4000",
"epoch": 2.1569156106767324,
"eval_steps": 1000,
"global_step": 4000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 5.000000000000001e-07,
"loss": 5.8743,
"step": 25
},
{
"epoch": 0.03,
"learning_rate": 1.0000000000000002e-06,
"loss": 4.5217,
"step": 50
},
{
"epoch": 0.04,
"learning_rate": 1.5e-06,
"loss": 3.1166,
"step": 75
},
{
"epoch": 0.05,
"learning_rate": 2.0000000000000003e-06,
"loss": 2.3201,
"step": 100
},
{
"epoch": 0.07,
"learning_rate": 2.5e-06,
"loss": 2.0321,
"step": 125
},
{
"epoch": 0.08,
"learning_rate": 3e-06,
"loss": 1.8315,
"step": 150
},
{
"epoch": 0.09,
"learning_rate": 3.5e-06,
"loss": 1.6767,
"step": 175
},
{
"epoch": 0.11,
"learning_rate": 4.000000000000001e-06,
"loss": 1.5361,
"step": 200
},
{
"epoch": 0.12,
"learning_rate": 4.5e-06,
"loss": 1.4069,
"step": 225
},
{
"epoch": 0.13,
"learning_rate": 5e-06,
"loss": 1.3365,
"step": 250
},
{
"epoch": 0.15,
"learning_rate": 5.500000000000001e-06,
"loss": 1.2859,
"step": 275
},
{
"epoch": 0.16,
"learning_rate": 6e-06,
"loss": 1.2073,
"step": 300
},
{
"epoch": 0.18,
"learning_rate": 6.5000000000000004e-06,
"loss": 1.1188,
"step": 325
},
{
"epoch": 0.19,
"learning_rate": 7e-06,
"loss": 1.0234,
"step": 350
},
{
"epoch": 0.2,
"learning_rate": 7.500000000000001e-06,
"loss": 0.8149,
"step": 375
},
{
"epoch": 0.22,
"learning_rate": 8.000000000000001e-06,
"loss": 0.773,
"step": 400
},
{
"epoch": 0.23,
"learning_rate": 8.5e-06,
"loss": 0.7559,
"step": 425
},
{
"epoch": 0.24,
"learning_rate": 9e-06,
"loss": 0.7063,
"step": 450
},
{
"epoch": 0.26,
"learning_rate": 9.5e-06,
"loss": 0.7092,
"step": 475
},
{
"epoch": 0.27,
"learning_rate": 1e-05,
"loss": 0.6875,
"step": 500
},
{
"epoch": 0.28,
"learning_rate": 9.92857142857143e-06,
"loss": 0.6551,
"step": 525
},
{
"epoch": 0.3,
"learning_rate": 9.857142857142859e-06,
"loss": 0.6519,
"step": 550
},
{
"epoch": 0.31,
"learning_rate": 9.785714285714286e-06,
"loss": 0.6382,
"step": 575
},
{
"epoch": 0.32,
"learning_rate": 9.714285714285715e-06,
"loss": 0.6039,
"step": 600
},
{
"epoch": 0.34,
"learning_rate": 9.642857142857144e-06,
"loss": 0.6033,
"step": 625
},
{
"epoch": 0.35,
"learning_rate": 9.571428571428573e-06,
"loss": 0.5866,
"step": 650
},
{
"epoch": 0.36,
"learning_rate": 9.5e-06,
"loss": 0.603,
"step": 675
},
{
"epoch": 0.38,
"learning_rate": 9.42857142857143e-06,
"loss": 0.5935,
"step": 700
},
{
"epoch": 0.39,
"learning_rate": 9.357142857142859e-06,
"loss": 0.557,
"step": 725
},
{
"epoch": 0.4,
"learning_rate": 9.285714285714288e-06,
"loss": 0.5639,
"step": 750
},
{
"epoch": 0.42,
"learning_rate": 9.214285714285715e-06,
"loss": 0.5286,
"step": 775
},
{
"epoch": 0.43,
"learning_rate": 9.142857142857144e-06,
"loss": 0.538,
"step": 800
},
{
"epoch": 0.44,
"learning_rate": 9.071428571428573e-06,
"loss": 0.4961,
"step": 825
},
{
"epoch": 0.46,
"learning_rate": 9e-06,
"loss": 0.5278,
"step": 850
},
{
"epoch": 0.47,
"learning_rate": 8.92857142857143e-06,
"loss": 0.5256,
"step": 875
},
{
"epoch": 0.49,
"learning_rate": 8.857142857142858e-06,
"loss": 0.5184,
"step": 900
},
{
"epoch": 0.5,
"learning_rate": 8.785714285714286e-06,
"loss": 0.4817,
"step": 925
},
{
"epoch": 0.51,
"learning_rate": 8.714285714285715e-06,
"loss": 0.5209,
"step": 950
},
{
"epoch": 0.53,
"learning_rate": 8.642857142857144e-06,
"loss": 0.4935,
"step": 975
},
{
"epoch": 0.54,
"learning_rate": 8.571428571428571e-06,
"loss": 0.4794,
"step": 1000
},
{
"epoch": 0.54,
"eval_loss": 0.45038458704948425,
"eval_runtime": 3123.7684,
"eval_samples_per_second": 3.934,
"eval_steps_per_second": 0.492,
"eval_wer": 42.07215239147632,
"step": 1000
},
{
"epoch": 0.55,
"learning_rate": 8.5e-06,
"loss": 0.4779,
"step": 1025
},
{
"epoch": 0.57,
"learning_rate": 8.428571428571429e-06,
"loss": 0.4896,
"step": 1050
},
{
"epoch": 0.58,
"learning_rate": 8.357142857142858e-06,
"loss": 0.4915,
"step": 1075
},
{
"epoch": 0.59,
"learning_rate": 8.285714285714287e-06,
"loss": 0.4944,
"step": 1100
},
{
"epoch": 0.61,
"learning_rate": 8.214285714285714e-06,
"loss": 0.4802,
"step": 1125
},
{
"epoch": 0.62,
"learning_rate": 8.142857142857143e-06,
"loss": 0.5046,
"step": 1150
},
{
"epoch": 0.63,
"learning_rate": 8.071428571428572e-06,
"loss": 0.4581,
"step": 1175
},
{
"epoch": 0.65,
"learning_rate": 8.000000000000001e-06,
"loss": 0.4465,
"step": 1200
},
{
"epoch": 0.66,
"learning_rate": 7.928571428571429e-06,
"loss": 0.4576,
"step": 1225
},
{
"epoch": 0.67,
"learning_rate": 7.857142857142858e-06,
"loss": 0.4614,
"step": 1250
},
{
"epoch": 0.69,
"learning_rate": 7.785714285714287e-06,
"loss": 0.4498,
"step": 1275
},
{
"epoch": 0.7,
"learning_rate": 7.714285714285716e-06,
"loss": 0.4521,
"step": 1300
},
{
"epoch": 0.71,
"learning_rate": 7.642857142857143e-06,
"loss": 0.4555,
"step": 1325
},
{
"epoch": 0.73,
"learning_rate": 7.571428571428572e-06,
"loss": 0.4856,
"step": 1350
},
{
"epoch": 0.74,
"learning_rate": 7.500000000000001e-06,
"loss": 0.4377,
"step": 1375
},
{
"epoch": 0.75,
"learning_rate": 7.428571428571429e-06,
"loss": 0.4529,
"step": 1400
},
{
"epoch": 0.77,
"learning_rate": 7.357142857142858e-06,
"loss": 0.4282,
"step": 1425
},
{
"epoch": 0.78,
"learning_rate": 7.285714285714286e-06,
"loss": 0.4244,
"step": 1450
},
{
"epoch": 0.8,
"learning_rate": 7.2142857142857145e-06,
"loss": 0.4424,
"step": 1475
},
{
"epoch": 0.81,
"learning_rate": 7.1428571428571436e-06,
"loss": 0.4303,
"step": 1500
},
{
"epoch": 0.82,
"learning_rate": 7.0714285714285726e-06,
"loss": 0.411,
"step": 1525
},
{
"epoch": 0.84,
"learning_rate": 7e-06,
"loss": 0.4003,
"step": 1550
},
{
"epoch": 0.85,
"learning_rate": 6.928571428571429e-06,
"loss": 0.4187,
"step": 1575
},
{
"epoch": 0.86,
"learning_rate": 6.857142857142858e-06,
"loss": 0.4248,
"step": 1600
},
{
"epoch": 0.88,
"learning_rate": 6.785714285714287e-06,
"loss": 0.4264,
"step": 1625
},
{
"epoch": 0.89,
"learning_rate": 6.714285714285714e-06,
"loss": 0.4193,
"step": 1650
},
{
"epoch": 0.9,
"learning_rate": 6.642857142857143e-06,
"loss": 0.425,
"step": 1675
},
{
"epoch": 0.92,
"learning_rate": 6.571428571428572e-06,
"loss": 0.3975,
"step": 1700
},
{
"epoch": 0.93,
"learning_rate": 6.5000000000000004e-06,
"loss": 0.4302,
"step": 1725
},
{
"epoch": 0.94,
"learning_rate": 6.4285714285714295e-06,
"loss": 0.4033,
"step": 1750
},
{
"epoch": 0.96,
"learning_rate": 6.357142857142858e-06,
"loss": 0.3912,
"step": 1775
},
{
"epoch": 0.97,
"learning_rate": 6.285714285714286e-06,
"loss": 0.4112,
"step": 1800
},
{
"epoch": 0.98,
"learning_rate": 6.214285714285715e-06,
"loss": 0.4381,
"step": 1825
},
{
"epoch": 1.0,
"learning_rate": 6.142857142857144e-06,
"loss": 0.4085,
"step": 1850
},
{
"epoch": 1.01,
"learning_rate": 6.071428571428571e-06,
"loss": 0.3286,
"step": 1875
},
{
"epoch": 1.02,
"learning_rate": 6e-06,
"loss": 0.3104,
"step": 1900
},
{
"epoch": 1.04,
"learning_rate": 5.928571428571429e-06,
"loss": 0.3033,
"step": 1925
},
{
"epoch": 1.05,
"learning_rate": 5.857142857142858e-06,
"loss": 0.302,
"step": 1950
},
{
"epoch": 1.06,
"learning_rate": 5.785714285714286e-06,
"loss": 0.3034,
"step": 1975
},
{
"epoch": 1.08,
"learning_rate": 5.7142857142857145e-06,
"loss": 0.313,
"step": 2000
},
{
"epoch": 1.08,
"eval_loss": 0.3821106553077698,
"eval_runtime": 3103.0984,
"eval_samples_per_second": 3.961,
"eval_steps_per_second": 0.495,
"eval_wer": 38.93918655901151,
"step": 2000
},
{
"epoch": 1.09,
"learning_rate": 5.6428571428571435e-06,
"loss": 0.3135,
"step": 2025
},
{
"epoch": 1.11,
"learning_rate": 5.571428571428572e-06,
"loss": 0.3213,
"step": 2050
},
{
"epoch": 1.12,
"learning_rate": 5.500000000000001e-06,
"loss": 0.3173,
"step": 2075
},
{
"epoch": 1.13,
"learning_rate": 5.428571428571429e-06,
"loss": 0.2961,
"step": 2100
},
{
"epoch": 1.15,
"learning_rate": 5.357142857142857e-06,
"loss": 0.3127,
"step": 2125
},
{
"epoch": 1.16,
"learning_rate": 5.285714285714286e-06,
"loss": 0.3217,
"step": 2150
},
{
"epoch": 1.17,
"learning_rate": 5.214285714285715e-06,
"loss": 0.3166,
"step": 2175
},
{
"epoch": 1.19,
"learning_rate": 5.142857142857142e-06,
"loss": 0.3018,
"step": 2200
},
{
"epoch": 1.2,
"learning_rate": 5.071428571428571e-06,
"loss": 0.3085,
"step": 2225
},
{
"epoch": 1.21,
"learning_rate": 5e-06,
"loss": 0.3006,
"step": 2250
},
{
"epoch": 1.23,
"learning_rate": 4.928571428571429e-06,
"loss": 0.2992,
"step": 2275
},
{
"epoch": 1.24,
"learning_rate": 4.857142857142858e-06,
"loss": 0.3049,
"step": 2300
},
{
"epoch": 1.25,
"learning_rate": 4.785714285714287e-06,
"loss": 0.3068,
"step": 2325
},
{
"epoch": 1.27,
"learning_rate": 4.714285714285715e-06,
"loss": 0.2947,
"step": 2350
},
{
"epoch": 1.28,
"learning_rate": 4.642857142857144e-06,
"loss": 0.3066,
"step": 2375
},
{
"epoch": 1.29,
"learning_rate": 4.571428571428572e-06,
"loss": 0.3068,
"step": 2400
},
{
"epoch": 1.31,
"learning_rate": 4.5e-06,
"loss": 0.32,
"step": 2425
},
{
"epoch": 1.32,
"learning_rate": 4.428571428571429e-06,
"loss": 0.2943,
"step": 2450
},
{
"epoch": 1.33,
"learning_rate": 4.357142857142857e-06,
"loss": 0.3142,
"step": 2475
},
{
"epoch": 1.35,
"learning_rate": 4.2857142857142855e-06,
"loss": 0.3189,
"step": 2500
},
{
"epoch": 1.36,
"learning_rate": 4.2142857142857145e-06,
"loss": 0.3004,
"step": 2525
},
{
"epoch": 1.38,
"learning_rate": 4.1428571428571435e-06,
"loss": 0.308,
"step": 2550
},
{
"epoch": 1.39,
"learning_rate": 4.071428571428572e-06,
"loss": 0.3011,
"step": 2575
},
{
"epoch": 1.4,
"learning_rate": 4.000000000000001e-06,
"loss": 0.3101,
"step": 2600
},
{
"epoch": 1.42,
"learning_rate": 3.928571428571429e-06,
"loss": 0.3057,
"step": 2625
},
{
"epoch": 1.43,
"learning_rate": 3.857142857142858e-06,
"loss": 0.323,
"step": 2650
},
{
"epoch": 1.44,
"learning_rate": 3.785714285714286e-06,
"loss": 0.3076,
"step": 2675
},
{
"epoch": 1.46,
"learning_rate": 3.7142857142857146e-06,
"loss": 0.3101,
"step": 2700
},
{
"epoch": 1.47,
"learning_rate": 3.642857142857143e-06,
"loss": 0.2994,
"step": 2725
},
{
"epoch": 1.48,
"learning_rate": 3.5714285714285718e-06,
"loss": 0.3041,
"step": 2750
},
{
"epoch": 1.5,
"learning_rate": 3.5e-06,
"loss": 0.3126,
"step": 2775
},
{
"epoch": 1.51,
"learning_rate": 3.428571428571429e-06,
"loss": 0.3026,
"step": 2800
},
{
"epoch": 1.52,
"learning_rate": 3.357142857142857e-06,
"loss": 0.3005,
"step": 2825
},
{
"epoch": 1.54,
"learning_rate": 3.285714285714286e-06,
"loss": 0.2911,
"step": 2850
},
{
"epoch": 1.55,
"learning_rate": 3.2142857142857147e-06,
"loss": 0.2966,
"step": 2875
},
{
"epoch": 1.56,
"learning_rate": 3.142857142857143e-06,
"loss": 0.3015,
"step": 2900
},
{
"epoch": 1.58,
"learning_rate": 3.071428571428572e-06,
"loss": 0.2913,
"step": 2925
},
{
"epoch": 1.59,
"learning_rate": 3e-06,
"loss": 0.3016,
"step": 2950
},
{
"epoch": 1.6,
"learning_rate": 2.928571428571429e-06,
"loss": 0.2983,
"step": 2975
},
{
"epoch": 1.62,
"learning_rate": 2.8571428571428573e-06,
"loss": 0.2948,
"step": 3000
},
{
"epoch": 1.62,
"eval_loss": 0.3468478322029114,
"eval_runtime": 3105.1004,
"eval_samples_per_second": 3.958,
"eval_steps_per_second": 0.495,
"eval_wer": 35.427000012556974,
"step": 3000
},
{
"epoch": 1.63,
"learning_rate": 2.785714285714286e-06,
"loss": 0.2908,
"step": 3025
},
{
"epoch": 1.64,
"learning_rate": 2.7142857142857144e-06,
"loss": 0.2932,
"step": 3050
},
{
"epoch": 1.66,
"learning_rate": 2.642857142857143e-06,
"loss": 0.2841,
"step": 3075
},
{
"epoch": 1.67,
"learning_rate": 2.571428571428571e-06,
"loss": 0.2967,
"step": 3100
},
{
"epoch": 1.69,
"learning_rate": 2.5e-06,
"loss": 0.292,
"step": 3125
},
{
"epoch": 1.7,
"learning_rate": 2.428571428571429e-06,
"loss": 0.2984,
"step": 3150
},
{
"epoch": 1.71,
"learning_rate": 2.3571428571428574e-06,
"loss": 0.3032,
"step": 3175
},
{
"epoch": 1.73,
"learning_rate": 2.285714285714286e-06,
"loss": 0.2994,
"step": 3200
},
{
"epoch": 1.74,
"learning_rate": 2.2142857142857146e-06,
"loss": 0.3113,
"step": 3225
},
{
"epoch": 1.75,
"learning_rate": 2.1428571428571427e-06,
"loss": 0.2926,
"step": 3250
},
{
"epoch": 1.77,
"learning_rate": 2.0714285714285717e-06,
"loss": 0.2749,
"step": 3275
},
{
"epoch": 1.78,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.2947,
"step": 3300
},
{
"epoch": 1.79,
"learning_rate": 1.928571428571429e-06,
"loss": 0.2796,
"step": 3325
},
{
"epoch": 1.81,
"learning_rate": 1.8571428571428573e-06,
"loss": 0.2781,
"step": 3350
},
{
"epoch": 1.82,
"learning_rate": 1.7857142857142859e-06,
"loss": 0.296,
"step": 3375
},
{
"epoch": 1.83,
"learning_rate": 1.7142857142857145e-06,
"loss": 0.2912,
"step": 3400
},
{
"epoch": 1.85,
"learning_rate": 1.642857142857143e-06,
"loss": 0.2926,
"step": 3425
},
{
"epoch": 1.86,
"learning_rate": 1.5714285714285714e-06,
"loss": 0.2783,
"step": 3450
},
{
"epoch": 1.87,
"learning_rate": 1.5e-06,
"loss": 0.286,
"step": 3475
},
{
"epoch": 1.89,
"learning_rate": 1.4285714285714286e-06,
"loss": 0.2876,
"step": 3500
},
{
"epoch": 1.9,
"learning_rate": 1.3571428571428572e-06,
"loss": 0.2921,
"step": 3525
},
{
"epoch": 1.91,
"learning_rate": 1.2857142857142856e-06,
"loss": 0.2826,
"step": 3550
},
{
"epoch": 1.93,
"learning_rate": 1.2142857142857144e-06,
"loss": 0.2786,
"step": 3575
},
{
"epoch": 1.94,
"learning_rate": 1.142857142857143e-06,
"loss": 0.2856,
"step": 3600
},
{
"epoch": 1.95,
"learning_rate": 1.0714285714285714e-06,
"loss": 0.2834,
"step": 3625
},
{
"epoch": 1.97,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.2862,
"step": 3650
},
{
"epoch": 1.98,
"learning_rate": 9.285714285714287e-07,
"loss": 0.2818,
"step": 3675
},
{
"epoch": 2.0,
"learning_rate": 8.571428571428572e-07,
"loss": 0.2835,
"step": 3700
},
{
"epoch": 2.01,
"learning_rate": 7.857142857142857e-07,
"loss": 0.2625,
"step": 3725
},
{
"epoch": 2.02,
"learning_rate": 7.142857142857143e-07,
"loss": 0.2344,
"step": 3750
},
{
"epoch": 2.04,
"learning_rate": 6.428571428571428e-07,
"loss": 0.2297,
"step": 3775
},
{
"epoch": 2.05,
"learning_rate": 5.714285714285715e-07,
"loss": 0.2424,
"step": 3800
},
{
"epoch": 2.06,
"learning_rate": 5.000000000000001e-07,
"loss": 0.2417,
"step": 3825
},
{
"epoch": 2.08,
"learning_rate": 4.285714285714286e-07,
"loss": 0.2407,
"step": 3850
},
{
"epoch": 2.09,
"learning_rate": 3.5714285714285716e-07,
"loss": 0.2326,
"step": 3875
},
{
"epoch": 2.1,
"learning_rate": 2.8571428571428575e-07,
"loss": 0.2247,
"step": 3900
},
{
"epoch": 2.12,
"learning_rate": 2.142857142857143e-07,
"loss": 0.2334,
"step": 3925
},
{
"epoch": 2.13,
"learning_rate": 1.4285714285714287e-07,
"loss": 0.2205,
"step": 3950
},
{
"epoch": 2.14,
"learning_rate": 7.142857142857144e-08,
"loss": 0.2362,
"step": 3975
},
{
"epoch": 2.16,
"learning_rate": 0.0,
"loss": 0.249,
"step": 4000
},
{
"epoch": 2.16,
"eval_loss": 0.34158244729042053,
"eval_runtime": 3115.7686,
"eval_samples_per_second": 3.944,
"eval_steps_per_second": 0.493,
"eval_wer": 34.928488014365186,
"step": 4000
}
],
"logging_steps": 25,
"max_steps": 4000,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 1000,
"total_flos": 3.693258236215296e+19,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}