whisper-tiny-it-12 / checkpoint-4000 /trainer_state.json
GIanlucaRub's picture
Copied from Google Drive
158c81a
{
"best_metric": 42.24902939545203,
"best_model_checkpoint": "./whisper-tiny-it-12/checkpoint-4000",
"epoch": 1.5273004963726613,
"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": 4.0000000000000003e-07,
"loss": 3.7311,
"step": 25
},
{
"epoch": 0.02,
"learning_rate": 9.000000000000001e-07,
"loss": 3.2401,
"step": 50
},
{
"epoch": 0.03,
"learning_rate": 1.4000000000000001e-06,
"loss": 2.4666,
"step": 75
},
{
"epoch": 0.04,
"learning_rate": 1.9000000000000002e-06,
"loss": 1.7924,
"step": 100
},
{
"epoch": 0.05,
"learning_rate": 2.4000000000000003e-06,
"loss": 1.2235,
"step": 125
},
{
"epoch": 0.06,
"learning_rate": 2.9e-06,
"loss": 0.9162,
"step": 150
},
{
"epoch": 0.07,
"learning_rate": 3.4000000000000005e-06,
"loss": 0.8499,
"step": 175
},
{
"epoch": 0.08,
"learning_rate": 3.900000000000001e-06,
"loss": 0.8388,
"step": 200
},
{
"epoch": 0.09,
"learning_rate": 4.4e-06,
"loss": 0.7834,
"step": 225
},
{
"epoch": 0.1,
"learning_rate": 4.9000000000000005e-06,
"loss": 0.7666,
"step": 250
},
{
"epoch": 0.11,
"learning_rate": 5.400000000000001e-06,
"loss": 0.7621,
"step": 275
},
{
"epoch": 0.11,
"learning_rate": 5.9e-06,
"loss": 0.7067,
"step": 300
},
{
"epoch": 0.12,
"learning_rate": 6.4000000000000006e-06,
"loss": 0.745,
"step": 325
},
{
"epoch": 0.13,
"learning_rate": 6.9e-06,
"loss": 0.6709,
"step": 350
},
{
"epoch": 0.14,
"learning_rate": 7.4e-06,
"loss": 0.6614,
"step": 375
},
{
"epoch": 0.15,
"learning_rate": 7.9e-06,
"loss": 0.6952,
"step": 400
},
{
"epoch": 0.16,
"learning_rate": 8.400000000000001e-06,
"loss": 0.6711,
"step": 425
},
{
"epoch": 0.17,
"learning_rate": 8.900000000000001e-06,
"loss": 0.674,
"step": 450
},
{
"epoch": 0.18,
"learning_rate": 9.4e-06,
"loss": 0.631,
"step": 475
},
{
"epoch": 0.19,
"learning_rate": 9.9e-06,
"loss": 0.6322,
"step": 500
},
{
"epoch": 0.2,
"learning_rate": 9.942857142857145e-06,
"loss": 0.6427,
"step": 525
},
{
"epoch": 0.21,
"learning_rate": 9.871428571428572e-06,
"loss": 0.6419,
"step": 550
},
{
"epoch": 0.22,
"learning_rate": 9.800000000000001e-06,
"loss": 0.6484,
"step": 575
},
{
"epoch": 0.23,
"learning_rate": 9.72857142857143e-06,
"loss": 0.6228,
"step": 600
},
{
"epoch": 0.24,
"learning_rate": 9.657142857142859e-06,
"loss": 0.6652,
"step": 625
},
{
"epoch": 0.25,
"learning_rate": 9.585714285714286e-06,
"loss": 0.6331,
"step": 650
},
{
"epoch": 0.26,
"learning_rate": 9.514285714285715e-06,
"loss": 0.6281,
"step": 675
},
{
"epoch": 0.27,
"learning_rate": 9.442857142857144e-06,
"loss": 0.6164,
"step": 700
},
{
"epoch": 0.28,
"learning_rate": 9.371428571428572e-06,
"loss": 0.5798,
"step": 725
},
{
"epoch": 0.29,
"learning_rate": 9.3e-06,
"loss": 0.6095,
"step": 750
},
{
"epoch": 0.3,
"learning_rate": 9.22857142857143e-06,
"loss": 0.5941,
"step": 775
},
{
"epoch": 0.31,
"learning_rate": 9.157142857142857e-06,
"loss": 0.5803,
"step": 800
},
{
"epoch": 0.32,
"learning_rate": 9.085714285714286e-06,
"loss": 0.5709,
"step": 825
},
{
"epoch": 0.32,
"learning_rate": 9.014285714285715e-06,
"loss": 0.606,
"step": 850
},
{
"epoch": 0.33,
"learning_rate": 8.942857142857142e-06,
"loss": 0.5946,
"step": 875
},
{
"epoch": 0.34,
"learning_rate": 8.871428571428571e-06,
"loss": 0.5791,
"step": 900
},
{
"epoch": 0.35,
"learning_rate": 8.8e-06,
"loss": 0.6073,
"step": 925
},
{
"epoch": 0.36,
"learning_rate": 8.72857142857143e-06,
"loss": 0.5667,
"step": 950
},
{
"epoch": 0.37,
"learning_rate": 8.657142857142858e-06,
"loss": 0.5819,
"step": 975
},
{
"epoch": 0.38,
"learning_rate": 8.585714285714286e-06,
"loss": 0.5844,
"step": 1000
},
{
"epoch": 0.38,
"eval_loss": 0.8011854887008667,
"eval_runtime": 332.5485,
"eval_samples_per_second": 4.511,
"eval_steps_per_second": 0.565,
"eval_wer": 48.98779811425402,
"step": 1000
},
{
"epoch": 0.39,
"learning_rate": 8.514285714285715e-06,
"loss": 0.5654,
"step": 1025
},
{
"epoch": 0.4,
"learning_rate": 8.442857142857144e-06,
"loss": 0.5665,
"step": 1050
},
{
"epoch": 0.41,
"learning_rate": 8.371428571428573e-06,
"loss": 0.549,
"step": 1075
},
{
"epoch": 0.42,
"learning_rate": 8.3e-06,
"loss": 0.5919,
"step": 1100
},
{
"epoch": 0.43,
"learning_rate": 8.22857142857143e-06,
"loss": 0.5713,
"step": 1125
},
{
"epoch": 0.44,
"learning_rate": 8.157142857142858e-06,
"loss": 0.5451,
"step": 1150
},
{
"epoch": 0.45,
"learning_rate": 8.085714285714287e-06,
"loss": 0.5698,
"step": 1175
},
{
"epoch": 0.46,
"learning_rate": 8.014285714285715e-06,
"loss": 0.558,
"step": 1200
},
{
"epoch": 0.47,
"learning_rate": 7.942857142857144e-06,
"loss": 0.5724,
"step": 1225
},
{
"epoch": 0.48,
"learning_rate": 7.871428571428573e-06,
"loss": 0.5566,
"step": 1250
},
{
"epoch": 0.49,
"learning_rate": 7.800000000000002e-06,
"loss": 0.5568,
"step": 1275
},
{
"epoch": 0.5,
"learning_rate": 7.72857142857143e-06,
"loss": 0.5443,
"step": 1300
},
{
"epoch": 0.51,
"learning_rate": 7.657142857142858e-06,
"loss": 0.5527,
"step": 1325
},
{
"epoch": 0.52,
"learning_rate": 7.585714285714286e-06,
"loss": 0.5313,
"step": 1350
},
{
"epoch": 0.53,
"learning_rate": 7.514285714285715e-06,
"loss": 0.4991,
"step": 1375
},
{
"epoch": 0.53,
"learning_rate": 7.442857142857144e-06,
"loss": 0.5445,
"step": 1400
},
{
"epoch": 0.54,
"learning_rate": 7.371428571428571e-06,
"loss": 0.5457,
"step": 1425
},
{
"epoch": 0.55,
"learning_rate": 7.3e-06,
"loss": 0.5201,
"step": 1450
},
{
"epoch": 0.56,
"learning_rate": 7.2285714285714294e-06,
"loss": 0.5538,
"step": 1475
},
{
"epoch": 0.57,
"learning_rate": 7.1571428571428584e-06,
"loss": 0.5259,
"step": 1500
},
{
"epoch": 0.58,
"learning_rate": 7.085714285714286e-06,
"loss": 0.5295,
"step": 1525
},
{
"epoch": 0.59,
"learning_rate": 7.014285714285715e-06,
"loss": 0.5038,
"step": 1550
},
{
"epoch": 0.6,
"learning_rate": 6.942857142857144e-06,
"loss": 0.53,
"step": 1575
},
{
"epoch": 0.61,
"learning_rate": 6.871428571428572e-06,
"loss": 0.5386,
"step": 1600
},
{
"epoch": 0.62,
"learning_rate": 6.800000000000001e-06,
"loss": 0.5218,
"step": 1625
},
{
"epoch": 0.63,
"learning_rate": 6.728571428571429e-06,
"loss": 0.5296,
"step": 1650
},
{
"epoch": 0.64,
"learning_rate": 6.657142857142857e-06,
"loss": 0.5052,
"step": 1675
},
{
"epoch": 0.65,
"learning_rate": 6.585714285714286e-06,
"loss": 0.5032,
"step": 1700
},
{
"epoch": 0.66,
"learning_rate": 6.514285714285715e-06,
"loss": 0.4859,
"step": 1725
},
{
"epoch": 0.67,
"learning_rate": 6.442857142857143e-06,
"loss": 0.5267,
"step": 1750
},
{
"epoch": 0.68,
"learning_rate": 6.371428571428572e-06,
"loss": 0.5363,
"step": 1775
},
{
"epoch": 0.69,
"learning_rate": 6.300000000000001e-06,
"loss": 0.5312,
"step": 1800
},
{
"epoch": 0.7,
"learning_rate": 6.22857142857143e-06,
"loss": 0.531,
"step": 1825
},
{
"epoch": 0.71,
"learning_rate": 6.157142857142858e-06,
"loss": 0.5172,
"step": 1850
},
{
"epoch": 0.72,
"learning_rate": 6.085714285714286e-06,
"loss": 0.5018,
"step": 1875
},
{
"epoch": 0.73,
"learning_rate": 6.014285714285715e-06,
"loss": 0.4923,
"step": 1900
},
{
"epoch": 0.74,
"learning_rate": 5.942857142857143e-06,
"loss": 0.5049,
"step": 1925
},
{
"epoch": 0.74,
"learning_rate": 5.871428571428572e-06,
"loss": 0.4974,
"step": 1950
},
{
"epoch": 0.75,
"learning_rate": 5.8e-06,
"loss": 0.5182,
"step": 1975
},
{
"epoch": 0.76,
"learning_rate": 5.7285714285714285e-06,
"loss": 0.4961,
"step": 2000
},
{
"epoch": 0.76,
"eval_loss": 0.7139552235603333,
"eval_runtime": 342.7202,
"eval_samples_per_second": 4.377,
"eval_steps_per_second": 0.549,
"eval_wer": 46.32556849694953,
"step": 2000
},
{
"epoch": 0.77,
"learning_rate": 5.6571428571428576e-06,
"loss": 0.501,
"step": 2025
},
{
"epoch": 0.78,
"learning_rate": 5.5857142857142866e-06,
"loss": 0.4892,
"step": 2050
},
{
"epoch": 0.79,
"learning_rate": 5.514285714285714e-06,
"loss": 0.4988,
"step": 2075
},
{
"epoch": 0.8,
"learning_rate": 5.442857142857143e-06,
"loss": 0.476,
"step": 2100
},
{
"epoch": 0.81,
"learning_rate": 5.371428571428572e-06,
"loss": 0.5241,
"step": 2125
},
{
"epoch": 0.82,
"learning_rate": 5.300000000000001e-06,
"loss": 0.4917,
"step": 2150
},
{
"epoch": 0.83,
"learning_rate": 5.22857142857143e-06,
"loss": 0.495,
"step": 2175
},
{
"epoch": 0.84,
"learning_rate": 5.157142857142857e-06,
"loss": 0.4908,
"step": 2200
},
{
"epoch": 0.85,
"learning_rate": 5.085714285714286e-06,
"loss": 0.497,
"step": 2225
},
{
"epoch": 0.86,
"learning_rate": 5.014285714285715e-06,
"loss": 0.5083,
"step": 2250
},
{
"epoch": 0.87,
"learning_rate": 4.9428571428571435e-06,
"loss": 0.4824,
"step": 2275
},
{
"epoch": 0.88,
"learning_rate": 4.871428571428572e-06,
"loss": 0.5031,
"step": 2300
},
{
"epoch": 0.89,
"learning_rate": 4.800000000000001e-06,
"loss": 0.5148,
"step": 2325
},
{
"epoch": 0.9,
"learning_rate": 4.728571428571429e-06,
"loss": 0.5205,
"step": 2350
},
{
"epoch": 0.91,
"learning_rate": 4.657142857142857e-06,
"loss": 0.4803,
"step": 2375
},
{
"epoch": 0.92,
"learning_rate": 4.585714285714286e-06,
"loss": 0.5111,
"step": 2400
},
{
"epoch": 0.93,
"learning_rate": 4.514285714285714e-06,
"loss": 0.5055,
"step": 2425
},
{
"epoch": 0.94,
"learning_rate": 4.442857142857143e-06,
"loss": 0.4817,
"step": 2450
},
{
"epoch": 0.95,
"learning_rate": 4.371428571428572e-06,
"loss": 0.5025,
"step": 2475
},
{
"epoch": 0.95,
"learning_rate": 4.3e-06,
"loss": 0.5122,
"step": 2500
},
{
"epoch": 0.96,
"learning_rate": 4.228571428571429e-06,
"loss": 0.4851,
"step": 2525
},
{
"epoch": 0.97,
"learning_rate": 4.1571428571428575e-06,
"loss": 0.477,
"step": 2550
},
{
"epoch": 0.98,
"learning_rate": 4.0857142857142865e-06,
"loss": 0.4698,
"step": 2575
},
{
"epoch": 0.99,
"learning_rate": 4.014285714285715e-06,
"loss": 0.4762,
"step": 2600
},
{
"epoch": 1.0,
"learning_rate": 3.942857142857143e-06,
"loss": 0.4402,
"step": 2625
},
{
"epoch": 1.01,
"learning_rate": 3.871428571428572e-06,
"loss": 0.4024,
"step": 2650
},
{
"epoch": 1.02,
"learning_rate": 3.8000000000000005e-06,
"loss": 0.3709,
"step": 2675
},
{
"epoch": 1.03,
"learning_rate": 3.7285714285714286e-06,
"loss": 0.4004,
"step": 2700
},
{
"epoch": 1.04,
"learning_rate": 3.6571428571428576e-06,
"loss": 0.3876,
"step": 2725
},
{
"epoch": 1.05,
"learning_rate": 3.5857142857142862e-06,
"loss": 0.4016,
"step": 2750
},
{
"epoch": 1.06,
"learning_rate": 3.5142857142857144e-06,
"loss": 0.394,
"step": 2775
},
{
"epoch": 1.07,
"learning_rate": 3.4428571428571434e-06,
"loss": 0.3938,
"step": 2800
},
{
"epoch": 1.08,
"learning_rate": 3.3714285714285716e-06,
"loss": 0.3993,
"step": 2825
},
{
"epoch": 1.09,
"learning_rate": 3.3000000000000006e-06,
"loss": 0.381,
"step": 2850
},
{
"epoch": 1.1,
"learning_rate": 3.2285714285714288e-06,
"loss": 0.3793,
"step": 2875
},
{
"epoch": 1.11,
"learning_rate": 3.1571428571428573e-06,
"loss": 0.3875,
"step": 2900
},
{
"epoch": 1.12,
"learning_rate": 3.085714285714286e-06,
"loss": 0.363,
"step": 2925
},
{
"epoch": 1.13,
"learning_rate": 3.0142857142857145e-06,
"loss": 0.428,
"step": 2950
},
{
"epoch": 1.14,
"learning_rate": 2.9428571428571427e-06,
"loss": 0.3847,
"step": 2975
},
{
"epoch": 1.15,
"learning_rate": 2.8714285714285717e-06,
"loss": 0.4059,
"step": 3000
},
{
"epoch": 1.15,
"eval_loss": 0.680098295211792,
"eval_runtime": 337.3888,
"eval_samples_per_second": 4.446,
"eval_steps_per_second": 0.557,
"eval_wer": 43.254298391569606,
"step": 3000
},
{
"epoch": 1.16,
"learning_rate": 2.8000000000000003e-06,
"loss": 0.3816,
"step": 3025
},
{
"epoch": 1.16,
"learning_rate": 2.728571428571429e-06,
"loss": 0.3894,
"step": 3050
},
{
"epoch": 1.17,
"learning_rate": 2.6571428571428575e-06,
"loss": 0.4049,
"step": 3075
},
{
"epoch": 1.18,
"learning_rate": 2.5857142857142856e-06,
"loss": 0.3724,
"step": 3100
},
{
"epoch": 1.19,
"learning_rate": 2.5142857142857147e-06,
"loss": 0.3742,
"step": 3125
},
{
"epoch": 1.2,
"learning_rate": 2.442857142857143e-06,
"loss": 0.3796,
"step": 3150
},
{
"epoch": 1.21,
"learning_rate": 2.371428571428572e-06,
"loss": 0.3844,
"step": 3175
},
{
"epoch": 1.22,
"learning_rate": 2.3000000000000004e-06,
"loss": 0.3897,
"step": 3200
},
{
"epoch": 1.23,
"learning_rate": 2.228571428571429e-06,
"loss": 0.3847,
"step": 3225
},
{
"epoch": 1.24,
"learning_rate": 2.157142857142857e-06,
"loss": 0.4006,
"step": 3250
},
{
"epoch": 1.25,
"learning_rate": 2.0857142857142858e-06,
"loss": 0.4062,
"step": 3275
},
{
"epoch": 1.26,
"learning_rate": 2.0142857142857144e-06,
"loss": 0.3912,
"step": 3300
},
{
"epoch": 1.27,
"learning_rate": 1.942857142857143e-06,
"loss": 0.368,
"step": 3325
},
{
"epoch": 1.28,
"learning_rate": 1.8714285714285715e-06,
"loss": 0.3848,
"step": 3350
},
{
"epoch": 1.29,
"learning_rate": 1.8000000000000001e-06,
"loss": 0.3769,
"step": 3375
},
{
"epoch": 1.3,
"learning_rate": 1.7285714285714287e-06,
"loss": 0.3993,
"step": 3400
},
{
"epoch": 1.31,
"learning_rate": 1.657142857142857e-06,
"loss": 0.3846,
"step": 3425
},
{
"epoch": 1.32,
"learning_rate": 1.5857142857142857e-06,
"loss": 0.3898,
"step": 3450
},
{
"epoch": 1.33,
"learning_rate": 1.5142857142857145e-06,
"loss": 0.3715,
"step": 3475
},
{
"epoch": 1.34,
"learning_rate": 1.442857142857143e-06,
"loss": 0.3871,
"step": 3500
},
{
"epoch": 1.35,
"learning_rate": 1.3714285714285717e-06,
"loss": 0.3985,
"step": 3525
},
{
"epoch": 1.36,
"learning_rate": 1.3e-06,
"loss": 0.4056,
"step": 3550
},
{
"epoch": 1.37,
"learning_rate": 1.2285714285714286e-06,
"loss": 0.3898,
"step": 3575
},
{
"epoch": 1.37,
"learning_rate": 1.1571428571428572e-06,
"loss": 0.3924,
"step": 3600
},
{
"epoch": 1.38,
"learning_rate": 1.0857142857142858e-06,
"loss": 0.3571,
"step": 3625
},
{
"epoch": 1.39,
"learning_rate": 1.0142857142857144e-06,
"loss": 0.3998,
"step": 3650
},
{
"epoch": 1.4,
"learning_rate": 9.42857142857143e-07,
"loss": 0.3886,
"step": 3675
},
{
"epoch": 1.41,
"learning_rate": 8.714285714285716e-07,
"loss": 0.3789,
"step": 3700
},
{
"epoch": 1.42,
"learning_rate": 8.000000000000001e-07,
"loss": 0.3854,
"step": 3725
},
{
"epoch": 1.43,
"learning_rate": 7.285714285714287e-07,
"loss": 0.384,
"step": 3750
},
{
"epoch": 1.44,
"learning_rate": 6.571428571428571e-07,
"loss": 0.3721,
"step": 3775
},
{
"epoch": 1.45,
"learning_rate": 5.857142857142857e-07,
"loss": 0.4092,
"step": 3800
},
{
"epoch": 1.46,
"learning_rate": 5.142857142857143e-07,
"loss": 0.3684,
"step": 3825
},
{
"epoch": 1.47,
"learning_rate": 4.4285714285714286e-07,
"loss": 0.3723,
"step": 3850
},
{
"epoch": 1.48,
"learning_rate": 3.7142857142857145e-07,
"loss": 0.3805,
"step": 3875
},
{
"epoch": 1.49,
"learning_rate": 3.0000000000000004e-07,
"loss": 0.3733,
"step": 3900
},
{
"epoch": 1.5,
"learning_rate": 2.285714285714286e-07,
"loss": 0.4051,
"step": 3925
},
{
"epoch": 1.51,
"learning_rate": 1.5714285714285717e-07,
"loss": 0.3981,
"step": 3950
},
{
"epoch": 1.52,
"learning_rate": 8.571428571428573e-08,
"loss": 0.3732,
"step": 3975
},
{
"epoch": 1.53,
"learning_rate": 1.4285714285714288e-08,
"loss": 0.402,
"step": 4000
},
{
"epoch": 1.53,
"eval_loss": 0.6701788902282715,
"eval_runtime": 339.575,
"eval_samples_per_second": 4.417,
"eval_steps_per_second": 0.554,
"eval_wer": 42.24902939545203,
"step": 4000
}
],
"max_steps": 4000,
"num_train_epochs": 2,
"total_flos": 1.57553479102464e+18,
"trial_name": null,
"trial_params": null
}