whisper-small-ar / checkpoint-4000 /trainer_state.json
MMars's picture
Upload 53 files
cba1d2e
raw
history blame
20.5 kB
{
"best_metric": 44.976586172068686,
"best_model_checkpoint": "./whisper-small-ar/whisper-small-ar/checkpoint-4000",
"epoch": 1.6625103906899419,
"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": 4.2000000000000006e-07,
"loss": 3.262,
"step": 25
},
{
"epoch": 0.02,
"learning_rate": 9.200000000000001e-07,
"loss": 2.5217,
"step": 50
},
{
"epoch": 0.03,
"learning_rate": 1.42e-06,
"loss": 1.7958,
"step": 75
},
{
"epoch": 0.04,
"learning_rate": 1.9200000000000003e-06,
"loss": 1.2313,
"step": 100
},
{
"epoch": 0.05,
"learning_rate": 2.42e-06,
"loss": 1.0969,
"step": 125
},
{
"epoch": 0.06,
"learning_rate": 2.92e-06,
"loss": 0.9611,
"step": 150
},
{
"epoch": 0.07,
"learning_rate": 3.4200000000000007e-06,
"loss": 0.9175,
"step": 175
},
{
"epoch": 0.08,
"learning_rate": 3.920000000000001e-06,
"loss": 0.819,
"step": 200
},
{
"epoch": 0.09,
"learning_rate": 4.42e-06,
"loss": 0.7742,
"step": 225
},
{
"epoch": 0.1,
"learning_rate": 4.92e-06,
"loss": 0.6782,
"step": 250
},
{
"epoch": 0.11,
"learning_rate": 5.420000000000001e-06,
"loss": 0.5778,
"step": 275
},
{
"epoch": 0.12,
"learning_rate": 5.92e-06,
"loss": 0.456,
"step": 300
},
{
"epoch": 0.14,
"learning_rate": 6.42e-06,
"loss": 0.4114,
"step": 325
},
{
"epoch": 0.15,
"learning_rate": 6.92e-06,
"loss": 0.4197,
"step": 350
},
{
"epoch": 0.16,
"learning_rate": 7.420000000000001e-06,
"loss": 0.3929,
"step": 375
},
{
"epoch": 0.17,
"learning_rate": 7.92e-06,
"loss": 0.4104,
"step": 400
},
{
"epoch": 0.18,
"learning_rate": 8.42e-06,
"loss": 0.4269,
"step": 425
},
{
"epoch": 0.19,
"learning_rate": 8.920000000000001e-06,
"loss": 0.4303,
"step": 450
},
{
"epoch": 0.2,
"learning_rate": 9.42e-06,
"loss": 0.3523,
"step": 475
},
{
"epoch": 0.21,
"learning_rate": 9.920000000000002e-06,
"loss": 0.3633,
"step": 500
},
{
"epoch": 0.22,
"learning_rate": 9.940000000000001e-06,
"loss": 0.361,
"step": 525
},
{
"epoch": 0.23,
"learning_rate": 9.86857142857143e-06,
"loss": 0.3368,
"step": 550
},
{
"epoch": 0.24,
"learning_rate": 9.797142857142858e-06,
"loss": 0.329,
"step": 575
},
{
"epoch": 0.25,
"learning_rate": 9.725714285714287e-06,
"loss": 0.3671,
"step": 600
},
{
"epoch": 0.26,
"learning_rate": 9.654285714285716e-06,
"loss": 0.3832,
"step": 625
},
{
"epoch": 0.27,
"learning_rate": 9.582857142857143e-06,
"loss": 0.3494,
"step": 650
},
{
"epoch": 0.28,
"learning_rate": 9.511428571428572e-06,
"loss": 0.3334,
"step": 675
},
{
"epoch": 0.29,
"learning_rate": 9.440000000000001e-06,
"loss": 0.3466,
"step": 700
},
{
"epoch": 0.3,
"learning_rate": 9.368571428571428e-06,
"loss": 0.3464,
"step": 725
},
{
"epoch": 0.31,
"learning_rate": 9.297142857142857e-06,
"loss": 0.3481,
"step": 750
},
{
"epoch": 0.32,
"learning_rate": 9.225714285714286e-06,
"loss": 0.3188,
"step": 775
},
{
"epoch": 0.33,
"learning_rate": 9.154285714285715e-06,
"loss": 0.3396,
"step": 800
},
{
"epoch": 0.34,
"learning_rate": 9.082857142857143e-06,
"loss": 0.3187,
"step": 825
},
{
"epoch": 0.35,
"learning_rate": 9.011428571428572e-06,
"loss": 0.3079,
"step": 850
},
{
"epoch": 0.36,
"learning_rate": 8.94e-06,
"loss": 0.311,
"step": 875
},
{
"epoch": 0.37,
"learning_rate": 8.86857142857143e-06,
"loss": 0.3162,
"step": 900
},
{
"epoch": 0.38,
"learning_rate": 8.797142857142857e-06,
"loss": 0.3474,
"step": 925
},
{
"epoch": 0.39,
"learning_rate": 8.725714285714286e-06,
"loss": 0.3154,
"step": 950
},
{
"epoch": 0.41,
"learning_rate": 8.654285714285715e-06,
"loss": 0.3315,
"step": 975
},
{
"epoch": 0.42,
"learning_rate": 8.582857142857144e-06,
"loss": 0.2811,
"step": 1000
},
{
"epoch": 0.42,
"eval_loss": 0.3930181860923767,
"eval_runtime": 2355.0366,
"eval_samples_per_second": 4.433,
"eval_steps_per_second": 0.554,
"eval_wer": 53.77834909558351,
"step": 1000
},
{
"epoch": 0.43,
"learning_rate": 8.511428571428571e-06,
"loss": 0.2884,
"step": 1025
},
{
"epoch": 0.44,
"learning_rate": 8.44e-06,
"loss": 0.3166,
"step": 1050
},
{
"epoch": 0.45,
"learning_rate": 8.36857142857143e-06,
"loss": 0.2723,
"step": 1075
},
{
"epoch": 0.46,
"learning_rate": 8.297142857142859e-06,
"loss": 0.2841,
"step": 1100
},
{
"epoch": 0.47,
"learning_rate": 8.225714285714288e-06,
"loss": 0.3115,
"step": 1125
},
{
"epoch": 0.48,
"learning_rate": 8.154285714285715e-06,
"loss": 0.3161,
"step": 1150
},
{
"epoch": 0.49,
"learning_rate": 8.082857142857144e-06,
"loss": 0.2771,
"step": 1175
},
{
"epoch": 0.5,
"learning_rate": 8.011428571428573e-06,
"loss": 0.2826,
"step": 1200
},
{
"epoch": 0.51,
"learning_rate": 7.94e-06,
"loss": 0.3057,
"step": 1225
},
{
"epoch": 0.52,
"learning_rate": 7.86857142857143e-06,
"loss": 0.2849,
"step": 1250
},
{
"epoch": 0.53,
"learning_rate": 7.797142857142858e-06,
"loss": 0.2959,
"step": 1275
},
{
"epoch": 0.54,
"learning_rate": 7.725714285714286e-06,
"loss": 0.3015,
"step": 1300
},
{
"epoch": 0.55,
"learning_rate": 7.654285714285715e-06,
"loss": 0.2686,
"step": 1325
},
{
"epoch": 0.56,
"learning_rate": 7.5828571428571444e-06,
"loss": 0.2554,
"step": 1350
},
{
"epoch": 0.57,
"learning_rate": 7.511428571428572e-06,
"loss": 0.3009,
"step": 1375
},
{
"epoch": 0.58,
"learning_rate": 7.440000000000001e-06,
"loss": 0.2896,
"step": 1400
},
{
"epoch": 0.59,
"learning_rate": 7.36857142857143e-06,
"loss": 0.271,
"step": 1425
},
{
"epoch": 0.6,
"learning_rate": 7.297142857142858e-06,
"loss": 0.2695,
"step": 1450
},
{
"epoch": 0.61,
"learning_rate": 7.225714285714286e-06,
"loss": 0.287,
"step": 1475
},
{
"epoch": 0.62,
"learning_rate": 7.154285714285715e-06,
"loss": 0.2473,
"step": 1500
},
{
"epoch": 0.63,
"learning_rate": 7.082857142857143e-06,
"loss": 0.2454,
"step": 1525
},
{
"epoch": 0.64,
"learning_rate": 7.011428571428572e-06,
"loss": 0.2952,
"step": 1550
},
{
"epoch": 0.65,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.2765,
"step": 1575
},
{
"epoch": 0.67,
"learning_rate": 6.868571428571429e-06,
"loss": 0.2615,
"step": 1600
},
{
"epoch": 0.68,
"learning_rate": 6.797142857142858e-06,
"loss": 0.26,
"step": 1625
},
{
"epoch": 0.69,
"learning_rate": 6.725714285714287e-06,
"loss": 0.2687,
"step": 1650
},
{
"epoch": 0.7,
"learning_rate": 6.654285714285716e-06,
"loss": 0.2568,
"step": 1675
},
{
"epoch": 0.71,
"learning_rate": 6.582857142857143e-06,
"loss": 0.2634,
"step": 1700
},
{
"epoch": 0.72,
"learning_rate": 6.511428571428572e-06,
"loss": 0.2655,
"step": 1725
},
{
"epoch": 0.73,
"learning_rate": 6.440000000000001e-06,
"loss": 0.2457,
"step": 1750
},
{
"epoch": 0.74,
"learning_rate": 6.368571428571429e-06,
"loss": 0.2427,
"step": 1775
},
{
"epoch": 0.75,
"learning_rate": 6.297142857142857e-06,
"loss": 0.236,
"step": 1800
},
{
"epoch": 0.76,
"learning_rate": 6.225714285714286e-06,
"loss": 0.2553,
"step": 1825
},
{
"epoch": 0.77,
"learning_rate": 6.1542857142857145e-06,
"loss": 0.2521,
"step": 1850
},
{
"epoch": 0.78,
"learning_rate": 6.0828571428571435e-06,
"loss": 0.2536,
"step": 1875
},
{
"epoch": 0.79,
"learning_rate": 6.011428571428572e-06,
"loss": 0.2696,
"step": 1900
},
{
"epoch": 0.8,
"learning_rate": 5.94e-06,
"loss": 0.2436,
"step": 1925
},
{
"epoch": 0.81,
"learning_rate": 5.868571428571429e-06,
"loss": 0.2197,
"step": 1950
},
{
"epoch": 0.82,
"learning_rate": 5.797142857142858e-06,
"loss": 0.2526,
"step": 1975
},
{
"epoch": 0.83,
"learning_rate": 5.725714285714287e-06,
"loss": 0.2356,
"step": 2000
},
{
"epoch": 0.83,
"eval_loss": 0.3487941324710846,
"eval_runtime": 2297.3957,
"eval_samples_per_second": 4.544,
"eval_steps_per_second": 0.568,
"eval_wer": 47.79359103847213,
"step": 2000
},
{
"epoch": 0.84,
"learning_rate": 5.654285714285714e-06,
"loss": 0.2516,
"step": 2025
},
{
"epoch": 0.85,
"learning_rate": 5.582857142857143e-06,
"loss": 0.2467,
"step": 2050
},
{
"epoch": 0.86,
"learning_rate": 5.511428571428572e-06,
"loss": 0.23,
"step": 2075
},
{
"epoch": 0.87,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.2571,
"step": 2100
},
{
"epoch": 0.88,
"learning_rate": 5.368571428571429e-06,
"loss": 0.2424,
"step": 2125
},
{
"epoch": 0.89,
"learning_rate": 5.297142857142858e-06,
"loss": 0.2402,
"step": 2150
},
{
"epoch": 0.9,
"learning_rate": 5.225714285714286e-06,
"loss": 0.2466,
"step": 2175
},
{
"epoch": 0.91,
"learning_rate": 5.154285714285715e-06,
"loss": 0.2415,
"step": 2200
},
{
"epoch": 0.92,
"learning_rate": 5.082857142857144e-06,
"loss": 0.2228,
"step": 2225
},
{
"epoch": 0.94,
"learning_rate": 5.011428571428571e-06,
"loss": 0.2359,
"step": 2250
},
{
"epoch": 0.95,
"learning_rate": 4.94e-06,
"loss": 0.237,
"step": 2275
},
{
"epoch": 0.96,
"learning_rate": 4.868571428571429e-06,
"loss": 0.232,
"step": 2300
},
{
"epoch": 0.97,
"learning_rate": 4.797142857142857e-06,
"loss": 0.246,
"step": 2325
},
{
"epoch": 0.98,
"learning_rate": 4.725714285714286e-06,
"loss": 0.2365,
"step": 2350
},
{
"epoch": 0.99,
"learning_rate": 4.6542857142857145e-06,
"loss": 0.2492,
"step": 2375
},
{
"epoch": 1.0,
"learning_rate": 4.5828571428571435e-06,
"loss": 0.2309,
"step": 2400
},
{
"epoch": 1.01,
"learning_rate": 4.514285714285714e-06,
"loss": 0.2007,
"step": 2425
},
{
"epoch": 1.02,
"learning_rate": 4.442857142857143e-06,
"loss": 0.1757,
"step": 2450
},
{
"epoch": 1.03,
"learning_rate": 4.371428571428572e-06,
"loss": 0.1718,
"step": 2475
},
{
"epoch": 1.04,
"learning_rate": 4.3e-06,
"loss": 0.1514,
"step": 2500
},
{
"epoch": 1.05,
"learning_rate": 4.228571428571429e-06,
"loss": 0.1563,
"step": 2525
},
{
"epoch": 1.06,
"learning_rate": 4.1571428571428575e-06,
"loss": 0.1723,
"step": 2550
},
{
"epoch": 1.07,
"learning_rate": 4.0857142857142865e-06,
"loss": 0.163,
"step": 2575
},
{
"epoch": 1.08,
"learning_rate": 4.014285714285715e-06,
"loss": 0.1618,
"step": 2600
},
{
"epoch": 1.09,
"learning_rate": 3.942857142857143e-06,
"loss": 0.1746,
"step": 2625
},
{
"epoch": 1.1,
"learning_rate": 3.871428571428572e-06,
"loss": 0.1623,
"step": 2650
},
{
"epoch": 1.11,
"learning_rate": 3.8000000000000005e-06,
"loss": 0.1673,
"step": 2675
},
{
"epoch": 1.12,
"learning_rate": 3.7285714285714286e-06,
"loss": 0.1554,
"step": 2700
},
{
"epoch": 1.13,
"learning_rate": 3.6571428571428576e-06,
"loss": 0.1571,
"step": 2725
},
{
"epoch": 1.14,
"learning_rate": 3.5857142857142862e-06,
"loss": 0.1643,
"step": 2750
},
{
"epoch": 1.15,
"learning_rate": 3.5142857142857144e-06,
"loss": 0.1504,
"step": 2775
},
{
"epoch": 1.16,
"learning_rate": 3.4428571428571434e-06,
"loss": 0.1582,
"step": 2800
},
{
"epoch": 1.17,
"learning_rate": 3.3714285714285716e-06,
"loss": 0.1427,
"step": 2825
},
{
"epoch": 1.18,
"learning_rate": 3.3000000000000006e-06,
"loss": 0.1559,
"step": 2850
},
{
"epoch": 1.19,
"learning_rate": 3.2285714285714288e-06,
"loss": 0.1617,
"step": 2875
},
{
"epoch": 1.21,
"learning_rate": 3.1571428571428573e-06,
"loss": 0.156,
"step": 2900
},
{
"epoch": 1.22,
"learning_rate": 3.085714285714286e-06,
"loss": 0.1486,
"step": 2925
},
{
"epoch": 1.23,
"learning_rate": 3.0142857142857145e-06,
"loss": 0.1604,
"step": 2950
},
{
"epoch": 1.24,
"learning_rate": 2.9428571428571427e-06,
"loss": 0.1587,
"step": 2975
},
{
"epoch": 1.25,
"learning_rate": 2.8714285714285717e-06,
"loss": 0.1705,
"step": 3000
},
{
"epoch": 1.25,
"eval_loss": 0.3322072923183441,
"eval_runtime": 2282.7387,
"eval_samples_per_second": 4.573,
"eval_steps_per_second": 0.572,
"eval_wer": 45.75888348177394,
"step": 3000
},
{
"epoch": 1.26,
"learning_rate": 2.8000000000000003e-06,
"loss": 0.1556,
"step": 3025
},
{
"epoch": 1.27,
"learning_rate": 2.728571428571429e-06,
"loss": 0.1552,
"step": 3050
},
{
"epoch": 1.28,
"learning_rate": 2.6571428571428575e-06,
"loss": 0.1631,
"step": 3075
},
{
"epoch": 1.29,
"learning_rate": 2.5857142857142856e-06,
"loss": 0.1511,
"step": 3100
},
{
"epoch": 1.3,
"learning_rate": 2.5142857142857147e-06,
"loss": 0.1587,
"step": 3125
},
{
"epoch": 1.31,
"learning_rate": 2.442857142857143e-06,
"loss": 0.1544,
"step": 3150
},
{
"epoch": 1.32,
"learning_rate": 2.371428571428572e-06,
"loss": 0.1549,
"step": 3175
},
{
"epoch": 1.33,
"learning_rate": 2.3000000000000004e-06,
"loss": 0.1451,
"step": 3200
},
{
"epoch": 1.34,
"learning_rate": 2.228571428571429e-06,
"loss": 0.1467,
"step": 3225
},
{
"epoch": 1.35,
"learning_rate": 2.157142857142857e-06,
"loss": 0.1543,
"step": 3250
},
{
"epoch": 1.36,
"learning_rate": 2.0857142857142858e-06,
"loss": 0.1642,
"step": 3275
},
{
"epoch": 1.37,
"learning_rate": 2.0142857142857144e-06,
"loss": 0.1514,
"step": 3300
},
{
"epoch": 1.38,
"learning_rate": 1.942857142857143e-06,
"loss": 0.1405,
"step": 3325
},
{
"epoch": 1.39,
"learning_rate": 1.8714285714285715e-06,
"loss": 0.1615,
"step": 3350
},
{
"epoch": 1.4,
"learning_rate": 1.8000000000000001e-06,
"loss": 0.1473,
"step": 3375
},
{
"epoch": 1.41,
"learning_rate": 1.7285714285714287e-06,
"loss": 0.1626,
"step": 3400
},
{
"epoch": 1.42,
"learning_rate": 1.657142857142857e-06,
"loss": 0.1577,
"step": 3425
},
{
"epoch": 1.43,
"learning_rate": 1.5857142857142857e-06,
"loss": 0.1629,
"step": 3450
},
{
"epoch": 1.44,
"learning_rate": 1.5142857142857145e-06,
"loss": 0.1587,
"step": 3475
},
{
"epoch": 1.45,
"learning_rate": 1.442857142857143e-06,
"loss": 0.1451,
"step": 3500
},
{
"epoch": 1.47,
"learning_rate": 1.3714285714285717e-06,
"loss": 0.1485,
"step": 3525
},
{
"epoch": 1.48,
"learning_rate": 1.3e-06,
"loss": 0.1477,
"step": 3550
},
{
"epoch": 1.49,
"learning_rate": 1.2285714285714286e-06,
"loss": 0.154,
"step": 3575
},
{
"epoch": 1.5,
"learning_rate": 1.1571428571428572e-06,
"loss": 0.149,
"step": 3600
},
{
"epoch": 1.51,
"learning_rate": 1.0857142857142858e-06,
"loss": 0.1551,
"step": 3625
},
{
"epoch": 1.52,
"learning_rate": 1.0142857142857144e-06,
"loss": 0.1613,
"step": 3650
},
{
"epoch": 1.53,
"learning_rate": 9.42857142857143e-07,
"loss": 0.1529,
"step": 3675
},
{
"epoch": 1.54,
"learning_rate": 8.714285714285716e-07,
"loss": 0.1574,
"step": 3700
},
{
"epoch": 1.55,
"learning_rate": 8.000000000000001e-07,
"loss": 0.1534,
"step": 3725
},
{
"epoch": 1.56,
"learning_rate": 7.285714285714287e-07,
"loss": 0.148,
"step": 3750
},
{
"epoch": 1.57,
"learning_rate": 6.571428571428571e-07,
"loss": 0.1441,
"step": 3775
},
{
"epoch": 1.58,
"learning_rate": 5.857142857142857e-07,
"loss": 0.1518,
"step": 3800
},
{
"epoch": 1.59,
"learning_rate": 5.142857142857143e-07,
"loss": 0.1447,
"step": 3825
},
{
"epoch": 1.6,
"learning_rate": 4.4285714285714286e-07,
"loss": 0.153,
"step": 3850
},
{
"epoch": 1.61,
"learning_rate": 3.7142857142857145e-07,
"loss": 0.1543,
"step": 3875
},
{
"epoch": 1.62,
"learning_rate": 3.0000000000000004e-07,
"loss": 0.1404,
"step": 3900
},
{
"epoch": 1.63,
"learning_rate": 2.285714285714286e-07,
"loss": 0.1513,
"step": 3925
},
{
"epoch": 1.64,
"learning_rate": 1.5714285714285717e-07,
"loss": 0.1519,
"step": 3950
},
{
"epoch": 1.65,
"learning_rate": 8.571428571428573e-08,
"loss": 0.1468,
"step": 3975
},
{
"epoch": 1.66,
"learning_rate": 1.4285714285714288e-08,
"loss": 0.1476,
"step": 4000
},
{
"epoch": 1.66,
"eval_loss": 0.3225502073764801,
"eval_runtime": 2316.4817,
"eval_samples_per_second": 4.507,
"eval_steps_per_second": 0.563,
"eval_wer": 44.976586172068686,
"step": 4000
}
],
"max_steps": 4000,
"num_train_epochs": 2,
"total_flos": 1.84651368394752e+19,
"trial_name": null,
"trial_params": null
}