warmestman's picture
Training in progress, step 10000, checkpoint
2323d53 verified
raw
history blame
No virus
51.2 kB
{
"best_metric": 41.91738628238271,
"best_model_checkpoint": "warmestman/whisper-large-v3-mn-cv-fleurs/checkpoint-6000",
"epoch": 59.880239520958085,
"eval_steps": 1000,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.15,
"learning_rate": 5e-08,
"loss": 1.4341,
"step": 25
},
{
"epoch": 0.3,
"learning_rate": 1e-07,
"loss": 1.3997,
"step": 50
},
{
"epoch": 0.45,
"learning_rate": 1.5e-07,
"loss": 1.33,
"step": 75
},
{
"epoch": 0.6,
"learning_rate": 2e-07,
"loss": 1.2112,
"step": 100
},
{
"epoch": 0.75,
"learning_rate": 2.5e-07,
"loss": 1.1128,
"step": 125
},
{
"epoch": 0.9,
"learning_rate": 3e-07,
"loss": 1.0554,
"step": 150
},
{
"epoch": 1.05,
"learning_rate": 3.5e-07,
"loss": 0.9914,
"step": 175
},
{
"epoch": 1.2,
"learning_rate": 4e-07,
"loss": 0.9151,
"step": 200
},
{
"epoch": 1.35,
"learning_rate": 4.5e-07,
"loss": 0.8636,
"step": 225
},
{
"epoch": 1.5,
"learning_rate": 5e-07,
"loss": 0.7988,
"step": 250
},
{
"epoch": 1.65,
"learning_rate": 5.5e-07,
"loss": 0.755,
"step": 275
},
{
"epoch": 1.8,
"learning_rate": 6e-07,
"loss": 0.7075,
"step": 300
},
{
"epoch": 1.95,
"learning_rate": 6.5e-07,
"loss": 0.6772,
"step": 325
},
{
"epoch": 2.1,
"learning_rate": 7e-07,
"loss": 0.6182,
"step": 350
},
{
"epoch": 2.25,
"learning_rate": 7.5e-07,
"loss": 0.5912,
"step": 375
},
{
"epoch": 2.4,
"learning_rate": 8e-07,
"loss": 0.5273,
"step": 400
},
{
"epoch": 2.54,
"learning_rate": 8.499999999999999e-07,
"loss": 0.5237,
"step": 425
},
{
"epoch": 2.69,
"learning_rate": 9e-07,
"loss": 0.4872,
"step": 450
},
{
"epoch": 2.84,
"learning_rate": 9.499999999999999e-07,
"loss": 0.4826,
"step": 475
},
{
"epoch": 2.99,
"learning_rate": 1e-06,
"loss": 0.4555,
"step": 500
},
{
"epoch": 3.14,
"learning_rate": 9.987179487179487e-07,
"loss": 0.4109,
"step": 525
},
{
"epoch": 3.29,
"learning_rate": 9.974358974358974e-07,
"loss": 0.3968,
"step": 550
},
{
"epoch": 3.44,
"learning_rate": 9.961538461538461e-07,
"loss": 0.3976,
"step": 575
},
{
"epoch": 3.59,
"learning_rate": 9.948717948717949e-07,
"loss": 0.3741,
"step": 600
},
{
"epoch": 3.74,
"learning_rate": 9.935897435897436e-07,
"loss": 0.3633,
"step": 625
},
{
"epoch": 3.89,
"learning_rate": 9.923076923076923e-07,
"loss": 0.3697,
"step": 650
},
{
"epoch": 4.04,
"learning_rate": 9.91025641025641e-07,
"loss": 0.3382,
"step": 675
},
{
"epoch": 4.19,
"learning_rate": 9.897435897435898e-07,
"loss": 0.3004,
"step": 700
},
{
"epoch": 4.34,
"learning_rate": 9.884615384615385e-07,
"loss": 0.3067,
"step": 725
},
{
"epoch": 4.49,
"learning_rate": 9.871794871794872e-07,
"loss": 0.3106,
"step": 750
},
{
"epoch": 4.64,
"learning_rate": 9.85897435897436e-07,
"loss": 0.286,
"step": 775
},
{
"epoch": 4.79,
"learning_rate": 9.846153846153847e-07,
"loss": 0.2978,
"step": 800
},
{
"epoch": 4.94,
"learning_rate": 9.833333333333332e-07,
"loss": 0.2964,
"step": 825
},
{
"epoch": 5.09,
"learning_rate": 9.820512820512819e-07,
"loss": 0.2618,
"step": 850
},
{
"epoch": 5.24,
"learning_rate": 9.807692307692306e-07,
"loss": 0.2552,
"step": 875
},
{
"epoch": 5.39,
"learning_rate": 9.794871794871793e-07,
"loss": 0.2487,
"step": 900
},
{
"epoch": 5.54,
"learning_rate": 9.78205128205128e-07,
"loss": 0.2412,
"step": 925
},
{
"epoch": 5.69,
"learning_rate": 9.769230769230768e-07,
"loss": 0.2331,
"step": 950
},
{
"epoch": 5.84,
"learning_rate": 9.756410256410255e-07,
"loss": 0.2418,
"step": 975
},
{
"epoch": 5.99,
"learning_rate": 9.743589743589742e-07,
"loss": 0.24,
"step": 1000
},
{
"epoch": 5.99,
"eval_loss": 0.4358440637588501,
"eval_runtime": 584.1166,
"eval_samples_per_second": 0.717,
"eval_steps_per_second": 0.091,
"eval_wer": 47.676145426544196,
"step": 1000
},
{
"epoch": 6.14,
"learning_rate": 9.73076923076923e-07,
"loss": 0.2033,
"step": 1025
},
{
"epoch": 6.29,
"learning_rate": 9.717948717948717e-07,
"loss": 0.2085,
"step": 1050
},
{
"epoch": 6.44,
"learning_rate": 9.705128205128204e-07,
"loss": 0.2081,
"step": 1075
},
{
"epoch": 6.59,
"learning_rate": 9.692307692307691e-07,
"loss": 0.1931,
"step": 1100
},
{
"epoch": 6.74,
"learning_rate": 9.679487179487179e-07,
"loss": 0.2043,
"step": 1125
},
{
"epoch": 6.89,
"learning_rate": 9.666666666666666e-07,
"loss": 0.1946,
"step": 1150
},
{
"epoch": 7.04,
"learning_rate": 9.653846153846153e-07,
"loss": 0.1871,
"step": 1175
},
{
"epoch": 7.19,
"learning_rate": 9.64102564102564e-07,
"loss": 0.1628,
"step": 1200
},
{
"epoch": 7.34,
"learning_rate": 9.628205128205128e-07,
"loss": 0.1626,
"step": 1225
},
{
"epoch": 7.49,
"learning_rate": 9.615384615384615e-07,
"loss": 0.1659,
"step": 1250
},
{
"epoch": 7.63,
"learning_rate": 9.602564102564102e-07,
"loss": 0.1692,
"step": 1275
},
{
"epoch": 7.78,
"learning_rate": 9.58974358974359e-07,
"loss": 0.1708,
"step": 1300
},
{
"epoch": 7.93,
"learning_rate": 9.576923076923077e-07,
"loss": 0.1643,
"step": 1325
},
{
"epoch": 8.08,
"learning_rate": 9.564102564102564e-07,
"loss": 0.1476,
"step": 1350
},
{
"epoch": 8.23,
"learning_rate": 9.551282051282051e-07,
"loss": 0.1328,
"step": 1375
},
{
"epoch": 8.38,
"learning_rate": 9.538461538461538e-07,
"loss": 0.1337,
"step": 1400
},
{
"epoch": 8.53,
"learning_rate": 9.525641025641025e-07,
"loss": 0.1484,
"step": 1425
},
{
"epoch": 8.68,
"learning_rate": 9.512820512820512e-07,
"loss": 0.1384,
"step": 1450
},
{
"epoch": 8.83,
"learning_rate": 9.499999999999999e-07,
"loss": 0.1323,
"step": 1475
},
{
"epoch": 8.98,
"learning_rate": 9.487179487179486e-07,
"loss": 0.1402,
"step": 1500
},
{
"epoch": 9.13,
"learning_rate": 9.474358974358974e-07,
"loss": 0.1109,
"step": 1525
},
{
"epoch": 9.28,
"learning_rate": 9.461538461538461e-07,
"loss": 0.1134,
"step": 1550
},
{
"epoch": 9.43,
"learning_rate": 9.448717948717948e-07,
"loss": 0.1103,
"step": 1575
},
{
"epoch": 9.58,
"learning_rate": 9.435897435897435e-07,
"loss": 0.1112,
"step": 1600
},
{
"epoch": 9.73,
"learning_rate": 9.423076923076923e-07,
"loss": 0.1114,
"step": 1625
},
{
"epoch": 9.88,
"learning_rate": 9.41025641025641e-07,
"loss": 0.1199,
"step": 1650
},
{
"epoch": 10.03,
"learning_rate": 9.397435897435897e-07,
"loss": 0.1094,
"step": 1675
},
{
"epoch": 10.18,
"learning_rate": 9.384615384615384e-07,
"loss": 0.0867,
"step": 1700
},
{
"epoch": 10.33,
"learning_rate": 9.371794871794872e-07,
"loss": 0.0892,
"step": 1725
},
{
"epoch": 10.48,
"learning_rate": 9.358974358974359e-07,
"loss": 0.0911,
"step": 1750
},
{
"epoch": 10.63,
"learning_rate": 9.346153846153846e-07,
"loss": 0.0944,
"step": 1775
},
{
"epoch": 10.78,
"learning_rate": 9.333333333333333e-07,
"loss": 0.091,
"step": 1800
},
{
"epoch": 10.93,
"learning_rate": 9.32051282051282e-07,
"loss": 0.0931,
"step": 1825
},
{
"epoch": 11.08,
"learning_rate": 9.307692307692308e-07,
"loss": 0.0798,
"step": 1850
},
{
"epoch": 11.23,
"learning_rate": 9.294871794871795e-07,
"loss": 0.071,
"step": 1875
},
{
"epoch": 11.38,
"learning_rate": 9.282051282051282e-07,
"loss": 0.0722,
"step": 1900
},
{
"epoch": 11.53,
"learning_rate": 9.26923076923077e-07,
"loss": 0.0762,
"step": 1925
},
{
"epoch": 11.68,
"learning_rate": 9.256410256410257e-07,
"loss": 0.0734,
"step": 1950
},
{
"epoch": 11.83,
"learning_rate": 9.243589743589744e-07,
"loss": 0.0738,
"step": 1975
},
{
"epoch": 11.98,
"learning_rate": 9.230769230769231e-07,
"loss": 0.077,
"step": 2000
},
{
"epoch": 11.98,
"eval_loss": 0.4446885585784912,
"eval_runtime": 586.2687,
"eval_samples_per_second": 0.715,
"eval_steps_per_second": 0.09,
"eval_wer": 43.46845608052362,
"step": 2000
},
{
"epoch": 12.13,
"learning_rate": 9.217948717948717e-07,
"loss": 0.0593,
"step": 2025
},
{
"epoch": 12.28,
"learning_rate": 9.205641025641026e-07,
"loss": 0.0547,
"step": 2050
},
{
"epoch": 12.43,
"learning_rate": 9.192820512820513e-07,
"loss": 0.0598,
"step": 2075
},
{
"epoch": 12.57,
"learning_rate": 9.18e-07,
"loss": 0.0564,
"step": 2100
},
{
"epoch": 12.72,
"learning_rate": 9.167179487179488e-07,
"loss": 0.0574,
"step": 2125
},
{
"epoch": 12.87,
"learning_rate": 9.154358974358975e-07,
"loss": 0.0614,
"step": 2150
},
{
"epoch": 13.02,
"learning_rate": 9.141538461538462e-07,
"loss": 0.0575,
"step": 2175
},
{
"epoch": 13.17,
"learning_rate": 9.128717948717947e-07,
"loss": 0.0474,
"step": 2200
},
{
"epoch": 13.32,
"learning_rate": 9.115897435897435e-07,
"loss": 0.0434,
"step": 2225
},
{
"epoch": 13.47,
"learning_rate": 9.103076923076922e-07,
"loss": 0.0464,
"step": 2250
},
{
"epoch": 13.62,
"learning_rate": 9.090256410256409e-07,
"loss": 0.0437,
"step": 2275
},
{
"epoch": 13.77,
"learning_rate": 9.077435897435896e-07,
"loss": 0.0437,
"step": 2300
},
{
"epoch": 13.92,
"learning_rate": 9.064615384615384e-07,
"loss": 0.0457,
"step": 2325
},
{
"epoch": 14.07,
"learning_rate": 9.051794871794871e-07,
"loss": 0.042,
"step": 2350
},
{
"epoch": 14.22,
"learning_rate": 9.038974358974358e-07,
"loss": 0.0339,
"step": 2375
},
{
"epoch": 14.37,
"learning_rate": 9.026153846153845e-07,
"loss": 0.0342,
"step": 2400
},
{
"epoch": 14.52,
"learning_rate": 9.013333333333333e-07,
"loss": 0.0323,
"step": 2425
},
{
"epoch": 14.67,
"learning_rate": 9.00051282051282e-07,
"loss": 0.0348,
"step": 2450
},
{
"epoch": 14.82,
"learning_rate": 8.987692307692307e-07,
"loss": 0.0368,
"step": 2475
},
{
"epoch": 14.97,
"learning_rate": 8.974871794871794e-07,
"loss": 0.036,
"step": 2500
},
{
"epoch": 15.12,
"learning_rate": 8.962051282051282e-07,
"loss": 0.0252,
"step": 2525
},
{
"epoch": 15.27,
"learning_rate": 8.949230769230769e-07,
"loss": 0.0287,
"step": 2550
},
{
"epoch": 15.42,
"learning_rate": 8.936410256410256e-07,
"loss": 0.0275,
"step": 2575
},
{
"epoch": 15.57,
"learning_rate": 8.923589743589743e-07,
"loss": 0.0263,
"step": 2600
},
{
"epoch": 15.72,
"learning_rate": 8.91076923076923e-07,
"loss": 0.0268,
"step": 2625
},
{
"epoch": 15.87,
"learning_rate": 8.897948717948718e-07,
"loss": 0.0277,
"step": 2650
},
{
"epoch": 16.02,
"learning_rate": 8.885128205128205e-07,
"loss": 0.0263,
"step": 2675
},
{
"epoch": 16.17,
"learning_rate": 8.872307692307692e-07,
"loss": 0.0182,
"step": 2700
},
{
"epoch": 16.32,
"learning_rate": 8.85948717948718e-07,
"loss": 0.0222,
"step": 2725
},
{
"epoch": 16.47,
"learning_rate": 8.846666666666667e-07,
"loss": 0.0205,
"step": 2750
},
{
"epoch": 16.62,
"learning_rate": 8.833846153846154e-07,
"loss": 0.0216,
"step": 2775
},
{
"epoch": 16.77,
"learning_rate": 8.821025641025641e-07,
"loss": 0.0207,
"step": 2800
},
{
"epoch": 16.92,
"learning_rate": 8.808205128205127e-07,
"loss": 0.0223,
"step": 2825
},
{
"epoch": 17.07,
"learning_rate": 8.795384615384615e-07,
"loss": 0.0173,
"step": 2850
},
{
"epoch": 17.22,
"learning_rate": 8.782564102564102e-07,
"loss": 0.0144,
"step": 2875
},
{
"epoch": 17.37,
"learning_rate": 8.769743589743589e-07,
"loss": 0.0159,
"step": 2900
},
{
"epoch": 17.51,
"learning_rate": 8.756923076923076e-07,
"loss": 0.0146,
"step": 2925
},
{
"epoch": 17.66,
"learning_rate": 8.744102564102564e-07,
"loss": 0.0154,
"step": 2950
},
{
"epoch": 17.81,
"learning_rate": 8.731282051282051e-07,
"loss": 0.0152,
"step": 2975
},
{
"epoch": 17.96,
"learning_rate": 8.718461538461538e-07,
"loss": 0.0188,
"step": 3000
},
{
"epoch": 17.96,
"eval_loss": 0.5422356724739075,
"eval_runtime": 586.4584,
"eval_samples_per_second": 0.714,
"eval_steps_per_second": 0.09,
"eval_wer": 43.32544964523404,
"step": 3000
},
{
"epoch": 18.11,
"learning_rate": 8.705641025641025e-07,
"loss": 0.0133,
"step": 3025
},
{
"epoch": 18.26,
"learning_rate": 8.692820512820513e-07,
"loss": 0.0108,
"step": 3050
},
{
"epoch": 18.41,
"learning_rate": 8.68e-07,
"loss": 0.0131,
"step": 3075
},
{
"epoch": 18.56,
"learning_rate": 8.667179487179487e-07,
"loss": 0.0126,
"step": 3100
},
{
"epoch": 18.71,
"learning_rate": 8.654358974358974e-07,
"loss": 0.0128,
"step": 3125
},
{
"epoch": 18.86,
"learning_rate": 8.641538461538462e-07,
"loss": 0.012,
"step": 3150
},
{
"epoch": 19.01,
"learning_rate": 8.628717948717949e-07,
"loss": 0.0128,
"step": 3175
},
{
"epoch": 19.16,
"learning_rate": 8.615897435897436e-07,
"loss": 0.0098,
"step": 3200
},
{
"epoch": 19.31,
"learning_rate": 8.603076923076923e-07,
"loss": 0.0092,
"step": 3225
},
{
"epoch": 19.46,
"learning_rate": 8.590256410256411e-07,
"loss": 0.0092,
"step": 3250
},
{
"epoch": 19.61,
"learning_rate": 8.577435897435897e-07,
"loss": 0.009,
"step": 3275
},
{
"epoch": 19.76,
"learning_rate": 8.564615384615384e-07,
"loss": 0.0092,
"step": 3300
},
{
"epoch": 19.91,
"learning_rate": 8.551794871794871e-07,
"loss": 0.0091,
"step": 3325
},
{
"epoch": 20.06,
"learning_rate": 8.538974358974359e-07,
"loss": 0.0097,
"step": 3350
},
{
"epoch": 20.21,
"learning_rate": 8.526153846153846e-07,
"loss": 0.0075,
"step": 3375
},
{
"epoch": 20.36,
"learning_rate": 8.513333333333333e-07,
"loss": 0.0084,
"step": 3400
},
{
"epoch": 20.51,
"learning_rate": 8.500512820512819e-07,
"loss": 0.0069,
"step": 3425
},
{
"epoch": 20.66,
"learning_rate": 8.487692307692307e-07,
"loss": 0.0071,
"step": 3450
},
{
"epoch": 20.81,
"learning_rate": 8.474871794871794e-07,
"loss": 0.0077,
"step": 3475
},
{
"epoch": 20.96,
"learning_rate": 8.462051282051281e-07,
"loss": 0.0081,
"step": 3500
},
{
"epoch": 21.11,
"learning_rate": 8.449230769230768e-07,
"loss": 0.006,
"step": 3525
},
{
"epoch": 21.26,
"learning_rate": 8.436410256410255e-07,
"loss": 0.0058,
"step": 3550
},
{
"epoch": 21.41,
"learning_rate": 8.423589743589743e-07,
"loss": 0.0059,
"step": 3575
},
{
"epoch": 21.56,
"learning_rate": 8.41076923076923e-07,
"loss": 0.0053,
"step": 3600
},
{
"epoch": 21.71,
"learning_rate": 8.397948717948717e-07,
"loss": 0.0057,
"step": 3625
},
{
"epoch": 21.86,
"learning_rate": 8.385128205128204e-07,
"loss": 0.0053,
"step": 3650
},
{
"epoch": 22.01,
"learning_rate": 8.372307692307692e-07,
"loss": 0.0063,
"step": 3675
},
{
"epoch": 22.16,
"learning_rate": 8.359487179487179e-07,
"loss": 0.0045,
"step": 3700
},
{
"epoch": 22.31,
"learning_rate": 8.346666666666666e-07,
"loss": 0.0043,
"step": 3725
},
{
"epoch": 22.46,
"learning_rate": 8.333846153846153e-07,
"loss": 0.0041,
"step": 3750
},
{
"epoch": 22.6,
"learning_rate": 8.321025641025641e-07,
"loss": 0.0042,
"step": 3775
},
{
"epoch": 22.75,
"learning_rate": 8.308205128205128e-07,
"loss": 0.0044,
"step": 3800
},
{
"epoch": 22.9,
"learning_rate": 8.295384615384615e-07,
"loss": 0.0046,
"step": 3825
},
{
"epoch": 23.05,
"learning_rate": 8.282564102564102e-07,
"loss": 0.005,
"step": 3850
},
{
"epoch": 23.2,
"learning_rate": 8.26974358974359e-07,
"loss": 0.0036,
"step": 3875
},
{
"epoch": 23.35,
"learning_rate": 8.256923076923077e-07,
"loss": 0.0039,
"step": 3900
},
{
"epoch": 23.5,
"learning_rate": 8.244102564102564e-07,
"loss": 0.0033,
"step": 3925
},
{
"epoch": 23.65,
"learning_rate": 8.231282051282051e-07,
"loss": 0.0035,
"step": 3950
},
{
"epoch": 23.8,
"learning_rate": 8.218461538461539e-07,
"loss": 0.0041,
"step": 3975
},
{
"epoch": 23.95,
"learning_rate": 8.205641025641026e-07,
"loss": 0.004,
"step": 4000
},
{
"epoch": 23.95,
"eval_loss": 0.6144538521766663,
"eval_runtime": 590.5826,
"eval_samples_per_second": 0.709,
"eval_steps_per_second": 0.09,
"eval_wer": 42.95693306198779,
"step": 4000
},
{
"epoch": 24.1,
"learning_rate": 8.192820512820512e-07,
"loss": 0.0037,
"step": 4025
},
{
"epoch": 24.25,
"learning_rate": 8.179999999999999e-07,
"loss": 0.0031,
"step": 4050
},
{
"epoch": 24.4,
"learning_rate": 8.167179487179487e-07,
"loss": 0.0028,
"step": 4075
},
{
"epoch": 24.55,
"learning_rate": 8.154358974358974e-07,
"loss": 0.0034,
"step": 4100
},
{
"epoch": 24.7,
"learning_rate": 8.141538461538461e-07,
"loss": 0.0032,
"step": 4125
},
{
"epoch": 24.85,
"learning_rate": 8.128717948717948e-07,
"loss": 0.0031,
"step": 4150
},
{
"epoch": 25.0,
"learning_rate": 8.115897435897436e-07,
"loss": 0.0031,
"step": 4175
},
{
"epoch": 25.15,
"learning_rate": 8.103076923076923e-07,
"loss": 0.0023,
"step": 4200
},
{
"epoch": 25.3,
"learning_rate": 8.09025641025641e-07,
"loss": 0.0024,
"step": 4225
},
{
"epoch": 25.45,
"learning_rate": 8.077435897435897e-07,
"loss": 0.0025,
"step": 4250
},
{
"epoch": 25.6,
"learning_rate": 8.064615384615385e-07,
"loss": 0.0021,
"step": 4275
},
{
"epoch": 25.75,
"learning_rate": 8.051794871794872e-07,
"loss": 0.0022,
"step": 4300
},
{
"epoch": 25.9,
"learning_rate": 8.038974358974359e-07,
"loss": 0.0022,
"step": 4325
},
{
"epoch": 26.05,
"learning_rate": 8.026153846153845e-07,
"loss": 0.0024,
"step": 4350
},
{
"epoch": 26.2,
"learning_rate": 8.013333333333333e-07,
"loss": 0.0018,
"step": 4375
},
{
"epoch": 26.35,
"learning_rate": 8.00051282051282e-07,
"loss": 0.0016,
"step": 4400
},
{
"epoch": 26.5,
"learning_rate": 7.987692307692307e-07,
"loss": 0.0019,
"step": 4425
},
{
"epoch": 26.65,
"learning_rate": 7.974871794871794e-07,
"loss": 0.0021,
"step": 4450
},
{
"epoch": 26.8,
"learning_rate": 7.962051282051282e-07,
"loss": 0.0019,
"step": 4475
},
{
"epoch": 26.95,
"learning_rate": 7.949230769230769e-07,
"loss": 0.0019,
"step": 4500
},
{
"epoch": 27.1,
"learning_rate": 7.936410256410256e-07,
"loss": 0.0021,
"step": 4525
},
{
"epoch": 27.25,
"learning_rate": 7.923589743589743e-07,
"loss": 0.0019,
"step": 4550
},
{
"epoch": 27.4,
"learning_rate": 7.910769230769231e-07,
"loss": 0.002,
"step": 4575
},
{
"epoch": 27.54,
"learning_rate": 7.897948717948718e-07,
"loss": 0.0024,
"step": 4600
},
{
"epoch": 27.69,
"learning_rate": 7.885128205128205e-07,
"loss": 0.0028,
"step": 4625
},
{
"epoch": 27.84,
"learning_rate": 7.872307692307691e-07,
"loss": 0.0033,
"step": 4650
},
{
"epoch": 27.99,
"learning_rate": 7.859487179487178e-07,
"loss": 0.0036,
"step": 4675
},
{
"epoch": 28.14,
"learning_rate": 7.846666666666666e-07,
"loss": 0.0035,
"step": 4700
},
{
"epoch": 28.29,
"learning_rate": 7.833846153846153e-07,
"loss": 0.0031,
"step": 4725
},
{
"epoch": 28.44,
"learning_rate": 7.82102564102564e-07,
"loss": 0.0032,
"step": 4750
},
{
"epoch": 28.59,
"learning_rate": 7.808205128205127e-07,
"loss": 0.0037,
"step": 4775
},
{
"epoch": 28.74,
"learning_rate": 7.795384615384615e-07,
"loss": 0.0036,
"step": 4800
},
{
"epoch": 28.89,
"learning_rate": 7.782564102564102e-07,
"loss": 0.0046,
"step": 4825
},
{
"epoch": 29.04,
"learning_rate": 7.769743589743589e-07,
"loss": 0.0042,
"step": 4850
},
{
"epoch": 29.19,
"learning_rate": 7.756923076923076e-07,
"loss": 0.0032,
"step": 4875
},
{
"epoch": 29.34,
"learning_rate": 7.744102564102564e-07,
"loss": 0.0037,
"step": 4900
},
{
"epoch": 29.49,
"learning_rate": 7.731282051282051e-07,
"loss": 0.0039,
"step": 4925
},
{
"epoch": 29.64,
"learning_rate": 7.718461538461538e-07,
"loss": 0.0034,
"step": 4950
},
{
"epoch": 29.79,
"learning_rate": 7.705641025641025e-07,
"loss": 0.0034,
"step": 4975
},
{
"epoch": 29.94,
"learning_rate": 7.692820512820513e-07,
"loss": 0.0036,
"step": 5000
},
{
"epoch": 29.94,
"eval_loss": 0.6325389742851257,
"eval_runtime": 588.3841,
"eval_samples_per_second": 0.712,
"eval_steps_per_second": 0.09,
"eval_wer": 42.08789395522798,
"step": 5000
},
{
"epoch": 30.09,
"learning_rate": 7.68e-07,
"loss": 0.0032,
"step": 5025
},
{
"epoch": 30.24,
"learning_rate": 7.667179487179487e-07,
"loss": 0.003,
"step": 5050
},
{
"epoch": 30.39,
"learning_rate": 7.654358974358974e-07,
"loss": 0.0026,
"step": 5075
},
{
"epoch": 30.54,
"learning_rate": 7.641538461538462e-07,
"loss": 0.0026,
"step": 5100
},
{
"epoch": 30.69,
"learning_rate": 7.628717948717949e-07,
"loss": 0.0026,
"step": 5125
},
{
"epoch": 30.84,
"learning_rate": 7.615897435897436e-07,
"loss": 0.0027,
"step": 5150
},
{
"epoch": 30.99,
"learning_rate": 7.603076923076923e-07,
"loss": 0.0029,
"step": 5175
},
{
"epoch": 31.14,
"learning_rate": 7.590256410256411e-07,
"loss": 0.0024,
"step": 5200
},
{
"epoch": 31.29,
"learning_rate": 7.577435897435898e-07,
"loss": 0.0019,
"step": 5225
},
{
"epoch": 31.44,
"learning_rate": 7.564615384615384e-07,
"loss": 0.0023,
"step": 5250
},
{
"epoch": 31.59,
"learning_rate": 7.551794871794871e-07,
"loss": 0.0021,
"step": 5275
},
{
"epoch": 31.74,
"learning_rate": 7.538974358974359e-07,
"loss": 0.002,
"step": 5300
},
{
"epoch": 31.89,
"learning_rate": 7.526153846153846e-07,
"loss": 0.0021,
"step": 5325
},
{
"epoch": 32.04,
"learning_rate": 7.513333333333333e-07,
"loss": 0.0024,
"step": 5350
},
{
"epoch": 32.19,
"learning_rate": 7.50051282051282e-07,
"loss": 0.002,
"step": 5375
},
{
"epoch": 32.34,
"learning_rate": 7.487692307692308e-07,
"loss": 0.0021,
"step": 5400
},
{
"epoch": 32.49,
"learning_rate": 7.475384615384615e-07,
"loss": 0.0019,
"step": 5425
},
{
"epoch": 32.63,
"learning_rate": 7.462564102564101e-07,
"loss": 0.0019,
"step": 5450
},
{
"epoch": 32.78,
"learning_rate": 7.449743589743588e-07,
"loss": 0.0018,
"step": 5475
},
{
"epoch": 32.93,
"learning_rate": 7.436923076923076e-07,
"loss": 0.0018,
"step": 5500
},
{
"epoch": 33.08,
"learning_rate": 7.424102564102563e-07,
"loss": 0.0016,
"step": 5525
},
{
"epoch": 33.23,
"learning_rate": 7.41128205128205e-07,
"loss": 0.0019,
"step": 5550
},
{
"epoch": 33.38,
"learning_rate": 7.398461538461537e-07,
"loss": 0.0017,
"step": 5575
},
{
"epoch": 33.53,
"learning_rate": 7.385641025641025e-07,
"loss": 0.0019,
"step": 5600
},
{
"epoch": 33.68,
"learning_rate": 7.372820512820512e-07,
"loss": 0.0015,
"step": 5625
},
{
"epoch": 33.83,
"learning_rate": 7.359999999999999e-07,
"loss": 0.0016,
"step": 5650
},
{
"epoch": 33.98,
"learning_rate": 7.347179487179486e-07,
"loss": 0.0016,
"step": 5675
},
{
"epoch": 34.13,
"learning_rate": 7.334358974358974e-07,
"loss": 0.0013,
"step": 5700
},
{
"epoch": 34.28,
"learning_rate": 7.321538461538461e-07,
"loss": 0.0018,
"step": 5725
},
{
"epoch": 34.43,
"learning_rate": 7.308717948717948e-07,
"loss": 0.0015,
"step": 5750
},
{
"epoch": 34.58,
"learning_rate": 7.295897435897435e-07,
"loss": 0.0017,
"step": 5775
},
{
"epoch": 34.73,
"learning_rate": 7.283076923076923e-07,
"loss": 0.0017,
"step": 5800
},
{
"epoch": 34.88,
"learning_rate": 7.27025641025641e-07,
"loss": 0.0018,
"step": 5825
},
{
"epoch": 35.03,
"learning_rate": 7.257435897435897e-07,
"loss": 0.002,
"step": 5850
},
{
"epoch": 35.18,
"learning_rate": 7.244615384615384e-07,
"loss": 0.0016,
"step": 5875
},
{
"epoch": 35.33,
"learning_rate": 7.231794871794872e-07,
"loss": 0.0018,
"step": 5900
},
{
"epoch": 35.48,
"learning_rate": 7.218974358974359e-07,
"loss": 0.0016,
"step": 5925
},
{
"epoch": 35.63,
"learning_rate": 7.206153846153846e-07,
"loss": 0.0015,
"step": 5950
},
{
"epoch": 35.78,
"learning_rate": 7.193333333333333e-07,
"loss": 0.0014,
"step": 5975
},
{
"epoch": 35.93,
"learning_rate": 7.180512820512821e-07,
"loss": 0.0018,
"step": 6000
},
{
"epoch": 35.93,
"eval_loss": 0.6662390232086182,
"eval_runtime": 594.4694,
"eval_samples_per_second": 0.705,
"eval_steps_per_second": 0.089,
"eval_wer": 41.91738628238271,
"step": 6000
},
{
"epoch": 36.08,
"learning_rate": 7.167692307692307e-07,
"loss": 0.0017,
"step": 6025
},
{
"epoch": 36.23,
"learning_rate": 7.154871794871794e-07,
"loss": 0.0019,
"step": 6050
},
{
"epoch": 36.38,
"learning_rate": 7.142051282051281e-07,
"loss": 0.0015,
"step": 6075
},
{
"epoch": 36.53,
"learning_rate": 7.129230769230769e-07,
"loss": 0.0025,
"step": 6100
},
{
"epoch": 36.68,
"learning_rate": 7.116410256410256e-07,
"loss": 0.0017,
"step": 6125
},
{
"epoch": 36.83,
"learning_rate": 7.103589743589743e-07,
"loss": 0.0014,
"step": 6150
},
{
"epoch": 36.98,
"learning_rate": 7.09076923076923e-07,
"loss": 0.0015,
"step": 6175
},
{
"epoch": 37.13,
"learning_rate": 7.077948717948718e-07,
"loss": 0.0011,
"step": 6200
},
{
"epoch": 37.28,
"learning_rate": 7.065128205128205e-07,
"loss": 0.0012,
"step": 6225
},
{
"epoch": 37.43,
"learning_rate": 7.052307692307692e-07,
"loss": 0.0017,
"step": 6250
},
{
"epoch": 37.57,
"learning_rate": 7.039487179487179e-07,
"loss": 0.0015,
"step": 6275
},
{
"epoch": 37.72,
"learning_rate": 7.026666666666667e-07,
"loss": 0.0016,
"step": 6300
},
{
"epoch": 37.87,
"learning_rate": 7.013846153846154e-07,
"loss": 0.0018,
"step": 6325
},
{
"epoch": 38.02,
"learning_rate": 7.001025641025641e-07,
"loss": 0.0013,
"step": 6350
},
{
"epoch": 38.17,
"learning_rate": 6.988205128205128e-07,
"loss": 0.0012,
"step": 6375
},
{
"epoch": 38.32,
"learning_rate": 6.975384615384616e-07,
"loss": 0.0013,
"step": 6400
},
{
"epoch": 38.47,
"learning_rate": 6.962564102564103e-07,
"loss": 0.0009,
"step": 6425
},
{
"epoch": 38.62,
"learning_rate": 6.94974358974359e-07,
"loss": 0.0012,
"step": 6450
},
{
"epoch": 38.77,
"learning_rate": 6.936923076923077e-07,
"loss": 0.001,
"step": 6475
},
{
"epoch": 38.92,
"learning_rate": 6.924102564102565e-07,
"loss": 0.001,
"step": 6500
},
{
"epoch": 39.07,
"learning_rate": 6.911282051282052e-07,
"loss": 0.001,
"step": 6525
},
{
"epoch": 39.22,
"learning_rate": 6.898461538461539e-07,
"loss": 0.0012,
"step": 6550
},
{
"epoch": 39.37,
"learning_rate": 6.885641025641025e-07,
"loss": 0.0012,
"step": 6575
},
{
"epoch": 39.52,
"learning_rate": 6.872820512820513e-07,
"loss": 0.0014,
"step": 6600
},
{
"epoch": 39.67,
"learning_rate": 6.86e-07,
"loss": 0.0018,
"step": 6625
},
{
"epoch": 39.82,
"learning_rate": 6.847179487179486e-07,
"loss": 0.0014,
"step": 6650
},
{
"epoch": 39.97,
"learning_rate": 6.834358974358973e-07,
"loss": 0.0017,
"step": 6675
},
{
"epoch": 40.12,
"learning_rate": 6.82153846153846e-07,
"loss": 0.0015,
"step": 6700
},
{
"epoch": 40.27,
"learning_rate": 6.808717948717948e-07,
"loss": 0.0019,
"step": 6725
},
{
"epoch": 40.42,
"learning_rate": 6.795897435897435e-07,
"loss": 0.0014,
"step": 6750
},
{
"epoch": 40.57,
"learning_rate": 6.783076923076922e-07,
"loss": 0.0019,
"step": 6775
},
{
"epoch": 40.72,
"learning_rate": 6.770256410256409e-07,
"loss": 0.0016,
"step": 6800
},
{
"epoch": 40.87,
"learning_rate": 6.757435897435897e-07,
"loss": 0.0015,
"step": 6825
},
{
"epoch": 41.02,
"learning_rate": 6.744615384615384e-07,
"loss": 0.0014,
"step": 6850
},
{
"epoch": 41.17,
"learning_rate": 6.731794871794871e-07,
"loss": 0.0012,
"step": 6875
},
{
"epoch": 41.32,
"learning_rate": 6.718974358974358e-07,
"loss": 0.0013,
"step": 6900
},
{
"epoch": 41.47,
"learning_rate": 6.706153846153846e-07,
"loss": 0.0013,
"step": 6925
},
{
"epoch": 41.62,
"learning_rate": 6.693333333333333e-07,
"loss": 0.0012,
"step": 6950
},
{
"epoch": 41.77,
"learning_rate": 6.68051282051282e-07,
"loss": 0.0012,
"step": 6975
},
{
"epoch": 41.92,
"learning_rate": 6.667692307692307e-07,
"loss": 0.0013,
"step": 7000
},
{
"epoch": 41.92,
"eval_loss": 0.6843737363815308,
"eval_runtime": 592.3336,
"eval_samples_per_second": 0.707,
"eval_steps_per_second": 0.089,
"eval_wer": 42.35190583576261,
"step": 7000
},
{
"epoch": 42.07,
"learning_rate": 6.654871794871795e-07,
"loss": 0.0011,
"step": 7025
},
{
"epoch": 42.22,
"learning_rate": 6.642051282051282e-07,
"loss": 0.0009,
"step": 7050
},
{
"epoch": 42.37,
"learning_rate": 6.629230769230769e-07,
"loss": 0.001,
"step": 7075
},
{
"epoch": 42.51,
"learning_rate": 6.616410256410256e-07,
"loss": 0.0011,
"step": 7100
},
{
"epoch": 42.66,
"learning_rate": 6.603589743589744e-07,
"loss": 0.0014,
"step": 7125
},
{
"epoch": 42.81,
"learning_rate": 6.590769230769231e-07,
"loss": 0.001,
"step": 7150
},
{
"epoch": 42.96,
"learning_rate": 6.577948717948718e-07,
"loss": 0.0012,
"step": 7175
},
{
"epoch": 43.11,
"learning_rate": 6.565128205128205e-07,
"loss": 0.0011,
"step": 7200
},
{
"epoch": 43.26,
"learning_rate": 6.552307692307693e-07,
"loss": 0.001,
"step": 7225
},
{
"epoch": 43.41,
"learning_rate": 6.539487179487179e-07,
"loss": 0.0015,
"step": 7250
},
{
"epoch": 43.56,
"learning_rate": 6.526666666666666e-07,
"loss": 0.0016,
"step": 7275
},
{
"epoch": 43.71,
"learning_rate": 6.513846153846153e-07,
"loss": 0.0017,
"step": 7300
},
{
"epoch": 43.86,
"learning_rate": 6.501025641025641e-07,
"loss": 0.0014,
"step": 7325
},
{
"epoch": 44.01,
"learning_rate": 6.488205128205128e-07,
"loss": 0.0012,
"step": 7350
},
{
"epoch": 44.16,
"learning_rate": 6.475384615384615e-07,
"loss": 0.0014,
"step": 7375
},
{
"epoch": 44.31,
"learning_rate": 6.462564102564102e-07,
"loss": 0.001,
"step": 7400
},
{
"epoch": 44.46,
"learning_rate": 6.44974358974359e-07,
"loss": 0.0011,
"step": 7425
},
{
"epoch": 44.61,
"learning_rate": 6.436923076923077e-07,
"loss": 0.0012,
"step": 7450
},
{
"epoch": 44.76,
"learning_rate": 6.424102564102564e-07,
"loss": 0.0011,
"step": 7475
},
{
"epoch": 44.91,
"learning_rate": 6.411282051282051e-07,
"loss": 0.0012,
"step": 7500
},
{
"epoch": 45.06,
"learning_rate": 6.398461538461539e-07,
"loss": 0.0011,
"step": 7525
},
{
"epoch": 45.21,
"learning_rate": 6.385641025641026e-07,
"loss": 0.0008,
"step": 7550
},
{
"epoch": 45.36,
"learning_rate": 6.372820512820513e-07,
"loss": 0.001,
"step": 7575
},
{
"epoch": 45.51,
"learning_rate": 6.36e-07,
"loss": 0.0011,
"step": 7600
},
{
"epoch": 45.66,
"learning_rate": 6.347179487179488e-07,
"loss": 0.0009,
"step": 7625
},
{
"epoch": 45.81,
"learning_rate": 6.334358974358974e-07,
"loss": 0.0011,
"step": 7650
},
{
"epoch": 45.96,
"learning_rate": 6.321538461538461e-07,
"loss": 0.0009,
"step": 7675
},
{
"epoch": 46.11,
"learning_rate": 6.308717948717948e-07,
"loss": 0.0007,
"step": 7700
},
{
"epoch": 46.26,
"learning_rate": 6.295897435897435e-07,
"loss": 0.0006,
"step": 7725
},
{
"epoch": 46.41,
"learning_rate": 6.283076923076923e-07,
"loss": 0.0006,
"step": 7750
},
{
"epoch": 46.56,
"learning_rate": 6.27025641025641e-07,
"loss": 0.0008,
"step": 7775
},
{
"epoch": 46.71,
"learning_rate": 6.257435897435897e-07,
"loss": 0.0007,
"step": 7800
},
{
"epoch": 46.86,
"learning_rate": 6.244615384615384e-07,
"loss": 0.0008,
"step": 7825
},
{
"epoch": 47.01,
"learning_rate": 6.231794871794872e-07,
"loss": 0.0007,
"step": 7850
},
{
"epoch": 47.16,
"learning_rate": 6.218974358974358e-07,
"loss": 0.0005,
"step": 7875
},
{
"epoch": 47.31,
"learning_rate": 6.206153846153845e-07,
"loss": 0.0005,
"step": 7900
},
{
"epoch": 47.46,
"learning_rate": 6.193333333333332e-07,
"loss": 0.0005,
"step": 7925
},
{
"epoch": 47.6,
"learning_rate": 6.18051282051282e-07,
"loss": 0.0004,
"step": 7950
},
{
"epoch": 47.75,
"learning_rate": 6.167692307692307e-07,
"loss": 0.0004,
"step": 7975
},
{
"epoch": 47.9,
"learning_rate": 6.154871794871794e-07,
"loss": 0.0005,
"step": 8000
},
{
"epoch": 47.9,
"eval_loss": 0.7143049240112305,
"eval_runtime": 593.312,
"eval_samples_per_second": 0.706,
"eval_steps_per_second": 0.089,
"eval_wer": 42.22540014300644,
"step": 8000
},
{
"epoch": 48.05,
"learning_rate": 6.142051282051281e-07,
"loss": 0.0005,
"step": 8025
},
{
"epoch": 48.2,
"learning_rate": 6.129230769230769e-07,
"loss": 0.0004,
"step": 8050
},
{
"epoch": 48.35,
"learning_rate": 6.116410256410256e-07,
"loss": 0.0004,
"step": 8075
},
{
"epoch": 48.5,
"learning_rate": 6.103589743589743e-07,
"loss": 0.0004,
"step": 8100
},
{
"epoch": 48.65,
"learning_rate": 6.09076923076923e-07,
"loss": 0.0004,
"step": 8125
},
{
"epoch": 48.8,
"learning_rate": 6.077948717948718e-07,
"loss": 0.0004,
"step": 8150
},
{
"epoch": 48.95,
"learning_rate": 6.065128205128205e-07,
"loss": 0.0005,
"step": 8175
},
{
"epoch": 49.1,
"learning_rate": 6.052307692307692e-07,
"loss": 0.0004,
"step": 8200
},
{
"epoch": 49.25,
"learning_rate": 6.039487179487179e-07,
"loss": 0.0004,
"step": 8225
},
{
"epoch": 49.4,
"learning_rate": 6.026666666666667e-07,
"loss": 0.0004,
"step": 8250
},
{
"epoch": 49.55,
"learning_rate": 6.013846153846154e-07,
"loss": 0.0004,
"step": 8275
},
{
"epoch": 49.7,
"learning_rate": 6.001025641025641e-07,
"loss": 0.0004,
"step": 8300
},
{
"epoch": 49.85,
"learning_rate": 5.988205128205128e-07,
"loss": 0.0004,
"step": 8325
},
{
"epoch": 50.0,
"learning_rate": 5.975384615384616e-07,
"loss": 0.0004,
"step": 8350
},
{
"epoch": 50.15,
"learning_rate": 5.962564102564103e-07,
"loss": 0.0003,
"step": 8375
},
{
"epoch": 50.3,
"learning_rate": 5.94974358974359e-07,
"loss": 0.0004,
"step": 8400
},
{
"epoch": 50.45,
"learning_rate": 5.936923076923077e-07,
"loss": 0.0003,
"step": 8425
},
{
"epoch": 50.6,
"learning_rate": 5.924102564102565e-07,
"loss": 0.0003,
"step": 8450
},
{
"epoch": 50.75,
"learning_rate": 5.911282051282051e-07,
"loss": 0.0003,
"step": 8475
},
{
"epoch": 50.9,
"learning_rate": 5.898461538461538e-07,
"loss": 0.0003,
"step": 8500
},
{
"epoch": 51.05,
"learning_rate": 5.885641025641025e-07,
"loss": 0.0003,
"step": 8525
},
{
"epoch": 51.2,
"learning_rate": 5.872820512820513e-07,
"loss": 0.0003,
"step": 8550
},
{
"epoch": 51.35,
"learning_rate": 5.86e-07,
"loss": 0.0003,
"step": 8575
},
{
"epoch": 51.5,
"learning_rate": 5.847179487179487e-07,
"loss": 0.0003,
"step": 8600
},
{
"epoch": 51.65,
"learning_rate": 5.834358974358974e-07,
"loss": 0.0003,
"step": 8625
},
{
"epoch": 51.8,
"learning_rate": 5.821538461538462e-07,
"loss": 0.0003,
"step": 8650
},
{
"epoch": 51.95,
"learning_rate": 5.808717948717949e-07,
"loss": 0.0003,
"step": 8675
},
{
"epoch": 52.1,
"learning_rate": 5.795897435897436e-07,
"loss": 0.0003,
"step": 8700
},
{
"epoch": 52.25,
"learning_rate": 5.783076923076922e-07,
"loss": 0.0003,
"step": 8725
},
{
"epoch": 52.4,
"learning_rate": 5.77025641025641e-07,
"loss": 0.0003,
"step": 8750
},
{
"epoch": 52.54,
"learning_rate": 5.757435897435897e-07,
"loss": 0.0003,
"step": 8775
},
{
"epoch": 52.69,
"learning_rate": 5.744615384615384e-07,
"loss": 0.0003,
"step": 8800
},
{
"epoch": 52.84,
"learning_rate": 5.731794871794871e-07,
"loss": 0.0003,
"step": 8825
},
{
"epoch": 52.99,
"learning_rate": 5.718974358974358e-07,
"loss": 0.0003,
"step": 8850
},
{
"epoch": 53.14,
"learning_rate": 5.706153846153846e-07,
"loss": 0.0003,
"step": 8875
},
{
"epoch": 53.29,
"learning_rate": 5.693333333333333e-07,
"loss": 0.0003,
"step": 8900
},
{
"epoch": 53.44,
"learning_rate": 5.68051282051282e-07,
"loss": 0.0003,
"step": 8925
},
{
"epoch": 53.59,
"learning_rate": 5.667692307692307e-07,
"loss": 0.0003,
"step": 8950
},
{
"epoch": 53.74,
"learning_rate": 5.654871794871795e-07,
"loss": 0.0003,
"step": 8975
},
{
"epoch": 53.89,
"learning_rate": 5.642051282051282e-07,
"loss": 0.0003,
"step": 9000
},
{
"epoch": 53.89,
"eval_loss": 0.7420342564582825,
"eval_runtime": 596.5217,
"eval_samples_per_second": 0.702,
"eval_steps_per_second": 0.089,
"eval_wer": 42.20889940047302,
"step": 9000
},
{
"epoch": 54.04,
"learning_rate": 5.629230769230769e-07,
"loss": 0.0003,
"step": 9025
},
{
"epoch": 54.19,
"learning_rate": 5.616410256410256e-07,
"loss": 0.0003,
"step": 9050
},
{
"epoch": 54.34,
"learning_rate": 5.603589743589743e-07,
"loss": 0.0003,
"step": 9075
},
{
"epoch": 54.49,
"learning_rate": 5.59076923076923e-07,
"loss": 0.0003,
"step": 9100
},
{
"epoch": 54.64,
"learning_rate": 5.577948717948717e-07,
"loss": 0.0003,
"step": 9125
},
{
"epoch": 54.79,
"learning_rate": 5.565128205128204e-07,
"loss": 0.0003,
"step": 9150
},
{
"epoch": 54.94,
"learning_rate": 5.552307692307692e-07,
"loss": 0.0003,
"step": 9175
},
{
"epoch": 55.09,
"learning_rate": 5.539487179487179e-07,
"loss": 0.0002,
"step": 9200
},
{
"epoch": 55.24,
"learning_rate": 5.526666666666666e-07,
"loss": 0.0002,
"step": 9225
},
{
"epoch": 55.39,
"learning_rate": 5.513846153846153e-07,
"loss": 0.0002,
"step": 9250
},
{
"epoch": 55.54,
"learning_rate": 5.501025641025641e-07,
"loss": 0.0002,
"step": 9275
},
{
"epoch": 55.69,
"learning_rate": 5.488205128205128e-07,
"loss": 0.0003,
"step": 9300
},
{
"epoch": 55.84,
"learning_rate": 5.475384615384615e-07,
"loss": 0.0002,
"step": 9325
},
{
"epoch": 55.99,
"learning_rate": 5.462564102564102e-07,
"loss": 0.0002,
"step": 9350
},
{
"epoch": 56.14,
"learning_rate": 5.44974358974359e-07,
"loss": 0.0002,
"step": 9375
},
{
"epoch": 56.29,
"learning_rate": 5.436923076923077e-07,
"loss": 0.0002,
"step": 9400
},
{
"epoch": 56.44,
"learning_rate": 5.424102564102564e-07,
"loss": 0.0002,
"step": 9425
},
{
"epoch": 56.59,
"learning_rate": 5.411282051282051e-07,
"loss": 0.0002,
"step": 9450
},
{
"epoch": 56.74,
"learning_rate": 5.398461538461539e-07,
"loss": 0.0002,
"step": 9475
},
{
"epoch": 56.89,
"learning_rate": 5.385641025641026e-07,
"loss": 0.0002,
"step": 9500
},
{
"epoch": 57.04,
"learning_rate": 5.372820512820513e-07,
"loss": 0.0002,
"step": 9525
},
{
"epoch": 57.19,
"learning_rate": 5.36e-07,
"loss": 0.0002,
"step": 9550
},
{
"epoch": 57.34,
"learning_rate": 5.347179487179488e-07,
"loss": 0.0002,
"step": 9575
},
{
"epoch": 57.49,
"learning_rate": 5.334358974358975e-07,
"loss": 0.0002,
"step": 9600
},
{
"epoch": 57.63,
"learning_rate": 5.321538461538462e-07,
"loss": 0.0002,
"step": 9625
},
{
"epoch": 57.78,
"learning_rate": 5.308717948717949e-07,
"loss": 0.0002,
"step": 9650
},
{
"epoch": 57.93,
"learning_rate": 5.295897435897437e-07,
"loss": 0.0002,
"step": 9675
},
{
"epoch": 58.08,
"learning_rate": 5.283076923076923e-07,
"loss": 0.0002,
"step": 9700
},
{
"epoch": 58.23,
"learning_rate": 5.27025641025641e-07,
"loss": 0.0002,
"step": 9725
},
{
"epoch": 58.38,
"learning_rate": 5.257435897435897e-07,
"loss": 0.0002,
"step": 9750
},
{
"epoch": 58.53,
"learning_rate": 5.244615384615385e-07,
"loss": 0.0002,
"step": 9775
},
{
"epoch": 58.68,
"learning_rate": 5.231794871794871e-07,
"loss": 0.0002,
"step": 9800
},
{
"epoch": 58.83,
"learning_rate": 5.218974358974358e-07,
"loss": 0.0002,
"step": 9825
},
{
"epoch": 58.98,
"learning_rate": 5.206153846153845e-07,
"loss": 0.0002,
"step": 9850
},
{
"epoch": 59.13,
"learning_rate": 5.193333333333332e-07,
"loss": 0.0002,
"step": 9875
},
{
"epoch": 59.28,
"learning_rate": 5.18051282051282e-07,
"loss": 0.0002,
"step": 9900
},
{
"epoch": 59.43,
"learning_rate": 5.167692307692307e-07,
"loss": 0.0002,
"step": 9925
},
{
"epoch": 59.58,
"learning_rate": 5.154871794871794e-07,
"loss": 0.0002,
"step": 9950
},
{
"epoch": 59.73,
"learning_rate": 5.142051282051281e-07,
"loss": 0.0002,
"step": 9975
},
{
"epoch": 59.88,
"learning_rate": 5.129230769230769e-07,
"loss": 0.0002,
"step": 10000
},
{
"epoch": 59.88,
"eval_loss": 0.7664361596107483,
"eval_runtime": 596.9914,
"eval_samples_per_second": 0.702,
"eval_steps_per_second": 0.089,
"eval_wer": 42.69842142896431,
"step": 10000
}
],
"logging_steps": 25,
"max_steps": 20000,
"num_input_tokens_seen": 0,
"num_train_epochs": 120,
"save_steps": 1000,
"total_flos": 5.429983718493389e+20,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}