agnesluhtaru's picture
Model save
13067d9
raw
history blame
30.6 kB
{
"best_metric": 17.25914154421838,
"best_model_checkpoint": "models/whisper-medium-et-ERR2020/checkpoint-4000",
"epoch": 0.6,
"global_step": 6000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.6000000000000004e-07,
"loss": 1.538,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.600000000000001e-07,
"loss": 1.1539,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.46e-06,
"loss": 0.904,
"step": 75
},
{
"epoch": 0.01,
"learning_rate": 1.9600000000000003e-06,
"loss": 0.8887,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 2.46e-06,
"loss": 0.9294,
"step": 125
},
{
"epoch": 0.01,
"learning_rate": 2.96e-06,
"loss": 0.8664,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 3.46e-06,
"loss": 0.791,
"step": 175
},
{
"epoch": 0.02,
"learning_rate": 3.96e-06,
"loss": 0.7603,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 4.4600000000000005e-06,
"loss": 0.6953,
"step": 225
},
{
"epoch": 0.03,
"learning_rate": 4.960000000000001e-06,
"loss": 0.6213,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 5.460000000000001e-06,
"loss": 0.6298,
"step": 275
},
{
"epoch": 0.03,
"learning_rate": 5.9600000000000005e-06,
"loss": 0.5772,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 6.460000000000001e-06,
"loss": 0.5241,
"step": 325
},
{
"epoch": 0.04,
"learning_rate": 6.96e-06,
"loss": 0.5019,
"step": 350
},
{
"epoch": 0.04,
"learning_rate": 7.4600000000000006e-06,
"loss": 0.4879,
"step": 375
},
{
"epoch": 0.04,
"learning_rate": 7.960000000000002e-06,
"loss": 0.4563,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 8.46e-06,
"loss": 0.4498,
"step": 425
},
{
"epoch": 0.04,
"learning_rate": 8.96e-06,
"loss": 0.4136,
"step": 450
},
{
"epoch": 0.05,
"learning_rate": 9.460000000000001e-06,
"loss": 0.4136,
"step": 475
},
{
"epoch": 0.05,
"learning_rate": 9.960000000000001e-06,
"loss": 0.3882,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 9.975789473684211e-06,
"loss": 0.3335,
"step": 525
},
{
"epoch": 0.06,
"learning_rate": 9.949473684210526e-06,
"loss": 0.3474,
"step": 550
},
{
"epoch": 0.06,
"learning_rate": 9.923157894736844e-06,
"loss": 0.333,
"step": 575
},
{
"epoch": 0.06,
"learning_rate": 9.89684210526316e-06,
"loss": 0.3255,
"step": 600
},
{
"epoch": 0.06,
"learning_rate": 9.870526315789474e-06,
"loss": 0.2756,
"step": 625
},
{
"epoch": 0.07,
"learning_rate": 9.84421052631579e-06,
"loss": 0.281,
"step": 650
},
{
"epoch": 0.07,
"learning_rate": 9.817894736842106e-06,
"loss": 0.2666,
"step": 675
},
{
"epoch": 0.07,
"learning_rate": 9.791578947368422e-06,
"loss": 0.2592,
"step": 700
},
{
"epoch": 0.07,
"learning_rate": 9.765263157894737e-06,
"loss": 0.2504,
"step": 725
},
{
"epoch": 0.07,
"learning_rate": 9.738947368421054e-06,
"loss": 0.2398,
"step": 750
},
{
"epoch": 0.08,
"learning_rate": 9.712631578947369e-06,
"loss": 0.2283,
"step": 775
},
{
"epoch": 0.08,
"learning_rate": 9.686315789473684e-06,
"loss": 0.2377,
"step": 800
},
{
"epoch": 0.08,
"learning_rate": 9.66e-06,
"loss": 0.2262,
"step": 825
},
{
"epoch": 0.09,
"learning_rate": 9.633684210526316e-06,
"loss": 0.1998,
"step": 850
},
{
"epoch": 0.09,
"learning_rate": 9.607368421052632e-06,
"loss": 0.2151,
"step": 875
},
{
"epoch": 0.09,
"learning_rate": 9.581052631578947e-06,
"loss": 0.2035,
"step": 900
},
{
"epoch": 0.09,
"learning_rate": 9.554736842105264e-06,
"loss": 0.189,
"step": 925
},
{
"epoch": 0.1,
"learning_rate": 9.52842105263158e-06,
"loss": 0.1786,
"step": 950
},
{
"epoch": 0.1,
"learning_rate": 9.502105263157896e-06,
"loss": 0.1877,
"step": 975
},
{
"epoch": 0.1,
"learning_rate": 9.475789473684212e-06,
"loss": 0.1828,
"step": 1000
},
{
"epoch": 0.1,
"eval_loss": 0.3546636998653412,
"eval_runtime": 11312.8233,
"eval_samples_per_second": 0.699,
"eval_steps_per_second": 0.044,
"eval_wer": 20.882880202910577,
"step": 1000
},
{
"epoch": 0.1,
"learning_rate": 9.449473684210527e-06,
"loss": 0.1761,
"step": 1025
},
{
"epoch": 0.1,
"learning_rate": 9.423157894736842e-06,
"loss": 0.1627,
"step": 1050
},
{
"epoch": 0.11,
"learning_rate": 9.396842105263159e-06,
"loss": 0.1522,
"step": 1075
},
{
"epoch": 0.11,
"learning_rate": 9.370526315789474e-06,
"loss": 0.1618,
"step": 1100
},
{
"epoch": 0.11,
"learning_rate": 9.34421052631579e-06,
"loss": 0.1485,
"step": 1125
},
{
"epoch": 0.12,
"learning_rate": 9.317894736842105e-06,
"loss": 0.1703,
"step": 1150
},
{
"epoch": 0.12,
"learning_rate": 9.291578947368422e-06,
"loss": 0.148,
"step": 1175
},
{
"epoch": 0.12,
"learning_rate": 9.265263157894737e-06,
"loss": 0.152,
"step": 1200
},
{
"epoch": 0.12,
"learning_rate": 9.238947368421052e-06,
"loss": 0.1459,
"step": 1225
},
{
"epoch": 0.12,
"learning_rate": 9.21263157894737e-06,
"loss": 0.1618,
"step": 1250
},
{
"epoch": 0.13,
"learning_rate": 9.186315789473685e-06,
"loss": 0.1345,
"step": 1275
},
{
"epoch": 0.13,
"learning_rate": 9.16e-06,
"loss": 0.1467,
"step": 1300
},
{
"epoch": 0.13,
"learning_rate": 9.133684210526317e-06,
"loss": 0.1281,
"step": 1325
},
{
"epoch": 0.14,
"learning_rate": 9.107368421052632e-06,
"loss": 0.137,
"step": 1350
},
{
"epoch": 0.14,
"learning_rate": 9.081052631578949e-06,
"loss": 0.1389,
"step": 1375
},
{
"epoch": 0.14,
"learning_rate": 9.054736842105264e-06,
"loss": 0.1306,
"step": 1400
},
{
"epoch": 0.14,
"learning_rate": 9.02842105263158e-06,
"loss": 0.1378,
"step": 1425
},
{
"epoch": 0.14,
"learning_rate": 9.002105263157895e-06,
"loss": 0.1246,
"step": 1450
},
{
"epoch": 0.15,
"learning_rate": 8.97578947368421e-06,
"loss": 0.1325,
"step": 1475
},
{
"epoch": 0.15,
"learning_rate": 8.949473684210527e-06,
"loss": 0.1301,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 8.923157894736842e-06,
"loss": 0.1163,
"step": 1525
},
{
"epoch": 0.15,
"learning_rate": 8.896842105263159e-06,
"loss": 0.1233,
"step": 1550
},
{
"epoch": 0.16,
"learning_rate": 8.870526315789474e-06,
"loss": 0.1312,
"step": 1575
},
{
"epoch": 0.16,
"learning_rate": 8.84421052631579e-06,
"loss": 0.115,
"step": 1600
},
{
"epoch": 0.16,
"learning_rate": 8.817894736842107e-06,
"loss": 0.1176,
"step": 1625
},
{
"epoch": 0.17,
"learning_rate": 8.791578947368422e-06,
"loss": 0.109,
"step": 1650
},
{
"epoch": 0.17,
"learning_rate": 8.765263157894739e-06,
"loss": 0.1184,
"step": 1675
},
{
"epoch": 0.17,
"learning_rate": 8.738947368421053e-06,
"loss": 0.1055,
"step": 1700
},
{
"epoch": 0.17,
"learning_rate": 8.712631578947368e-06,
"loss": 0.0995,
"step": 1725
},
{
"epoch": 0.17,
"learning_rate": 8.686315789473685e-06,
"loss": 0.1171,
"step": 1750
},
{
"epoch": 0.18,
"learning_rate": 8.66e-06,
"loss": 0.1222,
"step": 1775
},
{
"epoch": 0.18,
"learning_rate": 8.633684210526317e-06,
"loss": 0.1066,
"step": 1800
},
{
"epoch": 0.18,
"learning_rate": 8.607368421052632e-06,
"loss": 0.1074,
"step": 1825
},
{
"epoch": 0.18,
"learning_rate": 8.581052631578948e-06,
"loss": 0.1064,
"step": 1850
},
{
"epoch": 0.19,
"learning_rate": 8.554736842105263e-06,
"loss": 0.1118,
"step": 1875
},
{
"epoch": 0.19,
"learning_rate": 8.528421052631578e-06,
"loss": 0.1008,
"step": 1900
},
{
"epoch": 0.19,
"learning_rate": 8.502105263157897e-06,
"loss": 0.1069,
"step": 1925
},
{
"epoch": 0.2,
"learning_rate": 8.475789473684212e-06,
"loss": 0.1126,
"step": 1950
},
{
"epoch": 0.2,
"learning_rate": 8.449473684210527e-06,
"loss": 0.1061,
"step": 1975
},
{
"epoch": 0.2,
"learning_rate": 8.423157894736843e-06,
"loss": 0.09,
"step": 2000
},
{
"epoch": 0.2,
"eval_loss": 0.3476346433162689,
"eval_runtime": 11632.5005,
"eval_samples_per_second": 0.68,
"eval_steps_per_second": 0.043,
"eval_wer": 19.00955840333174,
"step": 2000
},
{
"epoch": 0.2,
"learning_rate": 8.396842105263158e-06,
"loss": 0.1139,
"step": 2025
},
{
"epoch": 0.2,
"learning_rate": 8.370526315789475e-06,
"loss": 0.0995,
"step": 2050
},
{
"epoch": 0.21,
"learning_rate": 8.34421052631579e-06,
"loss": 0.1025,
"step": 2075
},
{
"epoch": 0.21,
"learning_rate": 8.317894736842107e-06,
"loss": 0.099,
"step": 2100
},
{
"epoch": 0.21,
"learning_rate": 8.291578947368422e-06,
"loss": 0.0932,
"step": 2125
},
{
"epoch": 0.21,
"learning_rate": 8.265263157894737e-06,
"loss": 0.0998,
"step": 2150
},
{
"epoch": 0.22,
"learning_rate": 8.238947368421053e-06,
"loss": 0.0948,
"step": 2175
},
{
"epoch": 0.22,
"learning_rate": 8.212631578947368e-06,
"loss": 0.0976,
"step": 2200
},
{
"epoch": 0.22,
"learning_rate": 8.186315789473685e-06,
"loss": 0.0936,
"step": 2225
},
{
"epoch": 0.23,
"learning_rate": 8.16e-06,
"loss": 0.0906,
"step": 2250
},
{
"epoch": 0.23,
"learning_rate": 8.133684210526316e-06,
"loss": 0.0918,
"step": 2275
},
{
"epoch": 0.23,
"learning_rate": 8.107368421052633e-06,
"loss": 0.0974,
"step": 2300
},
{
"epoch": 0.23,
"learning_rate": 8.081052631578948e-06,
"loss": 0.1078,
"step": 2325
},
{
"epoch": 0.23,
"learning_rate": 8.054736842105265e-06,
"loss": 0.0928,
"step": 2350
},
{
"epoch": 0.24,
"learning_rate": 8.02842105263158e-06,
"loss": 0.1048,
"step": 2375
},
{
"epoch": 0.24,
"learning_rate": 8.002105263157895e-06,
"loss": 0.0869,
"step": 2400
},
{
"epoch": 0.24,
"learning_rate": 7.975789473684211e-06,
"loss": 0.0872,
"step": 2425
},
{
"epoch": 0.24,
"learning_rate": 7.949473684210526e-06,
"loss": 0.104,
"step": 2450
},
{
"epoch": 0.25,
"learning_rate": 7.923157894736843e-06,
"loss": 0.0897,
"step": 2475
},
{
"epoch": 0.25,
"learning_rate": 7.896842105263158e-06,
"loss": 0.0798,
"step": 2500
},
{
"epoch": 0.25,
"learning_rate": 7.870526315789475e-06,
"loss": 0.0949,
"step": 2525
},
{
"epoch": 0.26,
"learning_rate": 7.84421052631579e-06,
"loss": 0.085,
"step": 2550
},
{
"epoch": 0.26,
"learning_rate": 7.817894736842105e-06,
"loss": 0.0841,
"step": 2575
},
{
"epoch": 0.26,
"learning_rate": 7.791578947368423e-06,
"loss": 0.0795,
"step": 2600
},
{
"epoch": 0.26,
"learning_rate": 7.765263157894738e-06,
"loss": 0.0849,
"step": 2625
},
{
"epoch": 0.27,
"learning_rate": 7.738947368421053e-06,
"loss": 0.0878,
"step": 2650
},
{
"epoch": 0.27,
"learning_rate": 7.71263157894737e-06,
"loss": 0.0858,
"step": 2675
},
{
"epoch": 0.27,
"learning_rate": 7.686315789473685e-06,
"loss": 0.0987,
"step": 2700
},
{
"epoch": 0.27,
"learning_rate": 7.660000000000001e-06,
"loss": 0.088,
"step": 2725
},
{
"epoch": 0.28,
"learning_rate": 7.633684210526316e-06,
"loss": 0.0801,
"step": 2750
},
{
"epoch": 0.28,
"learning_rate": 7.607368421052632e-06,
"loss": 0.089,
"step": 2775
},
{
"epoch": 0.28,
"learning_rate": 7.581052631578948e-06,
"loss": 0.0908,
"step": 2800
},
{
"epoch": 0.28,
"learning_rate": 7.554736842105264e-06,
"loss": 0.0686,
"step": 2825
},
{
"epoch": 0.28,
"learning_rate": 7.5284210526315794e-06,
"loss": 0.0873,
"step": 2850
},
{
"epoch": 0.29,
"learning_rate": 7.502105263157895e-06,
"loss": 0.0913,
"step": 2875
},
{
"epoch": 0.29,
"learning_rate": 7.475789473684211e-06,
"loss": 0.0844,
"step": 2900
},
{
"epoch": 0.29,
"learning_rate": 7.449473684210526e-06,
"loss": 0.0931,
"step": 2925
},
{
"epoch": 0.29,
"learning_rate": 7.4231578947368436e-06,
"loss": 0.0758,
"step": 2950
},
{
"epoch": 0.3,
"learning_rate": 7.3968421052631585e-06,
"loss": 0.0772,
"step": 2975
},
{
"epoch": 0.3,
"learning_rate": 7.370526315789474e-06,
"loss": 0.083,
"step": 3000
},
{
"epoch": 0.3,
"eval_loss": 0.3385966122150421,
"eval_runtime": 11712.6787,
"eval_samples_per_second": 0.676,
"eval_steps_per_second": 0.042,
"eval_wer": 18.130435803696543,
"step": 3000
},
{
"epoch": 0.3,
"learning_rate": 7.34421052631579e-06,
"loss": 0.0862,
"step": 3025
},
{
"epoch": 0.3,
"learning_rate": 7.317894736842106e-06,
"loss": 0.0801,
"step": 3050
},
{
"epoch": 0.31,
"learning_rate": 7.291578947368422e-06,
"loss": 0.0798,
"step": 3075
},
{
"epoch": 0.31,
"learning_rate": 7.265263157894738e-06,
"loss": 0.0854,
"step": 3100
},
{
"epoch": 0.31,
"learning_rate": 7.2389473684210534e-06,
"loss": 0.088,
"step": 3125
},
{
"epoch": 0.32,
"learning_rate": 7.212631578947369e-06,
"loss": 0.0875,
"step": 3150
},
{
"epoch": 0.32,
"learning_rate": 7.186315789473684e-06,
"loss": 0.0801,
"step": 3175
},
{
"epoch": 0.32,
"learning_rate": 7.16e-06,
"loss": 0.0828,
"step": 3200
},
{
"epoch": 0.32,
"learning_rate": 7.133684210526316e-06,
"loss": 0.082,
"step": 3225
},
{
"epoch": 0.33,
"learning_rate": 7.107368421052632e-06,
"loss": 0.0881,
"step": 3250
},
{
"epoch": 0.33,
"learning_rate": 7.0810526315789475e-06,
"loss": 0.0789,
"step": 3275
},
{
"epoch": 0.33,
"learning_rate": 7.054736842105264e-06,
"loss": 0.0835,
"step": 3300
},
{
"epoch": 0.33,
"learning_rate": 7.02842105263158e-06,
"loss": 0.0769,
"step": 3325
},
{
"epoch": 0.34,
"learning_rate": 7.002105263157896e-06,
"loss": 0.0923,
"step": 3350
},
{
"epoch": 0.34,
"learning_rate": 6.975789473684212e-06,
"loss": 0.0765,
"step": 3375
},
{
"epoch": 0.34,
"learning_rate": 6.9494736842105275e-06,
"loss": 0.0867,
"step": 3400
},
{
"epoch": 0.34,
"learning_rate": 6.9231578947368424e-06,
"loss": 0.0761,
"step": 3425
},
{
"epoch": 0.34,
"learning_rate": 6.896842105263158e-06,
"loss": 0.0863,
"step": 3450
},
{
"epoch": 0.35,
"learning_rate": 6.870526315789474e-06,
"loss": 0.0722,
"step": 3475
},
{
"epoch": 0.35,
"learning_rate": 6.84421052631579e-06,
"loss": 0.0799,
"step": 3500
},
{
"epoch": 0.35,
"learning_rate": 6.817894736842106e-06,
"loss": 0.0826,
"step": 3525
},
{
"epoch": 0.35,
"learning_rate": 6.792631578947368e-06,
"loss": 0.0774,
"step": 3550
},
{
"epoch": 0.36,
"learning_rate": 6.766315789473685e-06,
"loss": 0.0791,
"step": 3575
},
{
"epoch": 0.36,
"learning_rate": 6.740000000000001e-06,
"loss": 0.0866,
"step": 3600
},
{
"epoch": 0.36,
"learning_rate": 6.7136842105263165e-06,
"loss": 0.0798,
"step": 3625
},
{
"epoch": 0.36,
"learning_rate": 6.687368421052632e-06,
"loss": 0.0775,
"step": 3650
},
{
"epoch": 0.37,
"learning_rate": 6.661052631578948e-06,
"loss": 0.0833,
"step": 3675
},
{
"epoch": 0.37,
"learning_rate": 6.634736842105264e-06,
"loss": 0.0795,
"step": 3700
},
{
"epoch": 0.37,
"learning_rate": 6.60842105263158e-06,
"loss": 0.0849,
"step": 3725
},
{
"epoch": 0.38,
"learning_rate": 6.582105263157896e-06,
"loss": 0.0768,
"step": 3750
},
{
"epoch": 0.38,
"learning_rate": 6.5557894736842106e-06,
"loss": 0.0685,
"step": 3775
},
{
"epoch": 0.38,
"learning_rate": 6.529473684210526e-06,
"loss": 0.082,
"step": 3800
},
{
"epoch": 0.38,
"learning_rate": 6.503157894736842e-06,
"loss": 0.0738,
"step": 3825
},
{
"epoch": 0.39,
"learning_rate": 6.476842105263158e-06,
"loss": 0.0822,
"step": 3850
},
{
"epoch": 0.39,
"learning_rate": 6.450526315789474e-06,
"loss": 0.0796,
"step": 3875
},
{
"epoch": 0.39,
"learning_rate": 6.42421052631579e-06,
"loss": 0.0691,
"step": 3900
},
{
"epoch": 0.39,
"learning_rate": 6.397894736842106e-06,
"loss": 0.0769,
"step": 3925
},
{
"epoch": 0.4,
"learning_rate": 6.371578947368422e-06,
"loss": 0.0787,
"step": 3950
},
{
"epoch": 0.4,
"learning_rate": 6.345263157894738e-06,
"loss": 0.0846,
"step": 3975
},
{
"epoch": 0.4,
"learning_rate": 6.318947368421054e-06,
"loss": 0.0765,
"step": 4000
},
{
"epoch": 0.4,
"eval_loss": 0.3365060091018677,
"eval_runtime": 11429.3476,
"eval_samples_per_second": 0.692,
"eval_steps_per_second": 0.043,
"eval_wer": 17.25914154421838,
"step": 4000
},
{
"epoch": 0.4,
"learning_rate": 6.292631578947369e-06,
"loss": 0.0736,
"step": 4025
},
{
"epoch": 0.41,
"learning_rate": 6.266315789473685e-06,
"loss": 0.0884,
"step": 4050
},
{
"epoch": 0.41,
"learning_rate": 6.24e-06,
"loss": 0.0795,
"step": 4075
},
{
"epoch": 0.41,
"learning_rate": 6.213684210526316e-06,
"loss": 0.0778,
"step": 4100
},
{
"epoch": 0.41,
"learning_rate": 6.187368421052632e-06,
"loss": 0.0757,
"step": 4125
},
{
"epoch": 0.41,
"learning_rate": 6.161052631578948e-06,
"loss": 0.0866,
"step": 4150
},
{
"epoch": 0.42,
"learning_rate": 6.134736842105264e-06,
"loss": 0.0767,
"step": 4175
},
{
"epoch": 0.42,
"learning_rate": 6.108421052631579e-06,
"loss": 0.0725,
"step": 4200
},
{
"epoch": 0.42,
"learning_rate": 6.0821052631578945e-06,
"loss": 0.0757,
"step": 4225
},
{
"epoch": 0.42,
"learning_rate": 6.055789473684212e-06,
"loss": 0.08,
"step": 4250
},
{
"epoch": 0.43,
"learning_rate": 6.029473684210527e-06,
"loss": 0.0723,
"step": 4275
},
{
"epoch": 0.43,
"learning_rate": 6.003157894736843e-06,
"loss": 0.0801,
"step": 4300
},
{
"epoch": 0.43,
"learning_rate": 5.976842105263159e-06,
"loss": 0.0697,
"step": 4325
},
{
"epoch": 0.43,
"learning_rate": 5.950526315789474e-06,
"loss": 0.0897,
"step": 4350
},
{
"epoch": 0.44,
"learning_rate": 5.92421052631579e-06,
"loss": 0.0798,
"step": 4375
},
{
"epoch": 0.44,
"learning_rate": 5.897894736842106e-06,
"loss": 0.0733,
"step": 4400
},
{
"epoch": 0.44,
"learning_rate": 5.871578947368422e-06,
"loss": 0.0941,
"step": 4425
},
{
"epoch": 0.45,
"learning_rate": 5.845263157894737e-06,
"loss": 0.0676,
"step": 4450
},
{
"epoch": 0.45,
"learning_rate": 5.818947368421053e-06,
"loss": 0.0797,
"step": 4475
},
{
"epoch": 0.45,
"learning_rate": 5.7926315789473685e-06,
"loss": 0.0684,
"step": 4500
},
{
"epoch": 0.45,
"learning_rate": 5.766315789473684e-06,
"loss": 0.0832,
"step": 4525
},
{
"epoch": 0.46,
"learning_rate": 5.74e-06,
"loss": 0.0773,
"step": 4550
},
{
"epoch": 0.46,
"learning_rate": 5.713684210526316e-06,
"loss": 0.0697,
"step": 4575
},
{
"epoch": 0.46,
"learning_rate": 5.687368421052633e-06,
"loss": 0.0806,
"step": 4600
},
{
"epoch": 0.46,
"learning_rate": 5.661052631578948e-06,
"loss": 0.0752,
"step": 4625
},
{
"epoch": 0.47,
"learning_rate": 5.634736842105264e-06,
"loss": 0.0678,
"step": 4650
},
{
"epoch": 0.47,
"learning_rate": 5.60842105263158e-06,
"loss": 0.0754,
"step": 4675
},
{
"epoch": 0.47,
"learning_rate": 5.582105263157895e-06,
"loss": 0.0694,
"step": 4700
},
{
"epoch": 0.47,
"learning_rate": 5.555789473684211e-06,
"loss": 0.0731,
"step": 4725
},
{
"epoch": 0.47,
"learning_rate": 5.529473684210527e-06,
"loss": 0.0779,
"step": 4750
},
{
"epoch": 0.48,
"learning_rate": 5.5031578947368425e-06,
"loss": 0.0847,
"step": 4775
},
{
"epoch": 0.48,
"learning_rate": 5.476842105263158e-06,
"loss": 0.0719,
"step": 4800
},
{
"epoch": 0.48,
"learning_rate": 5.450526315789474e-06,
"loss": 0.0697,
"step": 4825
},
{
"epoch": 0.48,
"learning_rate": 5.42421052631579e-06,
"loss": 0.0643,
"step": 4850
},
{
"epoch": 0.49,
"learning_rate": 5.397894736842105e-06,
"loss": 0.0712,
"step": 4875
},
{
"epoch": 0.49,
"learning_rate": 5.371578947368421e-06,
"loss": 0.078,
"step": 4900
},
{
"epoch": 0.49,
"learning_rate": 5.345263157894738e-06,
"loss": 0.0843,
"step": 4925
},
{
"epoch": 0.49,
"learning_rate": 5.318947368421053e-06,
"loss": 0.0792,
"step": 4950
},
{
"epoch": 0.5,
"learning_rate": 5.292631578947369e-06,
"loss": 0.0722,
"step": 4975
},
{
"epoch": 0.5,
"learning_rate": 5.266315789473685e-06,
"loss": 0.0592,
"step": 5000
},
{
"epoch": 0.5,
"eval_loss": 0.3534376323223114,
"eval_runtime": 13205.6363,
"eval_samples_per_second": 0.599,
"eval_steps_per_second": 0.037,
"eval_wer": 19.021300913567295,
"step": 5000
},
{
"epoch": 0.5,
"learning_rate": 5.240000000000001e-06,
"loss": 0.0754,
"step": 5025
},
{
"epoch": 0.51,
"learning_rate": 5.2136842105263165e-06,
"loss": 0.0759,
"step": 5050
},
{
"epoch": 0.51,
"learning_rate": 5.187368421052632e-06,
"loss": 0.0667,
"step": 5075
},
{
"epoch": 0.51,
"learning_rate": 5.161052631578948e-06,
"loss": 0.0649,
"step": 5100
},
{
"epoch": 0.51,
"learning_rate": 5.134736842105263e-06,
"loss": 0.0723,
"step": 5125
},
{
"epoch": 0.52,
"learning_rate": 5.108421052631579e-06,
"loss": 0.0756,
"step": 5150
},
{
"epoch": 0.52,
"learning_rate": 5.082105263157895e-06,
"loss": 0.0787,
"step": 5175
},
{
"epoch": 0.52,
"learning_rate": 5.0557894736842105e-06,
"loss": 0.0752,
"step": 5200
},
{
"epoch": 0.52,
"learning_rate": 5.029473684210526e-06,
"loss": 0.0755,
"step": 5225
},
{
"epoch": 0.53,
"learning_rate": 5.003157894736842e-06,
"loss": 0.0773,
"step": 5250
},
{
"epoch": 0.53,
"learning_rate": 4.976842105263158e-06,
"loss": 0.063,
"step": 5275
},
{
"epoch": 0.53,
"learning_rate": 4.950526315789474e-06,
"loss": 0.0701,
"step": 5300
},
{
"epoch": 0.53,
"learning_rate": 4.92421052631579e-06,
"loss": 0.0673,
"step": 5325
},
{
"epoch": 0.54,
"learning_rate": 4.897894736842106e-06,
"loss": 0.0742,
"step": 5350
},
{
"epoch": 0.54,
"learning_rate": 4.871578947368421e-06,
"loss": 0.0754,
"step": 5375
},
{
"epoch": 0.54,
"learning_rate": 4.845263157894737e-06,
"loss": 0.0878,
"step": 5400
},
{
"epoch": 0.54,
"learning_rate": 4.818947368421053e-06,
"loss": 0.0663,
"step": 5425
},
{
"epoch": 0.55,
"learning_rate": 4.792631578947369e-06,
"loss": 0.0711,
"step": 5450
},
{
"epoch": 0.55,
"learning_rate": 4.7663157894736845e-06,
"loss": 0.0636,
"step": 5475
},
{
"epoch": 0.55,
"learning_rate": 4.74e-06,
"loss": 0.0719,
"step": 5500
},
{
"epoch": 0.55,
"learning_rate": 4.713684210526316e-06,
"loss": 0.0711,
"step": 5525
},
{
"epoch": 0.56,
"learning_rate": 4.687368421052632e-06,
"loss": 0.0656,
"step": 5550
},
{
"epoch": 0.56,
"learning_rate": 4.661052631578948e-06,
"loss": 0.0741,
"step": 5575
},
{
"epoch": 0.56,
"learning_rate": 4.634736842105264e-06,
"loss": 0.0639,
"step": 5600
},
{
"epoch": 0.56,
"learning_rate": 4.6084210526315794e-06,
"loss": 0.0716,
"step": 5625
},
{
"epoch": 0.56,
"learning_rate": 4.582105263157895e-06,
"loss": 0.0735,
"step": 5650
},
{
"epoch": 0.57,
"learning_rate": 4.55578947368421e-06,
"loss": 0.0677,
"step": 5675
},
{
"epoch": 0.57,
"learning_rate": 4.529473684210527e-06,
"loss": 0.0615,
"step": 5700
},
{
"epoch": 0.57,
"learning_rate": 4.503157894736843e-06,
"loss": 0.0692,
"step": 5725
},
{
"epoch": 0.57,
"learning_rate": 4.4768421052631585e-06,
"loss": 0.0764,
"step": 5750
},
{
"epoch": 0.58,
"learning_rate": 4.450526315789474e-06,
"loss": 0.0661,
"step": 5775
},
{
"epoch": 0.58,
"learning_rate": 4.424210526315789e-06,
"loss": 0.0703,
"step": 5800
},
{
"epoch": 0.58,
"learning_rate": 4.397894736842105e-06,
"loss": 0.075,
"step": 5825
},
{
"epoch": 0.58,
"learning_rate": 4.371578947368421e-06,
"loss": 0.0679,
"step": 5850
},
{
"epoch": 0.59,
"learning_rate": 4.345263157894738e-06,
"loss": 0.0685,
"step": 5875
},
{
"epoch": 0.59,
"learning_rate": 4.3189473684210535e-06,
"loss": 0.0673,
"step": 5900
},
{
"epoch": 0.59,
"learning_rate": 4.2926315789473684e-06,
"loss": 0.0702,
"step": 5925
},
{
"epoch": 0.59,
"learning_rate": 4.266315789473684e-06,
"loss": 0.0726,
"step": 5950
},
{
"epoch": 0.6,
"learning_rate": 4.24e-06,
"loss": 0.0595,
"step": 5975
},
{
"epoch": 0.6,
"learning_rate": 4.213684210526316e-06,
"loss": 0.0672,
"step": 6000
},
{
"epoch": 0.6,
"eval_loss": 0.36219528317451477,
"eval_runtime": 13383.7299,
"eval_samples_per_second": 0.591,
"eval_steps_per_second": 0.037,
"eval_wer": 18.426347061632523,
"step": 6000
}
],
"max_steps": 10000,
"num_train_epochs": 9223372036854775807,
"total_flos": 1.9595614224384e+20,
"trial_name": null,
"trial_params": null
}