whisper-small-it / trainer_state.json
luigisaetta's picture
End of training
45fd2d9
{
"best_metric": 0.11720343274797772,
"best_model_checkpoint": "./whisper-small-it/checkpoint-3000",
"epoch": 8.388935456831517,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.04,
"learning_rate": 4.4e-07,
"loss": 1.1848,
"step": 25
},
{
"epoch": 0.08,
"learning_rate": 9.400000000000001e-07,
"loss": 1.0188,
"step": 50
},
{
"epoch": 0.13,
"learning_rate": 1.44e-06,
"loss": 0.6728,
"step": 75
},
{
"epoch": 0.17,
"learning_rate": 1.94e-06,
"loss": 0.3915,
"step": 100
},
{
"epoch": 0.21,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.3182,
"step": 125
},
{
"epoch": 0.25,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.2981,
"step": 150
},
{
"epoch": 0.29,
"learning_rate": 3.44e-06,
"loss": 0.2823,
"step": 175
},
{
"epoch": 0.34,
"learning_rate": 3.94e-06,
"loss": 0.2744,
"step": 200
},
{
"epoch": 0.38,
"learning_rate": 4.440000000000001e-06,
"loss": 0.2547,
"step": 225
},
{
"epoch": 0.42,
"learning_rate": 4.94e-06,
"loss": 0.2547,
"step": 250
},
{
"epoch": 0.46,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.2367,
"step": 275
},
{
"epoch": 0.5,
"learning_rate": 5.94e-06,
"loss": 0.2369,
"step": 300
},
{
"epoch": 0.54,
"learning_rate": 6.440000000000001e-06,
"loss": 0.2288,
"step": 325
},
{
"epoch": 0.59,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.2285,
"step": 350
},
{
"epoch": 0.63,
"learning_rate": 7.440000000000001e-06,
"loss": 0.2209,
"step": 375
},
{
"epoch": 0.67,
"learning_rate": 7.94e-06,
"loss": 0.2227,
"step": 400
},
{
"epoch": 0.71,
"learning_rate": 8.44e-06,
"loss": 0.2187,
"step": 425
},
{
"epoch": 0.75,
"learning_rate": 8.94e-06,
"loss": 0.2099,
"step": 450
},
{
"epoch": 0.8,
"learning_rate": 9.440000000000001e-06,
"loss": 0.2082,
"step": 475
},
{
"epoch": 0.84,
"learning_rate": 9.940000000000001e-06,
"loss": 0.2008,
"step": 500
},
{
"epoch": 0.88,
"learning_rate": 9.951111111111111e-06,
"loss": 0.2049,
"step": 525
},
{
"epoch": 0.92,
"learning_rate": 9.895555555555557e-06,
"loss": 0.2014,
"step": 550
},
{
"epoch": 0.96,
"learning_rate": 9.84e-06,
"loss": 0.1973,
"step": 575
},
{
"epoch": 1.01,
"learning_rate": 9.784444444444445e-06,
"loss": 0.1968,
"step": 600
},
{
"epoch": 1.05,
"learning_rate": 9.72888888888889e-06,
"loss": 0.1576,
"step": 625
},
{
"epoch": 1.09,
"learning_rate": 9.673333333333334e-06,
"loss": 0.153,
"step": 650
},
{
"epoch": 1.13,
"learning_rate": 9.617777777777778e-06,
"loss": 0.1539,
"step": 675
},
{
"epoch": 1.17,
"learning_rate": 9.562222222222223e-06,
"loss": 0.1524,
"step": 700
},
{
"epoch": 1.22,
"learning_rate": 9.506666666666667e-06,
"loss": 0.1551,
"step": 725
},
{
"epoch": 1.26,
"learning_rate": 9.451111111111112e-06,
"loss": 0.1523,
"step": 750
},
{
"epoch": 1.3,
"learning_rate": 9.395555555555556e-06,
"loss": 0.1506,
"step": 775
},
{
"epoch": 1.34,
"learning_rate": 9.340000000000002e-06,
"loss": 0.1549,
"step": 800
},
{
"epoch": 1.38,
"learning_rate": 9.284444444444444e-06,
"loss": 0.1501,
"step": 825
},
{
"epoch": 1.43,
"learning_rate": 9.22888888888889e-06,
"loss": 0.1483,
"step": 850
},
{
"epoch": 1.47,
"learning_rate": 9.173333333333334e-06,
"loss": 0.152,
"step": 875
},
{
"epoch": 1.51,
"learning_rate": 9.117777777777778e-06,
"loss": 0.1475,
"step": 900
},
{
"epoch": 1.55,
"learning_rate": 9.062222222222224e-06,
"loss": 0.1461,
"step": 925
},
{
"epoch": 1.59,
"learning_rate": 9.006666666666666e-06,
"loss": 0.1479,
"step": 950
},
{
"epoch": 1.64,
"learning_rate": 8.951111111111112e-06,
"loss": 0.1464,
"step": 975
},
{
"epoch": 1.68,
"learning_rate": 8.895555555555556e-06,
"loss": 0.1441,
"step": 1000
},
{
"epoch": 1.68,
"eval_loss": 0.19120633602142334,
"eval_runtime": 1809.5025,
"eval_samples_per_second": 8.288,
"eval_steps_per_second": 0.13,
"eval_wer": 0.1255868383953135,
"step": 1000
},
{
"epoch": 1.72,
"learning_rate": 8.84e-06,
"loss": 0.1448,
"step": 1025
},
{
"epoch": 1.76,
"learning_rate": 8.784444444444446e-06,
"loss": 0.1397,
"step": 1050
},
{
"epoch": 1.8,
"learning_rate": 8.72888888888889e-06,
"loss": 0.1398,
"step": 1075
},
{
"epoch": 1.84,
"learning_rate": 8.673333333333334e-06,
"loss": 0.1396,
"step": 1100
},
{
"epoch": 1.89,
"learning_rate": 8.617777777777778e-06,
"loss": 0.1385,
"step": 1125
},
{
"epoch": 1.93,
"learning_rate": 8.562222222222224e-06,
"loss": 0.1372,
"step": 1150
},
{
"epoch": 1.97,
"learning_rate": 8.506666666666668e-06,
"loss": 0.1396,
"step": 1175
},
{
"epoch": 2.01,
"learning_rate": 8.451111111111112e-06,
"loss": 0.1267,
"step": 1200
},
{
"epoch": 2.06,
"learning_rate": 8.395555555555557e-06,
"loss": 0.0982,
"step": 1225
},
{
"epoch": 2.1,
"learning_rate": 8.34e-06,
"loss": 0.0999,
"step": 1250
},
{
"epoch": 2.14,
"learning_rate": 8.284444444444446e-06,
"loss": 0.0978,
"step": 1275
},
{
"epoch": 2.18,
"learning_rate": 8.22888888888889e-06,
"loss": 0.1003,
"step": 1300
},
{
"epoch": 2.22,
"learning_rate": 8.173333333333334e-06,
"loss": 0.0985,
"step": 1325
},
{
"epoch": 2.26,
"learning_rate": 8.11777777777778e-06,
"loss": 0.0974,
"step": 1350
},
{
"epoch": 2.31,
"learning_rate": 8.062222222222222e-06,
"loss": 0.0998,
"step": 1375
},
{
"epoch": 2.35,
"learning_rate": 8.006666666666667e-06,
"loss": 0.0988,
"step": 1400
},
{
"epoch": 2.39,
"learning_rate": 7.951111111111111e-06,
"loss": 0.096,
"step": 1425
},
{
"epoch": 2.43,
"learning_rate": 7.895555555555557e-06,
"loss": 0.0985,
"step": 1450
},
{
"epoch": 2.47,
"learning_rate": 7.840000000000001e-06,
"loss": 0.0981,
"step": 1475
},
{
"epoch": 2.52,
"learning_rate": 7.784444444444445e-06,
"loss": 0.0962,
"step": 1500
},
{
"epoch": 2.56,
"learning_rate": 7.72888888888889e-06,
"loss": 0.0952,
"step": 1525
},
{
"epoch": 2.6,
"learning_rate": 7.673333333333333e-06,
"loss": 0.0959,
"step": 1550
},
{
"epoch": 2.64,
"learning_rate": 7.617777777777778e-06,
"loss": 0.0983,
"step": 1575
},
{
"epoch": 2.68,
"learning_rate": 7.562222222222223e-06,
"loss": 0.0977,
"step": 1600
},
{
"epoch": 2.73,
"learning_rate": 7.506666666666668e-06,
"loss": 0.0933,
"step": 1625
},
{
"epoch": 2.77,
"learning_rate": 7.451111111111111e-06,
"loss": 0.0957,
"step": 1650
},
{
"epoch": 2.81,
"learning_rate": 7.395555555555556e-06,
"loss": 0.0941,
"step": 1675
},
{
"epoch": 2.85,
"learning_rate": 7.340000000000001e-06,
"loss": 0.0963,
"step": 1700
},
{
"epoch": 2.89,
"learning_rate": 7.284444444444445e-06,
"loss": 0.099,
"step": 1725
},
{
"epoch": 2.94,
"learning_rate": 7.22888888888889e-06,
"loss": 0.0938,
"step": 1750
},
{
"epoch": 2.98,
"learning_rate": 7.173333333333335e-06,
"loss": 0.0931,
"step": 1775
},
{
"epoch": 3.02,
"learning_rate": 7.117777777777778e-06,
"loss": 0.0871,
"step": 1800
},
{
"epoch": 3.06,
"learning_rate": 7.062222222222223e-06,
"loss": 0.0695,
"step": 1825
},
{
"epoch": 3.1,
"learning_rate": 7.006666666666667e-06,
"loss": 0.0693,
"step": 1850
},
{
"epoch": 3.15,
"learning_rate": 6.951111111111112e-06,
"loss": 0.0672,
"step": 1875
},
{
"epoch": 3.19,
"learning_rate": 6.8955555555555565e-06,
"loss": 0.0656,
"step": 1900
},
{
"epoch": 3.23,
"learning_rate": 6.8400000000000014e-06,
"loss": 0.0661,
"step": 1925
},
{
"epoch": 3.27,
"learning_rate": 6.784444444444445e-06,
"loss": 0.0667,
"step": 1950
},
{
"epoch": 3.31,
"learning_rate": 6.7288888888888895e-06,
"loss": 0.0646,
"step": 1975
},
{
"epoch": 3.36,
"learning_rate": 6.6733333333333335e-06,
"loss": 0.0653,
"step": 2000
},
{
"epoch": 3.36,
"eval_loss": 0.18448910117149353,
"eval_runtime": 1809.4982,
"eval_samples_per_second": 8.288,
"eval_steps_per_second": 0.13,
"eval_wer": 0.11822997221499842,
"step": 2000
},
{
"epoch": 3.4,
"learning_rate": 6.617777777777778e-06,
"loss": 0.0684,
"step": 2025
},
{
"epoch": 3.44,
"learning_rate": 6.562222222222223e-06,
"loss": 0.0671,
"step": 2050
},
{
"epoch": 3.48,
"learning_rate": 6.5066666666666665e-06,
"loss": 0.0674,
"step": 2075
},
{
"epoch": 3.52,
"learning_rate": 6.451111111111111e-06,
"loss": 0.0695,
"step": 2100
},
{
"epoch": 3.56,
"learning_rate": 6.395555555555556e-06,
"loss": 0.0661,
"step": 2125
},
{
"epoch": 3.61,
"learning_rate": 6.34e-06,
"loss": 0.0679,
"step": 2150
},
{
"epoch": 3.65,
"learning_rate": 6.284444444444445e-06,
"loss": 0.0659,
"step": 2175
},
{
"epoch": 3.69,
"learning_rate": 6.22888888888889e-06,
"loss": 0.0668,
"step": 2200
},
{
"epoch": 3.73,
"learning_rate": 6.173333333333333e-06,
"loss": 0.0675,
"step": 2225
},
{
"epoch": 3.77,
"learning_rate": 6.117777777777778e-06,
"loss": 0.0667,
"step": 2250
},
{
"epoch": 3.82,
"learning_rate": 6.062222222222223e-06,
"loss": 0.0689,
"step": 2275
},
{
"epoch": 3.86,
"learning_rate": 6.006666666666667e-06,
"loss": 0.0675,
"step": 2300
},
{
"epoch": 3.9,
"learning_rate": 5.951111111111112e-06,
"loss": 0.0667,
"step": 2325
},
{
"epoch": 3.94,
"learning_rate": 5.895555555555557e-06,
"loss": 0.0661,
"step": 2350
},
{
"epoch": 3.98,
"learning_rate": 5.84e-06,
"loss": 0.0681,
"step": 2375
},
{
"epoch": 4.03,
"learning_rate": 5.784444444444445e-06,
"loss": 0.0596,
"step": 2400
},
{
"epoch": 4.07,
"learning_rate": 5.72888888888889e-06,
"loss": 0.0486,
"step": 2425
},
{
"epoch": 4.11,
"learning_rate": 5.673333333333334e-06,
"loss": 0.0466,
"step": 2450
},
{
"epoch": 4.15,
"learning_rate": 5.617777777777779e-06,
"loss": 0.0484,
"step": 2475
},
{
"epoch": 4.19,
"learning_rate": 5.562222222222222e-06,
"loss": 0.0465,
"step": 2500
},
{
"epoch": 4.24,
"learning_rate": 5.506666666666667e-06,
"loss": 0.0481,
"step": 2525
},
{
"epoch": 4.28,
"learning_rate": 5.451111111111112e-06,
"loss": 0.0476,
"step": 2550
},
{
"epoch": 4.32,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.0474,
"step": 2575
},
{
"epoch": 4.36,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.0471,
"step": 2600
},
{
"epoch": 4.4,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.0491,
"step": 2625
},
{
"epoch": 4.45,
"learning_rate": 5.228888888888889e-06,
"loss": 0.0469,
"step": 2650
},
{
"epoch": 4.49,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.0491,
"step": 2675
},
{
"epoch": 4.53,
"learning_rate": 5.117777777777778e-06,
"loss": 0.0467,
"step": 2700
},
{
"epoch": 4.57,
"learning_rate": 5.062222222222222e-06,
"loss": 0.0493,
"step": 2725
},
{
"epoch": 4.61,
"learning_rate": 5.006666666666667e-06,
"loss": 0.0463,
"step": 2750
},
{
"epoch": 4.66,
"learning_rate": 4.951111111111111e-06,
"loss": 0.0456,
"step": 2775
},
{
"epoch": 4.7,
"learning_rate": 4.895555555555556e-06,
"loss": 0.0487,
"step": 2800
},
{
"epoch": 4.74,
"learning_rate": 4.84e-06,
"loss": 0.0463,
"step": 2825
},
{
"epoch": 4.78,
"learning_rate": 4.784444444444445e-06,
"loss": 0.0488,
"step": 2850
},
{
"epoch": 4.82,
"learning_rate": 4.728888888888889e-06,
"loss": 0.0464,
"step": 2875
},
{
"epoch": 4.87,
"learning_rate": 4.673333333333333e-06,
"loss": 0.0477,
"step": 2900
},
{
"epoch": 4.91,
"learning_rate": 4.617777777777778e-06,
"loss": 0.0475,
"step": 2925
},
{
"epoch": 4.95,
"learning_rate": 4.562222222222222e-06,
"loss": 0.0463,
"step": 2950
},
{
"epoch": 4.99,
"learning_rate": 4.506666666666667e-06,
"loss": 0.0453,
"step": 2975
},
{
"epoch": 5.03,
"learning_rate": 4.451111111111112e-06,
"loss": 0.0374,
"step": 3000
},
{
"epoch": 5.03,
"eval_loss": 0.19193173944950104,
"eval_runtime": 1809.3674,
"eval_samples_per_second": 8.289,
"eval_steps_per_second": 0.13,
"eval_wer": 0.11720343274797772,
"step": 3000
},
{
"epoch": 5.08,
"learning_rate": 4.395555555555556e-06,
"loss": 0.0338,
"step": 3025
},
{
"epoch": 5.12,
"learning_rate": 4.34e-06,
"loss": 0.0327,
"step": 3050
},
{
"epoch": 5.16,
"learning_rate": 4.284444444444445e-06,
"loss": 0.0331,
"step": 3075
},
{
"epoch": 5.2,
"learning_rate": 4.228888888888889e-06,
"loss": 0.0346,
"step": 3100
},
{
"epoch": 5.24,
"learning_rate": 4.173333333333334e-06,
"loss": 0.0332,
"step": 3125
},
{
"epoch": 5.28,
"learning_rate": 4.117777777777779e-06,
"loss": 0.0326,
"step": 3150
},
{
"epoch": 5.33,
"learning_rate": 4.062222222222223e-06,
"loss": 0.0336,
"step": 3175
},
{
"epoch": 5.37,
"learning_rate": 4.006666666666667e-06,
"loss": 0.0334,
"step": 3200
},
{
"epoch": 5.41,
"learning_rate": 3.951111111111112e-06,
"loss": 0.0332,
"step": 3225
},
{
"epoch": 5.45,
"learning_rate": 3.895555555555556e-06,
"loss": 0.0335,
"step": 3250
},
{
"epoch": 5.49,
"learning_rate": 3.8400000000000005e-06,
"loss": 0.0327,
"step": 3275
},
{
"epoch": 5.54,
"learning_rate": 3.784444444444445e-06,
"loss": 0.0336,
"step": 3300
},
{
"epoch": 5.58,
"learning_rate": 3.728888888888889e-06,
"loss": 0.0336,
"step": 3325
},
{
"epoch": 5.62,
"learning_rate": 3.673333333333334e-06,
"loss": 0.0325,
"step": 3350
},
{
"epoch": 5.66,
"learning_rate": 3.617777777777778e-06,
"loss": 0.0328,
"step": 3375
},
{
"epoch": 5.7,
"learning_rate": 3.5622222222222224e-06,
"loss": 0.0332,
"step": 3400
},
{
"epoch": 5.75,
"learning_rate": 3.5066666666666673e-06,
"loss": 0.0347,
"step": 3425
},
{
"epoch": 5.79,
"learning_rate": 3.4511111111111113e-06,
"loss": 0.0333,
"step": 3450
},
{
"epoch": 5.83,
"learning_rate": 3.3955555555555558e-06,
"loss": 0.0346,
"step": 3475
},
{
"epoch": 5.87,
"learning_rate": 3.3400000000000006e-06,
"loss": 0.0334,
"step": 3500
},
{
"epoch": 5.91,
"learning_rate": 3.2844444444444447e-06,
"loss": 0.0342,
"step": 3525
},
{
"epoch": 5.96,
"learning_rate": 3.228888888888889e-06,
"loss": 0.035,
"step": 3550
},
{
"epoch": 6.0,
"learning_rate": 3.173333333333334e-06,
"loss": 0.0314,
"step": 3575
},
{
"epoch": 6.04,
"learning_rate": 3.117777777777778e-06,
"loss": 0.0264,
"step": 3600
},
{
"epoch": 6.08,
"learning_rate": 3.0622222222222225e-06,
"loss": 0.0247,
"step": 3625
},
{
"epoch": 6.12,
"learning_rate": 3.0066666666666674e-06,
"loss": 0.0229,
"step": 3650
},
{
"epoch": 6.17,
"learning_rate": 2.9511111111111114e-06,
"loss": 0.0244,
"step": 3675
},
{
"epoch": 6.21,
"learning_rate": 2.895555555555556e-06,
"loss": 0.0249,
"step": 3700
},
{
"epoch": 6.25,
"learning_rate": 2.84e-06,
"loss": 0.0233,
"step": 3725
},
{
"epoch": 6.29,
"learning_rate": 2.784444444444445e-06,
"loss": 0.024,
"step": 3750
},
{
"epoch": 6.33,
"learning_rate": 2.7288888888888893e-06,
"loss": 0.0235,
"step": 3775
},
{
"epoch": 6.38,
"learning_rate": 2.6733333333333333e-06,
"loss": 0.025,
"step": 3800
},
{
"epoch": 6.42,
"learning_rate": 2.617777777777778e-06,
"loss": 0.0249,
"step": 3825
},
{
"epoch": 6.46,
"learning_rate": 2.5622222222222226e-06,
"loss": 0.0244,
"step": 3850
},
{
"epoch": 6.5,
"learning_rate": 2.5066666666666667e-06,
"loss": 0.0234,
"step": 3875
},
{
"epoch": 6.54,
"learning_rate": 2.451111111111111e-06,
"loss": 0.0251,
"step": 3900
},
{
"epoch": 6.59,
"learning_rate": 2.3955555555555556e-06,
"loss": 0.0247,
"step": 3925
},
{
"epoch": 6.63,
"learning_rate": 2.3400000000000005e-06,
"loss": 0.0245,
"step": 3950
},
{
"epoch": 6.67,
"learning_rate": 2.2844444444444445e-06,
"loss": 0.0241,
"step": 3975
},
{
"epoch": 6.71,
"learning_rate": 2.228888888888889e-06,
"loss": 0.0238,
"step": 4000
},
{
"epoch": 6.71,
"eval_loss": 0.2068559229373932,
"eval_runtime": 1847.5087,
"eval_samples_per_second": 8.117,
"eval_steps_per_second": 0.127,
"eval_wer": 0.120207771588125,
"step": 4000
},
{
"epoch": 6.75,
"learning_rate": 2.1733333333333334e-06,
"loss": 0.0249,
"step": 4025
},
{
"epoch": 6.79,
"learning_rate": 2.117777777777778e-06,
"loss": 0.0248,
"step": 4050
},
{
"epoch": 6.84,
"learning_rate": 2.0622222222222223e-06,
"loss": 0.0252,
"step": 4075
},
{
"epoch": 6.88,
"learning_rate": 2.006666666666667e-06,
"loss": 0.0247,
"step": 4100
},
{
"epoch": 6.92,
"learning_rate": 1.9511111111111113e-06,
"loss": 0.0247,
"step": 4125
},
{
"epoch": 6.96,
"learning_rate": 1.8955555555555557e-06,
"loss": 0.0238,
"step": 4150
},
{
"epoch": 7.01,
"learning_rate": 1.8400000000000002e-06,
"loss": 0.0244,
"step": 4175
},
{
"epoch": 7.05,
"learning_rate": 1.7844444444444444e-06,
"loss": 0.0189,
"step": 4200
},
{
"epoch": 7.09,
"learning_rate": 1.728888888888889e-06,
"loss": 0.0184,
"step": 4225
},
{
"epoch": 7.13,
"learning_rate": 1.6733333333333335e-06,
"loss": 0.0189,
"step": 4250
},
{
"epoch": 7.17,
"learning_rate": 1.6177777777777778e-06,
"loss": 0.0184,
"step": 4275
},
{
"epoch": 7.21,
"learning_rate": 1.5622222222222225e-06,
"loss": 0.0183,
"step": 4300
},
{
"epoch": 7.26,
"learning_rate": 1.506666666666667e-06,
"loss": 0.0179,
"step": 4325
},
{
"epoch": 7.3,
"learning_rate": 1.4511111111111112e-06,
"loss": 0.0185,
"step": 4350
},
{
"epoch": 7.34,
"learning_rate": 1.3955555555555556e-06,
"loss": 0.0193,
"step": 4375
},
{
"epoch": 7.38,
"learning_rate": 1.34e-06,
"loss": 0.0185,
"step": 4400
},
{
"epoch": 7.42,
"learning_rate": 1.2844444444444445e-06,
"loss": 0.0192,
"step": 4425
},
{
"epoch": 7.47,
"learning_rate": 1.228888888888889e-06,
"loss": 0.0186,
"step": 4450
},
{
"epoch": 7.51,
"learning_rate": 1.1733333333333335e-06,
"loss": 0.0196,
"step": 4475
},
{
"epoch": 7.55,
"learning_rate": 1.117777777777778e-06,
"loss": 0.0192,
"step": 4500
},
{
"epoch": 7.59,
"learning_rate": 1.0622222222222222e-06,
"loss": 0.0193,
"step": 4525
},
{
"epoch": 7.63,
"learning_rate": 1.0066666666666668e-06,
"loss": 0.0194,
"step": 4550
},
{
"epoch": 7.68,
"learning_rate": 9.511111111111111e-07,
"loss": 0.0194,
"step": 4575
},
{
"epoch": 7.72,
"learning_rate": 8.955555555555557e-07,
"loss": 0.0191,
"step": 4600
},
{
"epoch": 7.76,
"learning_rate": 8.400000000000001e-07,
"loss": 0.0189,
"step": 4625
},
{
"epoch": 7.8,
"learning_rate": 7.844444444444445e-07,
"loss": 0.0182,
"step": 4650
},
{
"epoch": 7.84,
"learning_rate": 7.28888888888889e-07,
"loss": 0.0185,
"step": 4675
},
{
"epoch": 7.89,
"learning_rate": 6.733333333333334e-07,
"loss": 0.019,
"step": 4700
},
{
"epoch": 7.93,
"learning_rate": 6.177777777777778e-07,
"loss": 0.0191,
"step": 4725
},
{
"epoch": 7.97,
"learning_rate": 5.622222222222223e-07,
"loss": 0.0189,
"step": 4750
},
{
"epoch": 8.01,
"learning_rate": 5.066666666666667e-07,
"loss": 0.0184,
"step": 4775
},
{
"epoch": 8.05,
"learning_rate": 4.511111111111111e-07,
"loss": 0.0158,
"step": 4800
},
{
"epoch": 8.1,
"learning_rate": 3.9555555555555557e-07,
"loss": 0.016,
"step": 4825
},
{
"epoch": 8.14,
"learning_rate": 3.4000000000000003e-07,
"loss": 0.0168,
"step": 4850
},
{
"epoch": 8.18,
"learning_rate": 2.844444444444445e-07,
"loss": 0.0163,
"step": 4875
},
{
"epoch": 8.22,
"learning_rate": 2.2888888888888892e-07,
"loss": 0.0168,
"step": 4900
},
{
"epoch": 8.26,
"learning_rate": 1.7333333333333335e-07,
"loss": 0.0164,
"step": 4925
},
{
"epoch": 8.31,
"learning_rate": 1.1777777777777778e-07,
"loss": 0.0157,
"step": 4950
},
{
"epoch": 8.35,
"learning_rate": 6.222222222222223e-08,
"loss": 0.0164,
"step": 4975
},
{
"epoch": 8.39,
"learning_rate": 6.666666666666667e-09,
"loss": 0.0162,
"step": 5000
},
{
"epoch": 8.39,
"eval_loss": 0.21840792894363403,
"eval_runtime": 1846.1471,
"eval_samples_per_second": 8.123,
"eval_steps_per_second": 0.127,
"eval_wer": 0.12232244289018765,
"step": 5000
},
{
"epoch": 8.39,
"step": 5000,
"total_flos": 3.694724272114526e+20,
"train_loss": 0.09243068100810051,
"train_runtime": 70070.7243,
"train_samples_per_second": 18.267,
"train_steps_per_second": 0.071
}
],
"max_steps": 5000,
"num_train_epochs": 9,
"total_flos": 3.694724272114526e+20,
"trial_name": null,
"trial_params": null
}