malmarz's picture
End of training
cff82aa
raw
history blame contribute delete
No virus
25.8 kB
{
"best_metric": 51.491735086544,
"best_model_checkpoint": "./checkpoint-4000",
"epoch": 5.1411999999999995,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 4.6000000000000004e-07,
"loss": 0.8711,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.600000000000001e-07,
"loss": 0.8279,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.46e-06,
"loss": 0.6855,
"step": 75
},
{
"epoch": 0.02,
"learning_rate": 1.9600000000000003e-06,
"loss": 0.5674,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 2.46e-06,
"loss": 0.5478,
"step": 125
},
{
"epoch": 0.03,
"learning_rate": 2.96e-06,
"loss": 0.517,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 3.46e-06,
"loss": 0.5029,
"step": 175
},
{
"epoch": 0.04,
"learning_rate": 3.96e-06,
"loss": 0.5215,
"step": 200
},
{
"epoch": 0.04,
"learning_rate": 4.4600000000000005e-06,
"loss": 0.5113,
"step": 225
},
{
"epoch": 0.05,
"learning_rate": 4.960000000000001e-06,
"loss": 0.4756,
"step": 250
},
{
"epoch": 0.06,
"learning_rate": 5.460000000000001e-06,
"loss": 0.4076,
"step": 275
},
{
"epoch": 0.06,
"learning_rate": 5.9600000000000005e-06,
"loss": 0.49,
"step": 300
},
{
"epoch": 0.07,
"learning_rate": 6.460000000000001e-06,
"loss": 0.4474,
"step": 325
},
{
"epoch": 0.07,
"learning_rate": 6.96e-06,
"loss": 0.4227,
"step": 350
},
{
"epoch": 0.07,
"learning_rate": 7.4600000000000006e-06,
"loss": 0.4112,
"step": 375
},
{
"epoch": 0.08,
"learning_rate": 7.960000000000002e-06,
"loss": 0.4052,
"step": 400
},
{
"epoch": 0.09,
"learning_rate": 8.46e-06,
"loss": 0.3887,
"step": 425
},
{
"epoch": 0.09,
"learning_rate": 8.96e-06,
"loss": 0.4037,
"step": 450
},
{
"epoch": 0.1,
"learning_rate": 9.460000000000001e-06,
"loss": 0.4464,
"step": 475
},
{
"epoch": 0.1,
"learning_rate": 9.960000000000001e-06,
"loss": 0.4757,
"step": 500
},
{
"epoch": 0.1,
"learning_rate": 9.94888888888889e-06,
"loss": 0.4526,
"step": 525
},
{
"epoch": 0.11,
"learning_rate": 9.893333333333334e-06,
"loss": 0.4185,
"step": 550
},
{
"epoch": 0.12,
"learning_rate": 9.837777777777778e-06,
"loss": 0.4357,
"step": 575
},
{
"epoch": 0.12,
"learning_rate": 9.782222222222222e-06,
"loss": 0.4111,
"step": 600
},
{
"epoch": 0.12,
"learning_rate": 9.726666666666668e-06,
"loss": 0.408,
"step": 625
},
{
"epoch": 0.13,
"learning_rate": 9.671111111111112e-06,
"loss": 0.4335,
"step": 650
},
{
"epoch": 0.14,
"learning_rate": 9.615555555555558e-06,
"loss": 0.4112,
"step": 675
},
{
"epoch": 0.14,
"learning_rate": 9.56e-06,
"loss": 0.3756,
"step": 700
},
{
"epoch": 0.14,
"learning_rate": 9.504444444444446e-06,
"loss": 0.375,
"step": 725
},
{
"epoch": 0.15,
"learning_rate": 9.44888888888889e-06,
"loss": 0.3662,
"step": 750
},
{
"epoch": 0.15,
"learning_rate": 9.393333333333334e-06,
"loss": 0.4707,
"step": 775
},
{
"epoch": 0.16,
"learning_rate": 9.33777777777778e-06,
"loss": 0.4977,
"step": 800
},
{
"epoch": 0.17,
"learning_rate": 9.282222222222222e-06,
"loss": 0.4868,
"step": 825
},
{
"epoch": 0.17,
"learning_rate": 9.226666666666668e-06,
"loss": 0.4123,
"step": 850
},
{
"epoch": 1.0,
"learning_rate": 9.171111111111112e-06,
"loss": 0.3552,
"step": 875
},
{
"epoch": 1.01,
"learning_rate": 9.115555555555556e-06,
"loss": 0.4024,
"step": 900
},
{
"epoch": 1.01,
"learning_rate": 9.060000000000001e-06,
"loss": 0.392,
"step": 925
},
{
"epoch": 1.02,
"learning_rate": 9.004444444444445e-06,
"loss": 0.3572,
"step": 950
},
{
"epoch": 1.02,
"learning_rate": 8.94888888888889e-06,
"loss": 0.3441,
"step": 975
},
{
"epoch": 1.03,
"learning_rate": 8.893333333333333e-06,
"loss": 0.3629,
"step": 1000
},
{
"epoch": 1.03,
"eval_loss": 0.4917178452014923,
"eval_runtime": 6498.5407,
"eval_samples_per_second": 1.798,
"eval_steps_per_second": 0.225,
"eval_wer": 53.12913829903252,
"step": 1000
},
{
"epoch": 1.03,
"learning_rate": 8.83777777777778e-06,
"loss": 0.2955,
"step": 1025
},
{
"epoch": 1.04,
"learning_rate": 8.782222222222223e-06,
"loss": 0.3233,
"step": 1050
},
{
"epoch": 1.04,
"learning_rate": 8.726666666666667e-06,
"loss": 0.3108,
"step": 1075
},
{
"epoch": 1.05,
"learning_rate": 8.671111111111113e-06,
"loss": 0.3776,
"step": 1100
},
{
"epoch": 1.05,
"learning_rate": 8.615555555555555e-06,
"loss": 0.4024,
"step": 1125
},
{
"epoch": 1.06,
"learning_rate": 8.560000000000001e-06,
"loss": 0.3219,
"step": 1150
},
{
"epoch": 1.06,
"learning_rate": 8.504444444444445e-06,
"loss": 0.3406,
"step": 1175
},
{
"epoch": 1.07,
"learning_rate": 8.448888888888889e-06,
"loss": 0.3008,
"step": 1200
},
{
"epoch": 1.07,
"learning_rate": 8.393333333333335e-06,
"loss": 0.281,
"step": 1225
},
{
"epoch": 1.08,
"learning_rate": 8.337777777777777e-06,
"loss": 0.2815,
"step": 1250
},
{
"epoch": 1.08,
"learning_rate": 8.282222222222223e-06,
"loss": 0.2789,
"step": 1275
},
{
"epoch": 1.09,
"learning_rate": 8.226666666666667e-06,
"loss": 0.2588,
"step": 1300
},
{
"epoch": 1.09,
"learning_rate": 8.171111111111113e-06,
"loss": 0.3163,
"step": 1325
},
{
"epoch": 1.1,
"learning_rate": 8.115555555555557e-06,
"loss": 0.3036,
"step": 1350
},
{
"epoch": 1.1,
"learning_rate": 8.06e-06,
"loss": 0.3028,
"step": 1375
},
{
"epoch": 1.11,
"learning_rate": 8.004444444444445e-06,
"loss": 0.2849,
"step": 1400
},
{
"epoch": 1.11,
"learning_rate": 7.948888888888889e-06,
"loss": 0.3225,
"step": 1425
},
{
"epoch": 1.12,
"learning_rate": 7.893333333333335e-06,
"loss": 0.3546,
"step": 1450
},
{
"epoch": 1.12,
"learning_rate": 7.837777777777779e-06,
"loss": 0.3541,
"step": 1475
},
{
"epoch": 1.13,
"learning_rate": 7.782222222222223e-06,
"loss": 0.3413,
"step": 1500
},
{
"epoch": 1.13,
"learning_rate": 7.726666666666667e-06,
"loss": 0.2851,
"step": 1525
},
{
"epoch": 1.14,
"learning_rate": 7.67111111111111e-06,
"loss": 0.2558,
"step": 1550
},
{
"epoch": 1.14,
"learning_rate": 7.6155555555555564e-06,
"loss": 0.2515,
"step": 1575
},
{
"epoch": 1.15,
"learning_rate": 7.5600000000000005e-06,
"loss": 0.2788,
"step": 1600
},
{
"epoch": 1.15,
"learning_rate": 7.504444444444445e-06,
"loss": 0.3439,
"step": 1625
},
{
"epoch": 1.16,
"learning_rate": 7.44888888888889e-06,
"loss": 0.3047,
"step": 1650
},
{
"epoch": 1.16,
"learning_rate": 7.393333333333333e-06,
"loss": 0.3683,
"step": 1675
},
{
"epoch": 1.17,
"learning_rate": 7.337777777777778e-06,
"loss": 0.3039,
"step": 1700
},
{
"epoch": 2.0,
"learning_rate": 7.282222222222222e-06,
"loss": 0.2689,
"step": 1725
},
{
"epoch": 2.01,
"learning_rate": 7.226666666666667e-06,
"loss": 0.2152,
"step": 1750
},
{
"epoch": 2.01,
"learning_rate": 7.171111111111112e-06,
"loss": 0.202,
"step": 1775
},
{
"epoch": 2.02,
"learning_rate": 7.115555555555557e-06,
"loss": 0.2195,
"step": 1800
},
{
"epoch": 2.02,
"learning_rate": 7.06e-06,
"loss": 0.2246,
"step": 1825
},
{
"epoch": 2.03,
"learning_rate": 7.004444444444445e-06,
"loss": 0.2481,
"step": 1850
},
{
"epoch": 2.03,
"learning_rate": 6.948888888888889e-06,
"loss": 0.1741,
"step": 1875
},
{
"epoch": 2.04,
"learning_rate": 6.893333333333334e-06,
"loss": 0.1981,
"step": 1900
},
{
"epoch": 2.04,
"learning_rate": 6.837777777777779e-06,
"loss": 0.1986,
"step": 1925
},
{
"epoch": 2.05,
"learning_rate": 6.782222222222222e-06,
"loss": 0.2184,
"step": 1950
},
{
"epoch": 2.05,
"learning_rate": 6.726666666666667e-06,
"loss": 0.2498,
"step": 1975
},
{
"epoch": 2.06,
"learning_rate": 6.671111111111112e-06,
"loss": 0.289,
"step": 2000
},
{
"epoch": 2.06,
"eval_loss": 0.4746549129486084,
"eval_runtime": 6488.3214,
"eval_samples_per_second": 1.801,
"eval_steps_per_second": 0.225,
"eval_wer": 61.38548304173742,
"step": 2000
},
{
"epoch": 2.06,
"learning_rate": 6.615555555555556e-06,
"loss": 0.2892,
"step": 2025
},
{
"epoch": 2.07,
"learning_rate": 6.560000000000001e-06,
"loss": 0.3032,
"step": 2050
},
{
"epoch": 2.07,
"learning_rate": 6.504444444444446e-06,
"loss": 0.2558,
"step": 2075
},
{
"epoch": 2.08,
"learning_rate": 6.448888888888889e-06,
"loss": 0.2977,
"step": 2100
},
{
"epoch": 2.08,
"learning_rate": 6.393333333333334e-06,
"loss": 0.2597,
"step": 2125
},
{
"epoch": 2.09,
"learning_rate": 6.3377777777777786e-06,
"loss": 0.2635,
"step": 2150
},
{
"epoch": 2.09,
"learning_rate": 6.282222222222223e-06,
"loss": 0.3013,
"step": 2175
},
{
"epoch": 2.1,
"learning_rate": 6.2266666666666675e-06,
"loss": 0.3035,
"step": 2200
},
{
"epoch": 2.1,
"learning_rate": 6.171111111111112e-06,
"loss": 0.3006,
"step": 2225
},
{
"epoch": 2.11,
"learning_rate": 6.1155555555555555e-06,
"loss": 0.2888,
"step": 2250
},
{
"epoch": 2.11,
"learning_rate": 6.0600000000000004e-06,
"loss": 0.3038,
"step": 2275
},
{
"epoch": 2.12,
"learning_rate": 6.004444444444445e-06,
"loss": 0.2599,
"step": 2300
},
{
"epoch": 2.12,
"learning_rate": 5.951111111111112e-06,
"loss": 0.2985,
"step": 2325
},
{
"epoch": 2.13,
"learning_rate": 5.895555555555557e-06,
"loss": 0.3001,
"step": 2350
},
{
"epoch": 2.13,
"learning_rate": 5.84e-06,
"loss": 0.2833,
"step": 2375
},
{
"epoch": 2.14,
"learning_rate": 5.784444444444445e-06,
"loss": 0.3002,
"step": 2400
},
{
"epoch": 2.14,
"learning_rate": 5.72888888888889e-06,
"loss": 0.2697,
"step": 2425
},
{
"epoch": 2.15,
"learning_rate": 5.673333333333334e-06,
"loss": 0.308,
"step": 2450
},
{
"epoch": 2.15,
"learning_rate": 5.617777777777779e-06,
"loss": 0.2892,
"step": 2475
},
{
"epoch": 2.16,
"learning_rate": 5.562222222222222e-06,
"loss": 0.3042,
"step": 2500
},
{
"epoch": 2.16,
"learning_rate": 5.506666666666667e-06,
"loss": 0.3623,
"step": 2525
},
{
"epoch": 2.17,
"learning_rate": 5.451111111111112e-06,
"loss": 0.3683,
"step": 2550
},
{
"epoch": 2.17,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.3044,
"step": 2575
},
{
"epoch": 3.0,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.1678,
"step": 2600
},
{
"epoch": 3.01,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.1571,
"step": 2625
},
{
"epoch": 3.01,
"learning_rate": 5.228888888888889e-06,
"loss": 0.163,
"step": 2650
},
{
"epoch": 3.02,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.1765,
"step": 2675
},
{
"epoch": 3.02,
"learning_rate": 5.117777777777778e-06,
"loss": 0.1875,
"step": 2700
},
{
"epoch": 3.03,
"learning_rate": 5.062222222222222e-06,
"loss": 0.1495,
"step": 2725
},
{
"epoch": 3.03,
"learning_rate": 5.006666666666667e-06,
"loss": 0.1402,
"step": 2750
},
{
"epoch": 3.04,
"learning_rate": 4.951111111111111e-06,
"loss": 0.1472,
"step": 2775
},
{
"epoch": 3.04,
"learning_rate": 4.895555555555556e-06,
"loss": 0.1626,
"step": 2800
},
{
"epoch": 3.05,
"learning_rate": 4.84e-06,
"loss": 0.1648,
"step": 2825
},
{
"epoch": 3.05,
"learning_rate": 4.784444444444445e-06,
"loss": 0.2344,
"step": 2850
},
{
"epoch": 3.06,
"learning_rate": 4.728888888888889e-06,
"loss": 0.2556,
"step": 2875
},
{
"epoch": 3.06,
"learning_rate": 4.673333333333333e-06,
"loss": 0.2603,
"step": 2900
},
{
"epoch": 3.07,
"learning_rate": 4.617777777777778e-06,
"loss": 0.2184,
"step": 2925
},
{
"epoch": 3.07,
"learning_rate": 4.562222222222222e-06,
"loss": 0.2225,
"step": 2950
},
{
"epoch": 3.08,
"learning_rate": 4.506666666666667e-06,
"loss": 0.2332,
"step": 2975
},
{
"epoch": 3.08,
"learning_rate": 4.451111111111112e-06,
"loss": 0.2996,
"step": 3000
},
{
"epoch": 3.08,
"eval_loss": 0.45417240262031555,
"eval_runtime": 6499.9416,
"eval_samples_per_second": 1.798,
"eval_steps_per_second": 0.225,
"eval_wer": 55.46917600137099,
"step": 3000
},
{
"epoch": 3.09,
"learning_rate": 4.395555555555556e-06,
"loss": 0.2554,
"step": 3025
},
{
"epoch": 3.09,
"learning_rate": 4.34e-06,
"loss": 0.2476,
"step": 3050
},
{
"epoch": 3.1,
"learning_rate": 4.284444444444445e-06,
"loss": 0.2497,
"step": 3075
},
{
"epoch": 3.1,
"learning_rate": 4.228888888888889e-06,
"loss": 0.2573,
"step": 3100
},
{
"epoch": 3.11,
"learning_rate": 4.173333333333334e-06,
"loss": 0.254,
"step": 3125
},
{
"epoch": 3.11,
"learning_rate": 4.117777777777779e-06,
"loss": 0.2484,
"step": 3150
},
{
"epoch": 3.12,
"learning_rate": 4.062222222222223e-06,
"loss": 0.2646,
"step": 3175
},
{
"epoch": 3.12,
"learning_rate": 4.006666666666667e-06,
"loss": 0.2615,
"step": 3200
},
{
"epoch": 3.13,
"learning_rate": 3.951111111111112e-06,
"loss": 0.2479,
"step": 3225
},
{
"epoch": 3.13,
"learning_rate": 3.895555555555556e-06,
"loss": 0.1851,
"step": 3250
},
{
"epoch": 3.14,
"learning_rate": 3.8400000000000005e-06,
"loss": 0.218,
"step": 3275
},
{
"epoch": 3.14,
"learning_rate": 3.784444444444445e-06,
"loss": 0.1965,
"step": 3300
},
{
"epoch": 3.15,
"learning_rate": 3.728888888888889e-06,
"loss": 0.2042,
"step": 3325
},
{
"epoch": 3.15,
"learning_rate": 3.673333333333334e-06,
"loss": 0.2345,
"step": 3350
},
{
"epoch": 3.16,
"learning_rate": 3.617777777777778e-06,
"loss": 0.2584,
"step": 3375
},
{
"epoch": 3.16,
"learning_rate": 3.5622222222222224e-06,
"loss": 0.2436,
"step": 3400
},
{
"epoch": 3.17,
"learning_rate": 3.5066666666666673e-06,
"loss": 0.1972,
"step": 3425
},
{
"epoch": 4.0,
"learning_rate": 3.4511111111111113e-06,
"loss": 0.2143,
"step": 3450
},
{
"epoch": 4.01,
"learning_rate": 3.3955555555555558e-06,
"loss": 0.3159,
"step": 3475
},
{
"epoch": 4.01,
"learning_rate": 3.3400000000000006e-06,
"loss": 0.2327,
"step": 3500
},
{
"epoch": 4.02,
"learning_rate": 3.2844444444444447e-06,
"loss": 0.2172,
"step": 3525
},
{
"epoch": 4.02,
"learning_rate": 3.228888888888889e-06,
"loss": 0.2576,
"step": 3550
},
{
"epoch": 4.03,
"learning_rate": 3.173333333333334e-06,
"loss": 0.2174,
"step": 3575
},
{
"epoch": 4.03,
"learning_rate": 3.117777777777778e-06,
"loss": 0.2701,
"step": 3600
},
{
"epoch": 4.04,
"learning_rate": 3.0622222222222225e-06,
"loss": 0.2382,
"step": 3625
},
{
"epoch": 4.04,
"learning_rate": 3.0066666666666674e-06,
"loss": 0.2204,
"step": 3650
},
{
"epoch": 4.05,
"learning_rate": 2.9511111111111114e-06,
"loss": 0.22,
"step": 3675
},
{
"epoch": 4.05,
"learning_rate": 2.895555555555556e-06,
"loss": 0.2753,
"step": 3700
},
{
"epoch": 4.06,
"learning_rate": 2.84e-06,
"loss": 0.2387,
"step": 3725
},
{
"epoch": 4.06,
"learning_rate": 2.784444444444445e-06,
"loss": 0.2577,
"step": 3750
},
{
"epoch": 4.07,
"learning_rate": 2.7288888888888893e-06,
"loss": 0.2344,
"step": 3775
},
{
"epoch": 4.07,
"learning_rate": 2.6733333333333333e-06,
"loss": 0.2134,
"step": 3800
},
{
"epoch": 4.08,
"learning_rate": 2.617777777777778e-06,
"loss": 0.2278,
"step": 3825
},
{
"epoch": 4.08,
"learning_rate": 2.5622222222222226e-06,
"loss": 0.2383,
"step": 3850
},
{
"epoch": 4.09,
"learning_rate": 2.5066666666666667e-06,
"loss": 0.2244,
"step": 3875
},
{
"epoch": 4.09,
"learning_rate": 2.451111111111111e-06,
"loss": 0.2577,
"step": 3900
},
{
"epoch": 4.1,
"learning_rate": 2.3955555555555556e-06,
"loss": 0.256,
"step": 3925
},
{
"epoch": 4.1,
"learning_rate": 2.3400000000000005e-06,
"loss": 0.2287,
"step": 3950
},
{
"epoch": 4.11,
"learning_rate": 2.2844444444444445e-06,
"loss": 0.2493,
"step": 3975
},
{
"epoch": 4.11,
"learning_rate": 2.228888888888889e-06,
"loss": 0.2331,
"step": 4000
},
{
"epoch": 4.11,
"eval_loss": 0.4353369474411011,
"eval_runtime": 5849.8513,
"eval_samples_per_second": 1.997,
"eval_steps_per_second": 0.25,
"eval_wer": 51.491735086544,
"step": 4000
},
{
"epoch": 4.12,
"learning_rate": 2.1733333333333334e-06,
"loss": 0.2927,
"step": 4025
},
{
"epoch": 4.12,
"learning_rate": 2.117777777777778e-06,
"loss": 0.2381,
"step": 4050
},
{
"epoch": 4.13,
"learning_rate": 2.0622222222222223e-06,
"loss": 0.2415,
"step": 4075
},
{
"epoch": 4.13,
"learning_rate": 2.006666666666667e-06,
"loss": 0.2123,
"step": 4100
},
{
"epoch": 4.14,
"learning_rate": 1.9511111111111113e-06,
"loss": 0.2367,
"step": 4125
},
{
"epoch": 4.14,
"learning_rate": 1.8955555555555557e-06,
"loss": 0.2168,
"step": 4150
},
{
"epoch": 4.15,
"learning_rate": 1.8400000000000002e-06,
"loss": 0.2272,
"step": 4175
},
{
"epoch": 4.15,
"learning_rate": 1.7844444444444444e-06,
"loss": 0.2308,
"step": 4200
},
{
"epoch": 4.16,
"learning_rate": 1.728888888888889e-06,
"loss": 0.2506,
"step": 4225
},
{
"epoch": 4.16,
"learning_rate": 1.6733333333333335e-06,
"loss": 0.2852,
"step": 4250
},
{
"epoch": 4.17,
"learning_rate": 1.6177777777777778e-06,
"loss": 0.2009,
"step": 4275
},
{
"epoch": 5.0,
"learning_rate": 1.5622222222222225e-06,
"loss": 0.2282,
"step": 4300
},
{
"epoch": 5.01,
"learning_rate": 1.506666666666667e-06,
"loss": 0.207,
"step": 4325
},
{
"epoch": 5.01,
"learning_rate": 1.4511111111111112e-06,
"loss": 0.2593,
"step": 4350
},
{
"epoch": 5.02,
"learning_rate": 1.3955555555555556e-06,
"loss": 0.2235,
"step": 4375
},
{
"epoch": 5.02,
"learning_rate": 1.34e-06,
"loss": 0.2476,
"step": 4400
},
{
"epoch": 5.03,
"learning_rate": 1.2844444444444445e-06,
"loss": 0.2279,
"step": 4425
},
{
"epoch": 5.03,
"learning_rate": 1.228888888888889e-06,
"loss": 0.2165,
"step": 4450
},
{
"epoch": 5.04,
"learning_rate": 1.1733333333333335e-06,
"loss": 0.2597,
"step": 4475
},
{
"epoch": 5.04,
"learning_rate": 1.117777777777778e-06,
"loss": 0.2637,
"step": 4500
},
{
"epoch": 5.05,
"learning_rate": 1.0622222222222222e-06,
"loss": 0.2284,
"step": 4525
},
{
"epoch": 5.05,
"learning_rate": 1.0066666666666668e-06,
"loss": 0.2037,
"step": 4550
},
{
"epoch": 5.06,
"learning_rate": 9.511111111111111e-07,
"loss": 0.2445,
"step": 4575
},
{
"epoch": 5.06,
"learning_rate": 8.977777777777778e-07,
"loss": 0.2194,
"step": 4600
},
{
"epoch": 5.07,
"learning_rate": 8.422222222222224e-07,
"loss": 0.25,
"step": 4625
},
{
"epoch": 5.07,
"learning_rate": 7.866666666666667e-07,
"loss": 0.2255,
"step": 4650
},
{
"epoch": 5.08,
"learning_rate": 7.311111111111112e-07,
"loss": 0.2722,
"step": 4675
},
{
"epoch": 5.08,
"learning_rate": 6.755555555555555e-07,
"loss": 0.2149,
"step": 4700
},
{
"epoch": 5.09,
"learning_rate": 6.200000000000001e-07,
"loss": 0.2058,
"step": 4725
},
{
"epoch": 5.09,
"learning_rate": 5.644444444444445e-07,
"loss": 0.193,
"step": 4750
},
{
"epoch": 5.1,
"learning_rate": 5.088888888888889e-07,
"loss": 0.215,
"step": 4775
},
{
"epoch": 5.1,
"learning_rate": 4.533333333333334e-07,
"loss": 0.1588,
"step": 4800
},
{
"epoch": 5.11,
"learning_rate": 3.9777777777777783e-07,
"loss": 0.173,
"step": 4825
},
{
"epoch": 5.11,
"learning_rate": 3.422222222222223e-07,
"loss": 0.1919,
"step": 4850
},
{
"epoch": 5.12,
"learning_rate": 2.866666666666667e-07,
"loss": 0.2126,
"step": 4875
},
{
"epoch": 5.12,
"learning_rate": 2.3111111111111112e-07,
"loss": 0.1365,
"step": 4900
},
{
"epoch": 5.13,
"learning_rate": 1.7555555555555558e-07,
"loss": 0.1245,
"step": 4925
},
{
"epoch": 5.13,
"learning_rate": 1.2000000000000002e-07,
"loss": 0.1461,
"step": 4950
},
{
"epoch": 5.14,
"learning_rate": 6.444444444444445e-08,
"loss": 0.1436,
"step": 4975
},
{
"epoch": 5.14,
"learning_rate": 8.88888888888889e-09,
"loss": 0.1566,
"step": 5000
},
{
"epoch": 5.14,
"eval_loss": 0.442874938249588,
"eval_runtime": 6113.7384,
"eval_samples_per_second": 1.911,
"eval_steps_per_second": 0.239,
"eval_wer": 52.75678875784816,
"step": 5000
},
{
"epoch": 5.14,
"step": 5000,
"total_flos": 9.230230877995008e+19,
"train_loss": 0.2917347553253174,
"train_runtime": 73075.0167,
"train_samples_per_second": 4.379,
"train_steps_per_second": 0.068
}
],
"max_steps": 5000,
"num_train_epochs": 9223372036854775807,
"total_flos": 9.230230877995008e+19,
"trial_name": null,
"trial_params": null
}