TheRains's picture
End of training
640d4d1
raw
history blame contribute delete
No virus
25.8 kB
{
"best_metric": 11.893259719346675,
"best_model_checkpoint": "./output/small/cv9-special-batch8-small-concat-Fleur/checkpoint-5000",
"epoch": 3.57653791130186,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02,
"learning_rate": 4.4e-07,
"loss": 1.2627,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 9.400000000000001e-07,
"loss": 1.2478,
"step": 50
},
{
"epoch": 0.05,
"learning_rate": 1.44e-06,
"loss": 0.9832,
"step": 75
},
{
"epoch": 0.07,
"learning_rate": 1.94e-06,
"loss": 0.717,
"step": 100
},
{
"epoch": 0.09,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.5363,
"step": 125
},
{
"epoch": 0.11,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.4513,
"step": 150
},
{
"epoch": 0.13,
"learning_rate": 3.44e-06,
"loss": 0.4034,
"step": 175
},
{
"epoch": 0.14,
"learning_rate": 3.94e-06,
"loss": 0.452,
"step": 200
},
{
"epoch": 0.16,
"learning_rate": 4.440000000000001e-06,
"loss": 0.4309,
"step": 225
},
{
"epoch": 0.18,
"learning_rate": 4.94e-06,
"loss": 0.4271,
"step": 250
},
{
"epoch": 0.2,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.3725,
"step": 275
},
{
"epoch": 0.21,
"learning_rate": 5.94e-06,
"loss": 0.3585,
"step": 300
},
{
"epoch": 0.23,
"learning_rate": 6.440000000000001e-06,
"loss": 0.3585,
"step": 325
},
{
"epoch": 0.25,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.3534,
"step": 350
},
{
"epoch": 0.27,
"learning_rate": 7.440000000000001e-06,
"loss": 0.3924,
"step": 375
},
{
"epoch": 0.29,
"learning_rate": 7.94e-06,
"loss": 0.3302,
"step": 400
},
{
"epoch": 0.3,
"learning_rate": 8.44e-06,
"loss": 0.3358,
"step": 425
},
{
"epoch": 0.32,
"learning_rate": 8.94e-06,
"loss": 0.3685,
"step": 450
},
{
"epoch": 0.34,
"learning_rate": 9.440000000000001e-06,
"loss": 0.2963,
"step": 475
},
{
"epoch": 0.36,
"learning_rate": 9.940000000000001e-06,
"loss": 0.3619,
"step": 500
},
{
"epoch": 0.38,
"learning_rate": 9.951111111111111e-06,
"loss": 0.3839,
"step": 525
},
{
"epoch": 0.39,
"learning_rate": 9.895555555555557e-06,
"loss": 0.3504,
"step": 550
},
{
"epoch": 0.41,
"learning_rate": 9.84e-06,
"loss": 0.3357,
"step": 575
},
{
"epoch": 0.43,
"learning_rate": 9.784444444444445e-06,
"loss": 0.3258,
"step": 600
},
{
"epoch": 0.45,
"learning_rate": 9.72888888888889e-06,
"loss": 0.3375,
"step": 625
},
{
"epoch": 0.46,
"learning_rate": 9.673333333333334e-06,
"loss": 0.3298,
"step": 650
},
{
"epoch": 0.48,
"learning_rate": 9.617777777777778e-06,
"loss": 0.3225,
"step": 675
},
{
"epoch": 0.5,
"learning_rate": 9.562222222222223e-06,
"loss": 0.3503,
"step": 700
},
{
"epoch": 0.52,
"learning_rate": 9.506666666666667e-06,
"loss": 0.3107,
"step": 725
},
{
"epoch": 0.54,
"learning_rate": 9.451111111111112e-06,
"loss": 0.3314,
"step": 750
},
{
"epoch": 0.55,
"learning_rate": 9.395555555555556e-06,
"loss": 0.2912,
"step": 775
},
{
"epoch": 0.57,
"learning_rate": 9.340000000000002e-06,
"loss": 0.3116,
"step": 800
},
{
"epoch": 0.59,
"learning_rate": 9.284444444444444e-06,
"loss": 0.327,
"step": 825
},
{
"epoch": 0.61,
"learning_rate": 9.22888888888889e-06,
"loss": 0.2992,
"step": 850
},
{
"epoch": 0.63,
"learning_rate": 9.173333333333334e-06,
"loss": 0.3357,
"step": 875
},
{
"epoch": 0.64,
"learning_rate": 9.117777777777778e-06,
"loss": 0.3406,
"step": 900
},
{
"epoch": 0.66,
"learning_rate": 9.062222222222224e-06,
"loss": 0.282,
"step": 925
},
{
"epoch": 0.68,
"learning_rate": 9.006666666666666e-06,
"loss": 0.2642,
"step": 950
},
{
"epoch": 0.7,
"learning_rate": 8.951111111111112e-06,
"loss": 0.2994,
"step": 975
},
{
"epoch": 0.72,
"learning_rate": 8.895555555555556e-06,
"loss": 0.2987,
"step": 1000
},
{
"epoch": 0.72,
"eval_loss": 0.25958719849586487,
"eval_runtime": 596.17,
"eval_samples_per_second": 6.075,
"eval_steps_per_second": 1.52,
"eval_wer": 15.136876006441224,
"step": 1000
},
{
"epoch": 0.73,
"learning_rate": 8.84e-06,
"loss": 0.2816,
"step": 1025
},
{
"epoch": 0.75,
"learning_rate": 8.784444444444446e-06,
"loss": 0.3069,
"step": 1050
},
{
"epoch": 0.77,
"learning_rate": 8.72888888888889e-06,
"loss": 0.2945,
"step": 1075
},
{
"epoch": 0.79,
"learning_rate": 8.673333333333334e-06,
"loss": 0.2717,
"step": 1100
},
{
"epoch": 0.8,
"learning_rate": 8.617777777777778e-06,
"loss": 0.2621,
"step": 1125
},
{
"epoch": 0.82,
"learning_rate": 8.562222222222224e-06,
"loss": 0.2826,
"step": 1150
},
{
"epoch": 0.84,
"learning_rate": 8.506666666666668e-06,
"loss": 0.2395,
"step": 1175
},
{
"epoch": 0.86,
"learning_rate": 8.451111111111112e-06,
"loss": 0.2545,
"step": 1200
},
{
"epoch": 0.88,
"learning_rate": 8.395555555555557e-06,
"loss": 0.2686,
"step": 1225
},
{
"epoch": 0.89,
"learning_rate": 8.34e-06,
"loss": 0.2821,
"step": 1250
},
{
"epoch": 0.91,
"learning_rate": 8.284444444444446e-06,
"loss": 0.2329,
"step": 1275
},
{
"epoch": 0.93,
"learning_rate": 8.22888888888889e-06,
"loss": 0.2539,
"step": 1300
},
{
"epoch": 0.95,
"learning_rate": 8.173333333333334e-06,
"loss": 0.29,
"step": 1325
},
{
"epoch": 0.97,
"learning_rate": 8.11777777777778e-06,
"loss": 0.2801,
"step": 1350
},
{
"epoch": 0.98,
"learning_rate": 8.062222222222222e-06,
"loss": 0.2421,
"step": 1375
},
{
"epoch": 1.0,
"learning_rate": 8.006666666666667e-06,
"loss": 0.2551,
"step": 1400
},
{
"epoch": 1.02,
"learning_rate": 7.951111111111111e-06,
"loss": 0.1138,
"step": 1425
},
{
"epoch": 1.04,
"learning_rate": 7.895555555555557e-06,
"loss": 0.1413,
"step": 1450
},
{
"epoch": 1.06,
"learning_rate": 7.840000000000001e-06,
"loss": 0.1409,
"step": 1475
},
{
"epoch": 1.07,
"learning_rate": 7.784444444444445e-06,
"loss": 0.1475,
"step": 1500
},
{
"epoch": 1.09,
"learning_rate": 7.72888888888889e-06,
"loss": 0.1649,
"step": 1525
},
{
"epoch": 1.11,
"learning_rate": 7.673333333333333e-06,
"loss": 0.107,
"step": 1550
},
{
"epoch": 1.13,
"learning_rate": 7.617777777777778e-06,
"loss": 0.1333,
"step": 1575
},
{
"epoch": 1.14,
"learning_rate": 7.562222222222223e-06,
"loss": 0.1288,
"step": 1600
},
{
"epoch": 1.16,
"learning_rate": 7.506666666666668e-06,
"loss": 0.1252,
"step": 1625
},
{
"epoch": 1.18,
"learning_rate": 7.451111111111111e-06,
"loss": 0.1272,
"step": 1650
},
{
"epoch": 1.2,
"learning_rate": 7.395555555555556e-06,
"loss": 0.1301,
"step": 1675
},
{
"epoch": 1.22,
"learning_rate": 7.340000000000001e-06,
"loss": 0.1333,
"step": 1700
},
{
"epoch": 1.23,
"learning_rate": 7.284444444444445e-06,
"loss": 0.1443,
"step": 1725
},
{
"epoch": 1.25,
"learning_rate": 7.22888888888889e-06,
"loss": 0.1155,
"step": 1750
},
{
"epoch": 1.27,
"learning_rate": 7.173333333333335e-06,
"loss": 0.1386,
"step": 1775
},
{
"epoch": 1.29,
"learning_rate": 7.117777777777778e-06,
"loss": 0.1323,
"step": 1800
},
{
"epoch": 1.31,
"learning_rate": 7.062222222222223e-06,
"loss": 0.1287,
"step": 1825
},
{
"epoch": 1.32,
"learning_rate": 7.006666666666667e-06,
"loss": 0.1219,
"step": 1850
},
{
"epoch": 1.34,
"learning_rate": 6.951111111111112e-06,
"loss": 0.1153,
"step": 1875
},
{
"epoch": 1.36,
"learning_rate": 6.8955555555555565e-06,
"loss": 0.1402,
"step": 1900
},
{
"epoch": 1.38,
"learning_rate": 6.8400000000000014e-06,
"loss": 0.137,
"step": 1925
},
{
"epoch": 1.39,
"learning_rate": 6.784444444444445e-06,
"loss": 0.1378,
"step": 1950
},
{
"epoch": 1.41,
"learning_rate": 6.7288888888888895e-06,
"loss": 0.1438,
"step": 1975
},
{
"epoch": 1.43,
"learning_rate": 6.6733333333333335e-06,
"loss": 0.1152,
"step": 2000
},
{
"epoch": 1.43,
"eval_loss": 0.23717719316482544,
"eval_runtime": 590.6287,
"eval_samples_per_second": 6.132,
"eval_steps_per_second": 1.534,
"eval_wer": 12.61099608925696,
"step": 2000
},
{
"epoch": 1.45,
"learning_rate": 6.617777777777778e-06,
"loss": 0.1456,
"step": 2025
},
{
"epoch": 1.47,
"learning_rate": 6.562222222222223e-06,
"loss": 0.1275,
"step": 2050
},
{
"epoch": 1.48,
"learning_rate": 6.5066666666666665e-06,
"loss": 0.1347,
"step": 2075
},
{
"epoch": 1.5,
"learning_rate": 6.451111111111111e-06,
"loss": 0.1308,
"step": 2100
},
{
"epoch": 1.52,
"learning_rate": 6.395555555555556e-06,
"loss": 0.1548,
"step": 2125
},
{
"epoch": 1.54,
"learning_rate": 6.34e-06,
"loss": 0.135,
"step": 2150
},
{
"epoch": 1.56,
"learning_rate": 6.284444444444445e-06,
"loss": 0.1413,
"step": 2175
},
{
"epoch": 1.57,
"learning_rate": 6.22888888888889e-06,
"loss": 0.1351,
"step": 2200
},
{
"epoch": 1.59,
"learning_rate": 6.173333333333333e-06,
"loss": 0.1435,
"step": 2225
},
{
"epoch": 1.61,
"learning_rate": 6.117777777777778e-06,
"loss": 0.1277,
"step": 2250
},
{
"epoch": 1.63,
"learning_rate": 6.062222222222223e-06,
"loss": 0.1159,
"step": 2275
},
{
"epoch": 1.65,
"learning_rate": 6.006666666666667e-06,
"loss": 0.1273,
"step": 2300
},
{
"epoch": 1.66,
"learning_rate": 5.951111111111112e-06,
"loss": 0.1297,
"step": 2325
},
{
"epoch": 1.68,
"learning_rate": 5.895555555555557e-06,
"loss": 0.1366,
"step": 2350
},
{
"epoch": 1.7,
"learning_rate": 5.84e-06,
"loss": 0.1318,
"step": 2375
},
{
"epoch": 1.72,
"learning_rate": 5.784444444444445e-06,
"loss": 0.1338,
"step": 2400
},
{
"epoch": 1.73,
"learning_rate": 5.72888888888889e-06,
"loss": 0.128,
"step": 2425
},
{
"epoch": 1.75,
"learning_rate": 5.673333333333334e-06,
"loss": 0.1027,
"step": 2450
},
{
"epoch": 1.77,
"learning_rate": 5.617777777777779e-06,
"loss": 0.1418,
"step": 2475
},
{
"epoch": 1.79,
"learning_rate": 5.562222222222222e-06,
"loss": 0.1197,
"step": 2500
},
{
"epoch": 1.81,
"learning_rate": 5.506666666666667e-06,
"loss": 0.1313,
"step": 2525
},
{
"epoch": 1.82,
"learning_rate": 5.451111111111112e-06,
"loss": 0.1308,
"step": 2550
},
{
"epoch": 1.84,
"learning_rate": 5.3955555555555565e-06,
"loss": 0.1149,
"step": 2575
},
{
"epoch": 1.86,
"learning_rate": 5.3400000000000005e-06,
"loss": 0.133,
"step": 2600
},
{
"epoch": 1.88,
"learning_rate": 5.2844444444444454e-06,
"loss": 0.1381,
"step": 2625
},
{
"epoch": 1.9,
"learning_rate": 5.228888888888889e-06,
"loss": 0.1138,
"step": 2650
},
{
"epoch": 1.91,
"learning_rate": 5.1733333333333335e-06,
"loss": 0.1252,
"step": 2675
},
{
"epoch": 1.93,
"learning_rate": 5.117777777777778e-06,
"loss": 0.1441,
"step": 2700
},
{
"epoch": 1.95,
"learning_rate": 5.062222222222222e-06,
"loss": 0.1176,
"step": 2725
},
{
"epoch": 1.97,
"learning_rate": 5.006666666666667e-06,
"loss": 0.1263,
"step": 2750
},
{
"epoch": 1.98,
"learning_rate": 4.951111111111111e-06,
"loss": 0.1137,
"step": 2775
},
{
"epoch": 2.0,
"learning_rate": 4.895555555555556e-06,
"loss": 0.1053,
"step": 2800
},
{
"epoch": 2.02,
"learning_rate": 4.84e-06,
"loss": 0.0506,
"step": 2825
},
{
"epoch": 2.04,
"learning_rate": 4.784444444444445e-06,
"loss": 0.049,
"step": 2850
},
{
"epoch": 2.06,
"learning_rate": 4.728888888888889e-06,
"loss": 0.0576,
"step": 2875
},
{
"epoch": 2.07,
"learning_rate": 4.673333333333333e-06,
"loss": 0.048,
"step": 2900
},
{
"epoch": 2.09,
"learning_rate": 4.617777777777778e-06,
"loss": 0.0633,
"step": 2925
},
{
"epoch": 2.11,
"learning_rate": 4.562222222222222e-06,
"loss": 0.0562,
"step": 2950
},
{
"epoch": 2.13,
"learning_rate": 4.506666666666667e-06,
"loss": 0.0484,
"step": 2975
},
{
"epoch": 2.15,
"learning_rate": 4.451111111111112e-06,
"loss": 0.0544,
"step": 3000
},
{
"epoch": 2.15,
"eval_loss": 0.23555360734462738,
"eval_runtime": 590.8662,
"eval_samples_per_second": 6.13,
"eval_steps_per_second": 1.533,
"eval_wer": 12.08189556015643,
"step": 3000
},
{
"epoch": 2.16,
"learning_rate": 4.395555555555556e-06,
"loss": 0.0548,
"step": 3025
},
{
"epoch": 2.18,
"learning_rate": 4.34e-06,
"loss": 0.0647,
"step": 3050
},
{
"epoch": 2.2,
"learning_rate": 4.284444444444445e-06,
"loss": 0.0492,
"step": 3075
},
{
"epoch": 2.22,
"learning_rate": 4.228888888888889e-06,
"loss": 0.0527,
"step": 3100
},
{
"epoch": 2.24,
"learning_rate": 4.173333333333334e-06,
"loss": 0.0468,
"step": 3125
},
{
"epoch": 2.25,
"learning_rate": 4.117777777777779e-06,
"loss": 0.0524,
"step": 3150
},
{
"epoch": 2.27,
"learning_rate": 4.062222222222223e-06,
"loss": 0.048,
"step": 3175
},
{
"epoch": 2.29,
"learning_rate": 4.006666666666667e-06,
"loss": 0.0534,
"step": 3200
},
{
"epoch": 2.31,
"learning_rate": 3.951111111111112e-06,
"loss": 0.0512,
"step": 3225
},
{
"epoch": 2.32,
"learning_rate": 3.895555555555556e-06,
"loss": 0.0384,
"step": 3250
},
{
"epoch": 2.34,
"learning_rate": 3.8400000000000005e-06,
"loss": 0.0459,
"step": 3275
},
{
"epoch": 2.36,
"learning_rate": 3.784444444444445e-06,
"loss": 0.0491,
"step": 3300
},
{
"epoch": 2.38,
"learning_rate": 3.728888888888889e-06,
"loss": 0.0595,
"step": 3325
},
{
"epoch": 2.4,
"learning_rate": 3.673333333333334e-06,
"loss": 0.0604,
"step": 3350
},
{
"epoch": 2.41,
"learning_rate": 3.617777777777778e-06,
"loss": 0.0542,
"step": 3375
},
{
"epoch": 2.43,
"learning_rate": 3.5622222222222224e-06,
"loss": 0.0438,
"step": 3400
},
{
"epoch": 2.45,
"learning_rate": 3.5066666666666673e-06,
"loss": 0.05,
"step": 3425
},
{
"epoch": 2.47,
"learning_rate": 3.4511111111111113e-06,
"loss": 0.0452,
"step": 3450
},
{
"epoch": 2.49,
"learning_rate": 3.3955555555555558e-06,
"loss": 0.0363,
"step": 3475
},
{
"epoch": 2.5,
"learning_rate": 3.3400000000000006e-06,
"loss": 0.0514,
"step": 3500
},
{
"epoch": 2.52,
"learning_rate": 3.2844444444444447e-06,
"loss": 0.0473,
"step": 3525
},
{
"epoch": 2.54,
"learning_rate": 3.228888888888889e-06,
"loss": 0.041,
"step": 3550
},
{
"epoch": 2.56,
"learning_rate": 3.173333333333334e-06,
"loss": 0.0508,
"step": 3575
},
{
"epoch": 2.58,
"learning_rate": 3.117777777777778e-06,
"loss": 0.0395,
"step": 3600
},
{
"epoch": 2.59,
"learning_rate": 3.0622222222222225e-06,
"loss": 0.0472,
"step": 3625
},
{
"epoch": 2.61,
"learning_rate": 3.0066666666666674e-06,
"loss": 0.0474,
"step": 3650
},
{
"epoch": 2.63,
"learning_rate": 2.9511111111111114e-06,
"loss": 0.0538,
"step": 3675
},
{
"epoch": 2.65,
"learning_rate": 2.895555555555556e-06,
"loss": 0.0395,
"step": 3700
},
{
"epoch": 2.66,
"learning_rate": 2.84e-06,
"loss": 0.0377,
"step": 3725
},
{
"epoch": 2.68,
"learning_rate": 2.784444444444445e-06,
"loss": 0.0437,
"step": 3750
},
{
"epoch": 2.7,
"learning_rate": 2.7288888888888893e-06,
"loss": 0.0492,
"step": 3775
},
{
"epoch": 2.72,
"learning_rate": 2.6733333333333333e-06,
"loss": 0.0521,
"step": 3800
},
{
"epoch": 2.74,
"learning_rate": 2.617777777777778e-06,
"loss": 0.0465,
"step": 3825
},
{
"epoch": 2.75,
"learning_rate": 2.5622222222222226e-06,
"loss": 0.044,
"step": 3850
},
{
"epoch": 2.77,
"learning_rate": 2.5066666666666667e-06,
"loss": 0.0542,
"step": 3875
},
{
"epoch": 2.79,
"learning_rate": 2.451111111111111e-06,
"loss": 0.0436,
"step": 3900
},
{
"epoch": 2.81,
"learning_rate": 2.3955555555555556e-06,
"loss": 0.0502,
"step": 3925
},
{
"epoch": 2.83,
"learning_rate": 2.3400000000000005e-06,
"loss": 0.0452,
"step": 3950
},
{
"epoch": 2.84,
"learning_rate": 2.2844444444444445e-06,
"loss": 0.0431,
"step": 3975
},
{
"epoch": 2.86,
"learning_rate": 2.228888888888889e-06,
"loss": 0.0431,
"step": 4000
},
{
"epoch": 2.86,
"eval_loss": 0.23700806498527527,
"eval_runtime": 591.0734,
"eval_samples_per_second": 6.128,
"eval_steps_per_second": 1.533,
"eval_wer": 11.953071083505867,
"step": 4000
},
{
"epoch": 2.88,
"learning_rate": 2.1733333333333334e-06,
"loss": 0.0432,
"step": 4025
},
{
"epoch": 2.9,
"learning_rate": 2.117777777777778e-06,
"loss": 0.0437,
"step": 4050
},
{
"epoch": 2.91,
"learning_rate": 2.0622222222222223e-06,
"loss": 0.0384,
"step": 4075
},
{
"epoch": 2.93,
"learning_rate": 2.006666666666667e-06,
"loss": 0.0504,
"step": 4100
},
{
"epoch": 2.95,
"learning_rate": 1.9511111111111113e-06,
"loss": 0.0478,
"step": 4125
},
{
"epoch": 2.97,
"learning_rate": 1.8955555555555557e-06,
"loss": 0.0456,
"step": 4150
},
{
"epoch": 2.99,
"learning_rate": 1.8400000000000002e-06,
"loss": 0.0415,
"step": 4175
},
{
"epoch": 3.0,
"learning_rate": 1.7844444444444444e-06,
"loss": 0.0347,
"step": 4200
},
{
"epoch": 3.02,
"learning_rate": 1.728888888888889e-06,
"loss": 0.0238,
"step": 4225
},
{
"epoch": 3.04,
"learning_rate": 1.6733333333333335e-06,
"loss": 0.0244,
"step": 4250
},
{
"epoch": 3.06,
"learning_rate": 1.6177777777777778e-06,
"loss": 0.0208,
"step": 4275
},
{
"epoch": 3.08,
"learning_rate": 1.5622222222222225e-06,
"loss": 0.0188,
"step": 4300
},
{
"epoch": 3.09,
"learning_rate": 1.506666666666667e-06,
"loss": 0.0184,
"step": 4325
},
{
"epoch": 3.11,
"learning_rate": 1.4511111111111112e-06,
"loss": 0.0213,
"step": 4350
},
{
"epoch": 3.13,
"learning_rate": 1.3955555555555556e-06,
"loss": 0.0159,
"step": 4375
},
{
"epoch": 3.15,
"learning_rate": 1.3422222222222222e-06,
"loss": 0.0223,
"step": 4400
},
{
"epoch": 3.17,
"learning_rate": 1.286666666666667e-06,
"loss": 0.0269,
"step": 4425
},
{
"epoch": 3.18,
"learning_rate": 1.2311111111111112e-06,
"loss": 0.0215,
"step": 4450
},
{
"epoch": 3.2,
"learning_rate": 1.1755555555555556e-06,
"loss": 0.0194,
"step": 4475
},
{
"epoch": 3.22,
"learning_rate": 1.12e-06,
"loss": 0.018,
"step": 4500
},
{
"epoch": 3.24,
"learning_rate": 1.0644444444444445e-06,
"loss": 0.0226,
"step": 4525
},
{
"epoch": 3.25,
"learning_rate": 1.008888888888889e-06,
"loss": 0.0171,
"step": 4550
},
{
"epoch": 3.27,
"learning_rate": 9.533333333333335e-07,
"loss": 0.0305,
"step": 4575
},
{
"epoch": 3.29,
"learning_rate": 8.977777777777778e-07,
"loss": 0.0252,
"step": 4600
},
{
"epoch": 3.31,
"learning_rate": 8.422222222222224e-07,
"loss": 0.0172,
"step": 4625
},
{
"epoch": 3.33,
"learning_rate": 7.866666666666667e-07,
"loss": 0.0235,
"step": 4650
},
{
"epoch": 3.34,
"learning_rate": 7.311111111111112e-07,
"loss": 0.0111,
"step": 4675
},
{
"epoch": 3.36,
"learning_rate": 6.755555555555555e-07,
"loss": 0.0191,
"step": 4700
},
{
"epoch": 3.38,
"learning_rate": 6.200000000000001e-07,
"loss": 0.019,
"step": 4725
},
{
"epoch": 3.4,
"learning_rate": 5.644444444444445e-07,
"loss": 0.016,
"step": 4750
},
{
"epoch": 3.42,
"learning_rate": 5.088888888888889e-07,
"loss": 0.0183,
"step": 4775
},
{
"epoch": 3.43,
"learning_rate": 4.533333333333334e-07,
"loss": 0.0175,
"step": 4800
},
{
"epoch": 3.45,
"learning_rate": 3.9777777777777783e-07,
"loss": 0.0193,
"step": 4825
},
{
"epoch": 3.47,
"learning_rate": 3.422222222222223e-07,
"loss": 0.0205,
"step": 4850
},
{
"epoch": 3.49,
"learning_rate": 2.866666666666667e-07,
"loss": 0.0183,
"step": 4875
},
{
"epoch": 3.51,
"learning_rate": 2.3111111111111112e-07,
"loss": 0.0178,
"step": 4900
},
{
"epoch": 3.52,
"learning_rate": 1.7555555555555558e-07,
"loss": 0.0211,
"step": 4925
},
{
"epoch": 3.54,
"learning_rate": 1.2000000000000002e-07,
"loss": 0.0236,
"step": 4950
},
{
"epoch": 3.56,
"learning_rate": 6.444444444444445e-08,
"loss": 0.0171,
"step": 4975
},
{
"epoch": 3.58,
"learning_rate": 8.88888888888889e-09,
"loss": 0.0176,
"step": 5000
},
{
"epoch": 3.58,
"eval_loss": 0.24974194169044495,
"eval_runtime": 596.4514,
"eval_samples_per_second": 6.073,
"eval_steps_per_second": 1.519,
"eval_wer": 11.893259719346675,
"step": 5000
},
{
"epoch": 3.58,
"step": 5000,
"total_flos": 1.154168450039808e+19,
"train_loss": 0.1590483027100563,
"train_runtime": 7812.7244,
"train_samples_per_second": 5.12,
"train_steps_per_second": 0.64
}
],
"max_steps": 5000,
"num_train_epochs": 4,
"total_flos": 1.154168450039808e+19,
"trial_name": null,
"trial_params": null
}