yt-special-batch4-tiny / trainer_state.json
TheRains's picture
End of training
ac5500b
raw
history blame contribute delete
No virus
25.8 kB
{
"best_metric": 85.86991764477013,
"best_model_checkpoint": "./output/small/yt-special-batch4-tiny/checkpoint-5000",
"epoch": 0.4294425835265825,
"global_step": 5000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.0000000000000003e-07,
"loss": 3.1379,
"step": 25
},
{
"epoch": 0.0,
"learning_rate": 9.000000000000001e-07,
"loss": 2.9263,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.4000000000000001e-06,
"loss": 2.5574,
"step": 75
},
{
"epoch": 0.01,
"learning_rate": 1.9000000000000002e-06,
"loss": 2.5256,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 2.4000000000000003e-06,
"loss": 2.4408,
"step": 125
},
{
"epoch": 0.01,
"learning_rate": 2.9e-06,
"loss": 2.3863,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 3.4000000000000005e-06,
"loss": 2.0215,
"step": 175
},
{
"epoch": 0.02,
"learning_rate": 3.900000000000001e-06,
"loss": 1.9868,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 4.4e-06,
"loss": 1.9312,
"step": 225
},
{
"epoch": 0.02,
"learning_rate": 4.9000000000000005e-06,
"loss": 1.7566,
"step": 250
},
{
"epoch": 0.02,
"learning_rate": 5.400000000000001e-06,
"loss": 1.9415,
"step": 275
},
{
"epoch": 0.03,
"learning_rate": 5.9e-06,
"loss": 1.5096,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 6.4000000000000006e-06,
"loss": 1.5393,
"step": 325
},
{
"epoch": 0.03,
"learning_rate": 6.9e-06,
"loss": 1.6095,
"step": 350
},
{
"epoch": 0.03,
"learning_rate": 7.4e-06,
"loss": 1.5683,
"step": 375
},
{
"epoch": 0.03,
"learning_rate": 7.9e-06,
"loss": 1.621,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 8.400000000000001e-06,
"loss": 1.7335,
"step": 425
},
{
"epoch": 0.04,
"learning_rate": 8.900000000000001e-06,
"loss": 1.6016,
"step": 450
},
{
"epoch": 0.04,
"learning_rate": 9.4e-06,
"loss": 1.723,
"step": 475
},
{
"epoch": 0.04,
"learning_rate": 9.9e-06,
"loss": 1.539,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 9.955555555555556e-06,
"loss": 1.3885,
"step": 525
},
{
"epoch": 0.05,
"learning_rate": 9.9e-06,
"loss": 1.8232,
"step": 550
},
{
"epoch": 0.05,
"learning_rate": 9.844444444444446e-06,
"loss": 1.6468,
"step": 575
},
{
"epoch": 0.05,
"learning_rate": 9.78888888888889e-06,
"loss": 1.6155,
"step": 600
},
{
"epoch": 0.05,
"learning_rate": 9.733333333333334e-06,
"loss": 1.461,
"step": 625
},
{
"epoch": 0.06,
"learning_rate": 9.677777777777778e-06,
"loss": 1.5219,
"step": 650
},
{
"epoch": 0.06,
"learning_rate": 9.622222222222222e-06,
"loss": 1.6706,
"step": 675
},
{
"epoch": 0.06,
"learning_rate": 9.566666666666668e-06,
"loss": 1.4404,
"step": 700
},
{
"epoch": 0.06,
"learning_rate": 9.511111111111112e-06,
"loss": 1.7742,
"step": 725
},
{
"epoch": 0.06,
"learning_rate": 9.455555555555557e-06,
"loss": 1.7022,
"step": 750
},
{
"epoch": 0.07,
"learning_rate": 9.4e-06,
"loss": 1.5377,
"step": 775
},
{
"epoch": 0.07,
"learning_rate": 9.344444444444446e-06,
"loss": 1.4674,
"step": 800
},
{
"epoch": 0.07,
"learning_rate": 9.28888888888889e-06,
"loss": 1.3616,
"step": 825
},
{
"epoch": 0.07,
"learning_rate": 9.233333333333334e-06,
"loss": 1.3786,
"step": 850
},
{
"epoch": 0.08,
"learning_rate": 9.17777777777778e-06,
"loss": 1.5969,
"step": 875
},
{
"epoch": 0.08,
"learning_rate": 9.122222222222223e-06,
"loss": 1.6339,
"step": 900
},
{
"epoch": 0.08,
"learning_rate": 9.066666666666667e-06,
"loss": 1.4945,
"step": 925
},
{
"epoch": 0.08,
"learning_rate": 9.011111111111111e-06,
"loss": 1.5719,
"step": 950
},
{
"epoch": 0.08,
"learning_rate": 8.955555555555555e-06,
"loss": 1.3626,
"step": 975
},
{
"epoch": 0.09,
"learning_rate": 8.900000000000001e-06,
"loss": 1.6736,
"step": 1000
},
{
"epoch": 0.09,
"eval_loss": 1.4867454767227173,
"eval_runtime": 314.227,
"eval_samples_per_second": 7.8,
"eval_steps_per_second": 3.902,
"eval_wer": 126.28234226055379,
"step": 1000
},
{
"epoch": 0.09,
"learning_rate": 8.844444444444445e-06,
"loss": 1.4547,
"step": 1025
},
{
"epoch": 0.09,
"learning_rate": 8.788888888888891e-06,
"loss": 1.4689,
"step": 1050
},
{
"epoch": 0.09,
"learning_rate": 8.733333333333333e-06,
"loss": 1.6107,
"step": 1075
},
{
"epoch": 0.09,
"learning_rate": 8.677777777777779e-06,
"loss": 1.427,
"step": 1100
},
{
"epoch": 0.1,
"learning_rate": 8.622222222222223e-06,
"loss": 1.5093,
"step": 1125
},
{
"epoch": 0.1,
"learning_rate": 8.566666666666667e-06,
"loss": 1.4465,
"step": 1150
},
{
"epoch": 0.1,
"learning_rate": 8.511111111111113e-06,
"loss": 1.5751,
"step": 1175
},
{
"epoch": 0.1,
"learning_rate": 8.455555555555555e-06,
"loss": 1.6314,
"step": 1200
},
{
"epoch": 0.11,
"learning_rate": 8.400000000000001e-06,
"loss": 1.5775,
"step": 1225
},
{
"epoch": 0.11,
"learning_rate": 8.344444444444445e-06,
"loss": 1.3122,
"step": 1250
},
{
"epoch": 0.11,
"learning_rate": 8.288888888888889e-06,
"loss": 1.5478,
"step": 1275
},
{
"epoch": 0.11,
"learning_rate": 8.233333333333335e-06,
"loss": 1.4037,
"step": 1300
},
{
"epoch": 0.11,
"learning_rate": 8.177777777777779e-06,
"loss": 1.586,
"step": 1325
},
{
"epoch": 0.12,
"learning_rate": 8.122222222222223e-06,
"loss": 1.5164,
"step": 1350
},
{
"epoch": 0.12,
"learning_rate": 8.066666666666667e-06,
"loss": 1.5237,
"step": 1375
},
{
"epoch": 0.12,
"learning_rate": 8.011111111111113e-06,
"loss": 1.2344,
"step": 1400
},
{
"epoch": 0.12,
"learning_rate": 7.955555555555557e-06,
"loss": 1.4881,
"step": 1425
},
{
"epoch": 0.12,
"learning_rate": 7.9e-06,
"loss": 1.4276,
"step": 1450
},
{
"epoch": 0.13,
"learning_rate": 7.844444444444446e-06,
"loss": 1.5571,
"step": 1475
},
{
"epoch": 0.13,
"learning_rate": 7.788888888888889e-06,
"loss": 1.3892,
"step": 1500
},
{
"epoch": 0.13,
"learning_rate": 7.733333333333334e-06,
"loss": 1.4613,
"step": 1525
},
{
"epoch": 0.13,
"learning_rate": 7.677777777777778e-06,
"loss": 1.2587,
"step": 1550
},
{
"epoch": 0.14,
"learning_rate": 7.622222222222223e-06,
"loss": 1.4007,
"step": 1575
},
{
"epoch": 0.14,
"learning_rate": 7.566666666666667e-06,
"loss": 1.3246,
"step": 1600
},
{
"epoch": 0.14,
"learning_rate": 7.511111111111111e-06,
"loss": 1.4026,
"step": 1625
},
{
"epoch": 0.14,
"learning_rate": 7.455555555555556e-06,
"loss": 1.3185,
"step": 1650
},
{
"epoch": 0.14,
"learning_rate": 7.4e-06,
"loss": 1.3796,
"step": 1675
},
{
"epoch": 0.15,
"learning_rate": 7.344444444444445e-06,
"loss": 1.2145,
"step": 1700
},
{
"epoch": 0.15,
"learning_rate": 7.28888888888889e-06,
"loss": 1.4185,
"step": 1725
},
{
"epoch": 0.15,
"learning_rate": 7.233333333333334e-06,
"loss": 1.3927,
"step": 1750
},
{
"epoch": 0.15,
"learning_rate": 7.177777777777778e-06,
"loss": 1.4451,
"step": 1775
},
{
"epoch": 0.15,
"learning_rate": 7.122222222222222e-06,
"loss": 1.3688,
"step": 1800
},
{
"epoch": 0.16,
"learning_rate": 7.066666666666667e-06,
"loss": 1.5201,
"step": 1825
},
{
"epoch": 0.16,
"learning_rate": 7.011111111111112e-06,
"loss": 1.3194,
"step": 1850
},
{
"epoch": 0.16,
"learning_rate": 6.955555555555557e-06,
"loss": 1.3846,
"step": 1875
},
{
"epoch": 0.16,
"learning_rate": 6.9e-06,
"loss": 1.2325,
"step": 1900
},
{
"epoch": 0.17,
"learning_rate": 6.844444444444445e-06,
"loss": 1.4205,
"step": 1925
},
{
"epoch": 0.17,
"learning_rate": 6.788888888888889e-06,
"loss": 1.3469,
"step": 1950
},
{
"epoch": 0.17,
"learning_rate": 6.733333333333334e-06,
"loss": 1.4563,
"step": 1975
},
{
"epoch": 0.17,
"learning_rate": 6.677777777777779e-06,
"loss": 1.3354,
"step": 2000
},
{
"epoch": 0.17,
"eval_loss": 1.358388066291809,
"eval_runtime": 409.515,
"eval_samples_per_second": 5.985,
"eval_steps_per_second": 2.994,
"eval_wer": 103.02185331690552,
"step": 2000
},
{
"epoch": 0.17,
"learning_rate": 6.6222222222222236e-06,
"loss": 1.3947,
"step": 2025
},
{
"epoch": 0.18,
"learning_rate": 6.566666666666667e-06,
"loss": 1.6198,
"step": 2050
},
{
"epoch": 0.18,
"learning_rate": 6.511111111111112e-06,
"loss": 1.3082,
"step": 2075
},
{
"epoch": 0.18,
"learning_rate": 6.455555555555556e-06,
"loss": 1.2893,
"step": 2100
},
{
"epoch": 0.18,
"learning_rate": 6.4000000000000006e-06,
"loss": 1.3989,
"step": 2125
},
{
"epoch": 0.18,
"learning_rate": 6.3444444444444454e-06,
"loss": 1.4822,
"step": 2150
},
{
"epoch": 0.19,
"learning_rate": 6.28888888888889e-06,
"loss": 1.3797,
"step": 2175
},
{
"epoch": 0.19,
"learning_rate": 6.2333333333333335e-06,
"loss": 1.3175,
"step": 2200
},
{
"epoch": 0.19,
"learning_rate": 6.177777777777778e-06,
"loss": 1.3237,
"step": 2225
},
{
"epoch": 0.19,
"learning_rate": 6.1222222222222224e-06,
"loss": 1.3438,
"step": 2250
},
{
"epoch": 0.2,
"learning_rate": 6.066666666666667e-06,
"loss": 1.4311,
"step": 2275
},
{
"epoch": 0.2,
"learning_rate": 6.011111111111112e-06,
"loss": 1.5698,
"step": 2300
},
{
"epoch": 0.2,
"learning_rate": 5.955555555555555e-06,
"loss": 1.3753,
"step": 2325
},
{
"epoch": 0.2,
"learning_rate": 5.9e-06,
"loss": 1.6377,
"step": 2350
},
{
"epoch": 0.2,
"learning_rate": 5.844444444444445e-06,
"loss": 1.4404,
"step": 2375
},
{
"epoch": 0.21,
"learning_rate": 5.788888888888889e-06,
"loss": 1.2067,
"step": 2400
},
{
"epoch": 0.21,
"learning_rate": 5.733333333333334e-06,
"loss": 1.3581,
"step": 2425
},
{
"epoch": 0.21,
"learning_rate": 5.677777777777779e-06,
"loss": 1.336,
"step": 2450
},
{
"epoch": 0.21,
"learning_rate": 5.622222222222222e-06,
"loss": 1.4442,
"step": 2475
},
{
"epoch": 0.21,
"learning_rate": 5.566666666666667e-06,
"loss": 1.4843,
"step": 2500
},
{
"epoch": 0.22,
"learning_rate": 5.511111111111112e-06,
"loss": 1.2732,
"step": 2525
},
{
"epoch": 0.22,
"learning_rate": 5.455555555555556e-06,
"loss": 1.1735,
"step": 2550
},
{
"epoch": 0.22,
"learning_rate": 5.400000000000001e-06,
"loss": 1.221,
"step": 2575
},
{
"epoch": 0.22,
"learning_rate": 5.344444444444446e-06,
"loss": 1.5046,
"step": 2600
},
{
"epoch": 0.23,
"learning_rate": 5.288888888888889e-06,
"loss": 1.3763,
"step": 2625
},
{
"epoch": 0.23,
"learning_rate": 5.233333333333334e-06,
"loss": 1.2299,
"step": 2650
},
{
"epoch": 0.23,
"learning_rate": 5.177777777777779e-06,
"loss": 1.5209,
"step": 2675
},
{
"epoch": 0.23,
"learning_rate": 5.122222222222223e-06,
"loss": 1.1238,
"step": 2700
},
{
"epoch": 0.23,
"learning_rate": 5.0666666666666676e-06,
"loss": 1.3985,
"step": 2725
},
{
"epoch": 0.24,
"learning_rate": 5.011111111111111e-06,
"loss": 1.3149,
"step": 2750
},
{
"epoch": 0.24,
"learning_rate": 4.9555555555555565e-06,
"loss": 1.3007,
"step": 2775
},
{
"epoch": 0.24,
"learning_rate": 4.9000000000000005e-06,
"loss": 1.3239,
"step": 2800
},
{
"epoch": 0.24,
"learning_rate": 4.8444444444444446e-06,
"loss": 1.1774,
"step": 2825
},
{
"epoch": 0.24,
"learning_rate": 4.7888888888888894e-06,
"loss": 1.5724,
"step": 2850
},
{
"epoch": 0.25,
"learning_rate": 4.7333333333333335e-06,
"loss": 1.4087,
"step": 2875
},
{
"epoch": 0.25,
"learning_rate": 4.677777777777778e-06,
"loss": 1.1499,
"step": 2900
},
{
"epoch": 0.25,
"learning_rate": 4.622222222222222e-06,
"loss": 1.376,
"step": 2925
},
{
"epoch": 0.25,
"learning_rate": 4.566666666666667e-06,
"loss": 1.2999,
"step": 2950
},
{
"epoch": 0.26,
"learning_rate": 4.511111111111111e-06,
"loss": 1.1805,
"step": 2975
},
{
"epoch": 0.26,
"learning_rate": 4.455555555555555e-06,
"loss": 1.4841,
"step": 3000
},
{
"epoch": 0.26,
"eval_loss": 1.293558120727539,
"eval_runtime": 374.795,
"eval_samples_per_second": 6.54,
"eval_steps_per_second": 3.271,
"eval_wer": 86.33032877245314,
"step": 3000
},
{
"epoch": 0.26,
"learning_rate": 4.4e-06,
"loss": 1.4241,
"step": 3025
},
{
"epoch": 0.26,
"learning_rate": 4.344444444444445e-06,
"loss": 1.5764,
"step": 3050
},
{
"epoch": 0.26,
"learning_rate": 4.288888888888889e-06,
"loss": 1.2202,
"step": 3075
},
{
"epoch": 0.27,
"learning_rate": 4.233333333333334e-06,
"loss": 1.2979,
"step": 3100
},
{
"epoch": 0.27,
"learning_rate": 4.177777777777778e-06,
"loss": 1.3351,
"step": 3125
},
{
"epoch": 0.27,
"learning_rate": 4.122222222222222e-06,
"loss": 1.3778,
"step": 3150
},
{
"epoch": 0.27,
"learning_rate": 4.066666666666667e-06,
"loss": 1.3542,
"step": 3175
},
{
"epoch": 0.27,
"learning_rate": 4.011111111111111e-06,
"loss": 1.2818,
"step": 3200
},
{
"epoch": 0.28,
"learning_rate": 3.955555555555556e-06,
"loss": 1.1038,
"step": 3225
},
{
"epoch": 0.28,
"learning_rate": 3.900000000000001e-06,
"loss": 1.3161,
"step": 3250
},
{
"epoch": 0.28,
"learning_rate": 3.844444444444445e-06,
"loss": 1.286,
"step": 3275
},
{
"epoch": 0.28,
"learning_rate": 3.7888888888888893e-06,
"loss": 1.2946,
"step": 3300
},
{
"epoch": 0.29,
"learning_rate": 3.7333333333333337e-06,
"loss": 1.255,
"step": 3325
},
{
"epoch": 0.29,
"learning_rate": 3.6777777777777778e-06,
"loss": 1.3425,
"step": 3350
},
{
"epoch": 0.29,
"learning_rate": 3.6222222222222226e-06,
"loss": 1.2132,
"step": 3375
},
{
"epoch": 0.29,
"learning_rate": 3.566666666666667e-06,
"loss": 1.2291,
"step": 3400
},
{
"epoch": 0.29,
"learning_rate": 3.511111111111111e-06,
"loss": 1.4048,
"step": 3425
},
{
"epoch": 0.3,
"learning_rate": 3.455555555555556e-06,
"loss": 1.2253,
"step": 3450
},
{
"epoch": 0.3,
"learning_rate": 3.4000000000000005e-06,
"loss": 1.3325,
"step": 3475
},
{
"epoch": 0.3,
"learning_rate": 3.3444444444444445e-06,
"loss": 1.1411,
"step": 3500
},
{
"epoch": 0.3,
"learning_rate": 3.2888888888888894e-06,
"loss": 1.4232,
"step": 3525
},
{
"epoch": 0.3,
"learning_rate": 3.2333333333333334e-06,
"loss": 1.2711,
"step": 3550
},
{
"epoch": 0.31,
"learning_rate": 3.177777777777778e-06,
"loss": 1.2715,
"step": 3575
},
{
"epoch": 0.31,
"learning_rate": 3.1222222222222228e-06,
"loss": 1.1375,
"step": 3600
},
{
"epoch": 0.31,
"learning_rate": 3.066666666666667e-06,
"loss": 1.2082,
"step": 3625
},
{
"epoch": 0.31,
"learning_rate": 3.0111111111111113e-06,
"loss": 1.4208,
"step": 3650
},
{
"epoch": 0.32,
"learning_rate": 2.955555555555556e-06,
"loss": 1.3165,
"step": 3675
},
{
"epoch": 0.32,
"learning_rate": 2.9e-06,
"loss": 1.3529,
"step": 3700
},
{
"epoch": 0.32,
"learning_rate": 2.8444444444444446e-06,
"loss": 1.2213,
"step": 3725
},
{
"epoch": 0.32,
"learning_rate": 2.788888888888889e-06,
"loss": 1.1965,
"step": 3750
},
{
"epoch": 0.32,
"learning_rate": 2.7333333333333336e-06,
"loss": 1.3351,
"step": 3775
},
{
"epoch": 0.33,
"learning_rate": 2.677777777777778e-06,
"loss": 1.2318,
"step": 3800
},
{
"epoch": 0.33,
"learning_rate": 2.6222222222222225e-06,
"loss": 1.3696,
"step": 3825
},
{
"epoch": 0.33,
"learning_rate": 2.566666666666667e-06,
"loss": 1.2649,
"step": 3850
},
{
"epoch": 0.33,
"learning_rate": 2.5111111111111114e-06,
"loss": 1.3345,
"step": 3875
},
{
"epoch": 0.33,
"learning_rate": 2.455555555555556e-06,
"loss": 1.3604,
"step": 3900
},
{
"epoch": 0.34,
"learning_rate": 2.4000000000000003e-06,
"loss": 1.3837,
"step": 3925
},
{
"epoch": 0.34,
"learning_rate": 2.3444444444444448e-06,
"loss": 1.1288,
"step": 3950
},
{
"epoch": 0.34,
"learning_rate": 2.2888888888888892e-06,
"loss": 1.274,
"step": 3975
},
{
"epoch": 0.34,
"learning_rate": 2.2333333333333333e-06,
"loss": 1.2807,
"step": 4000
},
{
"epoch": 0.34,
"eval_loss": 1.2523268461227417,
"eval_runtime": 336.5203,
"eval_samples_per_second": 7.283,
"eval_steps_per_second": 3.643,
"eval_wer": 85.94773361001232,
"step": 4000
},
{
"epoch": 0.35,
"learning_rate": 2.1777777777777777e-06,
"loss": 1.2628,
"step": 4025
},
{
"epoch": 0.35,
"learning_rate": 2.1222222222222226e-06,
"loss": 1.2199,
"step": 4050
},
{
"epoch": 0.35,
"learning_rate": 2.0666666666666666e-06,
"loss": 1.1674,
"step": 4075
},
{
"epoch": 0.35,
"learning_rate": 2.011111111111111e-06,
"loss": 1.456,
"step": 4100
},
{
"epoch": 0.35,
"learning_rate": 1.955555555555556e-06,
"loss": 1.2439,
"step": 4125
},
{
"epoch": 0.36,
"learning_rate": 1.9000000000000002e-06,
"loss": 1.3638,
"step": 4150
},
{
"epoch": 0.36,
"learning_rate": 1.8444444444444445e-06,
"loss": 1.0733,
"step": 4175
},
{
"epoch": 0.36,
"learning_rate": 1.788888888888889e-06,
"loss": 1.3064,
"step": 4200
},
{
"epoch": 0.36,
"learning_rate": 1.7333333333333336e-06,
"loss": 1.1729,
"step": 4225
},
{
"epoch": 0.37,
"learning_rate": 1.6777777777777779e-06,
"loss": 1.1621,
"step": 4250
},
{
"epoch": 0.37,
"learning_rate": 1.6222222222222223e-06,
"loss": 1.1784,
"step": 4275
},
{
"epoch": 0.37,
"learning_rate": 1.566666666666667e-06,
"loss": 1.487,
"step": 4300
},
{
"epoch": 0.37,
"learning_rate": 1.5111111111111112e-06,
"loss": 1.216,
"step": 4325
},
{
"epoch": 0.37,
"learning_rate": 1.4555555555555557e-06,
"loss": 1.3695,
"step": 4350
},
{
"epoch": 0.38,
"learning_rate": 1.4000000000000001e-06,
"loss": 1.2919,
"step": 4375
},
{
"epoch": 0.38,
"learning_rate": 1.3444444444444446e-06,
"loss": 1.2615,
"step": 4400
},
{
"epoch": 0.38,
"learning_rate": 1.288888888888889e-06,
"loss": 1.3932,
"step": 4425
},
{
"epoch": 0.38,
"learning_rate": 1.2333333333333335e-06,
"loss": 1.1616,
"step": 4450
},
{
"epoch": 0.38,
"learning_rate": 1.1777777777777778e-06,
"loss": 1.156,
"step": 4475
},
{
"epoch": 0.39,
"learning_rate": 1.1222222222222222e-06,
"loss": 1.2862,
"step": 4500
},
{
"epoch": 0.39,
"learning_rate": 1.066666666666667e-06,
"loss": 1.3127,
"step": 4525
},
{
"epoch": 0.39,
"learning_rate": 1.0111111111111111e-06,
"loss": 1.0732,
"step": 4550
},
{
"epoch": 0.39,
"learning_rate": 9.555555555555556e-07,
"loss": 1.3942,
"step": 4575
},
{
"epoch": 0.4,
"learning_rate": 9.000000000000001e-07,
"loss": 1.3173,
"step": 4600
},
{
"epoch": 0.4,
"learning_rate": 8.444444444444445e-07,
"loss": 1.3052,
"step": 4625
},
{
"epoch": 0.4,
"learning_rate": 7.888888888888889e-07,
"loss": 1.256,
"step": 4650
},
{
"epoch": 0.4,
"learning_rate": 7.333333333333334e-07,
"loss": 1.2368,
"step": 4675
},
{
"epoch": 0.4,
"learning_rate": 6.777777777777779e-07,
"loss": 1.4347,
"step": 4700
},
{
"epoch": 0.41,
"learning_rate": 6.222222222222223e-07,
"loss": 1.3749,
"step": 4725
},
{
"epoch": 0.41,
"learning_rate": 5.666666666666667e-07,
"loss": 1.2192,
"step": 4750
},
{
"epoch": 0.41,
"learning_rate": 5.111111111111112e-07,
"loss": 1.2106,
"step": 4775
},
{
"epoch": 0.41,
"learning_rate": 4.5555555555555563e-07,
"loss": 1.2094,
"step": 4800
},
{
"epoch": 0.41,
"learning_rate": 4.0000000000000003e-07,
"loss": 1.1827,
"step": 4825
},
{
"epoch": 0.42,
"learning_rate": 3.444444444444445e-07,
"loss": 1.262,
"step": 4850
},
{
"epoch": 0.42,
"learning_rate": 2.888888888888889e-07,
"loss": 1.3733,
"step": 4875
},
{
"epoch": 0.42,
"learning_rate": 2.3333333333333336e-07,
"loss": 1.1658,
"step": 4900
},
{
"epoch": 0.42,
"learning_rate": 1.777777777777778e-07,
"loss": 1.2891,
"step": 4925
},
{
"epoch": 0.43,
"learning_rate": 1.2222222222222225e-07,
"loss": 1.1855,
"step": 4950
},
{
"epoch": 0.43,
"learning_rate": 6.888888888888889e-08,
"loss": 1.2086,
"step": 4975
},
{
"epoch": 0.43,
"learning_rate": 1.3333333333333334e-08,
"loss": 1.2095,
"step": 5000
},
{
"epoch": 0.43,
"eval_loss": 1.2324541807174683,
"eval_runtime": 387.4959,
"eval_samples_per_second": 6.325,
"eval_steps_per_second": 3.164,
"eval_wer": 85.86991764477013,
"step": 5000
},
{
"epoch": 0.43,
"step": 5000,
"total_flos": 4.923777024e+17,
"train_loss": 1.4253231132507325,
"train_runtime": 2913.599,
"train_samples_per_second": 6.864,
"train_steps_per_second": 1.716
}
],
"max_steps": 5000,
"num_train_epochs": 1,
"total_flos": 4.923777024e+17,
"trial_name": null,
"trial_params": null
}