malmarz's picture
Training in progress, step 1000
d09a13e
raw
history blame
No virus
51.2 kB
{
"best_metric": 20.95505699346784,
"best_model_checkpoint": "./checkpoint-8000",
"epoch": 1.4375,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4.4e-07,
"loss": 0.8245,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 9.400000000000001e-07,
"loss": 0.7486,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 1.44e-06,
"loss": 0.7559,
"step": 75
},
{
"epoch": 0.01,
"learning_rate": 1.94e-06,
"loss": 0.6703,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 2.4400000000000004e-06,
"loss": 0.5852,
"step": 125
},
{
"epoch": 0.01,
"learning_rate": 2.9400000000000002e-06,
"loss": 0.5794,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 3.44e-06,
"loss": 0.5279,
"step": 175
},
{
"epoch": 0.02,
"learning_rate": 3.94e-06,
"loss": 0.4985,
"step": 200
},
{
"epoch": 0.02,
"learning_rate": 4.440000000000001e-06,
"loss": 0.5065,
"step": 225
},
{
"epoch": 0.03,
"learning_rate": 4.94e-06,
"loss": 0.5098,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 5.4400000000000004e-06,
"loss": 0.46,
"step": 275
},
{
"epoch": 0.03,
"learning_rate": 5.94e-06,
"loss": 0.4321,
"step": 300
},
{
"epoch": 0.03,
"learning_rate": 6.440000000000001e-06,
"loss": 0.4862,
"step": 325
},
{
"epoch": 0.04,
"learning_rate": 6.9400000000000005e-06,
"loss": 0.4822,
"step": 350
},
{
"epoch": 0.04,
"learning_rate": 7.440000000000001e-06,
"loss": 0.485,
"step": 375
},
{
"epoch": 0.04,
"learning_rate": 7.94e-06,
"loss": 0.466,
"step": 400
},
{
"epoch": 0.04,
"learning_rate": 8.44e-06,
"loss": 0.4402,
"step": 425
},
{
"epoch": 0.04,
"learning_rate": 8.94e-06,
"loss": 0.4634,
"step": 450
},
{
"epoch": 0.05,
"learning_rate": 9.440000000000001e-06,
"loss": 0.4424,
"step": 475
},
{
"epoch": 0.05,
"learning_rate": 9.940000000000001e-06,
"loss": 0.414,
"step": 500
},
{
"epoch": 0.05,
"learning_rate": 9.976842105263158e-06,
"loss": 0.4266,
"step": 525
},
{
"epoch": 0.06,
"learning_rate": 9.950526315789475e-06,
"loss": 0.4099,
"step": 550
},
{
"epoch": 0.06,
"learning_rate": 9.92421052631579e-06,
"loss": 0.4076,
"step": 575
},
{
"epoch": 0.06,
"learning_rate": 9.897894736842107e-06,
"loss": 0.4467,
"step": 600
},
{
"epoch": 0.06,
"learning_rate": 9.871578947368422e-06,
"loss": 0.4543,
"step": 625
},
{
"epoch": 0.07,
"learning_rate": 9.845263157894738e-06,
"loss": 0.4051,
"step": 650
},
{
"epoch": 0.07,
"learning_rate": 9.818947368421053e-06,
"loss": 0.4589,
"step": 675
},
{
"epoch": 0.07,
"learning_rate": 9.79263157894737e-06,
"loss": 0.4,
"step": 700
},
{
"epoch": 0.07,
"learning_rate": 9.766315789473685e-06,
"loss": 0.4261,
"step": 725
},
{
"epoch": 0.07,
"learning_rate": 9.74e-06,
"loss": 0.3796,
"step": 750
},
{
"epoch": 0.08,
"learning_rate": 9.713684210526317e-06,
"loss": 0.3903,
"step": 775
},
{
"epoch": 0.08,
"learning_rate": 9.687368421052632e-06,
"loss": 0.3795,
"step": 800
},
{
"epoch": 0.08,
"learning_rate": 9.661052631578948e-06,
"loss": 0.3979,
"step": 825
},
{
"epoch": 0.09,
"learning_rate": 9.634736842105265e-06,
"loss": 0.3664,
"step": 850
},
{
"epoch": 0.09,
"learning_rate": 9.60842105263158e-06,
"loss": 0.3989,
"step": 875
},
{
"epoch": 0.09,
"learning_rate": 9.582105263157897e-06,
"loss": 0.3673,
"step": 900
},
{
"epoch": 0.09,
"learning_rate": 9.555789473684211e-06,
"loss": 0.3911,
"step": 925
},
{
"epoch": 0.1,
"learning_rate": 9.529473684210528e-06,
"loss": 0.4083,
"step": 950
},
{
"epoch": 0.1,
"learning_rate": 9.503157894736843e-06,
"loss": 0.3819,
"step": 975
},
{
"epoch": 0.1,
"learning_rate": 9.476842105263158e-06,
"loss": 0.3559,
"step": 1000
},
{
"epoch": 0.1,
"eval_loss": 0.9147018194198608,
"eval_runtime": 2177.5738,
"eval_samples_per_second": 2.597,
"eval_steps_per_second": 0.325,
"eval_wer": 29.325203134278638,
"step": 1000
},
{
"epoch": 0.1,
"learning_rate": 9.450526315789475e-06,
"loss": 0.3193,
"step": 1025
},
{
"epoch": 0.1,
"learning_rate": 9.42421052631579e-06,
"loss": 0.3426,
"step": 1050
},
{
"epoch": 0.11,
"learning_rate": 9.397894736842106e-06,
"loss": 0.4137,
"step": 1075
},
{
"epoch": 0.11,
"learning_rate": 9.371578947368421e-06,
"loss": 0.3362,
"step": 1100
},
{
"epoch": 0.11,
"learning_rate": 9.345263157894738e-06,
"loss": 0.3951,
"step": 1125
},
{
"epoch": 0.12,
"learning_rate": 9.318947368421053e-06,
"loss": 0.3939,
"step": 1150
},
{
"epoch": 0.12,
"learning_rate": 9.292631578947368e-06,
"loss": 0.3739,
"step": 1175
},
{
"epoch": 0.12,
"learning_rate": 9.266315789473685e-06,
"loss": 0.3616,
"step": 1200
},
{
"epoch": 0.12,
"learning_rate": 9.240000000000001e-06,
"loss": 0.3247,
"step": 1225
},
{
"epoch": 0.12,
"learning_rate": 9.213684210526316e-06,
"loss": 0.3423,
"step": 1250
},
{
"epoch": 0.13,
"learning_rate": 9.187368421052633e-06,
"loss": 0.3659,
"step": 1275
},
{
"epoch": 0.13,
"learning_rate": 9.161052631578948e-06,
"loss": 0.3295,
"step": 1300
},
{
"epoch": 0.13,
"learning_rate": 9.134736842105265e-06,
"loss": 0.3563,
"step": 1325
},
{
"epoch": 0.14,
"learning_rate": 9.10842105263158e-06,
"loss": 0.3258,
"step": 1350
},
{
"epoch": 0.14,
"learning_rate": 9.082105263157896e-06,
"loss": 0.3003,
"step": 1375
},
{
"epoch": 0.14,
"learning_rate": 9.055789473684211e-06,
"loss": 0.3747,
"step": 1400
},
{
"epoch": 0.14,
"learning_rate": 9.029473684210526e-06,
"loss": 0.3225,
"step": 1425
},
{
"epoch": 0.14,
"learning_rate": 9.003157894736843e-06,
"loss": 0.3518,
"step": 1450
},
{
"epoch": 0.15,
"learning_rate": 8.976842105263158e-06,
"loss": 0.3846,
"step": 1475
},
{
"epoch": 0.15,
"learning_rate": 8.950526315789474e-06,
"loss": 0.3811,
"step": 1500
},
{
"epoch": 0.15,
"learning_rate": 8.92421052631579e-06,
"loss": 0.3452,
"step": 1525
},
{
"epoch": 0.15,
"learning_rate": 8.897894736842106e-06,
"loss": 0.3514,
"step": 1550
},
{
"epoch": 0.16,
"learning_rate": 8.871578947368423e-06,
"loss": 0.35,
"step": 1575
},
{
"epoch": 0.16,
"learning_rate": 8.845263157894738e-06,
"loss": 0.3526,
"step": 1600
},
{
"epoch": 0.16,
"learning_rate": 8.818947368421054e-06,
"loss": 0.3386,
"step": 1625
},
{
"epoch": 0.17,
"learning_rate": 8.79263157894737e-06,
"loss": 0.3179,
"step": 1650
},
{
"epoch": 0.17,
"learning_rate": 8.766315789473684e-06,
"loss": 0.3247,
"step": 1675
},
{
"epoch": 0.17,
"learning_rate": 8.740000000000001e-06,
"loss": 0.3672,
"step": 1700
},
{
"epoch": 0.17,
"learning_rate": 8.713684210526316e-06,
"loss": 0.3134,
"step": 1725
},
{
"epoch": 0.17,
"learning_rate": 8.687368421052633e-06,
"loss": 0.3344,
"step": 1750
},
{
"epoch": 0.18,
"learning_rate": 8.661052631578948e-06,
"loss": 0.3481,
"step": 1775
},
{
"epoch": 0.18,
"learning_rate": 8.634736842105264e-06,
"loss": 0.3191,
"step": 1800
},
{
"epoch": 0.18,
"learning_rate": 8.60842105263158e-06,
"loss": 0.337,
"step": 1825
},
{
"epoch": 0.18,
"learning_rate": 8.582105263157894e-06,
"loss": 0.3192,
"step": 1850
},
{
"epoch": 0.19,
"learning_rate": 8.555789473684213e-06,
"loss": 0.3355,
"step": 1875
},
{
"epoch": 0.19,
"learning_rate": 8.529473684210528e-06,
"loss": 0.3019,
"step": 1900
},
{
"epoch": 0.19,
"learning_rate": 8.503157894736843e-06,
"loss": 0.3267,
"step": 1925
},
{
"epoch": 0.2,
"learning_rate": 8.47684210526316e-06,
"loss": 0.2862,
"step": 1950
},
{
"epoch": 0.2,
"learning_rate": 8.450526315789474e-06,
"loss": 0.3319,
"step": 1975
},
{
"epoch": 0.2,
"learning_rate": 8.42421052631579e-06,
"loss": 0.3154,
"step": 2000
},
{
"epoch": 0.2,
"eval_loss": 1.1352506875991821,
"eval_runtime": 2066.5011,
"eval_samples_per_second": 2.737,
"eval_steps_per_second": 0.342,
"eval_wer": 26.57184653042307,
"step": 2000
},
{
"epoch": 0.2,
"learning_rate": 8.397894736842106e-06,
"loss": 0.3281,
"step": 2025
},
{
"epoch": 0.2,
"learning_rate": 8.371578947368422e-06,
"loss": 0.3229,
"step": 2050
},
{
"epoch": 0.21,
"learning_rate": 8.345263157894737e-06,
"loss": 0.3385,
"step": 2075
},
{
"epoch": 0.21,
"learning_rate": 8.318947368421052e-06,
"loss": 0.2919,
"step": 2100
},
{
"epoch": 0.21,
"learning_rate": 8.292631578947369e-06,
"loss": 0.3417,
"step": 2125
},
{
"epoch": 0.21,
"learning_rate": 8.266315789473684e-06,
"loss": 0.32,
"step": 2150
},
{
"epoch": 0.22,
"learning_rate": 8.24e-06,
"loss": 0.344,
"step": 2175
},
{
"epoch": 0.22,
"learning_rate": 8.213684210526316e-06,
"loss": 0.323,
"step": 2200
},
{
"epoch": 0.22,
"learning_rate": 8.187368421052632e-06,
"loss": 0.3176,
"step": 2225
},
{
"epoch": 0.23,
"learning_rate": 8.161052631578949e-06,
"loss": 0.3049,
"step": 2250
},
{
"epoch": 0.23,
"learning_rate": 8.134736842105264e-06,
"loss": 0.3285,
"step": 2275
},
{
"epoch": 0.23,
"learning_rate": 8.10842105263158e-06,
"loss": 0.3404,
"step": 2300
},
{
"epoch": 0.23,
"learning_rate": 8.082105263157896e-06,
"loss": 0.3253,
"step": 2325
},
{
"epoch": 0.23,
"learning_rate": 8.05578947368421e-06,
"loss": 0.3023,
"step": 2350
},
{
"epoch": 0.24,
"learning_rate": 8.029473684210527e-06,
"loss": 0.3258,
"step": 2375
},
{
"epoch": 0.24,
"learning_rate": 8.003157894736842e-06,
"loss": 0.3421,
"step": 2400
},
{
"epoch": 0.24,
"learning_rate": 7.976842105263159e-06,
"loss": 0.3451,
"step": 2425
},
{
"epoch": 0.24,
"learning_rate": 7.950526315789474e-06,
"loss": 0.3183,
"step": 2450
},
{
"epoch": 0.25,
"learning_rate": 7.92421052631579e-06,
"loss": 0.2939,
"step": 2475
},
{
"epoch": 0.25,
"learning_rate": 7.897894736842106e-06,
"loss": 0.3006,
"step": 2500
},
{
"epoch": 0.25,
"learning_rate": 7.87157894736842e-06,
"loss": 0.3299,
"step": 2525
},
{
"epoch": 0.26,
"learning_rate": 7.845263157894739e-06,
"loss": 0.3364,
"step": 2550
},
{
"epoch": 0.26,
"learning_rate": 7.818947368421054e-06,
"loss": 0.2966,
"step": 2575
},
{
"epoch": 0.26,
"learning_rate": 7.792631578947369e-06,
"loss": 0.3104,
"step": 2600
},
{
"epoch": 0.26,
"learning_rate": 7.766315789473685e-06,
"loss": 0.2972,
"step": 2625
},
{
"epoch": 0.27,
"learning_rate": 7.74e-06,
"loss": 0.3223,
"step": 2650
},
{
"epoch": 0.27,
"learning_rate": 7.713684210526317e-06,
"loss": 0.3307,
"step": 2675
},
{
"epoch": 0.27,
"learning_rate": 7.687368421052632e-06,
"loss": 0.285,
"step": 2700
},
{
"epoch": 0.27,
"learning_rate": 7.661052631578949e-06,
"loss": 0.3155,
"step": 2725
},
{
"epoch": 0.28,
"learning_rate": 7.634736842105264e-06,
"loss": 0.299,
"step": 2750
},
{
"epoch": 0.28,
"learning_rate": 7.6084210526315795e-06,
"loss": 0.3102,
"step": 2775
},
{
"epoch": 0.28,
"learning_rate": 7.582105263157895e-06,
"loss": 0.3608,
"step": 2800
},
{
"epoch": 0.28,
"learning_rate": 7.555789473684211e-06,
"loss": 0.3169,
"step": 2825
},
{
"epoch": 0.28,
"learning_rate": 7.529473684210526e-06,
"loss": 0.3362,
"step": 2850
},
{
"epoch": 0.29,
"learning_rate": 7.503157894736842e-06,
"loss": 0.3038,
"step": 2875
},
{
"epoch": 0.29,
"learning_rate": 7.476842105263159e-06,
"loss": 0.271,
"step": 2900
},
{
"epoch": 0.29,
"learning_rate": 7.4505263157894744e-06,
"loss": 0.3373,
"step": 2925
},
{
"epoch": 0.29,
"learning_rate": 7.42421052631579e-06,
"loss": 0.2813,
"step": 2950
},
{
"epoch": 0.3,
"learning_rate": 7.397894736842106e-06,
"loss": 0.3078,
"step": 2975
},
{
"epoch": 0.3,
"learning_rate": 7.371578947368422e-06,
"loss": 0.359,
"step": 3000
},
{
"epoch": 0.3,
"eval_loss": 0.9208301305770874,
"eval_runtime": 2090.7548,
"eval_samples_per_second": 2.705,
"eval_steps_per_second": 0.338,
"eval_wer": 25.398664600321542,
"step": 3000
},
{
"epoch": 0.3,
"learning_rate": 7.345263157894738e-06,
"loss": 0.3106,
"step": 3025
},
{
"epoch": 0.3,
"learning_rate": 7.3189473684210535e-06,
"loss": 0.3049,
"step": 3050
},
{
"epoch": 0.31,
"learning_rate": 7.292631578947369e-06,
"loss": 0.2846,
"step": 3075
},
{
"epoch": 0.31,
"learning_rate": 7.266315789473684e-06,
"loss": 0.2966,
"step": 3100
},
{
"epoch": 0.31,
"learning_rate": 7.24e-06,
"loss": 0.3259,
"step": 3125
},
{
"epoch": 0.32,
"learning_rate": 7.213684210526316e-06,
"loss": 0.2896,
"step": 3150
},
{
"epoch": 0.32,
"learning_rate": 7.187368421052632e-06,
"loss": 0.2745,
"step": 3175
},
{
"epoch": 0.32,
"learning_rate": 7.161052631578948e-06,
"loss": 0.2603,
"step": 3200
},
{
"epoch": 0.32,
"learning_rate": 7.134736842105264e-06,
"loss": 0.2842,
"step": 3225
},
{
"epoch": 0.33,
"learning_rate": 7.10842105263158e-06,
"loss": 0.2787,
"step": 3250
},
{
"epoch": 0.33,
"learning_rate": 7.082105263157896e-06,
"loss": 0.3063,
"step": 3275
},
{
"epoch": 0.33,
"learning_rate": 7.055789473684212e-06,
"loss": 0.2811,
"step": 3300
},
{
"epoch": 0.33,
"learning_rate": 7.0294736842105275e-06,
"loss": 0.2832,
"step": 3325
},
{
"epoch": 0.34,
"learning_rate": 7.0031578947368425e-06,
"loss": 0.3053,
"step": 3350
},
{
"epoch": 0.34,
"learning_rate": 6.976842105263158e-06,
"loss": 0.293,
"step": 3375
},
{
"epoch": 0.34,
"learning_rate": 6.950526315789474e-06,
"loss": 0.302,
"step": 3400
},
{
"epoch": 0.34,
"learning_rate": 6.92421052631579e-06,
"loss": 0.3331,
"step": 3425
},
{
"epoch": 0.34,
"learning_rate": 6.897894736842106e-06,
"loss": 0.3388,
"step": 3450
},
{
"epoch": 0.35,
"learning_rate": 6.871578947368422e-06,
"loss": 0.2805,
"step": 3475
},
{
"epoch": 0.35,
"learning_rate": 6.845263157894737e-06,
"loss": 0.3347,
"step": 3500
},
{
"epoch": 0.35,
"learning_rate": 6.818947368421052e-06,
"loss": 0.3385,
"step": 3525
},
{
"epoch": 0.35,
"learning_rate": 6.792631578947368e-06,
"loss": 0.2826,
"step": 3550
},
{
"epoch": 0.36,
"learning_rate": 6.766315789473685e-06,
"loss": 0.3156,
"step": 3575
},
{
"epoch": 0.36,
"learning_rate": 6.740000000000001e-06,
"loss": 0.313,
"step": 3600
},
{
"epoch": 0.36,
"learning_rate": 6.7136842105263165e-06,
"loss": 0.3042,
"step": 3625
},
{
"epoch": 0.36,
"learning_rate": 6.687368421052632e-06,
"loss": 0.2986,
"step": 3650
},
{
"epoch": 0.37,
"learning_rate": 6.661052631578948e-06,
"loss": 0.2638,
"step": 3675
},
{
"epoch": 0.37,
"learning_rate": 6.634736842105264e-06,
"loss": 0.3109,
"step": 3700
},
{
"epoch": 0.37,
"learning_rate": 6.60842105263158e-06,
"loss": 0.3027,
"step": 3725
},
{
"epoch": 0.38,
"learning_rate": 6.582105263157896e-06,
"loss": 0.3539,
"step": 3750
},
{
"epoch": 0.38,
"learning_rate": 6.5557894736842106e-06,
"loss": 0.3286,
"step": 3775
},
{
"epoch": 0.38,
"learning_rate": 6.529473684210526e-06,
"loss": 0.32,
"step": 3800
},
{
"epoch": 0.38,
"learning_rate": 6.503157894736842e-06,
"loss": 0.2646,
"step": 3825
},
{
"epoch": 0.39,
"learning_rate": 6.476842105263158e-06,
"loss": 0.2704,
"step": 3850
},
{
"epoch": 0.39,
"learning_rate": 6.450526315789474e-06,
"loss": 0.2641,
"step": 3875
},
{
"epoch": 0.39,
"learning_rate": 6.42421052631579e-06,
"loss": 0.266,
"step": 3900
},
{
"epoch": 0.39,
"learning_rate": 6.397894736842106e-06,
"loss": 0.2955,
"step": 3925
},
{
"epoch": 0.4,
"learning_rate": 6.371578947368422e-06,
"loss": 0.2825,
"step": 3950
},
{
"epoch": 0.4,
"learning_rate": 6.345263157894738e-06,
"loss": 0.2949,
"step": 3975
},
{
"epoch": 0.4,
"learning_rate": 6.318947368421054e-06,
"loss": 0.273,
"step": 4000
},
{
"epoch": 0.4,
"eval_loss": 0.9590798616409302,
"eval_runtime": 2052.518,
"eval_samples_per_second": 2.755,
"eval_steps_per_second": 0.344,
"eval_wer": 24.387700418579726,
"step": 4000
},
{
"epoch": 0.4,
"learning_rate": 6.292631578947369e-06,
"loss": 0.2907,
"step": 4025
},
{
"epoch": 0.41,
"learning_rate": 6.266315789473685e-06,
"loss": 0.2763,
"step": 4050
},
{
"epoch": 0.41,
"learning_rate": 6.24e-06,
"loss": 0.3175,
"step": 4075
},
{
"epoch": 0.41,
"learning_rate": 6.213684210526316e-06,
"loss": 0.2665,
"step": 4100
},
{
"epoch": 0.41,
"learning_rate": 6.187368421052632e-06,
"loss": 0.2795,
"step": 4125
},
{
"epoch": 0.41,
"learning_rate": 6.161052631578948e-06,
"loss": 0.2602,
"step": 4150
},
{
"epoch": 0.42,
"learning_rate": 6.13578947368421e-06,
"loss": 0.2979,
"step": 4175
},
{
"epoch": 0.42,
"learning_rate": 6.109473684210527e-06,
"loss": 0.3092,
"step": 4200
},
{
"epoch": 0.42,
"learning_rate": 6.083157894736843e-06,
"loss": 0.2529,
"step": 4225
},
{
"epoch": 0.42,
"learning_rate": 6.056842105263159e-06,
"loss": 0.2596,
"step": 4250
},
{
"epoch": 0.43,
"learning_rate": 6.0305263157894745e-06,
"loss": 0.2945,
"step": 4275
},
{
"epoch": 0.43,
"learning_rate": 6.00421052631579e-06,
"loss": 0.2554,
"step": 4300
},
{
"epoch": 0.43,
"learning_rate": 5.977894736842106e-06,
"loss": 0.2743,
"step": 4325
},
{
"epoch": 0.43,
"learning_rate": 5.951578947368422e-06,
"loss": 0.2752,
"step": 4350
},
{
"epoch": 0.44,
"learning_rate": 5.925263157894737e-06,
"loss": 0.3109,
"step": 4375
},
{
"epoch": 0.44,
"learning_rate": 5.898947368421053e-06,
"loss": 0.2907,
"step": 4400
},
{
"epoch": 0.44,
"learning_rate": 5.8726315789473685e-06,
"loss": 0.3005,
"step": 4425
},
{
"epoch": 0.45,
"learning_rate": 5.846315789473684e-06,
"loss": 0.2794,
"step": 4450
},
{
"epoch": 0.45,
"learning_rate": 5.82e-06,
"loss": 0.2705,
"step": 4475
},
{
"epoch": 0.45,
"learning_rate": 5.793684210526316e-06,
"loss": 0.2672,
"step": 4500
},
{
"epoch": 0.45,
"learning_rate": 5.767368421052632e-06,
"loss": 0.256,
"step": 4525
},
{
"epoch": 0.46,
"learning_rate": 5.7410526315789485e-06,
"loss": 0.2613,
"step": 4550
},
{
"epoch": 0.46,
"learning_rate": 5.714736842105264e-06,
"loss": 0.3101,
"step": 4575
},
{
"epoch": 0.46,
"learning_rate": 5.68842105263158e-06,
"loss": 0.2407,
"step": 4600
},
{
"epoch": 0.46,
"learning_rate": 5.662105263157895e-06,
"loss": 0.3038,
"step": 4625
},
{
"epoch": 0.47,
"learning_rate": 5.635789473684211e-06,
"loss": 0.2651,
"step": 4650
},
{
"epoch": 0.47,
"learning_rate": 5.609473684210527e-06,
"loss": 0.2673,
"step": 4675
},
{
"epoch": 0.47,
"learning_rate": 5.5831578947368425e-06,
"loss": 0.2945,
"step": 4700
},
{
"epoch": 0.47,
"learning_rate": 5.556842105263158e-06,
"loss": 0.2931,
"step": 4725
},
{
"epoch": 0.47,
"learning_rate": 5.530526315789474e-06,
"loss": 0.2794,
"step": 4750
},
{
"epoch": 0.48,
"learning_rate": 5.50421052631579e-06,
"loss": 0.2774,
"step": 4775
},
{
"epoch": 0.48,
"learning_rate": 5.477894736842105e-06,
"loss": 0.2998,
"step": 4800
},
{
"epoch": 0.48,
"learning_rate": 5.451578947368421e-06,
"loss": 0.3083,
"step": 4825
},
{
"epoch": 0.48,
"learning_rate": 5.425263157894737e-06,
"loss": 0.2635,
"step": 4850
},
{
"epoch": 0.49,
"learning_rate": 5.398947368421053e-06,
"loss": 0.2944,
"step": 4875
},
{
"epoch": 0.49,
"learning_rate": 5.372631578947369e-06,
"loss": 0.3059,
"step": 4900
},
{
"epoch": 0.49,
"learning_rate": 5.346315789473685e-06,
"loss": 0.2948,
"step": 4925
},
{
"epoch": 0.49,
"learning_rate": 5.320000000000001e-06,
"loss": 0.2948,
"step": 4950
},
{
"epoch": 0.5,
"learning_rate": 5.2936842105263166e-06,
"loss": 0.2736,
"step": 4975
},
{
"epoch": 0.5,
"learning_rate": 5.267368421052632e-06,
"loss": 0.2326,
"step": 5000
},
{
"epoch": 0.5,
"eval_loss": 0.9206530451774597,
"eval_runtime": 2043.8418,
"eval_samples_per_second": 2.767,
"eval_steps_per_second": 0.346,
"eval_wer": 21.90518951957476,
"step": 5000
},
{
"epoch": 0.5,
"learning_rate": 5.241052631578948e-06,
"loss": 0.2992,
"step": 5025
},
{
"epoch": 0.51,
"learning_rate": 5.214736842105263e-06,
"loss": 0.253,
"step": 5050
},
{
"epoch": 0.51,
"learning_rate": 5.188421052631579e-06,
"loss": 0.3138,
"step": 5075
},
{
"epoch": 0.51,
"learning_rate": 5.162105263157895e-06,
"loss": 0.316,
"step": 5100
},
{
"epoch": 0.51,
"learning_rate": 5.135789473684211e-06,
"loss": 0.3327,
"step": 5125
},
{
"epoch": 0.52,
"learning_rate": 5.1094736842105264e-06,
"loss": 0.2795,
"step": 5150
},
{
"epoch": 0.52,
"learning_rate": 5.083157894736842e-06,
"loss": 0.251,
"step": 5175
},
{
"epoch": 0.52,
"learning_rate": 5.056842105263158e-06,
"loss": 0.3183,
"step": 5200
},
{
"epoch": 0.52,
"learning_rate": 5.030526315789475e-06,
"loss": 0.3054,
"step": 5225
},
{
"epoch": 0.53,
"learning_rate": 5.0042105263157906e-06,
"loss": 0.3024,
"step": 5250
},
{
"epoch": 0.53,
"learning_rate": 4.9778947368421055e-06,
"loss": 0.2721,
"step": 5275
},
{
"epoch": 0.53,
"learning_rate": 4.951578947368421e-06,
"loss": 0.344,
"step": 5300
},
{
"epoch": 0.53,
"learning_rate": 4.925263157894737e-06,
"loss": 0.3312,
"step": 5325
},
{
"epoch": 0.54,
"learning_rate": 4.898947368421053e-06,
"loss": 0.2819,
"step": 5350
},
{
"epoch": 0.54,
"learning_rate": 4.872631578947369e-06,
"loss": 0.2647,
"step": 5375
},
{
"epoch": 0.54,
"learning_rate": 4.846315789473685e-06,
"loss": 0.2615,
"step": 5400
},
{
"epoch": 0.54,
"learning_rate": 4.8200000000000004e-06,
"loss": 0.279,
"step": 5425
},
{
"epoch": 0.55,
"learning_rate": 4.793684210526316e-06,
"loss": 0.3251,
"step": 5450
},
{
"epoch": 0.55,
"learning_rate": 4.767368421052632e-06,
"loss": 0.2577,
"step": 5475
},
{
"epoch": 0.55,
"learning_rate": 4.741052631578948e-06,
"loss": 0.2741,
"step": 5500
},
{
"epoch": 0.55,
"learning_rate": 4.714736842105264e-06,
"loss": 0.2644,
"step": 5525
},
{
"epoch": 0.56,
"learning_rate": 4.6884210526315795e-06,
"loss": 0.2595,
"step": 5550
},
{
"epoch": 0.56,
"learning_rate": 4.662105263157895e-06,
"loss": 0.2391,
"step": 5575
},
{
"epoch": 0.56,
"learning_rate": 4.63578947368421e-06,
"loss": 0.3202,
"step": 5600
},
{
"epoch": 0.56,
"learning_rate": 4.609473684210526e-06,
"loss": 0.3272,
"step": 5625
},
{
"epoch": 1.0,
"learning_rate": 4.583157894736843e-06,
"loss": 0.2806,
"step": 5650
},
{
"epoch": 1.0,
"learning_rate": 4.556842105263159e-06,
"loss": 0.2698,
"step": 5675
},
{
"epoch": 1.01,
"learning_rate": 4.5305263157894744e-06,
"loss": 0.2504,
"step": 5700
},
{
"epoch": 1.01,
"learning_rate": 4.504210526315789e-06,
"loss": 0.2277,
"step": 5725
},
{
"epoch": 1.01,
"learning_rate": 4.477894736842105e-06,
"loss": 0.3136,
"step": 5750
},
{
"epoch": 1.01,
"learning_rate": 4.451578947368421e-06,
"loss": 0.2795,
"step": 5775
},
{
"epoch": 1.02,
"learning_rate": 4.425263157894737e-06,
"loss": 0.2631,
"step": 5800
},
{
"epoch": 1.02,
"learning_rate": 4.3989473684210535e-06,
"loss": 0.2879,
"step": 5825
},
{
"epoch": 1.02,
"learning_rate": 4.3726315789473685e-06,
"loss": 0.2986,
"step": 5850
},
{
"epoch": 1.02,
"learning_rate": 4.346315789473684e-06,
"loss": 0.2783,
"step": 5875
},
{
"epoch": 1.03,
"learning_rate": 4.32e-06,
"loss": 0.2782,
"step": 5900
},
{
"epoch": 1.03,
"learning_rate": 4.293684210526316e-06,
"loss": 0.2824,
"step": 5925
},
{
"epoch": 1.03,
"learning_rate": 4.267368421052632e-06,
"loss": 0.2649,
"step": 5950
},
{
"epoch": 1.03,
"learning_rate": 4.241052631578948e-06,
"loss": 0.2562,
"step": 5975
},
{
"epoch": 1.04,
"learning_rate": 4.214736842105263e-06,
"loss": 0.2992,
"step": 6000
},
{
"epoch": 1.04,
"eval_loss": 0.944497287273407,
"eval_runtime": 2010.034,
"eval_samples_per_second": 2.813,
"eval_steps_per_second": 0.352,
"eval_wer": 22.45557116579523,
"step": 6000
},
{
"epoch": 1.04,
"learning_rate": 4.188421052631579e-06,
"loss": 0.2545,
"step": 6025
},
{
"epoch": 1.04,
"learning_rate": 4.162105263157895e-06,
"loss": 0.2461,
"step": 6050
},
{
"epoch": 1.04,
"learning_rate": 4.135789473684211e-06,
"loss": 0.2751,
"step": 6075
},
{
"epoch": 1.05,
"learning_rate": 4.109473684210527e-06,
"loss": 0.2377,
"step": 6100
},
{
"epoch": 1.05,
"learning_rate": 4.0831578947368425e-06,
"loss": 0.2351,
"step": 6125
},
{
"epoch": 1.05,
"learning_rate": 4.056842105263158e-06,
"loss": 0.2368,
"step": 6150
},
{
"epoch": 1.05,
"learning_rate": 4.030526315789474e-06,
"loss": 0.2677,
"step": 6175
},
{
"epoch": 1.06,
"learning_rate": 4.00421052631579e-06,
"loss": 0.2332,
"step": 6200
},
{
"epoch": 1.06,
"learning_rate": 3.977894736842106e-06,
"loss": 0.2833,
"step": 6225
},
{
"epoch": 1.06,
"learning_rate": 3.951578947368422e-06,
"loss": 0.2811,
"step": 6250
},
{
"epoch": 1.06,
"learning_rate": 3.9252631578947366e-06,
"loss": 0.2553,
"step": 6275
},
{
"epoch": 1.07,
"learning_rate": 3.898947368421052e-06,
"loss": 0.2434,
"step": 6300
},
{
"epoch": 1.07,
"learning_rate": 3.872631578947369e-06,
"loss": 0.228,
"step": 6325
},
{
"epoch": 1.07,
"learning_rate": 3.846315789473685e-06,
"loss": 0.2552,
"step": 6350
},
{
"epoch": 1.07,
"learning_rate": 3.820000000000001e-06,
"loss": 0.2407,
"step": 6375
},
{
"epoch": 1.08,
"learning_rate": 3.793684210526316e-06,
"loss": 0.2587,
"step": 6400
},
{
"epoch": 1.08,
"learning_rate": 3.767368421052632e-06,
"loss": 0.2516,
"step": 6425
},
{
"epoch": 1.08,
"learning_rate": 3.7410526315789473e-06,
"loss": 0.2363,
"step": 6450
},
{
"epoch": 1.08,
"learning_rate": 3.714736842105263e-06,
"loss": 0.2176,
"step": 6475
},
{
"epoch": 1.09,
"learning_rate": 3.6884210526315794e-06,
"loss": 0.2694,
"step": 6500
},
{
"epoch": 1.09,
"learning_rate": 3.662105263157895e-06,
"loss": 0.2671,
"step": 6525
},
{
"epoch": 1.09,
"learning_rate": 3.635789473684211e-06,
"loss": 0.2987,
"step": 6550
},
{
"epoch": 1.09,
"learning_rate": 3.6094736842105264e-06,
"loss": 0.2549,
"step": 6575
},
{
"epoch": 1.1,
"learning_rate": 3.5831578947368422e-06,
"loss": 0.3003,
"step": 6600
},
{
"epoch": 1.1,
"learning_rate": 3.556842105263158e-06,
"loss": 0.2688,
"step": 6625
},
{
"epoch": 1.1,
"learning_rate": 3.5305263157894743e-06,
"loss": 0.2259,
"step": 6650
},
{
"epoch": 1.1,
"learning_rate": 3.50421052631579e-06,
"loss": 0.2361,
"step": 6675
},
{
"epoch": 1.11,
"learning_rate": 3.4778947368421055e-06,
"loss": 0.226,
"step": 6700
},
{
"epoch": 1.11,
"learning_rate": 3.4515789473684213e-06,
"loss": 0.2519,
"step": 6725
},
{
"epoch": 1.11,
"learning_rate": 3.425263157894737e-06,
"loss": 0.2198,
"step": 6750
},
{
"epoch": 1.11,
"learning_rate": 3.4000000000000005e-06,
"loss": 0.2704,
"step": 6775
},
{
"epoch": 1.12,
"learning_rate": 3.3736842105263163e-06,
"loss": 0.234,
"step": 6800
},
{
"epoch": 1.12,
"learning_rate": 3.347368421052632e-06,
"loss": 0.2106,
"step": 6825
},
{
"epoch": 1.12,
"learning_rate": 3.3210526315789475e-06,
"loss": 0.2367,
"step": 6850
},
{
"epoch": 1.12,
"learning_rate": 3.2947368421052633e-06,
"loss": 0.2525,
"step": 6875
},
{
"epoch": 1.13,
"learning_rate": 3.268421052631579e-06,
"loss": 0.2236,
"step": 6900
},
{
"epoch": 1.13,
"learning_rate": 3.2421052631578945e-06,
"loss": 0.2313,
"step": 6925
},
{
"epoch": 1.13,
"learning_rate": 3.215789473684211e-06,
"loss": 0.2222,
"step": 6950
},
{
"epoch": 1.14,
"learning_rate": 3.1894736842105266e-06,
"loss": 0.2302,
"step": 6975
},
{
"epoch": 1.14,
"learning_rate": 3.1631578947368424e-06,
"loss": 0.2265,
"step": 7000
},
{
"epoch": 1.14,
"eval_loss": 0.9660412669181824,
"eval_runtime": 1990.1001,
"eval_samples_per_second": 2.842,
"eval_steps_per_second": 0.355,
"eval_wer": 21.223005952812017,
"step": 7000
},
{
"epoch": 1.14,
"learning_rate": 3.1368421052631582e-06,
"loss": 0.2694,
"step": 7025
},
{
"epoch": 1.14,
"learning_rate": 3.1105263157894736e-06,
"loss": 0.2459,
"step": 7050
},
{
"epoch": 1.15,
"learning_rate": 3.0842105263157895e-06,
"loss": 0.246,
"step": 7075
},
{
"epoch": 1.15,
"learning_rate": 3.0578947368421053e-06,
"loss": 0.2275,
"step": 7100
},
{
"epoch": 1.15,
"learning_rate": 3.0315789473684215e-06,
"loss": 0.2191,
"step": 7125
},
{
"epoch": 1.15,
"learning_rate": 3.0052631578947373e-06,
"loss": 0.2222,
"step": 7150
},
{
"epoch": 1.16,
"learning_rate": 2.9789473684210527e-06,
"loss": 0.206,
"step": 7175
},
{
"epoch": 1.16,
"learning_rate": 2.9526315789473685e-06,
"loss": 0.2457,
"step": 7200
},
{
"epoch": 1.16,
"learning_rate": 2.9263157894736844e-06,
"loss": 0.2221,
"step": 7225
},
{
"epoch": 1.16,
"learning_rate": 2.9e-06,
"loss": 0.2246,
"step": 7250
},
{
"epoch": 1.17,
"learning_rate": 2.8736842105263164e-06,
"loss": 0.2307,
"step": 7275
},
{
"epoch": 1.17,
"learning_rate": 2.847368421052632e-06,
"loss": 0.2388,
"step": 7300
},
{
"epoch": 1.17,
"learning_rate": 2.8210526315789476e-06,
"loss": 0.2545,
"step": 7325
},
{
"epoch": 1.17,
"learning_rate": 2.7947368421052635e-06,
"loss": 0.2644,
"step": 7350
},
{
"epoch": 1.18,
"learning_rate": 2.7684210526315793e-06,
"loss": 0.2429,
"step": 7375
},
{
"epoch": 1.18,
"learning_rate": 2.7421052631578947e-06,
"loss": 0.2082,
"step": 7400
},
{
"epoch": 1.18,
"learning_rate": 2.7157894736842105e-06,
"loss": 0.2232,
"step": 7425
},
{
"epoch": 1.18,
"learning_rate": 2.6894736842105267e-06,
"loss": 0.24,
"step": 7450
},
{
"epoch": 1.19,
"learning_rate": 2.6631578947368426e-06,
"loss": 0.2488,
"step": 7475
},
{
"epoch": 1.19,
"learning_rate": 2.6368421052631584e-06,
"loss": 0.2522,
"step": 7500
},
{
"epoch": 1.19,
"learning_rate": 2.6105263157894738e-06,
"loss": 0.1958,
"step": 7525
},
{
"epoch": 1.19,
"learning_rate": 2.5842105263157896e-06,
"loss": 0.2651,
"step": 7550
},
{
"epoch": 1.2,
"learning_rate": 2.5578947368421054e-06,
"loss": 0.2305,
"step": 7575
},
{
"epoch": 1.2,
"learning_rate": 2.5315789473684212e-06,
"loss": 0.2472,
"step": 7600
},
{
"epoch": 1.2,
"learning_rate": 2.5052631578947375e-06,
"loss": 0.2132,
"step": 7625
},
{
"epoch": 1.2,
"learning_rate": 2.478947368421053e-06,
"loss": 0.2457,
"step": 7650
},
{
"epoch": 1.21,
"learning_rate": 2.4526315789473687e-06,
"loss": 0.2185,
"step": 7675
},
{
"epoch": 1.21,
"learning_rate": 2.4263157894736845e-06,
"loss": 0.2759,
"step": 7700
},
{
"epoch": 1.21,
"learning_rate": 2.4000000000000003e-06,
"loss": 0.2439,
"step": 7725
},
{
"epoch": 1.21,
"learning_rate": 2.373684210526316e-06,
"loss": 0.2381,
"step": 7750
},
{
"epoch": 1.22,
"learning_rate": 2.347368421052632e-06,
"loss": 0.2445,
"step": 7775
},
{
"epoch": 1.22,
"learning_rate": 2.3210526315789473e-06,
"loss": 0.2466,
"step": 7800
},
{
"epoch": 1.22,
"learning_rate": 2.294736842105263e-06,
"loss": 0.2205,
"step": 7825
},
{
"epoch": 1.22,
"learning_rate": 2.268421052631579e-06,
"loss": 0.2081,
"step": 7850
},
{
"epoch": 1.23,
"learning_rate": 2.242105263157895e-06,
"loss": 0.2358,
"step": 7875
},
{
"epoch": 1.23,
"learning_rate": 2.2157894736842106e-06,
"loss": 0.2322,
"step": 7900
},
{
"epoch": 1.23,
"learning_rate": 2.1894736842105264e-06,
"loss": 0.2478,
"step": 7925
},
{
"epoch": 1.23,
"learning_rate": 2.1631578947368423e-06,
"loss": 0.2256,
"step": 7950
},
{
"epoch": 1.23,
"learning_rate": 2.136842105263158e-06,
"loss": 0.2251,
"step": 7975
},
{
"epoch": 1.24,
"learning_rate": 2.110526315789474e-06,
"loss": 0.2059,
"step": 8000
},
{
"epoch": 1.24,
"eval_loss": 0.9784936308860779,
"eval_runtime": 1965.3123,
"eval_samples_per_second": 2.877,
"eval_steps_per_second": 0.36,
"eval_wer": 20.95505699346784,
"step": 8000
},
{
"epoch": 1.24,
"learning_rate": 2.0842105263157897e-06,
"loss": 0.205,
"step": 8025
},
{
"epoch": 1.24,
"learning_rate": 2.0578947368421055e-06,
"loss": 0.2555,
"step": 8050
},
{
"epoch": 1.25,
"learning_rate": 2.031578947368421e-06,
"loss": 0.2235,
"step": 8075
},
{
"epoch": 1.25,
"learning_rate": 2.005263157894737e-06,
"loss": 0.2046,
"step": 8100
},
{
"epoch": 1.25,
"learning_rate": 1.978947368421053e-06,
"loss": 0.2597,
"step": 8125
},
{
"epoch": 1.25,
"learning_rate": 1.9526315789473684e-06,
"loss": 0.2513,
"step": 8150
},
{
"epoch": 1.25,
"learning_rate": 1.9263157894736846e-06,
"loss": 0.2596,
"step": 8175
},
{
"epoch": 1.26,
"learning_rate": 1.9000000000000002e-06,
"loss": 0.2895,
"step": 8200
},
{
"epoch": 1.26,
"learning_rate": 1.8736842105263158e-06,
"loss": 0.2735,
"step": 8225
},
{
"epoch": 1.26,
"learning_rate": 1.8473684210526319e-06,
"loss": 0.231,
"step": 8250
},
{
"epoch": 1.27,
"learning_rate": 1.8210526315789475e-06,
"loss": 0.2129,
"step": 8275
},
{
"epoch": 1.27,
"learning_rate": 1.7947368421052633e-06,
"loss": 0.2343,
"step": 8300
},
{
"epoch": 1.27,
"learning_rate": 1.768421052631579e-06,
"loss": 0.2982,
"step": 8325
},
{
"epoch": 1.27,
"learning_rate": 1.742105263157895e-06,
"loss": 0.2518,
"step": 8350
},
{
"epoch": 1.27,
"learning_rate": 1.7157894736842107e-06,
"loss": 0.2576,
"step": 8375
},
{
"epoch": 1.28,
"learning_rate": 1.6894736842105264e-06,
"loss": 0.2473,
"step": 8400
},
{
"epoch": 1.28,
"learning_rate": 1.6631578947368424e-06,
"loss": 0.27,
"step": 8425
},
{
"epoch": 1.28,
"learning_rate": 1.636842105263158e-06,
"loss": 0.2305,
"step": 8450
},
{
"epoch": 1.28,
"learning_rate": 1.6105263157894738e-06,
"loss": 0.2605,
"step": 8475
},
{
"epoch": 1.29,
"learning_rate": 1.5842105263157894e-06,
"loss": 0.2228,
"step": 8500
},
{
"epoch": 1.29,
"learning_rate": 1.5578947368421054e-06,
"loss": 0.207,
"step": 8525
},
{
"epoch": 1.29,
"learning_rate": 1.5315789473684213e-06,
"loss": 0.2513,
"step": 8550
},
{
"epoch": 1.29,
"learning_rate": 1.5052631578947369e-06,
"loss": 0.2414,
"step": 8575
},
{
"epoch": 1.3,
"learning_rate": 1.478947368421053e-06,
"loss": 0.2334,
"step": 8600
},
{
"epoch": 1.3,
"learning_rate": 1.4526315789473685e-06,
"loss": 0.2348,
"step": 8625
},
{
"epoch": 1.3,
"learning_rate": 1.4263157894736843e-06,
"loss": 0.1972,
"step": 8650
},
{
"epoch": 1.3,
"learning_rate": 1.4000000000000001e-06,
"loss": 0.2396,
"step": 8675
},
{
"epoch": 1.31,
"learning_rate": 1.373684210526316e-06,
"loss": 0.21,
"step": 8700
},
{
"epoch": 1.31,
"learning_rate": 1.3473684210526316e-06,
"loss": 0.2737,
"step": 8725
},
{
"epoch": 1.31,
"learning_rate": 1.3210526315789474e-06,
"loss": 0.275,
"step": 8750
},
{
"epoch": 1.31,
"learning_rate": 1.2947368421052634e-06,
"loss": 0.246,
"step": 8775
},
{
"epoch": 1.32,
"learning_rate": 1.268421052631579e-06,
"loss": 0.2425,
"step": 8800
},
{
"epoch": 1.32,
"learning_rate": 1.2421052631578948e-06,
"loss": 0.2505,
"step": 8825
},
{
"epoch": 1.32,
"learning_rate": 1.2157894736842107e-06,
"loss": 0.2364,
"step": 8850
},
{
"epoch": 1.32,
"learning_rate": 1.1894736842105265e-06,
"loss": 0.2066,
"step": 8875
},
{
"epoch": 1.33,
"learning_rate": 1.163157894736842e-06,
"loss": 0.2162,
"step": 8900
},
{
"epoch": 1.33,
"learning_rate": 1.136842105263158e-06,
"loss": 0.2258,
"step": 8925
},
{
"epoch": 1.33,
"learning_rate": 1.110526315789474e-06,
"loss": 0.2341,
"step": 8950
},
{
"epoch": 1.33,
"learning_rate": 1.0842105263157895e-06,
"loss": 0.206,
"step": 8975
},
{
"epoch": 1.34,
"learning_rate": 1.0578947368421054e-06,
"loss": 0.2239,
"step": 9000
},
{
"epoch": 1.34,
"eval_loss": 0.9636589288711548,
"eval_runtime": 2026.1646,
"eval_samples_per_second": 2.791,
"eval_steps_per_second": 0.349,
"eval_wer": 21.629998696464522,
"step": 9000
},
{
"epoch": 1.34,
"learning_rate": 1.0315789473684212e-06,
"loss": 0.2594,
"step": 9025
},
{
"epoch": 1.34,
"learning_rate": 1.005263157894737e-06,
"loss": 0.2278,
"step": 9050
},
{
"epoch": 1.34,
"learning_rate": 9.789473684210526e-07,
"loss": 0.2382,
"step": 9075
},
{
"epoch": 1.35,
"learning_rate": 9.526315789473685e-07,
"loss": 0.2278,
"step": 9100
},
{
"epoch": 1.35,
"learning_rate": 9.263157894736844e-07,
"loss": 0.242,
"step": 9125
},
{
"epoch": 1.35,
"learning_rate": 9.000000000000001e-07,
"loss": 0.2362,
"step": 9150
},
{
"epoch": 1.35,
"learning_rate": 8.736842105263159e-07,
"loss": 0.2435,
"step": 9175
},
{
"epoch": 1.36,
"learning_rate": 8.473684210526316e-07,
"loss": 0.2022,
"step": 9200
},
{
"epoch": 1.36,
"learning_rate": 8.210526315789474e-07,
"loss": 0.219,
"step": 9225
},
{
"epoch": 1.36,
"learning_rate": 7.947368421052632e-07,
"loss": 0.2466,
"step": 9250
},
{
"epoch": 1.36,
"learning_rate": 7.684210526315789e-07,
"loss": 0.2219,
"step": 9275
},
{
"epoch": 1.37,
"learning_rate": 7.421052631578949e-07,
"loss": 0.2385,
"step": 9300
},
{
"epoch": 1.37,
"learning_rate": 7.16842105263158e-07,
"loss": 0.2361,
"step": 9325
},
{
"epoch": 1.37,
"learning_rate": 6.905263157894737e-07,
"loss": 0.2095,
"step": 9350
},
{
"epoch": 1.38,
"learning_rate": 6.642105263157895e-07,
"loss": 0.2209,
"step": 9375
},
{
"epoch": 1.38,
"learning_rate": 6.378947368421053e-07,
"loss": 0.2193,
"step": 9400
},
{
"epoch": 1.38,
"learning_rate": 6.115789473684211e-07,
"loss": 0.2541,
"step": 9425
},
{
"epoch": 1.38,
"learning_rate": 5.852631578947369e-07,
"loss": 0.2273,
"step": 9450
},
{
"epoch": 1.39,
"learning_rate": 5.589473684210526e-07,
"loss": 0.2083,
"step": 9475
},
{
"epoch": 1.39,
"learning_rate": 5.326315789473684e-07,
"loss": 0.2171,
"step": 9500
},
{
"epoch": 1.39,
"learning_rate": 5.063157894736842e-07,
"loss": 0.2107,
"step": 9525
},
{
"epoch": 1.39,
"learning_rate": 4.800000000000001e-07,
"loss": 0.2662,
"step": 9550
},
{
"epoch": 1.4,
"learning_rate": 4.5368421052631583e-07,
"loss": 0.2258,
"step": 9575
},
{
"epoch": 1.4,
"learning_rate": 4.273684210526316e-07,
"loss": 0.2248,
"step": 9600
},
{
"epoch": 1.4,
"learning_rate": 4.0105263157894736e-07,
"loss": 0.2816,
"step": 9625
},
{
"epoch": 1.4,
"learning_rate": 3.7473684210526323e-07,
"loss": 0.2457,
"step": 9650
},
{
"epoch": 1.41,
"learning_rate": 3.48421052631579e-07,
"loss": 0.2405,
"step": 9675
},
{
"epoch": 1.41,
"learning_rate": 3.2210526315789476e-07,
"loss": 0.2344,
"step": 9700
},
{
"epoch": 1.41,
"learning_rate": 2.9578947368421053e-07,
"loss": 0.222,
"step": 9725
},
{
"epoch": 1.41,
"learning_rate": 2.6947368421052635e-07,
"loss": 0.271,
"step": 9750
},
{
"epoch": 1.42,
"learning_rate": 2.431578947368421e-07,
"loss": 0.227,
"step": 9775
},
{
"epoch": 1.42,
"learning_rate": 2.168421052631579e-07,
"loss": 0.2296,
"step": 9800
},
{
"epoch": 1.42,
"learning_rate": 1.9052631578947372e-07,
"loss": 0.2319,
"step": 9825
},
{
"epoch": 1.42,
"learning_rate": 1.642105263157895e-07,
"loss": 0.2075,
"step": 9850
},
{
"epoch": 1.43,
"learning_rate": 1.3789473684210528e-07,
"loss": 0.2759,
"step": 9875
},
{
"epoch": 1.43,
"learning_rate": 1.1157894736842106e-07,
"loss": 0.2432,
"step": 9900
},
{
"epoch": 1.43,
"learning_rate": 8.526315789473685e-08,
"loss": 0.3003,
"step": 9925
},
{
"epoch": 1.43,
"learning_rate": 5.8947368421052637e-08,
"loss": 0.2365,
"step": 9950
},
{
"epoch": 1.44,
"learning_rate": 3.263157894736842e-08,
"loss": 0.2082,
"step": 9975
},
{
"epoch": 1.44,
"learning_rate": 6.315789473684211e-09,
"loss": 0.2163,
"step": 10000
},
{
"epoch": 1.44,
"eval_loss": 0.9750341773033142,
"eval_runtime": 1994.645,
"eval_samples_per_second": 2.835,
"eval_steps_per_second": 0.354,
"eval_wer": 21.369291600886402,
"step": 10000
},
{
"epoch": 1.44,
"step": 10000,
"total_flos": 1.8469407903399936e+20,
"train_loss": 0.2953554097175598,
"train_runtime": 94208.4118,
"train_samples_per_second": 6.793,
"train_steps_per_second": 0.106
}
],
"max_steps": 10000,
"num_train_epochs": 9223372036854775807,
"total_flos": 1.8469407903399936e+20,
"trial_name": null,
"trial_params": null
}