swedish_asr_model_training / trainer_state.json
LudvigDoeser's picture
training after 11000 epochs (~13 epochs)
d812819
raw
history blame
51 kB
{
"best_metric": 19.932241671372104,
"best_model_checkpoint": "./swedish_asr_model_training/checkpoint-8000",
"epoch": 12.936610608020699,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 4.800000000000001e-07,
"loss": 0.1757,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 9.800000000000001e-07,
"loss": 0.1454,
"step": 50
},
{
"epoch": 0.1,
"learning_rate": 1.48e-06,
"loss": 0.1368,
"step": 75
},
{
"epoch": 0.13,
"learning_rate": 1.98e-06,
"loss": 0.1151,
"step": 100
},
{
"epoch": 0.16,
"learning_rate": 2.4800000000000004e-06,
"loss": 0.1015,
"step": 125
},
{
"epoch": 0.19,
"learning_rate": 2.9800000000000003e-06,
"loss": 0.0901,
"step": 150
},
{
"epoch": 0.23,
"learning_rate": 3.48e-06,
"loss": 0.0746,
"step": 175
},
{
"epoch": 0.26,
"learning_rate": 3.980000000000001e-06,
"loss": 0.0688,
"step": 200
},
{
"epoch": 0.29,
"learning_rate": 4.48e-06,
"loss": 0.0585,
"step": 225
},
{
"epoch": 0.32,
"learning_rate": 4.980000000000001e-06,
"loss": 0.0545,
"step": 250
},
{
"epoch": 0.36,
"learning_rate": 5.480000000000001e-06,
"loss": 0.0543,
"step": 275
},
{
"epoch": 0.39,
"learning_rate": 5.98e-06,
"loss": 0.0375,
"step": 300
},
{
"epoch": 0.42,
"learning_rate": 6.480000000000001e-06,
"loss": 0.0419,
"step": 325
},
{
"epoch": 0.45,
"learning_rate": 6.98e-06,
"loss": 0.0349,
"step": 350
},
{
"epoch": 0.49,
"learning_rate": 7.48e-06,
"loss": 0.0338,
"step": 375
},
{
"epoch": 0.52,
"learning_rate": 7.960000000000002e-06,
"loss": 0.0398,
"step": 400
},
{
"epoch": 0.55,
"learning_rate": 8.46e-06,
"loss": 0.033,
"step": 425
},
{
"epoch": 0.58,
"learning_rate": 8.96e-06,
"loss": 0.0317,
"step": 450
},
{
"epoch": 0.61,
"learning_rate": 9.460000000000001e-06,
"loss": 0.0254,
"step": 475
},
{
"epoch": 0.65,
"learning_rate": 9.960000000000001e-06,
"loss": 0.0308,
"step": 500
},
{
"epoch": 0.68,
"learning_rate": 9.975789473684211e-06,
"loss": 0.0306,
"step": 525
},
{
"epoch": 0.71,
"learning_rate": 9.949473684210526e-06,
"loss": 0.0289,
"step": 550
},
{
"epoch": 0.74,
"learning_rate": 9.923157894736844e-06,
"loss": 0.0294,
"step": 575
},
{
"epoch": 0.78,
"learning_rate": 9.89684210526316e-06,
"loss": 0.0325,
"step": 600
},
{
"epoch": 0.81,
"learning_rate": 9.870526315789474e-06,
"loss": 0.0375,
"step": 625
},
{
"epoch": 0.84,
"learning_rate": 9.84421052631579e-06,
"loss": 0.0421,
"step": 650
},
{
"epoch": 0.87,
"learning_rate": 9.817894736842106e-06,
"loss": 0.0366,
"step": 675
},
{
"epoch": 0.91,
"learning_rate": 9.791578947368422e-06,
"loss": 0.0419,
"step": 700
},
{
"epoch": 0.94,
"learning_rate": 9.765263157894737e-06,
"loss": 0.0476,
"step": 725
},
{
"epoch": 0.97,
"learning_rate": 9.738947368421054e-06,
"loss": 0.0478,
"step": 750
},
{
"epoch": 1.0,
"learning_rate": 9.712631578947369e-06,
"loss": 0.0489,
"step": 775
},
{
"epoch": 1.03,
"learning_rate": 9.686315789473684e-06,
"loss": 0.0171,
"step": 800
},
{
"epoch": 1.07,
"learning_rate": 9.66e-06,
"loss": 0.0171,
"step": 825
},
{
"epoch": 1.1,
"learning_rate": 9.633684210526316e-06,
"loss": 0.0152,
"step": 850
},
{
"epoch": 1.13,
"learning_rate": 9.607368421052632e-06,
"loss": 0.0187,
"step": 875
},
{
"epoch": 1.16,
"learning_rate": 9.581052631578947e-06,
"loss": 0.0211,
"step": 900
},
{
"epoch": 1.2,
"learning_rate": 9.554736842105264e-06,
"loss": 0.0213,
"step": 925
},
{
"epoch": 1.23,
"learning_rate": 9.52842105263158e-06,
"loss": 0.0297,
"step": 950
},
{
"epoch": 1.26,
"learning_rate": 9.502105263157896e-06,
"loss": 0.028,
"step": 975
},
{
"epoch": 1.29,
"learning_rate": 9.475789473684212e-06,
"loss": 0.0327,
"step": 1000
},
{
"epoch": 1.29,
"eval_loss": 0.33639949560165405,
"eval_runtime": 980.265,
"eval_samples_per_second": 5.171,
"eval_steps_per_second": 0.647,
"eval_wer": 21.868193917883357,
"step": 1000
},
{
"epoch": 1.33,
"learning_rate": 9.449473684210527e-06,
"loss": 0.0536,
"step": 1025
},
{
"epoch": 1.36,
"learning_rate": 9.423157894736842e-06,
"loss": 0.0572,
"step": 1050
},
{
"epoch": 1.39,
"learning_rate": 9.396842105263159e-06,
"loss": 0.0554,
"step": 1075
},
{
"epoch": 1.42,
"learning_rate": 9.370526315789474e-06,
"loss": 0.0579,
"step": 1100
},
{
"epoch": 1.46,
"learning_rate": 9.34421052631579e-06,
"loss": 0.058,
"step": 1125
},
{
"epoch": 1.49,
"learning_rate": 9.317894736842105e-06,
"loss": 0.0628,
"step": 1150
},
{
"epoch": 1.52,
"learning_rate": 9.291578947368422e-06,
"loss": 0.0483,
"step": 1175
},
{
"epoch": 1.55,
"learning_rate": 9.265263157894737e-06,
"loss": 0.0665,
"step": 1200
},
{
"epoch": 1.58,
"learning_rate": 9.238947368421052e-06,
"loss": 0.0714,
"step": 1225
},
{
"epoch": 1.62,
"learning_rate": 9.21263157894737e-06,
"loss": 0.0536,
"step": 1250
},
{
"epoch": 1.65,
"learning_rate": 9.186315789473685e-06,
"loss": 0.0646,
"step": 1275
},
{
"epoch": 1.68,
"learning_rate": 9.16e-06,
"loss": 0.0568,
"step": 1300
},
{
"epoch": 1.71,
"learning_rate": 9.133684210526317e-06,
"loss": 0.0594,
"step": 1325
},
{
"epoch": 1.75,
"learning_rate": 9.107368421052632e-06,
"loss": 0.0591,
"step": 1350
},
{
"epoch": 1.78,
"learning_rate": 9.081052631578949e-06,
"loss": 0.0663,
"step": 1375
},
{
"epoch": 1.81,
"learning_rate": 9.054736842105264e-06,
"loss": 0.0526,
"step": 1400
},
{
"epoch": 1.84,
"learning_rate": 9.02842105263158e-06,
"loss": 0.0663,
"step": 1425
},
{
"epoch": 1.88,
"learning_rate": 9.002105263157895e-06,
"loss": 0.0619,
"step": 1450
},
{
"epoch": 1.91,
"learning_rate": 8.97578947368421e-06,
"loss": 0.0528,
"step": 1475
},
{
"epoch": 1.94,
"learning_rate": 8.949473684210527e-06,
"loss": 0.0633,
"step": 1500
},
{
"epoch": 1.97,
"learning_rate": 8.923157894736842e-06,
"loss": 0.0602,
"step": 1525
},
{
"epoch": 2.01,
"learning_rate": 8.896842105263159e-06,
"loss": 0.0512,
"step": 1550
},
{
"epoch": 2.04,
"learning_rate": 8.870526315789474e-06,
"loss": 0.0165,
"step": 1575
},
{
"epoch": 2.07,
"learning_rate": 8.84421052631579e-06,
"loss": 0.0195,
"step": 1600
},
{
"epoch": 2.1,
"learning_rate": 8.817894736842107e-06,
"loss": 0.0236,
"step": 1625
},
{
"epoch": 2.13,
"learning_rate": 8.791578947368422e-06,
"loss": 0.0179,
"step": 1650
},
{
"epoch": 2.17,
"learning_rate": 8.765263157894739e-06,
"loss": 0.0242,
"step": 1675
},
{
"epoch": 2.2,
"learning_rate": 8.738947368421053e-06,
"loss": 0.02,
"step": 1700
},
{
"epoch": 2.23,
"learning_rate": 8.712631578947368e-06,
"loss": 0.016,
"step": 1725
},
{
"epoch": 2.26,
"learning_rate": 8.686315789473685e-06,
"loss": 0.0209,
"step": 1750
},
{
"epoch": 2.3,
"learning_rate": 8.66e-06,
"loss": 0.0206,
"step": 1775
},
{
"epoch": 2.33,
"learning_rate": 8.633684210526317e-06,
"loss": 0.0247,
"step": 1800
},
{
"epoch": 2.36,
"learning_rate": 8.607368421052632e-06,
"loss": 0.0193,
"step": 1825
},
{
"epoch": 2.39,
"learning_rate": 8.581052631578948e-06,
"loss": 0.0196,
"step": 1850
},
{
"epoch": 2.43,
"learning_rate": 8.554736842105263e-06,
"loss": 0.0261,
"step": 1875
},
{
"epoch": 2.46,
"learning_rate": 8.528421052631578e-06,
"loss": 0.0233,
"step": 1900
},
{
"epoch": 2.49,
"learning_rate": 8.502105263157897e-06,
"loss": 0.026,
"step": 1925
},
{
"epoch": 2.52,
"learning_rate": 8.475789473684212e-06,
"loss": 0.0198,
"step": 1950
},
{
"epoch": 2.55,
"learning_rate": 8.449473684210527e-06,
"loss": 0.0259,
"step": 1975
},
{
"epoch": 2.59,
"learning_rate": 8.423157894736843e-06,
"loss": 0.0224,
"step": 2000
},
{
"epoch": 2.59,
"eval_loss": 0.3422081768512726,
"eval_runtime": 979.073,
"eval_samples_per_second": 5.177,
"eval_steps_per_second": 0.648,
"eval_wer": 21.107257132101854,
"step": 2000
},
{
"epoch": 2.62,
"learning_rate": 8.396842105263158e-06,
"loss": 0.0226,
"step": 2025
},
{
"epoch": 2.65,
"learning_rate": 8.370526315789475e-06,
"loss": 0.0223,
"step": 2050
},
{
"epoch": 2.68,
"learning_rate": 8.34421052631579e-06,
"loss": 0.0204,
"step": 2075
},
{
"epoch": 2.72,
"learning_rate": 8.317894736842107e-06,
"loss": 0.0218,
"step": 2100
},
{
"epoch": 2.75,
"learning_rate": 8.291578947368422e-06,
"loss": 0.0191,
"step": 2125
},
{
"epoch": 2.78,
"learning_rate": 8.265263157894737e-06,
"loss": 0.0216,
"step": 2150
},
{
"epoch": 2.81,
"learning_rate": 8.238947368421053e-06,
"loss": 0.0205,
"step": 2175
},
{
"epoch": 2.85,
"learning_rate": 8.212631578947368e-06,
"loss": 0.0227,
"step": 2200
},
{
"epoch": 2.88,
"learning_rate": 8.186315789473685e-06,
"loss": 0.0207,
"step": 2225
},
{
"epoch": 2.91,
"learning_rate": 8.16e-06,
"loss": 0.0247,
"step": 2250
},
{
"epoch": 2.94,
"learning_rate": 8.133684210526316e-06,
"loss": 0.0259,
"step": 2275
},
{
"epoch": 2.98,
"learning_rate": 8.107368421052633e-06,
"loss": 0.0216,
"step": 2300
},
{
"epoch": 3.01,
"learning_rate": 8.081052631578948e-06,
"loss": 0.0158,
"step": 2325
},
{
"epoch": 3.04,
"learning_rate": 8.054736842105265e-06,
"loss": 0.0115,
"step": 2350
},
{
"epoch": 3.07,
"learning_rate": 8.02842105263158e-06,
"loss": 0.0097,
"step": 2375
},
{
"epoch": 3.1,
"learning_rate": 8.002105263157895e-06,
"loss": 0.0107,
"step": 2400
},
{
"epoch": 3.14,
"learning_rate": 7.975789473684211e-06,
"loss": 0.0091,
"step": 2425
},
{
"epoch": 3.17,
"learning_rate": 7.949473684210526e-06,
"loss": 0.0089,
"step": 2450
},
{
"epoch": 3.2,
"learning_rate": 7.923157894736843e-06,
"loss": 0.0099,
"step": 2475
},
{
"epoch": 3.23,
"learning_rate": 7.896842105263158e-06,
"loss": 0.0099,
"step": 2500
},
{
"epoch": 3.27,
"learning_rate": 7.870526315789475e-06,
"loss": 0.0084,
"step": 2525
},
{
"epoch": 3.3,
"learning_rate": 7.84421052631579e-06,
"loss": 0.0132,
"step": 2550
},
{
"epoch": 3.33,
"learning_rate": 7.817894736842105e-06,
"loss": 0.0115,
"step": 2575
},
{
"epoch": 3.36,
"learning_rate": 7.791578947368423e-06,
"loss": 0.0092,
"step": 2600
},
{
"epoch": 3.4,
"learning_rate": 7.765263157894738e-06,
"loss": 0.0109,
"step": 2625
},
{
"epoch": 3.43,
"learning_rate": 7.738947368421053e-06,
"loss": 0.0158,
"step": 2650
},
{
"epoch": 3.46,
"learning_rate": 7.71263157894737e-06,
"loss": 0.0091,
"step": 2675
},
{
"epoch": 3.49,
"learning_rate": 7.686315789473685e-06,
"loss": 0.0086,
"step": 2700
},
{
"epoch": 3.53,
"learning_rate": 7.660000000000001e-06,
"loss": 0.0091,
"step": 2725
},
{
"epoch": 3.56,
"learning_rate": 7.633684210526316e-06,
"loss": 0.0126,
"step": 2750
},
{
"epoch": 3.59,
"learning_rate": 7.607368421052632e-06,
"loss": 0.0108,
"step": 2775
},
{
"epoch": 3.62,
"learning_rate": 7.581052631578948e-06,
"loss": 0.0081,
"step": 2800
},
{
"epoch": 3.65,
"learning_rate": 7.554736842105264e-06,
"loss": 0.01,
"step": 2825
},
{
"epoch": 3.69,
"learning_rate": 7.5284210526315794e-06,
"loss": 0.009,
"step": 2850
},
{
"epoch": 3.72,
"learning_rate": 7.502105263157895e-06,
"loss": 0.0071,
"step": 2875
},
{
"epoch": 3.75,
"learning_rate": 7.475789473684211e-06,
"loss": 0.0102,
"step": 2900
},
{
"epoch": 3.78,
"learning_rate": 7.449473684210526e-06,
"loss": 0.0094,
"step": 2925
},
{
"epoch": 3.82,
"learning_rate": 7.4231578947368436e-06,
"loss": 0.0127,
"step": 2950
},
{
"epoch": 3.85,
"learning_rate": 7.3968421052631585e-06,
"loss": 0.0126,
"step": 2975
},
{
"epoch": 3.88,
"learning_rate": 7.370526315789474e-06,
"loss": 0.0101,
"step": 3000
},
{
"epoch": 3.88,
"eval_loss": 0.3693341314792633,
"eval_runtime": 979.4692,
"eval_samples_per_second": 5.175,
"eval_steps_per_second": 0.647,
"eval_wer": 21.228254147508807,
"step": 3000
},
{
"epoch": 3.91,
"learning_rate": 7.34421052631579e-06,
"loss": 0.0122,
"step": 3025
},
{
"epoch": 3.95,
"learning_rate": 7.317894736842106e-06,
"loss": 0.0113,
"step": 3050
},
{
"epoch": 3.98,
"learning_rate": 7.291578947368422e-06,
"loss": 0.01,
"step": 3075
},
{
"epoch": 4.01,
"learning_rate": 7.265263157894738e-06,
"loss": 0.0094,
"step": 3100
},
{
"epoch": 4.04,
"learning_rate": 7.2389473684210534e-06,
"loss": 0.0051,
"step": 3125
},
{
"epoch": 4.08,
"learning_rate": 7.212631578947369e-06,
"loss": 0.0044,
"step": 3150
},
{
"epoch": 4.11,
"learning_rate": 7.186315789473684e-06,
"loss": 0.0041,
"step": 3175
},
{
"epoch": 4.14,
"learning_rate": 7.16e-06,
"loss": 0.0054,
"step": 3200
},
{
"epoch": 4.17,
"learning_rate": 7.133684210526316e-06,
"loss": 0.0055,
"step": 3225
},
{
"epoch": 4.2,
"learning_rate": 7.107368421052632e-06,
"loss": 0.0067,
"step": 3250
},
{
"epoch": 4.24,
"learning_rate": 7.0810526315789475e-06,
"loss": 0.0062,
"step": 3275
},
{
"epoch": 4.27,
"learning_rate": 7.054736842105264e-06,
"loss": 0.0069,
"step": 3300
},
{
"epoch": 4.3,
"learning_rate": 7.02842105263158e-06,
"loss": 0.0054,
"step": 3325
},
{
"epoch": 4.33,
"learning_rate": 7.002105263157896e-06,
"loss": 0.0055,
"step": 3350
},
{
"epoch": 4.37,
"learning_rate": 6.975789473684212e-06,
"loss": 0.0059,
"step": 3375
},
{
"epoch": 4.4,
"learning_rate": 6.9494736842105275e-06,
"loss": 0.0073,
"step": 3400
},
{
"epoch": 4.43,
"learning_rate": 6.9231578947368424e-06,
"loss": 0.0039,
"step": 3425
},
{
"epoch": 4.46,
"learning_rate": 6.896842105263158e-06,
"loss": 0.0038,
"step": 3450
},
{
"epoch": 4.5,
"learning_rate": 6.870526315789474e-06,
"loss": 0.0056,
"step": 3475
},
{
"epoch": 4.53,
"learning_rate": 6.84421052631579e-06,
"loss": 0.0043,
"step": 3500
},
{
"epoch": 4.56,
"learning_rate": 6.817894736842106e-06,
"loss": 0.0047,
"step": 3525
},
{
"epoch": 4.59,
"learning_rate": 6.7915789473684215e-06,
"loss": 0.0069,
"step": 3550
},
{
"epoch": 4.62,
"learning_rate": 6.765263157894737e-06,
"loss": 0.0065,
"step": 3575
},
{
"epoch": 4.66,
"learning_rate": 6.738947368421052e-06,
"loss": 0.0053,
"step": 3600
},
{
"epoch": 4.69,
"learning_rate": 6.71263157894737e-06,
"loss": 0.0051,
"step": 3625
},
{
"epoch": 4.72,
"learning_rate": 6.686315789473685e-06,
"loss": 0.0039,
"step": 3650
},
{
"epoch": 4.75,
"learning_rate": 6.660000000000001e-06,
"loss": 0.0073,
"step": 3675
},
{
"epoch": 4.79,
"learning_rate": 6.6336842105263164e-06,
"loss": 0.0047,
"step": 3700
},
{
"epoch": 4.82,
"learning_rate": 6.607368421052632e-06,
"loss": 0.005,
"step": 3725
},
{
"epoch": 4.85,
"learning_rate": 6.581052631578948e-06,
"loss": 0.0066,
"step": 3750
},
{
"epoch": 4.88,
"learning_rate": 6.554736842105264e-06,
"loss": 0.0085,
"step": 3775
},
{
"epoch": 4.92,
"learning_rate": 6.52842105263158e-06,
"loss": 0.0051,
"step": 3800
},
{
"epoch": 4.95,
"learning_rate": 6.5021052631578955e-06,
"loss": 0.0045,
"step": 3825
},
{
"epoch": 4.98,
"learning_rate": 6.4757894736842105e-06,
"loss": 0.0056,
"step": 3850
},
{
"epoch": 5.01,
"learning_rate": 6.449473684210526e-06,
"loss": 0.0046,
"step": 3875
},
{
"epoch": 5.05,
"learning_rate": 6.423157894736842e-06,
"loss": 0.003,
"step": 3900
},
{
"epoch": 5.08,
"learning_rate": 6.396842105263158e-06,
"loss": 0.0028,
"step": 3925
},
{
"epoch": 5.11,
"learning_rate": 6.370526315789474e-06,
"loss": 0.0025,
"step": 3950
},
{
"epoch": 5.14,
"learning_rate": 6.3442105263157904e-06,
"loss": 0.0044,
"step": 3975
},
{
"epoch": 5.17,
"learning_rate": 6.317894736842106e-06,
"loss": 0.0023,
"step": 4000
},
{
"epoch": 5.17,
"eval_loss": 0.3898623287677765,
"eval_runtime": 983.4324,
"eval_samples_per_second": 5.154,
"eval_steps_per_second": 0.645,
"eval_wer": 20.486139119679493,
"step": 4000
},
{
"epoch": 5.21,
"learning_rate": 6.291578947368422e-06,
"loss": 0.0019,
"step": 4025
},
{
"epoch": 5.24,
"learning_rate": 6.265263157894738e-06,
"loss": 0.0032,
"step": 4050
},
{
"epoch": 5.27,
"learning_rate": 6.238947368421054e-06,
"loss": 0.0025,
"step": 4075
},
{
"epoch": 5.3,
"learning_rate": 6.212631578947369e-06,
"loss": 0.0031,
"step": 4100
},
{
"epoch": 5.34,
"learning_rate": 6.1863157894736845e-06,
"loss": 0.0039,
"step": 4125
},
{
"epoch": 5.37,
"learning_rate": 6.16e-06,
"loss": 0.0023,
"step": 4150
},
{
"epoch": 5.4,
"learning_rate": 6.133684210526316e-06,
"loss": 0.0022,
"step": 4175
},
{
"epoch": 5.43,
"learning_rate": 6.107368421052632e-06,
"loss": 0.0036,
"step": 4200
},
{
"epoch": 5.47,
"learning_rate": 6.081052631578948e-06,
"loss": 0.0037,
"step": 4225
},
{
"epoch": 5.5,
"learning_rate": 6.054736842105264e-06,
"loss": 0.0026,
"step": 4250
},
{
"epoch": 5.53,
"learning_rate": 6.0284210526315786e-06,
"loss": 0.0039,
"step": 4275
},
{
"epoch": 5.56,
"learning_rate": 6.002105263157896e-06,
"loss": 0.0042,
"step": 4300
},
{
"epoch": 5.6,
"learning_rate": 5.975789473684212e-06,
"loss": 0.0029,
"step": 4325
},
{
"epoch": 5.63,
"learning_rate": 5.949473684210527e-06,
"loss": 0.0032,
"step": 4350
},
{
"epoch": 5.66,
"learning_rate": 5.923157894736843e-06,
"loss": 0.0026,
"step": 4375
},
{
"epoch": 5.69,
"learning_rate": 5.8968421052631585e-06,
"loss": 0.0017,
"step": 4400
},
{
"epoch": 5.72,
"learning_rate": 5.870526315789474e-06,
"loss": 0.0026,
"step": 4425
},
{
"epoch": 5.76,
"learning_rate": 5.84421052631579e-06,
"loss": 0.0022,
"step": 4450
},
{
"epoch": 5.79,
"learning_rate": 5.817894736842106e-06,
"loss": 0.0022,
"step": 4475
},
{
"epoch": 5.82,
"learning_rate": 5.791578947368422e-06,
"loss": 0.0018,
"step": 4500
},
{
"epoch": 5.85,
"learning_rate": 5.765263157894737e-06,
"loss": 0.0026,
"step": 4525
},
{
"epoch": 5.89,
"learning_rate": 5.7389473684210526e-06,
"loss": 0.0033,
"step": 4550
},
{
"epoch": 5.92,
"learning_rate": 5.712631578947368e-06,
"loss": 0.0019,
"step": 4575
},
{
"epoch": 5.95,
"learning_rate": 5.686315789473684e-06,
"loss": 0.0027,
"step": 4600
},
{
"epoch": 5.98,
"learning_rate": 5.66e-06,
"loss": 0.002,
"step": 4625
},
{
"epoch": 6.02,
"learning_rate": 5.633684210526317e-06,
"loss": 0.0027,
"step": 4650
},
{
"epoch": 6.05,
"learning_rate": 5.6073684210526325e-06,
"loss": 0.0017,
"step": 4675
},
{
"epoch": 6.08,
"learning_rate": 5.581052631578948e-06,
"loss": 0.001,
"step": 4700
},
{
"epoch": 6.11,
"learning_rate": 5.554736842105264e-06,
"loss": 0.0045,
"step": 4725
},
{
"epoch": 6.14,
"learning_rate": 5.52842105263158e-06,
"loss": 0.0015,
"step": 4750
},
{
"epoch": 6.18,
"learning_rate": 5.502105263157895e-06,
"loss": 0.0015,
"step": 4775
},
{
"epoch": 6.21,
"learning_rate": 5.475789473684211e-06,
"loss": 0.0022,
"step": 4800
},
{
"epoch": 6.24,
"learning_rate": 5.4494736842105266e-06,
"loss": 0.0021,
"step": 4825
},
{
"epoch": 6.27,
"learning_rate": 5.423157894736842e-06,
"loss": 0.0038,
"step": 4850
},
{
"epoch": 6.31,
"learning_rate": 5.396842105263158e-06,
"loss": 0.0012,
"step": 4875
},
{
"epoch": 6.34,
"learning_rate": 5.370526315789474e-06,
"loss": 0.0017,
"step": 4900
},
{
"epoch": 6.37,
"learning_rate": 5.34421052631579e-06,
"loss": 0.002,
"step": 4925
},
{
"epoch": 6.4,
"learning_rate": 5.317894736842105e-06,
"loss": 0.0011,
"step": 4950
},
{
"epoch": 6.44,
"learning_rate": 5.291578947368422e-06,
"loss": 0.0009,
"step": 4975
},
{
"epoch": 6.47,
"learning_rate": 5.265263157894738e-06,
"loss": 0.0009,
"step": 5000
},
{
"epoch": 6.47,
"eval_loss": 0.40303418040275574,
"eval_runtime": 977.0375,
"eval_samples_per_second": 5.188,
"eval_steps_per_second": 0.649,
"eval_wer": 20.25758920168858,
"step": 5000
},
{
"epoch": 6.5,
"learning_rate": 5.238947368421053e-06,
"loss": 0.0015,
"step": 5025
},
{
"epoch": 6.53,
"learning_rate": 5.212631578947369e-06,
"loss": 0.001,
"step": 5050
},
{
"epoch": 6.57,
"learning_rate": 5.186315789473685e-06,
"loss": 0.0014,
"step": 5075
},
{
"epoch": 6.6,
"learning_rate": 5.1600000000000006e-06,
"loss": 0.0022,
"step": 5100
},
{
"epoch": 6.63,
"learning_rate": 5.133684210526316e-06,
"loss": 0.0009,
"step": 5125
},
{
"epoch": 6.66,
"learning_rate": 5.107368421052632e-06,
"loss": 0.0014,
"step": 5150
},
{
"epoch": 6.69,
"learning_rate": 5.081052631578948e-06,
"loss": 0.002,
"step": 5175
},
{
"epoch": 6.73,
"learning_rate": 5.054736842105263e-06,
"loss": 0.0016,
"step": 5200
},
{
"epoch": 6.76,
"learning_rate": 5.028421052631579e-06,
"loss": 0.0017,
"step": 5225
},
{
"epoch": 6.79,
"learning_rate": 5.002105263157895e-06,
"loss": 0.0015,
"step": 5250
},
{
"epoch": 6.82,
"learning_rate": 4.975789473684211e-06,
"loss": 0.0011,
"step": 5275
},
{
"epoch": 6.86,
"learning_rate": 4.949473684210527e-06,
"loss": 0.001,
"step": 5300
},
{
"epoch": 6.89,
"learning_rate": 4.923157894736842e-06,
"loss": 0.0009,
"step": 5325
},
{
"epoch": 6.92,
"learning_rate": 4.896842105263158e-06,
"loss": 0.0014,
"step": 5350
},
{
"epoch": 6.95,
"learning_rate": 4.870526315789474e-06,
"loss": 0.0033,
"step": 5375
},
{
"epoch": 6.99,
"learning_rate": 4.84421052631579e-06,
"loss": 0.001,
"step": 5400
},
{
"epoch": 7.02,
"learning_rate": 4.817894736842106e-06,
"loss": 0.0012,
"step": 5425
},
{
"epoch": 7.05,
"learning_rate": 4.791578947368421e-06,
"loss": 0.0009,
"step": 5450
},
{
"epoch": 7.08,
"learning_rate": 4.765263157894737e-06,
"loss": 0.0007,
"step": 5475
},
{
"epoch": 7.12,
"learning_rate": 4.738947368421053e-06,
"loss": 0.0007,
"step": 5500
},
{
"epoch": 7.15,
"learning_rate": 4.712631578947369e-06,
"loss": 0.0007,
"step": 5525
},
{
"epoch": 7.18,
"learning_rate": 4.6863157894736845e-06,
"loss": 0.0012,
"step": 5550
},
{
"epoch": 7.21,
"learning_rate": 4.66e-06,
"loss": 0.0009,
"step": 5575
},
{
"epoch": 7.24,
"learning_rate": 4.633684210526316e-06,
"loss": 0.0019,
"step": 5600
},
{
"epoch": 7.28,
"learning_rate": 4.607368421052632e-06,
"loss": 0.0016,
"step": 5625
},
{
"epoch": 7.31,
"learning_rate": 4.581052631578948e-06,
"loss": 0.0027,
"step": 5650
},
{
"epoch": 7.34,
"learning_rate": 4.5547368421052636e-06,
"loss": 0.0016,
"step": 5675
},
{
"epoch": 7.37,
"learning_rate": 4.528421052631579e-06,
"loss": 0.0008,
"step": 5700
},
{
"epoch": 7.41,
"learning_rate": 4.502105263157895e-06,
"loss": 0.0014,
"step": 5725
},
{
"epoch": 7.44,
"learning_rate": 4.475789473684211e-06,
"loss": 0.0011,
"step": 5750
},
{
"epoch": 7.47,
"learning_rate": 4.449473684210527e-06,
"loss": 0.0009,
"step": 5775
},
{
"epoch": 7.5,
"learning_rate": 4.423157894736843e-06,
"loss": 0.0008,
"step": 5800
},
{
"epoch": 7.54,
"learning_rate": 4.3968421052631585e-06,
"loss": 0.0008,
"step": 5825
},
{
"epoch": 7.57,
"learning_rate": 4.370526315789474e-06,
"loss": 0.0013,
"step": 5850
},
{
"epoch": 7.6,
"learning_rate": 4.344210526315789e-06,
"loss": 0.0006,
"step": 5875
},
{
"epoch": 7.63,
"learning_rate": 4.317894736842105e-06,
"loss": 0.0007,
"step": 5900
},
{
"epoch": 7.66,
"learning_rate": 4.291578947368422e-06,
"loss": 0.0009,
"step": 5925
},
{
"epoch": 7.7,
"learning_rate": 4.2652631578947376e-06,
"loss": 0.0006,
"step": 5950
},
{
"epoch": 7.73,
"learning_rate": 4.238947368421053e-06,
"loss": 0.0006,
"step": 5975
},
{
"epoch": 7.76,
"learning_rate": 4.212631578947368e-06,
"loss": 0.0006,
"step": 6000
},
{
"epoch": 7.76,
"eval_loss": 0.40835636854171753,
"eval_runtime": 978.8484,
"eval_samples_per_second": 5.179,
"eval_steps_per_second": 0.648,
"eval_wer": 20.117770428329436,
"step": 6000
},
{
"epoch": 7.79,
"learning_rate": 4.186315789473684e-06,
"loss": 0.0006,
"step": 6025
},
{
"epoch": 7.83,
"learning_rate": 4.16e-06,
"loss": 0.0006,
"step": 6050
},
{
"epoch": 7.86,
"learning_rate": 4.133684210526316e-06,
"loss": 0.0006,
"step": 6075
},
{
"epoch": 7.89,
"learning_rate": 4.1073684210526325e-06,
"loss": 0.0006,
"step": 6100
},
{
"epoch": 7.92,
"learning_rate": 4.0810526315789474e-06,
"loss": 0.0006,
"step": 6125
},
{
"epoch": 7.96,
"learning_rate": 4.054736842105263e-06,
"loss": 0.001,
"step": 6150
},
{
"epoch": 7.99,
"learning_rate": 4.028421052631579e-06,
"loss": 0.0011,
"step": 6175
},
{
"epoch": 8.02,
"learning_rate": 4.002105263157895e-06,
"loss": 0.0006,
"step": 6200
},
{
"epoch": 8.05,
"learning_rate": 3.975789473684211e-06,
"loss": 0.0006,
"step": 6225
},
{
"epoch": 8.09,
"learning_rate": 3.9494736842105265e-06,
"loss": 0.0014,
"step": 6250
},
{
"epoch": 8.12,
"learning_rate": 3.923157894736842e-06,
"loss": 0.0005,
"step": 6275
},
{
"epoch": 8.15,
"learning_rate": 3.896842105263158e-06,
"loss": 0.0006,
"step": 6300
},
{
"epoch": 8.18,
"learning_rate": 3.870526315789474e-06,
"loss": 0.0005,
"step": 6325
},
{
"epoch": 8.21,
"learning_rate": 3.84421052631579e-06,
"loss": 0.0005,
"step": 6350
},
{
"epoch": 8.25,
"learning_rate": 3.817894736842106e-06,
"loss": 0.0005,
"step": 6375
},
{
"epoch": 8.28,
"learning_rate": 3.791578947368421e-06,
"loss": 0.0006,
"step": 6400
},
{
"epoch": 8.31,
"learning_rate": 3.7652631578947373e-06,
"loss": 0.0006,
"step": 6425
},
{
"epoch": 8.34,
"learning_rate": 3.738947368421053e-06,
"loss": 0.0004,
"step": 6450
},
{
"epoch": 8.38,
"learning_rate": 3.712631578947369e-06,
"loss": 0.0005,
"step": 6475
},
{
"epoch": 8.41,
"learning_rate": 3.6863157894736847e-06,
"loss": 0.0004,
"step": 6500
},
{
"epoch": 8.44,
"learning_rate": 3.66e-06,
"loss": 0.0004,
"step": 6525
},
{
"epoch": 8.47,
"learning_rate": 3.633684210526316e-06,
"loss": 0.0004,
"step": 6550
},
{
"epoch": 8.51,
"learning_rate": 3.6073684210526317e-06,
"loss": 0.0006,
"step": 6575
},
{
"epoch": 8.54,
"learning_rate": 3.581052631578948e-06,
"loss": 0.0004,
"step": 6600
},
{
"epoch": 8.57,
"learning_rate": 3.554736842105264e-06,
"loss": 0.0005,
"step": 6625
},
{
"epoch": 8.6,
"learning_rate": 3.528421052631579e-06,
"loss": 0.0005,
"step": 6650
},
{
"epoch": 8.64,
"learning_rate": 3.502105263157895e-06,
"loss": 0.0005,
"step": 6675
},
{
"epoch": 8.67,
"learning_rate": 3.475789473684211e-06,
"loss": 0.0006,
"step": 6700
},
{
"epoch": 8.7,
"learning_rate": 3.4494736842105262e-06,
"loss": 0.0008,
"step": 6725
},
{
"epoch": 8.73,
"learning_rate": 3.423157894736842e-06,
"loss": 0.0005,
"step": 6750
},
{
"epoch": 8.76,
"learning_rate": 3.3968421052631583e-06,
"loss": 0.0005,
"step": 6775
},
{
"epoch": 8.8,
"learning_rate": 3.370526315789474e-06,
"loss": 0.0005,
"step": 6800
},
{
"epoch": 8.83,
"learning_rate": 3.34421052631579e-06,
"loss": 0.0006,
"step": 6825
},
{
"epoch": 8.86,
"learning_rate": 3.3178947368421053e-06,
"loss": 0.0005,
"step": 6850
},
{
"epoch": 8.89,
"learning_rate": 3.291578947368421e-06,
"loss": 0.0006,
"step": 6875
},
{
"epoch": 8.93,
"learning_rate": 3.265263157894737e-06,
"loss": 0.0006,
"step": 6900
},
{
"epoch": 8.96,
"learning_rate": 3.238947368421053e-06,
"loss": 0.0005,
"step": 6925
},
{
"epoch": 8.99,
"learning_rate": 3.212631578947369e-06,
"loss": 0.0005,
"step": 6950
},
{
"epoch": 9.02,
"learning_rate": 3.1863157894736844e-06,
"loss": 0.0004,
"step": 6975
},
{
"epoch": 9.06,
"learning_rate": 3.1600000000000002e-06,
"loss": 0.0004,
"step": 7000
},
{
"epoch": 9.06,
"eval_loss": 0.42490312457084656,
"eval_runtime": 978.1581,
"eval_samples_per_second": 5.182,
"eval_steps_per_second": 0.648,
"eval_wer": 20.16348041192762,
"step": 7000
},
{
"epoch": 9.09,
"learning_rate": 3.133684210526316e-06,
"loss": 0.0004,
"step": 7025
},
{
"epoch": 9.12,
"learning_rate": 3.107368421052632e-06,
"loss": 0.0003,
"step": 7050
},
{
"epoch": 9.15,
"learning_rate": 3.0810526315789473e-06,
"loss": 0.0004,
"step": 7075
},
{
"epoch": 9.18,
"learning_rate": 3.0547368421052635e-06,
"loss": 0.0004,
"step": 7100
},
{
"epoch": 9.22,
"learning_rate": 3.0284210526315793e-06,
"loss": 0.0003,
"step": 7125
},
{
"epoch": 9.25,
"learning_rate": 3.002105263157895e-06,
"loss": 0.0003,
"step": 7150
},
{
"epoch": 9.28,
"learning_rate": 2.975789473684211e-06,
"loss": 0.0003,
"step": 7175
},
{
"epoch": 9.31,
"learning_rate": 2.9494736842105264e-06,
"loss": 0.0003,
"step": 7200
},
{
"epoch": 9.35,
"learning_rate": 2.923157894736842e-06,
"loss": 0.0003,
"step": 7225
},
{
"epoch": 9.38,
"learning_rate": 2.896842105263158e-06,
"loss": 0.0003,
"step": 7250
},
{
"epoch": 9.41,
"learning_rate": 2.8705263157894742e-06,
"loss": 0.0004,
"step": 7275
},
{
"epoch": 9.44,
"learning_rate": 2.84421052631579e-06,
"loss": 0.0003,
"step": 7300
},
{
"epoch": 9.48,
"learning_rate": 2.8178947368421055e-06,
"loss": 0.0003,
"step": 7325
},
{
"epoch": 9.51,
"learning_rate": 2.7915789473684213e-06,
"loss": 0.0003,
"step": 7350
},
{
"epoch": 9.54,
"learning_rate": 2.765263157894737e-06,
"loss": 0.0003,
"step": 7375
},
{
"epoch": 9.57,
"learning_rate": 2.7389473684210525e-06,
"loss": 0.0003,
"step": 7400
},
{
"epoch": 9.61,
"learning_rate": 2.7126315789473683e-06,
"loss": 0.0003,
"step": 7425
},
{
"epoch": 9.64,
"learning_rate": 2.6863157894736845e-06,
"loss": 0.0003,
"step": 7450
},
{
"epoch": 9.67,
"learning_rate": 2.6600000000000004e-06,
"loss": 0.0003,
"step": 7475
},
{
"epoch": 9.7,
"learning_rate": 2.633684210526316e-06,
"loss": 0.0006,
"step": 7500
},
{
"epoch": 9.73,
"learning_rate": 2.6073684210526316e-06,
"loss": 0.001,
"step": 7525
},
{
"epoch": 9.77,
"learning_rate": 2.5810526315789474e-06,
"loss": 0.0004,
"step": 7550
},
{
"epoch": 9.8,
"learning_rate": 2.5547368421052632e-06,
"loss": 0.0004,
"step": 7575
},
{
"epoch": 9.83,
"learning_rate": 2.528421052631579e-06,
"loss": 0.0003,
"step": 7600
},
{
"epoch": 9.86,
"learning_rate": 2.5021052631578953e-06,
"loss": 0.0003,
"step": 7625
},
{
"epoch": 9.9,
"learning_rate": 2.4757894736842107e-06,
"loss": 0.0003,
"step": 7650
},
{
"epoch": 9.93,
"learning_rate": 2.4494736842105265e-06,
"loss": 0.0004,
"step": 7675
},
{
"epoch": 9.96,
"learning_rate": 2.4231578947368423e-06,
"loss": 0.0004,
"step": 7700
},
{
"epoch": 9.99,
"learning_rate": 2.396842105263158e-06,
"loss": 0.0004,
"step": 7725
},
{
"epoch": 10.03,
"learning_rate": 2.370526315789474e-06,
"loss": 0.0003,
"step": 7750
},
{
"epoch": 10.06,
"learning_rate": 2.3442105263157898e-06,
"loss": 0.0003,
"step": 7775
},
{
"epoch": 10.09,
"learning_rate": 2.317894736842105e-06,
"loss": 0.0003,
"step": 7800
},
{
"epoch": 10.12,
"learning_rate": 2.2915789473684214e-06,
"loss": 0.0003,
"step": 7825
},
{
"epoch": 10.16,
"learning_rate": 2.2652631578947372e-06,
"loss": 0.0003,
"step": 7850
},
{
"epoch": 10.19,
"learning_rate": 2.2389473684210526e-06,
"loss": 0.0003,
"step": 7875
},
{
"epoch": 10.22,
"learning_rate": 2.2126315789473684e-06,
"loss": 0.0003,
"step": 7900
},
{
"epoch": 10.25,
"learning_rate": 2.1863157894736843e-06,
"loss": 0.0003,
"step": 7925
},
{
"epoch": 10.28,
"learning_rate": 2.16e-06,
"loss": 0.0003,
"step": 7950
},
{
"epoch": 10.32,
"learning_rate": 2.133684210526316e-06,
"loss": 0.0003,
"step": 7975
},
{
"epoch": 10.35,
"learning_rate": 2.1073684210526317e-06,
"loss": 0.0003,
"step": 8000
},
{
"epoch": 10.35,
"eval_loss": 0.4340263605117798,
"eval_runtime": 978.9502,
"eval_samples_per_second": 5.178,
"eval_steps_per_second": 0.648,
"eval_wer": 19.932241671372104,
"step": 8000
},
{
"epoch": 10.38,
"learning_rate": 2.0810526315789475e-06,
"loss": 0.0003,
"step": 8025
},
{
"epoch": 10.41,
"learning_rate": 2.0547368421052633e-06,
"loss": 0.0004,
"step": 8050
},
{
"epoch": 10.45,
"learning_rate": 2.028421052631579e-06,
"loss": 0.0003,
"step": 8075
},
{
"epoch": 10.48,
"learning_rate": 2.002105263157895e-06,
"loss": 0.0003,
"step": 8100
},
{
"epoch": 10.51,
"learning_rate": 1.975789473684211e-06,
"loss": 0.0003,
"step": 8125
},
{
"epoch": 10.54,
"learning_rate": 1.949473684210526e-06,
"loss": 0.0003,
"step": 8150
},
{
"epoch": 10.58,
"learning_rate": 1.9231578947368424e-06,
"loss": 0.0003,
"step": 8175
},
{
"epoch": 10.61,
"learning_rate": 1.896842105263158e-06,
"loss": 0.0003,
"step": 8200
},
{
"epoch": 10.64,
"learning_rate": 1.8705263157894737e-06,
"loss": 0.0003,
"step": 8225
},
{
"epoch": 10.67,
"learning_rate": 1.8442105263157897e-06,
"loss": 0.0003,
"step": 8250
},
{
"epoch": 10.71,
"learning_rate": 1.8178947368421055e-06,
"loss": 0.0003,
"step": 8275
},
{
"epoch": 10.74,
"learning_rate": 1.7915789473684211e-06,
"loss": 0.0003,
"step": 8300
},
{
"epoch": 10.77,
"learning_rate": 1.7652631578947371e-06,
"loss": 0.0003,
"step": 8325
},
{
"epoch": 10.8,
"learning_rate": 1.7389473684210527e-06,
"loss": 0.0003,
"step": 8350
},
{
"epoch": 10.83,
"learning_rate": 1.7126315789473686e-06,
"loss": 0.0003,
"step": 8375
},
{
"epoch": 10.87,
"learning_rate": 1.6863157894736842e-06,
"loss": 0.0002,
"step": 8400
},
{
"epoch": 10.9,
"learning_rate": 1.6600000000000002e-06,
"loss": 0.0003,
"step": 8425
},
{
"epoch": 10.93,
"learning_rate": 1.633684210526316e-06,
"loss": 0.0003,
"step": 8450
},
{
"epoch": 10.96,
"learning_rate": 1.6073684210526316e-06,
"loss": 0.0003,
"step": 8475
},
{
"epoch": 11.0,
"learning_rate": 1.5810526315789477e-06,
"loss": 0.0003,
"step": 8500
},
{
"epoch": 11.03,
"learning_rate": 1.5547368421052633e-06,
"loss": 0.0003,
"step": 8525
},
{
"epoch": 11.06,
"learning_rate": 1.528421052631579e-06,
"loss": 0.0002,
"step": 8550
},
{
"epoch": 11.09,
"learning_rate": 1.5021052631578947e-06,
"loss": 0.0002,
"step": 8575
},
{
"epoch": 11.13,
"learning_rate": 1.4757894736842107e-06,
"loss": 0.0002,
"step": 8600
},
{
"epoch": 11.16,
"learning_rate": 1.4494736842105263e-06,
"loss": 0.0002,
"step": 8625
},
{
"epoch": 11.19,
"learning_rate": 1.4231578947368421e-06,
"loss": 0.0002,
"step": 8650
},
{
"epoch": 11.22,
"learning_rate": 1.3968421052631582e-06,
"loss": 0.0002,
"step": 8675
},
{
"epoch": 11.25,
"learning_rate": 1.3705263157894738e-06,
"loss": 0.0002,
"step": 8700
},
{
"epoch": 11.29,
"learning_rate": 1.3442105263157896e-06,
"loss": 0.0002,
"step": 8725
},
{
"epoch": 11.32,
"learning_rate": 1.3178947368421054e-06,
"loss": 0.0002,
"step": 8750
},
{
"epoch": 11.35,
"learning_rate": 1.2915789473684212e-06,
"loss": 0.0002,
"step": 8775
},
{
"epoch": 11.38,
"learning_rate": 1.2652631578947368e-06,
"loss": 0.0002,
"step": 8800
},
{
"epoch": 11.42,
"learning_rate": 1.2389473684210527e-06,
"loss": 0.0002,
"step": 8825
},
{
"epoch": 11.45,
"learning_rate": 1.2126315789473685e-06,
"loss": 0.0002,
"step": 8850
},
{
"epoch": 11.48,
"learning_rate": 1.1863157894736843e-06,
"loss": 0.0002,
"step": 8875
},
{
"epoch": 11.51,
"learning_rate": 1.1600000000000001e-06,
"loss": 0.0002,
"step": 8900
},
{
"epoch": 11.55,
"learning_rate": 1.133684210526316e-06,
"loss": 0.0002,
"step": 8925
},
{
"epoch": 11.58,
"learning_rate": 1.1073684210526318e-06,
"loss": 0.0002,
"step": 8950
},
{
"epoch": 11.61,
"learning_rate": 1.0810526315789474e-06,
"loss": 0.0002,
"step": 8975
},
{
"epoch": 11.64,
"learning_rate": 1.0547368421052632e-06,
"loss": 0.0002,
"step": 9000
},
{
"epoch": 11.64,
"eval_loss": 0.43969836831092834,
"eval_runtime": 976.3124,
"eval_samples_per_second": 5.192,
"eval_steps_per_second": 0.649,
"eval_wer": 19.97526283240569,
"step": 9000
},
{
"epoch": 11.68,
"learning_rate": 1.028421052631579e-06,
"loss": 0.0002,
"step": 9025
},
{
"epoch": 11.71,
"learning_rate": 1.0021052631578948e-06,
"loss": 0.0002,
"step": 9050
},
{
"epoch": 11.74,
"learning_rate": 9.757894736842106e-07,
"loss": 0.0002,
"step": 9075
},
{
"epoch": 11.77,
"learning_rate": 9.494736842105263e-07,
"loss": 0.0002,
"step": 9100
},
{
"epoch": 11.8,
"learning_rate": 9.231578947368422e-07,
"loss": 0.0002,
"step": 9125
},
{
"epoch": 11.84,
"learning_rate": 8.968421052631579e-07,
"loss": 0.0002,
"step": 9150
},
{
"epoch": 11.87,
"learning_rate": 8.705263157894737e-07,
"loss": 0.0002,
"step": 9175
},
{
"epoch": 11.9,
"learning_rate": 8.442105263157896e-07,
"loss": 0.0002,
"step": 9200
},
{
"epoch": 11.93,
"learning_rate": 8.178947368421053e-07,
"loss": 0.0002,
"step": 9225
},
{
"epoch": 11.97,
"learning_rate": 7.915789473684212e-07,
"loss": 0.0002,
"step": 9250
},
{
"epoch": 12.0,
"learning_rate": 7.652631578947369e-07,
"loss": 0.0002,
"step": 9275
},
{
"epoch": 12.03,
"learning_rate": 7.389473684210527e-07,
"loss": 0.0002,
"step": 9300
},
{
"epoch": 12.06,
"learning_rate": 7.126315789473685e-07,
"loss": 0.0002,
"step": 9325
},
{
"epoch": 12.1,
"learning_rate": 6.863157894736842e-07,
"loss": 0.0002,
"step": 9350
},
{
"epoch": 12.13,
"learning_rate": 6.6e-07,
"loss": 0.0002,
"step": 9375
},
{
"epoch": 12.16,
"learning_rate": 6.336842105263157e-07,
"loss": 0.0002,
"step": 9400
},
{
"epoch": 12.19,
"learning_rate": 6.073684210526317e-07,
"loss": 0.0002,
"step": 9425
},
{
"epoch": 12.23,
"learning_rate": 5.810526315789474e-07,
"loss": 0.0002,
"step": 9450
},
{
"epoch": 12.26,
"learning_rate": 5.547368421052632e-07,
"loss": 0.0002,
"step": 9475
},
{
"epoch": 12.29,
"learning_rate": 5.284210526315789e-07,
"loss": 0.0002,
"step": 9500
},
{
"epoch": 12.32,
"learning_rate": 5.021052631578948e-07,
"loss": 0.0002,
"step": 9525
},
{
"epoch": 12.35,
"learning_rate": 4.757894736842106e-07,
"loss": 0.0002,
"step": 9550
},
{
"epoch": 12.39,
"learning_rate": 4.4947368421052637e-07,
"loss": 0.0002,
"step": 9575
},
{
"epoch": 12.42,
"learning_rate": 4.2315789473684214e-07,
"loss": 0.0002,
"step": 9600
},
{
"epoch": 12.45,
"learning_rate": 3.968421052631579e-07,
"loss": 0.0002,
"step": 9625
},
{
"epoch": 12.48,
"learning_rate": 3.7052631578947377e-07,
"loss": 0.0002,
"step": 9650
},
{
"epoch": 12.52,
"learning_rate": 3.4421052631578954e-07,
"loss": 0.0002,
"step": 9675
},
{
"epoch": 12.55,
"learning_rate": 3.178947368421053e-07,
"loss": 0.0002,
"step": 9700
},
{
"epoch": 12.58,
"learning_rate": 2.9157894736842107e-07,
"loss": 0.0002,
"step": 9725
},
{
"epoch": 12.61,
"learning_rate": 2.6526315789473684e-07,
"loss": 0.0002,
"step": 9750
},
{
"epoch": 12.65,
"learning_rate": 2.3894736842105265e-07,
"loss": 0.0002,
"step": 9775
},
{
"epoch": 12.68,
"learning_rate": 2.1263157894736842e-07,
"loss": 0.0002,
"step": 9800
},
{
"epoch": 12.71,
"learning_rate": 1.8631578947368424e-07,
"loss": 0.0002,
"step": 9825
},
{
"epoch": 12.74,
"learning_rate": 1.6e-07,
"loss": 0.0002,
"step": 9850
},
{
"epoch": 12.77,
"learning_rate": 1.3368421052631582e-07,
"loss": 0.0002,
"step": 9875
},
{
"epoch": 12.81,
"learning_rate": 1.0736842105263159e-07,
"loss": 0.0002,
"step": 9900
},
{
"epoch": 12.84,
"learning_rate": 8.105263157894738e-08,
"loss": 0.0002,
"step": 9925
},
{
"epoch": 12.87,
"learning_rate": 5.473684210526316e-08,
"loss": 0.0002,
"step": 9950
},
{
"epoch": 12.9,
"learning_rate": 2.842105263157895e-08,
"loss": 0.0002,
"step": 9975
},
{
"epoch": 12.94,
"learning_rate": 2.105263157894737e-09,
"loss": 0.0002,
"step": 10000
},
{
"epoch": 12.94,
"eval_loss": 0.44238531589508057,
"eval_runtime": 980.543,
"eval_samples_per_second": 5.17,
"eval_steps_per_second": 0.647,
"eval_wer": 19.942996961630502,
"step": 10000
}
],
"max_steps": 10000,
"num_train_epochs": 13,
"total_flos": 4.614595985276928e+19,
"trial_name": null,
"trial_params": null
}