learning_rate: 1e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 4000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
1.7549	0.4630	100	1.3537	89.9316
0.9026	0.9259	200	0.7020	52.0959
0.7153	1.3889	300	0.5401	41.5519
0.5719	1.8519	400	0.4612	36.8348
0.4224	2.3148	500	0.4259	33.6538
0.4071	2.7778	600	0.3982	32.1871
0.2888	3.2407	700	0.3870	31.1862
0.2681	3.7037	800	0.3840	31.1961
0.1823	4.1667	900	0.3938	30.8988
0.1472	4.6296	1000	0.3931	29.5412
0.1011	5.0926	1100	0.4019	30.3241
0.1173	5.5556	1200	0.4080	29.4123
0.0879	6.0185	1300	0.4151	29.1448
0.0607	6.4815	1400	0.4308	30.5916
0.0573	6.9444	1500	0.4310	29.5214
0.0336	7.4074	1600	0.4542	29.4520
0.0376	7.8704	1700	0.4568	29.7493
0.0150	8.3333	1800	0.4701	29.8583
0.0211	8.7963	1900	0.4696	29.7988
0.0086	9.2593	2000	0.4828	29.3529
0.0111	9.7222	2100	0.4893	29.2042
0.0059	10.1852	2200	0.5054	28.9169
0.0063	10.6481	2300	0.5054	29.2042
0.0044	11.1111	2400	0.5182	29.1745
0.0036	11.5741	2500	0.5230	29.5412
0.0028	12.0370	2600	0.5253	29.2042
0.0018	12.5	2700	0.5344	29.0655
0.0028	12.9630	2800	0.5364	28.7880
0.0017	13.4259	2900	0.5398	28.8970
0.0014	13.8889	3000	0.5440	28.7088
0.0011	14.3519	3100	0.5488	28.9069
0.0012	14.8148	3200	0.5479	29.0853
0.0009	15.2778	3300	0.5522	28.9367
0.0011	15.7407	3400	0.5529	28.8673
0.0009	16.2037	3500	0.5561	28.9961
0.0009	16.6667	3600	0.5577	29.0655
0.0008	17.1296	3700	0.5589	29.0853
0.0008	17.5926	3800	0.5602	29.0655
0.0007	18.0556	3900	0.5605	28.9961
0.0007	18.5185	4000	0.5609	29.0160

dvm14/whisper-small-sd

whisper-small-sd