learning_rate: 0.0003
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH 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
num_epochs: 30
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
5.9857	0.8782	400	1.1527	0.6884
1.3889	1.7552	800	0.9422	0.5880
1.187	2.6323	1200	0.8545	0.5501
1.1338	3.5093	1600	0.7957	0.5231
1.0482	4.3864	2000	0.7578	0.4994
1.0305	5.2634	2400	0.7273	0.4851
0.9609	6.1405	2800	0.7110	0.4754
0.9499	7.0176	3200	0.6915	0.4636
0.9284	7.8957	3600	0.6732	0.4544
0.9071	8.7728	4000	0.6631	0.4484
0.8788	9.6498	4400	0.6616	0.4447
0.8719	10.5269	4800	0.6410	0.4271
0.8536	11.4040	5200	0.6324	0.4229
0.8389	12.2810	5600	0.6156	0.4139
0.8155	13.1581	6000	0.6139	0.4076
0.8226	14.0351	6400	0.6111	0.4054
0.8108	14.9133	6800	0.5977	0.3992
0.7765	15.7903	7200	0.5962	0.3981
0.8176	16.6674	7600	0.5899	0.3939
0.7887	17.5445	8000	0.5890	0.3897
0.761	18.4215	8400	0.5807	0.3812
0.7703	19.2986	8800	0.5763	0.3866
0.7602	20.1756	9200	0.5717	0.3782
0.7642	21.0527	9600	0.5643	0.3744
0.7558	21.9308	10000	0.5686	0.3746
0.7238	22.8079	10400	0.5643	0.3716
0.7797	23.6850	10800	0.5620	0.3691
0.7402	24.5620	11200	0.5528	0.3665
0.7404	25.4391	11600	0.5508	0.3659
0.7125	26.3161	12000	0.5567	0.3650
0.724	27.1932	12400	0.5471	0.3633
0.7325	28.0703	12800	0.5467	0.3623
0.7247	28.9484	13200	0.5481	0.3617
0.7307	29.8255	13600	0.5484	0.3622

Tashi-Projects/DZO_ASR_mms1ball

DZO_ASR_mms1ball