learning_rate: 0.0004
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use 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: 132
num_epochs: 30
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	59	0.4281	266.4391
1.3358	2.0	118	1.5591	60.6371
1.3358	3.0	177	0.5338	308.3049
0.4635	4.0	236	0.3016	44.3686
0.4635	5.0	295	0.2477	34.8123
0.2439	6.0	354	0.3004	45.3925
0.1766	7.0	413	0.2947	38.2253
0.1766	8.0	472	0.4231	63.8225
0.1258	9.0	531	0.2170	37.5427
0.1258	10.0	590	0.2435	104.2093
0.1033	11.0	649	0.2698	36.7463
0.0805	12.0	708	0.2398	35.1536
0.0805	13.0	767	0.2139	54.9488
0.0553	14.0	826	0.3215	41.5245
0.0553	15.0	885	0.2489	39.5904
0.0559	16.0	944	0.1868	52.3322
0.0377	17.0	1003	0.2378	56.3140
0.0377	18.0	1062	0.2427	30.2617
0.029	19.0	1121	0.1880	25.4835
0.029	20.0	1180	0.1866	27.7588
0.0111	21.0	1239	0.2203	22.8669
0.0111	22.0	1298	0.2507	28.4414
0.0037	23.0	1357	0.2227	26.9625
0.0025	24.0	1416	0.2157	24.4596
0.0025	25.0	1475	0.2189	24.6871
0.0005	26.0	1534	0.2184	24.2321
0.0005	27.0	1593	0.2184	25.1422
0.0002	28.0	1652	0.2184	24.2321
0.0001	29.0	1711	0.2184	24.2321
0.0001	29.4957	1740	0.2184	24.2321

susmitabhatt/whisper_nmc_nomimose_30

whisper_nmc_nomimose_30