whisper-cv-grain-lg_both

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

Model description

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 5e-05
train_batch_size: 8
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Use adamw_hf with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 100
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1.8815	1.0	5406	0.2234	0.6850	0.3766
0.5333	2.0	10812	0.1012	0.1432	0.0389
0.3488	3.0	16218	0.0827	0.1022	0.0269
0.2577	4.0	21624	0.0675	0.0844	0.0245
0.2069	5.0	27030	0.0660	0.0866	0.0260
0.1778	6.0	32436	0.0612	0.0766	0.0230
0.1634	7.0	37842	0.0636	0.0705	0.0227
0.1531	8.0	43248	0.0663	0.0675	0.0210
0.1493	9.0	48654	0.0565	0.0479	0.0175
0.1456	10.0	54060	0.0625	0.0603	0.0196
0.1346	11.0	59466	0.0587	0.0558	0.0220
0.1119	12.0	64872	0.0611	0.0558	0.0202
0.0947	13.0	70278	0.0482	0.0438	0.0174
0.0824	14.0	75684	0.0458	0.0412	0.0158
0.0722	15.0	81090	0.0529	0.0499	0.0190
0.0641	16.0	86496	0.0546	0.0438	0.0150
0.0585	17.0	91902	0.0481	0.0423	0.0154
0.0529	18.0	97308	0.0433	0.0399	0.0168
0.0484	19.0	102714	0.0482	0.0430	0.0152
0.0449	20.0	108120	0.0492	0.0408	0.0155
0.0416	21.0	113526	0.0449	0.0375	0.0144
0.0383	22.0	118932	0.0464	0.0323	0.0134
0.0359	23.0	124338	0.0485	0.0386	0.0155
0.0333	24.0	129744	0.0454	0.0484	0.0213
0.0305	25.0	135150	0.0505	0.0388	0.0139
0.0289	26.0	140556	0.0428	0.0367	0.0147
0.0267	27.0	145962	0.0477	0.0362	0.0142
0.0254	28.0	151368	0.0396	0.0299	0.0125
0.0239	29.0	156774	0.0569	0.0373	0.0135
0.0226	30.0	162180	0.0525	0.0328	0.0121
0.0216	31.0	167586	0.0418	0.0291	0.0117
0.0195	32.0	172992	0.0544	0.0328	0.0122
0.0191	33.0	178398	0.0536	0.0315	0.0136
0.0178	34.0	183804	0.0453	0.0312	0.0123
0.0171	35.0	189210	0.0504	0.0395	0.0149
0.0154	36.0	194616	0.0475	0.0319	0.0133
0.0149	37.0	200022	0.0488	0.0321	0.0125
0.0144	38.0	205428	0.0499	0.0358	0.0142