whisper-base.en-fsc-h

This model is a fine-tuned version of openai/whisper-base.en on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0700
• Accuracy: 0.9863

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0005
• train_batch_size: 48
• eval_batch_size: 48
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 192
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9959	120	0.7203	0.7332
No log	2.0	241	0.3685	0.8505
No log	2.9959	361	0.2024	0.9293
No log	4.0	482	0.1429	0.9552
No log	4.9959	602	0.1143	0.9636
No log	6.0	723	0.1738	0.9565
No log	6.9959	843	0.1295	0.9560
No log	8.0	964	0.1407	0.9599
0.3435	8.9959	1084	0.1104	0.9694
0.3435	10.0	1205	0.1054	0.9718
0.3435	10.9959	1325	0.1042	0.9728
0.3435	12.0	1446	0.1084	0.9710
0.3435	12.9959	1566	0.0926	0.9789
0.3435	14.0	1687	0.1120	0.9763
0.3435	14.9959	1807	0.0903	0.9797
0.3435	16.0	1928	0.0957	0.9800
0.0255	16.9959	2048	0.0851	0.9815
0.0255	18.0	2169	0.0723	0.9844
0.0255	18.9959	2289	0.0784	0.9834
0.0255	20.0	2410	0.0673	0.9860
0.0255	20.9959	2530	0.0700	0.9863
0.0255	22.0	2651	0.0707	0.9858
0.0255	22.9959	2771	0.0712	0.9858
0.0255	24.0	2892	0.0715	0.9858
0.0015	24.8963	3000	0.0716	0.9858

Framework versions

• Transformers 4.43.3
• Pytorch 2.2.2+cu121
• Datasets 2.18.0
• Tokenizers 0.19.1