whisper-small-alif

This model is a fine-tuned version of openai/whisper-small on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.2298
• Wer: 0.0

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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: 50
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	4	3.3199	100.0
6.3947	2.0	8	3.2238	100.0
6.3570	3.0	12	1.7064	100.0
3.6043	4.0	16	0.8654	75.0
1.3034	5.0	20	0.1943	0.0
1.3034	6.0	24	0.0241	0.0
0.1757	7.0	28	0.0071	0.0
0.0055	8.0	32	0.0000	0.0
0.0002	9.0	36	0.0000	0.0
0.0000	10.0	40	0.0000	0.0
0.0000	11.0	44	0.0000	0.0
0.0000	12.0	48	0.0000	0.0
0.0000	13.0	52	0.0000	0.0
0.0000	14.0	56	0.0000	0.0
0.0000	15.0	60	0.0000	0.0
0.0000	16.0	64	0.0000	0.0
0.0000	17.0	68	0.0000	0.0
0.0000	18.0	72	0.0000	0.0
0.0000	19.0	76	0.0000	0.0
0.0000	20.0	80	0.0000	0.0

Framework versions

• Transformers 5.12.0
• Pytorch 2.11.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2