whisper-nm-no-ls

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

• Loss: 0.1565
• Wer: 11.1562

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: 0.0004
• 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: 11
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	33	0.1768	8.9249
No log	2.0	66	0.2439	20.8925
No log	3.0	99	0.4285	255.7809
1.2232	4.0	132	0.2449	26.3692
1.2232	5.0	165	0.5282	37.9310
1.2232	6.0	198	0.2170	25.3550
0.3525	7.0	231	0.1980	74.4422
0.3525	8.0	264	0.1585	14.6045
0.3525	9.0	297	0.1963	18.0527
0.0919	10.0	330	0.1787	12.9817
0.0919	11.0	363	0.1565	11.1562

Framework versions

• Transformers 4.47.0.dev0
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0