whisper-fine-tuned

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

• Loss: 1.2411
• Wer: 82.1906

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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: 500
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
5.1882	0.4193	100	1.2411	129.5950
4.9978	0.8386	200	1.2093	76.9245
3.6713	1.2558	300	1.2233	75.6337
3.8984	1.6751	400	1.2455	88.2705
2.2193	2.0922	500	1.3534	102.7032
1.9958	2.5115	600	1.3091	82.6396
1.7498	2.9308	700	1.2411	82.1906

Framework versions

• Transformers 4.47.1
• Pytorch 2.4.1+cu118
• Datasets 3.2.0
• Tokenizers 0.21.0