whisper-fine_tuning

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: 4.4096
• Wer: 89.7704

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-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• training_steps: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
5.127	0.04	10	6.5305	89.7704
4.9407	0.08	20	5.6702	88.1002
3.9127	0.12	30	5.2648	85.1775
3.4678	0.16	40	5.0057	84.7599
3.7416	0.2	50	4.8397	85.3862
3.1575	0.24	60	4.6961	86.4301
3.3175	0.28	70	4.5819	87.2651
2.9554	0.32	80	4.4950	88.1002
3.0291	0.36	90	4.4375	89.7704
3.0219	0.4	100	4.4096	89.7704

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1