whisper_medium_phase2

This model is a fine-tuned version of PThi35/whisper_medium_phase2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8362
• Cer: 18.6667
• Wer: 29.4586

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: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Cer	Validation Loss	Wer
0.5275	0.9999	4709	22.0874	0.7764	35.0416
0.5415	2.0	9419	20.0444	0.7521	31.8854
0.4568	2.9999	14128	19.6955	0.7660	30.9815
0.3943	4.0	18838	19.4005	0.7752	30.7362
0.3481	4.9995	23545	18.6947	0.7911	29.7321
0.351	5.9999	28254	0.8078	18.9302	29.9617
0.3149	7.0	32964	0.8259	18.9425	29.8582
0.2879	7.9997	37672	0.8362	18.6667	29.4586

Framework versions

• Transformers 4.41.2
• Pytorch 2.1.2+cu118
• Datasets 2.19.0
• Tokenizers 0.19.1