whisper-small-reverse-ml-mft-1

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: 0.0806
• Wer: 44.7778
• Cer: 14.6248

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: 0.0002
• train_batch_size: 64
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 593
• training_steps: 5928

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.0667	1.0	1482	0.0940	51.0450	17.3373
0.0523	2.0	2964	0.0779	45.0478	14.6619
0.0344	3.0	4446	0.0767	44.3816	14.4052
0.0279	4.0	5928	0.0806	44.7778	14.6248

Framework versions

• Transformers 4.51.1
• Pytorch 2.8.0+cu128
• Datasets 4.4.1
• Tokenizers 0.21.4