whisper-tiny-transcription

This model is a fine-tuned version of openai/whisper-tiny on the covost2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6300
• Wer: 90.7309

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: 8
• eval_batch_size: 4
• 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
• training_steps: 6000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4768	1.3407	1000	0.6468	92.3030
0.3982	2.6814	2000	0.6164	90.8942
0.2697	4.0215	3000	0.6099	90.2613
0.2219	5.3622	4000	0.6209	90.2409
0.1836	6.7029	5000	0.6272	90.5063
0.1622	8.0429	6000	0.6300	90.7309

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.1
• Tokenizers 0.21.0