hip94/whisper-small-fr-full

Whisper Small Fr - IMT Atlantique X 52 Hertz Full

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

• Loss: 0.7958
• Wer: 29.5191

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: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0003	47.6265	1000	0.7117	34.4942
0.0001	95.2410	2000	0.7619	29.8507
0.0001	142.8675	3000	0.7865	29.5191
0.0	190.4819	4000	0.7958	29.5191

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 4.4.1
• Tokenizers 0.22.1