StephaneBah/whisper-small-rad-FR1

Whisper Small FR - Radiologie

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: 0.7451
• Wer: 35.5881

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 6
• seed: 3407
• optimizer: Use OptimizerNames.ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	6.25	100	0.8071	99.6953
No log	12.5	200	0.7781	33.0896
No log	18.75	300	0.7461	31.9317
No log	25.0	400	0.7410	35.6490
0.0584	31.25	500	0.7419	35.8318
0.0584	37.5	600	0.7449	35.7099
0.0584	43.75	700	0.7448	35.7709
0.0584	50.0	800	0.7452	35.5881
0.0584	56.25	900	0.7456	35.5881
0.0003	62.5	1000	0.7451	35.5881

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.2