StephaneBah/whisper-small-rad-FR2

Whisper Small FR - Radiologie

This model is a fine-tuned version of StephaneBah/whisper-small-rad-FR1 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7650
• Wer: 33.3943

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	0.7386	29.1895
No log	12.5	200	0.7423	33.2115
No log	18.75	300	0.7461	33.5771
No log	25.0	400	0.7524	30.8349
0.0005	31.25	500	0.7564	33.6990
0.0005	37.5	600	0.7596	33.8818
0.0005	43.75	700	0.7628	33.6380
0.0005	50.0	800	0.7629	33.0286
0.0005	56.25	900	0.7645	33.4552
0.0002	62.5	1000	0.7650	33.3943

Framework versions

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