deepdml/whisper-small-mix-fr

Whisper Small Mixed-French

This model is a fine-tuned version of openai/whisper-small on the fr datasets:

• mozilla-foundation/common_voice_17_0
• google/fleurs
• facebook/multilingual_librispeech
• facebook/voxpopuli

It achieves the following results on the evaluation set:

• Loss: 0.3092
• Wer: 15.0158

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: 64
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.187	0.2	1000	0.3653	17.3498
0.1445	0.4	2000	0.3379	16.0480
0.1659	0.6	3000	0.3255	15.3772
0.1594	0.8	4000	0.3136	15.1959
0.1371	1.0	5000	0.3092	15.0158

Framework versions

• Transformers 4.42.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1