wav2vec2-xls-r-300m-fr-30m

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7387
• Wer: 0.3945
• Cer: 0.1314

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: 7e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.98) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
3.0009	17.8571	500	2.9972	1.0	1.0
2.7032	35.7143	1000	2.4955	0.9906	0.8773
1.123	53.5714	1500	0.8270	0.5961	0.1858
0.7079	71.4286	2000	0.6459	0.4674	0.1450
0.496	89.2857	2500	0.6496	0.4271	0.1385
0.4189	107.1429	3000	0.6491	0.4134	0.1335
0.3499	125.0	3500	0.7011	0.4022	0.1332
0.3269	142.8571	4000	0.7218	0.3997	0.1324
0.2987	160.7143	4500	0.7287	0.3868	0.1277
0.2957	178.5714	5000	0.7387	0.3945	0.1314

Framework versions

• Transformers 4.50.0
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1