wav2vec2-xls-r-300m-oc-10min

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: 1.0358
• Wer: 0.5585
• Cer: 0.1390

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
3.08	55.5556	500	2.9377	1.0	1.0
2.6754	111.1111	1000	2.3376	1.0	0.8038
1.0095	166.6667	1500	0.8765	0.7114	0.1870
0.5376	222.2222	2000	0.8198	0.6131	0.1545
0.3166	277.7778	2500	0.8888	0.5866	0.1485
0.2414	333.3333	3000	0.9621	0.5928	0.1518
0.1924	388.8889	3500	0.9775	0.5601	0.1425
0.1064	444.4444	4000	1.0213	0.5569	0.1409
0.0988	500.0	4500	1.0345	0.5523	0.1377
0.1062	555.5556	5000	1.0358	0.5585	0.1390

Framework versions

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