wav2vec2-large-xlsr-53-Bhojpuri-Version6

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.9380
• Wer: 0.4226

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.5686	2.3108	2000	1.2936	0.6546
1.1325	4.6216	4000	1.0153	0.4864
1.3378	6.9324	6000	0.9690	0.4716
0.9299	9.2432	8000	0.9459	0.4465
0.9488	11.5540	10000	0.9376	0.4407
1.1409	13.8648	12000	0.9411	0.4288
1.3824	16.1756	14000	0.9388	0.4372
0.6866	18.4864	16000	0.9380	0.4226

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.1+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1