wav2vec2-hindi-3

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.0900
• Wer: 0.7281

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• 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: 40
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.609	6.41	1000	1.2290	0.7497
0.3754	12.82	2000	1.5350	0.7128
0.1587	19.23	3000	1.8671	0.7322
0.103	25.64	4000	1.9383	0.7300
0.0761	32.05	5000	2.0767	0.7306
0.0616	38.46	6000	2.0900	0.7281

Framework versions

• Transformers 4.20.0.dev0
• Pytorch 1.11.0+cu113
• Datasets 2.2.3.dev0
• Tokenizers 0.12.1