wav2vec2-mms-1b-CV17.0

This model is a fine-tuned version of facebook/mms-1b-all on the common_voice_17_0 dataset. Adapters for several languages were trained.

Intended uses & limitations

Speech-to-text transciption of Malayalam, Tamil, Telugu, and Yoruba.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.001
• 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_ratio: 0.15
• training_steps: 2000
• mixed_precision_training: Native AMP

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1