Wav2vec2 Large Xls R 300m Hi Cv8

🚀 wav2vec2-large-xls-r-300m-hi-cv8

This model is a fine - tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA - FOUNDATION/COMMON_VOICE_8_0 - HI dataset. It is designed for automatic speech recognition tasks, providing a reliable solution for converting spoken language into text.

✨ Features

• Multilingual Adaptability: Based on a large - scale pre - trained model, it can be adapted to different languages, including Hindi in this case.
• High - Quality Results: Achieves relatively low Word Error Rate (WER) and Character Error Rate (CER) on evaluation datasets.

📦 Installation

No installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

No code examples are provided in the original document, so this section is skipped.

📚 Documentation

Model Information

Evaluation Results

This model achieves the following results on the evaluation set:

• Loss: 0.6510
• Wer: 0.3179

Evaluation Commands

1. To evaluate on mozilla-foundation/common_voice_8_0 with test split

python eval.py --model_id DrishtiSharma/wav2vec2-large-xls-r-300m-hi-cv8 --dataset mozilla-foundation/common_voice_8_0 --config hi --split test --log_outputs

2. To evaluate on speech-recognition-community-v2/dev_data

python eval.py --model_id DrishtiSharma/wav2vec2-large-xls-r-300m-hi-cv8 --dataset speech-recognition-community-v2/dev_data --config hi --split validation --chunk_length_s 10 --stride_length_s 1

Note: Hindi language not found in speech-recognition-community-v2/dev_data

Training Hyperparameters

The following hyperparameters were used during training:

• 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: 2000
• num_epochs: 50
• mixed_precision_training: Native AMP

Training Results

Framework Versions

• Transformers 4.16.2
• Pytorch 1.10.0+cu111
• Datasets 1.18.3
• Tokenizers 0.11.0

📄 License

This model is licensed under the Apache - 2.0 license.