Wav2vec2 Large Xls R 300m Latvian

🚀 wav2vec2-large-xls-r-300m-latvian

This model is a fine - tuned version of facebook/wav2vec2-xls-r-300m for automatic speech recognition in Latvian, offering high - quality speech recognition capabilities.

🚀 Quick Start

This model is a fine - tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA - FOUNDATION/COMMON_VOICE_7_0 - LV dataset. It achieves the following results on the evaluation set:

• Loss: 0.1892
• Wer: 0.1698

✨ Features

• Multilingual Adaptability: Based on the pre - trained model facebook/wav2vec2-xls-r-300m, it can adapt to different language environments.
• High - precision Recognition: Achieved low WER and CER on multiple datasets, demonstrating high - precision speech recognition capabilities.

📦 Installation

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

📚 Documentation

Model Information

Model Performance

The model has been evaluated on multiple datasets, and the results are as follows:

• Common Voice 7:
  • Test WER: 16.977
  • Test CER: 4.23
• Robust Speech Event - Dev Data:
  • Test WER: 45.247
  • Test CER: 16.924
• Robust Speech Event - Test Data:
  • Test WER: 56.16

Training Procedure

Training Hyperparameters

The following hyperparameters were used during training:

• learning_rate: 7e - 05
• train_batch_size: 32
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• num_epochs: 100.0
• mixed_precision_training: Native AMP

Training Results

Framework Versions

• Transformers 4.16.0.dev0
• Pytorch 1.10.1+cu102
• Datasets 1.17.1.dev0
• Tokenizers 0.11.0

🔧 Technical Details

The model is based on the pre - trained model facebook/wav2vec2-xls-r-300m and fine - tuned on the mozilla - foundation/common_voice_7_0 dataset. During the training process, a series of hyperparameters were used to adjust the model, such as learning rate, batch size, etc. Through multiple epochs of training, the model gradually converges and achieves good performance on the evaluation set.

📄 License

This model is licensed under the apache - 2.0 license.