Breeze 7B Cantonese V0.1

🚀 Model Card for Breeze-7B-Cantonese-v0.1

Breeze-7B-Cantonese-v0.1 is a language model designed for Cantonese interaction. It is based on the Breeze-7B family, which in turn builds on Mistral-7B, and is specifically optimized for Traditional Chinese usage. This project is made possible thanks to the contributions from MediaTek-Research.

📚 Documentation

Model Background

Breeze-7B is a language model family built upon Mistral-7B, specifically crafted for Traditional Chinese use. It is developed by MediaTek-Research.

[Breeze-7B-Cantonese] is derived from the base model Breeze-7B-Base. It is fine - tuned using datasets from hon9kon9ize, enabling the model to engage in conversations in Cantonese.

The [Breeze-7B-Base] model was chosen as the base because of its extended vocabulary coverage for Traditional Chinese. Its language style closely matches the characteristics of Hong Kong discourse, making it a suitable choice for this project.

Open - Source Sharing

The axolotl configuration file axolotl-config.yml is shared as open - source, allowing users to train the model on their own.

Acknowledgments

This project is highly indebted to the excellent datasets provided by hon9kon9ize and indiejoseph. Their contributions are crucial for the existence of this project.

📄 License

The model is released under the Apache 2.0 license.

📦 Related Information