BELLE

The goal of this project is to promote the development of an open-source community for Chinese conversational large language models, with the vision of becoming an LLM Engine that can help everyone.
Rather than focusing on how to effectively pre-train large language models, BELLE is more concerned with how to build on the foundation of open-source pre-trained large language models to help everyone obtain their own high-performing, instruction-driven language model, thereby lowering the barriers to research and application of large language models, especially Chinese ones. To this end, the BELLE project will continuously provide access to instruction training data, related models, training code, application scenarios, and more, while also evaluating the impact of different training data and training algorithms on model performance. BELLE is optimized for Chinese and the model fine-tuning uses only data produced by ChatGPT (without incorporating any other data).