On September 5, 2024, Beijing Lingyi Wanwu Technology Co., Ltd. announced the official open-source release of the Yi-Coder series models, marking another significant advancement in the field of artificial intelligence for the company. The Yi-Coder series models focus on programming tasks, offering models with 1.5 billion and 9 billion parameters, with the Yi-Coder-9B demonstrating exceptional performance in code generation, comprehension, debugging, and completion.
The Yi-Coder series models excel in supporting multiple programming languages and can handle context content of up to 128K tokens, effectively capturing long-term dependencies, making them suitable for understanding and generating complex project-level code. In the LiveCodeBench code generation benchmark test, the Yi-Coder-9B leads among models with less than 10 billion parameters, even surpassing some models with larger parameter sizes.
Additionally, Yi-Coder is equally robust in code editing and completion capabilities, achieving excellent results in the CodeEditorBench evaluation. In terms of long sequence modeling and mathematical reasoning, Yi-Coder also demonstrates its powerful performance, with outstanding results in both long text comprehension and mathematical problem-solving.
Project Address: https://github.com/01-ai/Yi-Coder