Qwen

/

Qwen2.5-Coder-7B-Instruct

like 95

In my case, this model evaluation score is 85.71 with CodeStral-22B, compared to 85.24.

BTW:
Blog page has an error in the description:
「甚至打败了更大尺寸的 DeepSeek-Coder-V2-Lite 和 Codestral-20B」
正确的写法应该是 CodeStral-22B