登录
洞察 - Instruction Tuning Data Collection
Semi-Instruct: Bridging Natural-Instruct and Self-Instruct for Code Large Language Models
Semi-Instruct bridges the gap between Natural-Instruct and Self-Instruct to improve code Large Language Models by converting diverse but improper codes into proper instruction-code pairs.
Semi-Instruct: Bridging Natural-Instruct and Self-Instruct for Code Large Language Models
The author proposes Semi-Instruct as a method to combine the strengths of Natural-Instruct and Self-Instruct for code large language models, addressing issues of diversity and correctness in instruction tuning data.
