LLM Merging Methods for Coding Tasks

This table analyzes LLM merging approaches applied to coding tasks, covering standard evaluation suites such as HumanEval and MBPP. It examines training dependencies, models used and computational overhead.

本表格针对应用于编码任务的大语言模型（LLM）合并方法展开分析，涵盖HumanEval与MBPP等标准评测套件。该分析同时考察了训练依赖关系、所用模型以及计算开销。