EffiBench/effibench-x

EffiBench-X是一个多语言编程竞赛问题数据集，旨在评估LLM生成的代码的效率。该数据集包含623个问题，涵盖Python、C++、Java、JavaScript、Ruby和Golang六种编程语言，每个问题都配有人类专家的解决方案。数据集支持多语言评估，并提供效率指标和人类专家基线。EffiBench-X is a multi-language competitive programming problem dataset designed to evaluate the efficiency of LLM-generated code. The dataset includes 623 problems covering six programming languages: Python, C++, Java, JavaScript, Ruby, and Golang, each with expert solutions provided by human experts. The dataset supports multi-language evaluation and provides efficiency metrics and human expert baselines.