Back/
LiveCodeBench

LiveCodeBench

FreeCode Assistant

LiveCodeBench是一个面向代码大语言模型的全面且无污染的评估基准。它持续收集最新编程竞赛题目,支持代码生成、自我修复、执行与测试预测等多场景评测,帮助研究者客观衡量模型的泛化能力与真实水平。

Visit Website
LiveCodeBench

Our Verdict

AI is analyzing...

Who's Using It

personas.forDevelopers🎓Student Friendly

Features

Continuously collect the latest competition questions to prevent training data contamination
Cover multiple dimensions such as code generation, self-repair, execution, and test prediction
Provide dynamic evaluation of model generalization ability over time
Open-source submission mechanism, supporting custom model integration and leaderboard updates
Deep comparison of the performance of open-source and closed-source models in complex code tasks

Use Cases

Evaluate the true generalization ability of large language models on unseen programming problemsCompare performance differences among different code models in generation, repair, and execution tasksDetect and analyze potential overfitting issues of models in traditional benchmark tests

Best For

AI large model researchersCode large model developersAlgorithm competition and programming education professionals

Pricing

Free

Similar Tools

Related Tools