We introduce PaperBench, a benchmark evaluating the ability of AI agents to replicate state-of-the-art AI research. Agents must replicate 20 ICML 2024 Spotlight and Oral papers from scratch, including understanding paper contributions, developing a codebase, and successfully executing experiments. For objective evaluation, we develop rubrics that hierarchically decompose each replication task into smaller sub-tasks with clear grading criteria. In total, PaperBench contains 8,316 individually gradable tasks. Rubrics are co-developed with the author(s) of each ICML paper for accuracy and realism. To enable scalable evaluation, we also develop an LLM-based judge to automatically grade replication attempts against rubrics, and assess our judge's performance by creating a separate benchmark for judges. We evaluate several frontier models on PaperBench, finding that the best-performing tested agent, Claude 3.5 Sonnet (New) with open-source scaffolding, achieves an average replication score of 21.0\%. Finally, we recruit top ML PhDs to attempt a subset of PaperBench, finding that models do not yet outperform the human baseline. We \href{https://github.com/openai/preparedness}{open-source our code} to facilitate future research in understanding the AI engineering capabilities of AI agents.

翻译：本文介绍PaperBench，这是一个评估AI智能体复制前沿AI研究能力的基准测试。智能体必须从零开始复制20篇ICML 2024焦点报告与口头报告论文，包括理解论文贡献、开发代码库以及成功执行实验。为实现客观评估，我们开发了分层评估标准，将每个复制任务分解为具有明确评分标准的子任务。PaperBench共包含8,316个可独立评分的任务。评估标准与各ICML论文作者共同开发以确保准确性和真实性。为实现可扩展评估，我们还开发了基于LLM的自动评分器，依据评估标准对复制尝试进行自动评分，并通过创建独立的评分器基准来评估其性能。我们在PaperBench上评估了多个前沿模型，发现表现最佳的测试智能体（配备开源框架的Claude 3.5 Sonnet (New)）平均复制得分仅为21.0%。最后，我们招募顶尖机器学习博士生尝试PaperBench的子集任务，发现当前模型性能尚未超越人类基线。我们\href{https://github.com/openai/preparedness}{开源代码}以促进未来对AI智能体工程化能力的研究。