PaperBench: Evaluating AI’s Ability to Replicate AI Research
来自 OpenAI News
· 2025-04-02
精选
We introduce PaperBench, a benchmark evaluating the ability of AI agents to replicate state-of-the-art AI research.