原帖引用 · AI 翻译
Rohan Paul@rohanpaul_ai
OpenAI 现已发布其药物化学基准测试 MedChemBench。
其题目来自 117 篇论文和专利,涵盖 761 个接受评分的分子结构,并隐藏了研究人员后来报告的结果。
每道题都停在一个真实的决策节点,提供生物学靶点、截至该节点已制备的化合物以及部分实验测定结果。
表现最好的模型 GPT-6.2 Astra 总体得分约为 84%,在理解化学内容方面几乎完美,但在真正关键的决策上,正确率约为一半。
原文
OpenAI has now published MedChemBench, its medicinal chemistry benchmark. Its problems come from 117 papers and patents, span 761 graded molecular structures, and withhold the results researchers reported later. Each prompt stops at a real decision point, supplying the biological target, the compounds made so far and selected assay results. the best model GPT-6.2 Astra gets about 84% overall, is near perfect at reading the chemistry, and is right about half the time on the decision that actually matters.
· UTC+8