TypeSafe AI 被用于学术引用检查工具
用户将 @typesafeai 加入其引用检查器并给予好评,可在 PaperTrellis 免费试用,标签涉及 Jev 与医学 AI。 链接
该帖子引述腾讯研究论文,评估Jev在选项顺序变化下的判断能力,结果显示Jev在可解决问题上常出错,与作者经验相符。
A paper by Tencent comparing the judgments of Jev with those of GPT, Claude, Gemini, and others. The results show that Jev often gets even solvable problems wrong when the order of the options is changed (Figure d). Somehow, this feels like it aligns with my experience.