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
研究者 AVB 测试 Jev 后发现相同提示会产生不同的输出概率,质疑其“无幻觉”声明。
TypeSafe AI's Jev model is facing new scrutiny today over its "no-hallucination" claims. Here's what you need to know. Researcher AVB, who posts as @neural_avb, tested Jev and found it doesn't run deterministically. The exact same prompt produces different output probabilities