精选理由
Perplexity AI 搞了个新评估方法，软硬分数区别很清晰，基准和工具都开源了，做评测的可以看看。
Perplexity AI 引入了软分数（soft scores）和硬分数（hard scores）两种评估机制，软分数允许部分正确给分，硬分数要求成员完全正确。相关基准任务和评估工具已开源至 GitHub。这套方法为细粒度评估提供了新思路。
原文 · Perplexity
Soft scores give partial credit. Hard scores require every component for a member to be complete and...
Soft scores give partial credit. Hard scores require every component for a member to be complete and correct. The benchmark tasks and evaluation harness are available at: github.com/perplexityai/w… 💬 1 🔄 4 ❤️ 9 👀 1795 📊 3 ⚡