改PDF文件名能提高LLM评审分数,建议改用二元判断

LLM review weirdness indeed. Avoid using scores with LLM judges, or be extremely careful if you do...

精选理由

把论文名改成最终送审版,GPT-5.6-Terra给的分数就涨了。以后让LLM审稿别要分数,直接让它说过/不过。

AI 摘要

Zachary Horvitz发现,把上传的论文从paper.pdf改名为paper_final_draft_pdf_ready_for_review.pdf,GPT-5.6-Terra给出的平均评审分数就会上涨。这个针对GPT-5.6-Terra的实验显示,LLM评审会受文件名等无关因素干扰。他因此建议,在GPT-5.6-Terra这类模型上不要用数值评分做评审,尽量改用二元标签,比如通过/不通过。

原文 · elvis

LLM review weirdness indeed. Avoid using scores with LLM judges, or be extremely careful if you do...

LLM review weirdness indeed. Avoid using scores with LLM judges, or be extremely careful if you do. Use binary labels where possible. Zachary Horvitz @zachary_horvitz LLM review weirdness... Just renaming an uploaded pdf from "paper.pdf" to "paper_final_draft_pdf_ready_for_review.pdf" boosts average scores (gpt-5.6-terra) 🔗 View Quoted Tweet 💬 0 🔄 0 ❤️ 4 👀 1880 📊 1 ⚡