• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI
    Announcement

    Priced Guidance is a proposed '20 Questions' test for LLM research ideas

    Its creators say the method has an LLM recover a future paper’s idea, charging each hint by how surprising it is.

    CLSCL
    You JiachengYJ
    Kaiyue WenKW
    3 Sources, ,

    TLDR

    The researchers propose Priced Guidance to assess whether LLMs can generate novel research ideas. Instead of sampling ideas until one matches a future paper’s idea, they have the model play '20 Questions' to recover it, charging each hint by how surprising it is. They say this gives a lower bound on the probability that the LLM would forecast that idea.

    Combined views

    9.8K

    3 Sources, first seen 2h ago

    Combined views

    9.8K

    3 Sources, first seen 2h ago

    169 likes
    2h ago
    first seen 2h ago
    169 likes
    7 comments
    112 saves
    46 reposts
    Featured Source
    7 comments
    112 saves
    46 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    3 Sources

    Kaiyue Wen@wen_kaiyueCan LLMs generate novel research ideas? We propose Priced Guidance to quantify this via compression! We lower bound P(LLM forecasts a future paper’s idea X). Instead of sampling ideas until X is matched, the LLM plays “20 Questions” to recover X, charging each hint by surprisal.2h
    CLS@ChengleiSiRT @wen_kaiyue: Can LLMs generate novel research ideas? We propose Priced Guidance to quantify this via compression! We lower bound P(LLM f…2h
    You Jiacheng@YouJiachengcool work1h

    3 Sources

    Kaiyue Wen@wen_kaiyueCan LLMs generate novel research ideas? We propose Priced Guidance to quantify this via compression! We lower bound P(LLM forecasts a future paper’s idea X). Instead of sampling ideas until X is matched, the LLM plays “20 Questions” to recover X, charging each hint by surprisal.2h
    CLS@ChengleiSiRT @wen_kaiyue: Can LLMs generate novel research ideas? We propose Priced Guidance to quantify this via compression! We lower bound P(LLM f…2h
    You Jiacheng@YouJiachengcool work1h