Log inSign up
Image
user avatar
Gabrielle Kaili-May Liu (Kaili)
@pybeebee
CS PhD Candidate @Yale | Research Intern @Google
pybeebee.com
Joined September 2015
173
Following
149
Followers
RepliesMedia
  • Pinned
    user avatar
    Gabrielle Kaili-May Liu (Kaili)
    @pybeebee
    Jul 8
    ๐ŸŽ‰Thrilled to share that RLMF is headed to #COLM2026! In this work we introduce ๐—ฅ๐—Ÿ ๐˜„๐—ถ๐˜๐—ต ๐—บ๐—ฒ๐˜๐—ฎ๐—ฐ๐—ผ๐—ด๐—ป๐—ถ๐˜๐—ถ๐˜ƒ๐—ฒ ๐—ณ๐—ฒ๐—ฒ๐—ฑ๐—ฏ๐—ฎ๐—ฐ๐—ธ (๐—ฅ๐—Ÿ๐— ๐—™) & metacognitive data selection. It outperforms standard RL while improving metacognitive monitoring in LLMs ๐Ÿง โœจ Details below!๐Ÿ‘‡
    user avatar
    Gabrielle Kaili-May Liu (Kaili)
    @pybeebee
    Jul 1
    ๐Ÿ”ฅ New paper alert!! Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs ๐Ÿ”ฅ Can a model that learns to ๐˜ซ๐˜ถ๐˜ฅ๐˜จ๐˜ฆ ๐˜ช๐˜ต๐˜ด๐˜ฆ๐˜ญ๐˜ง get even better at tasksโ€”and learn to be honest about what it doesn't know? Turns out: yes๐Ÿคฏ Details๐Ÿงต๐Ÿ‘‡
    Image
  • user avatar
    Gabrielle Kaili-May Liu (Kaili)
    @pybeebee
    Jul 1
    ๐Ÿ”ฅ New paper alert!! Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs ๐Ÿ”ฅ Can a model that learns to ๐˜ซ๐˜ถ๐˜ฅ๐˜จ๐˜ฆ ๐˜ช๐˜ต๐˜ด๐˜ฆ๐˜ญ๐˜ง get even better at tasksโ€”and learn to be honest about what it doesn't know? Turns out: yes๐Ÿคฏ Details๐Ÿงต๐Ÿ‘‡
    Image
  • user avatar
    Gabrielle Kaili-May Liu (Kaili)
    @pybeebee
    Jun 3
    Also glad to share this was accepted as a โœจspotlightโœจto the Trustworthy AI4GOOD Workshop @ #ICML 2026!
    user avatar
    Gabrielle Kaili-May Liu (Kaili)
    @pybeebee
    Jun 3
    ๐Ÿ”ฅExcited to share our paper: Quantifying Faithful Confidence Expression in Large Reasoning Models (LRMs)!๐Ÿ”ฅ We trust reasoning models partly because they show their work. But do their words reflect how confident they really are? ๐Ÿค” Check our preprint to find out! Details ๐Ÿงต๐Ÿ‘‡
    Image