Want to create an interactive transcript for this episode?
Podcast: Arxiv Papers
Episode: [QA] Exploring the Hidden Reasoning Process of Large Language Models by Misleading Them
Description: The study investigates if LLMs/VLMs engage in abstract reasoning using Misleading Fine-Tuning, revealing their ability to apply contradictory rules in solving math problems, indicating internal abstraction mechanisms.https://arxiv.org/abs//2503.16401YouTube: https://www.youtube.com/@ArxivPapersTikTok: https://www.tiktok.com/@arxiv_papersApple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers