Think-to-Talk or Talk-to-Think? When LLMs Come Up with an Answer in Multi-Step Reasoning

Autor:	Kudo, Keito, Aoki, Yoichi, Kuribayashi, Tatsuki, Sone, Shusaku, Taniguchi, Masaya, Brassard, Ana, Sakaguchi, Keisuke, Inui, Kentaro
Rok vydání:	2024
Předmět:	Computer Science - Computation and Language
Druh dokumentu:	Working Paper
Popis:	This study investigates the internal reasoning mechanism of language models during symbolic multi-step reasoning, motivated by the question of whether chain-of-thought (CoT) outputs are faithful to the model's internals. Specifically, we inspect when they internally determine their answers, particularly before or after CoT begins, to determine whether models follow a post-hoc "think-to-talk" mode or a step-by-step "talk-to-think" mode of explanation. Through causal probing experiments in controlled arithmetic reasoning tasks, we found systematic internal reasoning patterns across models; for example, simple subproblems are solved before CoT begins, and more complicated multi-hop calculations are performed during CoT.
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2412.01113 Zobrazit plný text záznamu View this record from Arxiv