Large Language Models Can Learn Temporal Reasoning

arXiv:2401.06853v2 Announce Type: replace
Abstract: While large language models (LLMs) have demonstrated remarkable reasoning capabilities, they are not without their flaws and inaccuracies. Recent studies have introduced various methods to mitigate these limitations. Temporal reasoning (TR), in particular, presents a significant challenge for LLMs due to its reliance on diverse temporal expressions and intricate contextual details. In this paper, we propose TG-LLM, a new framework towards language-based TR. To be specific, we first teach LLM to translate the context into …

abstract arxiv capabilities challenge cs.cl diverse flaws language language models large language large language models learn limitations llms reasoning reliance studies temporal type