Xi Ye

Learning Composable Chains-of-Thought

2025 · arXiv (Cornell University)

A common approach for teaching large language models (LLMs) to reason is to train on chain-of-thought (CoT) traces of in-distribution reasoning problems, but such annotated data is costly to obtain for every problem of interest. …

RNG-KBQA: Generation Augmented Iterative Ranking for Knowledge Base Question Answering

2022 · Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

Existing KBQA approaches, despite achieving strong performance on i.i.d. test data, often struggle in generalizing to questions involving unseen KB schema items. Prior rankingbased approaches have shown some success in generalization, but suffer from the …