Poster
in
Workshop: 2nd AI for Math Workshop @ ICML 2025
Direct Induction Proof Challenge: Evaluating Large Language Models on Deeply Nested Mathematical Induction
Risako Ando · Koji Mineshima · Mitsuhiro Okada
Abstract:
We introduce a challenge designed to evaluate the capability of Large Language Models (LLMs) in performing mathematical induction proofs, with a particular focus on nested induction.Our task requires models to construct direct induction proofs in both formal and informal settings, without relying on any preexisting lemmas. Experimental results indicate that current models struggle with generating direct induction proofs, suggesting that there remains significant room for improvement.
Chat is not available.