AQuA

The 'AQUA' dataset/benchmark is used to evaluate the mathematical reasoning capabilities of large language models (LLMs) by providing a set of problems that require formal verification through theorem provers.

Papers using AQuA (2)

Theorem Prover as a Judge for Synthetic Data Generation2025 · 1 citesLEMON: Learning Executable Multi-Agent Orchestration via Counterfactual Reinforcement Learning2026