HMMT-25

The 'HMMT-25' dataset is a benchmark used to evaluate the reasoning capabilities of Large Language Models (LLMs) by analyzing their confidence dynamics during the inference process.

Papers using HMMT-25 (3)

Inference Time Optimization with Confidence Dynamics2026Chronos: Learning Temporal Dynamics of Reasoning Chains for Test-Time Scaling2026PromptCoT 2.0: Scaling Prompt Synthesis for Large Language Model Reasoning2025