ALL INTELLIGENCE
Hugging Face BlogT1·Models

NPHardEval Leaderboard Launches for LLM Reasoning Assessment

Original title: NPHardEval Leaderboard: Unveiling the Reasoning Abilities of Large Language Models through Complexity Classes and Dynamic Updates
45AI Score

The NPHardEval leaderboard introduces a dynamic benchmark using computational complexity classes (e.g., NP-Hard) to assess LLM reasoning limits. It aims to overcome overfitting in static benchmarks by providing a more realistic capability profile through continuous updates.

Full Report

NPHardEval Leaderboard: Unveiling the Reasoning Abilities of Large Language Models through Complexity Classes and Dynamic Updates