The official repository of "LooGLE v2: Are LLMs Ready for Real World Long Dependency Challenges?"
NeurIPS DB Track 2025
📋 Overview
LooGLE v2 is a comprehensive benchmark designed to evaluate large language models on their ability to understand and process long-context documents with complex dependencies. The benchmark covers diverse domains including Finance, Law, Code, and Game.
🚀 Quick Start
📦… See the full description on the dataset page: https://huggingface.co/datasets/MuLabPKU/LooGLE-v2.
