LongBench

eval
Z.ai
Comprehensive benchmark for long-context understanding in LLMs.
GitHubPaper (arXiv)HuggingFace

Dataset

GitHub
benchmarkscaling