dataset
active
dataset:longbench-v2

LongBench v2

Source of 24K–32K-token documents for long-context tests.

Neighborhood — ranked by edge-count