Skip to content

benchmark

HRM8K

Korean and English benchmark suite designed to separate a model's reasoning ability from its comprehension of non-English input.

Current clusters