TunsayBench is a human-reviewed, multi-task benchmark for evaluating Khmer
language models across commonsense reasoning, general knowledge, natural
language inference, cloze completion, and machine reading comprehension. It
was created by Mengsay Loem as part of Tunsay Labs.
The public v0.2.0 release contains only items accepted during human review.
