tw-formosa-bench
A difficulty-graded evaluation benchmark generated from community corpora and focused on contemporary Taiwanese daily life. Its questions are filtered through a frozen model pool to control the difficulty window.
Timeline (last 90 days)
OpenTWBench is an open evaluation suite built around two benchmarks: tw-exam-bench and tw-formosa-bench.
tw-formosa-bench is generated from community-curated corpus, containing 1,696 difficulty-graded questions about contemporary Taiwanese daily life, and no model participates in writing the reference answers.
The question set of tw-formosa-bench was filtered against a frozen pool of seven models so that their scores were kept at the passing margin (46.9%–74.6%, median about 54%).
Within six weeks after tw-formosa-bench was released, frontier models broke through its calibrated difficulty window; on the public leaderboard, 22 of 108 models exceeded the benchmark's ceiling, with the best reaching 91.6%.
All Facts (4)
OpenTWBench is an open evaluation suite built around two benchmarks: tw-exam-bench and tw-formosa-bench.
50%Unverifiedtw-formosa-bench is generated from community-curated corpus, containing 1,696 difficulty-graded questions about contemporary Taiwanese daily life, and no model participates in writing the reference answers.
50%UnverifiedThe question set of tw-formosa-bench was filtered against a frozen pool of seven models so that their scores were kept at the passing margin (46.9%–74.6%, median about 54%).
50%UnverifiedWithin six weeks after tw-formosa-bench was released, frontier models broke through its calibrated difficulty window; on the public leaderboard, 22 of 108 models exceeded the benchmark's ceiling, with the best reaching 91.6%.
50%