Benchmark Almanac

Koji Mineshima

We track 2 benchmarks that Koji Mineshima made or helped make.

0 in use 2 with public code 1 with a leaderboard here

Benchmarks

Name Kind Released Best score Status
JSICK Models judge entailment and semantic relatedness in Japanese sentence pairs. An additional stress test examines sensitivity to word order and case marking. Reasoning in many languages May 2021 89.4% Not in cards
JaNLI Models judge entailment between Japanese sentences designed to expose shortcuts in language inference. Reasoning in many languages Nov 2021 Not in cards