BenchmarkCross-lingual Natural Language Inference
XNLI
A multilingual natural language inference benchmark used to evaluate cross-lingual understanding capabilities of models. It covers multiple languages but does not distinguish regional variants within individual languages.
Timeline (last 90 days)
Oct 11
Existing evaluation benchmarks such as XNLI and FLORES-200 mostly operate at the granularity of language codes and do not distinguish regional variants within the same language code.
Unverified50%