SEA-NLI benchmark evaluates LLMs on Southeast Asian culture

Arxiv CS.CL (NLP/language models)

SEA-NLI is a new benchmark for evaluating cultural understanding in LLMs via natural language inference, focusing on Southeast Asian contexts. Existing NLI benchmarks are largely Western-centric, limiting assessment of underrepresented cultures.

Arxiv CS.CL (NLP/language models)·Peerawat ChomphooyodJian Gang NguiYosephine SusantoAttapol T. RutherfordAlham Fikri AjiSarana NutanongCan UdomcharoenchaikitPeerat Limkonchotiwat

8 days ago