Shaping the Future with AI: Research, Education, and Community Impact.

Open research data

EquiText-Nepali: A Dataset for Gender, Race, and Sociocultural Bias in Nepali Text

EquiText-Nepali is a curated dataset designed to evaluate and expose social biases, specifically related to gender, race, and sociocultural or religious identity within Nepali-language texts. It aims to support the development of fair and inclusive language …

Dataset notes

EquiText-Nepali is a curated dataset designed to evaluate and expose social biases, specifically related to genderrace, and sociocultural or religious identity within Nepali-language texts. It aims to support the development of fair and inclusive language technologies by offering annotated examples for bias detection, masked language model probing, and fairness benchmarking.

Related publications

Dual-Metric Evaluation of Social Bias in Large Language Models: Evidence from an Underrepresented Nepali Cultural Context →

Search CAIR-Nepal

What are you looking for?

ESC