New Filipino benchmark pairs show multilingual models choose sexist and anti-queer sentences measurably more often than neutral alternatives, with models trained on more Filipino text showing the strongest bias.
In Proceed- ings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 2887–2896, Singapore
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Filipino Benchmarks for Measuring Sexist and Homophobic Bias in Multilingual Language Models from Southeast Asia
New Filipino benchmark pairs show multilingual models choose sexist and anti-queer sentences measurably more often than neutral alternatives, with models trained on more Filipino text showing the strongest bias.