Intrinsic Bias is Predicted by Pretraining Data and Correlates with Downstream Performance in Vision-Language Encoders

Kshitish Ghate, Isaac Slaughter, Kyra Wilson, Mona T. Diab, Aylin Caliskan. Intrinsic Bias is Predicted by Pretraining Data and Correlates with Downstream Performance in Vision-Language Encoders. In Luis Chiruzzo, Alan Ritter, Lu Wang, editors, Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL 2025 - Volume 1: Long Papers, Albuquerque, New Mexico, USA, April 29 - May 4, 2025. pages 2899-2915, Association for Computational Linguistics, 2025. [doi]

Authors

Kshitish Ghate

This author has not been identified. Look up 'Kshitish Ghate' in Google

Isaac Slaughter

This author has not been identified. Look up 'Isaac Slaughter' in Google

Kyra Wilson

This author has not been identified. Look up 'Kyra Wilson' in Google

Mona T. Diab

This author has not been identified. Look up 'Mona T. Diab' in Google

Aylin Caliskan

This author has not been identified. Look up 'Aylin Caliskan' in Google