You Never Know: Quantization Induces Inconsistent Biases in Vision-Language Foundation Models

Eric Slyman, Anirudh Kanneganti, Sanghyun Hong, Stefan Lee
10/26/2024
cs.CVcs.CYcs.LG

No confident code implementation found for this paper. 2 possible matches can be shown below.

Abstract

We study the impact of a standard practice in compressing foundation vision-language models - quantization - on the models' ability to produce socially-fair outputs. In contrast to prior findings with unimodal models that compression consistently amplifies social biases, our extensive evaluation of four quantization settings across three datasets and three CLIP variants yields a surprising result: while individual models demonstrate bias, we find no consistent change in bias magnitude or direction across a population of compressed models due to quantization.

View on arXivView PDF

Code Implementations

No confident code match yet

We couldn't find an author-owned or strongly-evidenced community implementation for this paper. 2 weaker matches are hidden by default — verify before relying on them.

No code implementations found yet.

Know of an implementation? Let us know in the comments below!

Similar Papers

Uncertainty Drives Social Bias Changes in Quantized Large Language Models

87% match2026

How Quantization Shapes Bias in Large Language Models

86% match1 repo72025

Data Matters Most: Auditing Social Bias in Contrastive Vision Language Models

84% match2025

Images Speak Louder than Words: Understanding and Mitigating Bias in Vision-Language Model from a Causal Mediation Perspective

84% match2024

Biases Propagate in Encoder-based Vision-Language Models: A Systematic Analysis From Intrinsic Measures to Zero-shot Retrieval Outcomes

84% match2025

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

83% match1 repo2025

Cite this paper

@article{slyman2024never,
  title  = {You Never Know: Quantization Induces Inconsistent Biases in Vision-Language Foundation Models},
  author = {Eric Slyman and Anirudh Kanneganti and Sanghyun Hong and Stefan Lee},
  year   = {2024},
  eprint = {2410.20265},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2410.20265v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2410.20265.svg)](https://codeofpaper.com/paper/2410.20265)

Discussion