# Runway — Mitigating stereotypical biases in text to image generative systems

- Company: Runway (runway.com)
- Announced: 2023-10-10T22:24:14.964000+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://runway.com/research/mitigating-stereotypical-biases-in-text-to-image-generative-systems
- Record: https://forck.live/items/7934-mitigating-stereotypical-biases-in-text-to-image-generative-systems
- Subject: Gen / Aleph

Runway proposes a method to mitigate social biases in text-to-image generative models by finetuning on synthetic data that varies in perceived skin tones and genders. Their diversity finetuned (DFT) model improves group fairness by 150% for perceived skin tone and 97.7% for perceived gender. The company will release all text prompts and code to generate training images.

## Evidence

Verbatim from https://runway.com/research/mitigating-stereotypical-biases-in-text-to-image-generative-systems:

> In this work, we propose a method to mitigate such biases and ensure that the outcomes are fair across different groups of people. We do this by finetuning text-to-image models on synthetic data that varies in perceived skin tones and genders constructed from diverse text prompts.

---

Record: https://forck.live/items/7934-mitigating-stereotypical-biases-in-text-to-image-generative-systems
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md