lm-sys / lm-sys/FastChat

Excessive False Positives in Image Generation (Works in ChatGPT, fails in Lmarena)

Open
#3,768 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
39.5k
Forks
4.8k
PR merge metrics
No merged PRs in 30d

Description

Image The image generation safety filter is broken. It's blocking harmless poses (hands on hips) and standard anime character designs that previously worked. It is making the tool unusable due to false positives.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

The issue names no files, tests, or entry points, so first locate the image-generation safety filter and reproduce the reported false positives for hands-on-hips poses and standard anime designs. Done means harmless prompts that previously worked are no longer blocked, while genuinely unsafe requests remain filtered.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai, security
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.