lm-sys / lm-sys/FastChat

Incorrect Voting in Image Battle Mode of LMArena - Request for Correction or Feedback Adjustment

Open
#3,741 1 comment 1 reaction 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

Hello, team!

In my first and third rounds of Battle Mode, I wrongly voted for "both images are bad" because I initially thought the generations didn't have the girl's head. After enlarging the images, I realized that:

  • Round 1: the image on the right was better;
  • Round 3: the image on the right was excellent, and the image on the left was perfect (fully matched the prompt and upload).

Unfortunately, I was unable to undo my votes.

I request that, if possible, these two votes be corrected or disregarded from any classification/aggregation/training.

I also suggest implementing a reminder for users to enlarge images before voting. This would help prevent similar errors.

The images mentioned are attached.

Thank you for your incredible work, and I appreciate your efforts!

Image Image Image

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. First locate the LMArena Battle Mode vote-handling and image-comparison entry points, then determine whether the two submitted votes can be corrected or excluded and where a zoom reminder belongs. Done means the correction or reminder scope is agreed and verified against the voting workflow.

Written by the indexing model from the issue text.

Assessment

Domain
data, frontend
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.