Skip to content

Latest commit

 

History

6 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

visARC

ARC puzzle visualizer for verifying test integrity and comparing AI responses


What it does

  • Validates ARC quiz files against full puzzle datasets
  • Visualizes AI model outputs in human-readable color grids with numbers
  • Highlights mismatched grids between quiz/full and AI/actual answers
  • Fully client-side – works offline or on GitHub Pages

How to Use

  1. Visit: https://g4te-keep3r.github.io/visARC/
  2. Upload your ARC quiz JSON and/or full puzzle JSON files
  3. (Optional) Paste AI-generated answer JSON to compare against correct solution

Puzzle Files

Under sample_puzzles are 3 random selections from ARC-AGI. Each is included in full and quiz form. Also included is a puzzle of my creation that has produced 11 different wrong answers acros ChatGPT and Claude with various models (GPT-5 and Opus 4.1 now included).


Local Use

Open index.html in your browser. No setup needed.


Example Usage

Puzzle Rendering (with Color + Numbers)

Puzzle with cell numbers Grids are rendered with ARC-standard colors and also display the cell values for easy debugging or annotation comparison.

Full Layout + Answer Comparison

Full puzzle view with wrong answer highlighted

This example shows:

  • Quiz puzzle (top) and Full puzzle with test output (bottom)
  • Submitted AI answer (bottom-right)
  • The red border highlights a mismatch between AI answer and correct answer

If any training examples differ between quiz and full files (aside from the quiz lacking a test output), they’ll also be shown with red outlines.


TODO (other than a list of UX improvements)


License

MIT

About

ARC puzzle visualizer for verifying test integrity and comparing AI responses. (Shall we play… a puzzle?)

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages