yuhui-zh15/drml

Official Code Release for "Diagnosing and Rectifying Vision Models using Language" (ICLR 2023)

15
/ 100
Experimental

This project helps machine learning engineers and researchers understand why their computer vision models make mistakes. It takes an existing vision model and uses natural language descriptions to pinpoint specific types of images or situations where the model fails. The output identifies problematic data categories and suggests ways to fix these errors, all without needing to collect or label more visual data.

No commits in the last 6 months.

Use this if you need to quickly diagnose and understand the failure modes of your image classification models using natural language, rather than manually sifting through images.

Not ideal if your primary goal is to train a new vision model from scratch or if you don't have access to multi-modal language-vision embeddings.

computer-vision model-debugging ML-diagnostics model-rectification AI-explainability
No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 7 / 25
Maturity 8 / 25
Community 0 / 25

How are scores calculated?

Stars

34

Forks

Language

Jupyter Notebook

License

Last pushed

Jun 08, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/yuhui-zh15/drml"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.