evandez/neuron-descriptions
Natural Language Descriptions of Deep Visual Features, ICLR 2022
This project, MILAN, helps visual AI developers and researchers understand the 'concepts' individual neurons in their deep learning models have learned. It takes a computer vision model and its top-activating image regions for specific neurons as input. The output is a natural language description, like 'detects furry animals' or 'responds to building edges,' for each neuron, which previously required tedious manual analysis.
No commits in the last 6 months.
Use this if you need to quickly and automatically generate clear, specific natural language explanations for what individual neurons in your image classification or generative AI models are doing.
Not ideal if your models are not based on PyTorch or if your focus is on non-visual deep learning tasks, as this tool is specifically designed for computer vision models.
Stars
65
Forks
7
Language
Python
License
MIT
Last pushed
Jun 29, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/evandez/neuron-descriptions"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
obss/sahi
Framework agnostic sliced/tiled inference + interactive ui + error analysis plots
tensorflow/tcav
Code for the TCAV ML interpretability project
MAIF/shapash
🔅 Shapash: User-friendly Explainability and Interpretability to Develop Reliable and Transparent...
TeamHG-Memex/eli5
A library for debugging/inspecting machine learning classifiers and explaining their predictions
csinva/imodels
Interpretable ML package 🔍 for concise, transparent, and accurate predictive modeling...