fursovia/tcav_nlp
"Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)" paper implementation
This project helps data scientists and researchers understand why their natural language processing (NLP) models make certain predictions. You provide text data and a trained text classification model, along with specific words or phrases (concepts) you want to investigate. The tool then reveals how strongly these concepts influence the model's decisions for different categories, giving you interpretable insights beyond just overall accuracy.
No commits in the last 6 months.
Use this if you need to explain the reasoning behind your NLP model's classifications, particularly how specific real-world ideas or entities (like 'democracy' or 'Russia') impact its predictions.
Not ideal if you are looking for simple feature importance scores, or if you don't have a trained NLP text classification model and concept examples ready.
Stars
8
Forks
3
Language
Jupyter Notebook
License
—
Category
Last pushed
Mar 22, 2019
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/nlp/fursovia/tcav_nlp"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
rmovva/HypotheSAEs
HypotheSAEs: hypothesizing interpretable relationships in text datasets using sparse...
interpretml/interpret-text
A library that incorporates state-of-the-art explainers for text-based machine learning models...
fdalvi/NeuroX
A Python library that encapsulates various methods for neuron interpretation and analysis in...
jalammar/ecco
Explain, analyze, and visualize NLP language models. Ecco creates interactive visualizations...
alexdyysp/ESIM-pytorch
中国高校计算机大赛--大数据挑战赛