lkopf/prism
[NeurIPS 2025] PRISM is a multi-concept feature description framework which can identify and score polysemantic features.
When analyzing how AI models make decisions, it can be hard to understand why certain features activate. This tool helps you go beyond single explanations for these features, providing multiple, human-understandable descriptions that capture the full complexity of what an AI feature represents. It takes model activations and outputs clear, clustered descriptions of the concepts the model is detecting. AI researchers and practitioners focused on model interpretability or explainable AI would use this.
No commits in the last 6 months.
Use this if you need to deeply understand the multiple concepts an AI model's internal features are responding to, moving beyond simplistic, single explanations.
Not ideal if you are looking for a simple, single-word explanation for every AI feature or if your primary goal is basic performance evaluation rather than interpretability.
Stars
8
Forks
1
Language
Jupyter Notebook
License
—
Category
Last pushed
Aug 21, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/lkopf/prism"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
opentensor/bittensor
Internet-scale Neural Networks
trailofbits/fickling
A Python pickling decompiler and static analyzer
benchopt/benchopt
A framework for reproducible, comparable benchmarks
BiomedSciAI/fuse-med-ml
A python framework accelerating ML based discovery in the medical field by encouraging code...
mosaicml/streaming
A Data Streaming Library for Efficient Neural Network Training