zjunlp/KnowledgeCircuits
[NeurIPS 2024] Knowledge Circuits in Pretrained Transformers
This project helps AI researchers and developers understand how large language models (LLMs) store and use specific pieces of knowledge, like country-capital relationships. By identifying the "knowledge circuits" within these complex models, you can gain insights into their internal workings. It takes a pretrained language model and a specific type of knowledge as input, then outputs a visual representation of the circuit responsible for that knowledge.
164 stars.
Use this if you are a researcher or developer working with large language models and need to investigate which internal components are responsible for specific factual knowledge within them.
Not ideal if you are an end-user simply looking to apply LLMs for tasks like content generation or data analysis, as this is a research tool for model introspection.
Stars
164
Forks
11
Language
Python
License
MIT
Category
Last pushed
Nov 14, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/zjunlp/KnowledgeCircuits"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
steering-vectors/steering-vectors
Steering vectors for transformer language models in Pytorch / Huggingface
jianghoucheng/AlphaEdit
AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models, ICLR 2025 (Outstanding Paper)
kmeng01/memit
Mass-editing thousands of facts into a transformer memory (ICLR 2023)
boyiwei/alignment-attribution-code
[ICML 2024] Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications
jianghoucheng/AnyEdit
AnyEdit: Edit Any Knowledge Encoded in Language Models, ICML 2025