abirharrasse / auto-circuit_martian Public

forked from UFO-101/auto-circuit

Notifications You must be signed in to change notification settings
Fork 0
Star 0

A library for efficient patching and automatic circuit discovery.

ufo-101.github.io/auto-circuit

0 stars 13 forks Branches Tags Activity

Notifications

Name		Name	Last commit message	Last commit date
Latest commit History 102 Commits
.github		.github
.measurement_cache		.measurement_cache
.prune_scores_cache		.prune_scores_cache
.vscode		.vscode
auto_circuit		auto_circuit
datasets		datasets
docs		docs
experiments		experiments
tests		tests
.dockerignore		.dockerignore
.gitignore		.gitignore
.pre-commit-config.yaml		.pre-commit-config.yaml
Dockerfile		Dockerfile
README.md		README.md
Transformer Circuit Metrics are not Robust.pdf		Transformer Circuit Metrics are not Robust.pdf
__init__.py		__init__.py
mkdocs.yml		mkdocs.yml
poetry.lock		poetry.lock
poetry.toml		poetry.toml
pyproject.toml		pyproject.toml
setup.py		setup.py

Repository files navigation

AutoCircuit

A library for efficient patching and automatic circuit discovery.

Read the paper

Transformer Circuit Metrics are not Robust (Oral spotlight, COLM 2024)

Getting Started

pip install auto-circuit

Easy and Efficient Edge Patching

https://github.com/UFO-101/auto-circuit/blob/03ce552ccdea98fc6fed3b05cfb2df9b0eb4e323/experiments/demos/zero_ablate_an_edge.py#L20-L26

Different Ablation Methods

https://github.com/UFO-101/auto-circuit/blob/03ce552ccdea98fc6fed3b05cfb2df9b0eb4e323/experiments/demos/patch_an_edge.py#L37

Automatic Circuit Discovery

https://github.com/UFO-101/auto-circuit/blob/03ce552ccdea98fc6fed3b05cfb2df9b0eb4e323/experiments/demos/patch_an_edge.py#L49-L55

Visualization

https://github.com/UFO-101/auto-circuit/blob/5777057b445d55b7d4695eed8aface9d6bcefbe3/experiments/demos/patch_some_edges.py#L49

About

A library for efficient patching and automatic circuit discovery.

UFO-101.github.io/auto-circuit

Report repository

Releases

No releases published

Packages

No packages published

Languages

Python 99.7%
Dockerfile 0.3%