Reproducible tools and evidence for sparse autoencoder research
AI & ML interests
Building open foundations for understanding, evaluating, and intervening in neural networks.
Recent Activity
View all activity
Organization Card
Open research in mechanistic interpretability
From features
to testable
explanations.
Research tools for understanding how model features influence behavior.
SAEBase
One place to study what features do.
Capture activations, intervene on features, and evaluate the results in a reproducible workflow.
Open the research lab1Capture
2Intervene
3Evaluate
18Model–SAE configurations
48Evaluated layer sites
Open science
Research should be easy to examine. And easier to build on.
models 0
None public yet