{"slug":"saelens","name":"Saelens","kind":"tool","deliveryMethods":["tool"],"kindLabel":"Tool","grade":"A","score":87,"tagline":"Training Sparse Autoencoders on Language Models","description":"SAELens inference works with any PyTorch-based model, not just TransformerLens. While we provide deep integration with TransformerLens via `HookedSAETransformer`, SAEs can be used with Hugging Face Transformers, NNsight, or any other framework by extracting activations and passing them to the SAE's `encode()` and `decode()` methods.","tags":["interpretability","sparse-autoencoders","autoencoders","transformerlens","pytorch"],"url":"https://way.space/tool/saelens","stars":"—","category":"training","categories":["training"],"repo":"https://github.com/decoderesearch/saelens","website":"https://decoderesearch.github.io/SAELens/","license":"MIT","maintainers":[{"login":"decoderesearch","url":"https://github.com/decoderesearch","isMember":false,"verified":false},{"login":"chanind","url":"https://github.com/chanind","isMember":false,"verified":false},{"login":"hijohnnylin","url":"https://github.com/hijohnnylin","isMember":false,"verified":false},{"login":"danra","url":"https://github.com/danra","isMember":false,"verified":false}],"scoreWhy":[{"label":"Contributors & bus-factor","detail":"24 recent contributors","direction":"positive"},{"label":"Commit recency","detail":"last commit 2d ago","direction":"positive"},{"label":"Commit cadence","detail":"80 commits in 90 days","direction":"positive"},{"label":"Issue close ratio","detail":"79% of issues closed","direction":"positive"},{"label":"Release cadence","detail":"100 releases in 12 months","direction":"positive"},{"label":"Stars (log-damped)","detail":"1,568 stars (log-damped)","direction":"positive"},{"label":"Star authenticity","detail":"star-velocity check unavailable — GitHub no longer publishes stargazer timestamps for this repo","direction":"neutral"},{"label":"Engagement-to-star ratio","detail":"healthy engagement (0.69 per star)","direction":"positive"}]}