SAE Lens
Sparse-autoencoder research tooling for examining reconstruction, sparsity and feature interventions on neural activations.
External resource. No execution or independent verification is claimed here.
How does a fixed sparsity setting trade reconstruction error against active features on an authored low-dimensional activation set?
Prespecify the objective and active-feature definition; independently recompute reconstruction and sparsity metrics from saved tensors.
What you could produce
- A version-pinned protocol stating inputs, rights, expected behavior, tolerances and resource limits before execution.
- A retained per-case result and failure ledger with an independently controlled comparison, if qualified execution is later authorized.
Before you use it
- Python and tensor-training dependencies; real activation collections, pretrained autoencoders and base weights have separate rights and GPU needs.
- A separately qualified isolated runtime with a reviewed, pinned dependency and input closure.
Limits to keep in view
- No source program, example, build hook, package, dataset, model or generated research code has been executed or downloaded as a payload.
- The documented self-contained Python 3.13, 90-second pilot does not establish support for this package, its compiled dependencies, GPUs, services or agent sandboxes.
- Installation success, scientific outcomes, runtime compatibility, latency, memory use, API costs and security properties are unmeasured.
Source and permission context
Preserve the upstream project name, repository, exact source commit and applicable contributor/notices; resolve upstream citation guidance for any later formal use.
Catalog listing reviewed. This review covers the description and source links displayed here.
Approved for catalog metadata and links. The displayed entry identifies SAE Lens with the concise collection-authored summary “Sparse-autoencoder research tooling for examining reconstruction, sparsity and feature interventions on neural activations.” and points to the public upstream repository https://github.com/jbloomAus/SAELens. The wording describes function and possible investigation without reproducing upstream source or documentation, claiming execution, or implying endorsement or rights beyond the recorded scopes.
Reviewed 2026-09-14. Copying or adapting source files remains subject to their own terms.
code · MIT
The pinned root file contains the MIT permission grant, notice condition and warranty disclaimer. Observation is limited to LICENSE at commit b5711e34d072846dc112881f4c8d209407a32b56; this is not blanket artifact clearance.
Inspect the license evidence ↗Before copying source material
- Only the cited license and README texts were observed; file exceptions, dependency closure, vendored components and submodules are not comprehensively audited.
- Dataset files, task prompts, generated outputs, model weights, tokenizer assets and hosted APIs are not cleared by a root code license.
- README/documentation reuse rights and version-specific citation guidance remain separately unresolved; only links and original descriptions are retained.