PySpect

Home

lists

Frequently asked questions

© 2025 PySpect

Package profile

sae-lens

  • Summary: Training and Analyzing Sparse Autoencoders (SAEs)
  • Author: Joseph Bloom
  • Homepage: https://jbloomaus.github.io/SAELens
  • Source: https://github.com/jbloomAus/SAELens (Repo profile)
  • Number of releases: 145
  • First release: 0.1.0 on 2024-04-06
  • Latest release: 5.11.0 on 2025-07-12

Releases

Dates and sizes of releasesJulyOctober2025AprilJulyRelease Date0.060.080.100.120.14Size in MB

PyPI Downloads

Weekly downloads over the last 3 monthsFebruaryMarchAprilMayJuneDate0123456789 thousand downloads per week

Dependencies

Sae-lens has 20 dependencies, 1 of which optional.
Dependencies of sae-lens (20).
DependencyOptional
automated-interpretabilityfalse
babefalse
datasetsfalse
matplotlibfalse
matplotlib-inlinefalse
nltkfalse
plotlyfalse
plotly-expressfalse
pytest-profilingfalse
python-dotenvfalse
pyyamlfalse
pyzmqfalse
safetensorsfalse
simple-parsingfalse
transformer-lensfalse
transformersfalse
typerfalse
typing-extensionsfalse
zstandardfalse
mamba-lenstrue

Details