Skip to content

saev

saev

Initializing search

Home
Users
Developers
API

saev

Home
Users
Users
- Guide
- Inference
- Glossary
- Sweeps
Developers
Developers
API
API
- saev saev
  Table of contents
  - saev
- saev.colors
- saev.configs
- saev.data
- saev.data.buffers
- saev.data.clip
- saev.data.datasets
- saev.data.dinov2
- saev.data.dinov3
- saev.data.fake_clip
- saev.data.indexed
- saev.data.models
- saev.data.ordered
- saev.data.shards
- saev.data.shuffled
- saev.data.siglip
- saev.data.transforms
- saev.disk
- saev.framework
- saev.framework.inference
- saev.framework.shards
- saev.helpers
- saev.nn
- saev.nn.modeling
- saev.nn.objectives
- saev.utils
- saev.utils.scheduling
- saev.utils.statistics
- saev.utils.wandb
- saev.viz

saev

saev is a Python package for training sparse autoencoders (SAEs) on vision transformers (ViTs) in PyTorch.

Made with Material for MkDocs