Why it matters
SAE shapes modern interpretability. Understanding shapes latest research.
Advertisement
The architecture
Train SAE on model activations.
Sparse + reconstructive.
Features often monosemantic.
Advertisement
How it works end to end
Bricken et al. 2023 Towards Monosemanticity.
Templeton et al. 2024 scaling SAEs.
Gated + Top-K SAE variants.