Gated SAEs decouple which features to use from how large their activations should be, applying the L1 penalty only to selection and thereby eliminating shrinkage while halving the number of firing features needed for good fidelity.
Curve circuits
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 3verdicts
UNVERDICTED 3roles
background 1polarities
background 1representative citing papers
Gradient-informed placement of LoRA parameters recovers full performance under GRPO while random placement does not, due to differences in gradient rank and stability across training regimes.
Sparse autoencoders applied to language model activations yield more interpretable and monosemantic features than alternative approaches, enabling finer causal analysis on the indirect object identification task.
citing papers explorer
-
Improving Dictionary Learning with Gated Sparse Autoencoders
Gated SAEs decouple which features to use from how large their activations should be, applying the L1 penalty only to selection and thereby eliminating shrinkage while halving the number of firing features needed for good fidelity.
-
Not How Many, But Which: Parameter Placement in Low-Rank Adaptation
Gradient-informed placement of LoRA parameters recovers full performance under GRPO while random placement does not, due to differences in gradient rank and stability across training regimes.
-
Sparse Autoencoders Find Highly Interpretable Features in Language Models
Sparse autoencoders applied to language model activations yield more interpretable and monosemantic features than alternative approaches, enabling finer causal analysis on the indirect object identification task.