Pith. sign in

REVIEW 37 cited by

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2412.06410 v1 pith:DJMEJBMP submitted 2024-12-09 cs.LG cs.AIstat.ML

BatchTopK Sparse Autoencoders

classification cs.LG cs.AIstat.ML
keywords batchtopksaeslatentsactivationsnumbersamplesparsetopk
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Sparse autoencoders (SAEs) have emerged as a powerful tool for interpreting language model activations by decomposing them into sparse, interpretable features. A popular approach is the TopK SAE, that uses a fixed number of the most active latents per sample to reconstruct the model activations. We introduce BatchTopK SAEs, a training method that improves upon TopK SAEs by relaxing the top-k constraint to the batch-level, allowing for a variable number of latents to be active per sample. As a result, BatchTopK adaptively allocates more or fewer latents depending on the sample, improving reconstruction without sacrificing average sparsity. We show that BatchTopK SAEs consistently outperform TopK SAEs in reconstructing activations from GPT-2 Small and Gemma 2 2B, and achieve comparable performance to state-of-the-art JumpReLU SAEs. However, an advantage of BatchTopK is that the average number of latents can be directly specified, rather than approximately tuned through a costly hyperparameter sweep. We provide code for training and evaluating BatchTopK SAEs at https://github.com/bartbussmann/BatchTopK

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 37 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Expander Sparse Autoencoders: Parameter-Efficient Dictionaries for Mechanistic Interpretability

    cs.LG 2026-07 conditional novelty 8.0

    Expander SAEs apply left-d-regular expander masks to TopK SAEs, learning only dn decoder parameters instead of mn and tracing a storage-fidelity frontier that reaches 293x compression with 84% retained performance on ...

  2. WriteSAE: Sparse Autoencoders for Recurrent State

    cs.LG 2026-05 unverdicted novelty 8.0

    WriteSAE is the first sparse autoencoder that factors decoder atoms into the native d_k x d_v cache write shape of recurrent models and supplies a closed-form per-token logit shift for atom substitution.

  3. WriteSAE: Sparse Autoencoders for Recurrent State

    cs.LG 2026-05 unverdicted novelty 8.0

    WriteSAE introduces sparse autoencoders with rank-1 matrix atoms for recurrent state updates, allowing replacement tests that outperform deletion on 92.4% of positions and a formula predicting logit changes with R²=0.98.

  4. WriteSAE: Sparse Autoencoders for Recurrent State

    cs.LG 2026-05 unverdicted novelty 8.0

    WriteSAE decomposes recurrent model cache writes into substitutable atoms with a closed-form logit shift, achieving high substitution success and targeted behavioral installs on models like Qwen3.5 and Mamba-2.

  5. Turn-Averaged SAEs for Feature Discovery and Long-Context Attribution

    cs.CL 2026-06 unverdicted novelty 7.0

    Turn-averaged SAEs reconstruct average activations over conversation turns to represent high-level turn characteristics with a fixed number of features, simplifying long-context interpretability compared to per-token SAEs.

  6. PairSAE: Mechanistic Interpretability from Pair Representations in Protein Co-Folding

    cs.LG 2026-06 unverdicted novelty 7.0

    Introduces PairSAE, a sparse autoencoder for pair representations in structural biology foundation models that produces features aligned with UniProt annotations and affinity predictions.

  7. Beyond the Hard Budget: Sparsity Regularizers for More Interpretable Top-k Sparse Autoencoders

    cs.LG 2026-06 unverdicted novelty 7.0

    Sparsity regularizers applied before Top-k selection in SAEs improve monosemanticity and make reconstruction robust to inference-time k across vision models and datasets.

  8. Size Doesn't Matter: Cosine-Scored Sparse Autoencoders

    cs.LG 2026-06 unverdicted novelty 7.0

    Cosine-scored SAEs with a learned direction-magnitude blend learn more concept-aligned features than standard inner-product SAEs at matched reconstruction quality.

  9. Rational Sparse Autoencoder

    cs.LG 2026-06 unverdicted novelty 7.0

    RSAE replaces fixed SAE encoder activations (ReLU, JumpReLU, TopK) with trainable rational functions, initialized from baselines and fine-tuned to improve reconstruction and downstream metrics on language-model residu...

  10. VFUSE: Virulent Feature Understanding with Sparse autoEncoders

    cs.LG 2026-06 unverdicted novelty 7.0

    VFUSE applies sparse autoencoders to diffusion-transformer activations in RoseTTAFold3 and RFDiffusion3 to find monosemantic features that detect hazardous protein designs with AUROC up to 0.84.

  11. Subspace-Aware Sparse Autoencoders for Effective Mechanistic Interpretability

    cs.LG 2026-06 conditional novelty 7.0

    SASA replaces single-vector decoders in SAEs with learned subspaces plus block sparsity and nuclear-norm regularization, proving that a single group becomes the global minimizer once block size meets intrinsic dimensi...

  12. Sign-Aware Gated Sparse Autoencoders: Modeling Anticorrelated Features with Bi-Jump-ReLU Activations

    cs.LG 2026-05 conditional novelty 7.0

    SA-GSAE with Bi-Jump-ReLU enables one latent to encode both polarities of anticorrelated features, Pareto-dominating or matching full-width gated SAEs while reducing dead latents by up to 500x on some LLM hookpoints.

  13. Conceptualizing Embeddings: Sparse Disentanglement for Vision-Language Models

    cs.CV 2026-05 unverdicted novelty 7.0

    CEDAR learns an invertible rotation of vision-language embeddings to concentrate semantics into sparse, axis-aligned coordinates for improved interpretability.

  14. WriteSAE: Sparse Autoencoders for Recurrent State

    cs.LG 2026-05 unverdicted novelty 7.0

    WriteSAE factors sparse autoencoder decoder atoms to the native d_k x d_v cache write shape in recurrent models, provides a closed-form logit shift, and demonstrates high success in atom substitution and behavioral ed...

  15. Disentangled Sparse Representations for Concept-Separated Diffusion Unlearning

    cs.LG 2026-05 unverdicted novelty 7.0

    SAEParate disentangles sparse representations in diffusion models via contrastive clustering and nonlinear encoding to enable more precise concept unlearning with reduced side effects.

  16. SoftSAE: Dynamic Top-K Selection for Adaptive Sparse Autoencoders

    cs.LG 2026-05 unverdicted novelty 7.0

    SoftSAE introduces a dynamic top-k selection mechanism in sparse autoencoders that learns an input-dependent sparsity level via a differentiable soft top-k operator.

  17. Beyond Semantics: Disentangling Information Scope in Sparse Autoencoders for CLIP

    cs.CV 2026-04 unverdicted novelty 7.0

    The paper proposes information scope as a new interpretability axis for SAE features in CLIP and introduces the Contextual Dependency Score to separate local from global scope features, showing they influence model pr...

  18. Steering Vision-Language Models with Joint Sparse Autoencoders

    cs.CV 2026-06 unverdicted novelty 6.0

    JSAE jointly factorizes pooled vision and language activations in VLMs into aligned interpretable features, revealing layer-dependent asymmetry in additive steering versus suppression on three models.

  19. ICA Lens: Interpreting Language Models Without Training Another Dictionary

    cs.LG 2026-06 unverdicted novelty 6.0

    ICALens applies an optimized ICA workflow to LLM activations and recovers compact interpretable directions that match or exceed public SAEs on SAEBench probing and perturbation tasks without per-layer dictionary training.

  20. Interactions Between Crosscoder Features: A Compact Proofs Perspective

    cs.LG 2026-06 unverdicted novelty 6.0

    Derives an interaction measure between crosscoder features from reconstruction error in compact proofs and applies it to produce computationally sparse crosscoders retaining 60% MLP performance with single-feature sel...

  21. Shared Latent Structures Enable Unified Backdoor Detection and Mitigation in LLMs

    cs.AI 2026-06 unverdicted novelty 6.0

    Sparse autoencoders identify shared latent features across diverse backdoor attacks in LLMs that enable unified detection via classifiers, causal control via steering, and mitigation via ablation fine-tuning.

  22. Are Sparse Autoencoder Benchmarks Reliable?

    cs.LG 2026-05 unverdicted novelty 6.0

    An audit of SAEBench reveals that Targeted Probe Perturbation and Spurious Correlation Removal metrics fail reliability tests and should not be used to evaluate sparse autoencoders.

  23. The Rate-Distortion-Polysemanticity Tradeoff in SAEs

    cs.LG 2026-05 unverdicted novelty 6.0

    SAEs exhibit a rate-distortion-polysemanticity tradeoff where monosemanticity increases rate and distortion, with optimal polysemanticity set by feature co-occurrence probabilities in the data.

  24. Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces

    cs.LG 2026-05 unverdicted novelty 6.0

    A latent mediation framework with sparse autoencoders enables non-additive token-level influence attribution in LLMs by learning orthogonal features and back-propagating attributions.

  25. Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space

    cs.CL 2026-05 unverdicted novelty 6.0

    LLMs perform in-context learning as trajectories through a structured low-dimensional conceptual belief space, with the structure visible in both behavior and internal representations and causally manipulable via inte...

  26. Tree SAE: Learning Hierarchical Feature Structures in Sparse Autoencoders

    cs.LG 2026-05 unverdicted novelty 6.0

    Tree SAE learns hierarchical feature pairs in sparse autoencoders by combining activation coverage with a new reconstruction condition, outperforming prior methods on hierarchy detection while remaining competitive on...

  27. Tree SAE: Learning Hierarchical Feature Structures in Sparse Autoencoders

    cs.LG 2026-05 unverdicted novelty 6.0

    Tree SAE learns hierarchical feature structures by combining activation coverage with a new reconstruction condition, outperforming prior SAEs on hierarchical pair detection while matching state-of-the-art benchmark p...

  28. SoftSAE: Dynamic Top-K Selection for Adaptive Sparse Autoencoders

    cs.LG 2026-05 unverdicted novelty 6.0

    SoftSAE replaces fixed-K sparsity in autoencoders with a learned, input-dependent number of active features via a soft top-k operator.

  29. Feature Starvation as Geometric Instability in Sparse Autoencoders

    cs.LG 2026-05 unverdicted novelty 6.0

    Adaptive elastic net SAEs (AEN-SAEs) mitigate feature starvation in SAEs by combining ℓ2 structural stability with adaptive ℓ1 reweighting, producing a Lipschitz-continuous sparse coding map that recovers global featu...

  30. Dictionary-Aligned Concept Control for Safeguarding Multimodal LLMs

    cs.LG 2026-04 unverdicted novelty 6.0

    DACO curates a 15,000-concept dictionary from 400K image-caption pairs and uses it to initialize an SAE that enables granular, concept-specific steering of MLLM activations, raising safety scores on MM-SafetyBench and...

  31. Improving Robustness In Sparse Autoencoders via Masked Regularization

    cs.LG 2026-04 unverdicted novelty 6.0

    Masked regularization in sparse autoencoders disrupts token co-occurrences to reduce feature absorption, enhance probing, and narrow OOD gaps across architectures and sparsity levels.

  32. TADA! Tuning Audio Diffusion Models through Activation Steering

    cs.SD 2026-02 unverdicted novelty 6.0

    Activation steering at a semantic bottleneck in audio diffusion models achieves state-of-the-art control over musical attributes such as instruments, vocals, and genres.

  33. Crosscoding Through Time: Tracking Emergence & Consolidation Of Linguistic Representations Throughout LLM Pretraining

    cs.CL 2025-09 unverdicted novelty 6.0

    Sparse crosscoders on LLM checkpoint triplets track emergence, maintenance, and discontinuation of linguistic features during pretraining via a new RelIE metric.

  34. Aligning Sentence Embeddings to Human Concepts via Sparse Autoencoders

    cs.IR 2026-06 unverdicted novelty 5.0

    Top-k sparse autoencoders disentangle sentence embeddings into semantic, syntactic, and pragmatic concepts and enable activation steering to re-rank retrieval results for better user alignment.

  35. Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models

    cs.CL 2026-01 unverdicted novelty 5.0

    The survey organizes mechanistic interpretability techniques into a Locate-Steer-Improve framework to enable actionable improvements in LLM alignment, capability, and efficiency.

  36. Safe-SAIL: Towards a Fine-grained Safety Landscape of Large Language Models via Sparse Autoencoder Interpretation Framework

    cs.LG 2025-09 unverdicted novelty 5.0

    Safe-SAIL supplies a pre-explanation metric and segment-level simulation to interpret 1758 safety SAE features across pornography, politics, violence, and terror, with public models and tools released.

  37. Mechanistic Interpretability for Neural Networks: Circuits, Sparse Features and Symbolic Reasoning

    cs.LG 2026-07 accept

    A scoping review surveying circuit analysis, sparse autoencoders, activation steering, and neurosymbolic frameworks for interpreting and controlling Transformer-based neural networks.