The paper revisits parsimoniously activated dictionary learning (PADL), a method that imposes a global regularization on the number of activated dictionary atoms. It shows that PADL can be reformulated as a maximum a posteriori estimation under a structured generative model, enabling derivation of generalization guarantees and an analytical tradeoff between sparsity, storage cost, and reconstruction accuracy. Leveraging this insight, the authors propose an efficient PADL algorithm that removes manual hyperparameter tuning, improves reconstruction performance on visual benchmarks, and accelerates inference for vision‑language models.
arXiv:2606. 02385v1 Announce Type: cross Abstract: Sparse Autoencoders (SAEs) have found success parsing neural representations into interpretable concepts, providing a basis for understanding and control.
The paper introduces Consistent Sparse Dictionary Learning (ConSDL), a method that replaces the linear encoder in sparse autoencoders with per‑sample sparse coding to improve dictionary learning. Synthetic experiments show ConSDL achieves near‑perfect recovery of dictionaries and coefficients, and in real vision and language tasks it boosts cross‑seed consistency by 17.6% for dictionaries and 35.9% for coefficients compared to six SAE baselines.
The paper argues that the linear representation hypothesis—where neural activations encode concepts as linear mixtures—breaks down under superposition, causing linear decision boundaries to lose linearity after projection. Classical sparse coding with per‑sample inference can recover latent factors, but sparse autoencoders (SAEs) amortise this inference into a fixed encoder, creating a systematic gap that persists across training sizes, latent dimensions, and sparsity levels. Experiments show that the failure of SAEs under out‑of‑distribution compositional shifts stems from poor dictionary learning rather than inference, and that a good dictionary can solve the problem at all tested scales, highlighting scalable dictionary learning as the key open challenge.
arXiv:2509. 10033v2 Announce Type: replace Abstract: Sparse dictionary coding represents signals as linear combinations of a few dictionary atoms.
arXiv:2610.07389v1 Announce Type: cross Abstract: Sparse autoencoders are widely used to uncover interpretable features in neural networks, yet reliable recovery remains difficult when features overl...
The paper introduces a dictionary learning framework for probability measure-valued data by embedding measures into Hilbert spaces using features derived from optimal transport geometry. It learns data‑adapted dictionaries through sparse autoencoder parameterizations, yielding fast, scalable representations of probability measures as sparse finite‑dimensional codes. The authors demonstrate the method on hyperspectral images, geometric point clouds, and document embeddings from large language models, and provide generalization bounds and covering number results for optimal transport and entropic transport map dictionaries.
arXiv:2607. 08605v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) have emerged as a promising technique for mechanistic interpretability by learning a set of sparse latent features in large models, each of which encodes a distinct concept.
arXiv:2606. 27802v1 Announce Type: new Abstract: Hierarchical predictive coding provides an interpretable framework for perception as error-driven inference in multi-layer generative models, while sparse coding imposes parsimonious latent representations through explicit sparsity constraints.
MoR-MLLM introduces a computation-sparse multimodal large language model that uses adaptive per-token recursion to dynamically adjust recursive depth. This approach allocates more computation to visually or linguistically challenging tokens while skipping redundant operations for simpler ones. A three-stage MoR-Tuning strategy and entropy-regularized loss stabilize training, enabling the model to reduce training memory and computation complexity while maintaining high performance on vision-language tasks.
arXiv:2606. 30609v1 Announce Type: cross Abstract: Sparse Autoencoders (SAEs) are widely used to interpret large language models by decomposing activations into sparse, human-understandable features, but scaling to large dictionaries exposes fundamental challenges.
arXiv:2607. 20652v1 Announce Type: cross Abstract: Language models are thought to exhibit the phenomenon of superposition, representing many more features than dimensions in their residual streams.
arXiv:2508.03351v3 Announce Type: replace-cross Abstract: Large language models (LLMs) have demonstrated remarkable capabilities across diverse language tasks, motivating their extension to vision-la...