The Rate-Distortion-Polysemanticity Tradeoff in SAEs

• URL: http://arxiv.org/abs/2605.14694v1
• Date: Thu, 14 May 2026 11:10:52 GMT
• Title: The Rate-Distortion-Polysemanticity Tradeoff in SAEs
• Abstract summary: This paper introduces the Rate-Distortion-Polysemanticity tradeoff in SAEs.<n>We show that restricting the SAE to be monosemantic necessarily comes with an increase in rate and distortion.<n>We extend the analysis to real-world settings by deriving necessary conditions that a polysemanticity measure should satisfy.
• Score: 24.387030280550732
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Sparse Autoencoders (SAEs) that can accurately reconstruct their input (minimizing distortion) by making efficient use of few features (minimizing the rate) often fail to learn monosemantic representations (highly interpretable), limiting their usefulness for mechanistic interpretability. In this paper, we characterise this tension in learning faithful, efficient, and interpretable explanations, introducing the Rate-Distortion-Polysemanticity tradeoff in SAEs. Under toy-modeling assumptions, we theoretically and empirically show that restricting the SAE to be monosemantic necessarily comes with an increase in rate and distortion. Assuming a generative model behind the input observations, we further demonstrate that the degree of polysemanticity of optimal SAEs is determined by the training data distribution, especially by the probability of features to co-occur. Finally, we extend the analysis to real-world settings by deriving necessary conditions that a polysemanticity measure should satisfy when the data-generating process is unknown, and we benchmark existing proxy metrics on SAEs trained on Large Language Models. Taken together, our findings show that polysemanticity is a data problem that should be accounted for when addressing it at the architectural and optimization level.

Related papers

• From Sparse Features to Trustworthy Proxies: Certifying SAE-Based Interpretability [53.681678236115836]
  We study when can an SAE-based explanation be treated as a faithful view of an underlying frozen LM.<n>Our framework derives an upper bound on the base model's expected risk using four measurable quantities.<n>We show that the bound becomes non-vacuous on GPT-2 Small, Gemma-2B, and Llama-3-8B at practical sample sizes.
  arXiv  Detail & Related papers  (2026-06-16T18:28:23Z)
• Improving Machine Learning Performance with Synthetic Augmentation [0.0]
  We formalize synthetic augmentation as a modification of the effective training distribution.<n>Additional samples may reduce estimation error, but they may also shift the population objective.<n>We show that synthetic augmentation is beneficial only in variance-dominant regimes.<n>Rare-regime targeting can improve domain-specific metrics but may conflict with unconditional permutation inference.
  arXiv  Detail & Related papers  (2026-04-16T00:23:01Z)
• Did Models Sufficient Learn? Attribution-Guided Training via Subset-Selected Counterfactual Augmentation [61.248535801314375]
  Subset-Selected Counterfactual Augmentation (SS-CA)<n>We develop Counterfactual LIMA to identify minimal spatial region sets whose removal can selectively alter model predictions.<n>Experiments show that SS-CA improves generalization on in-distribution (ID) test data and achieves superior performance on out-of-distribution (OOD) benchmarks.
  arXiv  Detail & Related papers  (2025-11-15T08:39:22Z)
• On the Theoretical Understanding of Identifiable Sparse Autoencoders and Beyond [36.107366496809675]
  Sparse autoencoders (SAEs) have emerged as a powerful tool for interpreting features learned by large language models (LLMs)<n>It aims to recover complex superposed polysemantic features into interpretable monosemantic ones through feature reconstruction via sparsely activated neural networks.<n>Despite the wide applications of SAEs, it remains unclear under what conditions an SAE can fully recover the ground truth monosemantic features from the superposed polysemantic ones.
  arXiv  Detail & Related papers  (2025-06-19T02:16:08Z)
• Taming Polysemanticity in LLMs: Provable Feature Recovery via Sparse Autoencoders [50.52694757593443]
  Existing SAE training algorithms often lack rigorous mathematical guarantees and suffer from practical limitations.<n>We first propose a novel statistical framework for the feature recovery problem, which includes a new notion of feature identifiability.<n>We introduce a new SAE training algorithm based on bias adaptation'', a technique that adaptively adjusts neural network bias parameters to ensure appropriate activation sparsity.
  arXiv  Detail & Related papers  (2025-06-16T20:58:05Z)
• Beyond Interpretability: The Gains of Feature Monosemanticity on Model Robustness [68.69369585600698]
  Deep learning models often suffer from a lack of interpretability due to polysemanticity. Recent advances in monosemanticity, where neurons correspond to consistent and distinct semantics, have significantly improved interpretability. We show that monosemantic features not only enhance interpretability but also bring concrete gains in model performance.
  arXiv  Detail & Related papers  (2024-10-27T18:03:20Z)
• A Likelihood Based Approach to Distribution Regression Using Conditional Deep Generative Models [8.862614615192578]
  We study the large-sample properties of a likelihood-based approach for estimating conditional deep generative models.<n>Our results lead to the convergence rate of a sieve maximum likelihood estimator for estimating the conditional distribution.
  arXiv  Detail & Related papers  (2024-10-02T20:46:21Z)
• Quantifying Prediction Consistency Under Fine-Tuning Multiplicity in Tabular LLMs [10.494477811252034]
  Fine-tuning multiplicity can arise in Tabular LLMs on classification tasks.<n>Our work formalizes this unique challenge of fine-tuning multiplicity in Tabular LLMs.<n>We propose a novel measure to quantify consistency of individual predictions without expensive model retraining.
  arXiv  Detail & Related papers  (2024-07-04T22:22:09Z)
• Semi-supervised Regression Analysis with Model Misspecification and High-dimensional Data [8.619243141968886]
  We present an inference framework for estimating regression coefficients in conditional mean models. We develop an augmented inverse probability weighted (AIPW) method, employing regularized estimators for both propensity score (PS) and outcome regression (OR) models. Our theoretical findings are verified through extensive simulation studies and a real-world data application.
  arXiv  Detail & Related papers  (2024-06-20T00:34:54Z)
• Cycles of Thought: Measuring LLM Confidence through Stable Explanations [53.15438489398938]
  Large language models (LLMs) can reach and even surpass human-level accuracy on a variety of benchmarks, but their overconfidence in incorrect responses is still a well-documented failure mode. We propose a framework for measuring an LLM's uncertainty with respect to the distribution of generated explanations for an answer.
  arXiv  Detail & Related papers  (2024-06-05T16:35:30Z)
• Federated Conformal Predictors for Distributed Uncertainty Quantification [83.50609351513886]
  Conformal prediction is emerging as a popular paradigm for providing rigorous uncertainty quantification in machine learning. In this paper, we extend conformal prediction to the federated learning setting. We propose a weaker notion of partial exchangeability, better suited to the FL setting, and use it to develop the Federated Conformal Prediction framework.
  arXiv  Detail & Related papers  (2023-05-27T19:57:27Z)
• Regularizing Variational Autoencoder with Diversity and Uncertainty Awareness [61.827054365139645]
  Variational Autoencoder (VAE) approximates the posterior of latent variables based on amortized variational inference. We propose an alternative model, DU-VAE, for learning a more Diverse and less Uncertain latent space.
  arXiv  Detail & Related papers  (2021-10-24T07:58:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.