Semantic Representations

The standard sparse-autoencoder (SAE) interpretability protocol labels each feature from its top-activating contexts and validates by single-feature steering. We propose the pairwise matrix protocol, co-varying steering coefficient with joint condition, and report three findings the standard one-corner protocol misses on Qwen3-1.7B-Instruct, replicated on Gemma-2-2B-it. First, a feature labelled "AI self-disclaimer" from its top contexts produces an inverted U-shape under a coefficient sweep: at c=+500 the model substitutes a fluent contemplative-philosopher voice for the disclaimer. Two further features anchor the criterion (one monotonic, one pure breakdown). Second, three near-orthogonal cluster-specific features that individually steer a philosophy-of-mind register, jointly suppressed at c=-500, damage grounded composition on recipes and engine explanations as well as introspective prompts; single-feature suppression at the same magnitude leaves controls intact. Third, a matched-geometry comparison of single-feature, joint, and random-direction perturbations (norm ~1.55, cosine ~0.64) yields three distinct output regimes: single-feature substitutes strategy filler, random direction substitutes diverse content, joint suppression alone produces placeholder text. Coherence loss is direction-pattern-dependent, not magnitude-dependent. All three findings reproduce on Gemma with model-specific damage signatures; the matched-geometry control is CI-separated by ~10x. The pipeline also locates a top causally responsible feature in Llama-3.1-8B-Instruct.

We show that the geometric relations between semantic features in large language models' hidden states closely mirror human psychological associations. We construct feature vectors corresponding to 360 words and project them on 32 semantic axes (e.g. beautiful-ugly, soft-hard), and find that these projections correlate highly with human ratings of those words on the respective semantic scales. Second, we find that the cosine similarities between the semantic axes themselves are highly predictive of the correlations between these scales in the survey. Third, we show that substantial variance across the 32 semantic axes lies on a low-dimensional subspace, reproducing patterns typical of human semantic associations. Finally, we demonstrate that steering a word on one semantic axis causes spillover effects on the model's rating of that word on other semantic scales proportionate to the cosine similarity between those semantic axes. These findings suggest that features should be understood not only in isolation but through their geometric relations and the meaningful subspaces they form.

Sparse autoencoders (SAEs) extract millions of interpretable features from a language model, but flat feature inventories aren't very useful on their own. Domain concepts get mixed with generic and weakly grounded features, while related ideas are scattered across many units, and there's no way to understand relationships between features. We address this by first constructing a strict domain-specific concept universe from a large SAE inventory using contrastive activations and a multi-stage filtering process. Next, we build two aligned graph views on the filtered set: a co-occurrence graph for corpus-level conceptual structure, organized at multiple levels of granularity, and a transcoder-based mechanism graph that links source-layer and target-layer features through sparse latent pathways. Automated edge labeling then turns these graph views into readable knowledge graphs rather than unlabeled layouts. In a case study on a biology textbook, these graphs recover coherent chapter and subchapter-level structure, reveal concepts that bridge neighboring topics, and transform messy sentence-level activity containing thousands of features into compact, readable views that illustrate the model's local activity. Taken together, this reframes a flat SAE inventory as an internal knowledge graph that converts feature-level interpretability into a global map of model knowledge and enables audits of reasoning faithfulness.

Learned Sparse IR models, such as SPLADE, offer an excellent efficiency-effectiveness tradeoff. However, they rely on the underlying backbone vocabulary, which might hinder performance (polysemicity and synonymy) and pose a challenge for multi-lingual and multi-modal usages. To solve this limitation, we propose to replace the backbone vocabulary with a latent space of semantic concepts learned using Sparse Auto-Encoders (SAE). Throughout this paper, we study the compatibility of these 2 concepts, explore training approaches, and analyze the differences between our SAE-SPLADE model and traditional SPLADE models. Our experiments demonstrate that SAE-SPLADE achieves retrieval performance comparable to SPLADE on both in-domain and out-of-domain tasks while offering improved efficiency.

This paper reformulates Transformer/Attention mechanisms in Large Language Models (LLMs) through measure theory and frequency analysis, theoretically demonstrating that hallucination is an inevitable structural limitation. The embedding space functions as a conditional expectation over a σ-algebra, and its failure to be isomorphic to the semantic truth set fundamentally causes logical consistency breakdown. WavePhaseNet Method The authors propose WavePhaseNet, which explicitly constructs a Semantic Conceptual Hierarchy Structure (SCHS) using Discrete Fourier Transform (DFT). By applying DFT along the sequence dimension, semantic information is decomposed into frequency bands: low-frequency components capture global meaning and intent, while high-frequency components represent local syntax and expression. This staged separation enables precise semantic manipulation in diagonalized space. Dimensionality Reduction GPT-4's 24,576-dimensional embedding space exhibits a 1/f spectral structure based on language self-similarity and Zipf's law. Through cumulative energy analysis, the authors derive that approximately 3,000 dimensions constitute the lower bound for "complete representation." This demonstrates that reduction from 24,576 to 3,000 dimensions preserves meaning and intent while enabling rigorous reasoning and suppressing hallucination. Cohomological Consistency Control The reduced embedding space, constructed via cohomological regularization over overlapping local windows, allows defining a graph structure and cochain complex. This quantifies inconsistencies among local inferences as coboundary-based losses. Applying harmonic projection based on Hodge theory positions cohomology as a computable regularization principle for controlling semantic consistency, extracting maximally consistent global representations.

Sparse Autoencoders (SAEs) extract features from LLM internal activations, meant to correspond to interpretable concepts. A core SAE training hyperparameter is L0: how many SAE features should fire per token on average. Existing work compares SAE algorithms using sparsity-reconstruction tradeoff plots, implying L0 is a free parameter with no inherently correct value aside from its effect on reconstruction. In this work we study the effect of L0 on SAEs, and show that if L0 is not set correctly, the SAE fails to disentangle the underlying features of the LLM. If L0 is too low, the SAE will mix correlated features to improve reconstruction. If L0 is too high, the SAE finds degenerate solutions that also mix features. Further, we present a proxy metric that can help guide the search for the correct L0 for an SAE on a given training distribution. We show that our method finds the correct L0 in toy models and coincides with peak sparse probing performance in LLM SAEs. We find that most commonly used SAEs have an L0 that is too low. Our work shows that practitioners must set L0 correctly to train SAEs with monosemantic features.

While sparse autoencoders (SAEs) have generated significant excitement, a series of negative results have added to skepticism about their usefulness. Here, we establish a conceptual distinction that reconciles competing narratives surrounding SAEs. We argue that even if SAEs may be less effective for \textit{acting on known concepts}, SAEs are especially powerful tools for \textit{discovering unknown concepts}. This distinction separates existing negative results from positive results, and suggests several classes of SAE applications. Specifically, we outline use cases for SAEs in (i) ML interpretability, explainability, fairness, auditing, and safety, and (ii) social and health sciences.

Conversational Recommender Systems (CRS) powered by Large Language Models (LLMs) enable users to articulate explicit and dynamic preferences, overcoming the limitations of fixed templates. However, despite their superior semantic proficiency, LLMs have not yet achieved corresponding improvements in recommendation accuracy. This discrepancy arises from a fundamental representation gap: while LLMs operate within a semantic space, they lack the behavioral grounding needed to encode user behavioral patterns, such as item co-occurrences, which are crucial for accurate recommendations. To address this, we propose a model-agnostic Variational EM Framework for Bridging Semantics and Behavior in LLM-based Conversational Recommendation (LatentCRS). Based on the observation that dialogue and interactions reflect the same latent intent, LatentCRS uses a variational expectation-maximization (EM) procedure, where user intent connects semantic representations with behavioral patterns. Extensive experiments on real-world datasets demonstrate that LatentCRS effectively bridges the representation gap and outperforms baselines.

The basic problem of semantic compression is to minimize the length of a message while preserving its meaning. This differs from classical notions of compression in that the distortion is not measured directly at the level of bits, but rather in an abstract semantic space. In order to make this precise, we take inspiration from cognitive neuroscience and machine learning and model semantic space as a continuous Euclidean vector space. In such a space, stimuli like speech, images, or even ideas, are mapped to high-dimensional real vectors, and the location of these embeddings determines their meaning relative to other embeddings. This suggests that a natural metric for semantic similarity is just the Euclidean distance, which is what we use in this work. We map the optimization problem of determining the minimal-length, meaning-preserving message to a spin glass Hamiltonian and solve the resulting statistical mechanics problem using replica theory. We map out the replica symmetric phase diagram, identifying distinct phases of semantic compression: a first-order transition occurs between phases marked by the emergence of paraphrases, whereas a continuous crossover is seen from extractive to abstractive compression. We speculate on which features of the phase diagram are captured by replica symmetry, and which features change under replica symmetry breaking. We conclude by showing numerical simulations of compressions obtained by simulated annealing and greedy algorithms, and argue that while the problem of finding a meaning-preserving compression is computationally hard in the worst case, there exist efficient algorithms which achieve near optimal performance in the typical case.

Although Sparse Autoencoders (SAEs) have mitigated the opacity of large language models (LLMs) by decomposing dense representations into sparse features, explaining these features still remains a central challenge. Current explanation methods, however, typically operate within an open-loop paradigm, failing to leverage mechanistic feedback for further refinement. In this paper, we propose SAEExplainer, a training framework that utilizes activation scores as an objective reward signal to train the model for self-correction and iterative bootstrapping. By iteratively verifying and correcting foundational explanations through a two-round optimization process, SAEExplainer achieves continuous improvement in its explanatory capabilities. This mechanism significantly reduces explanation hallucinations and reinforces causal triggering patterns. Extensive experiments demonstrate our approach improves upon established baselines across most metrics, especially in causal triggering and discriminative activation. The code is available at https://github.com/he-jingyi/SAEExplainer.