# Concepts Whisper: Spectral Anti-Concentration and the Dual Geometry of Transformer Representations

arXiv: https://arxiv.org/abs/2605.01609 · HTML: https://mlsanity.com/paper/2605.01609/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Sparse autoencoder training and evaluation, one of 13 directions in Mechanistic interpretability.
- Builds on 27 of its 44 references that are in the map.
- First posted on arXiv 2 May 2026.

## Placed in

- Mechanistic interpretability / [Sparse autoencoder training and evaluation](https://mlsanity.com/mechinterp/sparse-autoencoder-training-and-evaluation/)

## Builds on (most cited first, 27 of 27)

- [Efficient Estimation of Word Representations in Vector Space](https://mlsanity.com/paper/1301.3781/) (2013)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Refusal in Language Models Is Mediated by a Single Direction](https://mlsanity.com/paper/2406.11717/) (2024)
- [How Contextual are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddings](https://mlsanity.com/paper/1909.00512/) (2019)
- [Prevalence of Neural Collapse during the terminal phase of deep learning training](https://mlsanity.com/paper/2008.08186/) (2020)
- [The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets](https://mlsanity.com/paper/2310.06824/) (2023)
- [Word Translation Without Parallel Data](https://mlsanity.com/paper/1710.04087/) (2017)
- [Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2](https://mlsanity.com/paper/2408.05147/) (2024)
- [Emergent Linear Representations in World Models of Self-Supervised Sequence Models](https://mlsanity.com/paper/2309.00941/) (2023)
- [Do Llamas Work in English? On the Latent Language of Multilingual Transformers](https://mlsanity.com/paper/2402.10588/) (2024)
- [Representation Degeneration Problem in Training Natural Language Generation Models](https://mlsanity.com/paper/1907.12009/) (2019)
- [Implicit Self-Regularization in Deep Neural Networks: Evidence from Random Matrix Theory and Implications for Learning](https://mlsanity.com/paper/1810.01075/) (2018)
- [Not All Language Model Features Are One-Dimensionally Linear](https://mlsanity.com/paper/2405.14860/) (2024)
- [Whitening Sentence Representations for Better Semantics and Faster Retrieval](https://mlsanity.com/paper/2103.15316/) (2021)
- [Emerging Cross-lingual Structure in Pretrained Language Models](https://mlsanity.com/paper/1911.01464/) (2019)
- [Linear Representations of Sentiment in Large Language Models](https://mlsanity.com/paper/2310.15154/) (2023)
- [All Bark and No Bite: Rogue Dimensions in Transformer Language Models Obscure Representational Quality](https://mlsanity.com/paper/2109.04404/) (2021)
- [Llama Scope: Extracting Millions of Features from Llama-3.1-8B with Sparse Autoencoders](https://mlsanity.com/paper/2410.20526/) (2024)
- [The Truth is in There: Improving Reasoning in Language Models with Layer-Selective Rank Reduction](https://mlsanity.com/paper/2312.13558/) (2023)
- [Predicting trends in the quality of state-of-the-art neural networks without access to training or testing data](https://mlsanity.com/paper/2002.06716/) (2020)
- [BERT Busters: Outlier Dimensions that Disrupt Transformers](https://mlsanity.com/paper/2105.06990/) (2021)
- [Traditional and Heavy-Tailed Self Regularization in Neural Network Models](https://mlsanity.com/paper/1901.08276/) (2019)
- [WhiteningBERT: An Easy Unsupervised Sentence Embedding Approach](https://mlsanity.com/paper/2104.01767/) (2021)
- [Small Singular Values Matter: A Random Matrix Analysis of Transformer Models](https://mlsanity.com/paper/2410.17770/) (2024)
- [Language-Specific Latent Process Hinders Cross-Lingual Performance](https://mlsanity.com/paper/2505.13141/) (2025)

## Abstract (arXiv)

We find that transformer concept representations systematically anti-concentrate in the spectral tail of the unembedding covariance, encoding word-level concepts in low-variance directions across a 17-model core suite and an expanded set of 22 semantic concept categories, with convergent replications from three independent extraction methods. Residual-stream difference-of-means vectors anti-concentrate in all 17 models (model-level one-sample t-test, p = 3.8e-9), and remain more tail-aligned than norm-matched random directions in 13 of 17; convergent support comes from sparse autoencoder (SAE) features (p = 4.5e-19 across concepts within a model) and linear probes on Llama and Qwen. We identify a dual geometry: activation-space concept directions anti-concentrate while static unembedding-row contrasts concentrate in high-variance directions (p < 10^-4). This investigation arose from testing whether the causal inner product of Park et al. (2024) aids cross-lingual concept transport; a matched-spectrum randomization across 17 models and four language pairs finds no evidence that Whitened Causal Alignment improves over spectral regularization alone (p = 0.95). Split-injection interventions, restricted to steering strengths at which both arms remain interpretable, show the predicted interference asymmetry in four of five models (paired Cohen's d_z up to 1.19) with no significant reversal inside that regime, and POS-tag probing across eight models shows syntax preferentially encoded in the high-variance subspace in six of eight architectures, with a significant reversal in the Qwen 2.5 family. These results suggest transformers rotate semantic content into spectrally quiet regions during contextualized processing, where, in some architectures, interventions may reduce grammatical disruption relative to high-variance steering.