mlsanityData to 14 Oct 2026

EigenTrack: Spectral Activation Feature Tracking for Hallucination and Out-of-Distribution Detection in LLMs and VLMs

arXiv 2509.15735 ↗First posted 19 Sep 2025cs.LGcited by 10

Not on the map. Why?

Abstract (arXiv)

Large language models (LLMs) offer broad utility but remain prone to hallucination and out-of-distribution (OOD) errors. We propose EigenTrack, an interpretable real-time detector that uses the spectral geometry of hidden activations, a compact global signature of model dynamics. By streaming covariance-spectrum statistics such as entropy, eigenvalue gaps, and KL divergence from random baselines into a lightweight recurrent classifier, EigenTrack tracks temporal shifts in representation structure that signal hallucination and OOD drift before surface errors appear. Unlike black- and grey-box methods, it needs only a single forward pass without resampling. Unlike existing white-box detectors, it preserves temporal context, aggregates global signals, and offers interpretable accuracy-latency trade-offs.

Key facts. Cited by 10 papers in the map (data through 14 Oct 2026). Builds on 19 of its 30 references that are in the map. First posted on arXiv 19 Sep 2025.

Builds on

19 of 30 references.

References · 19

  1. 2016A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networkscited by 2,385
  2. 2023SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Modelscited by 810
  3. 2023DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvaturecited by 726
  4. 2023Fast-DetectGPT: Efficient Zero-Shot Detection of Machine-Generated Text via Conditional Probability Curvaturecited by 287
  5. 2018Implicit Self-Regularization in Deep Neural Networks: Evidence from Random Matrix Theory and Implications for Learningcited by 215
  6. 2024ReDeEP: Detecting Hallucination in Retrieval-Augmented Generation via Mechanistic Interpretabilitycited by 101
  7. 2024Unsupervised Real-Time Hallucination Detection based on the Internal States of Large Language Modelscited by 86
  8. 2019How Does BERT Answer Questions? A Layer-Wise Analysis of Transformer Representationscited by 63
  9. 2022RankFeat: Rank-1 Feature Removal for Out-of-distribution Detectioncited by 52
  10. 2025Hallucination Detection in LLMs Using Spectral Features of Attention Mapscited by 43
  11. 2023Chain of Natural Language Inference for Reducing Large Language Model Ungrounded Hallucinationscited by 35
  12. 2024Weight decay induces low-rank attention layerscited by 27
7 more
  1. 2024Cost-Effective Hallucination Detection for LLMscited by 18
  2. 2024Glimpse: Enabling White-Box Methods to Use Proprietary Models for Zero-Shot LLM-Generated Text Detectioncited by 14
  3. 2023Lightweight, Uncertainty-Aware Conformalized Visual Odometrycited by 13
  4. 2023Mutual Information-calibrated Conformal Feature Fusion for Uncertainty-Aware Multimodal 3D Object Detection at the Edgecited by 9
  5. 2025SPARC: Subspace-Aware Prompt Adaptation for Robust Continual Learning in LLMscited by 3
  6. 2025EigenShield: Causal Subspace Filtering via Random Matrix Theory for Adversarially Robust Vision-Language Modelscited by 2
  7. 2025SpectralGap: Graph-Level Out-of-Distribution Detection via Laplacian Eigenvalue Gapscited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 9.

10 papers.

  1. 2025Improving Semantic Uncertainty Quantification in LVLMs with Semantic Gaussian Processescited by 3
  2. 2026Algebraic Dead Directions in LayerNorm Transformers: A Forward-Pass-Only Diagnostic at LLM Scalecited by 3
  3. 2026Dead-Direction Signatures: A Cheap Spectral Reading of Singular Complexitycited by 2
  4. 2026Does Bielik Know What It Doesn't Know? Activation Dispersion Separates Entity Familiarity from Factual Reliability Across Model Scalecited by 1
  5. 2026When Context Bites: Detecting RAG Poisoning via Document-Level Attention Collapsecited by 1
  6. 2026Spectral Archaeology: The Causal Topology of Model Evolutioncited by 0
  7. 2026Self-Attention as Transport: Limits of Symmetric Spectral Diagnosticscited by 0
  8. 2026Leveraging Visual Signals for Robust Token-Level Uncertainty in Vision-Language Generationcited by 0
  9. 2026Graded Entity-Familiarity Readouts in Language Models: Polish Adaptation, Cross-Language Robustness, and Refusal Steeringcited by 0
  10. 2026Prediction of Prediction (PoP): Inter-Layer Activation Fusion for Single-Pass Hallucination Detection in Large Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Structure and Redundancy in Large Language Models: A Spectral Study via Random Matrix Theory
  2. 2026Spectral Geometry for Deep Learning: Compression and Hallucination Detection via Random Matrix Theory
  3. 2024INSIDE: LLMs' Internal States Retain the Power of Hallucination Detection
  4. 2026AURORA: Asymmetry and Update-Induced Rotation for Robust Hallucination Detection in Large Language Models
  5. 2026Density Ridge Selective Prediction for LLM and VLM Hallucination Detection under Calibration Label Scarcity
  6. 2026External Observers May See More Clearly: Cross-Model Span-Level Hallucination Detection in Large Language Models via Hidden State Probing
  7. 2026Predictive Coding and Information Bottleneck for Hallucination Detection in Large Language Models
  8. 2026Thermodynamic Signatures of Reasoning: Free-Energy and Spectral-Form-Factor Diagnostics for Hallucination Detection in Large Language Models