mlsanityData to 7 Oct 2026

In-Context Sharpness as Alerts: An Inner Representation Perspective for Hallucination Mitigation

arXiv 2403.01548 ↗First posted 3 Mar 2024cs.CL cs.AI cs.LGcited by 43

Not on the map. Why?

Abstract (arXiv)

Large language models (LLMs) frequently hallucinate and produce factual errors, yet our understanding of why they make these errors remains limited. In this study, we delve into the underlying mechanisms of LLM hallucinations from the perspective of inner representations, and discover a salient pattern associated with hallucinations: correct generations tend to have sharper context activations in the hidden states of the in-context tokens, compared to the incorrect ones. Leveraging this insight, we propose an entropy-based metric to quantify the ``sharpness'' among the in-context hidden states and incorporate it into the decoding process to formulate a constrained decoding approach. Experiments on various knowledge-seeking and hallucination benchmarks demonstrate our approach's consistent effectiveness, for example, achieving up to an 8.6 point improvement on TruthfulQA. We believe this study can improve our understanding of hallucinations and serve as a practical solution for hallucination mitigation.

Key facts. Cited by 43 papers in the map (data through 7 Oct 2026). Builds on 30 of its 35 references that are in the map. First posted on arXiv 3 Mar 2024.

Builds on

30 of 35 references.

References · 30

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  3. 2018HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answeringcited by 3,686
  4. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,919
  5. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,895
  6. 2022Locating and Editing Factual Associations in GPTcited by 2,609
  7. 2022Survey of Hallucination in Natural Language Generationcited by 2,056
  8. 2023Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflectioncited by 1,800
  9. 2022Language Models (Mostly) Know What They Knowcited by 1,783
  10. 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,582
  11. 2020Transformer Feed-Forward Layers Are Key-Value Memoriescited by 1,307
  12. 2023Inference-Time Intervention: Eliciting Truthful Answers from a Language Modelcited by 1,259
18 more
  1. 2023Representation Engineering: A Top-Down Approach to AI Transparencycited by 1,239
  2. 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 946
  3. 2023Progress measures for grokking via mechanistic interpretabilitycited by 822
  4. 2023In-Context Retrieval-Augmented Language Modelscited by 694
  5. 2022Contrastive Decoding: Open-ended Text Generation as Optimizationcited by 598
  6. 2021Knowledge Neurons in Pretrained Transformerscited by 590
  7. 2023Dissecting Recall of Factual Associations in Auto-Regressive Language Modelscited by 512
  8. 2023Active Retrieval Augmented Generationcited by 506
  9. 2023DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Modelscited by 367
  10. 2023Challenges and Applications of Large Language Modelscited by 251
  11. 2023Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategiescited by 236
  12. 2023Survey on Factuality in Large Language Models: Knowledge, Retrieval and Domain-Specificitycited by 204
  13. 2022Analyzing Transformers in Embedding Spacecited by 131
  14. 2023Improving Language Models via Plug-and-Play Retrieval Feedbackcited by 87
  15. 2023Overthinking the Truth: Understanding how Language Models Process False Demonstrationscited by 78
  16. 2023Attention Satisfies: A Constraint-Satisfaction Lens on Factual Errors of Language Modelscited by 67
  17. 2023FELM: Benchmarking Factuality Evaluation of Large Language Modelscited by 66
  18. 2022What Are You Token About? Dense Retrieval as Distributions Over the Vocabularycited by 28

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 16; 2025: 16; 2026: 11.

43 papers.

Sort:

By area:

  1. 2024Lookback Lens: Detecting and Mitigating Contextual Hallucinations in Large Language Models Using Only Attention Mapscited by 140
  2. 2024To Believe or Not to Believe Your LLMcited by 120
  3. 2025Why Is Spatial Reasoning Hard for VLMs? An Attention Mechanism Perspective on Focus Areascited by 93
  4. 2024MLLM can see? Dynamic Correction Decoding for Hallucination Mitigationcited by 87
  5. 2024A Primer on the Inner Workings of Transformer-based Language Modelscited by 83
  6. 2024Look Twice Before You Answer: Memory-Space Visual Retracing for Hallucination Mitigation in Multimodal Large Language Modelscited by 72
  7. 2024Usable XAI: 10 Strategies Towards Exploiting Explainability in the LLM Eracited by 45
  8. 2024Knowledge Boundary of Large Language Models: A Surveycited by 36
  9. 2024DeCoRe: Decoding by Contrasting Retrieval Heads to Mitigate Hallucinationscited by 23
  10. 2026Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Modelscited by 22
  11. 2024A Survey on the Honesty of Large Language Modelscited by 21
  12. 2025Chain-of-Thought Prompting Obscures Hallucination Cues in Large Language Models: An Empirical Evaluationcited by 20
31 more
  1. 2025SelfElicit: Your Language Model Secretly Knows Where is the Relevant Evidencecited by 18
  2. 2024Mitigating Hallucinations in Large Language Models via Self-Refinement-Enhanced Knowledge Retrievalcited by 16
  3. 2024SLED: Self Logits Evolution Decoding for Improving Factuality in Large Language Modelscited by 16
  4. 2025Control the Temperature: Selective Sampling for Diverse and High-Quality LLM Outputscited by 15
  5. 2024VaLiD: Mitigating the Hallucination of Large Vision Language Models by Visual Layer Fusion Contrastive Decodingcited by 12
  6. 2025Active Layer-Contrastive Decoding Reduces Hallucination in Large Language Model Generationcited by 10
  7. 2024Integrative Decoding: Improve Factuality via Implicit Self-consistencycited by 8
  8. 2025Improve Decoding Factuality by Token-wise Cross Layer Entropy of Large Language Modelscited by 7
  9. 2024SED: Self-Evaluation Decoding Enhances Large Language Models for Better Generationcited by 5
  10. 2025Enhancing Visual Reliance in Text Generation: A Bayesian Perspective on Mitigating Hallucination in Large Vision-Language Modelscited by 5
  11. 2025Training-free Uncertainty Guidance for Complex Visual Tasks with MLLMscited by 4
  12. 2025Gradient-guided Attention Map Editing: Towards Efficient Contextual Hallucination Mitigationcited by 3
  13. 2025LayerCake: Token-Aware Contrastive Decoding within Large Language Model Layerscited by 3
  14. 2025PruneCD: Contrasting Pruned Self Model to Improve Decoding Factualitycited by 3
  15. 2025What are Models Thinking about? Understanding Large Language Model Hallucinations "Psychology" through Model Inner State Analysiscited by 2
  16. 2025DSVD: Dynamic Self-Verify Decoding for Faithful Generation in Large Language Modelscited by 2
  17. 2025TPC: Cross-Temporal Prediction Connection for Vision-Language Model Hallucination Reductioncited by 2
  18. 2026Hallucination Basins: A Dynamic Framework for Understanding and Controlling LLM Hallucinationscited by 2
  19. 2026Beyond In-Domain Detection: SpikeScore for Cross-Domain Hallucination Detectioncited by 1
  20. 2026Attn-GS: Attention-Guided Context Compression for Efficient Personalized LLMscited by 1
  21. 2026SAKED: Mitigating Hallucination in Large Vision-Language Models via Stability-Aware Knowledge Enhanced Decodingcited by 1
  22. 2026Hallucinations as Orthogonal Noise: Inference-Time Manifold Alignment via Dynamic Contextual Orthogonalizationcited by 1
  23. 2024FacLens: Transferable Probe for Foreseeing Non-Factuality in Fact-Seeking Question Answering of Large Language Modelscited by 0
  24. 2024Maintaining Informative Coherence: Migrating Hallucinations in Large Language Models via Absorbing Markov Chainscited by 0
  25. 2025URECA: The Chain of Two Minimum Set Cover Problems exists behind Adaptation to Shifts in Semantic Code Searchcited by 0
  26. 2025Extracting Visual Facts from Intermediate Layers for Mitigating Hallucinations in Multimodal Large Language Modelscited by 0
  27. 2026LANCET: Neural Intervention via Structural Entropy for Mitigating Faithfulness Hallucinations in LLMscited by 0
  28. 2026Entropy Centroids as Intrinsic Rewards for Test-Time Scalingcited by 0
  29. 2026APCD: Adaptive Path-Contrastive Decoding for Reliable Large Language Model Generationcited by 0
  30. 2026Mitigating Hallucinations in Large Language Models Via Decoder Layer Skippingcited by 0
  31. 2026LOCKR: A Hidden-State Trajectory-Guided Planner for Detecting and Repairing Stable-but-Wrong Lock-In in Diffusion Language Modelscited by 0