# Layer Importance and Hallucination Analysis in Large Language Models via Enhanced Activation Variance-Sparsity

arXiv: https://arxiv.org/abs/2411.10069 · HTML: https://mlsanity.com/paper/2411.10069/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 43 references that are in the map.
- First posted on arXiv 15 Nov 2024.

## Builds on (most cited first, 20 of 20)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022)
- [A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions](https://mlsanity.com/paper/2311.05232/) (2023)
- [Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks](https://mlsanity.com/paper/1706.02690/) (2017)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generation](https://mlsanity.com/paper/2302.09664/) (2023)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [Are Emergent Abilities of Large Language Models a Mirage?](https://mlsanity.com/paper/2304.15004/) (2023)
- [Self-critiquing models for assisting human evaluators](https://mlsanity.com/paper/2206.05802/) (2022)
- [DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Models](https://mlsanity.com/paper/2309.03883/) (2023)
- [INSIDE: LLMs' Internal States Retain the Power of Hallucination Detection](https://mlsanity.com/paper/2402.03744/) (2024)
- [Unsupervised Real-Time Hallucination Detection based on the Internal States of Large Language Models](https://mlsanity.com/paper/2403.06448/) (2024)
- [Datasheet for the Pile](https://mlsanity.com/paper/2201.07311/) (2022)
- [Uncertainty Estimation in Autoregressive Structured Prediction](https://mlsanity.com/paper/2002.07650/) (2020)
- [Enhancing Large Language Models Against Inductive Instructions with Dual-critique Prompting](https://mlsanity.com/paper/2305.13733/) (2023)
- [AlpaGasus: Training A Better Alpaca with Fewer Data](https://mlsanity.com/paper/2307.08701/) (2023)

## Built on by (most cited first, 5 of 5)

- [Correctness Assessment of Code Generated by Large Language Models Using Internal Representations](https://mlsanity.com/paper/2501.12934/) (2025)
- [EM-MIAs: Enhancing Membership Inference Attacks in Large Language Models through Ensemble Modeling](https://mlsanity.com/paper/2412.17249/) (2024)
- [LightMoE: Reducing Mixture-of-Experts Redundancy through Expert Replacing](https://mlsanity.com/paper/2603.12645/) (2026)
- [SymLoc: Symbolic Localization of Hallucination across HaluEval and TruthfulQA](https://mlsanity.com/paper/2511.14172/) (2025)
- [Curvature-Weighted Capacity Allocation: A Minimum Description Length Framework for Layer-Adaptive Large Language Model Optimization](https://mlsanity.com/paper/2603.00910/) (2026)

## Abstract (arXiv)

Evaluating the importance of different layers in large language models (LLMs) is crucial for optimizing model performance and interpretability. This paper first explores layer importance using the Activation Variance-Sparsity Score (AVSS), which combines normalized activation variance and sparsity to quantify each layer's contribution to overall model performance. By ranking layers based on AVSS and pruning the least impactful 25%, our experiments on tasks such as question answering, language modeling, and sentiment classification show that over 90% of the original performance is retained, highlighting potential redundancies in LLM architectures. Building on AVSS, we propose an enhanced version tailored to assess hallucination propensity across layers (EAVSS). This improved approach introduces Hallucination-Specific Activation Variance (HSAV) and Hallucination-Specific Sparsity (HSS) metrics, allowing precise identification of hallucination-prone layers. By incorporating contrastive learning on these layers, we effectively mitigate hallucination generation, contributing to more robust and efficient LLMs(The maximum performance improvement is 12%). Our results on the NQ, SciQ, TriviaQA, TruthfulQA, and WikiQA datasets demonstrate the efficacy of this method, offering a comprehensive framework for both layer importance evaluation and hallucination mitigation in LLMs.