From Confidence to Collapse in LLM Factual Robustness

• URL: http://arxiv.org/abs/2508.16267v2
• Date: Tue, 26 Aug 2025 11:54:16 GMT
• Title: From Confidence to Collapse in LLM Factual Robustness
• Authors: Alina Fastowski, Bardh Prenkaj, Gjergji Kasneci,
• Abstract summary: We introduce a principled approach to measure factual robustness from the perspective of the generation process.<n>The Factual Robustness Score (FRS) is a novel metric which quantifies the stability of a fact against perturbations in decoding conditions.
• Score: 21.27503954808115
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Ensuring the robustness of factual knowledge in LLMs is critical for reliable applications in tasks such as question answering and reasoning. However, existing evaluation methods predominantly focus on performance-based metrics, often investigating from the perspective of prompt perturbations, which captures only the externally triggered side of knowledge robustness. To bridge this gap, we introduce a principled approach to measure factual robustness from the perspective of the generation process by analyzing token distribution entropy in combination with temperature scaling sensitivity. These two factors build the Factual Robustness Score (FRS), a novel metric which quantifies the stability of a fact against perturbations in decoding conditions, given its initial uncertainty. To validate our approach, we conduct extensive experiments on 5 LLMs across 3 closed-book QA datasets (SQuAD, TriviaQA, and HotpotQA). We show that factual robustness varies significantly -- smaller models report an FRS of $0.76$, larger ones $0.93$ -- with accuracy degrading by ~$60\%$ under increased uncertainty. These insights demonstrate how entropy and temperature scaling impact factual accuracy, and lay a foundation for developing more robust knowledge retention and retrieval in future models.

Related papers

• ReasonBENCH: Benchmarking the (In)Stability of LLM Reasoning [2.1461777157838724]
  We introduce ReasonBENCH, the first benchmark designed to quantify the underlying instability in large language models (LLMs) reasoning.<n>Across tasks from different domains, we find that the vast majority of reasoning strategies and models exhibit high instability.<n>We further analyze the impact of prompts, model families, and scale on the trade-off between solve rate and stability.
  arXiv  Detail & Related papers  (2025-12-08T18:26:58Z)
• Confidence-Based Response Abstinence: Improving LLM Trustworthiness via Activation-Based Uncertainty Estimation [7.3923284353934875]
  We propose a method for confidence estimation in retrieval-augmented generation (RAG) systems that aligns closely with the correctness of large language model (LLM) outputs.<n>Our approach extends prior uncertainty quantification methods by leveraging raw feed-forward network (FFN) activations as auto-regressive signals.<n>Our results demonstrate that activation-based confidence modeling offers a scalable, architecture-aware path toward trustworthy RAG deployment.
  arXiv  Detail & Related papers  (2025-10-15T16:55:56Z)
• Towards Harmonized Uncertainty Estimation for Large Language Models [22.58034272573749]
  It is essential to quantify the reliability of their generations through uncertainty estimation.<n>We propose CUE (Corrector for Uncertainty Estimation): A straightforward yet effective method that employs a lightweight model trained on data aligned with the target LLM's performance to adjust uncertainty scores.
  arXiv  Detail & Related papers  (2025-05-25T10:17:57Z)
• Token-Level Uncertainty Estimation for Large Language Model Reasoning [24.56760223952017]
  Large Language Models (LLMs) have demonstrated impressive capabilities, but their output quality remains inconsistent across various application scenarios.<n>We propose a token-level uncertainty estimation framework to enable LLMs to self-assess and self-improve their generation quality in mathematical reasoning.
  arXiv  Detail & Related papers  (2025-05-16T22:47:32Z)
• Enhancing LLM Reliability via Explicit Knowledge Boundary Modeling [48.15636223774418]
  Large language models (LLMs) are prone to hallucination stemming from misaligned self-awareness.<n>We propose the Explicit Knowledge Boundary Modeling framework to integrate fast and slow reasoning systems to harmonize reliability and usability.
  arXiv  Detail & Related papers  (2025-03-04T03:16:02Z)
• Towards Fully Exploiting LLM Internal States to Enhance Knowledge Boundary Perception [58.62352010928591]
  Large language models (LLMs) exhibit impressive performance across diverse tasks but often struggle to accurately gauge their knowledge boundaries.<n>This paper explores leveraging LLMs' internal states to enhance their perception of knowledge boundaries from efficiency and risk perspectives.
  arXiv  Detail & Related papers  (2025-02-17T11:11:09Z)
• Estimating LLM Uncertainty with Evidence [66.51144261657983]
  We present Logits-induced token uncertainty (LogTokU) as a framework for estimating decoupled token uncertainty in Large Language Models.<n>We employ evidence modeling to implement LogTokU and use the estimated uncertainty to guide downstream tasks.
  arXiv  Detail & Related papers  (2025-02-01T03:18:02Z)
• Uncertainty is Fragile: Manipulating Uncertainty in Large Language Models [79.76293901420146]
  Large Language Models (LLMs) are employed across various high-stakes domains, where the reliability of their outputs is crucial. Our research investigates the fragility of uncertainty estimation and explores potential attacks. We demonstrate that an attacker can embed a backdoor in LLMs, which, when activated by a specific trigger in the input, manipulates the model's uncertainty without affecting the final output.
  arXiv  Detail & Related papers  (2024-07-15T23:41:11Z)
• Rigorous Probabilistic Guarantees for Robust Counterfactual Explanations [80.86128012438834]
  We show for the first time that computing the robustness of counterfactuals with respect to plausible model shifts is NP-complete. We propose a novel probabilistic approach which is able to provide tight estimates of robustness with strong guarantees.
  arXiv  Detail & Related papers  (2024-07-10T09:13:11Z)
• Decomposing Uncertainty for Large Language Models through Input Clarification Ensembling [69.83976050879318]
  In large language models (LLMs), identifying sources of uncertainty is an important step toward improving reliability, trustworthiness, and interpretability. In this paper, we introduce an uncertainty decomposition framework for LLMs, called input clarification ensembling. Our approach generates a set of clarifications for the input, feeds them into an LLM, and ensembles the corresponding predictions.
  arXiv  Detail & Related papers  (2023-11-15T05:58:35Z)
• Improving the Reliability of Large Language Models by Leveraging Uncertainty-Aware In-Context Learning [76.98542249776257]
  Large-scale language models often face the challenge of "hallucination" We introduce an uncertainty-aware in-context learning framework to empower the model to enhance or reject its output in response to uncertainty.
  arXiv  Detail & Related papers  (2023-10-07T12:06:53Z)
• A Survey on Uncertainty Toolkits for Deep Learning [3.113304966059062]
  We present the first survey on toolkits for uncertainty estimation in deep learning (DL) We investigate 11 toolkits with respect to modeling and evaluation capabilities. While the first two provide a large degree of flexibility and seamless integration into their respective framework, the last one has the larger methodological scope.
  arXiv  Detail & Related papers  (2022-05-02T17:23:06Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.