# Fine-Tuned LLMs Know They Don't Know: A Parameter-Efficient Approach to Recovering Honesty

arXiv: https://arxiv.org/abs/2511.12991 · HTML: https://mlsanity.com/paper/2511.12991/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 30 references that are in the map.
- First posted on arXiv 17 Nov 2025.

## Builds on (most cited first, 18 of 18)

- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [A General Language Assistant as a Laboratory for Alignment](https://mlsanity.com/paper/2112.00861/) (2021)
- [Knowledge Neurons in Pretrained Transformers](https://mlsanity.com/paper/2104.08696/) (2021)
- [ORPO: Monolithic Preference Optimization without Reference Model](https://mlsanity.com/paper/2403.07691/) (2024)
- [Improving alignment of dialogue agents via targeted human judgements](https://mlsanity.com/paper/2209.14375/) (2022)
- [SaySelf: Teaching LLMs to Express Confidence with Self-Reflective Rationales](https://mlsanity.com/paper/2405.20974/) (2024)
- [Medical Hallucinations in Foundation Models and Their Impact on Healthcare](https://mlsanity.com/paper/2503.05777/) (2025)
- [Catastrophic Failure of LLM Unlearning via Quantization](https://mlsanity.com/paper/2410.16454/) (2024)
- [Confidence Regulation Neurons in Language Models](https://mlsanity.com/paper/2406.16254/) (2024)
- [What does the Knowledge Neuron Thesis Have to do with Knowledge?](https://mlsanity.com/paper/2405.02421/) (2024)
- [IRCAN: Mitigating Knowledge Conflicts in LLM Generation via Identifying and Reweighting Context-Aware Neurons](https://mlsanity.com/paper/2406.18406/) (2024)
- [NLSR: Neuron-Level Safety Realignment of Large Language Models Against Harmful Fine-Tuning](https://mlsanity.com/paper/2412.12497/) (2024)
- [Examining LLMs' Uncertainty Expression Towards Questions Outside Parametric Knowledge](https://mlsanity.com/paper/2311.09731/) (2023)
- [Fine-Tuning is Fine, if Calibrated](https://mlsanity.com/paper/2409.16223/) (2024)
- [On the Loss of Context-awareness in General Instruction Fine-tuning](https://mlsanity.com/paper/2411.02688/) (2024)
- [Smoothing Out Hallucinations: Mitigating LLM Hallucination with Smoothed Knowledge Distillation](https://mlsanity.com/paper/2502.11306/) (2025)

## Built on by (most cited first, 3 of 3)

- [Probing the Knowledge Boundary: An Interactive Agentic Framework for Deep Knowledge Extraction](https://mlsanity.com/paper/2602.00959/) (2026)
- [Introspecting Alignment Shifts Beyond Behaviors Implanted Through Fine-Tuning](https://mlsanity.com/paper/2608.04347/) (2026)
- [The Commit-Abstain Circuit: Why Language Models Hallucinate Instead of Abstaining](https://mlsanity.com/paper/2609.32964/) (2026)

## Abstract (arXiv)

The honesty of Large Language Models (LLMs) is increasingly important for safe deployment in high-stakes domains. However, this crucial trait is severely undermined by supervised fine-tuning (SFT), a common technique for model specialization. Existing recovery methods rely on data-intensive global parameter adjustments, implicitly assuming that SFT deeply corrupts the models' ability to recognize their knowledge boundaries. However, we observe that fine-tuned LLMs still preserve this ability; what is damaged is their capacity to faithfully express that awareness. Building on this, we propose Honesty-Critical Neurons Restoration (HCNR) to surgically repair this suppressed capacity. HCNR identifies and restores key expression-governing neurons to their pre-trained state while harmonizing them with task-oriented neurons via Hessian-guided compensation. Experiments on four QA tasks and five LLM families demonstrate that HCNR effectively recovers 33.25% of the compromised honesty while achieving at least 2.23x speedup with over 10x less data compared to baseline methods, offering a practical solution for trustworthy LLM deployment.