AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Feeling of Knowing in Large Language Models

Zichuan Fu, Xian Wu, Jingtong Gao, Wenlin Zhang, Jiaxuan Li, Binhao Wang, Yimin Deng, Guojing Li, Xiaopeng Li, Derong Xu, Yefeng Zheng, Xiangyu Zhao

NeurIPS 2026 · Large language models › Retrieval and knowledge · approach: Knowledge-augmented question answering · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Studies the feeling of knowing (FoK), an LLM's self-assessment of whether it knows an answer, and how post-training updates affect its reliability for selective generation, retrieval triggering, and model routing.

Problem

LLMs are frequently updated via post-training, and their self-assessment of whether they know what they know (feeling of knowing) can become miscalibrated, which harms selective generation, retrieval triggering, and model routing.

Approach

Frames post-training-induced changes in the feeling of knowing as a reliability problem for LLM self-assessment; the abstract provides only the motivation, so the specific method is not stated.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
apply vanilla knowledge SFT (K-SFT) updateQwen2.5-7B · TriviaQA held-outFoK AUROC: 0.688 -> 0.519Hurt
apply vanilla K-SFT updateQwen2.5 and LLaMA models · 7 models, TriviaQAFoK AUROC: Qwen2.5-3B -0.220Hurt
add judgment SFT (J-SFT) after the knowledge updateQwen2.5-7B · TriviaQAFoK AUROC: -0.042 -> +0.100 change from BaseImproved
use on-policy SFT supervising only corrected answer-relevant suffix instead of vanilla K-SFT (ablation)Qwen2.5-7B · TriviaQA held-outFoK AUROC: 0.519 -> 0.603Improved
add knowledge-delta FoK training (KΔ-FT) after OP-SFT (ablation)Qwen2.5-7B · TriviaQA held-outFoK AUROC: 0.603 -> 0.770Improved
add FGPO after KΔ-FT (ablation)Qwen2.5-7B · TriviaQA held-outFoK AUROC: 0.770 -> 0.809Improved
add FGPO after KΔ-FT (ablation)Qwen2.5-7B · TriviaQA held-outanswer accuracy: 0.699 -> 0.692No clear effect
retrain the FoK channel each round instead of freezing after round 1 (ablation)KΔ-FT · three sequential updatesFoK AUROC: 0.754 -> 0.797Improved
run three sequential knowledge updatesQwen2.5-7B · TriviaQA held-outanswer accuracy: 0.672 -> 0.661No clear effect
apply a MEMIT knowledge editQwen2.5-7B · MEMIT zsRE editFoK AUROC on edited facts: 0.594 -> 0.553Hurt
add FGPO policy training (ablation)Qwen3-8B (thinking) · coverage-accuracy curvesthreshold sensitivityMixed
use post-hoc estimators on the base model with no knowledge update (ablation)Base + EliCal · TriviaQA held-outFoK AUROC: 0.810 vs 0.809 for full methodNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy