AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Know More, Know Clearer: A Meta-Cognitive Framework for Knowledge Augmentation in Large Language Models

Hao Chen, Ye He, Yuchun Fan, Yukun Yan, Zhenghao Liu, Qingfu Zhu, Maosong Sun, Wanxiang Che

ICML 2026 · Large language models › Retrieval and knowledge · approach: Knowledge editing in LLMs · 0 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add differentiated region-specific knowledge augmentation (CGKE) versus vanilla SFTQwen2.5-7B-Instruct · 11 QA benchmarks averageaccuracy: 27.98 -> 29.15Improved
add Knowledge Confused data in CGKE (ablation)CGKE · 11 QA average, Qwen2.5-7B-Instructaccuracy: 28.23 -> 29.15Improved
add Knowledge Missing data in CGKE (ablation)CGKE · 11 QA average, Qwen2.5-7B-Instructaccuracy: 28.38 -> 29.15Improved
add Knowledge Mastered data in CGKE (ablation)CGKE · 11 QA average, Qwen2.5-7B-Instructaccuracy: 29.04 -> 29.15Improved
add cognitive consistency alignment (entropy calibration loss) to GRPO (ablation)CDKC · 11 QA average, Qwen2.5-7B-Instructaccuracy: 43.91 -> 44.97Improved
add CGKE knowledge expansion before calibration (ablation)CDKC · 11 QA average, Qwen2.5-7B-Instructaccuracy: 42.86 -> 44.97Improved
run second expansion-calibration roundCDKC · 11 QA average, Qwen2.5-7B-Instructaccuracy: 44.97 -> 48.29Improved
run second expansion-calibration roundCDKC · 11 QA average, Llama-3.1-8B-Instructaccuracy: 51.44 -> 55.50Improved
apply CDKC calibration instead of GRPOCGKE model · 11 QA average, Llama-3.1-8B-Instructaccuracy: 49.71 -> 51.44Improved
use CDKC instead of retrieval (RAG) on PopQAQwen2.5-7B-Instruct · PopQAaccuracy: 43.44 -> 28.27Hurt
use Know What SFT abstention methodQwen2.5-7B-Instruct · self-knowledge benchmark answerableFN rate: 97.95% FNHurt
apply CDKC calibrationQwen2.5-7B-Instruct · self-knowledge benchmarkboundary recognition (TN): 12.21% increase over vanilla, 63.37% TPImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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