AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Diagnosing and Remedying Knowledge Deficiencies in LLMs via Label-free Curricular Meaningful Learning

Kai Xiong, Xiao Ding, Yixin Cao, Li Du, Jiahao Ying, Yang Zhao, Bing Qin, Ting Liu

ICLR 2026 · Large language models › Retrieval and knowledge · approach: Knowledge-augmented question answering · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

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 curricular ordering of synthesized data (ablation)LaMer on Mistral-7B · Seven-benchmark averageaccuracy: 55.64 -> 56.09Improved
add curricular ordering of synthesized data (ablation)LaMer on LLaMA-3-8B · Seven-benchmark averageaccuracy: 64.47 -> 64.70Improved
add curricular ordering of synthesized data (ablation)LaMer on Qwen2-7B · Seven-benchmark averageaccuracy: 65.50 -> 66.13Improved
add curricular ordering of synthesized data (ablation)LaMer on Gemma-1.1-2B · Seven-benchmark averageaccuracy: 33.80 -> 34.08Improved
use relative entropy instead of perplexity for diagnosisdeficiency diagnosis · e-CARE (Mistral-7B-Instruct-v0.2)F1: 40.10 -> 49.58Improved
use relative entropy instead of random selectiondeficiency diagnosis · e-CARE (Mistral-7B-Instruct-v0.2)F1: 35.23 -> 49.58Improved
use LaMer instead of Naive augmentationMistral-7B · Seven-benchmark averageaccuracy: 52.52 -> 56.09Improved
apply Naive augmentationLLaMA-3-8B · Seven-benchmark averageaccuracy: 62.82 -> 61.90Hurt
apply Single augmentationGemma-1.1-2B · Seven-benchmark averageaccuracy: 33.07 -> 31.51Hurt
apply AugGPT augmentationLLaMA-3-8B · GSM-Plusaccuracy: 61.93 -> 14.92Hurt
apply AugGPT augmentationMistral-7B · GSM-Plusaccuracy: 29.91 -> 13.78Hurt
use LaMer instead of label-reliant LLM2LLMMistral-7B · Seven-benchmark averageaccuracy: 54.00 -> 56.09Improved

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