AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Learning from Scoring Disagreements: Contrastive Error Mining for Efficient and Robust LLM-based Assessment

Lei Chen, Tengteng Cheng, BoYu Gao, Zitao Liu, Weiqi Luo

AAAI 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning for reasoning data · 0 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 contrastive hard-sample LoRA tuning after full fine-tuningQwen3-4B · SciEntsbank Unseen AnswersACC: 0.7593 -> 0.7685Improved
add contrastive hard-sample LoRA tuning after full fine-tuningQwen3-4B · SciEntsbank Unseen QuestionsACC: 0.6248 -> 0.6968Improved
add contrastive hard-sample LoRA tuning after full fine-tuningQwen3-4B · SciEntsbank Unseen DomainsQWK: 0.6183 -> 0.6309Improved
add contrastive hard-sample LoRA tuning after full fine-tuningQwen3-4B · Beetle Unseen AnswersACC: 0.7677 -> 0.7995Improved
add contrastive hard-sample LoRA tuning after full fine-tuningQwen3-4B · Beetle Unseen QuestionsQWK: 0.6549 -> 0.6554Improved
add contrastive hard-sample LoRA tuning after full fine-tuningQwen3-1.7B · Mohler Unseen AnswersMAE: 0.4216 -> 0.4019Improved
add contrastive hard-sample LoRA tuning after full fine-tuningQwen3-1.7B · Mohler Unseen QuestionsMAE: 0.5116 -> 0.5109No clear effect
full fine-tuning instead of few-shot promptingQwen3-4B · SciEntsbank Unseen AnswersACC: 0.5241 -> 0.7593Improved
apply CEM-FT across five Qwen base modelsQwen3-0.6B to Qwen3-8B · Mohler (average over tasks)TAA: improves by 1.78% to 2.98%Improved
raise beta from 1 to 2 (alpha=1) in hard-sample threshold (ablation)CEM-FT · SciEntsbankACC: 0.6891 -> 0.7003Improved
raise beta from 1 to 2 (alpha=1) in hard-sample threshold (ablation)CEM-FT · BeetleACC: 0.7311 -> 0.7447Improved
raise beta from 1 to 1.5 (alpha=1) in hard-sample threshold (ablation)CEM-FT · MohlerMAE: 0.4584 -> 0.4564Improved

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