AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

REFO: Reinforced Evolutionary Faithfulness Optimization for Large Language Models

Yi Wang, Xiaqiang Tang, Keyu Hu, Haojie Lu, Sihong Xie

AAAI 2026 · Large language models › Retrieval and knowledge · approach: Knowledge editing in LLMs · 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
apply REFO iterative self-evolutionLlama-3-8B · NQ-Swapspan EM: 73.54% -> 90.08%Improved
apply REFOLlama-3-8B · ConFiQA-MCspan EM: 33.68% -> 74.03%Improved
apply REFOMistral-7B · Memo-Trapspan EM: 34.34% -> 54.75%Improved
apply REFOQwen2.5-7B · FollowBenchCSL: 2.90 -> 2.92No clear effect
apply REFOLlama-3-8B · FollowBenchCSL: 2.75 -> 2.81Improved
add attention-guided DPO loss (Att-REFO vs REFO) (ablation)REFO Llama-3-8B · ELI5ROUGE-1 F1: 30.97% -> 34.03%Improved
add attention-guided DPO loss (ablation)REFO Qwen2.5-7B · ConFiQA-MCspan EM: 43.18% -> 50.38%Improved
add attention-guided DPO loss (ablation)REFO Mistral-7B · Memo-Trapspan EM: 54.75% -> 42.73%Hurt
add attention-guided DPO loss (ablation)REFO Llama-3-8B · NQ-Swapspan EM: 90.08% -> 88.22%Mixed
add attention-guided DPO loss (training speed) (ablation)REFO · avg NQ, NQ-Swap, MemoTrapspan EM per iterationImproved
add relative preference optimization (Step 2) (ablation)REFO Llama and Mistral · avg NQ, NQ-Swap, MemoTrapspan EMImproved
add faithfulness score model filtering (ablation)REFO Llama and Qwen · avg NQ, NQ-Swap, MemoTrapspan EMMixed

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