AI Paper Atlas

Research map / Large language models / Alignment and preferences

Targeting Misalignment: A Conflict-Aware Framework for Reward-Model-based LLM Alignment

Zixuan Liu, Siavash H. Khajavi, Guangkai Jiang, Xinru Liu

AAAI 2026 · Large language models › Alignment and preferences · approach: Steering and correcting LLM behavior · 1 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
refine reward with human feedback on high-conflict pairs (SHF-CAS) instead of vanilla proxyPPO · PKU-SafeRLHFgold reward (lower safer): 3.92 -> -2.90Improved
refine reward with human feedback on high-conflict pairs (SHF-CAS)PPO · Anthropic HH-RLHFgold reward: -2.32 -> 3.36Improved
select high-conflict pairs instead of random pairs (ablation)SHF-CAS · PKU-SafeRLHF (delta=1.6)gold reward (lower safer): -0.64 -> -2.90Improved
select high-conflict pairs instead of random pairs (ablation)SHF-CAS · HH-RLHF (delta=1.5)gold reward: 1.03 -> 3.36Improved
raise PACS threshold delta (fewer, more extreme conflicts) (ablation)SHF-CAS · PKU-SafeRLHFgold reward (lower safer): 1.59 -> -2.90 (delta 1.4 -> 1.6)Improved
raise PACS threshold delta (ablation)SHF-CAS · HH-RLHFgold reward: 2.33 -> 3.36 (delta 1 -> 1.5)Improved
use SHF-CAS instead of RSORSO · PKU-SafeRLHFgold reward (lower safer): 2.84 -> -2.90Improved
use SHF-CAS instead of RSORSO · HH-RLHFgold reward: -1.02 -> 3.36Improved
use GPT-4o oracle instead of reward-model oracle (ablation)SHF-CAS · PKU-SafeRLHF (delta=1.6)gold reward (lower safer): -2.90 -> -3.53Mixed
use GPT-4o oracle instead of reward-model oracle (ablation)SHF-CAS · HH-RLHF (delta=1.5)gold reward: 3.36 -> 3.31No 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