AI Paper Atlas

Research map / Large language models / Alignment and preferences

Reward Modeling from Natural Language Human Feedback

Zongqi Wang, Rui Wang, Yuchuan Wu, Yiyao Yu, Pinyi Zhang, Shaoning Sun, Yujiu Yang, Yongbin Li

ICML 2026 · Large language models › Alignment and preferences · approach: Reward modeling for diverse preferences · 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 human-critique process reward to outcome reward (ablation)RM-NLHF-Qwen-7B · 7-benchmark averageaverage accuracy: 0.6158 -> 0.6481Improved
add outcome regularization (no process reward when outcome is wrong) (ablation)RM-NLHF-Qwen-7B · 7-benchmark averageaverage accuracy: 0.6282 -> 0.6481Improved
train MetaRM only on HelpSteer3 instead of also using outcome data D_o (ablation)RM-NLHF-Qwen-7B · 7-benchmark averageaverage accuracy: 0.6332 -> 0.6481Improved
use online MetaRM instead of offline MetaRM (ablation)MetaRM on Qwen-7B · MetaRM ablation overalloverall accuracy: 0.5493 -> 0.5552Improved
use online MetaRM instead of offline MetaRM (ablation)MetaRM on Llama-8B · MetaRM ablation overalloverall accuracy: 0.6273 -> 0.6356Improved
add MetaRM process reward to outcome-only reward (No-MetaRM -> Online-MetaRM) (ablation)GRM on Qwen-7B · MetaRM ablation overalloverall accuracy: 0.5347 -> 0.5552Improved
use only MetaRM reward without outcome signal instead of combining both (ablation)MetaRM on Llama-8B · MetaRM ablation overalloverall accuracy: 0.6356 -> 0.6143Hurt
use only MetaRM reward without outcome signal instead of combining both (ablation)MetaRM on Qwen-7B · MetaRM ablation overalloverall accuracy: 0.5552 -> 0.5579Mixed
update MetaRM aggressively (lr 1e-5, 2 epochs per round) (ablation)MetaRM on Qwen-7B · MetaRM ablation overalloverall accuracy: 0.5552 -> 0.5407Hurt
discretize MetaRM output to binary instead of continuous regression (ablation)MetaRM on Qwen-7B · MetaRM ablation overalloverall accuracy: 0.5552 -> 0.5469Hurt
use classification head instead of regression for MetaRM (ablation)MetaRM on Qwen-7B · MetaRM ablation overalloverall accuracy: 0.5552 -> 0.5400Hurt
use similarity with core human-critique arguments instead of LLM-as-a-Meta-Judge for process reward (ablation)process reward proxy (gpt-5-mini) · 49 annotated HelpSteer3 samplesaccuracy: 0.4898 -> 0.8571Improved

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