AI Paper Atlas

Research map / Large language models / Alignment and preferences

mR3: Multilingual Rubric-Agnostic Reward Reasoning Models

David Anugraha, Shou-Yi Hung, Zilu Tang, En-Shiun Annie Lee, Derry Tanti Wijaya, Genta Indra Winata

ICLR 2026 · Large language models › Alignment and preferences · approach: LLM-as-judge and feedback data · 9 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
multilingual SFT on 100k rubric-based reasoning dataQwen3-14B · m-RewardBenchaccuracy: 88.46 -> 89.18Improved
multilingual SFTQwen3-4B · MM-Evalaccuracy: 80.07 -> 82.62Improved
multilingual SFTQwen3-14B · RTP-LXF1: 78.36 -> 90.19Improved
multilingual SFT versus English-only R3 trainingQwen3-14B · MM-Evalaccuracy: 84.04 -> 86.05Improved
multilingual SFT versus English-only R3 trainingQwen3-14B · RewardBench (English)accuracy: 91.00 -> 90.79Hurt
scale model from 4B to 14BmR3 · m-RewardBenchaccuracy: 87.61 -> 89.18Improved
use English reasoning instead of target-language reasoning (ablation)mR3 SFT · pairwise benchmarksaverage accuracyImproved
train with target-language reasoning (ablation)mR3 SFT · pairwise benchmarksaccuracyImproved
easy-to-hard curriculum vs random and other orderings (ablation)mR3 SFT · HelpSteer3 validationaccuracyImproved
DPO with mR3-Qwen3-14B rewardQwen3-30B-A3B-Instruct-2507 · m-ArenaHard-v2.0 (23 langs)win rate: 39.1 -> 45.2Improved
DPO with mR3-Qwen3-14B rewardQwen3-30B-A3B-Instruct-2507 · MCLM MT-MATH100accuracy: 90.47 -> 92.08Improved
DPO with mR3 instead of Nemotron-Multilingual-49B rewardQwen3-30B-A3B-Instruct-2507 · m-ArenaHard-v2.0 (23 langs)win rate: 47.0 -> 45.2Hurt

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