AI Paper Atlas

Research map / Large language models / Alignment and preferences

MENLO: From Preferences to Proficiency – Evaluating and Modeling Native-like Quality Across 47 Languages

Chenxi Whitehouse, Sebastian Ruder, Tony Zhiyang Lin, Oksana Kurylo, Haruka Takagi, Janice Lam, Nicolò Busetto, Denise Diaz

ICLR 2026 · Large language models › Alignment and preferences · approach: LLM-as-judge and feedback data

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
grade two responses side by side (pairwise) instead of one at a time (ablation)zero-shot LLM judges · MENLO testPreference accuracy: Qwen3-4B 40.54 -> 57.13Improved
add three graded few-shot examples to pointwise judging (ablation)LLM judges · MENLO testPreference accuracy: Qwen3-4B 40.54 -> 39.35No clear effect
add detailed 5-point grading rubrics (ablation)pointwise LLM judges · MENLO testMacro-F1: Qwen3-4B 16.00 -> 23.06Improved
add rubrics to pairwise judging (ablation)pairwise LLM judges · MENLO testMacro-F1: Qwen3-32B 38.10 -> 37.48No clear effect
train judge with SFT (no CoT) instead of zero-shot (ablation)Qwen3-4B · MENLO testMacro-F1: 35.46 -> 33.44Hurt
train judge with SFT instead of zero-shot (ablation)Llama4-Scout · MENLO testMacro-F1: 36.11 -> 44.17Improved
train judge with GRPO RL instead of zero-shot (ablation)Qwen3-4B · MENLO testMacro-F1: 35.46 -> 39.44Improved
initialize RL from the SFT checkpoint (ablation)Qwen3-4B / Llama4-Scout · MENLO testMacro-F1: 39.44 -> 39.33 (Qwen3-4B)No clear effect
add reward smoothing alone to binary reward (ablation)Qwen3-4B RL judge · MENLO testPreference accuracy: 58.27 -> 51.47Hurt
add smoothing and preference bonus to binary reward (ablation)Qwen3-4B RL judge · MENLO testMacro-F1: 37.11 -> 39.44Improved
train RL on English data only (ablation)Qwen3-4B RL judge · 47 language varietiesperformanceHurt
use RL on Localized Factuality dimension (ablation)Qwen3-4B · Localized FactualityMacro-F1: +0.6 single-task RL; regressions multitaskNo 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