AI Paper Atlas

Research map / Large language models / Alignment and preferences

Closing the Expression Gap in LLM Instructions via Socratic Questioning

Jianwen Sun, Yukang Feng, Yifan Chang, Chuanhao Li, Zizhen Li, Jiaxin Ai, Fanrui Zhang, Yu Dai, Kaipeng Zhang

ICML 2026 · Large language models › Alignment and preferences · approach: LLM-as-judge and feedback data · 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 information-gain entropy reward instead of slot-counting reward (ablation)Nous (OfG) · Diagram specification dialogue (100 test diagrams)total information gain: 97.11 -> 120.53Improved
add information-gain entropy reward instead of slot-counting reward (ablation)Nous (OfG) · Final chart quality, VisPainterweighted score: 0.63 -> 0.76Improved
add information-gain entropy reward instead of slot-counting reward (ablation)Nous (OfG) · Diagram specification dialogue (100 test diagrams)dialogue turns: 13.6 -> 20.3Hurt
use offline GRPO instead of SFTNous · Diagram specification dialogue (100 test diagrams)total information gain: 94.3 -> 120.5Improved
use offline GRPO instead of online GRPONous · Diagram specification dialogue (100 test diagrams)training resource hours: 93 -> 28.5Improved
use offline GRPO instead of online GRPONous · Diagram specification dialogue (100 test diagrams)total information gain: 115.8 -> 120.5Improved
use offline GRPO instead of DPONous · Diagram specification dialogue (100 test diagrams)total information gain: 111.3 -> 120.5Improved
train on Vague Oracle dialogues instead of Template Oracle (ablation)Nous (OfG) · Diagram specification dialogue (100 test diagrams)total information gain: 120.5 -> 117.1No clear effect
train on Noisy Oracle dialogues instead of Template Oracle (ablation)Nous (OfG) · Diagram specification dialogue (100 test diagrams)total information gain: 120.5 -> 115.8No clear effect
start from a partial draft instead of from scratchNous (OfG) · Human usersdialogue turns: 17.3 -> 7.6Improved
human users instead of simulated oracleNous (OfG) · Human users, zero-startdialogue turns: 20.3 -> 17.3Improved
apply Nous (OfG) in novel writing instead of GPT-zeroNous (OfG) · Novel writing generalizationtotal information gain: 55.2 -> 65.4Improved

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