AI Paper Atlas

Research map / Large language models / Alignment and preferences

Towards Self-Robust LLMs: Intrinsic Prompt Noise Resistance via CoIPO

Xin Yang, Letian Li, Abudukelimu Wuerkaixi, Xuxin Cheng, Cao Liu, Ke Zeng, Xunliang Cai, Wenyuan Jiang

ICLR 2026 · Large language models › Alignment and preferences · approach: Active and token-level preference learning · 3 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
train with CoIPO instead of COINAlpaca (Llama-7B) · NoisyPromptBench avgaccuracy (%): 58.60 -> 63.90Improved
train with CoIPO instead of COINQwen2.5-7B · NoisyPromptBench avgaccuracy (%): 81.48 -> 83.45Improved
add inverse DPO to contrastive learning (ablation)CoIPO (Llama) · NoisyPromptBench avgaccuracy (%): 58.60 -> 63.90Improved
add contrastive learning to inverse DPO (ablation)CoIPO (Qwen) · NoisyPromptBench avgaccuracy (%): 82.86 -> 83.45Improved
use CoIPO instead of plain SFT on noisy promptsLlama · Avg accuracy drop under perturbationdrop (%): 3.20 -> 3.88Hurt
use CoIPO instead of COINQwen2.5-7B · Avg accuracy drop under perturbationdrop (%): 1.81 -> 0.54Improved
scale CoIPO to 14B and 72BQwen2.5 · NoisyPromptBench avgaccuracy (%): 83.45 -> 83.93 -> 85.00Improved
apply COIN at larger scaleCOIN on Qwen2.5-14B · NoisyPromptBench avgaccuracy (%): 81.48 (7B) -> 71.86 (14B)Hurt
train with CoIPO (not trained on these tasks)Qwen2.5-14B · GSM8Kexact match: 75.66 -> 84.76Improved
train with CoIPO (not trained on these tasks)Qwen2.5-7B · MBPPPass@1: 35.6 -> 32.8Hurt
train with CoIPOLlama-7B · GSM8Kexact match: 7.20 -> 6.29Hurt

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