AI Paper Atlas

Research map / Probabilistic and scientific modeling / Bayesian optimization

Online Black-Box Prompt Optimization with Regret Guarantees under Noisy Feedback

Jinjie Fang, Runwen You, Wanli Shi, Wenkang Wang, Ganyu Wang, Haozhen Zhang, Yi Chang, Bin Gu

ICLR 2026 · Probabilistic and scientific modeling › Bayesian optimization · approach: Offline black-box optimization with surrogates

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 adaptive uncertainty scale adjustment on top of online zeroth-order tuning (ablation)AOZPT · Anime / Painting, Dreamlike-2.0aesthetic score: 6.173 -> 6.656 (Painting), 5.861 -> 6.282 (Anime)Improved
add online zeroth-order tuning alone (without scale adjustment) (ablation)AOZPT · Anime / Painting, Dreamlike-2.0aesthetic score: 5.855 -> 5.861 (Anime), 6.179 -> 6.173 (Painting)No clear effect
add adaptive scale adjustment (w/ delta1 vs none) (ablation)AOZPT, Llama-3.1-8B · CNN/DailyMailF1: 23.500 -> 25.089 -> 27.966Improved
add adaptive scale adjustment (w/ delta1 vs none) (ablation)AOZPT, Llama-3.1-8B · GSM8Kaccuracy: 69.267 -> 74.000 -> 75.533Improved
use frozen open-source LLM to generate discrete prompts instead of mapping soft prompts to vocabulary (ablation)AOZPT · CNN/DailyMail, Llama-3.1-8BF1: 9.911 -> 24.707Improved
use frozen open-source LLM to generate discrete prompts instead of direct vocabulary mapping (ablation)AOZPT · CNN/DailyMail, Qwen2.5-14BF1: 21.84 -> 24.767Improved
AOZPT versus ZO-OGD (plain online zeroth-order GD)online prompt tuning · CNN/DailyMail and GSM8K, three LLMsF1 / accuracy: e.g. GSM8K GPT-3.5-turbo 69.533 -> 78.133; CNN/DM Llama 24.667 -> 24.707Improved
AOZPT versus BDPL on GSM8K with Qwen2.5-14Bonline prompt tuning · GSM8K, Qwen2.5-14Baccuracy: 89.000 -> 92.933Improved
AOZPT versus ZO-OGD under data drift with window L=150Stable Diffusion v1.5 · Anime/Painting data driftaesthetic score: 6.037 -> 6.117 (L=150)Improved
sliding-window AOZPT versus Adam, Nadam, RMSProp (ablation)adaptive gradient update · Anime/Painting, Table 12aesthetic score: Painting Dreamlike: 6.179 (Adam) vs 6.140 (w=10), 6.181 (w=50)Mixed
remove adaptive scale adjustment from single-point estimator variant (ablation)AOZPT single-point · Anime, Stable Diffusion v1.5aesthetic score: 5.815 -> 5.872Improved
AOZPT versus MP/ICL/ZO-OGD on translationLlama-3.1-8B · WMT14 De-EnBLEU: 30.722 -> 32.510 (vs ZO-OGD)Improved

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