AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Curriculum Reinforcement Learning for Black-Box Prompt Tuning via Large Language Models

Shuai Gong, Chaoran Cui, Xiaolin Dong, Chunyun Zhang, Linwei Fan

ICML 2026 · Large language models › Language, safety and interpretability · approach: In-context learning analysis

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 imitation loss Lim (ablation)CRL-BPT · Caltech/Flowers/UCF 16-shotaverage accuracy: 77.6 -> 79.7Improved
add innovation loss Lin (ablation)CRL-BPT · Caltech/Flowers/UCF 16-shotaverage accuracy: 78.0 -> 79.7Improved
add both imitation and innovation losses (ablation)CRL-BPT · Caltech/Flowers/UCF 16-shotaverage accuracy: 76.2 -> 79.7Improved
add historical loss normalization (ablation)CRL-BPT · Caltech/Flowers/UCF 16-shotaverage accuracy: 77.4 -> 79.7Improved
add relative reward calibration (ablation)CRL-BPT · Caltech/Flowers/UCF 16-shotaverage accuracy: 77.8 -> 79.7Improved
add both HLN and RRC (ablation)CRL-BPT · Caltech/Flowers/UCF 16-shotaverage accuracy: 76.6 -> 79.7Improved
use exponential schedule instead of linear schedule (ablation)CRL-BPT · Caltech/Flowers/UCF 16-shotaverage accuracy: 78.2 -> 79.7Improved
use decaying curriculum schedule instead of constant weights (ablation)CRL-BPT · Caltech/Flowers/UCF 16-shotaverage accuracy: 76.9 -> 79.7Improved
use top-quality references (Top-20) instead of bottom-quality (Bottom-20) (ablation)CRL-BPT · Flowers/SUN/UCFaverage accuracy: 68.7 -> 70.2Improved
augment reference set with self-generated high-performing prompts (ablation)CRL-BPT · Flowers/SUN/UCFaverage accuracy: 70.2 -> 69.2Hurt
enlarge reference set from 20 to 40 prompts (ablation)CRL-BPT · Flowers/SUN/UCFaverage accuracy: 70.2 -> 69.5No clear effect
use in-context learning with references instead of imitation loss (ablation)CRL-BPT · Flowers102 convergenceaccuracy after 2,000 queries: around 72% vs 74.0%Hurt

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