AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Neural Prompt Search

Yuanhan Zhang, Kaiyang Zhou, Ziwei Liu

TPAMI 2024 · Architectures and efficiency › Neural architecture search · approach: Bi-level differentiable NAS · 163 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
search jointly over Adapter, LoRA and VPT design instead of using LoRA aloneLoRA · VTAB-1k averagetop-1 accuracy: 74.5 -> 75.5Improved
search combined prompt design instead of VPT aloneVPT · VTAB-1k averagetop-1 accuracy: 72.0 -> 75.5Improved
search combined prompt design instead of Adapter aloneAdapter · VTAB-1k averagetop-1 accuracy: 73.9 -> 75.5Improved
search design instead of full fine-tuningFull · VTAB-1k averagetop-1 accuracy: 68.9 -> 75.5Improved
search design instead of LoRA on Specialized groupLoRA · VTAB-1k Specializedtop-1 accuracy: 84.6 -> 84.9No clear effect
search design instead of LoRA on dSpr-LocLoRA · VTAB-1k dSpr-Loctop-1 accuracy: 75.7 -> 81.8Improved
search design instead of LoRA on EuroSATLoRA · VTAB-1k EuroSATtop-1 accuracy: 95.3 -> 95.4No clear effect
use NOAH instead of LoRA for domain generalizationLoRA · ImageNet-V2top-1 accuracy: 59.3 -> 66.1Improved
use NOAH instead of LoRA for domain generalizationLoRA · ImageNet-Atop-1 accuracy: 6.9 -> 11.9Improved
use NOAH instead of LoRA on source ImageNetLoRA · ImageNet (16-shot)top-1 accuracy: 70.8 -> 71.5Improved
retrain extracted subnet instead of inheriting supernet weights (ablation)NOAH (inherited) · VTAB-1k averagetop-1 accuracy: 74.1 -> 75.5Improved
retrain subnet on Structured group (ablation)NOAH (inherited) · VTAB-1k Structuredtop-1 accuracy: 58.7 -> 61.3Improved

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