AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Searching the Search Space of Vision Transformer

Minghao Chen, Kan Wu, Bolin Ni, Houwen Peng, Bei Liu, Jianlong Fu, Hongyang Chao, Haibin Ling

NeurIPS 2021 · Architectures and efficiency › Neural architecture search · approach: Latency-aware architecture search · 1 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
evolve search space from A to B (one step) (ablation)ViT search space · ImageNet-1k, top searched archtop-1 accuracy (%): 65.5 -> 78.0Improved
evolve search space from B to C (two steps) (ablation)ViT search space · ImageNet-1k, top searched archtop-1 accuracy (%): 78.0 -> 82.1Improved
evolve search space from A to C (ablation)ViT search space · ImageNet-1k, top searched archtop-1 accuracy (%): 65.5 -> 82.1Improved
use inherited supernet weights vs finetuning or retraining from scratch (ablation)sampled subnets (10 random) · ImageNet-1ktop-1 accuracyNo clear effect
replace Swin-T with S3-Tbackbone · ImageNet-1ktop-1 accuracy (%): 81.3 -> 82.1Improved
replace Swin-S with S3-Sbackbone · ImageNet-1ktop-1 accuracy (%): 83.0 -> 83.7Improved
replace Swin-B with S3-Bbackbone · ImageNet-1ktop-1 accuracy (%): 83.3 -> 84.0Improved
increase input resolution from 224 to 384 for S3-SS3-S · ImageNet-1ktop-1 accuracy (%): 83.7 -> 84.5Improved
increase input resolution from 224 to 384 for S3-BS3-B · ImageNet-1ktop-1 accuracy (%): 84.0 -> 84.7Improved
replace Swin-S with S3-S as detection backboneCascade Mask R-CNN 3x · COCObox AP: 51.8 -> 52.6Improved
replace Swin-T with S3-T as detection backboneCascade Mask R-CNN 3x · COCObox AP: 50.4 -> 50.5No clear effect
replace Swin-S with S3-SUperNet · ADE20KmIoU (%): 47.60 -> 48.04Improved

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