AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

RIFormer: Keep Your Vision Backbone Effective but Removing Token Mixer

Jiahao Wang, Songyang Zhang, Yong Liu, Taiqiang Wu, Yujiu Yang, Xihui Liu, Kai Chen, Ping Luo, Dahua Lin

CVPR 2023 · Architectures and efficiency › Attention and transformers · approach: Hybrid and interpretable ViT designs · 33 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
Remove the pooling token mixer and train with CE loss (ablation)MetaFormer-S12 · ImageNet-1KTop-1 accuracy (%): 75.01 -> 72.31Hurt
Use soft teacher distillation without ground-truth labels (ablation)RIFormer-S12 with identity mapping · ImageNet-1KTop-1 accuracy (%): 72.31 -> 74.05Improved
Use affine transformation without tailored distillation (ablation)RIFormer-S12 · ImageNet-1KTop-1 accuracy (%): 72.31 -> 72.25No clear effect
Replace identity mapping with affine transformation while using soft distillation without labels (ablation)RIFormer-S12 · ImageNet-1KTop-1 accuracy (%): 74.05 -> 74.07No clear effect
Add module imitation to the affine student (ablation)RIFormer-S12 · ImageNet-1KTop-1 accuracy (%): 74.07 -> 75.13Improved
Use a teacher with a large receptive field, GFNet-H-B, rather than other tested teachers (ablation)RIFormer-S12 · ImageNet-1KTop-1 accuracy (%): 73.63; 73.12; 73.25; 74.07 -> 74.83; 74.52; 74.25; 75.13Mixed
Initialize the student weights from the teacher, excluding the affine operator (ablation)RIFormer-S12 · ImageNet-1KTop-1 accuracy (%): 75.13% to 75.36%Improved
Replace module imitation with hidden-state feature distillation (ablation)RIFormer-S12 · ImageNet-1KTop-1 accuracy (%): 75.36 -> 74.90Hurt
Remove the teacher from training (ablation)RIFormer-S12 · ImageNet-1KTop-1 accuracy (%): 72.75 -> 75.62Hurt
Use RIFormer token-mixer removal instead of depth or width slimming (ablation)PoolFormer-S12 · ImageNet-1KTop-1 accuracy (%): PoolFormer-S9: 74.78; PoolFormer-XS12: 75.11; RIFormer-S12: 75.36Mixed
Remove token mixers from the RIFormer building blocksRIFormer-M36 · ImageNet-1K val, 224Top-1 accuracy (%): 82.1 -> 82.6Improved
Remove token mixers from the RIFormer building blocksRIFormer-S12 · ImageNet-1K val, 224Throughput (images/s): 4160.18 -> 4899.80Improved

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