Research map / Architectures and efficiency / Attention and transformers
EfficientFormer: Vision Transformers at MobileNet Speed
Yanyu Li, Geng Yuan, Yang Wen, Eric Hu, Georgios Evangelidis, Sergey Tulyakov, Yanzhi Wang, Jian Ren
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add 3D MHSA and latency-driven search (ablation) | EfficientFormer-L3 · ImageNet-1K | top-1 accuracy: 80.3 -> 82.4 | Improved |
| use conv stem instead of non-overlap patch embedding (ablation) | EfficientFormer-L3 (V1) · ImageNet-1K | iPhone latency: 5.8 -> 2.5 | Improved |
| use GN instead of BN in 4D blocks (ablation) | EfficientFormer-L3 (V1) · ImageNet-1K | top-1 accuracy: 80.3 -> 80.6 | Improved |
| use LN instead of BN in 4D blocks (ablation) | EfficientFormer-L3 (V1) · ImageNet-1K | iPhone latency: 2.5 -> 3.7 | Hurt |
| use ReLU instead of GeLU (ablation) | EfficientFormer-L3 (V1) · ImageNet-1K | top-1 accuracy: 80.3 -> 79.3 | Hurt |
| use HardSwish instead of GeLU (ablation) | EfficientFormer-L3 (V1) · ImageNet-1K | iPhone latency: 2.5 -> 32.4 | Hurt |
| dimension-consistent design vs non-D-C with reshapes (ablation) | EfficientFormer-L7 · A100 TensorRT | latency | Improved |
| latency-driven slimming vs random configs at same GMACs (ablation) | EfficientFormer-L1 · ImageNet-1K | accuracy/latency | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- LeViT-256 — LeViT: A Vision Transformer in ConvNet's Clothing for Faster Inference (ICCV 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- AutoViT: Achieving Real-Time Vision Transformers on Mobile via Latency-aware Coarse-to-Fine Search (IJCV 2025 · similarity 0.92)
- A Survey on Efficient Vision Transformers: Algorithms, Techniques, and Performance Benchmarking (TPAMI 2024 · similarity 0.91)
- RepViT: Revisiting Mobile CNN From ViT Perspective (CVPR 2024 · similarity 0.91)
- Real-time Core-Periphery Guided ViT with Smart Data Layout Selection on Mobile Devices (NeurIPS 2024 · similarity 0.90)
- SHViT: Single-Head Vision Transformer with Memory Efficient Macro Design (CVPR 2024 · similarity 0.88)
- Mobile Attention: Mobile-Friendly Linear-Attention for Vision Transformers (ICML 2024 · similarity 0.88)
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.
