Research map / Architectures and efficiency / Spiking networks
Spike-driven Transformer
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace Spikformer attention with SDSA | Spiking Transformer-8-512 · ImageNet-1K | top-1 accuracy: +1.2% | Helped |
| Use SDSA-based spike-driven attention | 8-768 self-attention · ImageNet-1K | energy ratio (E1/E2): 87.2 | Helped |
| Set inference timesteps to T=1 instead of T=4 | Spike-driven Transformer-8-512 · ImageNet-1K | power: 4.50 -> 1.13 | Helped |
| Set inference timesteps to T=1 instead of T=4 | Spike-driven Transformer-8-512 · ImageNet-1K | accuracy | Hurt |
| Add SDSA (ablation) | Spiking Transformer-2-512 · CIFAR-10 | accuracy: 93.12 -> 93.09 (-0.03) | Hurt |
| Add SDSA (ablation) | Spiking Transformer-2-512 · CIFAR-100 | accuracy: 73.17 -> 72.83 (-0.34) | Hurt |
| Add membrane shortcut (ablation) | Spiking Transformer-2-512 · CIFAR-10 | accuracy: 93.12 -> 93.93 (+0.81) | Helped |
| Add membrane shortcut (ablation) | Spiking Transformer-2-512 · CIFAR-100 | accuracy: 73.17 -> 74.63 (+1.46) | Helped |
| Use SDSA and membrane shortcuts together (ablation) | Spiking Transformer-2-512 · CIFAR-10 | accuracy: 93.12 -> 93.82 (+0.73) | Helped |
| Use SDSA and membrane shortcuts together (ablation) | Spiking Transformer-2-512 · CIFAR-100 | accuracy: 73.17 -> 74.41 (+1.24) | Helped |
| Replace soft vanilla attention with binary SDSA hard attention (ablation) | Spike-driven Transformer · CIFAR-10 and CIFAR-100 | accuracy: -0.03; -0.34 | Hurt |
Closest papers by meaning
- Spiking Transformer: Introducing Accurate Addition-Only Spiking Self-Attention for Transformer (CVPR 2025 · similarity 0.92)
- SpikedAttention: Training-Free and Fully Spike-Driven Transformer-to-SNN Conversion with Winner-Oriented Spike Shift for Softmax Operation (NeurIPS 2024 · similarity 0.92)
- Quantized Spike-driven Transformer (ICLR 2025 · similarity 0.91)
- Spike-driven Transformer V2: Meta Spiking Neural Network Architecture Inspiring the Design of Next-generation Neuromorphic Chips (ICLR 2024 · similarity 0.91)
- SpikeVideoFormer: An Efficient Spike-Driven Video Transformer with Hamming Attention and $\mathcal{O}(T)$ Complexity (ICML 2025 · similarity 0.90)
- SpikingResformer: Bridging ResNet and Vision Transformer in Spiking Neural Networks (CVPR 2024 · similarity 0.90)
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.