Research map / Architectures and efficiency / Spiking networks
Positional Encoding for Spiking Transformers
Zijian Zhou, Yu Liang, Honglin Cao, Ammar Belatreche, Jieyuan Zhang, Wenjie Wei, Shuai Wang, Malu Zhang, Yang Yang, Haizhou Li
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 PE-LIF position-dependent thresholds (positional encoding) (ablation) | 4-layer Spikingformer · 8 downstream NLP benchmarks avg | score: 66.72 -> 70.31 | Improved |
| add PE-LIF on STS-B (ablation) | 4-layer Spikingformer · STS-B | score: 27.03 -> 55.23 | Improved |
| add MPR-SG surrogate gradient (ablation) | SPE · 8 downstream NLP benchmarks avg | score: 70.31 -> 71.43 | Improved |
| add MPR-SG on QNLI (ablation) | SPE · QNLI | accuracy: 71.06 -> 74.01 | Improved |
| add MPR-SG on STS-B (ablation) | SPE · STS-B | score: 55.23 -> 61.78 | Improved |
| add MPR-SG on SST-5 (ablation) | SPE · SST-5 | accuracy: 43.79/43.48 -> 44.16 | No clear effect |
| replace CPG/Gray/Log-PE with SPE | Spikingformer · Long text (AGNEWS, IMDB) | accuracy avg: 83.62 (Log-PE) -> 91.05 | Improved |
| replace Log-PE with SPE | Spikingformer · Short text avg | accuracy avg: 76.46 -> 78.68 | Improved |
| add SPE to Spikingformer | Spikingformer · GLUE dev avg | score: 66.8 -> 72.6 | Improved |
| add SPE to Spikingformer (energy) | Spikingformer · GLUE | energy (mJ): 6.76 -> 4.92 | Improved |
| add SPE (higher Q/K thresholds) | Spikingformer · attention energy | mJ/sample: 0.0894 -> 0.0666 | Improved |
| add SPE on MNLI-m | Spikingformer · GLUE MNLI-m/mm | accuracy: 71.9/72.5 -> 70.7/72.0 | No clear effect |
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.
- Spikingformer — Spikingformer: A Key Foundation Model for Spiking Neural Networks (AAAI 2026) · better on 4 of 4 shared results
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
- Toward Relative Positional Encoding in Spiking Transformers (NeurIPS 2025 · similarity 0.93)
- Advancing Spiking Neural Networks for Sequential Modeling with Central Pattern Generators (NeurIPS 2024 · similarity 0.89)
- Towards Efficient Spiking Transformer: a Token Sparsification Framework for Training and Inference Acceleration (ICML 2024 · similarity 0.88)
- Spiking Transformer: Introducing Accurate Addition-Only Spiking Self-Attention for Transformer (CVPR 2025 · similarity 0.88)
- STEP: A Unified Spiking Transformer Evaluation Platform for Fair and Reproducible Benchmarking (NeurIPS 2025 · similarity 0.88)
- SpikF: Spiking Fourier Network for Efficient Long-term Prediction (ICML 2025 · similarity 0.87)
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.
