Image Classification on ImageNet-1K 1.0 (test) (Accuracy and Energy Efficiency)
85.7AccuracyQKFormer
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| QKFormerHardware technology=14nm Loihi-class, Effective dimensions (deff)=892026.04 | 85.7 | 0.4 | 44 | |
| SpikingResformerHardware technology=14nm Loihi-class, Effective dimensions (deff)=892026.04 | 85 | 0.35 | 50 | |
| ViT-B/16Hardware technology=14nm Loihi-class, Effective dimensions (deff)=892026.04 | 84.5 | 17.6 | 1 | |
| Theo. Min.Hardware technology=14nm Loihi-class, Effective dimensions (deff)=892026.04 | 84.5 | 0.16 | 110 | |
| Meta-SpikeFormerHardware technology=14nm Loihi-class, Effective dimensions (deff)=892026.04 | 80 | 0.31 | 57 | |
| Spikformer-8-512Hardware technology=14nm Loihi-class, Effective dimensions (deff)=892026.04 | 74.8 | 0.46 | 38 |