Image Classification on ImageNet-1K (Top-1 Accuracy, Throughput)
83.9Top-1 AccuracyVMamba-B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| VMamba-BToken Mixer=Mamba+2D Conv., Param. (M)=892026.03 | 83.9 | 315 | |
| STAR-B*††Token Mixer=Mamba, Param. (M)=85, Tail class token=true, Pre-training epochs=16002026.03 | 83.5 | 1,301 | |
| RegNetY-16GToken Mixer=2D Conv., Param. (M)=842026.03 | 82.9 | 870 | |
| STAR-B*Token Mixer=Mamba, Param. (M)=85, Tail class token=true2026.03 | 82.9 | 1,301 | |
| DeiT-BToken Mixer=Attention, Param. (M)=212026.03 | 81.2 | 1,073 | |
| Vim-BToken Mixer=Mamba, Param. (M)=982026.03 | 81.2 | 890 | |
| MambaMLP-BToken Mixer=Mamba, Param. (M)=852026.03 | 81.2 | 1,301 |