Model Zoo
March 14, 2024 ยท View on GitHub
ImageNet-1K pretrained (224x224)
| Model | Top-1 | #Param. | FLOPs | Model | Shell |
|---|---|---|---|---|---|
| VideoMamba-Ti | 76.9 | 7M | 1.1G | aliyun, :hugs:HF | run.sh |
| VideoMamba-S | 81.2 | 26M | 4.3G | aliyun, :hugs:HF | run.sh |
| VideoMamba-M w/ SD | 82.8 | 74M | 12.7G | aliyun, :hugs:HF | run.sh |
| VideoMamba-B w/ SD | 82.7 | 98M | 16.9G | aliyun, :hugs:HF | run.sh |
SD refers to Self-Distillation
Large resolution fine-tuning
| Model | Resolution | Top-1 | #Param. | FLOPs | Model | Shell |
|---|---|---|---|---|---|---|
| VideoMamba-Ti | 448 | 79.3 | 7M | 4.3G | aliyun, :hugs:HF | run.sh |
| VideoMamba-Ti | 576 | 79.6 | 7M | 7.1G | aliyun, :hugs:HF | run.sh |
| VideoMamba-S | 448 | 83.2 | 26M | 16.9G | aliyun, :hugs:HF | run.sh |
| VideoMamba-S | 576 | 83.5 | 26M | 28.0G | aliyun, :hugs:HF | run.sh |
| VideoMamba-M | 448 | 83.8 | 75M | 50.4G | aliyun, :hugs:HF | run.sh |
| VideoMamba-M | 576 | 84.0 | 75M | 83.1G | aliyun, :hugs:HF | run.sh |