PyTorch implementation of PNASNet-5. On the ImageNet classification task, the model achieves PnasNet-Mobile:74.2% top-1 accuracy and 91.9% top-5 accuracy. With PnasNet-Large: 82.9 top-1 accuracy and 96.2 top- 5 accuracy.
A PyTorch implementation of MnasNet. On the ImageNet classification task, the model achieves MnasNet:74.0% top-1 accuracy with 76ms latency on a Pixel phone, which is 1.5× faster than MobileNetV2 and Mnasnet-SE:75.42% top-1 accuracy.
- PyTorch 0.4.1
- torchvision 0.2.1