Hi,
Thank you for your great work. I found a potential problem and a performance issue in the get_resnet_preds function inside applications/imagenetV2/generate_csv.py.
Inside the with torch.no_grad(): block, resnet101(images) is called for both ResNet-50 and ResNet-101 predictions. The loaded resnet50 model is never actually used for inference.
Hi,
Thank you for your great work. I found a potential problem and a performance issue in the
get_resnet_predsfunction insideapplications/imagenetV2/generate_csv.py.Inside the
with torch.no_grad():block,resnet101(images)is called for both ResNet-50 and ResNet-101 predictions. The loadedresnet50model is never actually used for inference.