Replaced ToTensor with a combination of PILToTensor and ConvertImageDtype (#4481)

* Replaced ToTensor with a combination of PILToTensor and ConvertImageDtype

* Pass dtype

references/classification/presets.py

+import torch
 from torchvision.transforms import autoaugment, transforms
 from torchvision.transforms.functional import InterpolationMode
 
@@ -17,7 +18,8 @@ class ClassificationPresetTrain:
                 aa_policy = autoaugment.AutoAugmentPolicy(auto_augment_policy)
                 trans.append(autoaugment.AutoAugment(policy=aa_policy))
         trans.extend([
-            transforms.ToTensor(),
+            transforms.PILToTensor(),
+            transforms.ConvertImageDtype(torch.float),
             transforms.Normalize(mean=mean, std=std),
         ])
         if random_erase_prob > 0:
@@ -36,7 +38,8 @@ class ClassificationPresetEval:
         self.transforms = transforms.Compose([
             transforms.Resize(resize_size, interpolation=interpolation),
             transforms.CenterCrop(crop_size),
-            transforms.ToTensor(),
+            transforms.PILToTensor(),
+            transforms.ConvertImageDtype(torch.float),
             transforms.Normalize(mean=mean, std=std),
         ])