Update README.md

README.md

@@ -20,6 +20,14 @@ augment: True         # represents the medical data augmentation
 split: 'late'         # represents the late fusion strategy
 ```
 
+### Load Model
+
+```
+model = RadNet(num_cls=num_classes, backbone=backbone, depth=depth, ltype=ltype, augment=augment, fuse=fuse, ke=ke, encoded=encoded, adapter=adapter)
+pretrained_weights = torch.load("path/to/pytorch_model_32_late.bin")
+missing, unexpect = model.load_state_dict(pretrained_weights,strict=False)
+```
+
 ## Why we provide this checkpoint?
 
 All the early fusion checkpoint can be further finetuned from this checkpoint. If you need other checkpoints using different parameter settings, there are two possible ways: