Fixed bugs

1. Lack "model.model.eval()" before testing in train.py
2. Modify Readme
3. Add comments for cross attention module

README.md

@@ -90,14 +90,18 @@ After preprocessing, the data should be organized as following:
 
 There are four modes supported in F3-Net​.
 
-| Mode(string) |                                                         |
-| ------------ | ------------------------------------------------------- |
-| 'FAD'        | Use FAD branch only.                                    |
-| 'LFS'        | Use LFS branch only.                                    |
-| 'Both'       | Use both of branches and concate before classification. |
-| 'Mix'        | Use both of branches and MixBlock.                      |
-
-
+| Mode(string)       |                                                         |
+| ------------------ | ------------------------------------------------------- |
+| 'FAD'              | Use FAD branch only.                                    |
+| 'LFS'              | Use LFS branch only.                                    |
+| 'Both'             | Use both of branches and concate before classification. |
+| 'Mix'(unavailable) | Use both of branches and MixBlock.                      |
+
+ **Note**:
+
+Mode 'Mix' is unavailable yet. If you're interested in this part, check 'class Mixblock' in models.py. 
+
+
 
 #### Run
 

models.py

@@ -314,6 +314,8 @@ def fea_8_12(self, x):
     return x
 
 class MixBlock(nn.Module):
+    # An implementation of the cross attention module in F3-Net
+    # Haven't added into the whole network yet
     def __init__(self, c_in, width, height):
         super(MixBlock, self).__init__()
         self.FAD_query = nn.Conv2d(c_in, c_in, (1,1))

train.py

@@ -104,5 +104,6 @@
                 model.model.train()
         epoch = epoch + 1
 
+    model.model.eval()
     auc, r_acc, f_acc = evaluate(model, dataset_path, mode='test')
     logger.debug(f'(Test @ epoch {epoch}) auc: {auc}, r_acc: {r_acc}, f_acc:{f_acc}')