modify evaluate

@@ -16,7 +16,7 @@ dataloader:
 model:
   name: 'SLTransformer'
   args:
-    seq_lens:  200
+    seq_lens:  160
     num_classes: 10 
     embed_dim: 9 
     depth: 6

@@ -166,7 +166,7 @@ def build_dataset(img_dir, go_res_dir, label_dir, top_text_list, skip_list, save
         
         X = list()
         y_true = list()
-        for i in range(200):
+        for i in range(160):
             if i >= valid_lens:
                 X.append([0., 0., 0., 0., 0., 0., 0., 0., 0.])
                 y_true.append([0, 0, 0, 0, 0, 0, 0, 0, 0, 0])

@@ -65,7 +65,7 @@ class SLSolver(object):
         train_loss = torch.zeros(1).to(self.device)
         correct = torch.zeros(1).to(self.device)
         for batch, (X, y, valid_lens) in enumerate(self.train_loader):
-            X, y = X.to(self.device), y.to(self.device)
+            X, y, valid_lens = X.to(self.device), y.to(self.device), valid_lens.to(self.device)
             
             pred = self.model(X, valid_lens)
             # [batch_size, seq_len, num_classes]
@@ -97,7 +97,7 @@ class SLSolver(object):
         val_loss = torch.zeros(1).to(self.device)
         correct = torch.zeros(1).to(self.device)
         for X, y, valid_lens in self.val_loader:
-            X, y = X.to(self.device), y.to(self.device)
+            X, y, valid_lens = X.to(self.device), y.to(self.device), valid_lens.to(self.device)
 
             # pred = torch.nn.Sigmoid()(self.model(X))
             pred = self.model(X, valid_lens)
@@ -162,25 +162,30 @@ class SLSolver(object):
 
         label_true_list = []
         label_pred_list = []
-        for X, y in self.val_loader:
-            X, y_true = X.to(self.device), y.to(self.device)
+        for X, y, valid_lens in self.val_loader:
+            X, y_true, valid_lens = X.to(self.device), y.to(self.device), valid_lens.to(self.device)
 
             # pred = torch.nn.Sigmoid()(self.model(X))
-            pred = self.model(X)
-            
-            y_pred = torch.nn.Sigmoid()(pred)
+            y_pred = self.model(X, valid_lens)
             
-            y_pred_idx = torch.argmax(y_pred, dim=1) + 1
-            y_pred_is_other = (torch.amax(y_pred, dim=1) > 0.5).int()
+            # [batch_size, seq_len, num_classes]
+            y_pred_sigmoid = torch.nn.Sigmoid()(y_pred)
+            # [batch_size, seq_len]
+            y_pred_idx = torch.argmax(y_pred_sigmoid, dim=-1) + 1
+            # [batch_size, seq_len]
+            y_pred_is_other = (torch.amax(y_pred_sigmoid, dim=-1) > 0.5).int()
             y_pred_rebuild = torch.multiply(y_pred_idx, y_pred_is_other)
 
-            y_true_idx = torch.argmax(y_true, dim=1) + 1
-            y_true_is_other = torch.sum(y_true, dim=1)
+            y_true_idx = torch.argmax(y_true, dim=-1) + 1
+            y_true_is_other = torch.sum(y_true, dim=-1).int()
             y_true_rebuild = torch.multiply(y_true_idx, y_true_is_other)
 
-            label_true_list.extend(y_true_rebuild.cpu().numpy().tolist())
-            label_pred_list.extend(y_pred_rebuild.cpu().numpy().tolist())
+            # masked_y_true_rebuild = sequence_mask(y_true_rebuild, valid_lens, value=-1)
 
+            for idx, seq_result in enumerate(y_true_rebuild.cpu().numpy().tolist()):
+                label_true_list.extend(seq_result[: valid_lens.cpu().numpy()[idx]])
+            for idx, seq_result in enumerate(y_pred_rebuild.cpu().numpy().tolist()):
+                label_pred_list.extend(seq_result[: valid_lens.cpu().numpy()[idx]])
         
         acc = accuracy_score(label_true_list, label_pred_list)
         cm = confusion_matrix(label_true_list, label_pred_list)