Update main.py

main.py

@@ -89,7 +89,7 @@ def train(args, model: Union[Einet, EinetMixture], device, train_loader, optimiz
                     epoch,
                     batch_idx * len(data),
                     len(train_loader.dataset),
-                    loss.item() / args.batch_size,
+                    loss.item(),
                     acc_term,
                 )
             )
@@ -122,10 +122,6 @@ def test(model, device, loader, tag):
             test_loss += -1 * lls.sum()
             test_losses += lls.squeeze().cpu().tolist()
 
-            # Else compute negative log likelihoods
-            test_loss += -1 * outputs.sum()
-            test_losses += outputs.squeeze().cpu().tolist()
-
             if args.classification:
                 _, predicted = outputs.max(1)
                 total += target.size(0)
@@ -161,6 +157,9 @@ def test(model, device, loader, tag):
     elif args.dist == "normal":
         leaf_type = Normal
         leaf_kwargs = {}
+    elif args.dist == "normal_rat":
+        leaf_type = RatNormal
+        leaf_kwargs = {"min_sigma": args.min_sigma, "max_sigma": args.max_sigma}
     elif args.dist == "categorical":
         leaf_type = Categorical
         leaf_kwargs = {"num_bins": n_bins}
@@ -178,7 +177,7 @@ def test(model, device, loader, tag):
         num_classes=num_classes,
         leaf_type=leaf_type,
         leaf_kwargs=leaf_kwargs,
-        layer_type="linsum",
+        layer_type=args.layer,
         dropout=0.0,
     )
 
@@ -218,15 +217,19 @@ def test(model, device, loader, tag):
     if args.train:
         for epoch in range(1, args.epochs + 1):
             train(args, model, device, train_loader, optimizer, epoch)
-            lr_scheduler.step()
+            # lr_scheduler.step()
 
         torch.save(model.state_dict(), os.path.join(result_dir, "model.pth"))
+        test(model, device, train_loader, "Train")
+        test(model, device, val_loader, "Val")
+        test(model, device, test_loader, "Test")
 
     else:
         model.load_state_dict(torch.load(os.path.join(result_dir, "model.pth")))
 
-        # test(model, device, test_loader, "Train")
-        # test(model, device, test_loader, "Test")
+        test(model, device, train_loader, "Train")
+        test(model, device, val_loader, "Val")
+        test(model, device, test_loader, "Test")
 
     # Don't sample when doing classification
     if not args.classification: