Not displaying graph as expected in Jupyter Notebook (either during training or after training)

I followed the tutorial and put the HL functions in my validation part, but the graph is not showing at all... Matplotlib is working and running the demo pytorch_train.ipynb has no problem.

I packed the plotting part in my evaluation function and there are other string outputs during training. Should I delete those string outputs?

Code snippet

def evaluate(loader):
    model.eval()

    predictions = []
    labels = []
    correct = 0
    loss = 0
    with torch.no_grad():
        for i, data in enumerate(loader,0):
            data = data.to(device)
            pred = model(data.x,data.batch).detach().cpu()
            label = data.y.detach().cpu()

            predictions.append(pred.numpy())
            labels.append(label.numpy())
            loss += loss_func(pred, label).item()

            pred = pred.numpy().squeeze()
            label = label.numpy().squeeze()        
            correct += (abs(pred-label)<0.5).sum()

            global_iter_num = epoch * len(loader) + i + 1
            print(global_iter_num)
            if (global_iter_num % log_step_interval == 0):
                print("o")

                history.log((epoch, i),
                            val_loss=loss/(i+1),
                            val_acc=correct/(i+1),
                            hidden_weight1=model.conv1.weight,
                            hidden_weight2=model.conv2.weight
                           )
                with canvas:
                    canvas.draw_plot(history["val_loss"])
                    canvas.draw_plot(history["val_acc"])
                    canvas.draw_image(history["hidden_weight1"])
                    canvas.draw_image(history["hidden_weight2"])
                time.sleep(0.1)
        return predictions, labels, loss/len(loader.dataset), correct/len(loader.dataset)

history = hl.History()
canvas = hl.Canvas()
log_step_interval = 20

device = torch.device('cuda')
model = GCN().to(device)
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate, weight_decay = l2_reg)
loss_func = torch.nn.BCELoss()  # binary cross-entropy
train_loader = DataLoader(train_dataset, batch_size=batch_size, drop_last = True)
val_loader = DataLoader(val_dataset, batch_size=batch_size)
for epoch in range(num_epochs):
    gc.collect()
    train_loss, train_acc = train()
    _, _, val_loss, val_acc = evaluate(val_loader)
    print('Epoch: {:03d}, Train Loss: {:.5f}, Train Acc: {:.5f}, Val Loss: {:.5f}, Val Acc: {:.5f}'.
          format(epoch, train_loss, train_acc, val_loss, val_acc))

Typical output:

23
Epoch: 000, Train Loss: 0.02185, Train Acc: 0.52832, Val Loss: 0.02351, Val Acc: 0.51029
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
o
/public/workspace/z/miniconda3/envs/ST-Torch/lib/python3.7/site-packages/torch/nn/modules/loss.py:498: UserWarning: Using a target size (torch.Size([32])) that is different to the input size (torch.Size([32, 1])) is deprecated. Please ensure they have the same size.
  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)
41
42
43
44
45
46
Epoch: 001, Train Loss: 0.02081, Train Acc: 0.58771, Val Loss: 0.02304, Val Acc: 0.53498
/public/workspace/z/miniconda3/envs/ST-Torch/lib/python3.7/site-packages/torch/nn/modules/loss.py:498: UserWarning: Using a target size (torch.Size([25])) that is different to the input size (torch.Size([25, 1])) is deprecated. Please ensure they have the same size.
  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)
47
48
49
50
51
52
53
54
55
56
57
58
59
60
o

waleedka / hiddenlayer

Not displaying graph as expected in Jupyter Notebook (either during training or after training) #87