Accuracy on training in Pytorch

1

I would like to do binary classification with softmax. Am I calculating training accuracy in the code below wrong?

I am new to Pytorch and it is very strange I am getting accuracy above 100

correct = 0
train_loss = []


model.train()

for epoch in range(epochs):
    for xb, yb in train_dl:
        y_pred = model(xb)           
        loss = loss_func(y_pred, yb)  
        optimizer.zero_grad()       
        loss.backward()              
        optimizer.step()              
    train_loss.append(loss.item())
    
    _,y_pred_ = torch.max(y_pred, dim = 1)
    
    correct += (y_pred_ == yb).sum()
    

acc_ = 100 * correct / len(x_tensor)

print(correct)   
print("Accuracy = {}".format(acc_))
    
    
print('Last iteration loss value: '+str(loss.item()))

thanks in advance

2

What is x_tensor here?

You may want to do do a sanity check that len(x_tensor) matches the total number of datapoints seen doing evaluation, e.g., summing the total of y_pred.size(0).