edit

tutorial-contents/302_classification.py

@@ -51,7 +51,7 @@ plt.show()
 
 for t in range(100):
     out = net(x)                 # input x and predict based on x
-    loss = loss_func(out, y)     # must be (1. nn output, 2. target), the target label is not one-hotted
+    loss = loss_func(out, y)     # must be (1. nn output, 2. target), the target label is NOT one-hotted
 
     optimizer.zero_grad()   # clear gradients for next train
     loss.backward()         # backpropagation, compute gradients