problems in "08.rainbow.ipynb"

[915288938lx]

in "08.rainbow.ipynb" , DQNAgent.update_model() function:

PER: importance sampling before average

loss = torch.mean(elementwise_loss * weights)

elementwise_loss: shape(128,)
weights: shape(128,1)

elementwise_loss * weights: shape: (128, 128)
I think the expected shape of "elementwise_loss * weights" should be (128, )

hope your answer