See Book compatibility and corrections for the maintained chapter-by-chapter corrections, including gradient clearing, Bellman targets, rotation, environment limits, and modern-library APIs. Validation distinguishes successful execution from demonstrated learning.
Here we keep an updated list of book errata listed by chapter. The errata will be addressed periodically in the eBook and liveBook versions and less frequently in print versions. We highly recommend using the GitHub code when following along the book as it is very difficult to maintain code within the book text.
Code should be:
x = torch.Tensor([2,4]) #input data m = torch.randn(2, requires_grad=True) #parameter 1 b = torch.randn(1, requires_grad=True) #parameter 2 y = m*x+b #linear model y_known = torch.Tensor([5,9]) loss = (torch.sum(y_known - y))**2 #loss function loss.backward() #calculate gradients m.grad tensor([ -51.9402, -103.8803])
Code should be: def train(env, epochs=10000, learning_rate=1e-3):
Text should say: “When we train this network for 10,000 epochs”
“with probability ε – 1” should be “with probability 1 - ε “
model2=model2= copy.deepcopy(model)
First line of code should be: from IPython.display import clear_output