Ch 2 - Code doesn't run

[LMolr]

Almost all the inline code snippets in chapter 2 are not working.

E.g:

Page 34

Code

plt.xlabel("Plays")
plt.ylabel("Avg Reward")
for i in range(500):
  if random.random() > eps:
    choice = get_best_arm(pastRewards, actions)
  else:
    choice = np.where(arms == np.random.choice(arms))[0][0]
  thisAV = np.array([[choice, reward(arms[choice])]])
  av = np.vstack((av, thisAV))
  percCorrect = 100*(len(av[np.where(av[:,0] == np.argmax(arms))])/len(av))
  runningMean = np.mean(av[:,1])
  plt.scatter(i, runningMean)

Error

NameError: name 'pastRewards' is not defined

Page 37

Code

for i in range(500):
  choice = np.random.choice(arms, p=av_softmax)
  counts[choice] += 1
  k = counts[choice]
  rwd = reward(arms[choice])
  old_avg = av[choice]
  new_avg = old_avg + (1/k)*(rwd - old_avg)
  av[choice] = new_avg
  av_softmax = softmax(av)

Error

IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices

Page 41

Code

>>> 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
>>> loss = (torch.sum(y_known - y))**2 #loss function
>>> loss.backward() #calculate gradients
>>> m.grad
tensor([ 0.7734, -90.4993])

Error

NameError: name 'y_known' is not defined

Code

model = torch.nn.Sequential(
  torchh.nn.Linear(10, 150),
  torch.nn.ReLU(),
  torch.nn.Linear(150, 4),
  torch.nn.ReLU(),
)
loss_fn = torch.nn.MSELoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.01)

Error

NameError: name 'torchh' is not defined

[outlace]

We've uploaded new code Chx_book.ipynb notebooks that should all work for all chapters