Simple Neural Network Practice

1.代码截图

import matplotlib.pyplot as plt
plt.hist(torch.randn(1000).numpy(),100)

plt.hist(torch.randn(10**6).numpy(),100);

learning_rate = 1e-3
lambda_l2 = 1e-5
model = nn.Sequential(
  nn.Linear(D, H),
  nn.Linear(H, C)
)
model.to(device)
criterion = torch.nn.CrossEntropyLoss()

optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate, weight_decay=lambda_l2)

for t in range(1000):
  y_pred = model(X)
  loss = criterion(y_pred, Y)
  score, predicted = torch.max(y_pred, 1)
  acc = (Y == predicted).sum().float() / len(Y)
  print('[EPOCH]: %i, [LOSS]: %.6f, [ACCURACY]: %.3f' % (t, loss.item(), acc))
  display.clear_output(wait=True)
  optimizer.zero_grad()
  loss.backward()
  optimizer.step()

  
print(model)
plot_model(X, Y, model)

learning_rate = 1e-3
lambda_l2 = 1e-5

model = nn.Sequential(
    nn.Linear(D, H),
    nn.ReLU(),
    nn.Linear(H, C)
)
model.to(device)

criterion = torch.nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate, weight_decay=lambda_l2)

for t in range(1000):
  y_pred = model(X)
  loss = criterion(y_pred, Y)
  score, predicted = torch.max(y_pred, 1)
  acc = ((Y == predicted).sum().float() / len(Y))
  print("[EPOCH]: %i, [LOSS]: %.6f, [ACCURACY]: %.3f" % (t, loss.item(), acc))
  display.clear_output(wait=True)
  optimizer.zero_grad()
  loss.backward()
  optimizer.step()


print(model)
plot_model(X, Y, model)