plt.plot(batch_x.data.numpy(), predict.squeeze(1).data.numpy())

代码有一个小错误，第26行应该是`print(f"Epoch:{i+1}/{epochs}, loss:{loss.item()}")`，需要加上括号。另外，在第34行，应该是`loss = criterion(out, batch_y)`，而不是`loss = criterion(predict, batch_y)`。正确的代码如下所示： ``` import torch import matplotlib.pyplot as plt def features(x): x = x.unsqueeze(1) return torch.cat([x ** i for i in range(1, 5)], 1) x_weight = torch.Tensor([-1.13, -2.14, 3.15, -0.01]).unsqueeze(1) b = torch.Tensor([0.512]) def target(x): return x.mm(x_weight) + b.item() def get_batch_data(batch_size): batch_x = torch.rand(batch_size) features_x = features(batch_x) target_y = target(features_x) return features_x, target_y class PR(torch.nn.Module): def __init__(self): super(PR, self).__init__() self.poly = torch.nn.Linear(4, 1) def forward(self, x): return self.poly(x) epochs = 10000 batch_size = 32 model = PR() criterion = torch.nn.MSELoss() optimizer = torch.optim.SGD(model.parameters(), 0.01) for i in range(epochs): batch_x, batch_y = get_batch_data(batch_size) out = model(batch_x) loss = criterion(out, batch_y) optimizer.zero_grad() loss.backward() optimizer.step() if(i % 100 == 0): print(f"Epoch:{i+1}/{epochs}, loss:{loss.item()}") if(i%1000 == 0): predict = model(features(batch_x)) plt.plot(batch_x.data.numpy(), predict.squeeze().data.numpy(), "r") loss = criterion(out, batch_y) plt.title(f"Loss:{loss.item()}") plt.xlabel("X") plt.ylabel("Y") plt.scatter(batch_x, batch_y) plt.show() ```

在代码中，有一些小错误： 1. `torch.randon`应该是`torch.rand`，生成随机数的函数名称写错了。 2. `loss.item` 应该是`loss.item()`，需要调用这个方法来得到损失值。 3. 在if语句中，应该是`i % 100 == 0`而不是`epochs % 100 == 0`，因为epochs是循环次数，不会改变，而i是当前循环的次数。 修改后的代码如下： ``` import torch import matplotlib.pyplot as plt def features(x): x = x.unsqueeze(1) return torch.cat([x ** i for i in range(1, 5)], 1) x_weight = torch.Tensor([-1.13, -2.14, 3.15, -0.01]).unsqueeze(1) b = torch.Tensor([0.512]) def target(x): return x.mm(x_weight) + b.item() def get_batch_data(batch_size): batch_x = torch.rand(batch_size) # 将 torch.randon 改为 torch.rand features_x = features(batch_x) target_y = target(features_x) return features_x, target_y class PR(torch.nn.Module): def __init__(self): super(PR, self).__init__() self.poly = torch.nn.Linear(4, 1) def forward(self, x): return self.poly(x) epochs = 10000 batch_size = 32 model = PR() criterion = torch.nn.MSELoss() optimizer = torch.optim.SGD(model.parameters(), 0.01) for i in range(epochs): batch_x, batch_y = get_batch_data(batch_size) out = model(batch_x) loss = criterion(out, batch_y) optimizer.zero_grad() loss.backward() optimizer.step() if(i % 100 == 0): # 将 epochs % 100 == 0 改为 i % 100 == 0 print(f"Epoch:{i+1}/{epochs}, loss:{loss.item()}") if(i % 1000 == 0): predict = model(features(batch_x)) plt.plot(batch_x.data.numpy(), predict.squeeze(1).data.numpy(), c="r") loss = criterion(predict, batch_y) plt.title(f"Loss:{loss.item()}") plt.xlabel("X") plt.ylabel("Y") plt.scatter(batch_x, batch_y) plt.show() ```