num_epochs =3for epoch inrange(num_epochs):for X, y in data_iter:
l = loss(net(X),y)
trainer.zero_grad()
l.backward()
trainer.step()
l = loss(net(features), labels)print(f'epoch {
epoch +1}, loss {
l:f}')
比较生成数据集的真实参数和通过有限数据集训练获得的模型参数
从net访问所需的层,读取该层的权重和偏置
w = net[0].weight.data
print('w的估计误差:', true_w - w.reshape(true_w.shape))
b = net[0].bias.data
print('b的估计误差:', true_b - b)