1. #训练函数部分
    2. def train(i,train_iter, test_iter, net, loss, optimizer, device, num_epochs):
    3.     net = net.to(device)
    4.     print("training on ", device)
    5.     start = time.time()
    6.     test_acc_max_l = []
    7.     train_acc_max_l = []
    8.     train_l_min_l=[]
    9.     for epoch in range(num_epochs):  #迭代100次
    10.         batch_count = 0
    11.         train_l_sum, train_acc_sum, test_acc_sum, n = 0.0, 0.0, 0.0, 0
    12.         for X, y in train_iter:
    13.             X = X.to(device)
    14.             y = y.to(device)
    15.             y_hat = net(X)
    16.             l = loss(y_hat, y)
    17.             optimizer.zero_grad()
    18.             l.backward()
    19.             optimizer.step()
    20.             train_l_sum += l.cpu().item()
    21.             train_acc_sum += (y_hat.argmax(dim=1) == y).sum().cpu().item()
    22.             n += y.shape[0]
    23.             batch_count += 1
    24.         #至此，一个epoches完成
    25.         test_acc_sum= d2l.evaluate_accuracy(test_iter, net)
    26.         train_l_min_l.append(train_l_sum/batch_count)
    27.         train_acc_max_l.append(train_acc_sum/n)
    28.         test_acc_max_l.append(test_acc_sum)
    29.         print('fold %d epoch %d, loss %.4f, train acc %.3f, test acc %.3f'
    30.               % (i+1,epoch + 1, train_l_sum / batch_count, train_acc_sum / n, test_acc_sum))
    31.     #train_l_min_l.sort()
    32.     #￥train_acc_max_l.sort()
    33.     index_max=test_acc_max_l.index(max(test_acc_max_l))
    34.     f = open("./shallow/results.txt", "a")
    36.     if i==0:
    37.         f.write("%d fold"+"   "+"train_loss"+"       "+"train_acc"+"      "+"test_acc")
    38.     f.write('\n' +"fold"+str(i+1)+":"+str(train_l_min_l[index_max]) + " ;" + str(train_acc_max_l[index_max]) + " ;" + str(test_acc_max_l[index_max]))
    39.     f.close()
    40.     print('fold %d, train_loss_min %.4f, train acc max%.4f, test acc max %.4f, time %.1f sec'
    41.             % (i + 1, train_l_min_l[index_max], train_acc_max_l[index_max], test_acc_max_l[index_max], time.time() - start))
    42.     return train_l_min_l[index_max],train_acc_max_l[index_max],test_acc_max_l[index_max]