rnn的数据维度和cnn的数据维度

    for epoch in range(EPOCH):
        for step, (x, batch_y) in enumerate(train_loader):
            # print('x',x.shape) 
            # x torch.Size([64, 1, 28, 28])    #batch_size,channels,width,height          
            batch_x = x.view(-1, 28, 28)
            # print('batch_x',batch_x.shape)    
            # batch torch.Size([64, 28, 28])
            output = rnn(batch_x)
            loss = loss_func(output, batch_y)

            optimizer.zero_grad()
            loss.backward()
            optimizer.step()

rnn需要三个维度
batch_size
input_size
time_step

cnn需要四个维度
batch_size
channels
width
height