pytorch动态调整learning rate

optimizer = torch.optim.Adam(rnn.parameters(), lr=LR)  # optimize all cnn parameters
# scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=30, gamma=0.9)
scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer, mode='min', factor=0.9)
loss_func = nn.CrossEntropyLoss().cuda()  # the target label is not one-hotted

# train
for epoch in range(EPOCH):
    epoch_loss = 0
    for step, (b_x, b_y) in enumerate(train_loader):  # gives batch data
        b_x = b_x.view(-1, TIME_STEP, INPUT_SIZE).float()  # reshape x to (batch, time_step, input_size)
        output = rnn(b_x.cuda())  # rnn output
        loss = loss_func(output, b_y.cuda())  # cross entropy loss， output is vector, target is int
        optimizer.zero_grad()  # clear gradients for this training step
        loss.backward()  # backpropagation, compute gradients
        optimizer.step()  # apply gradients
        epoch_loss += loss.data.cpu().numpy()
    scheduler.step(epoch_loss)
    print(optimizer.state_dict()['param_groups'][0]['lr'])