PyTorch – How to get learning rate during training?

For only one parameter group like in the example you’ve given, you can use this function and call it during training to get the current learning rate:

def get_lr(optimizer):
    for param_group in optimizer.param_groups:
        return param_group['lr']