Created by: JiayiFeng

In the chaper6(understand_sentiment), we offer users two available models, one is CNN and the other one is LSTM. However, if we use LSTM model, some errors will occur:

Traceback (most recent call last):
  File "train.py", line 125, in <module>
    cost = stacked_lstm_net(dict_dim, class_dim=class_dim, stacked_num=3)
  File "train.py", line 69, in stacked_lstm_net
    input=fc1, act=relu, bias_attr=bias_attr, layer_attr=layer_attr)
  File "/usr/local/lib/python2.7/dist-packages/paddle/v2/config_base.py", line 52, in wrapped
    out = f(*args, **xargs)
  File "/usr/local/lib/python2.7/dist-packages/paddle/trainer_config_helpers/default_decorators.py", line 53, in __wrapper__
    return func(*args, **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/paddle/trainer_config_helpers/default_decorators.py", line 53, in __wrapper__
    return func(*args, **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/paddle/trainer_config_helpers/default_decorators.py", line 53, in __wrapper__
    return func(*args, **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/paddle/trainer_config_helpers/default_decorators.py", line 53, in __wrapper__
    return func(*args, **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/paddle/trainer_config_helpers/default_decorators.py", line 53, in __wrapper__
    return func(*args, **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/paddle/trainer_config_helpers/layers.py", line 396, in wrapper
    val.check(method.__name__)
  File "/usr/local/lib/python2.7/dist-packages/paddle/trainer_config_helpers/attrs.py", line 276, in check
    (layer_name, key))
NotImplementedError: Layer lstmemory does not support drop_rate

The reason is LSTM doesn't support drop_rate. So we shall remove drop_rate from LSTM's layer_attr.