lstm调参之旅一：dim_proj，validFreq，saveFreq，maxlen，valid_batch_size，modal_costs，recyl_maxlen

•     dim_proj：表示feature的种类个数
•     validFreq：表示多少次更新之后计算validation error
•     saveFreq：每次更新saveFreq的时候保存参数，表示保存参数的次数
•     maxlen：一般应该是和模态数相等
•     valid_batch_size：测试集的batch_size，一定要小于测试集
•     modal_costs：一维矩阵，个数是模态数

def train_lstm(
    dim_proj=72,  # word embeding dimension and LSTM number of hidden units.
    patience=15,  # Number of epoch to wait before early stop if no progress
    max_epochs=5000,  # The maximum number of epoch to run
    dispFreq=10,  # Display to stdout the training progress every N updates
    decay_c=0.,  # Weight decay for the classifier applied to the U weights.
    lrate=0.01,  # Learning rate for sgd (not used for adadelta and rmsprop)
    optimizer=adadelta,  # sgd, adadelta and rmsprop available, sgd very hard to use, not recommanded (probably need momentum and