met "Segmentation fault": added a lstm layer in the full_network, when run train

[liuhuang31]

Hi, thanks the share. when i add a lstm layer, met "Segmentation fault"

Add a simple lstm layer in the full_network, the code is below:

output = tf.squeeze(output) // origin code, doesn't change.
// lstm, the code i add.
lstm_cell = tf.contrib.rnn.BasicLSTMCell(64, forget_bias=1.0, state_is_tuple=True)
init_state = lstm_cell.zero_state(batch_size=config.batch_size, dtype=tf.float32)
lstm_out, final_state = tf.nn.dynamic_rnn(lstm_cell, output, initial_state=init_state, time_major=False)
print("-----------lstm_out.shape------------", lstm_out.shape) // (30, 128, 64)
return lstm_out

when train, i meet "Segmentation fault", and there is no other error information.
it's amazing, the code i add is simple and no fault.

[pc2752]

Hi,

Could you provide more details about which function of which file you are changing?

[liuhuang31]

Hi,
the change file is 'modules_tf.py'.
The code i added is below.

output = encoder_decoder_archi_gan(inputs, is_train)
output = tf.tanh(tf.layers.batch_normalization(tf.layers.dense(output, config.output_features, name = "Fu_F", kernel_initializer=tf.random_normal_initializer(stddev=0.02)), training = is_train, name = "bn_fu_out"))

output = tf.squeeze(output)

# add lstm layer
lstm_cell = tf.contrib.rnn.BasicLSTMCell(64, forget_bias=1.0, state_is_tuple=True)
init_state = lstm_cell.zero_state(batch_size=config.batch_size, dtype=tf.float32)
lstm_out, final_state = tf.nn.dynamic_rnn(lstm_cell, output, initial_state=init_state, time_major=False)
lstm_out = tf.tanh(tf.layers.batch_normalization(tf.layers.dense(lstm_out, config.output_features, name = "T_GRU", kernel_initializer=tf.random_normal_initializer(stddev=0.02)), training=is_train, name = "gru_out"))

return lstm_out

[pc2752]

I would define an LSTM function as follows:

def bi_static_stacked_RNN(x, scope='RNN', lstm_size = config.lstm_size):
"""
Input and output in batch major format
"""
with tf.variable_scope(scope):
x = tf.unstack(x, config.max_phr_len, 1)

    output = x
    num_layer = 2
    # for n in range(num_layer):
    lstm_fw = tf.nn.rnn_cell.LSTMCell(lstm_size, state_is_tuple=True)
    lstm_bw = tf.nn.rnn_cell.LSTMCell(lstm_size, state_is_tuple=True)

    _initial_state_fw = lstm_fw.zero_state(config.batch_size, tf.float32)
    _initial_state_bw = lstm_bw.zero_state(config.batch_size, tf.float32)

    output, _state1, _state2 = tf.contrib.rnn.static_bidirectional_rnn(lstm_fw, lstm_bw, output, 
                                              initial_state_fw=_initial_state_fw,
                                              initial_state_bw=_initial_state_bw, 
                                              scope='BLSTM_'+scope)
    output = tf.stack(output)
    output_fw = output[0]
    output_bw = output[1]
    output = tf.transpose(output, [1,0,2])

    return output

And then pass the output to it as:
output = bi_static_stacked_RNN(output, scope = "scope")