RNN's output is different from tensorflow lstm

[dhkim0225]

Description

Hello, Thanks for developing TensorRT Engine. I saw something strange outputs from RNNv2 layer.

It's really strange. I'm working with TRT python api to convert tf seq2seq model. My sequence_length of rnn is 33. The output shape is (batch_size, 33, 512). The first output [0, 0,:] was the same as the output [0,0 ,:] of tensorflow. However, values ​​started to differ from the next sequence.

The rnn input of INetwork have a same value as the rnn input of tensorflow.

The average errors of 33 time-steps are as follows.

for i in range(33): 
    print(i, np.mean(np.abs(inet_out[i, :] - tf_out[i, :])))

0 1.2609632e-07  <== wow ...!
1 0.0015718038
2 0.010808524
3 0.010692278
4 0.012011731
5 0.016582778
6 0.021841913
7 0.026234297
8 0.028343435
9 0.027489826
10 0.023572126
11 0.026155617
12 0.029127385
13 0.036912326
14 0.04262212
15 0.038910516
16 0.027831417
17 0.024144478
18 0.022921633
19 0.02656816
20 0.033705484
21 0.040616006
22 0.039923944
23 0.030223113
24 0.029302634
25 0.030033736
26 0.03281911
27 0.041052207
28 0.041627046
29 0.035327498
30 0.036981985
31 0.03735178
32 0.030929128

Environment

TensorRT Version: 7.0.0.11 GPU Type: T4 Nvidia Driver Version: 440.33 CUDA Version: 10.2.89 CUDNN Version: 7.6.5 Operating System + Version: Ubuntu 18.04 Python Version (if applicable): 3.6 TensorFlow Version (if applicable): 1.14.0

Relevant Files

TF RNN

    cell_fw = tf.contrib.rnn.LSTMCell(rnn_size, initializer=kernel_initializer)

    outputs_fw, state_fw = tf.nn.dynamic_rnn(
        cell_fw, bottom_sequence,
        sequence_length=sequence_length,
        time_major=True,
        dtype=tf.float32,
        scope=scope + '/fw')

    return outputs_fw

INetworkDefinition

def encoder(net, w_dict, cnn_out):
    permute_layer = net.add_shuffle(cnn_out)
    permute_layer.reshape_dims = trt.Dims3(0, 512, 33)  # (-1, 512, 33)
    permute_layer.second_transpose = trt.Permutation([0, 2, 1])  # (-1, 33, 512)

    bilstm_enc = net.add_rnn_v2(permute_layer.get_output(0),
                                layer_count=1,
                                hidden_size=512,
                                max_seq_length=33,
                                op=trt.RNNOperation.LSTM)
    bilstm_enc.direction = trt.RNNDirection.UNIDIRECTION
    bilstm_enc.input_mode = trt.RNNInputMode.LINEAR

In my opinion, the weight conversion seems to be correct since the first time-step output came out correctly.

Any Ideas..?

Happy to see you again Ryan :p

[rmccorm4]

Hi @dhkim0225 ,

Sorry for the delay, I'll start looking into this tomorrow.

[dhkim0225]

Yup, always, thanks a lot !! :)

[rmccorm4]

Hi @dhkim0225 ,

Would it be possible for you to make a similar repro but comparing TensorRT RNNv2 output to TF's cudnnLSTM op (https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/contrib/cudnn_rnn/CudnnLSTM), and see if that still has the issue?

[dhkim0225]

@rmccorm4

Sorry for late reply. Finally, I solved a problem. It was because of forget_bias.

I didn't understand following statement in tensorrt documentation.

And it means,

_If you your model learned with forget_bias=1 parameter,
(default param of tf.contrib.rnn.lstmcell)
!! You should add forgetbias 1 !!

like this.

Close Issue, since I solve a problem.

Thanks !!

[haiderasad]

@rmccorm4

Sorry for late reply. Finally, I solved a problem. It was because of forget_bias.

I didn't understand following statement in tensorrt documentation.

And it means,

_If you your model learned with forget_bias=1 parameter, (default param of tf.contrib.rnn.lstmcell) !! You should add forgetbias 1 !!

like this.

Close Issue, since I solve a problem.

Thanks !!

hi i am also having different outputs for LSTM in keras and tensorrt, to be more precise i am using bi directional LSTM .While keras lstm gives the correct ouput. so according to you i just need to set forget bias to 1 in tensorrt?