Retrieving tensor names from a pre trained model

[JRMeyer]

>>> boborbt
[January 19, 2018, 4:57pm]

Dear all, slash
I am trying to inspect the hidden states of the recurrent part of the
network, but I cannot find a way to retrieve the relevant tensors. Is
there any way to do that?

A good starting point for me could be to retrieve the tensors
corresponding to the output states as returned by
tf.nn.bidirectional_dynamic_rnn(...) as in:

github.com

#### mozilla/DeepSpeech/blob/94ab93c42dfd369eb42cd7952692f04dc0deb40a/DeepSpeech.py slash #L449

<br/> lstm_bw_cell = tf.contrib.rnn.DropoutWrapper(lstm_bw_cell,<br/> input_keep_prob=1.0 - dropout[4],<br/> output_keep_prob=1.0 - dropout[4],<br/> seed=FLAGS.random_seed)<br/> <br/> # `layer_3` is now reshaped into `[n_steps, batch_size, 2*n_cell_dim]`,<br/> # as the LSTM BRNN expects its input to be of shape `[max_time, batch_size, input_size]`.<br/> layer_3 = tf.reshape(layer_3, [-1, batch_x_shape[0], n_hidden_3])<br/> <br/> # Now we feed `layer_3` into the LSTM BRNN cell and obtain the LSTM BRNN output.<br/> outputs, output_states = tf.nn.bidirectional_dynamic_rnn(cell_fw=lstm_fw_cell,<br/> cell_bw=lstm_bw_cell,<br/> inputs=layer_3,<br/> dtype=tf.float32,<br/> time_major=True,<br/> sequence_length=seq_length)<br/> <br/> # Reshape outputs from two tensors each of shape [n_steps, batch_size, n_cell_dim]<br/> # to a single tensor of shape [n_steps*batch_size, 2*n_cell_dim]<br/> outputs = tf.concat(outputs, 2)<br/> outputs = tf.reshape(outputs, [-1, 2*n_cell_dim])<br/>

My plan is to use the tensor name to retrieve the tensor value using
tf.Graph.get_tensor_by_name(). I started by retrieving all available
names using

graph_def = tf.GraphDef()
graph_def.ParseFromString(f.read())

names = [n.name for n in graph_def.node]

but to the best of my understanding this list does not contain the name
of the tensors I'm looking for. Am I on the right track?

Thanks in advance.

Roberto

[This is an archived TTS discussion thread from discourse.mozilla.org/t/retrieving-tensor-names-from-a-pre-trained-model]

[JRMeyer]

>>> lissyx
[January 19, 2018, 5:28pm]

We only set names for input and output layers, though internally they
all have names. The way you do it should be okay. What you want is the
tensor outputs, right?

[Archived Post]

[JRMeyer]

>>> mattia.cerrato
[January 22, 2018, 3:19pm]

Hi,

thanks for your answer. What me and
would like to do is actually
to retrieve both the output_states and outputs tensors.

Right now we are looking at all the operation names defined in the
pre-trained model graph; however the only name we can see that seems to
be connected with the output_states tensor is called output_node.
However, this is not what we are looking for as it is the name that was
used for the tensor representing the decoded sequence:

https://github.com/mozilla/DeepSpeech/blob/9f26c64db66a04c702e50ce30d099732f6c0db05/DeepSpeech.py#L1703.

Per our understanding, the tensors we are looking for have not been
named explicitly in DeepSpeech.py. Is there any
way to understand which names have they been assigned by tensorflow?

Here is a pastebin of the operation
names we are retrieving. Does any of this stand out to you as possibly
being either outputs or output_states?

[Archived Post]

[JRMeyer]

>>> lissyx
[January 22, 2018, 3:26pm]

Have you tried having a look at the graphviz output of the graph? This
should help you identify much easily. Naming includes the scope, so for
example bidirectional_rnn/* is actually inside the bidirectional
layers.

With the graph representation, you should be able to see the nodes and
edges that are coming outside of those layers and getting into the
concat / reshape. You also can now that layer 3 / 5 elements will be
named with some h3, b3, h5 or b5.
[Maybe you could also just make your own graph, nam]ing those elements,
and then comparing the dumps between your graph and ours, thus
identifying the layers?

[Archived Post]

[JRMeyer]

>>> mattia.cerrato
[January 23, 2018, 11:38am]

Ah, thanks, we were not aware of
tfgraphviz - were you
recommending this? We tried to visualize the frozen graph that comes
with the 1.0 release, however the bidirectional_rnn function call is a
bit of a black box in the resulting image:

It does not seem to be showing the output tensors, and filtering all the
op names for 'bidirectional_rnn' does not really help as the majority of
ops are in the scope with that very name. Filtering for both
'bidirectional_rnn' and 'output' instead shows no results.

Your suggestion about creating a new graph is also very sensible. Could
you please give us a bit of direction in using DeepSpeech.py's BiRNN()
class definition to create a graph ourselves? We are currently looking
to use the export() function but it seems like we need a checkpoint
file for that since it is using the freeze_graph tensorflow utility. I
have a tentative script
here,
would you please be so kind as to tell me if I am going in the right
direction?

Thanks again!

[Archived Post]

[JRMeyer]

>>> lissyx
[January 23, 2018, 11:42am]

Hm no, I was thinking about
./tensorflow/tools/quantization/graph_to_dot.py:
https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tools/quantization/graph_to_dot.py

[Archived Post]

[JRMeyer]

>>> mattia.cerrato
[January 25, 2018, 4:56pm]

Alright, thanks! By looking at the output dot file we were able to
locate the outputs tensor in concat.outputs[0]. Thanks for all your
help!

We would be able to proceed forward with our work now, but we could not
help but notice that some tensors and operations are not named - in the
case of output_states, we believe this situation makes it very hard to
identify it while using visualization tools such as
./tensorflow/tools/quantization/graph_to_dot.py or tfgraphviz. Would a
pull request to include names for the output_states tensor and others

[Archived Post]

[JRMeyer]

>>> lissyx
[January 25, 2018, 5:26pm]

I guess that anything making people able to hack on top of our model is
good
![:slight_smile:](

[Archived Post]