No module named models.rnn.translate

[llsourcell]

No module named models.rnn.translate

[chrismpalmer]

@Graystripe17 That was the problem. I am training now.

So if anyone else comes accross this. I fixed it by changing seq2seq_model to

import data_utils instead of the tensor flow one.

[chrismpalmer]

@Graystripe17 thanks a lot. I do notice there is a local data_utils and seq2seq_model imports a tensor flow one. I wonder if this is intended.

[ImGokulM]

Worked!
As @Graystripe17 said, in seq2seq_model.py you need to replace the line #28
'from tensorflow.models.rnn.translate import data_utils' with 'import data_utils'.
Thanks @Graystripe17!

[athuldevin]

i have the same issue and i applied the fix it works ,but it throws another error for both train and test

Mode : test

Traceback (most recent call last):
File "C:\Users\win-7\Downloads\Compressed\tensorflow_chatbot-master\execute.py", line 322, in
decode()
File "C:\Users\win-7\Downloads\Compressed\tensorflow_chatbot-master\execute.py", line 211, in decode
model = create_model(sess, True)
File "C:\Users\win-7\Downloads\Compressed\tensorflow_chatbot-master\execute.py", line 104, in create_model
model = seq2seq_model.Seq2SeqModel( gConfig['enc_vocab_size'], gConfig['dec_vocab_size'], _buckets, gConfig['layer_size'], gConfig['num_layers'], gConfig['max_gradient_norm'], gConfig['batch_size'], gConfig['learning_rate'], gConfig['learning_rate_decay_factor'], forward_only=forward_only)
File "C:\Users\win-7\Downloads\Compressed\tensorflow_chatbot-master\seq2seq_model.py", line 98, in init
single_cell = tf.nn.rnn_cell.GRUCell(size)
AttributeError: module 'tensorflow.python.ops.nn' has no attribute 'rnn_cell'

[Graystripe17]

@RathanakSreang if you try to go in and change the import a dozen of other changes spring up. If you want to use the latest version, wait for TF1.1 because I messaged some of the workers at TF and they promised to move the modules back

[sujeet1747]

You can uninstall using the following shell script___
sudo pip uninstall tensorflow
and roll back by ____
sudo pip install tensorflow==0.12

[sujeet1747]

The conversation were gibberish. One I'd like to mention

Where are you ?
orgy orgy orgy orgy orgy orgy orgy orgy orgy orgy orgy orgy

Where is the error, folks ?

[ynakashio]

I'm getting the same problem....

[Graystripe17]

Use TensorFlow 0.12.0 instead of TensorFlow 1.0.0

[RathanakSreang]

@Graystripe17
do you have other solution beside using old version?

[chrismpalmer]

Friends! I am running 0.12.0 and getting this same problem. I appear to have the right config. I am using the tensorflow docker 0.12.0 so everything should be intact perhaps its built on a release candidate.

[Graystripe17]

@chrismpalmer You using python 2.7?

[chrismpalmer]

@Graystripe17 yes

root@5d8738fee03e:/notebooks# python --version
Python 2.7.6

[Graystripe17]

@chrismpalmer I'm sorry I don't really have a good solution for you. If it's the exact same installation, I don't think it should be giving you this particular error. I used virtualenv and am not familiar with docker. Maybe give that a shot?

[ImGokulM]

I'm also getting the same error,

Traceback (most recent call last):
File "execute.py", line 31, in
import seq2seq_model
File "\tensorflow_chatbot-master\seq2seq_model.py", line 28, in
from tensorflow.models.rnn.translate import data_utils
ImportError: No module named 'tensorflow.models'

Please help if anyone fixed this.

[Graystripe17]

@ImGokulM @chrismpalmer No problem! @AnupKumarGupta If this solved your issue, you can close this.

[roveneliah]

I'm getting the same thing as @athuldevin after fixing line 28. Any ideas on what this could possibly be?

[jmoney4769]

See #46 on how to get it working with 1.2. Also the Tensorflow release notes talk about where packages get moved to.

[LockGit]

I'm getting the same thing as @AnupKumarGupta .
use tensorflow docker container
Python 2.7.12

[LockGit]

thx