ptb_word_lm.py AttributeError: 'module' object has no attribute 'RNNParamsSaveable'

[larryfu]

when running ptb_word_lm.py with tf-nightly 1.5.0-dev20171103
I got this error

2017-11-04 16:48:59.759628: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1121] Creating     TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce GTX 1070, pci bus id: 0000:01:00.0, compute capability: 6.1)
2017-11-04 16:48:59.761131: E tensorflow/stream_executor/cuda/cuda_driver.cc:936] failed to allocate 221.62M (232390656 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY
WARNING:tensorflow:From /home/larry/workspace/tensorflow/models/tutorials/rnn/ptb/ptb_word_lm.py:165: get_or_create_global_step (from tensorflow.contrib.framework.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Please switch to tf.train.get_or_create_global_step
Traceback (most recent call last):
  File "/home/larry/workspace/tensorflow/models/tutorials/rnn/ptb/ptb_word_lm.py", line 527, in <module>
    tf.app.run()
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/platform/app.py", line 48, in run
    _sys.exit(main(_sys.argv[:1] + flags_passthrough))
  File "/home/larry/workspace/tensorflow/models/tutorials/rnn/ptb/ptb_word_lm.py", line 502, in main
    model.import_ops()
  File "/home/larry/workspace/tensorflow/models/tutorials/rnn/ptb/ptb_word_lm.py", line 272, in import_ops
    params_saveable = tf.contrib.cudnn_rnn.RNNParamsSaveable(
AttributeError: 'module' object has no attribute 'RNNParamsSaveable'

It seems that there is no RNNParamsSaveable in cudnn_rnn
Any idea how to fix it?

[angerson]

Please provide details about what platform you are using (operating system, architecture). Also include your TensorFlow version. Also, did you compile from source or install a binary? Make sure you also include the exact command if possible to produce the output included in your test case. If you are unclear what to include see the issue template displayed in the Github new issue template.

We ask for this in the issue submission template, because it is really difficult to help without that information. Thanks!

[rkuo2000]

Same problem here, (tensorflow 1.4.0 with python 3.6.3 with 1 GPU w2GB w CuDA8+CuDNN6.0,
Windows 7 Ultimate SP1)
$git clone https://github.com/tensorflow/models
$cd ~/models/tutorials/rnn/ptb
(download & unzip simple-samples.tar.gz)
$python ptb_word_lm.py --data_path=./simple-samples/data --model=small

Error messages as below:
Traceback (most recent call last):
File "ptb_word_lm.py", line 526, in
tf.app.run()
File "C:\Python36\lib\site-packages\tensorflow\python\platform\app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "ptb_word_lm.py", line 502, in main
model.import_ops()
File "ptb_word_lm.py", line 272, in import_ops
params_saveable = tf.contrib.cudnn_rnn.RNNParamsSaveable(
AttributeError: module 'tensorflow.contrib.cudnn_rnn' has no attribute 'RNNParamsSaveable'

[earlofyor714]

Same problem here, (tensorflow-gpu 1.4.0 with python 3.5.4 with 1 GPU w CuDA8+CUDNN6.0, Windows 10)

~/models/tutorials/rnn/ptb (master)
$ python ptb_word_lm.py --data_path=../../../simple-examples/data/ --model=small

Error messages as below:
WARNING:tensorflow:From ptb_word_lm.py:165: get_or_create_global_step (from tens orflow.contrib.framework.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Please switch to tf.train.get_or_create_global_step
Traceback (most recent call last):
File "ptb_word_lm.py", line 525, in
tf.app.run()
File "C:\Users\Enkimesh\Anaconda3\envs\tensorflow-gpu\lib\site-packages\tensor flow\python\platform\app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "ptb_word_lm.py", line 501, in main
model.import_ops()
File "ptb_word_lm.py", line 271, in import_ops
params_saveable = tf.contrib.cudnn_rnn.RNNParamsSaveable(
AttributeError: module 'tensorflow.contrib.cudnn_rnn' has no attribute 'RNNParamsSaveable'

[angerson]

tensorflow/tensorflow@b30ce47 changed RNNParamsSaveable into multiple classes, and ptb_word_lm hasn't been updated yet. @ebrevdo, can you take a look?

[David-Levinthal]

can somebody offer a suggestion on how to change the following block from ptb_word_lm.py (or should this just be commented out). It would be interesting to compare the cuda9/cudnn7 r 1.4 build with the older R1.3 cuda8/cudnn6 buiild
rnn_params = tf.get_collection_ref("rnn_params")
if self._cell and rnn_params:
params_saveable = tf.contrib.cudnn_rnn.RNNParamsSaveable(
self._cell,
self._cell.params_to_canonical,
self._cell.canonical_to_params,
rnn_params,
base_variable_scope="Model/RNN")
tf.add_to_collection(tf.GraphKeys.SAVEABLE_OBJECTS, params_saveable)

[David-Levinthal]

commenting out the if block and the tf.add_to_collection line will get the code to run

[withme6696]

I change the code as follows, then it works:
if self._cell and rnn_params:
params_saveable = tf.contrib.cudnn_rnn.CudnnLSTMSaveable(
rnn_params,
self._cell.num_layers,
self._cell.num_units,
self._cell.input_size,
self._cell.input_mode,
self._cell.direction,
scope="Model/RNN")
remove this line"tf.add_to_collection(tf.GraphKeys.SAVEABLE_OBJECTS, params_saveable)"

[saurabhpanwar67412]

Traceback (most recent call last):
File "ptb_word_lm.py", line 527, in
tf.app.run()
File "C:\Users\saurabh\Anaconda2\envs\tensorflow\lib\site-packages\tensorflow\python\platform\app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "ptb_word_lm.py", line 454, in main
% (len(gpus), FLAGS.num_gpus))
ValueError: Your machine has only 0 gpus which is less than the requested --num_gpus=1.
please help me to solve this issue . its kinda urgent . i am trying implementing on cpu version . and here is the code link on tensorflow to which i am trying to execute . " https://www.tensorflow.org/tutorials/recurrent" .

[David-Levinthal]

This is a different issue. I believe you have either built, installed a cuda build of TF. You need one built for a CPU. Possibly with MKL.

…

On Fri, Dec 8, 2017 at 11:40 AM, saurabhpanwar67412 < ***@***.***> wrote: Traceback (most recent call last): File "ptb_word_lm.py", line 527, in tf.app.run() File "C:\Users\saurabh\Anaconda2\envs\tensorflow\lib\site- packages\tensorflow\python\platform\app.py", line 48, in run _sys.exit(main(_sys.argv[:1] + flags_passthrough)) File "ptb_word_lm.py", line 454, in main % (len(gpus), FLAGS.num_gpus)) ValueError: Your machine has only 0 gpus which is less than the requested --num_gpus=1. please help me to solve this issue . its kinda urgent . i am trying implementing on cpu version . and here is the code link on tensorflow to which i am trying to execute . " https://www.tensorflow.org/ tutorials/recurrent" . please anyone run the code and tell me the errors to resolve it and send the the outcomes of this `tutorial on ***@***.***,com or just tell me the error. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2709 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AIUuT0CaqhPbWaqp3cfl4N_FCcwrJ3Khks5s-ZC6gaJpZM4QR8Kk> .

[ebrevdo]

Or you just run with the extra command line argument

…

--num_gpus=0 ? On Fri, Dec 8, 2017 at 12:51 PM, David Levinthal Ph.D. < notifications@github.com> wrote:

This is a different issue. I believe you have either built, installed a cuda build of TF. You need one built for a CPU. Possibly with MKL. On Fri, Dec 8, 2017 at 11:40 AM, saurabhpanwar67412 < ***@***.***> wrote: > Traceback (most recent call last): > File "ptb_word_lm.py", line 527, in > tf.app.run() > File "C:\Users\saurabh\Anaconda2\envs\tensorflow\lib\site- > packages\tensorflow\python\platform\app.py", line 48, in run > _sys.exit(main(_sys.argv[:1] + flags_passthrough)) > File "ptb_word_lm.py", line 454, in main > % (len(gpus), FLAGS.num_gpus)) > ValueError: Your machine has only 0 gpus which is less than the requested > --num_gpus=1. > please help me to solve this issue . its kinda urgent . i am trying > implementing on cpu version . and here is the code link on tensorflow to > which i am trying to execute . " https://www.tensorflow.org/ > tutorials/recurrent" . please anyone run the code and tell me the errors > to resolve it and send the the outcomes of this `tutorial on > ***@***.***,com or just tell me the error. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#2709 (comment) >, > or mute the thread > <https://github.com/notifications/unsubscribe- auth/AIUuT0CaqhPbWaqp3cfl4N_FCcwrJ3Khks5s-ZC6gaJpZM4QR8Kk> > . > — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#2709 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABtim_zIpPyxGGiVXOMsoTsMugNQN8irks5s-aE9gaJpZM4QR8Kk> .