Unable to use GPU

[sriram (sriram chaudhury)]

Hi,

My configuration is below

Os: Ubuntu 18.04.1 LTS

installed tensorflow-gpu in virtualenv

but while running the onmt-main

onmt-main train_and_eval --model_type NMTSmall --auto_config --config data.yml

getting the following error:

Traceback (most recent call last):
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/bin/onmt-main”, line 7, in
from opennmt.bin.main import main
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/opennmt/init.py”, line 5, in
from opennmt import decoders
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/opennmt/decoders/init.py”, line 3, in
from opennmt.decoders.rnn_decoder import RNNDecoder
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/opennmt/decoders/rnn_decoder.py”, line 5, in
import tensorflow as tf
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/tensorflow/init.py”, line 22, in
from tensorflow.python import pywrap_tensorflow # pylint: disable=unused-import
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/tensorflow/python/init.py”, line 49, in
from tensorflow.python import pywrap_tensorflow
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow.py”, line 74, in
raise ImportError(msg)
ImportError: Traceback (most recent call last):
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow.py”, line 58, in
from tensorflow.python.pywrap_tensorflow_internal import *
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow_internal.py”, line 28, in
_pywrap_tensorflow_internal = swig_import_helper()
File “/home/sriram/OpenNMT-tf-master/toy-ende-new/pyenv/local/lib/python2.7/site-packages/tensorflow/python/pywrap_tensorflow_internal.py”, line 24, in swig_import_helper
_mod = imp.load_module(’_pywrap_tensorflow_internal’, fp, pathname, description)
ImportError: libcublas.so.9.0: cannot open shared object file: No such file or directory

upon a google search found that it is something the tensorflow-gpu is needs cuda-9-0 libraries

But I have already installed the Cuda latest library and the OpenNMT-py with pytorch version works fine.

Can anybody please suggest what I have to resolve ?

Thanks,
Sriram

[guillaumekln (Guillaume Klein)]

Hi,

Which CUDA version did you precisely install?

If you have issues configuring your environment, I suggest using Docker instead. See for example:

Running OpenNMT-tf within Docker with nvidia-docker

Hi, It depends how you want to use it really. You could first build a custom image that includes OpenNMT-tf and invokes docker run as if you were invoking an actual script: FROM tensorflow/tensorflow:1.11.0-gpu WORKDIR /root RUN pip install OpenNMT-tf ENTRYPOINT [] docker build -t opennmt/opennmt-tf -f Dockerfile . nvidia-docker run -it --rm --entrypoint onmt-main opennmt/opennmt-tf -h Of course you need to mount the directories you want to access within the Docker container.

[sriram (sriram chaudhury)]

Thanks guillaumekln,

I used the Docker for tensoflow-gpu and inside it installed the openNMT-tf succesfully.

I am running the demo corpus, will report if I get any further error.

Thanks,
Sriram