DeepSpeech.PyTorch stops working after installing Torch to use also DeepSpeech.Torch

Dear friends,

My DeepSpeech.PyTorch stopped working after installing Torch to use also DeepSpeech.Torch. See the logs bellow. It is very similar with an another issue of the repo and they said we should use another gcc, but I am not sure exactly what is the REAL problem.

If a move the torch installation directory, DeepSpeech.PyTorch works again! If a move the torch installation directory back, DeepSpeech.PyTorch fails!

> dlm@vm001nc6:~/code/deepspeech.pytorch$
> dlm@vm001nc6:~/code/deepspeech.pytorch$
> dlm@vm001nc6:~/code/deepspeech.pytorch$ python train.py --train_manifest data/train_manifest.csv --val_manifest data/val_manifest.csv
> Traceback (most recent call last):
>   File "train.py", line 9, in <module>
>     from warpctc_pytorch import CTCLoss
>   File "/home/dlm/anaconda3/lib/python3.6/site-packages/warpctc_pytorch/__init__.py", line 7, in <module>
>     from ._warp_ctc import lib as _lib, ffi as _ffi
> ImportError: /home/dlm/anaconda3/lib/python3.6/site-packages/torch/lib/../../../../libgomp.so.1: version `GOMP_4.0' not found (required by /home/dlm/torch/install/lib/libwarpctc.so)
> dlm@vm001nc6:~/code/deepspeech.pytorch$

[mit456]

I will try removing everything and then installing from scratch if that helps I will let you know.

[xmfbit]

Share my solution here. However I don’t know why… The environment is Ubuntu16.04 with CUDA8.0, GCC version of system is 5.4.0. And GCC version of Anaconda is 4.8.5. I installed PyTorch from source following the instructions on its GitHub repo page.

1. I add the two lines in ~/.zshrc file (if you are using bash, modify ~/.bashrc instead).

export CC="/usr/bin/gcc"
export CXX="/usr/bin/g++"

Make sure that after you export Anaconda, the value of environment variable CC is /usr/bin/gcc.

# print the value of CC
echo $CC

If not, it will fail when building the warp-ctc C++ code.

1. Inspired by @skaae, but I didn’t install GCC 4.8 (because the server is owned by the lab and I can’t modify things at will). I just run the following command:

conda install libgcc

1. After building the warpctc code, you should export CUDA_HOME if you want to enable GPU before running python installation script.

export CUDA_HOME=/path/to/cuda
cd pytorch_binding
python setup.py install

1. Test! After all of this, you can test it!

cd pytorch_binding/tests
python test.py

If it works well, it will print the gradient information like this:

➜  tests git:(pytorch_bindings) ✗ python test.py 
Variable containing:
 0.1770 -0.7081  0.1770  0.1770  0.1770
-0.8230  0.1770  0.2919  0.1770  0.1770
 0.2919  0.1770 -0.8230  0.1770  0.1770
-0.8213  0.1787  0.2665  0.1975  0.1787
[torch.cuda.FloatTensor of size 4x5 (GPU 0)]

.Variable containing:
 0.1770 -0.7081  0.1770  0.1770  0.1770
 0.1770 -0.8230  0.2919  0.1770  0.1770
 0.2919  0.1770 -0.8230  0.1770  0.1770
 0.1787  0.1787 -0.7335  0.1975  0.1787
[torch.cuda.FloatTensor of size 4x5 (GPU 0)]

.Variable containing:
 0.0000  0.0000  0.0000  0.0000  0.0000
 0.0000 -1.0000  1.0000  0.0000  0.0000
 1.0000  0.0000 -1.0000  0.0000  0.0000
 0.0000  0.0000  0.0000  0.0000  0.0000
[torch.cuda.FloatTensor of size 4x5 (GPU 0)]

.Variable containing:
 0.1770 -0.7081  0.1770  0.1770  0.1770
 0.1770  0.1770 -0.7081  0.1770  0.1770
[torch.cuda.FloatTensor of size 2x5 (GPU 0)]

.
----------------------------------------------------------------------
Ran 4 tests in 2.832s

OK