Infrastructure/software/pytorch: Difference between revisions

From Nordic Language Processing Laboratory
Jump to navigation Jump to search
Line 11: Line 11:
in a Python 3.5 virtual environment.
in a Python 3.5 virtual environment.


<code>
<pre>
module purge
module purge
module load cuda/8.0
module load cuda/8.0
python3/3.5.0
module load python3/3.5.0
</code>
</pre>


<code>
<pre>
cd /projects/nlpl/software
cd /projects/nlpl/software
mkdir pytorch
mkdir pytorch
virtualenv pytorch/0.4.1
virtualenv pytorch/0.4.1
</code>
</pre>


Next, we need to create a module defintion, in this case
Next, we need to create a module defintion, in this case
<tt>/projects/nlpl/software/modulefiles/nlpl-pytorch/0.4.1</tt>.
<tt>/projects/nlpl/software/modulefiles/nlpl-pytorch/0.4.1</tt>.
<pre>
module load nlpl-pytorch/0.4.1
pip install --upgrade pip
pip install --upgrade numpy pillow six
pip install torch torchvision
</pre>

Revision as of 13:23, 14 September 2018

Background

PyTorch is one of the widely used programming frameworks for Deep Learning; in and of itself it is not NLP-specific, but it is a critical dependency for several of the NLPL strands of activity.

Installation on Abel

The module nlpl-pytorch provides a PyTorch installation in a Python 3.5 virtual environment.

module purge
module load cuda/8.0
module load python3/3.5.0
cd /projects/nlpl/software
mkdir pytorch
virtualenv pytorch/0.4.1

Next, we need to create a module defintion, in this case /projects/nlpl/software/modulefiles/nlpl-pytorch/0.4.1.

module load nlpl-pytorch/0.4.1
pip install --upgrade pip
pip install --upgrade numpy pillow six
pip install torch torchvision