Get started with Pytorch
Prerequisites
You need the following:
- Git
- Python >=3.8
- pip
Some libraries installed on Graal:
- tensorboardX==1.6.0
- aiohttp==3.8.1
- h5py==3.6.0
- pandas==1.4.2
- pyarrow==7.0.0
- protobuf==3.19.4
- prometheus-client==0.13.1
- python-socketio==5.4.1
- s3fs==2022.2.0
- gcsfs==2022.2.0
- adlfs==2022.2.0
- requests>=2.21.0
- urllib3>=1.21.1
Distributed feature
In order to use Pytorch distributed it is necessary to adapt your code. For more information here is the official documentation of Pytorch distributed_pytorch Graal takes care of installing and preparing the Pytorch operator, the job configuration and its creation in Kubernetes.
Example
Clone the example project and use pip to build it. The example project named pytorch_examples shows how to use Pytorch distributed in a sample code.