Machine Learning > Deep Learning Instance > User Guide

Create a Deep Learning Instance

To use the Deep Learning Instance, you must first create an instance.

deeplearninginstance_guide_en_01_20211013.png

Click the Create Deep Learning Instance button, and you will be taken to Machine Learning > Deep Learning Instance > Create Instance.

Deep Learning Framework Instance provides the following versions of software:

Software Version Installation method
TensorFlow 2.4.1 pip, Reference
PyTorch 1.7.1 conda, Reference
Python 3.8.11 conda
OS Ubuntu 18.04 LTS N/A
NVIDIA Driver 450.102.04 apt
NVIDIA CUDA 11.0 apt
NVIDIA cuDNN 8.0.4 apt
NVIDIA NCCL 2.7.8 apt
NVIDIA TensorRT 7.1.3 apt
Intel oneAPI MKL 2021.4.0 apt

deeplearninginstance_guide_en_02_20211013.png

After completing the setup, create an instance. For more information on instance creation, see Instance Overview.

Check Installed Development Environment

Use the conda command to check the development environment installed by Miniconda.

$ conda info --envs
# conda environments:
#
                         /opt/intel/oneapi/intelpython/latest
                         /opt/intel/oneapi/intelpython/latest/envs/2021.4.0
base                  *  /root/miniconda3
pt_py38                  /root/miniconda3/envs/pt_py38
tf2_py38                 /root/miniconda3/envs/tf2_py38

[Note]

For more detailed instructions, refer to Miniconda documentation.

How to Use TensorFlow

First, activate the TensorFlow environment.

(base) root@b64e6a035884:~# conda activate tf2_py38
(tf2_py38) root@b64e6a035884:~#

Test TensorFlow training as follows:

$ cd ~/
$ git clone https://github.com/tensorflow/models.git
$ cd models
$ git checkout tags/v2.4.0
$ git status
HEAD detached at v2.4.0
nothing to commit, working tree clean

$ mkdir $HOME/models/model
$ mkdir $HOME/models/dataset
$ vim train.sh
#!/bin/bash


export PYTHONPATH=$HOME/models
export NCCL_DEBUG=INFO
MODEL_DIR=$HOME/models/model
DATA_DIR=$HOME/models/dataset
# Set when one or more GPU is used
NUM_GPUS=1 # Example: NUM_GPUS=2

python $HOME/models/official/vision/image_classification/mnist_main.py \
  --model_dir=$MODEL_DIR \
  --data_dir=$DATA_DIR \
  --train_epochs=2 \
  --distribution_strategy=mirrored \ # Set when one or more GPU is used
  --num_gpus=$NUM_GPUS \ # Set when one or more GPU is used
  --download

$ chmod +x train.sh
$ ./train.sh

[Note]

For more detailed instructions, refer to TensorFlow Tutorial.

How to Use PyTorch

First, activate the PyTorch environment.

(tf2_py38) root@b64e6a035884:~# conda deactivate
(base) root@b64e6a035884:~# conda activate pt_py38
(pt_py38) root@b64e6a035884:~#

Test PyTorch training as follows:

$ cd ~/
$ git clone https://github.com/pytorch/examples.git
$ cd examples/mnist
$ python main.py --epochs 1

[Note]

For more detailed instructions, refer to PyTorch Tutorial.

Table of Contents
TOP