ModelArts DevEnviron provides Docker container images, which can run as preset containers. Certain preset images are built on common AI engine frameworks such as PyTorch, TensorFlow, and MindSpore. These images are named using the AI engines. Additionally, many common packages are preset in these images, relieving you from the package installation.
The images preset in ModelArts DevEnviron include:
Table 1 Preset x86 images
Engine |
Image |
PyTorch |
pytorch_2.1.0-cuda_12.1-py_3.10.6-ubuntu_22.04-x86_64 |
pytorch1.8-cuda10.2-cudnn7-ubuntu18.04 |
pytorch1.10-cuda10.2-cudnn7-ubuntu18.04 |
pytorch1.4-cuda10.1-cudnn7-ubuntu18.04 |
Tensorflow |
tensorflow2.1-cuda10.1-cudnn7-ubuntu18.04 |
tensorflow1.13-cuda10.0-cudnn7-ubuntu18.04 |
MindSpore |
mindspore1.7.0-cuda10.1-py3.7-ubuntu18.04 |
mindspore1.7.0-py3.7-ubuntu18.04 |
mindspore1.2.0-cuda10.1-cudnn7-ubuntu18.04 |
mindspore1.2.0-openmpi2.1.1-ubuntu18.04 |
No AI engine (base images dedicated for image customization) |
conda3-cuda10.2-cudnn7-ubuntu18.04 |
conda3-ubuntu18.04 |
x86-powered PyTorch Base Images
PyTorch contains three types of images:
Image 1: pytorch_2.1.0-cuda_12.1-py_3.10.6-ubuntu_22.04-x86_64
Table 2 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
PyTorch 2.1
|
Yes
(CUDA 10.2) |
swr.{region_id}.myhuaweicloud.com/atelier/pytorch_2_1:pytorch_2.1.0-cuda_12.1-py_3.10.6-ubuntu_22.04-x86_64-20250305173557-cb53968 |
PyPI package |
Ubuntu package |
torch 2.1.0+cu121
torchvision 0.16.0+cu121
torchaudio 2.1.0+cu121
ipykernel 6.7.0
ipython 8.33.0
jupyter-client 7.4.9
ma-cli 1.2.3
matplotlib 3.7.3
moxing-framework 2.2.10
numpy 1.24.2
opencv-python 4.8.0.76
pandas 2.2.2
Pillow 10.4.0
pip 22.0.4
psutil 6.0.0
PyYAML 6.0.2
scipy 1.12.0
scikit-learn 1.5.1
tornado 6.4.2
tensorboard 2.18.0 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libcudnn7
libcudnn7-dev
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
pandoc
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
Image 2: pytorch1.8-cuda10.2-cudnn7-ubuntu18.04
Table 3 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
PyTorch 1.8 |
Yes
(CUDA 10.2) |
swr.{region_id}.myhuaweicloud.com/atelier/pytorch_1_8:pytorch_1.8.0-cuda_10.2-py_3.7-ubuntu_18.04-x86_64-20220926104358-041ba2e |
PyPI package |
Ubuntu package |
torch 1.8.0
torchvision 0.9.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.4
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.1.0.5d9c87c8
numpy 1.19.5
opencv-python 4.1.2.30
pandas 1.1.5
Pillow 9.3.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.5.2
scikit-learn 0.22.1
tornado 6.2
tensorboard 2.1.1 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libcudnn7
libcudnn7-dev
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
pandoc
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
Image 3: pytorch1.10-cuda10.2-cudnn7-ubuntu18.04
Table 4 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
Pytorch 1.10 |
Yes
(CUDA 10.2) |
swr.{region_id}.myhuaweicloud.com/atelier/pytorch_1_10:pytorch_1.10.2-cuda_10.2-py_3.7-ubuntu_18.04-x86_64-20221008154718-2b3e39c |
PyPI package |
Ubuntu package |
torch 1.10.2
torchvision 0.11.3
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.4
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.1.0.5d9c87c8
numpy 1.19.5
opencv-python 4.1.2.30
pandas 1.1.5
Pillow 9.3.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.5.2
scikit-learn 0.22.1
tornado 6.2 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libcudnn7
libcudnn7-dev
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
pandoc
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
Image 4: pytorch1.4-cuda10.1-cudnn7-ubuntu18.04
Table 5 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
Pytorch 1.4 |
Yes
(CUDA 10.1) |
swr.{region_id}.myhuaweicloud.com/atelier/pytorch_1_4:pytorch_1.4-cuda_10.1-py37-ubuntu_18.04-x86_64-20220926104017-041ba2e |
PyPI package |
Ubuntu package |
torch 1.4.0
torchvision 0.5.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.7
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.1.0.5d9c87c8
numpy 1.19.5
opencv-python 4.1.2.30
pandas 1.1.5
Pillow 6.2.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.5.2
scikit-learn 0.22.1
tornado 6.2
tensorboard 2.1.1 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libcudnn7
libcudnn7-dev
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
pandoc
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
x86-powered TensorFlow Base Images
TensorFlow contains two types of images:
Image 1: tensorflow2.1-cuda10.1-cudnn7-ubuntu18.04
Table 6 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
Tensorflow 2.1 |
Yes
(CUDA 10.1) |
swr.{region_id}.myhuaweicloud.com/atelier/tensorflow_2_1:tensorflow_2.1.0-cuda_10.1-py_3.7-ubuntu_18.04-x86_64-20220926144607-041ba2e |
PyPI package |
Ubuntu package |
tensorflow 2.1.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.4
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.1.0.5d9c87c8
numpy 1.19.5
opencv-python 4.1.2.30
pandas 1.1.5
Pillow 9.3.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.5.2
scikit-learn 0.22.1
tornado 6.2
tensorboard 2.1.1 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libcudnn7
libcudnn7-dev
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
Image 2: tensorflow1.13-cuda10.0-cudnn7-ubuntu18.04
Table 7 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
Tensorflow 1.13-gpu |
Yes
(CUDA 10.0) |
swr.{region_id}.myhuaweicloud.com/atelier/tensorflow_1_13:tensorflow_1.13-cuda_10.0-py_3.7-ubuntu_18.04-x86_64-20220926104358-041ba2e |
PyPI package |
Ubuntu package |
tensorflow-gpu 1.13.1
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.6
ma-cli 1.2.3
matplotlib 3.5.2
modelarts 1.4.25
moxing-framework 2.0.1.rc0.ffd1c0c8
numpy 1.17.0
opencv-python 4.1.2.30
pandas 1.1.5
Pillow 6.2.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.2.2
scikit-learn 0.22.1
tornado 6.2 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libcudnn7
libcudnn7-dev
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
x86-powered MindSpore Base Images
MindSpore contains four types of images:
Image 1: mindspore1.7.0-cuda10.1-py3.7-ubuntu18.04
Table 8 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
MindSpore-gpu 1.7.0 |
Yes
(CUDA 10.1) |
swr.{region_id}.myhuaweicloud.com/atelier/mindspore_1_7_0:mindspore_1.7.0-cuda_10.1-py_3.7-ubuntu_18.04-x86_64-20220926104017-041ba2e |
PyPI package |
Ubuntu package |
mindspore-gpu 1.7.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.4
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.1.0.5d9c87c8
numpy 1.19.5
pandas 1.1.5
Pillow 9.3.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.5.2
scikit-learn 0.22.1
tornado 6.2
mindinsight 1.7.0
mindvision 0.1.0 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libcudnn7
libcudnn7-dev
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
Image 2: mindspore1.7.0-py3.7-ubuntu18.04
Table 9 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
MindSpore 1.7.0 |
No |
swr.{region_id}.myhuaweicloud.com/atelier/mindspore_1_7_0:mindspore_1.7.0-cpu-py_3.7-ubuntu_18.04-x86_64-20220926104017-041ba2e |
PyPI package |
Ubuntu package |
mindspore 1.7.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.6
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.1.0.5d9c87c8
numpy 1.19.5
pandas 1.1.5
Pillow 9.3.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.5.2
scikit-learn 0.22.1
tornado 6.2
mindinsight 1.7.0
mindvision 0.1.0 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
Image 3: mindspore1.2.0-cuda10.1-cudnn7-ubuntu18.04
Table 10 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
Mindspore-gpu 1.2.0 |
Yes
(CUDA 10.1) |
swr.{region_id}.myhuaweicloud.com/atelier/mindspore_1_2_0:mindspore_1.2.0-py_3.7-cuda_10.1-ubuntu_18.04-x86_64-20220926104106-041ba2e |
PyPI package |
Ubuntu package |
mindspore-gpu 1.2.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.6
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.1.0.5d9c87c8
numpy 1.19.5
pandas 1.1.5
Pillow 6.2.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.5.2
scikit-learn 0.22.1
tornado 6.2
mindinsight 1.2.0 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libcudnn7
libcudnn7-dev
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
Image 4: mindspore1.2.0-openmpi2.1.1-ubuntu18.04
Table 11 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
Mindspore 1.2.0 |
No |
swr.{region_id}.myhuaweicloud.com/atelier/mindspore_1_2_0:mindspore_1.2.0-py_3.7-ubuntu_18.04-x86_64-20220926104106-041ba2e |
PyPI package |
Ubuntu package |
mindspore 1.2.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.6
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.1.0.5d9c87c8
numpy 1.19.5
pandas 1.1.5
Pillow 6.2.0
pip 21.0.1
psutil 5.8.0
PyYAML 5.1
scipy 1.5.2
scikit-learn 0.22.1
tornado 6.2
mindinsight 1.2.0 |
automake
build-essential
ca-certificates
cmake
cpp
curl
ffmpeg
g++
gcc
gfortran
git
git-lfs
grep
libjpeg-dev:amd64
libjpeg8-dev:amd64
openssh-client
openssh-server
nginx
python3
rpm
screen
tar
tmux
unzip
vim
wget
zip |
x86-powered Custom Dedicated Base Images
ModelArts provides the following notebook base images powered by custom images (x86): conda3-cuda10.2-cudnn7-ubuntu18.04 and conda3-ubuntu18.04. These images do not have AI engines or related software packages. The image size is only 2 GB to 5 GB. You can use these images as base images and install your desired engine and dependency packages, improving scalability. In addition, these images are preset with some configurations required for starting the development environment. You can use these images after installing required software packages, without the need for any adaptations.
Such images are the most basic ones and have no component installed. They are small enough to facilitate image customization. If you need to use the OBS SDK, use ModelArts SDK instead to copy files. For details, see Transferring Files.
Image 1: conda3-cuda10.2-cudnn7-ubuntu18.04
Table 12 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
None |
Yes
(CUDA 10.2) |
swr.{region_id}.myhuaweicloud.com/atelier/user_defined_base:cuda_10.2-ubuntu_18.04-x86_64-20221008154718-2b3e39c |
PyPI package |
Ubuntu package |
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.9
ma-cli 1.2.3
matplotlib 3.5.2
modelarts 1.4.25
moxing-framework 2.1.6.879ab2f4
numpy 1.21.6
pandas 1.3.5
Pillow 9.5.0
pip 20.3.3
psutil 5.9.4
PyYAML 6.0
scipy 1.7.3
tornado 6.2 |
automake
build-essential
ca-certificates
cmake
cpp
curl
g++
gcc
gfortran
grep
libcudnn7
libcudnn7-dev
nginx
python3
rpm
tar
unzip
vim
wget
zip |
Image 2: conda3-ubuntu18.04
Table 13 Information about the image
AI Engine |
Whether to Use GPUs
(CUDA Version) |
URL |
Dependency |
None |
No |
swr.{region_id}.myhuaweicloud.com/atelier/user_defined_base:ubuntu_18.04-x86_64-20221008154718-2b3e39c
Example:
CN North-Beijing4
swr.cn-north-4.myhuaweicloud.com/atelier/user_defined_base:ubuntu_18.04-x86_64-20221008154718-2b3e39c
CN East-Shanghai1
swr.cn-east-3.myhuaweicloud.com/atelier/user_defined_base:ubuntu_18.04-x86_64-20221008154718-2b3e39c
CN South-Guangzhou
swr.cn-south-1.myhuaweicloud.com/atelier/user_defined_base:ubuntu_18.04-x86_64-20221008154718-2b3e39c |
PyPI package |
Ubuntu package |
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.9
ma-cli 1.2.3
matplotlib 3.5.2
modelarts 1.4.25
moxing-framework 2.1.6.879ab2f4
numpy 1.21.6
pandas 1.3.5
Pillow 9.5.0
pip 20.3.3
psutil 5.9.4
PyYAML 6.0
scipy 1.7.3
tornado 6.2 |
automake
build-essential
ca-certificates
cmake
cpp
curl
g++
gcc
gfortran
grep
nginx
python3
rpm
tar
unzip
vim
wget
zip |
Arm-powered MindSpore Base Images
Arm-powered MindSpore contains three types of images:
Image 1: mindspore_1.10.0-cann_6.0.1-py_3.7-euler_2.8.3
Table 14 Information about the image
AI Engine |
URL |
Dependency |
Mindspore-Ascend 1.10.0 |
{region_id}.myhuaweicloud.com/atelier/mindspore_1_10_ascend:mindspore_1.10.0-cann_6.0.1-py_3.7-euler_2.8.3-aarch64-d910-20230303173945-815d627
Example:
CN North-Beijing4
swr.cn-north-4.myhuaweicloud.com/atelier/mindspore_1_10_ascend:mindspore_1.10.0-cann_6.0.1-py_3.7-euler_2.8.3-aarch64-d910-20230303173945-815d627 |
PyPI package |
YUM package |
mindspore-ascend 1.10.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.5
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.0.1.rc0.ffd1c0c8
numpy 1.21.2
pandas 1.1.3
Pillow 9.4.0
pip 21.0.1
psutil 5.7.0
PyYAML 5.3.1
scipy 1.5.4
scikit-learn 0.24.0
tornado 6.2
mindinsight 1.9.0 |
cmake
cpp
curl
ffmpeg
g++
gcc
git
grep
python3
rpm
tar
unzip
wget
zip |
Image 2: mindspore_1.9.0-cann_6.0.0-py_3.7-euler_2.8.3
Table 15 Information about the image
AI Engine |
URL |
Dependency |
MindSpore 1.9.0 |
swr.{region_id}.myhuaweicloud.com/atelier/mindspore_1_9_ascend:mindspore_1.9.0-cann_6.0.0-py_3.7-euler_2.8.3-aarch64-d910-20221116111529 |
PyPI package |
YUM package |
mindspore-ascend 1.9.0
ipykernel 6.7.0
ipython 7.34.0
jupyter-client 7.4.5
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.0.1.rc0.ffd1c0c8
numpy 1.21.2
pandas 1.1.3
Pillow 9.3.0
pip 22.3.1
psutil 5.7.0
PyYAML 5.3.1
scipy 1.5.4
scikit-learn 0.24.0
tornado 6.2
mindinsight 1.9.0 |
cmake
cpp
curl
ffmpeg
g++
gcc
git
grep
python3
rpm
tar
unzip
wget
zip |
Image 3: mindspore1.7.0-cann5.1.0-py3.7-euler2.8.3
Table 16 Information about the image
AI Engine |
URL |
Dependency |
Mindspore-Ascend 1.7.0 |
swr.{region_id}.myhuaweicloud.com/atelier/mindspore_1_7_0:mindspore_1.7.0-cann_5.1.0-py_3.7-euler_2.8.3-aarch64-d910-20220906 |
PyPI package |
YUM package |
mindspore-ascend 1.7.0
ipykernel 5.3.4
ipython 7.34.0
jupyter-client 7.3.4
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.0.1.rc0.ffd1c0c8
numpy 1.21.2
pandas 1.1.3
Pillow 9.2.0
pip 22.1.2
psutil 5.7.0
PyYAML 5.3.1
scipy 1.5.4
scikit-learn 0.24.0
tornado 6.2
mindinsight 1.7.0 |
cmake
cpp
curl
ffmpeg
g++
gcc
git
grep
python3
rpm
tar
unzip
wget
zip |
Arm-powered TensorFlow Base Images
Arm-powered TensorFlow contains two types of images:
Image 1: tensorflow1.15-mindspore1.7.0-cann5.1.0-euler2.8-aarch64
Table 17 Information about the image
AI Engine |
URL |
Dependency |
Mindspore-Ascend 1.7.0 |
swr.{region_id}.myhuaweicloud.com/atelier/notebook2.0-mul-kernel-arm-ascend-cp37:5.0.1-c81-20220726 |
PyPI package |
YUM package |
mindspore-ascend 1.7.0
ipykernel 6.7.0
ipython 7.29.0
jupyter-client 7.0.6
ma-cli 1.2.3
matplotlib 3.1.2
modelarts 1.4.25
moxing-framework 2.0.0.rc2.4b57a67b
numpy 1.17.5
pandas 1.1.3
Pillow 7.0.0
pip 21.2.4
psutil 5.7.0
PyYAML 5.3.1
scipy 1.5.4
scikit-learn 0.24.0
tornado 6.1
mindinsight 1.7.0 |
cmake
cpp
curl
ffmpeg
g++
gcc
git
grep
python3
rpm
tar
unzip
wget
zip |
Image 2: tensorflow1.15-cann5.1.0-py3.7-euler2.8.3
Table 18 Information about the image
AI Engine |
Whether to Use Ascend
(CANN Version) |
URL |
Dependency |
Tensorflow 1.15 |
Yes
(CANN 5.1) |
swr.{Region ID}.{Region domain name}./atelier/
tensorflow_1_15_ascend:tensorflow_1.15-cann_5.1.0-py_3.7-euler_2.8.3-aarch64-d910-20220906 |
PyPI package |
YUM package |
tensorflow 1.15.0
tensorboard 1.15.0
ipykernel 5.3.4
ipython 7.34.0
jupyter-client 7.3.4
ma-cli 1.2.3
matplotlib 3.5.1
modelarts 1.4.25
moxing-framework 2.0.1.rc0.ffd1c0c8
numpy 1.17.5
pandas 0.24.2
Pillow 9.2.0
pip 22.1.2
psutil 5.7.0
PyYAML 5.3.1
scipy 1.3.3
scikit-learn 0.20.0
tornado 6.2 |
ca-certificates.noarch
cmake
cpp
curl
gcc-c++
gcc
gdb
grep
nginx
python3
rpm
tar
unzip
vim
wget
zip |