Installation Guidelines
Confirming Version Matching Relationship
The currently supported hardware is the Atlas 800T A2、Atlas 800I A2、Atlas 900 A3 SuperPoD.
The current recommended Python version for the suite is 3.11.4.
MindSpore Transformers |
MindSpore |
CANN |
Firmware & Drivers |
Mirror Links |
|---|---|---|---|---|
1.6.0 |
Currently MindSpore Transformers recommends using a software package relationship as above.
Historical version matching relationship:
MindSpore Transformers |
MindSpore |
CANN |
Firmware & Drivers |
|---|---|---|---|
1.5.0 |
|||
1.3.2 |
|||
1.3.0 |
|||
1.2.0 |
Installing Dependent Software
Install Firmware and Driver: Download the firmware and driver package through the Confirming Version Matching Relationship to download the installation package, and refer to the Ascend official tutorial for installation.
Install CANN and MindSpore: Follow the Manual Installation section on the MindSpore website for installation.
Installing MindSpore Transformers
MindSpore Transformers supports both source code compiled installation and pip installation.
Installation by Source Code Compilation
Users can compile and install MindSpore Transformers by executing the following command:
git clone -b r1.6.0 https://gitee.com/mindspore/mindformers.git
cd mindformers
bash build.sh
Installation by pip
Users can install MindSpore Transformers by executing the following command:
pip install https://ms-release.obs.cn-north-4.myhuaweicloud.com/2.7.0rc1/MindFormers/any/mindformers-1.6.0-py3-none-any.whl --trusted-host ms-release.obs.cn-north-4.myhuaweicloud.com -i https://repo.huaweicloud.com/repository/pypi/simple
This installation method requires access to the public network. If you are in an internal network environment, please ensure that the network connection is configured correctly.
This method only installs the basic software package of MindSpore Transformers. Model files and scripts, etc., please obtain from the MindSpore Transformers Gitee repository.
Installation Verification
To determine whether MindSpore Transformers has been successfully installed, execute the following code:
python -c "import mindformers as mf;mf.run_check()"
A similar result as below proves that the installation was successful:
- INFO - All checks passed, used **** seconds, the environment is correctly set up!