Working with Watson Machine Learning Accelerator notebooks in IBM® Cloud Pak for Data
Use IBM Watson® Machine Learning Accelerator notebooks to run elastic distributed training workloads in Cloud Pak for Data.
- Before using Watson Machine Learning Accelerator notebooks, complete the following tasks:
- Submitting workloads using Watson Machine Learning Accelerator notebooks, see Using elastic distributed training API in notebooks.
Considerations for upgrade
Before you upgrade Watson Machine Learning Accelerator, you can backup existing notebooks to your local system. After upgrading, you are able to restore these notebooks.
When upgrading all previously installed Watson Machine Learning Accelerator notebooks are removed.
- Login to the Watson Machine Learning Accelerator console.
- Go to and start the notebook server.
- In the JupyterLab page, navigate to the notebook file you want to migrate using the left menu.
- Right-click on the notebook file.
- Click Download to save the notebook file to your local system.
- Navigate to the user project page in Watson Studio.
- Click New Asset.
- Select Jupyter Notebook Editor.
- Click the From File tab, select the Watson Machine Learning Accelerator notebook runtime and upload the notebook file you just save.
Install the Watson Machine Learning Accelerator notebook runtime configuration
To be able to use the Watson Machine Learning Accelerator notebooks from Watson Studio, the administrator must install the Watson Machine Learning Accelerator notebook runtime configuration.
Before you begin
In order to install the Watson Machine Learning Accelerator notebook runtime configuration, you need to download the required scripts from IBM Git, these scripts configure the Watson Machine Learning Accelerator notebook runtime and the JupyterLab runtime configuration.
You must be a Cloud Pak for Data administrator to run the scripts.
Steps
- Download install_wmla_notebook_runtime.sh
- Install
jq
.yum install jq
- Run the install_wmla_notebook_runtime.sh
script:
bash install_wmla_notebook_runtime.sh bash install_wmla_notebook_runtime.sh -u <user> -c <cpd_host> bash install_wmla_notebook_runtime.sh -u <user> -x <password> -c <cpd_host> bash install_wmla_notebook_runtime.sh -h
- Download install_wmla_jupyterlab_runtime.sh
- Install
jq
.yum install jq
- Run the install_wmla_notebook_runtime.sh
script:
bash install_wmla_jupyterlab_runtime.sh bash install_wmla_jupyterlab_runtime.sh -u <user> -c <cpd_host> bash install_wmla_jupyterlab_runtime.sh -u <user> -x <password> -c <cpd_host> bash install_wmla_jupyterlab_runtime.sh -h
Create the Watson Machine Learning Accelerator environment template
- From the
Manage
tab of your project, select theEnvironments
page, and clickNew template
underTemplates
. - Enter a name and a description, for example:
wmla
. - Select the environment type as
Default
. - Choose the hardware configuration size and select
Runtime 23.1 on Python 3.10 (jupyter-wmla-py)
in theSoftware version
dropdown. This template has a hardware configuration of 1 CPU and 2 GP RAM. - Click the
Create
button.
After you create the Watson Machine Learning Accelerator environment template, you can select to run your notebook in that environment at the time you create the notebook.
In a project, you can run more than one notebook using the same Watson Machine Learning Accelerator environment template. This means that if you open a second notebook with the same environment template in the same project, a second kernel is started in the same runtime. The runtime resources are shared by the kernels that you start in the runtime. The runtime is started per single user and not per notebook.
Open a notebook
You can find and open notebooks from the Assets page of the project.
You can change the environment of the notebook to use the Watson Machine Learning Accelerator environment template.
To open the notebook, click on the notebook for it to open in edit mode.