Submitting Workloads

In the Environment pane select or create a new environment. Use the search box to find environments that are not listed.

1. In the Set the connection for your tool(s) pane, choose a tool for your environment (if available). In the Access pane, edit the field and choose a type of access. Everyone allows all users in the platform to access the selected tool. Group allows you a select a specific group of users (Identity provider group). Press +Group to add more groups. User allows you to grant access individual users (by user email) in the platform. Press +User to add more users. (optional)
2. In the Runtime settings field, Set commands and arguments for the container running in the pod. (optional)
3. In the Environment variable field, you can set one or more environment variables. (optional)

In the Compute resource pane, select resources for your trainings or create a new compute resource. Use the search box to find resources that are not listed. Press More settings to use Node Affinity to limit the resources to a specific node.

Open the Volume pane, and press Volume to add a volume to your training.

1. Select the Storage class from the dropdown.
2. Select the Access mode from the dropdown.
3. Enter a claim size, and select the units.
4. Select a Volume system, mode from the dropdown.
5. Enter the Container path for volume target location.
6. Select a Volume persistency.

In the Data sources pane, select a data source. If you need a new data source, press add a new data source. For more information, see Creating a new data source When complete press, Create Data Source.

In the General pane, add special settings for your training (optional):

1. Toggle the switch to allow the workspace to exceed the project's quota.
2. Set the backoff limit before workload failure, this can be changed, if necessary. Use integers only. (Default = 6, maximum = 100, minimum = 0).
3. Press Auto-deletion to delete the training automatically when it either completes or fails. You can configure the timeframe in days, hours, minuets, and seconds. If the timeframe is set to 0, the training will be deleted immediately after it completes or fails. (default = 30 days)
4. Press Annotation to a name and value to annotate the training. Repeat this step to add multiple annotations.
5. Press Label to a name and value to label the training. Repeat this step to add multiple labels.

When complete, press Create workspace.

1. In the Set the connection for your tool(s) pane, choose a tool for your environment (if available). In the Access pane, edit the field and choose a type of access. Everyone allows all users in the platform to access the selected tool. Group allows you a select a specific group of users (Identity provider group). Press +Group to add more groups. User allows you to grant access individual users (by user email) in the platform. Press +User to add more users. (optional)
2. In the Runtime settings field, Set commands and arguments for the container running in the pod. (optional)
3. In the Environment variable field, you can set one or more environment variables. (optional)

In the Compute resource pane:

1. Select the number of workers for your training.
2. Select Compute resources for your training or create a new compute resource. Use the search box to find resources that are not listed. Press More settings to use Node Affinity to limit the resources to a specific node.

(Optional) Open the Volume pane, and press Volume to add a volume to your training.

1. Select the Storage class from the dropdown.
2. Select the Access mode from the dropdown.
3. Enter a claim size, and select the units.
4. Select a Volume system, mode from the dropdown.
5. Enter the Container path for volume target location.
6. Select a Volume persistency. Choose Persistent or Ephemeral.

(Optional) In the Data sources pane, select a data source. If you need a new data source, press add a new data source. For more information, see Creating a new data source When complete press, Create Data Source.

(Optional) In the General pane, add special settings for your training (optional):

1. Set the backoff limit before workload failure, this can be changed, if necessary. Use integers only. (Default = 6, maximum = 100, minimum = 0).
2. Press Auto-deletion to delete the training automatically when it either completes or fails. You can configure the timeframe in days, hours, minuets, and seconds. If the timeframe is set to 0, the training will be deleted immediately after it completes or fails. (default = 30 days)
3. Press Annotation to a name and value to annotate the training. Repeat this step to add multiple annotations.
4. Press Label to a name and value to label the training. Repeat this step to add multiple labels.

If you if selected Workers & master Press Continue to Configure the master and go to the next step. If not, then press Create training.

If you do not want a different setup for the master, press Create training. If you would like to have a different setup for the master, toggle the switch to enable to enable a different setup.

1. In the Environment pane select or create a new environment. Use the search box to find environments that are not listed. Press More settings to add an Environment variable or to edit the Command and Arguments field for the environment you selected.
   1. In the Set the connection for your tool(s) pane, choose a tool for your environment (if available). In the Access pane, edit the field and choose a type of access. Everyone allows all users in the platform to access the selected tool. Group allows you a select a specific group of users (Identity provider group). Press +Group to add more groups. User allows you to grant access individual users (by user email) in the platform. Press +User to add more users. (optional)
   2. In the Runtime settings field, Set commands and arguments for the container running in the pod. (optional)
   3. In the Environment variable field, you can set one or more environment variables. (optional)
2. In the Compute resource pane, select a Compute resources for your training or create a new compute resource. Use the search box to find resources that are not listed. Press More settings to use Node Affinity to limit the resources to a specific node.

3. (Optional) Open the Volume pane, and press Volume to add a volume to your training.

   1. Select the Storage class from the dropdown.
   2. Select the Access mode from the dropdown.
   3. Enter a claim size, and select the units.
   4. Select a Volume system, mode from the dropdown.
   5. Enter the Container path for volume target location.
   6. Select a Volume persistency. Choose Persistent or Ephemeral.

4. (Optional) In the Data sources pane, select a data source. If you need a new data source, press add a new data source. For more information, see Creating a new data source When complete press, Create Data Source.

!!! Note * Data sources that have private credentials, which have the status of issues found, will be greyed out. * Data sources can now include Secrets.

1. (Optional) In the General pane, add special settings for your training (optional):

   1. Set the backoff limit before workload failure, this can be changed, if necessary. Use integers only. (Default = 6, maximum = 100, minimum = 0).
   2. Press Auto-deletion to delete the training automatically when it either completes or fails. You can configure the timeframe in days, hours, minuets, and seconds. If the timeframe is set to 0, the training will be deleted immediately after it completes or fails. (default = 30 days)
   3. Press Annotation to a name and value to annotate the training. Repeat this step to add multiple annotations.
   4. Press Label to a name and value to label the training. Repeat this step to add multiple labels.

When your training configuration is complete. press Create training.

In the Inference by type pane select Custom or model.

When you select Model:

1. Select a catalog. Choose from Run:ai or Hugging Face.
   1. If you choose Run:ai, select a model from the tiles. Use the search box to find a model that is not listed. If you can't find the model, see your system administrator.
   2. If you choose Hugging Face, go to the next step.
2. In the Inference name field, enter a name for the workload.
3. In the Credentials field, enter the token to access the model catalog.
4. If you selected Hugging Face, enter the name of the model in the Model Name section. This will not appear if you selected Run:ai.

5. In the Compute resource field, select a compute resource from the tiles.

   1. In the Replica autoscaling section, set the minimum and maximum replicas for your inference.
   2. In the Set conditions for creating a new replica section, use the drop down to select from Throughput (Requests/sec), Latency (milliseconds), or Concurrency (Requests/sec). Then set the value. (default = 100) This section will only appear if you have 2 or more set as the maximum.
   3. In the Set when replicas should be automatically scaled down to zero section, from the drop down select Never, After one, five, 15 or 30 minutes of inactivity.

   Note

   When automatic scaling to zero is enabled, the minimum number of replicas is 0.

   1. In the Nodes field, change the order of priority of the node pools, or add a new node pool to the list.

6. When complete, press Create inference.

When you select Custom:

1. In the Inference name field, enter a name for the workload.
2. In the Environment field, select an environment. Use the search box to find an environment that is not listed. If you can't find an environment, press New environment or see your system administrator.
   1. In the Set the connection for your tool(s) pane, choose a tool for your environment (if available). In the Access pane, edit the field and choose a type of access. Everyone allows all users in the platform to access the selected tool. Group allows you a select a specific group of users (Identity provider group). Press +Group to add more groups. User allows you to grant access individual users (by user email) in the platform. Press +User to add more users. (optional)
   2. In the Runtime settings field, Set commands and arguments for the container running in the pod. (optional)
   3. In the Environment variable field, you can set one or more environment variables. (optional)

3. In the Compute resource field, select a compute resource from the tiles. Use the search box to find a compute resource that is not listed. If you can't find an environment, press New compute resource or see your system administrator.

   1. In the Replica autoscaling section, set the minimum and maximum replicas for your inference.
   2. In the Set conditions for creating a new replica section, use the drop down to select from Throughput (Requests/sec), Latency (milliseconds), or Concurrency (Requests/sec). Then set the value. (default = 100) This section will only appear if you have 2 or more set as the maximum.
   3. In the Set when replicas should be automatically scaled down to zero section, from the drop down select Never, After one, five, 15 or 30 minutes of inactivity.

   Note

   When automatic scaling to zero is enabled, the minimum number of replicas is 0.

4. In the Data sources field, add a New data source. (optional)

   Note

   • Data sources that are not available will be greyed out.
   • Assets that are cluster syncing will be greyed out.
   • Only PVC, Git, and ConfigMap resources are supported.

5. In the General field you can:

   1. Add an Auto-deletion time. This sets the timeframe between inference completion/failure and auto-deletion. (optional) (default = 30 days)
   2. Add one or more Annotation. (optional)
   3. Add one or more Labels. (optional)
6. When complete, press Create inference.