For AI agents: the complete documentation index is available at https://docs.ovhcloud.com/en/llms.txt, the full documentation bundle is available at https://docs.ovhcloud.com/en/llms-full.txt, and this page is available as Markdown at https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-training-submit-job.md.
Skip to content
Appearance
Console(the OVHcloud Console, opens in a new tab)

AI Training - Getting started

View as Markdown

Learn how to submit an AI Training job via UI

Objective

This guide covers the initialisation of AI Training and the submission of jobs through the OVHcloud Console.

Requirements

  • a Public cloud project
  • optionally container objects to attach data to the job at step 8, see our create data container guide.

OVHcloud Console Access

  • Direct link:
  • Navigation path: Public Cloud > Select your project > AI > AI Training

Instructions

Step 1 - Going to the AI Training menu

In the left-hand menu of your Public Cloud project, click AI Training under AI.

Step 2 - Starting a job submission

If the project has not been authorised for AI services yet, click Authorise the project in the Authorise this project for AI services banner. Upon authorising the project you grant OVHcloud access to your Object Storage containers. This access is only used to synchronise your data within AI Training with your containers.

From the jobs list you can start the job submission by clicking the Create a job button. The Create a job page opens: a summary of your configuration and its estimated price per hour is displayed on the right-hand side.

Step 3 - Naming your job and selecting a region

Give a name to your job in the Name section. This will make it easier to manage when you have multiple jobs running.

Each job is executed in an OVHcloud region. Each region has its own AI Training cluster with potentially varying capabilities. For more information, see the capabilities.

Select the desired region in the Region section.

Step 4 - Specifying the amount of resources

In the Flavor section, select the CPU or GPU flavor you need for your training workload, then adjust the Quantity.

The max amount of GPUs or CPUs you can select for your job is region-dependent. If you choose a GPU, a fixed ratio of CPU is applied based on the number of GPUs. Similarly, there is a fixed ratio of Memory based on the number of CPUs. For more information see the capabilities.

Once the amount of resources is set you can see a preview of the billing rate in the summary.

Step 5 - Providing a Docker image

A job is basically a Docker container that is run within the OVHcloud infrastructure. You need to provide a Docker image to be executed in the Docker image section. There are several options you can choose from:

Preset Images

With Ready-to-use image catalogue, OVHcloud provides a set of images from which you can choose to ease the submission of your first jobs. Provided images are essentially a JupyterLab environment bundled with some Deep Learning technology such as Tensorflow or MXNet. These images are planned to be removed, as the product is more focused on using custom images.

Custom Images

Preset images cannot cover all your needs so you can specify your own image if necessary with Custom image. You can use any image that is accessible from AI Training.

This includes public images (e.g. Dockerhub), images within the shared registry or images in your added private registry. For more information, see how to add a private registry.

Step 6 - Privacy Settings

Next, select your privacy settings in the Privacy section: Private (access requires a token) or Public (access without a token).

Warning

Public access will expose your data and code to anyone getting the AI Training job link. Be careful and don't use it with sensitive data. On the other hand, Private access will ask a user and password combination or an AI token to access the job content, ensuring a secure environment.

Step 7 - Job lifecycle

Warning

Actually our main motivation is to keep the plateform up-to-date, in matter of security patches and new features alignment, so we need to update and restart hosts from time to time, which requires to get them free from any customer workload. This also allows us to address host-located incidents, e.g. when a single GPU fails due to faulty hardware (only 1 customer impacted, but we still need to stop the host for bringing it to PCI maintenance, so every remaining workload must be evicted or rescheduled elsewhere).

By default, your job will automatically shut down after 7 consecutive days of being in RUNNING status (Expiration section). Rest assured that all your settings and data will be preserved. You have the option to turn on Automatically restart after timeout, which will automatically restart your job every 7 days, ensuring minimal disruption to your workflow. Alternatively, you can also contact our support to extend this automatic restart period.

Step 8 - Advanced configuration

This step is optional.

In the Advanced configuration section, turn on the switch of each option you need.

First, the Docker image you provided in Step 5 includes an entrypoint for your container. You can override this entrypoint by specifying your own command (Commands).

Then, by default, your AI Training job comes with ephemeral storage (local storage). But in this step, you can also link Object Storage containers (Volumes) and Git repositories (Git Public) to your job, either as input for your training workload or as output for your results (e.g. model weights).

If you want to learn more about configuring containers and Git repositories in the job, you can refer to this model training example. For now, we will launch a classic job without any external volumes added to it.

You can also add Labels to your job.

Finally, SSH key allows you to access your job remotely.

Step 9 - Submitting your job

Before submission, check the overview of the job you configured in the summary. Click CLI equivalent to get the equivalent command to use with the ovhai CLI.

Click Create the job to confirm and launch the creation of your job on the cluster.

When your job is created, it will appear in the list of your AI Training jobs.

Step 10 - Consulting your job

From this list you can access your job details by clicking on its name. The Overview tab includes several components, like the job resources, status, life cycle and access URL (Connection details card). This URL is of the form https://<JOB-ID>.job.<REGION>.ai.cloud.ovh.net/. Moreover, you can access any available port by adding it to the job URL this way: https://<JOB-ID>-<PORT>.job.<REGION>.ai.cloud.ovh.net/.

You can also check your job logs in the Observability tab.

Step 11 - Stopping your job

If you are done using your job, if your model converged prematurely or if you just wish to interrupt your job you can do so from the jobs list and the job page.

From the list of jobs, click the ... button at the far right of the entry and interrupt the job by clicking Stop. Alternatively, from the Overview tab of the job, click Stop in the General information card. Confirm with Stop: the run is interrupted and cannot be resumed.

After that, if you no longer need your job, you can delete it. To do so, click the ... button on the row of the job, then select Delete (or click Delete job in the General information card) and confirm.

Step 12 - Cloning your job

You can also click Clone job (from the ... button of the job list or the General information card) to clone the job with the same configuration. This will launch a new job identical to the original (e.g., example-job), allowing you to rerun your workflow without re‑configuring the job.

Go further

The AI Training service is mainly supposed to be used through the ovhai CLI. The OVHcloud Console only offers a subset of the features and is meant to help you get started before using the CLI. Discover how to install the OVHcloud AI CLI.

For training or technical assistance implementing our solutions, contact your sales representative or visit our Professional Services page to request a quote and have your project analysed by our experts.

Feedback

Please send us your questions, feedback and suggestions to improve the service:

Was this page helpful?