CLI - Lancer un AI Notebook (EN)
Objective
This guide covers the submission of new AI Notebooks through the ovhai CLI.
Requirements
To follow this guide, ensure that you have the ovhai CLI installed and that you are logged in as a user of one of your Public Cloud Projects, with the necessary permissions (at least AI Training Operator and ObjectStore operator), to use OVHcloud AI Products.
For more information on installing and configuring the ovhai CLI, see the guide on Installing the OVHcloud AI CLI.
Instructions
Run a new notebook
To run a new notebook, the basic command is ovhai notebook run. This command needs to be followed by some arguments that will indicate the specifications of your notebook.
You can see the full list of options by running ovhai notebook run --help or ovhai notebook run -h:
ovhai notebook run -h
Run a new notebook
Usage: ovhai notebook run [OPTIONS] <FRAMEWORK_ID> <EDITOR_ID>
Arguments:
<FRAMEWORK_ID> Framework ID to use in the notebook The list of available frameworks is available under capabilities `ovhai capabilities framework`
<EDITOR_ID> Editor ID to use in the notebook The list of available editors is available under capabilities `ovhai capabilities editor`
Options:
--token <TOKEN>
Authentication using Token rather than OAuth
--unsecure-http
HTTP services inside job will not require authentication to be accessed from the outside
-g, --gpu <GPU>
Number of GPUs
-f, --flavor <flavor>
the flavor to use, `ovhai capabilities flavor list` to get the whole list
-c, --cpu <CPU>
Number of CPUs (ignored if GPUs is specified)
-v, --volume <VOLUME_DEFINITION>
Volumes mounted on the image (multiple)
-n, --name <NAME>
Optional name, only informative
-l, --label <name=value>
Optional labels, only informative
-o, --output <OUTPUT>
Command output format [possible values: json, yaml]
-s, --ssh-public-keys <ssh-public-key-file>
Enable the job ssh feature, specify each ssh public key files or give the public key directly
--framework-version <FRAMEWORK_VERSION>
Version of the selected framework
-e, --envvar <name=value>
Environment variable to be set inside the notebook
--restart-on-timeout [<RESTART_ON_TIMEOUT>]
Automatically restart on timeout [possible values: true, false]
--no-color
Remove colors from output
-h, --help
Print help (see more with '--help')
As you can see, <FRAMEWORK_ID> and <EDITOR_ID> are two mandatory arguments that will specify which framework and editor you want to use inside your notebook.
You can find a list of available frameworks and editors by running the ovhai capabilities framework list and ovhai capabilities editor list commands:
ovhai capabilities framework list
ID NAME DESCRIPTION DOC_URL VERSIONS
alicebob Alice & Bob Felis Run fault-tolerant quantum algorithms with logical qubits. Study quantum error correction with superconducting cat qubits. alice_and_bob0.4.1
myqlm Atos myQLM Quantum Computing Framework by Atos myqlm1.7.3-py39-cpu-v22-4
autogluon-mxnet AutoGluon + MXNet AutoGluon + MXNet 0.1.0-ovh.beta.1
c12 C12 C12 Quantum Electronics environment https://github.com/c12qe/c12simulator-clients/tree/master/docs c122.0.1-py311-cpu,c120.0.6-py310-cpu-v22-4
fastai fastai Course fastai simplifies training fast and accurate neural nets using modern best practices https://docs.fast.ai/ 2021-08-04-ovh.beta.1,fastai2.7.10-py310-cuda11.6-v22-4
huggingface Hugging Face Transformers State-of-the-art Machine Learning. Includes Transformers + Datasets https://huggingface.co/docs/transformers/index hf4.34.0-py311-cuda11.8,hf4.26.0-py310-cuda11.6-v22-4
jax JAX JAX is Autograd and XLA, brought together for high-performance numerical computing, including large-scale machine learning research. https://jax.readthedocs.io/en/latest/ jax0.4.20
conda Miniconda Images with conda available for you to install your requirements yourself https://docs.conda.io/en/latest/miniconda.html conda-py311-cudaDevel11.8,conda-py311-cpu,conda-py310-cpu-v22-4,conda-py310-cudaDevel11.8-v22-4,conda-py310-cuda11.8-v22-4,conda-py310-cudaDevel11.7-v22-4,conda-py310-cuda11.7-v22-4,conda-py310-cudaDevel11.6-v22-4,conda-py310-cuda11.6-v22-4,conda-py310-cudaDevel11.5-v22-4,conda-py310-cuda11.5-v22-4,conda-py310-cudaDevel11.4-v22-4,conda-py310-cuda11.4-v22-4,conda-py310-cudaDevel11.3-v22-4,conda-py310-cuda11.3-v22-4,conda-py310-cudaDevel11.2-v22-4,conda-py310-cuda11.2-v22-4,conda-py310-cudaDevel11.1-v22-4,conda-py310-cuda11.1-v22-4,conda-py310-cudaDevel11.0-v22-4,conda-py310-cuda11.0-v22-4,conda-py310-cudaDevel10.2-v22-4,conda-py310-cuda10.2-v22-4,conda-py310-cudaDevel10.1-v22-4,conda-py310-cuda10.1-v22-4,conda-py310-cudaDevel10.0-v22-4,conda-py310-cuda10.0-v22-4,conda-py39-cpu-v22-4,conda-py39-cudaDevel11.8-v22-4,conda-py39-cuda11.8-v22-4,conda-py39-cudaDevel11.7-v22-4,conda-py39-cuda11.7-v22-4,conda-py39-cudaDevel11.6-v22-4,conda-py39-cuda11.6-v22-4,conda-py39-cudaDevel11.5-v22-4,conda-py39-cuda11.5-v22-4,conda-py39-cudaDevel11.4-v22-4,conda-py39-cuda11.4-v22-4,conda-py39-cudaDevel11.3-v22-4,conda-py39-cuda11.3-v22-4,conda-py39-cudaDevel11.2-v22-4,conda-py39-cuda11.2-v22-4,conda-py39-cudaDevel11.1-v22-4,conda-py39-cuda11.1-v22-4,conda-py39-cudaDevel11.0-v22-4,conda-py39-cuda11.0-v22-4,conda-py39-cudaDevel10.2-v22-4,conda-py39-cuda10.2-v22-4,conda-py39-cudaDevel10.1-v22-4,conda-py39-cuda10.1-v22-4,conda-py39-cudaDevel10.0-v22-4,conda-py39-cuda10.0-v22-4,conda-py38-cpu-v22-4,conda-py38-cudaDevel11.8-v22-4,conda-py38-cuda11.8-v22-4,conda-py38-cudaDevel11.7-v22-4,conda-py38-cuda11.7-v22-4,conda-py38-cudaDevel11.6-v22-4,conda-py38-cuda11.6-v22-4,conda-py38-cudaDevel11.5-v22-4,conda-py38-cuda11.5-v22-4,conda-py38-cudaDevel11.4-v22-4,conda-py38-cuda11.4-v22-4,conda-py38-cudaDevel11.3-v22-4,conda-py38-cuda11.3-v22-4,conda-py38-cudaDevel11.2-v22-4,conda-py38-cuda11.2-v22-4,conda-py38-cudaDevel11.1-v22-4,conda-py38-cuda11.1-v22-4,conda-py38-cudaDevel11.0-v22-4,conda-py38-cuda11.0-v22-4,conda-py38-cudaDevel10.2-v22-4,conda-py38-cuda10.2-v22-4,conda-py38-cudaDevel10.1-v22-4,conda-py38-cuda10.1-v22-4,conda-py38-cudaDevel10.0-v22-4,conda-py38-cuda10.0-v22-4
colatible Miniconda with Colab compatibility This image is based on our miniconda image. It has everything preinstalled to be compatible with your Google Colab notebooks. colatible-py310-cudaDevel11.8,colatible-py37-cudaDevel11.1-v22-4
mlr3 MLR3 An open-source collection of R packages providing a unified interface for machine learning in the R language. Successor of mlr. https://mlr3book.mlr-org.com/ mlr3-py310-cudaDevel11.8
mxnet MXNet A flexible and efficient library for deep learning https://mxnet.apache.org/versions/1.5.0/ mxnet1.9.1-py310-cuda11.2-v22-4
one-for-all One image to rule them all Tensorflow 2 + Pytorch + Fast.ai + MXNet + Transformers + And many more (incompatible with JupyterLab Real-Time collaboration editor) v98-ovh.beta.1
pasqal Pulser Library for pulse-level/analog control of neutral atom devices. Emulator with QuTiP. pulser0.17.0
pytorch PyTorch PyTorch is an optimized tensor library for deep learning using GPUs and CPUs https://pytorch.org/docs/stable/index.html pytorch2.4.0-py311-cuda11.8,pytorch2.1.0-py311-cuda11.8,pytorch2.0.0-py310-cuda11.8-v22-4,pytorch1.12.1-py310-cuda11.6-v22-4,pytorch1.11.0-py310-cuda11.3-v22-4,pytorch1.10.1-py39-cuda10.2-v22-4,pytorch1.9.0-py39-cuda10.2-v22-4
qiskit Qiskit Qiskit is an open-source SDK for working with quantum computers at the level of extended quantum circuits, operators, and primitives. https://docs.quantum.ibm.com/ qiskit1.0.2
perceval Quandela Perceval Simulation and Control Library for quantum photonic circuits perceval0.10.2-py311,perceval0.9.1-py310-cpu-v22-4,perceval0.8.1-py310-cpu-v22-4
sklearn Scikit Learn Simple and efficient tools for predictive data analysis https://scikit-learn.org/stable/user_guide.html sklearn1.3.0-py311-cpu,sklearn1.0.2-py39-cpu-v22-4
tensorflow Tensorflow An end-to-end open source machine learning platform https://www.tensorflow.org/api_docs tf2.15-py311-cudaDevel12.2,tf2.12-py311-cudaDevel11.8,tf2.11-py310-cuda11.2-v22-4,tf2.10-py310-cuda11.2-v22-4,tf2.9-py310-cuda11.2-v22-4,tf2.8-py39-cuda11.2-v22-4,tf2.7-py39-cuda11.2-v22-4,tf2.6-py39-cuda11.2-v22-4,tf2.5-py39-cuda11.2-v22-4,tf2.4-py38-cuda11.0-v22-4,tf2.3-py38-cuda10.1-v22-4,tf2.2-py38-cuda10.1-v22-4
ovhai capabilities editor list
ID NAME DOC_URL
jupyterlab JupyterLab https://jupyterlab.readthedocs.io/en/stable/
jupyterlabcollaborative JupyterLab Experimental https://jupyterlab.readthedocs.io/en/stable/user/rtc.html
vscode VSCode https://code.visualstudio.com/docs
From there, you can launch a first basic notebook.
Here is an example command for launching a notebook with PyTorch and JupyterLab:
ovhai notebook run pytorch jupyterlab
For VSCode, you can run:
ovhai notebook run pytorch vscode
By default, this notebook will be ran on a single V100S GPU, in a restricted access, with the latest version available of the framework you have selected.
As we saw with the execution of ovhai notebook run -h, there are many options available for customizing your notebook. One of them (--flavor <FLAVOR_ID>) allows you to specify the flavor you want to use (Type of CPUs/GPUs) by specifying the flavor ID. The flavor list can be accessed by running:
ovhai capabilities flavor list
ID TYPE DESCRIPTION CPU-PER-UNIT MEMORY-PER-UNIT GPU_INFORMATION.INFO DEFAULT MAX
a100-1-gpu gpu NVIDIA Ampere A100 13 160.0 GiB Ampere-A100 NVIDIA false 2
ai1-1-cpu cpu Intel CPU vCores 1 4.0 GiB ~ true 12
ai1-1-gpu gpu NVIDIA Tesla V100S 13 40.0 GiB Tesla-V100S NVIDIA true 4
ai1-le-1-gpu gpu NVIDIA Tesla V100S 13 40.0 GiB Tesla-V100S NVIDIA false 4
h100-1-gpu gpu NVIDIA H100 28 350.0 GiB H100 NVIDIA false 4
l40s-1-gpu gpu NVIDIA L40s 13 80.0 GiB L40s NVIDIA false 4
l4-1-gpu gpu NVIDIA L4 20 80.0 GiB L4 NVIDIA false 4
For example, to launch a notebook with the a100-1-gpu flavor, you would use the following command:
ovhai notebook run pytorch jupyterlab \
--flavor a100-1-gpu
Then, you can add to your command the --cpu <NUMBER> or --gpu <NUMBER> flag to specify how many CPUs/GPUs you want. The following example will run the same notebook but on 2 A100 GPUs:
ovhai notebook run pytorch jupyterlab \
--flavor a100-1-gpu \
--gpu 2
An example of what a slightly more specific notebook launch would look like:
ovhai notebook run pytorch jupyterlab \
--name my_first_notebook \
--framework-version pytorch2.4.0-py311-cuda11.8 \
--flavor ai1-1-cpu \
--cpu 6 \
--label my_label_key=my_label_value
This will launch a new AI Notebook with PyTorch 2.4.0, Jupyterlab, and 6 CPUs, in a restricted access.
Accessing a new notebook
Launching a new notebook using ovhai notebook run will display the following information in the terminal, which are the specifications of your notebook:
Id: 7600dda5-27f7-4937-9260-e24c987ae3c6
Created At: 11-09-24 10:12:29
Updated At: 11-09-24 10:12:29
User: user-fu9ThaHERbAT
Spec:
Name: my_first_notebook
Labels: {}
Resources:
Cpu: 6
Ephemeral Storage: 240.0 GiB
Flavor: ai1-1-cpu
Gpu: 0
Gpu Brand: ~
Gpu Memory: ~
Gpu Model: ~
Memory: 24.0 GiB
Private Network: 0 bps
Public Network: 3.0 Gbps
Volumes: []
Unsecure Http: false
Env:
Framework Id: pytorch
Framework Version: pytorch2.4.0-py311-cuda11.8
Editor Id: jupyterlab
Env Vars: ~
Ssh Public Keys: ~
Timeout Auto Restart: false
Status:
Last Transition Date: ~
Info:
Message: Notebook is starting
State: STARTING
Duration: 0
Url: https://7600dda5-27f7-4937-9260-e24c987ae3c6.notebook.gra.ai.cloud.ovh.net
Grpc Address: 7600dda5-27f7-4937-9260-e24c987ae3c6.nb-grpc.gra.ai.cloud.ovh.net:443
Info Url: https://ui.gra.ai.cloud.ovh.net/notebook/7600dda5-27f7-4937-9260-e24c987ae3c6
Monitoring Url: ~
Ssh Url: ~
Last Started At: ~
Last Stopped At: ~
The first line in the output shows the new AI Notebook ID: 7600dda5-27f7-4937-9260-e24c987ae3c6.
In the last lines, the Url field corresponds to your JupyterLab server. You will need to wait a few seconds for the notebook to start and reach the RUNNING status. Once it is running, the notebook URL will be accessible.
To follow the notebook state, you can get the notebook information again using its ID.
If you don't have the ID of the notebook you want to access, you can list all your notebooks using:
To get the notebook information again using its ID, run the following command:
ovhai notebook get 7600dda5-27f7-4937-9260-e24c987ae3c6
Id: 7600dda5-27f7-4937-9260-e24c987ae3c6
Created At: 11-09-24 10:12:29
Updated At: 11-09-24 10:12:29
User: user-fu9ThaHERbAT
Spec:
Name: my_first_notebook
Labels: {}
Resources:
Cpu: 6
Ephemeral Storage: 240.0 GiB
Flavor: ai1-1-cpu
Gpu: 0
Gpu Brand: ~
Gpu Memory: ~
Gpu Model: ~
Memory: 24.0 GiB
Private Network: 0 bps
Public Network: 3.0 Gbps
Volumes: []
Unsecure Http: false
Env:
Framework Id: pytorch
Framework Version: pytorch2.4.0-py311-cuda11.8
Editor Id: jupyterlab
Status:
Last Transition Date: ~
Info:
Message: Notebook is running
State: RUNNING
Duration: 592
Url: https://7600dda5-27f7-4937-9260-e24c987ae3c6.notebook.gra.ai.cloud.ovh.net
Grpc Address: 7600dda5-27f7-4937-9260-e24c987ae3c6.nb-grpc.gra.ai.cloud.ovh.net:443
Info Url: https://ui.gra.ai.cloud.ovh.net/notebook/7600dda5-27f7-4937-9260-e24c987ae3c6
Monitoring Url: https://monitoring.gra.ai.cloud.ovh.net/d/job?var-notebook=7600dda5-27f7-4937-9260-e24c987ae3c6&from=1726049489545
Ssh Url: ~
Last Started At: 11-09-24 10:12:39
Last Stopped At: ~
…
Now that the notebook is in the RUNNING state, a https address is defined in the Url field. This URL corresponds to your JupyterLab server. Pasting this URL in your browser displays the following screen:
You can now start writing code in your notebook. Since we used the PyTorch framework in our example, we will be able to use it without having to install anything ourselves.
As you explore your notebook, you will find an ai-training-examples folder already included in your /workspace. This repository is automatically cloned for each notebook and includes a variety of examples that you can use with OVHcloud AI products, such as AI Notebooks.
Stopping a notebook
Once you are done with your notebook, you can stop it with the command below:
ovhai notebook stop <NOTEBOOK_ID>
In our example above, the NOTEBOOKD_ID is 7600dda5-27f7-4937-9260-e24c987ae3c6
When you stop a notebook, it becomes unavailable for a certain period of time, during which your data is synchronized on the Object Storage. This allows you to retrieve your work later.
Once the synchronization is complete, the notebook will reach the STOPPED state. At this point, you can either start the notebook again or delete it.
Restarting a stopped notebook
Being able to restart a notebook is one of the main differences compared to using AI Training jobs. Restarting a notebook will restore your notebook code as it was when you stopped it. However, you will need to re-run your code to reload your variables because the program state is not saved.
To restart a notebook, run this command:
ovhai notebook start <NOTEBOOK_ID>
You are billed for RUNNING notebooks but not for STARTING, STOPPING and STOPPED notebooks.
However, to restore your code when you restart a STOPPED notebook, it needs to be stored in your Object Storage, which you are billed for. You can find more information about the AI Notebooks billing in the Billing and Lifecycle guide.
This is useful when you work on a notebook for some time.
Deleting a notebook
If you know you will not use a notebook anymore, you can delete it:
ovhai notebook delete <NOTEBOOK_ID>
Note that even if a notebook is still running, you can delete it. Of course, a deleted notebook cannot be restarted.
INFO
The notebook state stored in the Object Storage (including your notebook files) is not cleaned up automatically after notebook deletion.
You can find it and delete it in the notebooks_workspace container of your Object Storage, under the notebook ID directory.
Going further
Learn how to access your Object Storage data and Git repositories from your notebooks here.
Learn how to share your notebooks with other people here.
If you need training or technical assistance to implement our solutions, contact your sales representative or click on this link to get a quote and ask our Professional Services experts for a custom analysis of your project.
Feedback
Please send us your questions, feedback and suggestions to improve the service:
