The IDRIS course registration website will be closed for maintenance for an indefinite period.

For people belonging to CNRS, EPSCP or university institutions, registration requests should be sent by email to e-cours@idris.fr until the website is back up.

Other people should go to the CNRS Formation Entreprise website. Some training are not part of CNRS Formation Entreprise catalogue:

• Kokkos
• Workshop Jean Zay
• Minerva supported trainings

In this case the registration has to be sent by mail to e-cours@idris.fr

The upcoming sessions are as follows (last update: 2026/11/03):

Contents:

• Upcoming FineWeb dataset update on DSDIR
• Reminder: SCRATCH space best practices
• Reminder: Security Awareness Conference for Research on 20 October 2026
• GENCI webinar - Jean Zay Grands Challenges 4: DynaRepo on Friday 23 October 2026
• IDRIS training courses

• Upcoming FineWeb dataset update on DSDIR

The FineWeb dataset is currently available in version 1.0.0 on the DSDIR ($DSDIR/HuggingFace/fineweb). This version is now obsolete and will be replaced on 2 November by the latest available version, currently 1.4.0.

If you use this dataset and think this update may cause issues, please contact user support (assist@idris.fr) as soon as possible.

• Reminder: SCRATCH space best practices

The available free space on the SCRATCH storage space is regularly low. We remind you that this space is shared by all Jean Zay users, so it is essential to follow these best practices:

• Quickly delete files that are no longer useful
• Move and archive important files to WORK or STORE after the post-processing phase
• Do not apply any procedure to artificially extend the lifespan of files.

We need your cooperation to ensure this storage space can work properly, we are counting on you!

You can find the description of the different storage spaces available on Jean Zay in our documentation: Disk spaces.

• Reminder: Security Awareness Conference for Research on 20 October 2026

IDRIS is organising an exchange with the "Direction Générale de la Sécurité intérieure" (DGSI, General Directorate for Internal Security) to support its users and scientific research partners in protecting their work.

For more information and to register: Security awareness conference in the research field on 20 October 2026.

• GENCI webinar - Jean Zay Grands Challenges 4: DynaRepo on Friday 23 October 2026

GENCI is organising a webinar - Jean Zay Grands Challenges 4: "DynaRepo: capture the hidden movements of life's molecules" on Friday 23 October 2026 at 14:30.

For more information and to register: GENCI webinar - Jean Zay Grands Challenges 4: DynaRepo

• IDRIS training courses

Remember to register now for IDRIS training courses scheduled until the end of the year:

• Practical Introduction to Deep Learning, from 2 to 3 November 2026.
• HPC Debugging, on 19 November 2026.
• Introduction to OpenACC and OpenMP GPU, from 25 to 27 November 2026.
• Introduction to ScaLAPACK and MAGMA libraries, from 2 to 3 December 2026.
• Introduction to the PETSc library, from 3 to 4 December 2026.

For more information on the IDRIS training catalogue and registration procedures: http://www.idris.fr/en/docs/formations/catalogue-formations.

For your information, GENCI is organizing a webinar "Grands Challenges Jean Zay 4: DynaRepo capturing the hidden movements of life’s molecules" on Friday, October 23, 2026 – 2:30 p.m. (CEST).

How can we better understand how the molecules of life work? By studying not only their shape, but also their movements.

On October 23, Yasaman Karami will present the results of the DynaRepo project, carried out as part of the Grand Challenges on the Jean Zay supercomputer.
Proteins, RNA and DNA play an essential role in how our cells function. But these molecules are not rigid structures: they fold, deform, change configuration and constantly interact with their environment. To understand them, knowing their three-dimensional structure is not enough. We also need to be able to observe their movements.\

That is precisely the goal of DynaRepo.

Thanks to 300,000 GPU computing hours on Jean Zay, Yasaman Karami’s team has carried out large-scale molecular dynamics simulations. These simulations make it possible to track the movements of the atoms that make up molecules over time and explore their behaviour.

Register here:

https://www.eventbrite.fr/e/webinaire-jean-zay-4-dyna-repo-tickets-2003210239752

Security Awareness in Scientific Projects – DGSI Conference (General Directorate for Internal Security)​

In a context where threats are multiplying—cyberattacks, foreign interference, leaks of knowledge—the General Directorate for Internal Security (DGSI) is coming to meet you for an exclusive and essential conference.

IDRIS is organising this exchange to support its users and scientific research partners in protecting their work.

💡 The conference is designed to explain and illustrate the risks your activities may be exposed to:

• Your trips abroad: How to recognise the targeting of researchers for information gathering?
• Leakage of your knowledge: How to avoid concealed collaborations or prevent unauthorised transfers of your knowledge and research?
• What are the real threats: Human, cyber, physical, legal threats... and damage to reputation?

🔍 A unique opportunity to exchange with the DGSI, understand the risks and adopt the right reflexes to secure your work and that of your teams.

Limited places – Be proactive in securing your research

Dear All

NVIDIA is bringing together Paris's developer and AI community to a Build a Claw event in Paris on October 13 — an in-person event for developers, engineers, and students who want to get hands-on with autonomous agents.

Here's what attendees will experience:

Live agent demos — including one that monitors your inbox and delivers a daily briefing, 24/7, with zero intervention. Physical AI in action — see how robots powered by NVIDIA make the leap from research lab to real-world deployment. Hands-on time with agent frameworks like OpenClaw and Hermes, running on NVIDIA hardware

Whether you're deep in AI development or just starting to explore autonomous systems, this is the kind of event the attendees will leave with something concrete.

The event is free and open to everyone in the Paris tech community. Register here!

Note that registration is waitlist-based. Once confirmed, attendees will receive the venue details directly by email.

AI Factory France and MINERVA have joined forces to host a hackathon dedicated to AI for Healthcare.

Hosted at CINES in Montpellier, France November 16 – November 18 2026 , this edition will bring together developers, researchers, clinicians, data scientists, and HPC/AI support teams to collaborate on real-world healthcare AI challenges. Participants will explore how to develop trustworthy AI solutions that combine performance, explainability, ethical principles, and data sovereignty.

Applications are open until 25 September, EOB: https://minerva4ai.eu/event/ai2f-minerva-hackathon-healthcare-ai/

For your information, GENCI is organizing a webinar "Grands Challenges Jean Zay 4: AION-1, a multimodal foundation model for astronomy" on Monday, September 14, 2026, at 4:30 PM.

This webinar focuses on AION-1, a multimodal foundation model for astronomy that was awarded 200,000 H100 GPU hours on the Jean Zay supercomputer through the "Grands Challenges" program. It was developed by an international team (including CNRS/IDRIS, Paris-Saclay University, Flatiron Institute, and the Universities of Cambridge and Oxford). François Lanusse, a member of the team, will present the project during the webinar.

In response to the explosion of astronomical data, AION-1 is the first model to unify the analysis of heterogeneous observations—images, spectra, and scalar measurements—within a single architecture comprising over 3 billion parameters. Trained on 120 terabytes of data covering more than 200 million celestial objects from five major astronomical surveys (Legacy Survey, Hyper Suprime-Cam, SDSS, DESI, and Gaia), it enables astrophysicists to interpret data from diverse instruments and rapidly adapt to new scientific tasks , with only around a hundred annotated examples. To date, AION-1 is the largest AI model ever trained in astrophysics , and more broadly in physics.

Join us to discover this project presented directly by the research team.

To register: https://www.eventbrite.fr/e/webinaire-aion-1-modele-de-fondation-multimodal-pour-lastronomie-tickets-1996101526399?aff=oddtdtcreator

Kind regards,

The GENCI team

Nicolas Belot Chief Communication Officer and European relations

A new module pytorch-gpu/py3/2.13.0 has been available since August 24, 2026 for the H100 partition of Jean Zay. This module is built around PyTorch 2.13 and compiled with CUDA 13. It includes the latest versions of the various packages typically found in our modules.

To load it: module load arch/h100 pytorch-gpu/py3/2.13.0.

Contents:

• Technical Downtime around Tuesday, 15 September 2026
• New Training at IDRIS: Introduction to Kokkos
• IDRIS Training Courses

• Technical Downtime around Tuesday, 15 September

A technical downtime is scheduled for Tuesday, 15 September 2026 to carry out the annual maintenance, as well as part of the preparatory work related to the expansion of the centre's electrical and cooling capacities. Throughout this day, production will be completely suspended, and access to your data stored at IDRIS will be closed. This downtime may be extended to Monday, 14 September in the afternoon and/or Wednesday, 16 September in the morning. Further details on these dates will be provided later.

• New Training at IDRIS: Introduction to Kokkos

An introductory training course on the Kokkos library is now available at IDRIS: http://www.idris.fr/docs/formations/hpc/formation-kokkos/. The first edition will take place from 28 to 30 September 2026. This training is free for all.

• IDRIS Training Courses

Remember to register now for the IDRIS training courses scheduled for next autumn:

• MPI, from 22/09/2026 to 25/09/2026.
• Introduction to Kokkos, from 28/09/2026 to 30/09/2026.
• Jean Zay Workshop, from 01/10/2026 to 2/10/2026.
• OpenMP, from 14/10/2026 to 16/10/2026.
• Practical Introduction to Deep Learning, from 02/11/2026 to 03/11/2026.
• Deep Learning Architectures, from 04/11/2026 to 05/11/2026.
• Deep Learning Kick-off Workshop, on 06/11/2026.
• HPC Debugging, on 19/11/2026.
• Introduction to OpenACC and OpenMP GPU, from 25/11/2026 to 27/11/2026.
• Introduction to ScaLAPACK and MAGMA Libraries, from 02/12/2026 to 03/12/2026.
• Introduction to the PETSc Library, from 03/12/2026 to 04/12/2026.

For more information on the IDRIS training catalogue and registration procedures: http://www.idris.fr/en/docs/formations/catalogue-formations

As part of the National AI Research Programme (PNRIA)*, CNRS Computer Science has the mission to build a network of engineers around the Jean Zay ecosystem and the user support missions of IDRIS.

To this end, a team of 9 PNRIA engineers is based at IDRIS, within the AI User Support team to be close to the Jean Zay supercomputer and the teams responsible for its operation. In parallel, smaller teams of 3 people are also established at GRICAD (Grenoble), IRISA (Rennes), IRIT (Toulouse) and LORIA (Nancy).

The PNRIA Deep Learning expert engineers have the mission to support the French academic community in adopting AI technologies to advance their research projects.

This service is intended for researchers from all scientific fields (environment, astrophysics, materials physics, fluid mechanics, chemistry, biology, health, humanities and social sciences, computer science, etc.).

Whether you are discovering the possibilities of applying Deep Learning in your research field, or you are already experienced in training deep neural networks, whether you want to exploit large databases or optimise the performance of your training algorithm, you can benefit from enhanced support for your project by the AI engineers of the programme for a period of up to six months.

* The PNRIA is a programme led by CNRS Computer Science on behalf of the CNRS. It is funded under the France 2030 plan and operated by the French National Research Agency (ANR).