The HPC for Learning working group (GT) of the CNRS GDR C4P (“paradigms, parallelism, performance, precision”) is launching a new in-person seminar series in February 2026, held about once a month in the SCAI building at Sorbonne Université in Paris, and supported by PEPR SHARP.

As part of the GDR C4P and hosted within the Paris ELLIS Unit, this series aims to build a collaborative community at the intersection of AI and high-performance computing, focusing on how HPC tools, architectures and accelerators are deployed to scale modern learning algorithms, improve memory and communication efficiency, and enable robust, reliable inference on current hardware.

Seminars are open to the broader AI and HPC community; attendance is free but requires registration via the website: https://gdr-hpc4learning.github.io/

The mission of the HPC for Learning GT (within the CNRS GDR C4P) is to build a collaborative community at the crossroads of AI and HPC, co-designing algorithms and architectures for scalable training, memory/communication-efficient learning, accelerator-optimized algorithms, and robust, reproducible inference on modern hardware.