26

Mar

Teaser image to LRZ Bootcamp with NIVIDA on Accelerated Distributed Computing

Bootcamp

LRZ Bootcamp With NIVIDA on Accelerated Distributed Computing

Train Multiple Systems and Learn About New Tools

   26.03.2025

   9:00 am - 2:15 pm

   Leibniz Rechenzentrum, Boltzmannstr. 1, 85748 Garching bei München

The LRZ (Leibniz Supercomputing Centre) BDAI Team, together with NVIDIA, is organizing a Bootcamp on Accelerated Distributed Computing - powered by NVIDIA on March 26, 2025, at 9 AM.

The key topics addressed are:
1. Training Parallelization - Various methods to split and optimize training across multiple systems, including data/model parallelism and advanced techniques like ZeRO and Mixture-of-Experts

2. Communication Systems - How components communicate within and between nodes using tools like NCCL

3. Implementation Tools - Frameworks like NeMo Megatron and Profiler

You can register yourself for the bootcamp under the following link, where you can also find a detailed agenda.


Related

Link to Workshop NLPOR @COLM 2025

Workshop  •  10.10.2025  •  Montreal, Canada

Workshop NLPOR @COLM 2025

NLPOR aims to strengthen the emerging connection between Natural Language Processing and Public Opinion Research.


Link to Part 4: The Ethics of  AI Agents

Workshop  •  25.09.2025  •  Carl Friedrich von Siemens Stiftung, Südliches Schloßrondell 23, 80638 München

Part 4: The Ethics of AI Agents

"The Ethics of AI Agents" on September 25 shows how personal assistants are rapidly moving from concept to deployment.


Link to MCML Workshop: Reproducibility and Scientific Computing

Workshop  •  24.09.2025 - 30.09.2025  •  Richard-Wagner-Str. 10, Room D 105

MCML Workshop: Reproducibility and Scientific Computing

MCML workshop offers hands-on training in reproducibility, HPC, and packaging for early-career researchers.


Link to InterACT Workshop 2025

Workshop  •  01.09.2025 - 04.09.2025  •  LMU Munich, Department of Statistics, Ludwigstraße 33, 80539 München

InterACT Workshop 2025

Four-day INTERACT workshop on Interpretable Machine Learning (IML), co-organized by the MCML.