Castiel2 Multi-GPU AI Train the Trainer workshop

Prerequisites

You are familiar with AI frameworks.

Schedule

The course will run across 5 days with a morning and afternoon session each day. Times in CET:

Day	Session	Topic
1	Morning	Introduction to GPU architectures and Access to HPC infrastructure used
1	Afternoon	Introduction to deep learning
2	Morning	Pytorch Distributed Data Parallel
2	Afternoon	Model parallelism with Pytorch
3	Morning	Pytorch Lightning
3	Afternoon	LLM, Finetuning & HuggingFace Accelerate, Deepspeed
4	Morning	Computer Vision
4	Afternoon	MLops
5	Morning	Ray + Retrieval Augmented Generation (RAG)
5	Afternoon	Hyperparameter tuning followed by Q&A

The lesson materials

Reference

Scientists, engineers, and PhD students who need to scale AI models efficiently or optimize distributed training workflows.

Instructors

Workshop helpers

…