Presenter: Christodoulos Stylianou, Research Engineer, CaSToRC, The Cyprus Institute     |     Mr Ivan Gentile, Data Scientist, IFAB (NCC Italy)

 

Overview: The tutorial aims to provide optimization techniques for Llama, a foundational Large Language Model (LLM) based on the Transformer Architecture, analogous to the GPT series. Noted for their human-like text generation capabilities, these models encounter challenges regarding efficiency and scalability due to their complexity and computational demands. The session intends to augment the operational efficiency of these models through PyTorch-native optimization strategies, including model compilation, GPU quantization, speculative decoding, and tensor parallelism. Participants will have the chance to evaluate the proposed optimizations in real-time on one of Europe’s largest supercomputers. These methods seek to significantly reduce inference times and optimize resource usage, thus expanding the advanced models' applicability across various computational frameworks and research initiatives.

 

Prerequisites: As an intermediate-level tutorial, we expect basic knowledge of Deep Learning, and programming in Python. Additionally, some experience in using HPC systems is helpful (Linux shell, Slurm) but not mandatory. Participants are expected to provide a laptop with which they can access the HPC system. Access will be facilitated via individual accounts using the Jupyter platform.

 

 

 

Presenter: Charalambos Chrysostomou, Associate Research Scientist, The Cyprus Institute 

 

Overview:  This tutorial aims to provide a concise yet comprehensive overview of scaling machine learning models in PyTorch. It covers the essentials of distributed training, focusing on practical examples to demonstrate the scalability and efficiency of PyTorch. The session is structured to transition smoothly from basic to intermediate topics.

 

Prerequisites: Python Skills: Basic to intermediate Python, including familiarity with its standard library; Machine Learning Basics: Understanding of core ML concepts and algorithms; PyTorch Fundamentals: Experience with model definition, data loading, and training in PyTorch; Deep Learning Knowledge: Familiarity with neural networks, including different layers and activation functions; Parallel Computing: Basic knowledge of parallel computing, multiprocessing, and GPUs; Hardware: Access to a machine with a CUDA-enabled GPU. 

Time	Description	Speaker
10:00 - 11:00	Optimizing Llama: Enhancing Efficiency and Scalability of Large Language Models with PyTorch	Mr Christodoulos Stylianou & Mr Ivan Gentile
11:00 - 11:15	Coffee break
11:15 - 12:15	Efficient Scaling of Machine Learning Models: Distributed Training	Dr Charalambos Chrysostomou
12:15 - 13:00	Coffee break