Optimizing LLMs through Quantization: A Hands-On Tutorial

In this talk, we will delve into the world of LLM Quantization, exploring how it can significantly enhance the efficiency and deployment of advanced deep learning models. Quantization reduces the computational and memory demands of LLMs by representing weights and activations with lower precision. This session will provide a practical, hands-on approach using Jupyter Notebooks to guide you from basic concepts to advanced quantization techniques. You will gain insights into the Post-Training Quantization (PTQ) and Quantization-Aware Training (QAT), understanding their applications, benefits, and challenges. By the end of the webinar, you'll be equipped with knowledge and tools to implement quantization strategies in your own projects, optimizing performance without sacrificing model accuracy.

Any Prerequisites for the Session -

-Basic understanding of ML concepts

-Familiarity with neural networks and deep learning fundamentals.

-Proficiency in Python programming.

-Good to have familiarity with PyTorch and HuggingFace Transformers (not mandatory)

-Basic experience in using Jupyter notebooks for code execution and visualization.