Quantization For Genai Models

Quantization For Genai Models
Published 10/2024
MP4 | Video: h264, 1920x1080 | Audio: AAC, 44.1 KHz
Language: English | Size: 731.43 MB | Duration: 2h 34m

Unlock the power of model optimization! Learn how to apply quantization and make your GenAI models efficient with Python

What you'll learn

Understand model optimization techniques: Pruning, Distillation, and Quantization

Learn the basics of data types like FP32, FP16, BFloat16, and INT8

Master downcasting from FP32 to BF16 and FP32 to INT8

Learn the difference between symmetric and asymmetric quantization

Implement quantization techniques in Python with real examples

Apply quantization to make models more efficient and deployment-ready

Gain practical skills to optimize models for edge devices and resource-constrained environments

Requirements

Basic Python knowledge is recommended, but no prior AI experience is required.

Description

If you are a developer, data scientist, or machine learning enthusiast who wants to optimize and deploy efficient AI models, this course is for you. Do you want to make your models faster and more resource-efficient while maintaining performance? Are you looking to learn how to apply quantization techniques for better model deployment? This course will teach you how to implement practical quantization techniques, making your models lean and deployable on edge devices.In this course, you will:Learn the core concepts of Quantization, Pruning, and Distillation.Understand different data types like FP32, FP16, BFloat16, and INT8.Explore how to convert FP32 to BF16 and INT8 for efficient model compression.Implement symmetric and asymmetric quantization in Python with real-world applications.Understand how to downcast model parameters from FP32 to INT8 for deployment.Gain hands-on experience with Python-based quantization, making your models suitable for mobile and IoT devices.Why learn quantization? Quantization allows you to reduce the size and computational load of models, making them suitable for resource-constrained devices like smartphones, IoT devices, and embedded systems. By mastering quantization, you can ensure your models are faster, more energy-efficient, and easier to deploy while maintaining accuracy.Throughout the course, you'll learn to implement quantization techniques and optimize your models for real-world applications. This course provides the perfect balance of theory and practical application for making machine learning models more efficient.By the end of the course, you'll have a deep understanding of quantization, and the ability to optimize and deploy efficient models on edge devices. Ready to optimize your AI models for efficiency and performance? Enroll now and start your journey!

Overview

Section 1: Introduction

Lecture 1 Introduction

Section 2: Gen AI model optimisation techniques

Lecture 2 Introduction to Gen AI models

Lecture 3 Model optimisation techniques - Introduction

Lecture 4 Introduction to Pruning

Lecture 5 Introduction to Knowledge Distillation

Lecture 6 Introduction to Quantization

Section 3: Data Types and Number Representation

Lecture 7 Data Types and Number Representation

Lecture 8 Integer Data types

Lecture 9 Integer Data typesin PyTorch

Lecture 10 8-Bit Fixed-Point Numbers

Lecture 11 Floating-Point Numbers

Lecture 12 Other Floating-Point formats

Lecture 13 Floating Point data types in PyTorch

Lecture 14 Other formats

Section 4: Quantization

Lecture 15 Downcasting FP32 to BF16

Lecture 16 Downcasting of tensors in Python

Lecture 17 Downcasting of an ML model in Python

Lecture 18 Downcasting FP32 to INT8

Lecture 19 Symmetrics quantization

Lecture 20 Asymmetrics quantization

Lecture 21 GPT Neo 125 quantization

Beginners in machine learning looking to learn practical model optimization techniques like quantization,AI professionals and students wanting to optimize models for deployment on resource-constrained devices