QLoRA is the first approach that allows the TRAINING of Large Language Models (LLMs) on a single GPU. It does this by using three contributions namely: 1. a novel NormalFloat data type, 2. Double Quantization, 3. Paged Optimizers.

In this video, we explain the idea behind all three approaches and explain how they are put together to arrive at QLoRA.

⌚ ⌚ ⌚ TIMESTAMPS ⌚ ⌚ ⌚
0:00 QLoRA
1:34 Quantization
3:10 Problem with Quantization
3:50 Blockwise Quantization
5:42 Normal Float
6:37 Double Quantization
7:25 Paged Optimizers
8:45 QLoRA Finetuning
9:30 LoRA vs QLoRA
10:28 Results

MY SOFTWARE TOOLS
✍ Notion https://affiliate.notion.so/aibitesyt
✍ Notion AI https://affiliate.notion.so/ys9rqzv2vdd8
OBS Studio for video editing https://obsproject.com
Manim for some animations https://www.manim.community

BOOKS I HAVE READ, REFER AND RECOMMEND
Deep Learning by Ian Goodfellow https://amzn.to/3Wnyixv
Pattern Recognition and Machine Learning by Christopher M. Bishop https://amzn.to/3ZVnQQA
Machine Learning: A Probabilistic Perspective by Kevin Murphy https://amzn.to/3kAqThb
Multiple View Geometry in Computer Vision by R Hartley and A Zisserman https://amzn.to/3XKVOWi

MY KEY LINKS
YouTube:    / @aibites  
Twitter:   / ai_bites​  
Patreon:   / ai_bites​  
Github: https://github.com/aibites​

WHO AM I?
I am a Machine Learning Researcher/Practitioner who has seen the grind of academia and startups equally. I started my career as a software engineer 15 years ago. Because of my love for Mathematics (coupled with a glimmer of luck), I graduated with a Master's in Computer Vision and Robotics in 2016 when the now happening AI revolution just started. Life has changed for the better ever since.

#machinelearning #deeplearning #aibites