Deep Dive into LLMs like ChatGPT (Andrej Karpathy)
Deep Dive into LLMs like ChatGPT (Andrej Karpathy)
A clear explanation of how ChatGPT is made; about 3.5 hours long.
Andrej Karpathy explains the whole process of how a model like ChatGPT is made: data, tokens, training, fine-tuning and why AI 'makes things up'.
Overview
A video for non-specialists that explains the entire process of creating a large language model: collecting data, tokenization, pretraining, fine-tuning into an assistant, reinforcement learning, and why models sometimes make up information.
After watching, you will understand why AI answers need to be double-checked and how to ask questions to reduce errors.
Highlights
For non-specialists
No math needed.
Explains the 'making things up' phenomenon
Understand it to guard against it.
A leading expert
Andrej Karpathy.
What it's for
Setting the right expectations for AI
Know what to and not to hand over.
Raising awareness in the team
Watch together, then discuss.
How to use it
- 1
Open the course page
Click Open the official page to go to the YouTube page.
- 2
Register an account
Create a free account to save your learning progress.
- 3
Learn lesson by lesson
Duration is about 3.5 hours; split it into 30–60 minutes a day.
- 4
Apply it to real work right away
After each lesson, try the technique you just learned on a specific task of yours.
FAQ
Is this course free?
Yes, the video is free on YouTube.
How long does it take?
About 3.5 hours.
Is there a Vietnamese version?
English, with automatic subtitles (translatable into Vietnamese).
Useful links
Want to apply this in your business?
2dm looks at your workflows, picks the right tools and sets them up with your team. The first call is free.
Same topic
DeepLearning.AI Short Courses
4.5
A library of short courses on generative AI, agents and prompting from Andrew Ng's DeepLearning.AI, free to try, each course only a few hours.
Anthropic Academy
4.5
Anthropic's online academy with free AI Fluency and Claude courses, with completion certificates, including a course made for small businesses.
Elements of AI
4.0
An introductory AI course for non-specialists from the University of Helsinki and MinnaLearn, free to take, no math or programming required.