Andrej Karpathy dropped a must-watch 3.5-hour video on YouTube that dives into all sorts of stuff about building large language models. He goes through the whole training process, shares some interesting ways to think about their “psychology,” and gives tips on how to make the most out of them for real-life use.

Andrej who? Andrej Karpathy is a computer scientist renowned for his work in deep learning and computer vision, as well as for being a co-founder of OpenAI and the former Senior Director of AI at Tesla. He founded Eureka Labs, which aims to build an AI-native school.

Yes, it’s a 3.5-hour video and yes, it’s totally going to be worth your time! Andrej is an amazing teacher who goes out of his way to make things understandable for a non-technical audience.

Chapters

00:00:00 introduction
00:01:00 pretraining data (internet)
00:07:47 tokenization
00:14:27 neural network I/O
00:20:11 neural network internals
00:26:01 inference
00:31:09 GPT-2: training and inference
00:42:52 Llama 3.1 base model inference
00:59:23 pretraining to post-training
01:01:06 post-training data (conversations)
01:20:32 hallucinations, tool use, knowledge/working memory
01:41:46 knowledge of self
01:46:56 models need tokens to think
02:01:11 tokenization revisited: models struggle with spelling
02:04:53 jagged intelligence
02:07:28 supervised finetuning to reinforcement learning
02:14:42 reinforcement learning
02:27:47 DeepSeek-R1
02:42:07 AlphaGo
02:48:26 reinforcement learning from human feedback (RLHF)
03:09:39 preview of things to come
03:15:15 keeping track of LLMs
03:18:34 where to find LLMs
03:21:46 grand summary