#reinforcement-learning
-
Deep Dive into LLMs like ChatGPT
Karpathy's 3.5-hour general-audience deep dive (9.7M views, Feb 2025) covers the full ChatGPT pipeline: pretraining (internet → tokens → next-token prediction), supervised fine-tuning (conversations → an assistant persona), and reinforcement learning (practice problems → emergent reasoning). Throughout, he maps the stages onto how children learn from textbooks: exposition, worked examples, practice problems.
-
AlphaGo - The Movie | Full award-winning documentary
DeepMind's award-winning 90-minute documentary (on YouTube since 2020, 38M views) on AlphaGo: from beating European champion Fan Hui 5-0 in secret (Oct 2015) to the Google DeepMind Challenge Match against Lee Sedol in Seoul (March 2016), with 80-100M people watching. AlphaGo won 4-1, about a decade before experts expected Go to fall.