Deep Dive into LLMs like ChatGPT

Andrej Karpathy explains how LLMs are pre-trained, fine-tuned and subjected to reinforcement learning.

June 14, 2026