Sources

This course is a mechanism spine, not a syllabus of lectures. The teachers below match the pages, with full names.

• Vaswani, Ashish, et al. Attention Is All You Need (2017). Transformer block, scaled dot-product attention, causal mask.
• OpenAI, InstructGPT (2022). Base model, then example chats, then preference picks.
• Grant Sanderson (3Blue1Brown), neural net videos — pictures of mix-clip-mix.
• Andrej Karpathy, tokenizer and GPT lectures — tokens and the next-piece loop.
• Tom Yeh, by-hand attention — the three-token napkin.
• Stephen Wolfram, essay on ChatGPT — philosophical aftertaste.

You should not need those to retell the core.