The Stanford LLM Map — All 18 Lectures of CS336 (Language Modeling from Scratch, Spring 2026) Triaged for Working Developers, With the 4-Week Laptop Plan

The companion to the full video. Stanford put its entire graduate LLM course — CS336: Language Modeling from Scratch (Percy Liang & Tatsunori Hashimoto, Spring 2026) — on YouTube for free: 18 lectures, 24.2 hours. Stanford's own site calls it a 5-unit class with 'an order of magnitude' more code than other classes — which is why most people who save the playlist never get past lecture two. This map fixes that. INSIDE: (1) THE VERDICT BOARD — all 18 lectures with a WATCH / IF / SKIP stamp, the minutes for each, why it earns its verdict, and a direct link to every lecture. (2) THE 8-LECTURE SPINE — about 11 of the 24 hours that carry ~90% of the value for a working developer (tokenization, architectures, scaling laws, inference, evaluation, data, post-training, and RLVR — how reasoning models are trained), in watch order, with what each unlocks. (3) THE HIDDEN HALF — Stanford's public assignment repos (assignment1-basics has 2.5k stars), labeled laptop-vs-rented-GPU, with the exact clone-and-test commands to start assignment 1 tonight. (4) THE 4-WEEK PLAN — about an hour a day, ending with a small language model you trained yourself on TinyStories, plus the honest prerequisites checklist. Lecture list and links pulled from the official Stanford Online playlist; star counts and course quotes captured live July 28, 2026. Independent PDF from Hyperautomation Labs — not affiliated with Stanford.

By submitting, you'll be subscribed to the Hyperautomation AI Report newsletter — one practical AI email per week. Unsubscribe anytime.

Free. No spam. Unsubscribe anytime.