Stanford CS 312 grades on predicting experiment outcomes, assignments worth 0%
yen_chen_lin · x · 2026-10-31
Stanford's new Fall 2026 course CS 312: Deep Learning Alchemy (instructors Tatsunori Hashimoto, Suhas Kotha) argues deep learning mastery requires running many experiments, not just math:
- Assignments: run experiments to build intuition, worth 0% of the grade
- Quizzes: students must predict held-out experiment outcomes; seven difficult quizzes, no inflated curves
- To "isn't it unfair to predict arbitrary experiments?": the staff pick relevant metrics, treat whole-class failure as their failure — but note "the real world is unfair: your experiments depend on your prediction"
A 5-unit class; all recordings and materials will be made public.
More from Companies & People
- Cisco gives engineer 'unlimited tokens' to refactor 98% of legacy code as Proton tells staff their job is now helping LLMs — sanjaykalra · 2026-11-01
- Failed a dozen elite startup interviews until a take-home: the take-home hiring debate — brandon_galang · 2026-11-01
- OpenAI's Astra 5.5 runs massive code patches just to list Devcon side events — devanshmehta · 2026-11-01
- Meta reportedly built humanoid robot prototype Max with Fourier in 3 months, second version underway — kimmonismus · 2026-11-01
- Career advice for breaking into applied AI: stop litigating fairness, find another way in — brandon_galang · 2026-11-01
- Stanford Data-Centric AI Workshop 2026 releases full schedule with NVIDIA, SAP, free registration — jure · 2026-11-01