Stanford CME 295 syllabus: Transformers, MoEs, scaling laws to KV caching

afshinea · x · 2026-09-29

The CME 295 syllabus spans three areas: Transformer foundations (tokenization, embeddings, attention, architecture), LLMs (pretraining, MoEs, decoding, scaling laws), and systems (distributed training, KV caching, speculative decoding), covering principles through engineering practice.

Related event: Stanford's CME 295 Returns: New LLM Course Hits YouTube(3 posts)→

Original post →

More from Companies & People

Companies & People channel →