Stanford CME 295: Transformers & Large Lang Models
Explains the evolution of NLP methods, the Transformer architecture, attention mechanisms, and how large language models are built, trained, tuned, and deployed.
Topics Covered Include:
- NLP basics (tokenization, embeddings)
- Transformer architecture & attention heads
- Variants like BERT, GPT, T5
- Training fundamentals
- Fine-tuning techniques (SFT, LoRA)
- Reinforcement-based tuning (RLHF)
- Retrieval-augmented generation & agent systems
- Evaluation and advanced LLM workflows