Stanford CS224R Deep Reinforcement Learning | Spring 2025 | Lecture 9: RL for LLMs

Name: Stanford CS224R Deep Reinforcement Learning | Spring 2025 | Lecture 9: RL for LLMs
Uploaded: 2025-12-08T00:00:00.000Z
Duration: 1 h 2 min 51 s
Description: Pre-training teaches next-token prediction and a lot of world knowledge, but assistants come from a later stack: instruction fine-tuning for format, then preference optimization via RLHF or DPO so the model actually f...

Stanford Online · 62:51

Pre-training teaches next-token prediction and a lot of world knowledge, but assistants come from a later stack: instruction fine-tuning for format, then preference optimization via RLHF or DPO so the model actually f...

Read the full summary on tuber