Assignment 5

Post-training

How a pretrained model learns from demonstrations, feedback, and rewards.

• Supervised fine-tuning
• Reward signals and evaluation
• Policy gradients for reasoning

Stanford assignment materials

← All tracks