Covers advanced reinforcement learning for agents: credit assignment via baselines and reward-to-go, turn-level rewards, and value functions learned with critics, using a bug-fixing coding agent as the running example.
Save to open this file
Carnegie Mellon 11-768: AI Agents
All
01-Course Overview - What Is an Agent
- 01-Course Overview - What Is an Agent
- 02-Agent Capabilities 1 - Tool Use
- 03-Agent Capabilities 2 - Context Management for Long-Context Agents
- 04-Agent Capabilities 3 - Skills and Memory
- 05-Agent Capabilities 4 - Planning, Task Decomposition, and Multi-Agent Coordination
- 06-Domains 1 - Coding Agents
- 07-Domains 2 - Computer Use Agents
- 08-Training 1 - Supervised Fine-Tuning (SFT)
- 09-Training 2 - Reinforcement Learning Basics
- 10-Domains 3 - Deep Research Agents
- 11-Training 3 - Advanced RL Algorithms
- 12-Training 4 - RL Systems
- 13-Safety 1 - Agent Safety
- 14-Frameworks 1 - OpenHands
- 15-Frameworks 2 - Orchestration and Monitoring
- 16-Agents and the Future of Work
- 17-Interaction 1 - Multi-Agent Interaction