OpenHands developer Graham Neubig and Daniel Fried announce a new CMU graduate course on building and evaluating LLM agents

This Fall at CMU we're teaching a new course on AI Agents!

The goal is that you learn how to create a scaffold, build evals, and train an agentic LLM using RL.

We'll try to balance theory and practice, and introduce modern frameworks and best practices.

Slides and videos will be posted online on the main site, so please follow along if you're interested:

https://www.cmu-agents.com/

We're also looking for compute sponsors, so if you can help provide infra support (sandboxes, GPUs, etc) please reach out!

The instructors are me (CMU prof and developer of @OpenHandsDev) and @dan_fried (CMU prof and formerly on Meta's agent team).

We have an all star team of TAs too! @yueqi_song @Jiarui_Liu_ @uilydna @apurvasgandhi @saujasv @sunweiwei12 @Aditya_Soni_8 @demisama_

Assignments (https://www.cmu-agents.com/#/assignments) will include: - Building an agentic harness - Creating agentic evaluations - Training with RL - A final research project

We're creating a new course on AI Agents at CMU this Fall!

We’re aiming to give students hands-on experience: from building agentic harnesses and evals to training with RL.

Check out our course site for the full schedule: http://cmu-agents.com

@gneubig This is super! Will the lectures be posted online? And looking forward to seeing cool demos from students!

@gneubig this is honestly the course I wish existed when I started building agents and bookmarking http://cmu-agents.com like it's a drop.

@gneubig amazing! will be following along, thank you for sharing

@gneubig cmu kids are gonna graduate knowing more about RL + evals than most working engineers rn

the gap between academia and industry is closing fast 🔥

@gneubig Sounds fun!

@gneubig students doing RL homework while most devs are still arguing about whether to use .strip()

@gneubig Thank you for this news! I wanted to ask you if you offer as well fully funded PhD positions in your lab. Thanks!

@gneubig what’s the smallest model you can rl on and it hill climbs? qwen math 7b?

@gneubig could you upload lectures on your yt too?

@gneubig Amazing! Thanks for sharing.

@gneubig Un curso de agentes que pone evals, memoria y sandboxing en el temario antes que la magia.

Dobby aprueba: primero la correa, luego el elfo corre por producción.

@gneubig Hi @gneubig , would this course be published online? Thanks!

@gneubig this is damn good ... wish i would have the accessibility too !!

@gneubig ngl the syllabus hitting harder than most bootcamps charging 3k

wonder how long before this is just... the default cs curriculum

@gneubig Wow much needed 🙌