Saujas Vaduguru

PhD student @LTIatCMU | Previous: Student @iiit_hyderabad, Intern @ChandarLab @Mila_Quebec | he/him/his

[Saujas Vaduguru @saujasv]

People adapt their language to communicate more efficiently over time. How can we make models do this? In our recent work, we trained models in self-play, and found that using the right incentives can make models adapt to communicate efficiently even without human demonstrations.

[Daniel Fried @dan_fried]

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: https://t.co/DW9J20fpZD

[Saujas Vaduguru @saujasv]

I'm also presenting this work at SCiL on the 4th, and excited to chat with the community!

[Saujas Vaduguru @saujasv]

I'll be at #ACL2026 presenting our work on convention formation! I'm thinking about collaborating and communicating with agents these days, and am excited to chat with others working on these problems!

Find us at Poster Session E on Monday, July 6 to check out the work!

[Saujas Vaduguru @saujasv]

People adapt their language to communicate more efficiently over time. How can we make models do this? In our recent work, we trained models in self-play, and found that using the right incentives can make models adapt to communicate efficiently even without human demonstrations.

[Nicholas Tomlin @NickATomlin]

I'm recruiting my first group of PhD students at TTIC! If you're interested, please apply! If you know people who might be interested, please spread the word!

Application deadline is Dec 9, 2025, and there is no application fee: https://t.co/kTqSsvV4EJ

[Saujas Vaduguru @saujasv]

@HuaYilun led a cool paper on a different approach that makes clever use of conversations between people to elicit this behavior.

https://t.co/BPtXuLYE2e

[Yilun Hua @HuaYilun]

Humans naturally communicate with increasing efficiency in interactions, by adapting language and forming conventions. Yet LLMs do not. We showed this in our COLM 2024 paper📜("Talk Less, Interact Better") https://t.co/hPDQJBBMsH
Now, we have an approach to fix this 🚀

[Saujas Vaduguru @saujasv]

People adapt their language to communicate more efficiently over time. How can we make models do this? In our recent work, we trained models in self-play, and found that using the right incentives can make models adapt to communicate efficiently even without human demonstrations.

[Saujas Vaduguru @saujasv]

More details in the paper at https://t.co/a7PL8Hfyes

This was a great opportunity to collaborate with @HuaYilun, @yoavartzi, and @dan_fried!

[evanthebouncy @evanthebouncy]

Going to present this at emnlp. Come say hi to me

Wednesday Nov5
13:00-14:00
Poster 4003 at Hall C

[Saujas Vaduguru @saujasv]

@evanthebouncy is presenting this at EMNLP next week! Find him in Hall C on Wednesday (Nov 5), 13:00–14:00 to chat about this!

[Saujas Vaduguru @saujasv]

When we instruct an agent to design something, its first output may not be precisely what we want. Humans collaborating refine their creations iteratively. Can we instruct an agent to refine its output? Is language the best medium for these instructions? We explore this in mrCAD.

[Saujas Vaduguru @saujasv]

Just look at these multimodal refinement instructions! How would we ground them into reasonable executions? Joint work with @wp_mccarthy @evanthebouncy @judyefan @dan_fried @KarlDD @JustinMatejka

Paper: https://t.co/mG6mnXUmad
Code: https://t.co/goQfARtU8W