Projects
Here are some projects I have worked on.
Unary Feedback (UFO)
June 2025A reinforcement learning framework for improving multi-turn reasoning in LLMs using minimal feedback like 'Let's try again'. Enhances answer diversity and error correction without dense supervision.
View Details →
RAGEN
May 2025A research framework for training LLM agents in stochastic, multi-turn question answering environments. Equips agents with trajectory-level feedback, enabling adaptive behavior over time rather than reliance on isolated responses.
View Details →