Escaping the Mode: Multi-Answer Reinforcement Learning in LMs

Isha Puri

Isha is a PhD student at MIT EECS and CSAIL where she is co-advised by Professor Marzyeh Ghassemi and Professor Yoon Kim. Her research focuses on building language models that can learn to reason like humans, as well as deployable, robust, and ethical AI. She graduated with her B.A. in Applied Mathematics and Computer Science from Harvard University in 2023, where she was an HBS Technology Innovation Fellow. She currently holds the MIT Great Educators Fellowship and the National Science Foundation’s Graduate Research Fellowship.