Back to all members...
Suhaas Bhat
Associate Member (PhD), started 2024
Suhaas Bhat is pursuing a DPhil in Theoretical Physics, supervised by Ard Louis and Yarin Gal. He is interested in the interface between physics, biology, and machine learning, and applying insights from these fields to medicine. Previously, he worked on machine learning applied to peptide drug design with Pranam Chatterjee, at the Church Lab at Harvard and the Chatterjee Lab at Duke. He has also worked on small molecule binding prediction at DE Shaw Research, and machine learning for automated DRAM verification at Normal Computing.
Suhaas holds an AB in Social Studies and Physics from Harvard, and wrote his undergraduate thesis on the ethics and viability of automating psychotherapy. He is funded by a Rhodes scholarship.
Publications while at OATML • News items mentioning Suhaas Bhat • Reproducibility and Code • Blog Posts
Publications while at OATML:
Building Reliable Long-Form Generation via Hallucination Rejection Sampling
Large language models (LLMs) have achieved remarkable progress in open-ended text generation, yet they remain prone to hallucinating incorrect or unsupported content, which undermines their reliability. This issue is exacerbated in long-form generation due to hallucination snowballing, a phenomenon where early errors propagate and compound into subsequent outputs. To address this challenge, we propose a novel inference-time hallucination mitigation framework, named Segment-wise HAllucination Rejection Sampling (SHARS), which uses an arbitrary hallucination detector to identify and reject hallucinated segments during generation and resample until faithful content is produced. By retaining only confident information and building subsequent generations upon it, the framework mitigates hallucination accumulation and enhances factual consistency. To instantiate this framework, we adopt semantic uncertainty as the detector and introduce several vital modifications to address its limitati... [full abstract]
Lin Li, Georgia Channing, Suhaas Bhat, Gabriel Jones, Yarin Gal
arxiv
[paper]
Blog Posts
OATML at ICML 2026
OATML group members and collaborators are proud to present 5 papers at ICML 2026. …
Full post...Yarin Gal, Sergio Calvo Ordoñez, Sören Mindermann, Panagiotis Tigas, Lin Li, Suhaas Bhat, Gabriel Jones, 08 Jul 2026