Dipendra Misra*

Latest

• Principled Fine-tuning of LLMs from User-Edits: A Medley of Preference, Supervision, and Reward
• Provable Interactive Learning with Hindsight Instruction Feedback