Inspiration Exchange #35
Query-Dependent Prompt Evaluation and Optimization with Offline Inverse RL (ICLR 2024)
Authors: Hao Sun, Alihan Hüyük, Mihaela van der Schaar(Paper Link)
This work introduces Prompt-OIRL, which harnesses offline inverse reinforcement learning to draw insights from offline prompting demonstration data to perform zero-shot prompt optimization.
Retrieval-Augmented Thought Process as Sequential Decision Making
Authors: Thomas Pouplin, Hao Sun, Samuel Holt, Mihaela van der Schaar (Paper Link)
This work introduces Retrieval-Augmented Thought Process (RATP) to improve retrieval-augmented generation. Given access to external knowledge, RATP formulates the thought generation of LLMs as a multiple-step decision process, using Monte-Carlo Tree Search to optimize this thought process.
We are excited to be presenting 10 papers at ICLR 2024, some of which will also be discussed in upcoming IE sessions. For now, to see the complete list, please visit https://www.vanderschaar-lab.com/van-der-schaar-lab-at-iclr-2024-10-papers-accepted/.
Additionally, we are excited to be joined by:
Dr. Vivek Natarajan (Research Lead, Google Health AI)
Dr. Chris Anagnostopoulos (Director of Life Sciences Innovation, QuantumBlack, AI by McKinsey)


Location and date/time
This event will take place on Zoom on March 11 at 16:00 GMT.



