Using Language Models for Instruction-Based Information Retrieval

Explore the innovative approach of using language models for instruction-based information retrieval in this 40-minute talk by Orion Weller from the Center for Language & Speech Processing at Johns Hopkins University. Discover how instructions in retrieval can enable complex information needs with zero-shot adaptation, similar to how prompts have revolutionized user tasks and optimization. Learn about the development of a promptable retriever that processes information like a language model. Examine the creation of an evaluation benchmark for testing retrieval models and gain insights into Promptriever, a new retriever showing significant improvements in instruction following and pioneering promptable retrieval. Delve into the intersection of natural language processing, information retrieval, and machine learning with Orion Weller, a fourth-year PhD student advised by Benjamin Van Durme and Dawn Lawrie.