Abstracts: October 9, 2023

Join the Microsoft research community on the cutting edge of innovation with Abstracts. In this episode, Dr. Sheng Zhang, Senior Researcher at Microsoft Research, discusses the paper and project, "UniversalNER: Targeted Distillation from Large Language Models for Open Named Entity Recognition" with host Dr. Gretchen Huizinga. The paper introduces a novel approach to distilling large language models into smaller, more efficient models for natural language processing tasks such as named entity recognition. The UniversalNER models developed using their approach produced state-of-the-art performance in this field. This model distillation technique can lead to more accessible and transparent NLP capabilities, especially in specialized domains like biomedicine. Check out the link to view the paper, explore the UniversalNER project website and demo, browse and download code on GitHub, and access datasets and models on Hugging Face.