Developing Language AI Models in Low-resource Scenarios | Hebrew University of Jerusalem | Clarifai
Автор: Clarifai
Загружено: 2021-10-20
Просмотров: 265
Join speaker Gabriel Stanovsky, Ph.D. from The Hebrew University of Jerusalem, and learn how large pre-trained language AI models are by now ubiquitous in natural language processing, yet it is still unclear how much data they require, and whether they can leverage multilingual signals. This talk presents work in developing a language model in an extremely low-resource scenario with a real-world application. See a model capable of filling in missing parts in ancient cuneiform tablets written thousands of years ago in now-extinct languages (Akkadian and Sumerian). Due to deterioration over time, these excavated tablets are often broken, faded, or cracked, making it hard for historians and archaeologists to read and interpret them. By leveraging large-scale language AI models pre-trained on modern texts we can restore missing parts in various domains and time periods, shown effective in both automatic as well as human evaluations.
Learn more at clarifai.com.
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
