Scaling Language-Image Learning in 100 Languages with PaLI
Have you ever wondered how teams across Google Research worked together to develop PaLI, the Pathways Language and Image Model, as a scalable multilingual approach to joint language and vision modeling? In this episode of #ResearchBytes, Research Scientists Radu Soricut and Neil Houlsby explain how PaLI works, share insights from their research combining language and vision models, and more! Watch this episode to learn more about PaLI and how it was created!
Resources:
Read the PaLI blog post → https://goo.gle/3sKmrgl
Read the paper → https://goo.gle/3WeCmkL
Learn how it works →https://goo.gle/3SRTtpF
Chapters:
0:00 - Intro
0:36 - What was the challenge with PaLI?
3:04 - What did we do?
4:14 - How did we do it?
9:40 - What’s next?
Watch more episodes of ResearchBytes → https://goo.gle/ResearchBytes
Subscribe to the Google Research Channel → https://goo.gle/GoogleResearch
#ResearchBytes