AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
4min snip
Unifying Vision and Language Models with Mohit Bansal - #636
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
U-DOP: Universal Document Processing for Text, Image, and Layout Generation
Text to image generation is now being used for infilling or in-pinting./nUniversal Document Processing (U-DOP) involves complex layouts with text, images, and tables./nThe UDAB model achieves state-of-the-art performance on the DUE benchmark for document understanding tasks./nThe model unifies text, image, vision, and layout by using joint pre-training objectives./nThe model makes layout information explicit and learns to predict the location of text on documents./nThe model can edit documents and generate new content while maintaining the same format or handwriting./nEditing and generating content improves the model's understanding of the input.
00:00
Transcript
Episode notes
Get the Snipd
podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode