Zhuang Liu@liuzhuang1234
Can a small academic team build a strong text-to-image model using only public datasets?
Introducing i1: a simple, fully open recipe for strong text-to-image models
4:28 PM · Jun 22, 2026 · 19.5K Views
Princeton's Zhuang Liu releases i1, an open text-to-image model trained entirely on public datasets
The project proves viable training within academic resource constraints.
72624611321K
Sentiment
Sentiment building, check back later.
Cluster Engagement
Digg Deeper
No Digg Deeper questions have been answered for this story yet.
Posts from X
Most Activity
Most Activity
VIEWS1.2KBOOKMARKS4RETWEETS1
Ravid Shwartz Ziv@ziv_ravid
Amazing work! It's also a great opportunity to remind that you can listen to Zhuang on the Information Bottleneck! https://www.the-information-bottleneck.com/what-actually-matters-in-ai-with-zhuang-liu-princeton/
Zhuang Liu@liuzhuang1234
Can a small academic team build a strong text-to-image model using only public datasets?
Introducing i1: a simple, fully open recipe for strong text-to-image models
10hViews 1.2KLikes 8Bookmarks 4
LIKES9
Lucas Beyer (bl16)@giffmana
@liuzhuang1234 @XingyuFu2 Nice work and write-up, congrats!
Zhuang Liu@liuzhuang1234
Can a small academic team build a strong text-to-image model using only public datasets?
Introducing i1: a simple, fully open recipe for strong text-to-image models
8hViews 485Likes 9Bookmarks 0