Iris-3B
Fully open 3B-parameter pixel-space text-to-image model and vision learner
Community-submitted · content unverified. This organization has not verified ownership or the information in this profile.
About
Iris-3B is Speridlabs' fully open 3B-parameter text-to-image model that generates directly in pixel space, bypassing the lossy compression of a VAE latent. The lab presents it as a general vision learner that can stand in for vision backbones such as DINOv2 on dense tasks like depth and restoration, and releases the paper, code, weights and a Hugging Face Space.
- Website
- speridlabs.com/research/iris
- Added via
- web
- Ownership
- unclaimed
Where to go
Product details
- Product kind
- Open Source
- Open source
- Yes
Social profiles
- Code (GitHub)GitHub