AuroraIMG2-50M-Preview Releases!

It's a tiny Text to Image model that's Denoiser is only 50M parameters, adding the custom text encoder, and the pretrained flux VAE, its total end to end inference parameters are around 90M It runs in: 27 Seconds @ 50 steps 11 Seconds @ 20 steps On a laptop CPU from 2016 (Intel i5 7200U) Samples and a quick start in the model card it is licensed under Apache 2.0 Questions, feedback, or criticism is welcome!