LibrePRX512-dc-ae-sft

PRX, Photoroom's 1.3B text-to-image transformer trained from scratch, with a T5Gemma 2B text encoder and the 32x DC-AE autoencoder. Native resolution 512; supervised fine-tune, ready to use. Mirrored for LibreYOLO.

from libreyolo import LibreGen

model = LibreGen("prx-512-dc-ae-sft")      # downloads this repo
model("a red fox sleeping in fresh snow").save("fox.png")
libreyolo generate prompt="a red fox" model=prx-512-dc-ae-sft

Install libreyolo[gen].

This is a mirror, not a new model

The weights are Photoroom's. Nothing was retrained, converted or re-encoded; every file is byte-identical to upstream and keeps upstream's diffusers pipeline layout, so diffusers loads this folder as it is.

File SHA-256
text_encoder/model-00001-of-00003.safetensors e7a37fef2b11c87a9a2a43c26fcd7199c1d1b203ef11e10863a56611c8fdb57f
text_encoder/model-00002-of-00003.safetensors 49da6c85ed4ff10f7f1afaa5a37a0b292fdb58aa855916ef584d697155a9bf58
text_encoder/model-00003-of-00003.safetensors 1a6361ba65805092625aa4e2b3b8d42b810ab836682ddc8277088c9a251b4084
transformer/diffusion_pytorch_model.safetensors 4d5c51f2f4abb7793b05d62434cb55652b0e3aa9a0b6f2465ca6bb16071c33de
vae/diffusion_pytorch_model.safetensors dfd991d1b54ffabf22745c5885589d8f2a7bc59930d95d92bd741c4fc64454bb

License

The PRX weights are Apache-2.0, as published by Photoroom. The bundled T5Gemma 2B text encoder is provided under and subject to the Gemma Terms of Use, and you may not use it for the restricted uses in the Gemma Prohibited Use Policy. The DC-AE autoencoder comes from mit-han-lab/dc-ae-f32c32-sana-1.1-diffusers (MIT); upstream's NOTICE names the FLUX.1 VAE for every checkpoint. See LICENSE and NOTICE.

Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for LibreYOLO/LibrePRX512-dc-ae-sft

Finetuned
(1)
this model

Collection including LibreYOLO/LibrePRX512-dc-ae-sft