dingshizhe/molmo-ae-minit2i-sft-latents download history

dingshizhe/molmo-ae-minit2i-sft-latents is a text to image dataset on the Hugging Face Hub. In the last 30 days it was downloaded 17 times (5 in the last 7 days), and 142 times in total. It ranks #417,790 among datasets by monthly downloads.

minit2i SFT latents — VGT-AE with a frozen SigLIP2 (Molmo) encoder Pre-encoded SFT images as (32, 16, 16) float16 latents, paired with their captions. WebDataset format: each member a <key>.npy + <key>.txt pair. Same encoder, same space, and same geometry as the pretraining set dingsh

Open dingshizhe/molmo-ae-minit2i-sft-latents on Hugging Face

Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.