Edit model card

Model description

SiT is a self-supervised learning model that combines masked image modeling and contrastive learning. The model is trained on ImageNet-1K.

Model Sources

Model Card Authors

Sara Atito, Muhammad Awais, Josef Kittler

How to use

from modeling_sit import ViTSiTForPreTraining
# reload
model = ViTSiTForPreTraining.from_pretrained("erow/SiT")

BibTeX entry and citation info

@inproceedings{atito2023sit,
  title={SiT is all you need},
  author={Atito, Sara and Awais, Muhammed and Nandam, Srinivasa and Kittler, Josef},
  booktitle={2023 IEEE International Conference on Image Processing (ICIP)},
  pages={2125--2129},
  year={2023},
  organization={IEEE}
}
Downloads last month
5
Safetensors
Model size
41.8M params
Tensor type
F32
·
Inference API
Inference API (serverless) has been turned off for this model.