A Vision Transformer (ViT) image model with a single channel.

This model is a variant of the vit_small_patch16_224.augreg_in21k_ft_in1k model from the timm library that contains a single channel instead of three.

The timm model was converted to a transformers model using a modified version of the convert_vit_timm_to_pytorch.py script which is included in this repository as convert.py.