vision-tranformer-coding

coding trying

This is on going. So there can be some mistakes.

At main.py model runs.

At Data_loading.py, only subset of CIFAR10 dataset will be loaded.

At input_processing.py, there are two functions

1. at image_to_patches, images divided into patches

   shape : B , N , embedding vector_size = patch_size^2 * channel_num

2. at preprocessing, cls_tocken and positional embedding will be added to image.

At multihead_attention.py, attention process will be runned.

At Encoder.py, encoder structure will be runned.(several encoding can be easily runned if you just add more encoders)