DIP-transformers

Preparations

• Create exps directory at project's root
• install requirements using pip install -r requirments.txt

Code walkthrough

• The main entrypoint is main_transformer.py
  Chaning EXP variable to run the desired architecture - the options are transformer or org
  All the visualiztions (training curves, intermediate results and the best result during the training will be under exps/YYYY_MM_DD_HH_mm/

• Our architecture is in new_nets.py
  Where all transformer related blocks are in vit_model.py

Run the code

To run the code:
python main_transformer.py