Thank you for opening my example experiment code. 

First, you can synchronize the necessary packages using the requirements file

I have prepared three experiments, and the path for saving the result files starts with 'main experiment~', so you should run this from the example code folder to avoid any path issues.

In experiment 4.1, it is important to note that to generate the left image, you must activate 'gradient2' in the following code:

for _ in range(iterations):
    grad = gradient(points) 
    # grad = gradient2(points) # Uncomment this line to use gradient2

The main code is optimized for hyperparameters. When executed, it will sequentially train models for 10,000 steps in the following order: ['CB', 'cic', 'diayn', 'dads', 'large_cb']. 

Additionally, result images will be automatically saved in the 'result' folder.

You can modify the parameter settings as desired within the 'configuration' folder.