Extracting Neural Network from a GitHub repository written in C++ version of TensorFlow -- 2

The project is about getting a Neural Network from a GitHub repository of a project at https://github.com/LeelaChessZero and written in the C++ and C++ version of TensorFlow and getting it ready in inference mode with a new code written in the Python version of the TensorFlow.The project is all about creating a chess engine that learns by self-play or reinforcement learning. We are only interested in extracting the neural network part of the engine with its trained weights. The Neural Network in python should have exactly the same architecture and weights borrowed from the provided repository. The entire description of the GitHub repository and the model is given below. While the solution of the contest should be a google collab or jupyter notebook with the model in inference mode.

Chess Engine Repository: https://github.com/LeelaChessZero
Additional Information: https://lczero.org/dev/

Tip: Visit sections 4.3 and 4.4 of the Book "Neural Network for Chess" available at "https://github.com/asdfjkl/neural_network_chess/releases/tag/v1.1" to understand the LCZero project repo and NN better and to complete the project in less than an hour.