Extracting Neural Network from a GitHub repository written in C++ version of TensorFlow
S-au achitat serviciile după ce au fost prestate
The project is about extracting a neural network from a GitHub repository written in C++ and C and even C++ version of TensorFlow to Jupyter notebook or google collab in python version of TensorFlow and in inference mode with accessible layer outputs. The project repo from which we need to extract the neural network is about creating a chess engine that uses a neural network and an MCTS search algorithm and learns through self-play or reinforcement learning . We are only interested in extracting the Neural Network architecture of the Neural Network part of the engine with its latest weights posted in the project's repo. The neural network in google collab should be created in python with the same architecture and weights borrowed from the provided repository.
LCZero Project Repo: [login to view URL]
Additional Help: [login to view URL]
Tip: Refer to sections 4.3 and 4.4 of the book "Neural Networks for Chess" available at "[login to view URL]" for understanding the neural network that is to be extracted and to do the project in less than an hour.
ID Proiect: #32035906