Pure Cuda/C++ implementation of AlphaZero using Graph Attention Networks
Note: Currently the code is kinda ugly and complex, I am going through and simplifying wherever possible and making itcleaner.
For the python implementation refer to https://github.com/vpareek2/pyGraphZero