Thanks for adjusting the title to clarify your question - you can also post some text in the main message too, not just the title. :)
You don't need "blank weights", pytorch should randomize things for you already. See documentation here https://github.com/lightvector/KataGo/blob/master/SelfplayTraining.md and the synchronous loop script it describes. If you want to implement different rules though and/or change how the training works, then you'll need to dive in and edit the code yourself, but a good starting point would probably be first making sure you can run training without any changes.
Thanks for adjusting the title to clarify your question - you can also post some text in the main message too, not just the title. :)
You don't need "blank weights", pytorch should randomize things for you already. See documentation here https://github.com/lightvector/KataGo/blob/master/SelfplayTraining.md and the synchronous loop script it describes. If you want to implement different rules though and/or change how the training works, then you'll need to dive in and edit the code yourself, but a good starting point would probably be first making sure you can run training without any changes.