👋 Hello there!
👀 I’m interested in machine learning and coding in general. Have a background in speech recognition (pre-deep learning revolution).
🌱 I’ve just finished replicating AlphaGo(Zero) on 5x5 Go. I promise I will write more of my adventure. Now working on 9x9 Go! Go is fun.
Well, so far I was successful in training a small policy network to predict expert moves in TF2 on Google's cloud GPU; got MCTS and selfplay roughly working. And along the way I learned of similar efforts like MuGo, MiniGo and KataGo. There are so many interesting aspects in AlphaZero. Other than 5x5, Go is far from solved!
💞️ If you are interested in collaboration, drop me a line via gmail, same username.