Ask HN: Is it feasible to recreate Alpha Zero?

2 points by slavaukraini ↗ HN
I'm looking for a toy project to learn more about AI and ML and I was thinking trying to recreate what Deepmind (following their paper) could be a good idea. Do you think it's feasible? Or what are some other good toy projects to get started?

1 comment

[ 3.3 ms ] story [ 11.3 ms ] thread
You won't be able to train it as much, it won't get so good -- but you would still make a pretty mean chess engine that way.