[Computer-go] Source code (Was: Reducing network size? (Was: AlphaGo Zero))
gcp at sjeng.org
Fri Oct 27 00:00:03 PDT 2017
On 27-10-17 00:33, Shawn Ligocki wrote:
> But the data should be different for different komi values, right?
> Iteratively producing self-play games and training with the goal of
> optimizing for komi 7 should converge to a different optimal player
> than optimizing for komi 5.
For the policy (head) network, yes, definitely. It makes no difference
to the value (head) network.
> But maybe having high quality data for komi 7 will still save a lot
> of the work for training a komi 5 (or komi agnostic) network?
I'd suspect so.
More information about the Computer-go