-
Hey,
I haven't had an opportunity to ask few of these:
- Is the comment about cache misses in `node.h` still relevant? Does it make sense to implement a tagged allocator or some other per-thread…
-
Running train.sh with the networks from http://lczero.org/networks fails with `Weights file is the wrong version.` It seems to expect the file to start with `1` on the first line, but the file seems t…
-
**AS A** machine learning engineer or a project manager
**I WANT TO** understand the performance of our MCTS and TF NN evaluations
**SO THAT I CAN** estimate the time and computation resource for a …
-
It ran for a couple of days and found several new best models. However, it also creates numerous files (502,586 items, totalling 5.6 GB). The models directory is large and the games directory has mo…
-
I am wondering how does the existing LZ handles Mirror Go. Especially by white. Has anyone tested it?
In addition, how do LZ learn to deal with this strategy in the future. I guess none of the tra…
-
Investigation in to possible reasons for the cascading failure of value head quality after v0.8 was released, suggested PUCT and fpu-reduction changes as likely causes.
If we believe that these value…
Tilps updated
6 years ago
-
From AlphaZero Paper :A general reinforcement learning algorithm that
masters chess, shogi and Go through self-play
_A move in chess may be described in two parts: first selecting the piece to mov…
-
the progress is very slow now , 40 blocks may reach the limit.
so, why not explore 50 or 60 blocks from now on?
-
Small note because I know many people will start "panicking" as we have 200k games with no notable strength progress (i.e. PASS).
One reason why progress dropped a bit is that the ELF games left th…
-
Hey @CR-Gjx Thanks for providing this open source code. Very helpful to study and I love the idea of hierarchical reinforcement learning.
In the recent AlphaGo Zero paper and [Thinking Fast and Slo…