-
https://github.com/gcp/leela-zero/blob/e1284477eb0c66d41af8c80088cbb488caf20341/src/UCTNodeRoot.cpp#L192
Apologies if I am not understanding correctly the code here, but it seems to me that the lin…
-
Hanayeol: "Look at the graph! Look at the graph! Ke Jie is in trouble!!" 😁 Looking good folks!
![image](https://user-images.githubusercontent.com/38073282/39351827-0a53908a-4a03-11e8-8e81-d9e43ef1d…
-
Great stuff! I got work in both Windows and Ubuntu... Now I get a question to ask, Since it write in Keras which should should support both Google Tensorflow and MS CNTK. Is any simple way to replac…
-
Goal is that with a given set of network weights, we should explore the exact same tree with one thread. This should allow us to refactor the search/eval code without fear. Same concept exists in SF…
-
I had similar cases to this output happening a couple of times now:
```
Created dirs data-1
PGN
1. d4 a6 2. a3 Nf6 3. Ra2 Ra7 4. Be3 b5 5. Kd2 Ng4 6. d5 Nxf2 7. Bb6 Nxd1
8. e3 h6 9. Be2 Rb7 10. B…
-
https://alphagoteach.deepmind.com/
It gives us the probabilities of the common openings. From this, we know that Black's best opening move has a 47.1% chance of winning (at Q16), which probably mea…
-
I can compile and make Leela-zero (go program). When trying chess, I get this:
```
brian@Tinker2Ubuntu:~/leela-chess/src$ make
Detected OS: Linux
make \
CXXFLAGS='-I/usr/include/openblas -I. -W…
-
I have compared training with the old pipeline (v1) and with the v2 pipeline, and it seems the v2 pipeline makes the value head extremely more overfit than the v1 pipeline for a reduced number of game…
-
https://github.com/Tencent/PhoenixGo
And their weights here.
https://github.com/Tencent/PhoenixGo/releases
-
From 311 self-play games I generated so far, I counted 157 draws, 139 white wins, and 15 black wins. This strikes me as odd, considering that the first move advantage should be worth very little for a…