-
DeepMind recently released a paper for their new program, AlphaZero. One of the major changes was "omitting the evaluation step and the selection of best player". If I'm interpreting this correctly, t…
jw098 updated
6 years ago
-
```
pratyush@pratyush-GS65-Stealth-Thin-8RF:~/dev/chess-alpha-zero$ python src/chess_zero/run.py self
2018-11-08 19:17:05,638@chess_zero.manager INFO # config type: mini
Using TensorFlow backend.
…
-
Seems like KataGo is the strongest free and open source out there at the moment, esp. for high handicap against AI.
Now beyond Sabaki and Lizzie gui there is KaTrain and q5Go interfaces...
Say …
-
As of yesterday, minusGo (funny how @amj and I came with such similar names despite being two completely independent projects) got to 9d OGS (https://online-go.com/player/481097/minusGo), which was th…
-
Everyone seems fixated on the lack of progress and the need to change every known possible parameter, but what is troubling is that no one is addressing the issue of oversampling of positions per game…
-
I have a 18 cores 36 threads Xeon E5-2686v3 (its an underclocked version of E5-2699v3) but the cpu version only yields around 1knps using all threads. After updating OpenBLAS to v0.2.19 gives me a hug…
-
New rtx3000 series don't support CUDA10.2
any plan to release new version to support RTX3000 series?
-
It would be interesting to have other more powerful versions having the same human-like style.
-
Bring back support for a neural network heuristic.
-
https://arxiv.org/abs/1809.03928
This is an alternate approach to handling the dynamic komi issue: instead of training one output for multiple komi, try to train 2 parameters that inform the situat…