-
### Endgame bots?
KataGo with its ownership maps and adjustable komi is presumably the best bot for playing a reasonable endgame. It mitigates the problems seen in other zero bots which don't care …
-
Some powerful contributors recently started helping this project. While we thank and kudo them for their resources, some changes are needed for the training pipeline, as the AlphaZero pipeline require…
-
This is the training result in my project. The ```fixup``` is Fixup Initialization (with gradient clipping). The ```renorm``` is Batch Renormalization. The two network both use the same training data …
-
Manjaro Linux, updated, kernel 6.7-rt; Nvidia goes well.
After standard installing Katrain via pipx katago engine doesn't start with 127 error code. Console output searching gives a hint.
Instal…
-
The new version of katago support human-like play,some command like:
./katago.exe gtp -model .bin.gz -human-model b18c384nbt-humanv0.bin.gz -config gtp_human5k_example.cfg
does this tool support i…
-
brew install katrain
brew install katago
brew install python (pip)
**When I look at KaTrain v1.13.0**
ERROR: KataGo config file/Applications/KaTrain.app/Contents/Resources/venv/bin does not exis…
-
@lightvector @y-ich
I used TITAN X (PASCAL) and automatically turned on the FP16 under the 1.3.2 program, but the performance is very low, which is far less efficient than the normal FP32.
-
The simplest approach is to RNN through the game: preserve some signals from some subset of layers from move n and feed them as input to the NN at move n+1. This would give an opportunity to the NN to…
-
![image](https://github.com/lightvector/KataGo/assets/59650171/31b64bb6-3160-462d-b51b-b83f8a253aa0)
In introduction K= 1/sqrt(variance) , but in code K= 1/sqrt(index of blocks) .
this is the co…
-
Endgame in Go often decomposes into independent regions.
Each region can have assigned mean and temperature as defined by CGT (Combinatorial Game Theory).
Playing a region with maximal temperature …