You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The engine has been written from the ground up. In early versions it used a simple PeSTO evaluation, which was replaced by the Stockfish NNUE for a short time. Since version 0.7, Caissa uses it's own efficiently updated neural network, trained with Caissa self-play games using a custom trainer. In a way, the first own Caissa network is based on Stockfish's network, but it was much weaker because of the small data set used back then (a few million positions). Currently (as of version 1.17) over 7.1 billion newly generated positions are used. Also, the old self-play games are successively purged, so that the newer networks are trained only on the most recent games generated by the most recent engine, and so on.
31
+
The engine has been written from the ground up. In early versions it used a simple PeSTO evaluation, which was replaced by the Stockfish NNUE for a short time. Since version 0.7, Caissa uses it's own efficiently updated neural network, trained with Caissa self-play games using a custom trainer. In a way, the first own Caissa network is based on Stockfish's network, but it was much weaker because of the small data set used back then (a few million positions). Currently (as of version 1.18) over 12 billion newly generated positions are used. Also, the old self-play games are successively purged, so that the newer networks are trained only on the most recent games generated by the most recent engine, and so on.
32
32
33
-
The runtime neural network evaluation code is located in [PackedNeuralNetwork.cpp](https://github.com/Witek902/Caissa/blob/devel/src/backend/PackedNeuralNetwork.cpp) and was inspired by [nnue.md document](https://github.com/glinscott/nnue-pytorch/blob/master/docs/nnue.md). The neural network trainer is written completely from scratch and is located in [NetworkTrainer.cpp](https://github.com/Witek902/Caissa/blob/devel/src/utils/NetworkTrainer.cpp), [NeuralNetwork.cpp](https://github.com/Witek902/Caissa/blob/devel/src/utils/NeuralNetwork.cpp) and other NeuralNetwork* files. The trainer is purely CPU-based and is heavily optimized to take advantage of many threads and AVX instructions as well as it exploits the sparse nature of the nets.
33
+
The runtime neural network evaluation code is located in [PackedNeuralNetwork.cpp](https://github.com/Witek902/Caissa/blob/master/src/backend/PackedNeuralNetwork.cpp) and was inspired by [nnue.md document](https://github.com/glinscott/nnue-pytorch/blob/master/docs/nnue.md). The neural network trainer is written completely from scratch and is located in [NetworkTrainer.cpp](https://github.com/Witek902/Caissa/blob/master/src/utils/NetworkTrainer.cpp), [NeuralNetwork.cpp](https://github.com/Witek902/Caissa/blob/master/src/utils/NeuralNetwork.cpp) and other NeuralNetwork* files. The trainer is purely CPU-based and is heavily optimized to take advantage of many threads and AVX instructions as well as it exploits the sparse nature of the nets.
34
34
35
-
The games are generated with the utility [SelfPlay.cpp](https://github.com/Witek902/Caissa/blob/devel/src/utils/SelfPlay.cpp), which generates games with a fixed number of nodes/depth and saves them in a custom binary game format to save space. The opening books used are either Stefan's Pohl [UHO books](https://www.sp-cc.de/downloads--links.htm) or DFRC openings with few random moves played at the beginning.
35
+
The games are generated with the utility [SelfPlay.cpp](https://github.com/Witek902/Caissa/blob/master/src/utils/SelfPlay.cpp), which generates games with a fixed number of nodes/depth and saves them in a custom binary game format to save space. The opening books used are either Stefan's Pohl [UHO books](https://www.sp-cc.de/downloads--links.htm) or DFRC openings with few random moves played at the beginning.
0 commit comments