Dear lichess.org community,
we are excited to announce that a preprint of our paper about the neural network engine CrazyAra is available on arXiv:
arxiv.org/abs/1908.06660
The publication is also accompanied with a full re-implementation of the MCTS search in C++.
CrazyAra was initially started as a semester project and exclusively learned to play crazyhouse based on your games.
It was first launched on lichess.org on September 7, 2018:
lichess.org/forum/general-chess-discussion/crazyhouse-ai-crazyara
using a full python backend.
@CrazyAra will be hosted for one day starting from Tue, 20 Aug 2019 14:00 (UTC):
www.timeanddate.com/countdown/generic?iso=20190820T16&p0=992&msg=CrazyAra+0.6.0+launch+on+lichess.org&font=cursive
an will be using our new efficient RISEv2-mobile architecture.
Later the same engine with a different neural network @CrazyAraFish will be online for one day starting from Wed, 21 Aug 2019 14:00 (UTC):
www.timeanddate.com/countdown/generic?iso=20190821T16&p0=992&msg=CrazyAraFish+0.6.0+launch+on+lichess.org&font=cursive
The neural network for CrazyAraFish was first trained on human and later fine-tuned on Stockfish self-play games using the RISEv1 architecture:
You can challenge CrazyAra(Fish) to a crazyhouse match in the blitz or classical time control format.
Moreover, both engines will be playing other engines (e.g. Stockfish) primarily at the end of their up-time.
The source code and binary release for Linux and Windows (CUDA/CuDNN, IntelMKL) can be found at:
github.com/QueensGambit/CrazyAra-Engine
Good luck at your games and we hope you enjoy reading our paper,
~IQ_QI (Johannes Czech)
we are excited to announce that a preprint of our paper about the neural network engine CrazyAra is available on arXiv:
arxiv.org/abs/1908.06660
The publication is also accompanied with a full re-implementation of the MCTS search in C++.
CrazyAra was initially started as a semester project and exclusively learned to play crazyhouse based on your games.
It was first launched on lichess.org on September 7, 2018:
lichess.org/forum/general-chess-discussion/crazyhouse-ai-crazyara
using a full python backend.
@CrazyAra will be hosted for one day starting from Tue, 20 Aug 2019 14:00 (UTC):
www.timeanddate.com/countdown/generic?iso=20190820T16&p0=992&msg=CrazyAra+0.6.0+launch+on+lichess.org&font=cursive
an will be using our new efficient RISEv2-mobile architecture.
Later the same engine with a different neural network @CrazyAraFish will be online for one day starting from Wed, 21 Aug 2019 14:00 (UTC):
www.timeanddate.com/countdown/generic?iso=20190821T16&p0=992&msg=CrazyAraFish+0.6.0+launch+on+lichess.org&font=cursive
The neural network for CrazyAraFish was first trained on human and later fine-tuned on Stockfish self-play games using the RISEv1 architecture:
You can challenge CrazyAra(Fish) to a crazyhouse match in the blitz or classical time control format.
Moreover, both engines will be playing other engines (e.g. Stockfish) primarily at the end of their up-time.
The source code and binary release for Linux and Windows (CUDA/CuDNN, IntelMKL) can be found at:
github.com/QueensGambit/CrazyAra-Engine
Good luck at your games and we hope you enjoy reading our paper,
~IQ_QI (Johannes Czech)
