Crazy Stone Deep Learning -The First Edition-
Crazy Stone has made a huge step forward by combining Deep Neural Networks with Monte Carlo Tree Search.
Estimated lifetime revenueLow confidence
$90.8K
$27.2Klow → high range$272.3K
Modeled estimate, not a reported figure. How we estimate →
Est. copies sold
1.7K
mid estimate
Reviews
35
Media
Screenshots & trailer
About
About the game
Crazy Stone has made a huge step forward by combining Deep Neural Networks with Monte Carlo Tree Search.
The new Crazy Stone employs Deep Convolutional neural networks as a replacement for the pattern database and has produced a significant improvement in strength.
The new program has scored a winning rate of over 90% against the previous program Crazy Stone 2013 !
We have provided 20 levels of play (13k-7d) for all the board sizes.
Crazy Stone has improved not only in strength, but also in his style of play and the lower levels are perfect for the average players.
Features
- 20 levels of play for each board size (9×9, 13×13, 19×19)
- Human vs Computer, Human vs Human (sharing a single device)
- Computer vs Computer Games
- Handicap games, variable options of Komi
- Hint (suggest)
- Instant Undo (available even when the computer is thinking)
- Automatic territory calculation
- Japanese / Chinese Rule
- Suspend / Re-start games
- Save / Load game record in sgf files
- Automatic and manual replay of a game record
- Highlight the last move
- COM resign feature
- Two types of Board and Stones
- Byoyomi games (Timed games)
(You will not be able to select computer level in timed games) - Analysis Mode (Move list, Histogram, Record analysis)
You can analyze your current game and also game records saved in sgf files. Some screen-shot samples of the analysis mode are shown above.