@omatamixcreated an amazing 4-Player chess engine. Most players know its name, Phoenix!
Phoenix is a ResNET bot, which is a residual neural network. What is a neural network you may ask? A residual neural network is an artificial neural network that skips connections, or shortcuts to jump over layers straight to the output. Input is basically the 4-Player Chess board, and the rules including: castling rights, enpassant, players turn, and all passed.@omatamixis working on Phoenix so it will contain 6 board states which means if it is in a certain position more than once it will play the same line, this makes it easier to recognize patterns. The output in basic terms is the winning percentage. 1- means blue and green won, 1 means red and yellow won. and 0 means a draw. There are also in between percentages including, 0.33 and -0.44 which means that either team is slightly ahead of the other. If it is -77, it means blue and green are winning by a lot, but it is not certain they will win the game, and vise versa. The winning percentage cannot exceed 1 or -1. All you need to know about how Phoenix learns how to play is... well it learns by playing more and more like a regular human.
The Phoenix Bots are written in the code language "C#".C# is a general-purpose, multi-paradigm programming language encompassing strong typing, lexically scoped, imperative, declarative, functional, generic, object-oriented, and component-oriented programming disciplines.Phoenix current rating is around 2300, but@icystun(ranked 2nd in teams) thinks that it is around 2400-2500 rating level. At the moment@omatamixis trying to get the bots to play more rated games so he can determine its true rating. He knows for sure that is above 2250 because it has beaten everyone below that level.
Currently Phoenix has 196 amazing openings, if you add up all the variations for each opening, it creates 594 possible openings! We will share some of the most common ones it uses.
First Opening:h2-h4..b7-c7..g13-g12..m8-l8 2.Ne1-f3..Qa7-b7..Qh14-f12..Qn8-m8 3.g2-g4..Na10-c9..Nj14-i12..m7-l7 4.Qg1-k5..Na5-c6.Contains 7 variations
These were the 4 games played, although Phoenix has a 1-3 record against Terminator, you should put in the consideration of rating differences. I think that it is amazing that anybody, human or bot was able to beat Terminator.
Thank you@omatamixfor giving me all the information needed to write this!
Everything you need to know about Phoenix!
@omatamix created an amazing 4-Player chess engine. Most players know its name, Phoenix!
Phoenix is a ResNET bot, which is a residual neural network. What is a neural network you may ask? A residual neural network is an artificial neural network that skips connections, or shortcuts to jump over layers straight to the output. Input is basically the 4-Player Chess board, and the rules including: castling rights, enpassant, players turn, and all passed. @omatamix is working on Phoenix so it will contain 6 board states which means if it is in a certain position more than once it will play the same line, this makes it easier to recognize patterns. The output in basic terms is the winning percentage. 1- means blue and green won, 1 means red and yellow won. and 0 means a draw. There are also in between percentages including, 0.33 and -0.44 which means that either team is slightly ahead of the other. If it is -77, it means blue and green are winning by a lot, but it is not certain they will win the game, and vise versa. The winning percentage cannot exceed 1 or -1. All you need to know about how Phoenix learns how to play is... well it learns by playing more and more like a regular human.
The Phoenix Bots are written in the code language "C#". C# is a general-purpose, multi-paradigm programming language encompassing strong typing, lexically scoped, imperative, declarative, functional, generic, object-oriented, and component-oriented programming disciplines. Phoenix current rating is around 2300, but @icystun (ranked 2nd in teams) thinks that it is around 2400-2500 rating level. At the moment @omatamix is trying to get the bots to play more rated games so he can determine its true rating. He knows for sure that is above 2250 because it has beaten everyone below that level.
Currently Phoenix has 196 amazing openings, if you add up all the variations for each opening, it creates 594 possible openings! We will share some of the most common ones it uses.
First Opening: h2-h4..b7-c7..g13-g12..m8-l8 2.Ne1-f3..Qa7-b7..Qh14-f12..Qn8-m8 3.g2-g4..Na10-c9..Nj14-i12..m7-l7 4.Qg1-k5..Na5-c6. Contains 7 variations
Second Opening: h2-h4 .. b7-c7 .. h13-h11 .. m8-l8 2. g2-g4 .. Ba6-b7+ .. Bi14-h13+ .. Bn9-m8 3. Bf1-g2. Contains 4 variations
Third Opening: h2-h3 .. b7-c7 .. g13-g12 .. m8-l8 2. Qg1-k5 .. Na5-c6 .. Qh14-e11 .. Qn8-m8 3. e2-e4 .. Ba6-b7. Contains 3 variations.
Phoenix is one of a few bots who play 4-Player chess, to watch it play against @Tony bots, the Terminators, you can go to these links.
Game 1: Winner: Terminator
https://www.chess.com/4-player-chess?g=3058796
Game 2: Winner: Phoenix
https://www.chess.com/4-player-chess?g=3058854
Game 3: Winner: Terminator
https://www.chess.com/4-player-chess?g=3058978
Game 4: Winner: Terminator
https://www.chess.com/4-player-chess?g=3059041
These were the 4 games played, although Phoenix has a 1-3 record against Terminator, you should put in the consideration of rating differences. I think that it is amazing that anybody, human or bot was able to beat Terminator.
Thank you @omatamix for giving me all the information needed to write this!