I toast to LeelaqueenOdds and its creators. I raise my wine glass high and shout: Long live the queen. We have a queen!!

[Father]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Good morning, Mr. Larry Kaufman. I would like to know, and therefore I ask, in the list of the 100 humans at the top of the LeelaQueenOdds Robot leaderboard, how many of these humans hold the title of Chess Master in any of its denominations. I would also like to know how many of these non-titled players hold a Master level. I appreciate your kindness in helping me with these questions. I would also like to know what you consider to be the "estimated average" of the 100 humans in the leaderboard. Thank you in advance. I wish you a happy day.

"Chess Master" is not a well-defined term; for FIDE it means FM (Normally reaching 2300 elo), for US it means reaching 2200 national rating, other countries may have varying standards. I think that the only practical definition we can use here is to say that anyone with a LiChess blitz rating over 2400 is at least as strong in blitz as a borderline US national master. Many LiChess players are anonymous, so that's the only practical definition we might use. You would then just look up the LIChess blitz ratings of each of them and call them masters if over 2400. As for average strength, if you average the Lichess blitz ratings and then subtract 200, that is a good estimate of the FIDE rating needed for equivalent skill in blitz.

Thank you very much, Mr. Larry Kaufman, for your teachings. I've done a rough comparison, and I believe that more than 50 of the 100 players on the list have all the characteristics of being National Masters with a FIDE ELO of 2200 or more. In addition, there is a significant group of people who also have the quality of being titled players or players of the highest level of play. I want to tell you, Mr. Larry Kaufman, that sometimes I feel "the LeelaQueenOdds machine" is alive and plays with me the way a cat plays with its prey. The large amount of play I have with the machine allows me, at a level of parallelism, to understand and see it mutate. I don't think it's my imagination. I truly perceive an immanent force in the computer. I feel like I'm LeelaQueenOdds's toy. I don't feel demoralized, but perhaps I'll leave things at that. I believe I have the ability and current skill to reach approximately 2710 Elo on the current leaderboard, but it seems I lack the patience, and my passion for playing leads me to play matches at low levels of my biorhythm, which isn't good for reaching a high Elo. I prefer to play more often than to have a higher Elo. I'll try to surprise you all, Mr. Larry Kaufman. If I don't make it, my ceiling would have been lower, Orichess88, last weekend. We'll see what happens. Thanks.

[lkaufman]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Good morning, Mr. Larry Kaufman. I would like to know, and therefore I ask, in the list of the 100 humans at the top of the LeelaQueenOdds Robot leaderboard, how many of these humans hold the title of Chess Master in any of its denominations. I would also like to know how many of these non-titled players hold a Master level. I appreciate your kindness in helping me with these questions. I would also like to know what you consider to be the "estimated average" of the 100 humans in the leaderboard. Thank you in advance. I wish you a happy day.

"Chess Master" is not a well-defined term; for FIDE it means FM (Normally reaching 2300 elo), for US it means reaching 2200 national rating, other countries may have varying standards. I think that the only practical definition we can use here is to say that anyone with a LiChess blitz rating over 2400 is at least as strong in blitz as a borderline US national master. Many LiChess players are anonymous, so that's the only practical definition we might use. You would then just look up the LIChess blitz ratings of each of them and call them masters if over 2400. As for average strength, if you average the Lichess blitz ratings and then subtract 200, that is a good estimate of the FIDE rating needed for equivalent skill in blitz.

Thank you very much, Mr. Larry Kaufman, for your teachings. I've done a rough comparison, and I believe that more than 50 of the 100 players on the list have all the characteristics of being National Masters with a FIDE ELO of 2200 or more. In addition, there is a significant group of people who also have the quality of being titled players or players of the highest level of play. I want to tell you, Mr. Larry Kaufman, that sometimes I feel "the LeelaQueenOdds machine" is alive and plays with me the way a cat plays with its prey. The large amount of play I have with the machine allows me, at a level of parallelism, to understand and see it mutate. I don't think it's my imagination. I truly perceive an immanent force in the computer. I feel like I'm LeelaQueenOdds's toy. I don't feel demoralized, but perhaps I'll leave things at that. I believe I have the ability and current skill to reach approximately 2710 Elo on the current leaderboard, but it seems I lack the patience, and my passion for playing leads me to play matches at low levels of my biorhythm, which isn't good for reaching a high Elo. I prefer to play more often than to have a higher Elo. I'll try to surprise you all, Mr. Larry Kaufman. If I don't make it, my ceiling would have been lower, Orichess88, last weekend. We'll see what happens. Thanks.

It seems that "Arka50" (whom you mentioned earlier), with a blitz rating about 2900 and a bullet rating of 3213, is the strongest GM on the list (at least the strongest with known or suspected identity), and the most successful in 1'0" bullet (playing Black) vs. Leela Queen odds of anyone with a meaningful number of games. We assume he is GM Arkady Naiditsch, although he doesn't actually say so. He is only 4 games behind out of 70, so he might very well soon reach a plus score under those conditions, which would be a first and quite impressive, since so many super-strong bullet players have failed to surpass even 25%.

[Father]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Good morning, Mr. Larry Kaufman. I would like to know, and therefore I ask, in the list of the 100 humans at the top of the LeelaQueenOdds Robot leaderboard, how many of these humans hold the title of Chess Master in any of its denominations. I would also like to know how many of these non-titled players hold a Master level. I appreciate your kindness in helping me with these questions. I would also like to know what you consider to be the "estimated average" of the 100 humans in the leaderboard. Thank you in advance. I wish you a happy day.

"Chess Master" is not a well-defined term; for FIDE it means FM (Normally reaching 2300 elo), for US it means reaching 2200 national rating, other countries may have varying standards. I think that the only practical definition we can use here is to say that anyone with a LiChess blitz rating over 2400 is at least as strong in blitz as a borderline US national master. Many LiChess players are anonymous, so that's the only practical definition we might use. You would then just look up the LIChess blitz ratings of each of them and call them masters if over 2400. As for average strength, if you average the Lichess blitz ratings and then subtract 200, that is a good estimate of the FIDE rating needed for equivalent skill in blitz.

Thank you very much, Mr. Larry Kaufman, for your teachings. I've done a rough comparison, and I believe that more than 50 of the 100 players on the list have all the characteristics of being National Masters with a FIDE ELO of 2200 or more. In addition, there is a significant group of people who also have the quality of being titled players or players of the highest level of play. I want to tell you, Mr. Larry Kaufman, that sometimes I feel "the LeelaQueenOdds machine" is alive and plays with me the way a cat plays with its prey. The large amount of play I have with the machine allows me, at a level of parallelism, to understand and see it mutate. I don't think it's my imagination. I truly perceive an immanent force in the computer. I feel like I'm LeelaQueenOdds's toy. I don't feel demoralized, but perhaps I'll leave things at that. I believe I have the ability and current skill to reach approximately 2710 Elo on the current leaderboard, but it seems I lack the patience, and my passion for playing leads me to play matches at low levels of my biorhythm, which isn't good for reaching a high Elo. I prefer to play more often than to have a higher Elo. I'll try to surprise you all, Mr. Larry Kaufman. If I don't make it, my ceiling would have been lower, Orichess88, last weekend. We'll see what happens. Thanks.

It seems that "Arka50" (whom you mentioned earlier), with a blitz rating about 2900 and a bullet rating of 3213, is the strongest GM on the list (at least the strongest with known or suspected identity), and the most successful in 1'0" bullet (playing Black) vs. Leela Queen odds of anyone with a meaningful number of games. We assume he is GM Arkady Naiditsch, although he doesn't actually say so. He is only 4 games behind out of 70, so he might very well soon reach a plus score under those conditions, which would be a first and quite impressive, since so many super-strong bullet players have failed to surpass even 25%.

A new day begins. I thank God for the life he gives us as human beings, and I give thanks for the algorithms that accompany our algorithms. Good morning, Mr. Larry Kaufman. I hope you are well and continue doing well. I've looked at the leaderboard on LeelaQueenOdds and see that today there's a new leader: "Arka50"!! I want to congratulate Arka50 and hope these words I'm writing reach him. Catecan and I feel like a mule and its rider competing against racehorses. I hope that our technique of moving "steadily and safely" will become very effective over time. When the queen "LeelaQueenOdds" is unbeatable, Catecan and I will shine like the great mule that Dad followed his dreams with: "Sterlina."

[lkaufman]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Good morning, Mr. Larry Kaufman. I would like to know, and therefore I ask, in the list of the 100 humans at the top of the LeelaQueenOdds Robot leaderboard, how many of these humans hold the title of Chess Master in any of its denominations. I would also like to know how many of these non-titled players hold a Master level. I appreciate your kindness in helping me with these questions. I would also like to know what you consider to be the "estimated average" of the 100 humans in the leaderboard. Thank you in advance. I wish you a happy day.

"Chess Master" is not a well-defined term; for FIDE it means FM (Normally reaching 2300 elo), for US it means reaching 2200 national rating, other countries may have varying standards. I think that the only practical definition we can use here is to say that anyone with a LiChess blitz rating over 2400 is at least as strong in blitz as a borderline US national master. Many LiChess players are anonymous, so that's the only practical definition we might use. You would then just look up the LIChess blitz ratings of each of them and call them masters if over 2400. As for average strength, if you average the Lichess blitz ratings and then subtract 200, that is a good estimate of the FIDE rating needed for equivalent skill in blitz.

Thank you very much, Mr. Larry Kaufman, for your teachings. I've done a rough comparison, and I believe that more than 50 of the 100 players on the list have all the characteristics of being National Masters with a FIDE ELO of 2200 or more. In addition, there is a significant group of people who also have the quality of being titled players or players of the highest level of play. I want to tell you, Mr. Larry Kaufman, that sometimes I feel "the LeelaQueenOdds machine" is alive and plays with me the way a cat plays with its prey. The large amount of play I have with the machine allows me, at a level of parallelism, to understand and see it mutate. I don't think it's my imagination. I truly perceive an immanent force in the computer. I feel like I'm LeelaQueenOdds's toy. I don't feel demoralized, but perhaps I'll leave things at that. I believe I have the ability and current skill to reach approximately 2710 Elo on the current leaderboard, but it seems I lack the patience, and my passion for playing leads me to play matches at low levels of my biorhythm, which isn't good for reaching a high Elo. I prefer to play more often than to have a higher Elo. I'll try to surprise you all, Mr. Larry Kaufman. If I don't make it, my ceiling would have been lower, Orichess88, last weekend. We'll see what happens. Thanks.

It seems that "Arka50" (whom you mentioned earlier), with a blitz rating about 2900 and a bullet rating of 3213, is the strongest GM on the list (at least the strongest with known or suspected identity), and the most successful in 1'0" bullet (playing Black) vs. Leela Queen odds of anyone with a meaningful number of games. We assume he is GM Arkady Naiditsch, although he doesn't actually say so. He is only 4 games behind out of 70, so he might very well soon reach a plus score under those conditions, which would be a first and quite impressive, since so many super-strong bullet players have failed to surpass even 25%.

A new day begins. I thank God for the life he gives us as human beings, and I give thanks for the algorithms that accompany our algorithms. Good morning, Mr. Larry Kaufman. I hope you are well and continue doing well. I've looked at the leaderboard on LeelaQueenOdds and see that today there's a new leader: "Arka50"!! I want to congratulate Arka50 and hope these words I'm writing reach him. Catecan and I feel like a mule and its rider competing against racehorses. I hope that our technique of moving "steadily and safely" will become very effective over time. When the queen "LeelaQueenOdds" is unbeatable, Catecan and I will shine like the great mule that Dad followed his dreams with: "Sterlina."

Yes, "Arka50" took over the lead by making a close score in 1'0" bullet, losing just by 44 to 38 plus 14 draws, while scoring about 85% in blitz. Meanwhile Turkish prodigy Ediz Gurel, with an incredible 3401 bullet rating on LiChess, achieved a similar result (losing 15 to 11 with five draws), but doesn't yet have enough games to have a high place on the standings. Both play only as Black. Who will be the first player to make a plus score playing Black at 1'+0" after at least twenty games? Also of interest, GM Andrey Esipenko, world number 42 on FIDE list, played five blitz games with LeelaKnightOdds today (at 3'1"), losing them all, plus three blitz games with LeelaRookOdds (same TC), losing two, winning one. The evidence is growing that Leela can make a plus score against almost anyone in the world giving rook odds in blitz, perhaps only Hikaru and Magnus might win a long match at 3'2" (Leela playing White). One GM, believed to be Devyatkin, has played way over a hundred RAPID games at 10'5" with only two wins! Even if he is just a "par" GM, that's pretty incredible.

[Ovyron]

The queen odds net is available, check on Leela Discord to see how to get it.

It is unavailable to the public because I can't access Discord:



Why is it a private net and not readily available to download as a public link? Why don't these nets appear at https://lczero.org/dev/wiki/best-nets-for-lc0/ ? It claims to feature the best nets but the best are obviously the LeelaQueenOdds nets and they don't appear in the page.

I wish now that people work on OpenLeelaQueenOdds weights that are shared openly and that people can use offline or continue training at home, and that people move from the current solution to open ones that have public links and not hoops people have to jump through to use. Imagine how far Stockfish would have gotten if the source could only have been downloaded from some private link on Discord, this is like one step forward, two steps back.

What if I want to see how badly LeelaQueenOdds destroys me from a normal chess game instead of queen odds? It's the strongest chess entity against humans but I can't play it in normal chess because the bot refuses my challenges? What if I want to analyze some chess position with it to know that's the trickiest continuation? Why are the doors closed?

The way has been shown, we need someone that opens the project to help the community as much as possible instead of only helping the interests of the creators of LeelaQueenOdds.

[Uri Blass]

The queen odds net is available, check on Leela Discord to see how to get it.

It is unavailable to the public because I can't access Discord:



Why is it a private net and not readily available to download as a public link? Why don't these nets appear at https://lczero.org/dev/wiki/best-nets-for-lc0/ ? It claims to feature the best nets but the best are obviously the LeelaQueenOdds nets and they don't appear in the page.

I wish now that people work on OpenLeelaQueenOdds weights that are shared openly and that people can use offline or continue training at home, and that people move from the current solution to open ones that have public links and not hoops people have to jump through to use. Imagine how far Stockfish would have gotten if the source could only have been downloaded from some private link on Discord, this is like one step forward, two steps back.

What if I want to see how badly LeelaQueenOdds destroys me from a normal chess game instead of queen odds? It's the strongest chess entity against humans but I can't play it in normal chess because the bot refuses my challenges? What if I want to analyze some chess position with it to know that's the trickiest continuation? Why are the doors closed?

The way has been shown, we need someone that opens the project to help the community as much as possible instead of only helping the interests of the creators of LeelaQueenOdds.

I think that there is a problem if you can play or analyze with LeelaQueenOdds when the programmers of Leela do not know about your experience.
It means that humans may find a weakness in Leela when they prepare for a match when the programmers of Leela do not know about the weakness in order to fix it and I consider it to be unfair advantage for the human side in playing against Leela.

[Father]

The queen odds net is available, check on Leela Discord to see how to get it.

It is unavailable to the public because I can't access Discord:



Why is it a private net and not readily available to download as a public link? Why don't these nets appear at https://lczero.org/dev/wiki/best-nets-for-lc0/ ? It claims to feature the best nets but the best are obviously the LeelaQueenOdds nets and they don't appear in the page.

I wish now that people work on OpenLeelaQueenOdds weights that are shared openly and that people can use offline or continue training at home, and that people move from the current solution to open ones that have public links and not hoops people have to jump through to use. Imagine how far Stockfish would have gotten if the source could only have been downloaded from some private link on Discord, this is like one step forward, two steps back.

What if I want to see how badly LeelaQueenOdds destroys me from a normal chess game instead of queen odds? It's the strongest chess entity against humans but I can't play it in normal chess because the bot refuses my challenges? What if I want to analyze some chess position with it to know that's the trickiest continuation? Why are the doors closed?

The way has been shown, we need someone that opens the project to help the community as much as possible instead of only helping the interests of the creators of LeelaQueenOdds.

I think that there is a problem if you can play or analyze with LeelaQueenOdds when the programmers of Leela do not know about your experience.
It means that humans may find a weakness in Leela when they prepare for a match when the programmers of Leela do not know about the weakness in order to fix it and I consider it to be unfair advantage for the human side in playing against Leela.

I share Mr. Uri Blass's concept. "The cards should be on the table." Furthermore, humans shouldn't be afraid to reveal their hand once we've played, draw the selected card, and strike in the middle of the ring. In my case, there are more than 10,000 games, including those I've played against the robots LeelaQueenOdds and LeelaRookOdds. This doesn't cause me apprehension or fear; if you prepare a few lines for me, I'll be able to shift my thoughts in another direction within the infinite finite number of move probabilities in the game of science. Incidentally, I've noticed that there are already tables of the top 100 humans in the positions of the various robots. I believe it's time for the best chess players in the world to come and claim the top spots. Creating an "open battle ring" has been an immense success that will provide a universal challenge in the objective of "man versus machine." For now, I'm ranked first on the LeelaRookOdds leaderboard, and 15th on the leaderboard of the beautiful cyber queen, "LeelaQueenOdds." I'd like all players to come and fight for a spot on the list. I've been incredibly surprised and very grateful to Mr. and Grandmaster Hikaru Nakamura for allowing us to watch his 14-match battle against the robot LeelaRookOdds. In my training system, to stay in shape for my battle against the machines, I strive daily to maintain in proper shape "the tripod on which I sustain my existence: "The body, the soul, and the mind." While I exercise my body, I have been watching the video of one hour, fifteen minutes and thirty-one seconds, through which we observe the 14 games between Mr. and GM Hikaru Nakamura against LeelaRookOdds. What a great lesson I have found in Mr. Nakamura's generocity in giving us his knowledge and his chess courage. Generocity has prevailed over ego, and we are all gaining courage.

[lkaufman]

The queen odds net is available, check on Leela Discord to see how to get it.

It is unavailable to the public because I can't access Discord:



Why is it a private net and not readily available to download as a public link? Why don't these nets appear at https://lczero.org/dev/wiki/best-nets-for-lc0/ ? It claims to feature the best nets but the best are obviously the LeelaQueenOdds nets and they don't appear in the page.

I wish now that people work on OpenLeelaQueenOdds weights that are shared openly and that people can use offline or continue training at home, and that people move from the current solution to open ones that have public links and not hoops people have to jump through to use. Imagine how far Stockfish would have gotten if the source could only have been downloaded from some private link on Discord, this is like one step forward, two steps back.

What if I want to see how badly LeelaQueenOdds destroys me from a normal chess game instead of queen odds? It's the strongest chess entity against humans but I can't play it in normal chess because the bot refuses my challenges? What if I want to analyze some chess position with it to know that's the trickiest continuation? Why are the doors closed?

The way has been shown, we need someone that opens the project to help the community as much as possible instead of only helping the interests of the creators of LeelaQueenOdds.

I have no idea why you can't get on Discord to get the net, I have no connection with Discord other than belonging to this Leela group. However I'll say that when you do finally get the queen odds net (which is public), you will be disappointed if you try to use it for normal chess. It was trained specifically to be strong at giving queen odds to humans of queen odds level (roughly FM level in blitz), and it is only really good at odds fairly close to that, like rook and knight or queen for knight. There is likely to be an upgrade to a new queen odds net very soon. We have no concerns about it being public, but the net used for knight odds and rook odds remains private as in this case we do have such concerns and also don't want to deprive the world of seeing the games GMs play against them. For standard chess I recommend just using a top-end Leela net with a high Contempt setting, that should be better for the purpose of how best to beat top human GMs in standard chess.

[Uri Blass]

The queen odds net is available, check on Leela Discord to see how to get it.

It is unavailable to the public because I can't access Discord:



Why is it a private net and not readily available to download as a public link? Why don't these nets appear at https://lczero.org/dev/wiki/best-nets-for-lc0/ ? It claims to feature the best nets but the best are obviously the LeelaQueenOdds nets and they don't appear in the page.

I wish now that people work on OpenLeelaQueenOdds weights that are shared openly and that people can use offline or continue training at home, and that people move from the current solution to open ones that have public links and not hoops people have to jump through to use. Imagine how far Stockfish would have gotten if the source could only have been downloaded from some private link on Discord, this is like one step forward, two steps back.

What if I want to see how badly LeelaQueenOdds destroys me from a normal chess game instead of queen odds? It's the strongest chess entity against humans but I can't play it in normal chess because the bot refuses my challenges? What if I want to analyze some chess position with it to know that's the trickiest continuation? Why are the doors closed?

The way has been shown, we need someone that opens the project to help the community as much as possible instead of only helping the interests of the creators of LeelaQueenOdds.

I have no idea why you can't get on Discord to get the net, I have no connection with Discord other than belonging to this Leela group. However I'll say that when you do finally get the queen odds net (which is public), you will be disappointed if you try to use it for normal chess. It was trained specifically to be strong at giving queen odds to humans of queen odds level (roughly FM level in blitz), and it is only really good at odds fairly close to that, like rook and knight or queen for knight. There is likely to be an upgrade to a new queen odds net very soon. We have no concerns about it being public, but the net used for knight odds and rook odds remains private as in this case we do have such concerns and also don't want to deprive the world of seeing the games GMs play against them. For standard chess I recommend just using a top-end Leela net with a high Contempt setting, that should be better for the purpose of how best to beat top human GMs in standard chess.

Most serious chess players do not play against top GMs so for them the question is not how to beat top human GM's but how to beat weaker chess players in 90+30(fide rating 1600 or 1800 or 2000)

[Uri Blass]

The queen odds net is available, check on Leela Discord to see how to get it.

It is unavailable to the public because I can't access Discord:



Why is it a private net and not readily available to download as a public link? Why don't these nets appear at https://lczero.org/dev/wiki/best-nets-for-lc0/ ? It claims to feature the best nets but the best are obviously the LeelaQueenOdds nets and they don't appear in the page.

I wish now that people work on OpenLeelaQueenOdds weights that are shared openly and that people can use offline or continue training at home, and that people move from the current solution to open ones that have public links and not hoops people have to jump through to use. Imagine how far Stockfish would have gotten if the source could only have been downloaded from some private link on Discord, this is like one step forward, two steps back.

What if I want to see how badly LeelaQueenOdds destroys me from a normal chess game instead of queen odds? It's the strongest chess entity against humans but I can't play it in normal chess because the bot refuses my challenges? What if I want to analyze some chess position with it to know that's the trickiest continuation? Why are the doors closed?

The way has been shown, we need someone that opens the project to help the community as much as possible instead of only helping the interests of the creators of LeelaQueenOdds.

I have no idea why you can't get on Discord to get the net, I have no connection with Discord other than belonging to this Leela group. However I'll say that when you do finally get the queen odds net (which is public), you will be disappointed if you try to use it for normal chess. It was trained specifically to be strong at giving queen odds to humans of queen odds level (roughly FM level in blitz), and it is only really good at odds fairly close to that, like rook and knight or queen for knight. There is likely to be an upgrade to a new queen odds net very soon. We have no concerns about it being public, but the net used for knight odds and rook odds remains private as in this case we do have such concerns and also don't want to deprive the world of seeing the games GMs play against them. For standard chess I recommend just using a top-end Leela net with a high Contempt setting, that should be better for the purpose of how best to beat top human GMs in standard chess.

I also get the same error unable to accept invite when I try the invite link in this link
https://lczero.org/about/community/

Edit:It seems it did not remember my passward after not using discord for a long time and I could fix the problem and could get into discord.