lkaufman wrote: ↑Mon Jan 27, 2025 9:52 pm
Joel leads 4 to 3, final game about to start on Silicon Road Youtube channel
Good evening Mr. Larry Kaufman. Thank you very much for the information. I wonder if it is possible to watch the games in Lichess? I would like to study them with the computer's analysis control.
Personally, I consider every tie a cyber scourge and every victory of the computer against the human with Odds, for me, is something unimaginable until recently. Now if the ribal is a GM human, the problem is much greater. Congratulations to the Leela team. Congratulations to GM Joel, defeating the machine is another epic feat. Battle between giants. Thank you for teaching us how great human thought is, and the creations of machine art and chess.
Final game was a miracle draw by Leela in a losing endgame, so the score was 2 wins for Joel (the first two games!!), one win for Leela, and five draws. All games can be seen either on Matthew Sadler's YouTube channel or simply by searching for the games of "SuperEasy" vs "LeelaKnightOdds" in the LiChess database (just select "tools" and "advanced search") at Classical time controls. After the first day (3 games), the node limit for Leela was raised from 20000 to 30000 (basically meaning instead of moving in one second it took about 1.5 seconds typically, still in bullet chess range), after which it made a plus score. Could be just a coincidence, hard to say, but since the increase it has played about 500 games without a single loss (excluding a forfeit due to a Windows update causing a reboot), despite playing many games with titled players and other players with ratings in the titled range (186 games vs players over 2600 LiChess plus Joel who doesn't play rated on LiChess plus one unrated who is clearly of IM or GM strength). In general Leela struggled quite a bit at b1 odds but was pretty competitive and made an even score at g1 odds. It seems that 1.e4 works pretty well at g1 odds, not so well at b1 odds. Joel played well and I congratulate him on his victory, but Leela made it close enough to prove that beating GMs in a Classical match at knight odds is a realistic near-term goal.
I see that before raising the nodes limit Leela lost against awesome_mule at 15+10 twice when this player is not close to 2600 in lichess.
For players above 2600 I found that IM MarioVillanueva could beat the bot but he lost 169 out of his last 177 games from 15.1.2025 and won twice(most of the games are 3+2 time control).
lkaufman wrote: ↑Mon Jan 27, 2025 9:52 pm
Joel leads 4 to 3, final game about to start on Silicon Road Youtube channel
Good evening Mr. Larry Kaufman. Thank you very much for the information. I wonder if it is possible to watch the games in Lichess? I would like to study them with the computer's analysis control.
Personally, I consider every tie a cyber scourge and every victory of the computer against the human with Odds, for me, is something unimaginable until recently. Now if the ribal is a GM human, the problem is much greater. Congratulations to the Leela team. Congratulations to GM Joel, defeating the machine is another epic feat. Battle between giants. Thank you for teaching us how great human thought is, and the creations of machine art and chess.
Final game was a miracle draw by Leela in a losing endgame, so the score was 2 wins for Joel (the first two games!!), one win for Leela, and five draws. All games can be seen either on Matthew Sadler's YouTube channel or simply by searching for the games of "SuperEasy" vs "LeelaKnightOdds" in the LiChess database (just select "tools" and "advanced search") at Classical time controls. After the first day (3 games), the node limit for Leela was raised from 20000 to 30000 (basically meaning instead of moving in one second it took about 1.5 seconds typically, still in bullet chess range), after which it made a plus score. Could be just a coincidence, hard to say, but since the increase it has played about 500 games without a single loss (excluding a forfeit due to a Windows update causing a reboot), despite playing many games with titled players and other players with ratings in the titled range (186 games vs players over 2600 LiChess plus Joel who doesn't play rated on LiChess plus one unrated who is clearly of IM or GM strength). In general Leela struggled quite a bit at b1 odds but was pretty competitive and made an even score at g1 odds. It seems that 1.e4 works pretty well at g1 odds, not so well at b1 odds. Joel played well and I congratulate him on his victory, but Leela made it close enough to prove that beating GMs in a Classical match at knight odds is a realistic near-term goal.
I see that before raising the nodes limit Leela lost against awesome_mule at 15+10 twice when this player is not close to 2600 in lichess.
For players above 2600 I found that IM MarioVillanueva could beat the bot but he lost 169 out of his last 177 games from 15.1.2025 and won twice(most of the games are 3+2 time control).
Today it has played nearly 200 games without a single draw (just one loss), with about half the games against players over 2200, several over 2600. It sure looks like the higher node limit helps. Note that M. Villanueva is a rather strong IM, 2454 FIDE but almost reached the 2500 GM standard in 2018 at 2496, so a 97% record against him is about a 3050 FIDE blitz performance, far above Carlsen's blitz rating.
lkaufman wrote: ↑Mon Jan 27, 2025 9:52 pm
Joel leads 4 to 3, final game about to start on Silicon Road Youtube channel
Good evening Mr. Larry Kaufman. Thank you very much for the information. I wonder if it is possible to watch the games in Lichess? I would like to study them with the computer's analysis control.
Personally, I consider every tie a cyber scourge and every victory of the computer against the human with Odds, for me, is something unimaginable until recently. Now if the ribal is a GM human, the problem is much greater. Congratulations to the Leela team. Congratulations to GM Joel, defeating the machine is another epic feat. Battle between giants. Thank you for teaching us how great human thought is, and the creations of machine art and chess.
Final game was a miracle draw by Leela in a losing endgame, so the score was 2 wins for Joel (the first two games!!), one win for Leela, and five draws. All games can be seen either on Matthew Sadler's YouTube channel or simply by searching for the games of "SuperEasy" vs "LeelaKnightOdds" in the LiChess database (just select "tools" and "advanced search") at Classical time controls. After the first day (3 games), the node limit for Leela was raised from 20000 to 30000 (basically meaning instead of moving in one second it took about 1.5 seconds typically, still in bullet chess range), after which it made a plus score. Could be just a coincidence, hard to say, but since the increase it has played about 500 games without a single loss (excluding a forfeit due to a Windows update causing a reboot), despite playing many games with titled players and other players with ratings in the titled range (186 games vs players over 2600 LiChess plus Joel who doesn't play rated on LiChess plus one unrated who is clearly of IM or GM strength). In general Leela struggled quite a bit at b1 odds but was pretty competitive and made an even score at g1 odds. It seems that 1.e4 works pretty well at g1 odds, not so well at b1 odds. Joel played well and I congratulate him on his victory, but Leela made it close enough to prove that beating GMs in a Classical match at knight odds is a realistic near-term goal.
I see that before raising the nodes limit Leela lost against awesome_mule at 15+10 twice when this player is not close to 2600 in lichess.
For players above 2600 I found that IM MarioVillanueva could beat the bot but he lost 169 out of his last 177 games from 15.1.2025 and won twice(most of the games are 3+2 time control).
Today it has played nearly 200 games without a single draw (just one loss), with about half the games against players over 2200, several over 2600. It sure looks like the higher node limit helps. Note that M. Villanueva is a rather strong IM, 2454 FIDE but almost reached the 2500 GM standard in 2018 at 2496, so a 97% record against him is about a 3050 FIDE blitz performance, far above Carlsen's blitz rating.
Volviendo a LeelaQueenOdds: Que una máquina sin su reina gane a un jugador por sobre 3000 elo Lichess en buler y de paso Maestro de e algo del otro mundo 🪽
I am thinking chess is in a coin.Human beings for ever playing in one face.Now I am playing in the other face:"Antichess". Computers are as a fortres where owner forgot to close a little door behind. You must enter across this door.Forget the front.
lkaufman wrote: ↑Mon Jan 27, 2025 9:52 pm
Joel leads 4 to 3, final game about to start on Silicon Road Youtube channel
Today it has played nearly 200 games without a single draw (just one loss), with about half the games against players over 2200, several over 2600. It sure looks like the higher node limit helps. Note that M. Villanueva is a rather strong IM, 2454 FIDE but almost reached the 2500 GM standard in 2018 at 2496, so a 97% record against him is about a 3050 FIDE blitz performance, far above Carlsen's blitz rating.
Volviendo a LeelaQueenOdds: Que una máquina sin su reina gane a un jugador por sobre 3000 elo Lichess en buler y de paso Maestro de e algo del otro mundo 🪽
In the last couple hours the world number 2 on official FIDE list, Fabiano Caruana, played nine games with Leela Knight Odds at 3'2", losing 8.5 to 0.5!!!!!
Volviendo a LeelaQueenOdds: Que una máquina sin su reina gane a un jugador por sobre 3000 elo Lichess en buler y de paso Maestro de e algo del otro mundo 🪽
The total score for LeelaQueenOdds playing White at 1'0" vs players rated over 3000 bullet since the Nov. 12 upgrade is: 338 wins for Leela, 50 draws, and 58 losses. Even the best players need a one second increment (or a second minute base) to make a plus score with queen odds (Leela White).
Volviendo a LeelaQueenOdds: Que una máquina sin su reina gane a un jugador por sobre 3000 elo Lichess en buler y de paso Maestro de e algo del otro mundo 🪽
The total score for LeelaQueenOdds playing White at 1'0" vs players rated over 3000 bullet since the Nov. 12 upgrade is: 338 wins for Leela, 50 draws, and 58 losses. Even the best players need a one second increment (or a second minute base) to make a plus score with queen odds (Leela White).
Good evening Mr. Larry Kaufman. I wonder: Could LeelaQueenOdds's rating at 1'0" playing the white computer mean that it is 3450?
I am thinking chess is in a coin.Human beings for ever playing in one face.Now I am playing in the other face:"Antichess". Computers are as a fortres where owner forgot to close a little door behind. You must enter across this door.Forget the front.
Volviendo a LeelaQueenOdds: Que una máquina sin su reina gane a un jugador por sobre 3000 elo Lichess en buler y de paso Maestro de e algo del otro mundo 🪽
The total score for LeelaQueenOdds playing White at 1'0" vs players rated over 3000 bullet since the Nov. 12 upgrade is: 338 wins for Leela, 50 draws, and 58 losses. Even the best players need a one second increment (or a second minute base) to make a plus score with queen odds (Leela White).
Good evening Mr. Larry Kaufman. I wonder: Could LeelaQueenOdds's rating at 1'0" playing the white computer mean that it is 3450?
The average opponent over 3000 is about 3100 (maybe a tad more), and the result is a +256 elo result, so 3350 is a reasonably conservative number to use. Of course this is a Lichess bullet elo; the Lichess bullet ratings run far above the Lichess blitz elos, which in turn run far above FIDE ratings. 3350 is roughly the top of the human scale for Lichess bullet (current number 1 is 3351), so perhaps LeelaQueenOdds vs Magnus Carlsen at 1'0", Leela White, would be a fair match.
Volviendo a LeelaQueenOdds: Que una máquina sin su reina gane a un jugador por sobre 3000 elo Lichess en buler y de paso Maestro de e algo del otro mundo 🪽
The total score for LeelaQueenOdds playing White at 1'0" vs players rated over 3000 bullet since the Nov. 12 upgrade is: 338 wins for Leela, 50 draws, and 58 losses. Even the best players need a one second increment (or a second minute base) to make a plus score with queen odds (Leela White).
Good evening Mr. Larry Kaufman. I wonder: Could LeelaQueenOdds's rating at 1'0" playing the white computer mean that it is 3450?
The average opponent over 3000 is about 3100 (maybe a tad more), and the result is a +256 elo result, so 3350 is a reasonably conservative number to use. Of course this is a Lichess bullet elo; the Lichess bullet ratings run far above the Lichess blitz elos, which in turn run far above FIDE ratings. 3350 is roughly the top of the human scale for Lichess bullet (current number 1 is 3351), so perhaps LeelaQueenOdds vs Magnus Carlsen at 1'0", Leela White, would be a fair match.
Mr. Larry Kaufman. Wonderful concept to consider that LeelaQueenOdda at 1'0" in a match with Magnus Carlsen would be a challenge between two titans. I do not lose faith and I do not lose hope that Magnus Carlsen will give us the immense joy of playing a 1'0 streak " vs. LeelaQueenOdds. I hope that Magnus Carlsen reads these posts and gives us the joy and gift of a good dose of happiness. Catecan, a simple wooden horse shouts to the sky: "Give us joy and happiness."
I am thinking chess is in a coin.Human beings for ever playing in one face.Now I am playing in the other face:"Antichess". Computers are as a fortres where owner forgot to close a little door behind. You must enter across this door.Forget the front.
I do not see a reason to assume that the best player in normal chess is going to be the best player with huge odds at bullet,
I will not be surprised if some player with fide rating 2600 may perform better than carlsen at bullet with queen odds against lc0.
I consider even without odds bullet and normal chess to be different games and I will not be surprised to see in the future a world champion in normal chess that is not a top bullet player at 1+0.
It is not important to know to move the mouse fast in long time control.
Uri Blass wrote: ↑Wed Jan 29, 2025 7:41 am
I do not see a reason to assume that the best player in normal chess is going to be the best player with huge odds at bullet,
I will not be surprised if some player with fide rating 2600 may perform better than carlsen at bullet with queen odds against lc0.
I consider even without odds bullet and normal chess to be different games and I will not be surprised to see in the future a world champion in normal chess that is not a top bullet player at 1+0.
It is not important to know to move the mouse fast in long time control.
Good morning Uri. I hope you are well and continue to be well. I agree with your message. However, the presence and catalyzing of the computer in front of an elite player is the best reference to qualify the computer's elo and resume. The Masters are "the salt of chess" without them chess would be incipient. And the champion of the human world is the great seasoning of the science game. The genius, talent and title of the reigning world champion are essential as a reference. Even in the case in which a qualified or unqualified player other than the world champion obtained the best results against the machine, the fight between the heavyweights of martial arts and boxing against the computer would always remain pending.
This is just my opinion Uri. That's all. Some sayings come to mind: "The habit does not make the monk." "The monkey, even if he dresses in cloth, remains a monkey." ""Caesar's wife must not only be honest, but also appear to be honest." I just invented this in the imagination that one day the "good catecan" my wooden horse would be number one on the LeelaQueenOdds standings: "A "Little wooden horse and its rider just run and win and are awarded the derby of dreams and ideals that live inside a storybook."
I am thinking chess is in a coin.Human beings for ever playing in one face.Now I am playing in the other face:"Antichess". Computers are as a fortres where owner forgot to close a little door behind. You must enter across this door.Forget the front.
🪽