Page 6 of 16

Re: Engine-Test Experiment on Lichess

Posted: Fri Jan 05, 2024 4:20 pm
by IbaiBuR
Great!

BotLi is excellent, in my opinion much better than Banksia GUI or other clients (at least for now). It is so customizable, you can even use books depending on the time control and the color and change their settings also based on this criteria. It also handles SyzygyTB (both online and local), ChessDB and time outs / disconnections with the server are very very rare unless you have a poor internet connection.

Just in case you want to know, my 2 BOTs are:

- @ReinforcementTrial (24/7 active)
- @GradualLearning

I had a lot more BOT accounts but I closed all of them recently excepting the ones I mention above.

Re: Engine-Test Experiment on Lichess

Posted: Fri Jan 05, 2024 5:17 pm
by Solista
These are things I can't understand. I just won a nice game against ArasanX with Evans Gambit, but I didn't get an Elo point for it. The game didn't count. You get the feeling that some bots on Lichess have a bonus, and others are cheated by the server. :(

Photo: I played 8 games against Arasan, but the last game doesn't count as you can see here:
Image :dunno:

Photo 2: Here you can see that I didn't get any Elo points with only a 100 Elo difference. It's also not easy to win against weak engines; against good ones, all games end in draws. And then you get cheated by the server, sorry! That's not nice.

Image

PGN with nice win (Evans Gambit): :)

[Event "Rated Blitz game"]
[Site "lichess.org"]
[Date "2024.01.05"]
[Round "?"]
[White "EdoBot"]
[Black "ArasanX"]
[Result "1-0"]
[ECO "C52"]
[WhiteElo "2954"]
[BlackElo "2849"]
[PlyCount "125"]
[EventDate "2024.??.??"]
[EventType "blitz"]
[TimeControl "180+1"]

1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. b4 Bxb4 5. c3 Ba5 6. d4 d6 7. Qb3 Qd7 8. O-O Bb6 9. Nbd2 Na5 10. Qc2 Ne7 11. dxe5 Ng6 12. exd6 Qxd6 13. Rd1 O-O 14. Nb3 Qc6 15. Be2 Nxb3 16. axb3 Bg4 17. Rd5 Bxf3 18. Bxf3 Rfe8 19. Bb2 Ne5 20. c4 Nxf3+ 21. gxf3 Rad8 22. Rad1 Rxd5 23. Rxd5 Qg6+ 24. Kf1 Qe6 25. c5 c6 26. Re5 Qh3+ 27. Ke2 Rxe5 28. cxb6 Re8 29. bxa7 Qd7 30. Qc4 Qc7 31. Qc5 f6 32. Bd4 h6 33. h4 Qd7 34. Kd2 Kh8 35. Kc2 Ra8 36. Be3 Rd8 37. Bd4 Qe8 38. Qb4 Qd7 39. Kb2 Kh7 40. Bb6 Ra8 41. Be3 h5 42. f4 Qc7 43. Kc3 Qf7 44. e5 fxe5 45. Qe4+ Kg8 46. Qxe5 Rd8 47. Bb6 Rc8 48. Qd6 c5 49. Bxc5 Qe8 50. Kb2 Qe2+ 51. Ka3 Qb5 52. Qb6 Qd7 53. Be3 Qe7+ 54. Qb4 Qd7 55. Qe4 Re8 56. Qc4+ Kh8 57. Qc5 Qf7 58. Bd4 Rg8 59. f5 Ra8 60. Qd6 Kh7 61. Qe6 Qxe6 62. fxe6 g5 63. Kb4 {Normal} 1-0

Re: Engine-Test Experiment on Lichess

Posted: Fri Jan 05, 2024 5:22 pm
by Solista
IbaiBuR wrote: Fri Jan 05, 2024 4:20 pm Great!

BotLi is excellent, in my opinion much better than Banksia GUI or other clients (at least for now). It is so customizable, you can even use books depending on the time control and the color and change their settings also based on this criteria. It also handles SyzygyTB (both online and local), ChessDB and time outs / disconnections with the server are very very rare unless you have a poor internet connection.

Just in case you want to know, my 2 BOTs are:

- @ReinforcementTrial (24/7 active)
- @GradualLearning

I had a lot more BOT accounts but I closed all of them recently excepting the ones I mention above.
Yes, I really like BotLi. It's really fun. Yes, a lot of things can be adjusted well and the tool is less complicated than the one from Lichess. Thanks again! :cool:

Re: Engine-Test Experiment on Lichess

Posted: Fri Jan 05, 2024 6:37 pm
by IbaiBuR
Solista wrote: Fri Jan 05, 2024 5:17 pm These are things I can't understand. I just won a nice game against ArasanX with Evans Gambit, but I didn't get an Elo point for it. The game didn't count. You get the feeling that some bots on Lichess have a bonus, and others are cheated by the server. :(

Photo: I played 8 games against Arasan, but the last game doesn't count as you can see here:
Image :dunno:

Photo 2: Here you can see that I didn't get any Elo points with only a 100 Elo difference. It's also not easy to win against weak engines; against good ones, all games end in draws. And then you get cheated by the server, sorry! That's not nice.

Image

PGN with nice win (Evans Gambit): :)

[Event "Rated Blitz game"]
[Site "lichess.org"]
[Date "2024.01.05"]
[Round "?"]
[White "EdoBot"]
[Black "ArasanX"]
[Result "1-0"]
[ECO "C52"]
[WhiteElo "2954"]
[BlackElo "2849"]
[PlyCount "125"]
[EventDate "2024.??.??"]
[EventType "blitz"]
[TimeControl "180+1"]

1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. b4 Bxb4 5. c3 Ba5 6. d4 d6 7. Qb3 Qd7 8. O-O Bb6 9. Nbd2 Na5 10. Qc2 Ne7 11. dxe5 Ng6 12. exd6 Qxd6 13. Rd1 O-O 14. Nb3 Qc6 15. Be2 Nxb3 16. axb3 Bg4 17. Rd5 Bxf3 18. Bxf3 Rfe8 19. Bb2 Ne5 20. c4 Nxf3+ 21. gxf3 Rad8 22. Rad1 Rxd5 23. Rxd5 Qg6+ 24. Kf1 Qe6 25. c5 c6 26. Re5 Qh3+ 27. Ke2 Rxe5 28. cxb6 Re8 29. bxa7 Qd7 30. Qc4 Qc7 31. Qc5 f6 32. Bd4 h6 33. h4 Qd7 34. Kd2 Kh8 35. Kc2 Ra8 36. Be3 Rd8 37. Bd4 Qe8 38. Qb4 Qd7 39. Kb2 Kh7 40. Bb6 Ra8 41. Be3 h5 42. f4 Qc7 43. Kc3 Qf7 44. e5 fxe5 45. Qe4+ Kg8 46. Qxe5 Rd8 47. Bb6 Rc8 48. Qd6 c5 49. Bxc5 Qe8 50. Kb2 Qe2+ 51. Ka3 Qb5 52. Qb6 Qd7 53. Be3 Qe7+ 54. Qb4 Qd7 55. Qe4 Re8 56. Qc4+ Kh8 57. Qc5 Qf7 58. Bd4 Rg8 59. f5 Ra8 60. Qd6 Kh7 61. Qe6 Qxe6 62. fxe6 g5 63. Kb4 {Normal} 1-0
I still don't know why this happens... but it has been happening for a long time now...

I never reported it to Lichess but it may be needed to open a pull request at their GitHub?

Or maybe it is due to the Glicko 2 rating system?

I just don't really know

Re: Engine-Test Experiment on Lichess

Posted: Sat Jan 06, 2024 3:51 am
by Solista
Ok, thank you!

But it is a mistake (BUG) when I played 8 games, but the result is not 6.5-1.5 but 5.5-1.5. You can see that I played 8 games, there is also a PGN. This is a BUG. What's strange is that this only happens to me against weaker bots; I haven't seen such errors against good ones. That's strange.

Another word about Open Chess and the secret moderators there:

Yesterday I asked that they delete my entire account. They didn't do that. But instead they blocked my account so that I can no longer write anything. Bravo. They don't respond to my request, but they want to forbid me from writing. They still want me there as a member but want to shut me up. What kind of sneaky people are these who want to rule over other people? The reason for the secret moderation is as follows: these moderators allow nasty insults in the forum. Really nasty personal insults. I have never read anything like that in any other forum. If the moderators responded to my request, they would admit that they read the postings. Then their bad style would be exposed. But if they secretly block posters, then no one knows what will happen. It's secret. I couldn't write a PM to the admin either because this option doesn't work. :dunno: Disgusting. :mad:

Re: Engine-Test Experiment on Lichess

Posted: Sat Jan 06, 2024 5:53 am
by Solista
Good Lichess news: :) Some of my another friends are interested in Lichess, and with your help Ibai, I can now help them set up a bot account too. He are currently still playing on PlayChess, but want to leave there. One of the reasons: the fraud and bragging of some accounts (like ProteusSF) and ChessBase do nothing about it. Someone just wrote me this via email.

Re: Engine-Test Experiment on Lichess

Posted: Sat Jan 06, 2024 9:37 am
by IbaiBuR
Great! Sooner or later I think that PlayChess (Engines) Server is going to disappear. Besides, Lichess is completely free (whereas PlayChess is paid) , you can run the BOT in whatever OS you want (whereas ChessBase is only available for Windows as far as I know) and the performance is likely to be better since the engines are driven throughout the terminal / CMD. And there is also the possibility to play against a lot of non-Stockfish / Stockfish derivatives BOTs, which give some interesting games.

I am happy that they are willing to move indeed!

Best wishes,

Ibai

Re: Engine-Test Experiment on Lichess

Posted: Sat Jan 06, 2024 5:58 pm
by Solista
I saw that the creator of the leaderboards (not Ibai) in the Open Chess Forum wrote to me. I will no longer post there, I have asked the admin there to delete my account. I would like to reiterate that there are no rules as to how someone has to play lichess. As long as he doesn't cheat, everyone can play as they want and challenge bots as they want. I won't let anything stop me, I'm not your monkey for your wishes.

I don't need private leaderboards. Anyone can watch all my games live on Lichess and also download them. Everyone can see who I'm playing against and what my opponent's rating is and what my rating is - all live.

Nevertheless, I have been playing Blitz in matchmaking mode for many hours a day for the last two days. I've played against the best Blitz bots in all this time and no bot has been able to beat me so far. Here is my review of the last two days. As I said, all games were played in matchmaking mode and I challenged all the bots that were available to challenge. I think I have a very good record, but I don't need your rankings for myself. What I see here is enough for me - and that is a great record against the best bots:

My own private Engine "Leptir" played with 5 Cores:
Image

Re: Engine-Test Experiment on Lichess

Posted: Sun Jan 07, 2024 5:51 pm
by Solista
Hello Ibai!

I have a question. I had just lost a game on lichess, because the engine in the Syzygy 6men no longer wanted to move, the pawn was on the 7th row before the conversion.

[Event "Rated Blitz game"]
[Site "lichess.org"]
[Date "2024.01.07"]
[Round "?"]
[White "EdoBot"]
[Black "Intron014"]
[Result "0-1"]
[ECO "C65"]
[WhiteElo "2958"]
[BlackElo "2939"]
[PlyCount "108"]
[EventDate "2024.??.??"]
[EventType "blitz"]
[TimeControl "180"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 Nf6 4. d3 Bc5 5. Nbd2 O-O 6. c3 Re8 7. O-O a6 8. Bc4 Bf8 9. a4 h6 10. Qb3 d5 11. Bxd5 Nxd5 12. exd5 Ne7 13. Nxe5 Qxd5 14. Re1 Bf5 15. d4 f6 16. Nef3 Qxb3 17. Nxb3 g5 18. Be3 b6 19. a5 Reb8 20. Nbd2 Nd5 21. Nf1 Kf7 22. Bd2 b5 23. Ng3 Bd3 24. h4 b4 25. hxg5 hxg5 26. Bxg5 bxc3 27. bxc3 Rb3 28. Bd2 Nxc3 29. Bxc3 Rxc3 30. Ne4 Bxe4 31. Rxe4 Re8 32. Rxe8 Kxe8 33. Kf1 Kd7 34. Nd2 Rc2 35. Nb3 Kc6 36. Re1 Kd5 37. Re3 Bb4 38. g4 Rc3 39. Ke2 Rc2+ 40. Kf1 Rc3 41. Ke2 Rc2+ 42. Kf3 Bc3 43. Nc5 Bxd4 44. Nxa6 Kd6 45. Re2 Rxe2 46. Kxe2 Bc3 47. f4 Bxa5 48. g5 fxg5 49. fxg5 c6 50. Kf1 c5 51. Ke2 c4 52. g6 c3 53. Nb8 c2 54. g7 c1=Q {Time forfeit} 0-1 :(

I got the following message (played with BotLi and newest Python):

self.check_magic(self.variant.tbz_magic, self.variant.pawnless_tbz_magic)
File "C:\Users\user\AppData\Local\Programs\Python\Python312\Lib\site-packages\chess\syzygy.py", line 627, in check_magic
raise IOError(f"invalid magic header: ensure {self.path!r} is a valid syzygy tablebase file")
OSError: invalid magic header: ensure 'E:\\3456men\\KQBvKNN.rtbz' is a valid syzygy tablebase file

I played with the following config:
- "E:/3456men"
max_pieces: 6
instant_play: true

I have played several hundred games without errors on lichess in the last 3 days.

Fritz GUI display is correct:

Image

1N6/6P1/3k4/b7/8/8/4K3/2q5 w - - 0 55

Analysis by Leptir MPV:

1. = (0.00): 55.g8Q
2. -+ (-2): 55.Nd7
3. -+ (-2): 55.Na6
4. -+ (-2): 55.Nc6
5. -+ (-2): 55.Kd3
6. -+ (-2): 55.Kf2
7. -+ (-2): 55.Kf3
8. -+ (-2): 55.g8R
9. -+ (-2): 55.g8B
10. -+ (-1): 55.g8N

What is wrong here?

Re: Engine-Test Experiment on Lichess

Posted: Sun Jan 07, 2024 10:06 pm
by IbaiBuR
Solista wrote: Sun Jan 07, 2024 5:51 pm Hello Ibai!

I have a question. I had just lost a game on lichess, because the engine in the Syzygy 6men no longer wanted to move, the pawn was on the 7th row before the conversion.

[Event "Rated Blitz game"]
[Site "lichess.org"]
[Date "2024.01.07"]
[Round "?"]
[White "EdoBot"]
[Black "Intron014"]
[Result "0-1"]
[ECO "C65"]
[WhiteElo "2958"]
[BlackElo "2939"]
[PlyCount "108"]
[EventDate "2024.??.??"]
[EventType "blitz"]
[TimeControl "180"]

1. e4 e5 2. Nf3 Nc6 3. Bb5 Nf6 4. d3 Bc5 5. Nbd2 O-O 6. c3 Re8 7. O-O a6 8. Bc4 Bf8 9. a4 h6 10. Qb3 d5 11. Bxd5 Nxd5 12. exd5 Ne7 13. Nxe5 Qxd5 14. Re1 Bf5 15. d4 f6 16. Nef3 Qxb3 17. Nxb3 g5 18. Be3 b6 19. a5 Reb8 20. Nbd2 Nd5 21. Nf1 Kf7 22. Bd2 b5 23. Ng3 Bd3 24. h4 b4 25. hxg5 hxg5 26. Bxg5 bxc3 27. bxc3 Rb3 28. Bd2 Nxc3 29. Bxc3 Rxc3 30. Ne4 Bxe4 31. Rxe4 Re8 32. Rxe8 Kxe8 33. Kf1 Kd7 34. Nd2 Rc2 35. Nb3 Kc6 36. Re1 Kd5 37. Re3 Bb4 38. g4 Rc3 39. Ke2 Rc2+ 40. Kf1 Rc3 41. Ke2 Rc2+ 42. Kf3 Bc3 43. Nc5 Bxd4 44. Nxa6 Kd6 45. Re2 Rxe2 46. Kxe2 Bc3 47. f4 Bxa5 48. g5 fxg5 49. fxg5 c6 50. Kf1 c5 51. Ke2 c4 52. g6 c3 53. Nb8 c2 54. g7 c1=Q {Time forfeit} 0-1 :(

I got the following message (played with BotLi and newest Python):

self.check_magic(self.variant.tbz_magic, self.variant.pawnless_tbz_magic)
File "C:\Users\user\AppData\Local\Programs\Python\Python312\Lib\site-packages\chess\syzygy.py", line 627, in check_magic
raise IOError(f"invalid magic header: ensure {self.path!r} is a valid syzygy tablebase file")
OSError: invalid magic header: ensure 'E:\\3456men\\KQBvKNN.rtbz' is a valid syzygy tablebase file

I played with the following config:
- "E:/3456men"
max_pieces: 6
instant_play: true

Hi Eduard!

Did you run the md5sum check of the TB files?

If you have not done it, please do it and tell me the result.

And secondly, do you have all the 6 man-syzygy tablebase files?

If the result is that all files are correct and you have them all, you should open an issue at BotLi's GitHub so the bug can be corrected :)

If you don't know how, don't worry, I'll do it for you, just tell me.

Regards,

Ibai