Lc0 51010

[EroSennin]

Over 3000 Elo at only half a million games? Wow, that's insane.

It is insane. Imagine poor Larry, all he learned in dozens of years is well surclassed by this dirt quick net using dirt basic monte carlo search.

Those dozens of years get to 1% of the amount of games Leela has played

[Uri Blass]

Over 3000 Elo at only half a million games? Wow, that's insane.

It is insane. Imagine poor Larry, all he learned in dozens of years is well surclassed by this dirt quick net using dirt basic monte carlo search.

This is not a sad thing. This is what many people in chess have strive to achieve for many reasons. As I chess player. I know we are flawed in our thinking when it comes to chess. And are assumptions are our worst enemy in chess play. This only makes us stronger for every class of chess player.

I don't think that flawed thinking and wrong assumptions are the reason I (or any strong player) lose to 3000+ rated engines. The reason is that I struggle to search even one node per second! Of course my thinking and assumptions are far from perfect, I just don't think that they are worse than what engines "think" or "assume". Although I'm not sure I could beat Lc0 in a serious match even with it set to use policy network only, i.e. one node per move, so maybe my last sentence only applies to A/B engines.

I think we make many wrong assumptions in chess. Lets start with the biggest one.

Chess is played out....

I do not understand.
What is the assumption that you mean in this post?

[mwyoung]

Over 3000 Elo at only half a million games? Wow, that's insane.

It is insane. Imagine poor Larry, all he learned in dozens of years is well surclassed by this dirt quick net using dirt basic monte carlo search.

This is not a sad thing. This is what many people in chess have strive to achieve for many reasons. As I chess player. I know we are flawed in our thinking when it comes to chess. And are assumptions are our worst enemy in chess play. This only makes us stronger for every class of chess player.

I don't think that flawed thinking and wrong assumptions are the reason I (or any strong player) lose to 3000+ rated engines. The reason is that I struggle to search even one node per second! Of course my thinking and assumptions are far from perfect, I just don't think that they are worse than what engines "think" or "assume". Although I'm not sure I could beat Lc0 in a serious match even with it set to use policy network only, i.e. one node per move, so maybe my last sentence only applies to A/B engines.

I think we make many wrong assumptions in chess. Lets start with the biggest one.

Chess is played out....

I do not understand.
What is the assumption that you mean in this post?

Have you ever asked yourself why Lc0 and A0 have stirred so much excitement with Titled players. It is because Lc0 and A0 have show us what we did not expect.

Or Lc0 and A0 would be like any other chess programs. Humans play chess with certain assumptions, and a thinking of the way chess should be played. Chess programs of the past, and now like Lc0 and A0 have show us things we never considered before.

Uri, you have been on CCC as long as I can remember. How many times over the years have you seen people say X chess program is rated X. And this is almost perfect play. I can remember this many times, and today those chess programs stand no chance against the best chess programs of today. And today's engine will stand no change against tomorrows.

Its like when you argued that chess could be solved, and not grasping the meaning of what 10 ^120 means in terms of solving chess.

Remember it was not that many years ago, players like Capablanca, Fischer, Seirawan... and other thought chess was played out. And needed new variants to keep chess alive. Like Capablanca Chess, Fischer random chess, Seirawan chess...

Chess programs have shown that many of our human assumptions about chess are wrong. And are assumptions are many, but all are subsets of Chess is played out...

[Nay Lin Tun]

1. T51 doesnt start from total random play ( they feed 100k game data from T50). Similarily, T40 start from seeded data from T35 and perform head to head vs 3200+ rated Laser within 2 weeks ( in tcec bonus tests)

2. It is smaller 10x128 net, it should learn 4x faster than 20x256 cos of 4x faster game generation. But those nets will saturate faster and cant compete 20x256 nets in top end hardware.

3. Probably due to technology advantage,
( a) Basic Brute force engines ( no prune, very bad at chess)could be 1500_2000
( b) Basic Alpha Beta engines ( benefits of selective search)could be 2200-2500
(c)Basic NN engines ( highly selective search thanks to NN) could be 2800-3000.

[dkappe]

3. Probably due to technology advantage,
( a) Basic Brute force engines ( no prune, very bad at chess)could be 1500_2000
( b) Basic Alpha Beta engines ( benefits of selective search)could be 2200-2500
(c)Basic NN engines ( highly selective search thanks to NN) could be 2800-3000.

“Basic” Alpha-Beta engines don’t perform selective search. The result of their search is provably the same as “brute force.” When you add forward pruning (null-move, etc.) these are selective, but very safe.

Right now the best ab engine is about even with the best nn engine. The arms race, including anti-nn play, is about to start (see the LeelaKiller config of ThothFish).

We live in exciting times for computer chess, but your notions about how these various engines work and what their relative ratings might be, are a bit off.

[Nay Lin Tun]

What I mean is that, AB pruning cut search trees massively and the PV looks like selective search in comparision with brute force search.
In fact SF can be called as highly selective search in comparision with other AB engines cos of her heuristic pruning.( massive narrow search in those bullet TC).

[lkaufman]

Over 3000 Elo at only half a million games? Wow, that's insane.

It is insane. Imagine poor Larry, all he learned in dozens of years is well surclassed by this dirt quick net using dirt basic monte carlo search.

This is not a sad thing. This is what many people in chess have strive to achieve for many reasons. As I chess player. I know we are flawed in our thinking when it comes to chess. And are assumptions are our worst enemy in chess play. This only makes us stronger for every class of chess player.

I don't think that flawed thinking and wrong assumptions are the reason I (or any strong player) lose to 3000+ rated engines. The reason is that I struggle to search even one node per second! Of course my thinking and assumptions are far from perfect, I just don't think that they are worse than what engines "think" or "assume". Although I'm not sure I could beat Lc0 in a serious match even with it set to use policy network only, i.e. one node per move, so maybe my last sentence only applies to A/B engines.

I think we make many wrong assumptions in chess. Lets start with the biggest one.

Chess is played out....

I do not understand.
What is the assumption that you mean in this post?

Have you ever asked yourself why Lc0 and A0 have stirred so much excitement with Titled players. It is because Lc0 and A0 have show us what we did not expect.

Or Lc0 and A0 would be like any other chess programs. Humans play chess with certain assumptions, and a thinking of the way chess should be played. Chess programs of the past, and now like Lc0 and A0 have show us things we never considered before.

Uri, you have been on CCC as long as I can remember. How many times over the years have you seen people say X chess program is rated X. And this is almost perfect play. I can remember this many times, and today those chess programs stand no chance against the best chess programs of today. And today's engine will stand no change against tomorrows.

Its like when you argued that chess could be solved, and not grasping the meaning of what 10 ^120 means in terms of solving chess.

Remember it was not that many years ago, players like Capablanca, Fischer, Seirawan... and other thought chess was played out. And needed new variants to keep chess alive. Like Capablanca Chess, Fischer random chess, Seirawan chess...

Chess programs have shown that many of our human assumptions about chess are wrong. And are assumptions are many, but all are subsets of Chess is played out...

Whether chess is played out depends on how you define the question. It seems pretty clear that if you take the strongest Stockfish and the strongest NN engine on TCEC type hardware, and have them play a long match at TCEC time controls, the results will depend heavily on the openings. If you give each side a really good, deep opening book and have them play only the optimum or near optimum openings, nearly every game will end in a draw. The only way the stronger engine can win games is to play suboptimal lines with White to exit from book early and hope to win despite the poor opening moves. The best moves in the opening are known with reasonable certainty to a great depth, but it is true that the best replies to inferior moves have not all been worked out. In a practical game between humans OTB this is not such a big issue, as we can't remember everything and so it is reasonably practical to get the opponent out of book without playing too suboptimally, but for engine vs engine games with opening books it is a huge problem. TCEC solves it by forcing inferior openings for one side or the other on the opponents; others solve it by disallowing opening books after a few moves. These solutions are an admission that chess is played out in one sense, but not in another. Anyway none of this seems relevant to me to the question of why computers beat humans. No matter how correct our beliefs about chess might be, we cannot compete with something that searches a million times faster than we can.

[supersharp77]

Over 3000 Elo at only half a million games? Wow, that's insane.

It is insane. Imagine poor Larry, all he learned in dozens of years is well surclassed by this dirt quick net using dirt basic monte carlo search.

This is not a sad thing. This is what many people in chess have strive to achieve for many reasons. As I chess player. I know we are flawed in our thinking when it comes to chess. And are assumptions are our worst enemy in chess play. This only makes us stronger for every class of chess player.

I don't think that flawed thinking and wrong assumptions are the reason I (or any strong player) lose to 3000+ rated engines. The reason is that I struggle to search even one node per second! Of course my thinking and assumptions are far from perfect, I just don't think that they are worse than what engines "think" or "assume". Although I'm not sure I could beat Lc0 in a serious match even with it set to use policy network only, i.e. one node per move, so maybe my last sentence only applies to A/B engines.

I think we make many wrong assumptions in chess. Lets start with the biggest one.

Chess is played out....

I do not understand.
What is the assumption that you mean in this post?

Have you ever asked yourself why Lc0 and A0 have stirred so much excitement with Titled players. It is because Lc0 and A0 have show us what we did not expect.

Or Lc0 and A0 would be like any other chess programs. Humans play chess with certain assumptions, and a thinking of the way chess should be played. Chess programs of the past, and now like Lc0 and A0 have show us things we never considered before.

Uri, you have been on CCC as long as I can remember. How many times over the years have you seen people say X chess program is rated X. And this is almost perfect play. I can remember this many times, and today those chess programs stand no chance against the best chess programs of today. And today's engine will stand no change against tomorrows.

Its like when you argued that chess could be solved, and not grasping the meaning of what 10 ^120 means in terms of solving chess.

Remember it was not that many years ago, players like Capablanca, Fischer, Seirawan... and other thought chess was played out. And needed new variants to keep chess alive. Like Capablanca Chess, Fischer random chess, Seirawan chess...

Chess programs have shown that many of our human assumptions about chess are wrong. And are assumptions are many, but all are subsets of Chess is played out...

A Brilliant Consideration! Well Said my Friend....As I view to games of the LC0 type engines...a definite new "Playing Style" Has arisen in the LC0 engine games without a doubt..LC0 throws many of the old conceptions "Out Of The Window" Attack..Defence..Castling...piece play...strategy..gain new meanings as this "Neural Net Engine" steps out from behind the curtain..."The Chess World Will have to adjust...and we will...So Far This Engine seems to Have a Aversion to Playing E-4 with the white pieces"..."Stockfish Still Reigns Supreme"..."Forget Expensive Hardware..I Prefer to Seek Out Good Moves"....

[crem]

1. T51 doesnt start from total random play ( they feed 100k game data from T50). Similarily, T40 start from seeded data from T35 and perform head to head vs 3200+ rated Laser within 2 weeks ( in tcec bonus tests)

Test51 started from random play, no test50 games were used.
Test51 did use the same random games that were used to seed test50, but nothing generated by test50 pipeline was used.

[jp]

Have you ever asked yourself why Lc0 and A0 have stirred so much excitement with Titled players. It is because Lc0 and A0 have show us what we did not expect.

Or Lc0 and A0 would be like any other chess programs. Humans play chess with certain assumptions, and a thinking of the way chess should be played. Chess programs of the past, and now like Lc0 and A0 have show us things we never considered before.
...

It's highly debatable that they have "stirred so much excitement with Titled players", at least compared with non-titled players.
It's even more debatable that they've "shown us things we never considered before", if those "things" mean chess things, except for maybe a concrete move in a concrete position.

Chess programs of the past and present have shown us that there's no substitute for brute-force calculation. I don't think the NN engines have changed or added to that insight.