AlphaZero beats AlphaGo Zero, Stockfish, and Elmo

[Lyudmil Tsvetkov]

I think you are confused. You seem to believe you have something to offer.

The confused one is you, and by much.

[abulmo2]

[
I guess ppl are a bit intimidated to ask question because it is Google, but many things are fishy and unfavourable to SF.
One big disadvantage was TC, 1min/move means SF spent only 1 minute for each of the opening moves while in normal TC like 40/40 it would spend easily 5-10 minutes per each of opening moves. That made it much weaker 20 maybe even 30Elo since most of loses for SF already happen in the opening.

Alpha0 may have a good time management too. This is quite trivial to implement. Given the data used by Alpha0, (moves with probabilities of goodness) it should even be possible to provide a better time management for alpha0 than it is for Stockfish.

Second is no-book play, where Alpha0 mainly forces openings and lines that it spent most of the time training and SF had no help from book whatsever, so in this case to make it at least a bit more fair one should use strong book such as Cerebellum as a support to SF.
Starting from 12 typical human openings (only 4 moves deep at max), the gap Alpha0 had over SF reduced from 100 to 77Elo which can be seen from the paper.

Alpha0 can use an opening book too. And with the computanional power available for it, probably a better one than the Cerebellum book.

Third even though they used last year TCEC winner, SF8 has untested behaviour on 64 cores, and on that hardware is at least 30 if not more Elo weaker than the current SFdev.
So taking all into consideration it is pretty safe to assume that latest Brainfish at normal TC like 40/40 would be at list on par if not stronger than Alpha0.

Right, SF has improved since SF8. But SF8 is the last official release and it is logical to test against it.

And all that on much weaker hardware.

I disagree here. Alpha0 uses more efficient hardware, but not bigger one.

[jhellis3]

7nm chips in 2018, give me a break.

I will take that bet, how much do you want to wager?

[Lyudmil Tsvetkov]

So, that, in reality, the hardware difference is not 16/1, as I thought initially, but more like 30/1.
Add to this the early opening advantage Alpha gets due to the simulated book, and conducting the test has been fully meaningless.

Alpha would play not stronger than 1850 on a single core.
Why would I care for such an engine?

What simulated book?

The point of this was to develop and demonstrate a new level of machine learning and AI using these processors...
AlphaZero obtained the knowledge it needed to beat SF in 4 hours!

No one at Google believes this was a super TCEC World Championship event,
it's simple science and research.

What was the base it started from, is it verifiable, and who will verify it?
You bet, simple science and research, Google are so modest about it, but all the forums world-wide are talking only of that, and in a short while, G will offer its kind support to a range of companies...

I would accept and acknowledge an achievement, but it is weaker than 2900.
Why would I consider a 2900-engine a big achievement?

Similarly, why don't we consider Jonny a big achievement, as it plays on equal terms with the top in Leiden and sometimes outplays them?

[Lyudmil Tsvetkov]

More than you with a 100. Very Happy

I'm sure. Well, you can't fix stupid. Bye.

Later.

[Lyudmil Tsvetkov]

What better time management, when the match was played at fixed TC?

[Milos]

Alpha0 may have a good time management too. This is quite trivial to implement. Given the data used by Alpha0, (moves with probabilities of goodness) it should even be possible to provide a better time management for alpha0 than it is for Stockfish.

Speculation, no proof of that.
If it had been possible they wouldn't play 1min/move but some real TC. You think those ppl in Google are stupid and don't know what TC is the most common in chess? Or it is the case that they have intentionally chosen ridiculous TC because it favours Alpha0 the most?

Alpha0 can use an opening book too. And with the computanional power available for it, probably a better one than the Cerebellum book.

Weights of NN are already behaving like a book, any additional book would just degrade the performance. You seems not to quite understand how Alpha0 operates.

And all that on much weaker hardware.

I disagree here. Alpha0 uses more efficient hardware, but not bigger one.

So you find it fair comparing Alpha0 on more efficient (specialized) hardware with SF on general purpose hardware.
SFs eval could easily be ported to FPGAs, there are already available solutions, you use 100s of Xilinx Ultrascale+ boards for eval and a single Haswell CPU for search, something like a DeepBlue. That kind of configuration could easily be 300-400Elo stronger than the current SF on a 64 core machine.

[Lyudmil Tsvetkov]

So what is so new?

Sigh.... I will try to put it terms you can understand.

AlphaZero is not like SF.

AlphaZero evaluated at 80 thousand nodes/sec while SF was at 80 million.

But the eval AlphaZero is using is a self constructed network. So the end result is more like Magnus Carlsen evaluating 80 thousand nodes per second with 0 mistakes. But actually it is worse than that because its "understanding" of the game is even better than Magnus's, it is beyond the human realm, and it is ever improving.

There are no gaps in its understanding for you to repeatedly exploit, and should a very large miracle occur and you were to find one. It would learn, on its own. It does not play chess as humans understand it, it plays chess as it understands it...

Never heard such meaningless harangue.

In order to be new, you must add value. Added value in chess today is strength, and that was what the Alpha team has been pursuing.
Did they succeed? Nope.
Playing at 2850 is not a success, but a failure.
The disgusting thing is that they would advertise their failure as success.

All your other statements are wrong too, just don't have the time to go over all of them.

[Lyudmil Tsvetkov]

I agree with you.
Also what people who claim the HW was much faster..... what they don’t understand is that the thing learned from itself in a very short time!

What if we now give it 1 Year to further learn?


Side note, I looked at the games and they are really impressive!

I would not be surprised, if in a year's time it also starts speaking and thinking of itself.
I, Robot.

[Lyudmil Tsvetkov]

Alpha had considerable hardware advantage

That comparison is not straightforward, but this claim does not seem to be true. SF had 64 threads. I'm not up on the latest scaling behavior of the engine but that has got to be near saturation.

-Carl

From what I gleaned from hardware comparisons, the advantage is 16/1.
Why would one want to run a similar very unfair match?
Only one thing comes to mind: that the company will want to advertise its colossal breakthrough with TPUs and artificial intelligence and then sell its products.

But then, the achievement is not there.

The fact that Google has created a chess playing entity that crushes SF is notable (and fascinating).

TPUs are not for sale, and (at the moment) are applied only to Googles deep learning and research projects,
except when Google donates them to research for free.

https://techcrunch.com/2017/05/17/the-t ... cientists/

What would be the score between SF on 64 cores and SF on 1024 cores out of 100 games?
You think the bigger-hardware SF would score less than 64 points?
I guess at least 80.

So what is so new?
They applied some big hardware, that is all.
The real strength of Alpha is 2850, so around spot 97 or so among engines.
97 is not such a bad achievement, after all.

I doubt if SF on 1024 cores is going to score even 50%
Maybe after some point more cores are counter productive for stockfish.

I also doubt if it is possible to get at least 80 points against stockfish with 64 cores at 1 minute per move.

Why not?
How much would SF 16-cores vs SF single core score, that is easily reproducible.
The experts claim the TPUs lack any SMP inefficiencies.