From now on we can start cloning AlphaZero instead of Stockfish. Problem is I still don't understand how this AlphZero works for article is very summier.
Can put your alpha beta search and your current evaluation code in the dustbin.
AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
Moderators: hgm, Rebel, chrisw
-
- Posts: 7218
- Joined: Mon May 27, 2013 10:31 am
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
Last edited by Henk on Wed Dec 06, 2017 5:26 pm, edited 1 time in total.
-
- Posts: 6052
- Joined: Tue Jun 12, 2012 12:41 pm
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
It uses eval, of course.MikeGL wrote:Come on Lyudmil, read the paper. AlphaZero don't use eval functions.Lyudmil Tsvetkov wrote:There is no deep eval, eval is static.kranium wrote:Milos wrote:It actually is, instead of 4TPUs required to run Alpha0 so far, on x64 hardware one would need around 2000 Haswell cores to achieve the same speed of NN (80k patterns evaluated per second). Since NNs are huge, with smaller resources matrix multiplication would have to be broken into smaller sub-matrices which would exponentially slow down the calculation.kranium wrote:As Daniel explains: no hard coded evaluation (software)...it's game play is based on learning (experience) from previous self-play games applied to a neural networkLyudmil Tsvetkov wrote:- Alpha had considerable hardware advantage
- SF played with version 8
- what was the code/software/evaluation base used for the first Alpha chess version, an advanced engine evaluation and search software or otherwise?
5,000 first-generation TPUs to generate self-play games
and 64 second-generation TPUs to train the neural networks
The hardware advantage is not such an important factor during gameplay as one would imagine.
AlphaZero very selectively evaluating 80k vs Stockfish's 70,000k positions/sec, probably achieving tremendous depths at such speeds,
but I'd guess it's the deep (learned) positional eval which is primarily adding strength...
How could an entity have 1000 times more terms than SF?
It probably don't know about a Rook being 5.0 or a pawn being 1.0 just like most engines.
It is using MCTS, and doing a lot of probability checks with 1-0, 0.5, 0-1 at the end of the
search Then choosing that path/route which has more wins or draws and least losses.
How would it pick the best move then?
-
- Posts: 186
- Joined: Fri Oct 10, 2014 10:05 pm
- Location: Berkeley, CA
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
The other difference is that Alpha uses MCTS not alpha-beta. Paper says nps used in these games is only 80,000!Daniel Shawul wrote:What is different is that alphazero's evaluation selects features of eval by itself (via a nerual network), while in the standard approach the programmer select features (e.g. passsed pawns, king safety, rook on open file etc) and just tunes the weights.
Also worth noting that Alpha trained for 4 hours, compared to many years of painstakingly tuning Stockfish!
-Carl
-
- Posts: 186
- Joined: Fri Oct 10, 2014 10:05 pm
- Location: Berkeley, CA
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
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.Lyudmil Tsvetkov wrote:Alpha had considerable hardware advantage
-Carl
-
- Posts: 4190
- Joined: Wed Nov 25, 2009 1:47 am
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
4 hours my ass (pardon my french). Try training it on state-of-the-art 1080.clumma wrote:The other difference is that Alpha uses MCTS not alpha-beta. Paper says nps used in these games is only 80,000!Daniel Shawul wrote:What is different is that alphazero's evaluation selects features of eval by itself (via a nerual network), while in the standard approach the programmer select features (e.g. passsed pawns, king safety, rook on open file etc) and just tunes the weights.
Also worth noting that Alpha trained for 4 hours, compared to many years of painstakingly tuning Stockfish!
Fully trained network requres 12h on 5000 gen1 TPUs for self-games and 64 gen2 TPUs for training itself.
Gen1 TPU is like 30x K80 which is like 5x 1080 in performance.
So you'd need like 375k training days with 1080, which is like 1000 years!!!
-
- Posts: 1010
- Joined: Thu Sep 01, 2011 2:49 pm
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
Excellent point.clumma wrote:... but that has got to be near saturation.Lyudmil Tsvetkov wrote:Alpha had considerable hardware advantage
-Carl
I mean at that hardware, there won't be any fluctuation on the best candidate move by SF8 anyway.
-
- Posts: 438
- Joined: Mon Apr 24, 2006 8:06 pm
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
Your math is wrong. I think it is doable with a distributed effort smaller than what was used for Stockfish.Milos wrote:4 hours my ass (pardon my french). Try training it on state-of-the-art 1080.
Fully trained network requres 12h on 5000 gen1 TPUs for self-games and 64 gen2 TPUs for training itself.
Gen1 TPU is like 30x K80 which is like 5x 1080 in performance.
So you'd need like 375k training days with 1080, which is like 1000 years!!!
-
- Posts: 6052
- Joined: Tue Jun 12, 2012 12:41 pm
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
From what I gleaned from hardware comparisons, the advantage is 16/1.clumma wrote: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.Lyudmil Tsvetkov wrote:Alpha had considerable hardware advantage
-Carl
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.
-
- Posts: 6052
- Joined: Tue Jun 12, 2012 12:41 pm
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
Alpha hardware equivalent was somewhere 1024 standard cores.MikeGL wrote:Excellent point.clumma wrote:... but that has got to be near saturation.Lyudmil Tsvetkov wrote:Alpha had considerable hardware advantage
-Carl
I mean at that hardware, there won't be any fluctuation on the best candidate move by SF8 anyway.
How 1024 cores compare with 64 cores?
How scientific is that.
I don't know what saturation you are talking about, from what I read, without fully understanding it, the TPUs are a very different architecture and quite differently affected by general computer chess concepts.
-
- Posts: 8755
- Joined: Sun Feb 26, 2006 4:07 pm
Re: AlphaZero beats AlphaGo Zero, Stockfish, and Elmo
so chess is doomed
and in 25 years, us.
and in 25 years, us.