Historic Milestone: AlphaZero

[Lyudmil Tsvetkov]

Even Stockfish on a Raspberry Pi is a strong chess player, though. So AlphaZero has decent performance, we just need a better comparison. I don't think there is any fundamental reason a NN based system such as AlphaZero couldn't run on commodity hardware, so maybe that can happen.

--Jon

Well, probably lower than 3000 elos.

[Lyudmil Tsvetkov]

Alpha chooses only 1.d4 and 1.Nf3, while Stockfish goes for 1.e4
Judging from this, I can say that Alpha is much weaker than SF in terms of software, and the only reason for the win is the very big hardware advantage.

I think Table 2 [ECO opennings] in the PDF would answer your argument.
All those 12 common opennings (on that Table 2) was played by AlphaZero against SF8, 100 times each. and only a total of 4 losses (out of 300 games) as white starting with 1.e4 (for AlphaZero) as shown on that table.

For some reason, I have no access to that table and the page as a whole.
Very weird.
It seems the page recognises its detractors.

The uppermost right diagram is a French Defence [C00], won't fit on my screen.

*
Just download the PDF (right-click then choose download) then upload again at gmail then view it from there.
Otherwise, just download and view it on your PC using any PDF reader.

The training matches are different from the 100 games match with Stockfish.
It is not at all clear to me where were books used and where not.
12 openings with reversed colours don't square in any way with 100 played games, so did they actually left some openings played more than others, or did not they flip colours?

[kranium]

The training matches are different from the 100 games match with Stockfish.
It is not at all clear to me where were books used and where not.
12 openings with reversed colours don't square in any way with 100 played games, so did they actually left some openings played more than others, or did not they flip colours?

I'm sure opening books were not used...
In the early self-play games things like 1.a3, 1.a4, etc. were probably tried by AlphaZero...
eventually it learned that 1. e4 or 1. d4 had the highest success rates.

[jhellis3]

Time to place an order for one of these: https://www.nvidia.com/en-us/data-center/dgx-station/

Unless Google starts selling TPU2s, which I doubt.....

[Milos]

Time to place an order for one of these: https://www.nvidia.com/en-us/data-center/dgx-station/

Unless Google starts selling TPU2s, which I doubt.....

If K80 is 5k$, that would be like 100k$ minimum starting price.

[JJJ]

Even Stockfish on a Raspberry Pi is a strong chess player, though. So AlphaZero has decent performance, we just need a better comparison. I don't think there is any fundamental reason a NN based system such as AlphaZero couldn't run on commodity hardware, so maybe that can happen.

--Jon

Well, probably lower than 3000 elos.

I don't think any engine below 3000 elo would crush Stockfish with a strong hardware.

[MikeGL]

The training matches are different from the 100 games match with Stockfish.

Yes, the plot on the diagram is the training game, but 100 games per openning was played. 50-50, and the score below the diagram is on AlphaZero perspective.

12 openings with reversed colours don't square in any way with 100 played games, so did they actually left some openings played more than others, or did not they flip colours?

12 opennings x 100 = 1,200 games total.

[jhellis3]

If K80 is 5k$, that would be like 100k$ minimum starting price.

It is $70,000 AFAIK.

[Lyudmil Tsvetkov]

The training matches are different from the 100 games match with Stockfish.
It is not at all clear to me where were books used and where not.
12 openings with reversed colours don't square in any way with 100 played games, so did they actually left some openings played more than others, or did not they flip colours?

I'm sure opening books were not used...
In the early self-play games things like 1.a3, 1.a4, etc. were probably tried by AlphaZero...
eventually it learned that 1. e4 or 1. d4 had the highest success rates.

How can you be sure if they don't specify it?
And it learned wrong. But 1.Nf3?

Is this engine still based on random choices? What perfect engine we are talking about then?

[Lyudmil Tsvetkov]

The training matches are different from the 100 games match with Stockfish.

Yes, the plot on the diagram is the training game, but 100 games per openning was played. 50-50, and the score below the diagram is on AlphaZero perspective.

12 openings with reversed colours don't square in any way with 100 played games, so did they actually left some openings played more than others, or did not they flip colours?

12 opennings x 100 = 1,200 games total.

Before we were talking about 300 and 100, now 1200 suddenly appears...
The 64/36 score certainly comes from 100 games, unless they assigned random points for a win.
And in that sample, I see Alpha playing just 1.d4 and 1.Nf3.