AlphaZero Chess test - SF8 16 GB vs. 512 MB

[CheckersGuy]

Did AlphaZero have a book?

Yes, the learning data can compare to a book.
It's a memory filled good score for lines that give good results (and bad score for lines that gives bad results). There's a small difference because opening lines are not hardcoded (= not a numerical value for each line) but store in a neural network. But with a lot of learning NN become a book with only good moves get only good eval.
This can be compare with old book learning. Example for HIARCS here : http://www.hiarcs.com/pc_uci_options.htm

Book Learning (ON)
This setting allows HIARCS to use its experiences with the current book to make decisions about which moves to make from the book. HIARCS has clever book learning so please use it! The default is ON and clearly best.

When the engine wins with a line, the score increase for this line.

One thing to add. The neural network doesn't score "lines"=sequences of moves but positions. I think saying that the neural network "stores" scores for positions/lines (or whatever...) makes it sound like A0 zero has a list and a score for each element in that list which is not true. Even if you show the neural network a position it has never seen before, it will still give a good result because it has learned concepts which is very different from an openingbook that can't say anything about a position which it has never seen before.

[CheckersGuy]

Did AlphaZero have a book?

Yes, the learning data can compare to a book.
It's a memory filled good score for lines that give good results (and bad score for lines that gives bad results). There's a small difference because opening lines are not hardcoded (= not a numerical value for each line) but store in a neural network. But with a lot of learning NN become a book with only good moves get only good eval.
This can be compare with old book learning. Example for HIARCS here : http://www.hiarcs.com/pc_uci_options.htm

Book Learning (ON)
This setting allows HIARCS to use its experiences with the current book to make decisions about which moves to make from the book. HIARCS has clever book learning so please use it! The default is ON and clearly best.

When the engine wins with a line, the score increase for this line.

One thing to add. The neural network doesn't score "lines"=sequences of moves but positions. I think saying that the neural network "stores" scores for positions/lines (or whatever...) makes it sound like A0 zero has a list and a score for each element in that list which is not true. Even if you show the neural network a position it has never seen before, it will still give a good result because it has learned concepts which is very different from an openingbook that can't say anything about a position which it has never seen before.

[CheckersGuy]

Did AlphaZero have a book?

Yes, the learning data can compare to a book.
It's a memory filled good score for lines that give good results (and bad score for lines that gives bad results). There's a small difference because opening lines are not hardcoded (= not a numerical value for each line) but store in a neural network. But with a lot of learning NN become a book with only good moves get only good eval.
This can be compare with old book learning. Example for HIARCS here : http://www.hiarcs.com/pc_uci_options.htm

Book Learning (ON)
This setting allows HIARCS to use its experiences with the current book to make decisions about which moves to make from the book. HIARCS has clever book learning so please use it! The default is ON and clearly best.

When the engine wins with a line, the score increase for this line.

[CheckersGuy]

Did AlphaZero have a book?

Yes, the learning data can compare to a book.
It's a memory filled good score for lines that give good results (and bad score for lines that gives bad results). There's a small difference because opening lines are not hardcoded (= not a numerical value for each line) but store in a neural network. But with a lot of learning NN become a book with only good moves get only good eval.
This can be compare with old book learning. Example for HIARCS here : http://www.hiarcs.com/pc_uci_options.htm

Book Learning (ON)
This setting allows HIARCS to use its experiences with the current book to make decisions about which moves to make from the book. HIARCS has clever book learning so please use it! The default is ON and clearly best.

When the engine wins with a line, the score increase for this line.

[CheckersGuy]

Did AlphaZero have a book?

Yes, the learning data can compare to a book.
It's a memory filled good score for lines that give good results (and bad score for lines that gives bad results). There's a small difference because opening lines are not hardcoded (= not a numerical value for each line) but store in a neural network. But with a lot of learning NN become a book with only good moves get only good eval.
This can be compare with old book learning. Example for HIARCS here : http://www.hiarcs.com/pc_uci_options.htm

Book Learning (ON)
This setting allows HIARCS to use its experiences with the current book to make decisions about which moves to make from the book. HIARCS has clever book learning so please use it! The default is ON and clearly best.

When the engine wins with a line, the score increase for this line.

One thing to add. The neural network doesn't score "lines"=sequences of moves but positions. I think saying that the neural network "stores" scores for positions/lines (or whatever...) makes it sound like A0 zero has a list and a score for each element in that list which is not true. Even if you show the neural network a position it has never seen before, it will still give a good result because it has learned concepts which is very different from an openingbook that can't say anything about a position which it has never seen before.

[CheckersGuy]

Did AlphaZero have a book?

Yes, the learning data can compare to a book.
It's a memory filled good score for lines that give good results (and bad score for lines that gives bad results). There's a small difference because opening lines are not hardcoded (= not a numerical value for each line) but store in a neural network. But with a lot of learning NN become a book with only good moves get only good eval.
This can be compare with old book learning. Example for HIARCS here : http://www.hiarcs.com/pc_uci_options.htm

Book Learning (ON)
This setting allows HIARCS to use its experiences with the current book to make decisions about which moves to make from the book. HIARCS has clever book learning so please use it! The default is ON and clearly best.

When the engine wins with a line, the score increase for this line.

One thing to add. The neural network doesn't score "lines"=sequences of moves but positions. I think saying that the neural network "stores" scores for positions/lines (or whatever...) makes it sound like A0 zero has a list and a score for each element in that list which is not true. Even if you show the neural network a position it has never seen before, it will still give a good result because it has learned concepts which is very different from an openingbook that can't say anything about a position which it has never seen before.

[Vinvin]

Did AlphaZero have a book?

Yes, the learning data can compare to a book.
It's a memory filled good score for lines that give good results (and bad score for lines that gives bad results). There's a small difference because opening lines are not hardcoded (= not a numerical value for each line) but store in a neural network. But with a lot of learning NN become a book with only good moves get only good eval.
This can be compare with old book learning. Example for HIARCS here : http://www.hiarcs.com/pc_uci_options.htm

Book Learning (ON)
This setting allows HIARCS to use its experiences with the current book to make decisions about which moves to make from the book. HIARCS has clever book learning so please use it! The default is ON and clearly best.

When the engine wins with a line, the score increase for this line.

+ The humans do the same thing : if they win a lot game with an opening line, they try to repeat this opening.

[Vinvin]

Did AlphaZero have a book?

Yes, the learning data can compare to a book.
It's a memory filled good score for lines that give good results (and bad score for lines that gives bad results). There's a small difference because opening lines are not hardcoded (= not a numerical value for each line) but store in a neural network. But with a lot of learning NN become a book with only good moves get only good eval.
This can be compare with old book learning. Example for HIARCS here : http://www.hiarcs.com/pc_uci_options.htm

Book Learning (ON)
This setting allows HIARCS to use its experiences with the current book to make decisions about which moves to make from the book. HIARCS has clever book learning so please use it! The default is ON and clearly best.

When the engine wins with a line, the score increase for this line.

+ The humans do the same thing : if they win a lot with an opening line, they try to repeat it.

[Milos]

One thing to add. The neural network doesn't score "lines"=sequences of moves but positions. I think saying that the neural network "stores" scores for positions/lines (or whatever...) makes it sound like A0 zero has a list and a score for each element in that list which is not true.

This is simply wrong. NN stores 8 moves in its history, i.e. 8 previous positions that were leading to the position that is being evaluated. I'm not gonna quote or argue. This is simply a fact. If you don't get it why, read the bloody paper.

Even if you show the neural network a position it has never seen before, it will still give a good result because it has learned concepts which is very different from an openingbook that can't say anything about a position which it has never seen before.

That is just your speculation. There is not enough evidence to support this.
As far as we know, NN might not even know how to win KRK without search.

[Adam Hair]

I did some single core test 1 MB vs. 64 MB. HERT500.PGN book. With fixed time=1 sec 1 MB was better than bigger? With 60s + 0,6s it changed totally:

Code: Select all

                   
1   SF 64MB   +42  +33/=158/-9 56.00%  112.0/200
2   SF 1MB    -42  +9/=158/-33 44.00%   88.0/200

Obviously fixed time per move DESTROYS SF SEARCH BADLY!

How does fixed time destroy SF's search? Does the search operate in a different manner based on the time control?