Search found 34 matches

by trulses
Tue Dec 18, 2018 8:38 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Policy training in Alpha Zero, LC0 ..
Replies: 26
Views: 1508

Re: Policy training in Alpha Zero, LC0 ..

I agree. the legal moves list is an attack map, and because of the way it is encoded, a weighted attack map, only for one side though. Unless you're talking about the policy label, you're not discriminating "bad" vs "good" moves by just providing the legal moves so I'm not sure what you mean by wei...
by trulses
Tue Dec 18, 2018 7:58 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Policy training in Alpha Zero, LC0 ..
Replies: 26
Views: 1508

Re: Policy training in Alpha Zero, LC0 ..

I agree. the legal moves list is an attack map, and because of the way it is encoded, a weighted attack map, only for one side though. Unless you're talking about the policy label, you're not discriminating "bad" vs "good" moves by just providing the legal moves so I'm not sure what you mean by wei...
by trulses
Tue Dec 18, 2018 6:38 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Policy training in Alpha Zero, LC0 ..
Replies: 26
Views: 1508

Re: Policy training in Alpha Zero, LC0 ..

chrisw wrote:
Tue Dec 18, 2018 6:00 pm
...

Just a passing thought, but isn’t this breaching the zero-rule?
I think knowing which moves are legal fall under "being given perfect knowledge of the game rules".
by trulses
Tue Dec 18, 2018 2:17 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Policy training in Alpha Zero, LC0 ..
Replies: 26
Views: 1508

Re: Policy training in Alpha Zero, LC0 ..

The label for the policy head is the visit count frequency from the tree search (potentially with a temperature).
by trulses
Fri Dec 07, 2018 4:09 pm
Forum: Computer Chess Club: General Topics
Topic: Alphazero news
Replies: 381
Views: 33099

Re: Alphazero news

matthewlai wrote:
Fri Dec 07, 2018 3:50 pm
[...]
Since you're here in these posts just wanted to say congrats on the publication and thanks for the pseudocode.

Is most of the code for alphazero python code, or is the pseudocode transcribed from a different language like C++?
by trulses
Thu Dec 06, 2018 8:34 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Extension stage in MCTS
Replies: 3
Views: 689

Re: Extension stage in MCTS

In "your typical MCTS" you use your tree policy to pick moves while you're in already expanded nodes. Like Gerd says this might be UCB1 or something to that effect. When you eventually reach a position that's outside your tree you then add it as a leaf node and run your evaluation function and back ...
by trulses
Thu Dec 06, 2018 7:42 pm
Forum: Computer Chess Club: General Topics
Topic: Alphazero news
Replies: 381
Views: 33099

Re: Alphazero news

http://science.sciencemag.org/content/s ... 0.full.pdf

not a lot of new info here but they tried using an opening book for sf at least
by trulses
Thu Dec 06, 2018 7:27 pm
Forum: Computer Chess Club: General Topics
Topic: Matthew Sadler analysis of A0 vs SF [Edit: A0 published in Science?]
Replies: 2
Views: 728

Matthew Sadler analysis of A0 vs SF [Edit: A0 published in Science?]

Not sure if these videos are being posted in another thread but it might be interesting,
https://www.youtube.com/watch?v=JacRX6cKIaY

edit: published here apparently http://science.sciencemag.org/content/362/6419/1140
by trulses
Tue Feb 13, 2018 8:38 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Deep misery
Replies: 13
Views: 4257

Re: Deep misery

I looked at batch normalization. But my network does not use mini-batches and computes gradient for loss over one training example. I read that probably ELU or SELU might help and is faster than using batch normalization. If not then I switch over to mini batches. I logged on for the sole purpose o...