Search found 1950 matches

by Michel
Fri Jul 13, 2018 7:01 pm
Forum: Computer Chess Club: General Topics
Topic: how will Leela fare at the WCCC?
Replies: 152
Views: 6375

Re: how will Leela fare at the WCCC?

Not sure if the result against K12.1 is really disappointing. The error bars indicate that both engines could still be equal strength...
by Michel
Sat May 19, 2018 8:07 am
Forum: Computer Chess Club: General Topics
Topic: LCZero: Progress and Scaling. Relation to CCRL Elo
Replies: 604
Views: 74971

Re: LCZero: Progress and Scaling. Relation to CCRL Elo

Guess what, why this happens is probably because of those who have a clue about how it may work probably wouldn't bother to argue with those self-complacent experts in the future way of developing chess engines. Come on. This is unfair. Lcero is the first public NN+MCTS engine. So nobody "has a clu...
by Michel
Sun May 13, 2018 12:51 pm
Forum: Computer Chess Club: General Topics
Topic: Bad day at the stock market?
Replies: 49
Views: 4045

Re: Bad day at the stock market?

Contributing users play >100,000 games per day so there should be sufficient significance to rule out any weird statistical explanations. I _think_ the training window is 500,000 games. So it may take a while before the effect of corrupted training data become visible. On the list of existing netwo...
by Michel
Sun May 13, 2018 11:58 am
Forum: Computer Chess Club: General Topics
Topic: Bad day at the stock market?
Replies: 49
Views: 4045

Re: Bad day at the stock market?

I don't see how a larger network can lead to better play all by itself. It will first have to be trained. I
I think it was trained for a while (using the existing training data) before it was released.
by Michel
Sun May 13, 2018 11:51 am
Forum: Computer Chess Club: General Topics
Topic: Bad day at the stock market?
Replies: 49
Views: 4045

Re: Bad day at the stock market?

Contributing users play >100,000 games per day so there should be sufficient significance to rule out any weird statistical explanations.
I _think_ the training window is 500,000 games. So it may take a while before the effect of corrupted training data become visible.
by Michel
Sun May 13, 2018 9:26 am
Forum: Computer Chess Club: General Topics
Topic: Lczero stalled in progress?
Replies: 34
Views: 3113

Re: Lczero stalled in progress?

I think the developers of Leela Chess Zero should learn methodology from the developers of Stockfish. The success of Stockfish originates mainly from its systematical development. That is easier said than done. The feedback loop of lczero is much longer than that of SF since it passes through the t...
by Michel
Mon May 07, 2018 9:43 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: MCTS with minmax backup operator?
Replies: 24
Views: 2522

Re: MCTS with minmax backup operator?

This is going to annoy you more but I am going to repeat: Whether it is a good/bad policy NN, it is going to fail. It better be good, but even in that case it is going to fail because traps are very common in chess. It doesn't annoy me :-) I just find your argument not as compelling as you do. But ...
by Michel
Sun May 06, 2018 1:57 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: MCTS with minmax backup operator?
Replies: 24
Views: 2522

Re: MCTS with minmax backup operator?

The policy network is a static rule (unlike an actual shallow depth search). If it is mostly good, then it won't give enough visits for trap moves that look bad initially but turn out to be good on exhaustive search. If it is bad, there is no hope. You keep repeating this but it does not have to be...
by Michel
Sat May 05, 2018 12:05 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: MCTS with minmax backup operator?
Replies: 24
Views: 2522

Re: MCTS with minmax backup operator?

Remi Coulom wrote:Averaging values produces a pessimistic estimation of the current board. So the move that leads to it will have an optimistic evaluation, which is necessary for the good dynamics of MCTS.
That's a very nice theory!
by Michel
Fri Apr 27, 2018 6:07 am
Forum: Computer Chess Club: General Topics
Topic: LC0 on the WAC test suite, bad results
Replies: 54
Views: 8457

Re: LC0 on the WAC test suite, bad results

I am pretty sure the NN alone can't do it, so the debate was if the policy NN can guide the MCTS search precisely to solve tactics. The task of the policy head should be to indicate "interesting" moves. I do not see why a NN would not be capable of doing something like that. For example: I read som...