TalkChess.com

Input of a training example consists of position value plus probability for each legal move And how do you know the probability for each legal move from a PGN? You don't. In the normal training of AlphaZero and LC0, your training samples are written at the end of a search, and the visit count of ea...

Henk, you are adding noise to the conversation.

What you are describing is the sort of reinforcement learning that was described in one stage of the initial AlphaGo paper. We are not talking about that here.

Input of a training example consists of position value plus probability for each legal move And how do you know the probability for each legal move from a PGN? You don't. In the normal training of AlphaZero and LC0, your training samples are written at the end of a search, and the visit count of ea...

The code for LC0 is publicly available. The point of the training is that the policy network is learning to guess the result of a search without searching. In order to learn something useful, you don't need the ultimate oracle; it's enough to have access to data that is of better quality that what y...

hgm wrote: ↑

Fri Dec 14, 2018 1:35 pm

But why spend 5 minutes if a solution already exists that works 'out of the box'...?

Well, by asking on the forum you are spending the time of a lot of us trying to help you.

Pretty sure you could do that with python-chess, without a lot of trouble. --Jon This is correct. Here's a 5-minute attempt, and I am not really a Python programmer: #!/usr/bin/python3 import sys import chess import chess.pgn game = chess.pgn.Game() node = game for line in sys.stdin: for move in li...

Here's what I do: Whenever I find a move that becomes the best move at the root, I move it to the top of the list. At the end, I just return the top move. This handles aborted searches very naturally. Move search_root(Board const &board) { std::vector<Move> moves = board.generate_moves(); try { for ...

Or simply keep your code open.

I am not sure comparing the accuracy of these different setups is very informative, since they really are different things. Having W/D/L probabilities has some practical advantages, though. For instance, you can assign a value for draws different than 0, and your engine will then make reasonable dec...

Funny, I was just randomly browsing the RuyDos repository, when I saw a new commit come up, the first one since March. The comment for the commit is: evaluate.cpp from J.M., unchecked; I might regret this ;) This is not an official version, just a development version 1.1.8. https://bitbucket.org/al...