12 ms·

A typical chess position has 20-40 legal moves. The complete space of moves for the model to predict from has about 1800 moves. For comparison Leela Zero gets

A typical chess position has 20-40 legal moves. The complete space of moves for the model to predict from has about 1800 moves. For comparison Leela Zero gets around 60% accuracy on predicting its own next move. With this sort of accuracy you can reduce the search part of the algorithm to an effective branching factor of 2-4 rather than 40, nearly for free, which is a pretty big win.

nuclearnice1 6y ago
I don’t understand the comment about Leela. Why isn’t own move prediction deterministic?
- thomasahle 6y ago
  Because Leela (like fastchess mentioned above) has two parts: A neural network predicting good moves, and a tree search exploring the moves suggested and evaluating the resulting positions (with a second net). If the prediction (policy) net had a 100% accuracy, you wouldn't need the tree search part at all.
  - heavenlyblue 6y ago
    You haven’t mentioned any nondeterministic behaviour, therefore Leela is supposed to predict it’s own moves with a 100% accuracy.
    - tel 6y ago
      It's not non-determinism, it's partial information. The NN part guesses the best move that will be found by search X% of the time. If you just ditched the search part, Leela would be faster and lose out on (1-X)% of the better moves.
  - nuclearnice1 6y ago
    Got it. Thanks for clarifying. Let me restate. Part one of Leela ranks several chess moves. Part two picks among those. 60% of the time part 2 chooses the #1 ranked move.
    - thomasahle 6y ago
      That works :-) One addition: The second part can be run for an arbitrary amount of time, gradually improving the quality of the returned move. The 60% figure comes from the training games, which are played very quickly, and so don't have a lot of time for refining, thus increasing the prediction accuracy. In real games, tcec-chess.com/ this "self accuracy" would probably be a bit lower.

nuclearnice1 6y ago

I don’t understand the comment about Leela. Why isn’t own move prediction deterministic?

thomasahle 6y ago
Because Leela (like fastchess mentioned above) has two parts: A neural network predicting good moves, and a tree search exploring the moves suggested and evaluating the resulting positions (with a second net). If the prediction (policy) net had a 100% accuracy, you wouldn't need the tree search part at all.
- heavenlyblue 6y ago
  You haven’t mentioned any nondeterministic behaviour, therefore Leela is supposed to predict it’s own moves with a 100% accuracy.
  - tel 6y ago
    It's not non-determinism, it's partial information. The NN part guesses the best move that will be found by search X% of the time. If you just ditched the search part, Leela would be faster and lose out on (1-X)% of the better moves.
- nuclearnice1 6y ago
  Got it. Thanks for clarifying. Let me restate. Part one of Leela ranks several chess moves. Part two picks among those. 60% of the time part 2 chooses the #1 ranked move.
  - thomasahle 6y ago
    That works :-) One addition: The second part can be run for an arbitrary amount of time, gradually improving the quality of the returned move. The 60% figure comes from the training games, which are played very quickly, and so don't have a lot of time for refining, thus increasing the prediction accuracy. In real games, tcec-chess.com/ this "self accuracy" would probably be a bit lower.