Faster SL worker for parsing large PGN files. It simply iterates through and adds one time every game in every pgn in play_data folder, instead of choosing one at random like in online learning.

Whenever I run AlphaZero chess for the second time after reinstalling python I get the error

File "h5py\h5f.pyx", line 78, in h5py.h5f.open OSError: Unable to open file (File signature not found)

How do I avoid getting that error. When I run the command src\ chess_zero\run.py self --distributed and want to stop execution I type ctrl-c. How else do I stop the command without getting the above error? Thanks.

Philip

Great stuff! I got work in both Windows and Ubuntu... Now I get a question to ask, Since it write in Keras which should should support both Google Tensorflow and MS CNTK. Is any simple way to replace Tensorflow with CNTK?

I am trying to train a network closed to the uploaded one, for getting some experience.

I have used the config of the uploaded one，downloaded games from ficsgames.org and made some attempts. But I can't train a network good enough.

Can you share your training data? Without them, I can't know if I do something wrong or I just use wrong games :(

Trained for about four days using supervised learning feeding large pgn files (CCRL) directly to new version of optimize.py with fit_generator.

Did several passes reducing learning rate from .01 to .001 and finally .0001
Used batch size of 2,048 and increased number of steps to 100,000 for LR .0001 (that took 3 days).

Win rate v prior best weights 96.6% after 29 games.

For example this is what I get: 2017-12-13 09:38:11,484@chess_zero.agent.model_chess DEBUG # model files does not exist at {config_path} and {weight_path} 2017-12-13 09:38:12,425@chess_zero.agent.model_chess DEBUG # save model to {config_path} 2017-12-13 09:38:13,099@chess_zero.agent.model_chess DEBUG # saved model digest {self.digest}

IDK if I'm missing something here, but I don't think logger has access to the calling scope.

@Akababa and @benediamond and @anyothers with comments or suggestions: I noticed some older efforts in your forks with the 8 (half) move history AZ input idea. After considerable, but far from exhaustive, testing with only the current board as input, I am starting to think that some move history is critical. LCZero uses it with a far smaller (64x6) NN than I have been training (256x7), and my net still drops material carelessly. This is after several hundred thousand supervised games.

Anyway, I may be missing other larger picture issues (random self-play works better than supervised?), but I would like try supervised training with some move history, perhaps starting with 4 half moves.

Accordingly, could you please point me to your "best" versions. One was with 110 and another with 105 (96+5), IIRC. I plan to just graft the input back on to the relatively stable Zeta master that I have been fiddling with to use pgn input.

Thanks, Brian

Traceback (most recent call last): File "src/chess_zero/run.py", line 17, in from chess_zero import manager File "src/chess_zero/manager.py", line 22 def setup(config: Config, args): ^ SyntaxError: invalid syntax

Hi, Zeta36!

I am trying to construct an alpha-zero-style AI for Twelve Shogi, a simple kind of Shogi. I am wondering that how you designed the board state for CNN input.

Thanks.

@Zeta36

When I run python src/chess_zero/run.py self under root, it says:

Traceback (most recent call last):
  File "src/chess_zero/run.py", line 16, in <module>
    from chess_zero import manager
  File "src/chess_zero/manager.py", line 6, in <module>
    from .config import Config
  File "src/chess_zero/config.py", line 2, in <module>
    import chess
ModuleNotFoundError: No module named 'chess'

Is there any python module named chess?

Error while running run.py --cmd self

Using TensorFlow backend.
2021-04-14 13:18:34,299@chess_zero.agent.model_chess DEBUG # loading model from C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\data\model\model_best_config.json
Traceback (most recent call last):
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\src\chess_zero\run.py", line 20, in <module>
    manager.start()
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\src\chess_zero\manager.py", line 64, in start
    return self_play.start(config)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\src\chess_zero\worker\self_play.py", line 25, in start
    return SelfPlayWorker(config).start()
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\src\chess_zero\worker\self_play.py", line 45, in __init__
    self.current_model = self.load_model()
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\src\chess_zero\worker\self_play.py", line 85, in load_model
    if self.config.opts.new or not load_best_model_weight(model):
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\src\chess_zero\lib\model_helper.py", line 15, in load_best_model_weight
    return model.load(model.config.resource.model_best_config_path, model.config.resource.model_best_weight_path)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\src\chess_zero\agent\model_chess.py", line 145, in load
    self.model = Model.from_config(json.load(f))
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\engine\training.py", line 2332, in from_config
    functional.reconstruct_from_config(config, custom_objects))
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\engine\functional.py", line 1274, in reconstruct_from_config
    process_layer(layer_data)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\engine\functional.py", line 1256, in process_layer
    layer = deserialize_layer(layer_data, custom_objects=custom_objects)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\layers\serialization.py", line 159, in deserialize
    return generic_utils.deserialize_keras_object(
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\utils\generic_utils.py", line 675, in deserialize_keras_object
    deserialized_obj = cls.from_config(cls_config)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\engine\base_layer.py", line 716, in from_config
    return cls(**config)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\engine\input_layer.py", line 152, in __init__
    super(InputLayer, self).__init__(dtype=dtype, name=name)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\tensorflow\python\training\tracking\base.py", line 522, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\engine\base_layer.py", line 350, in __init__
    self._init_set_name(name)
  File "C:\UTD\Sem 2\Machine Learning\chess-alpha-zero-master\venv\lib\site-packages\keras\engine\base_layer.py", line 2372, in _init_set_name
    backend.observe_object_name(name)
AttributeError: module 'keras.backend' has no attribute 'observe_object_name'

Bumps tensorflow-gpu from 1.15.2 to 2.9.3.

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

I noticed that the version of Tensorflow required is quite old. I'm not sure about this, but I think Ampere GPUs usually require at least CUDA 11, which Tensorflow 2.4.0 requires. Does anyone know if I can still run this repo on Ampere?

Traceback (most recent call last): File "src/chess_zero/run.py", line 20, in manager.start() File "src/chess_zero/manager.py", line 64, in start return self_play.start(config) File "src/chess_zero/worker/self_play.py", line 25, in start return SelfPlayWorker(config).start() File "src/chess_zero/worker/self_play.py", line 69, in start pretty_print(env, ("current_model", "current_model")) File "src/chess_zero/lib/data_helper.py", line 26, in pretty_print pyperclip.copy(env.board.fen()) File "/usr/local/lib/python3.6/dist-packages/pyperclip/init.py", line 659, in lazy_load_stub_copy return copy(text) File "/usr/local/lib/python3.6/dist-packages/pyperclip/init.py", line 336, in call raise PyperclipException(EXCEPT_MSG) pyperclip.PyperclipException: Pyperclip could not find a copy/paste mechanism for your system. For more information, please visit https://pyperclip.readthedocs.io/en/latest/index.html#not-implemented-error Exception in thread prediction_worker: Traceback (most recent call last): File "/usr/lib/python3.6/threading.py", line 916, in _bootstrap_inner self.run() File "/usr/lib/python3.6/threading.py", line 864, in run self._target(*self._args, **self._kwargs) File "src/chess_zero/agent/api_chess.py", line 63, in _predict_batch_worker data.append(pipe.recv()) File "/usr/lib/python3.6/multiprocessing/connection.py", line 250, in recv buf = self._recv_bytes() File "/usr/lib/python3.6/multiprocessing/connection.py", line 407, in _recv_bytes buf = self._recv(4) File "/usr/lib/python3.6/multiprocessing/connection.py", line 383, in _recv raise EOFError EOFError

The engine takes around 30s to 40s per move. It also plays just one opening(English Opening) as white. Is there any way to decrease the per move time? Any and all help would be appreciated.

Hello all,

Just git cloned the project and tried to play with it, the command for Supervised Learning is not python src/chess_zero/run.py sl

but

python src/chess_zero/run.py --cmd sl

else the command "sl" is not recognized