chipiron.players.boardevaluators.neural_networks.input_converters package

Submodules

chipiron.players.boardevaluators.neural_networks.input_converters.ModelInputRepresentationType module

class chipiron.players.boardevaluators.neural_networks.input_converters.ModelInputRepresentationType.InternalTensorRepresentationType(value, names=<not given>, *values, module=None, qualname=None, type=None, start=1, boundary=None)[source]

Bases: str, Enum

BUG364 = '364_bug'
NO = 'no'
NOBUG364 = '364_no_bug'
class chipiron.players.boardevaluators.neural_networks.input_converters.ModelInputRepresentationType.ModelInputRepresentationType(value, names=<not given>, *values, module=None, qualname=None, type=None, start=1, boundary=None)[source]

Bases: str, Enum

BOARD_PIECES_TWO_SIDES = 'board_pieces_two_sides'
BUG364 = '364_bug'
NOBUG364 = '364_no_bug'
PIECE_DIFFERENCE = 'pieces_difference'
PIECE_MAP = 'piece_map'
chipiron.players.boardevaluators.neural_networks.input_converters.ModelInputRepresentationType.assert_compatibilities_representation_type(model_input_representation_type: ModelInputRepresentationType, internal_tensor_representation_type: InternalTensorRepresentationType) → None[source]
chipiron.players.boardevaluators.neural_networks.input_converters.ModelInputRepresentationType.get_default_internal_representation(model_input_representation_type: ModelInputRepresentationType) → InternalTensorRepresentationType[source]

chipiron.players.boardevaluators.neural_networks.input_converters.board_representation module

Module defining the board representation interface and the 364 features board representation.

class chipiron.players.boardevaluators.neural_networks.input_converters.board_representation.BoardRepresentation(*args, **kwargs)[source]

Bases: Protocol

Protocol defining the interface for a board representation.

get_evaluator_input(color_to_play: bool) → Tensor[source]

Returns the evaluator input tensor for the given color to play.

Parameters:

color_to_play – The color to play, either chess.WHITE or chess.BLACK.

Returns:

The evaluator input tensor.

class chipiron.players.boardevaluators.neural_networks.input_converters.board_representation.Representation364(tensor_white: Tensor, tensor_black: Tensor, tensor_castling_white: Tensor, tensor_castling_black: Tensor)[source]

Bases: BoardRepresentation

Dataclass representing a board representation with 364 features.

get_evaluator_input(color_to_play: bool) → Tensor[source]

Returns the evaluator input tensor for the given color to play.

Parameters:

color_to_play – The color to play, either chess.WHITE or chess.BLACK.

Returns:

The evaluator input tensor.

tensor_black: Tensor
tensor_castling_black: Tensor
tensor_castling_white: Tensor
tensor_white: Tensor
class chipiron.players.boardevaluators.neural_networks.input_converters.board_representation.Representation364_2(tensor: Tensor)[source]

Bases: BoardRepresentation

Dataclass representing a board representation with 364 features.

get_evaluator_input(color_to_play: bool) → Tensor[source]

Returns the evaluator input tensor for the given color to play.

Parameters:

color_to_play – The color to play, either chess.WHITE or chess.BLACK.

Returns:

The evaluator input tensor.

tensor: Tensor

chipiron.players.boardevaluators.neural_networks.input_converters.board_to_input module

Module for the BoardToInput protocol and BoardToInputFunction protocol.

class chipiron.players.boardevaluators.neural_networks.input_converters.board_to_input.BoardToInput(*args, **kwargs)[source]

Bases: Protocol

Protocol for converting a chess board to a tensor input for a neural network.

convert(board: IBoard) → Tensor[source]

Converts the given chess board to a tensor input.

Parameters:

board (BoardChi) – The chess board to convert.

Returns:

The tensor input representing the chess board.

Return type:

torch.Tensor

class chipiron.players.boardevaluators.neural_networks.input_converters.board_to_input.BoardToInputFunction(*args, **kwargs)[source]

Bases: Protocol

Protocol for a callable object that converts a chess board to a tensor input for a neural network.

chipiron.players.boardevaluators.neural_networks.input_converters.board_to_input.create_board_to_input(model_input_representation_type: ModelInputRepresentationType) → BoardToInputFunction[source]
chipiron.players.boardevaluators.neural_networks.input_converters.board_to_input.create_board_to_input_from_representation(internal_tensor_representation_type: InternalTensorRepresentationType) → BoardToInputFunction[source]

chipiron.players.boardevaluators.neural_networks.input_converters.board_to_transformer_input module

chipiron.players.boardevaluators.neural_networks.input_converters.board_to_transformer_input.build_transformer_input(piece_map: dict[int, tuple[int, bool]], board_turn: bool, transformer_args: TransformerArgs) → Tensor[source]

chipiron.players.boardevaluators.neural_networks.input_converters.factory module

This module provides functions and classes for creating board representations used in neural networks.

The main classes in this module are: - Representation364: Represents the board state using a tensor of size 364. - Representation364Factory: Factory class for creating Representation364 objects.

The module also includes helper functions for converting board states to tensors and vice versa.

Note: This module is part of the chipiron package.

class chipiron.players.boardevaluators.neural_networks.input_converters.factory.CreateFromBoard(*args, **kwargs)[source]

Bases: Protocol, Generic

class chipiron.players.boardevaluators.neural_networks.input_converters.factory.CreateFromBoardAndFromParent(*args, **kwargs)[source]

Bases: Protocol, Generic

class chipiron.players.boardevaluators.neural_networks.input_converters.factory.RepresentationFactory(create_from_board: CreateFromBoard, create_from_board_and_from_parent: CreateFromBoardAndFromParent)[source]

Bases: Generic

Factory class for creating instances of BoardFactory

create_from_board: CreateFromBoard
create_from_board_and_from_parent: CreateFromBoardAndFromParent
create_from_transition(tree_node: TreeNode[Any], parent_node_representation: T_BoardRepresentation | None, modifications: BoardModificationP | None) → T_BoardRepresentation[source]

Create a Representation364 object from a transition.

Parameters:

  • tree_node (TreeNode) – The current tree node.

  • parent_node (AlgorithmNode | None) – The parent node of the current tree node.

  • modifications (board_mod.BoardModification | None) – The board modifications.

Returns:

The created Representation364 object.

Return type:

Representation364

chipiron.players.boardevaluators.neural_networks.input_converters.rep_364 module

chipiron.players.boardevaluators.neural_networks.input_converters.rep_364.a(black: Tensor, white: Tensor, piece_map: dict[int, tuple[int, bool]]) → None[source]
chipiron.players.boardevaluators.neural_networks.input_converters.rep_364.b(board: IBoard, castling_black: Tensor, castling_white: Tensor) → None[source]
chipiron.players.boardevaluators.neural_networks.input_converters.rep_364.c(castling_black: Tensor, castling_white: Tensor, black: Tensor, white: Tensor) → Representation364[source]
chipiron.players.boardevaluators.neural_networks.input_converters.rep_364.create_from_board(board: IBoard) → Representation364[source]

Create a Representation364 object from a board.

Parameters:

board (BoardChi) – The chess board.

Returns:

The created Representation364 object.

Return type:

Representation364

chipiron.players.boardevaluators.neural_networks.input_converters.rep_364.create_from_board_and_from_parent(board: IBoard, board_modifications: BoardModificationP, parent_node_board_representation: Representation364) → Representation364[source]

Converts the node, board modifications, and parent node into a tensor representation.

Parameters:

  • board (IBoard) – The current board in the tree.

  • board_modifications (board_mod.BoardModification) – The modifications made to the board.

  • parent_node (AlgorithmNode) – The parent node of the current node.

Returns:

The tensor representation of the node, board modifications, and parent node.

Return type:

Representation364

chipiron.players.boardevaluators.neural_networks.input_converters.rep_364.d() → tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor][source]
chipiron.players.boardevaluators.neural_networks.input_converters.rep_364.e(board: IBoard) → dict[int, tuple[int, bool]][source]

chipiron.players.boardevaluators.neural_networks.input_converters.rep_364_bug module

chipiron.players.boardevaluators.neural_networks.input_converters.rep_364_bug.a(black: Tensor, white: Tensor, piece_map: dict[int, tuple[int, bool]]) → None[source]
chipiron.players.boardevaluators.neural_networks.input_converters.rep_364_bug.b(board: IBoard, castling_black: Tensor, castling_white: Tensor) → None[source]
chipiron.players.boardevaluators.neural_networks.input_converters.rep_364_bug.c(castling_black: Tensor, castling_white: Tensor, black: Tensor, white: Tensor) → Representation364[source]
chipiron.players.boardevaluators.neural_networks.input_converters.rep_364_bug.create_from_board(board: IBoard) → Representation364[source]

Create a Representation364 object from a board.

Parameters:

board (BoardChi) – The chess board.

Returns:

The created Representation364 object.

Return type:

Representation364

chipiron.players.boardevaluators.neural_networks.input_converters.rep_364_bug.create_from_board_and_from_parent(board: IBoard, board_modifications: BoardModificationP, parent_node_board_representation: Representation364) → Representation364[source]

Converts the node, board modifications, and parent node into a tensor representation.

Parameters:

  • board (IBoard) – The current board in the tree.

  • board_modifications (board_mod.BoardModification) – The modifications made to the board.

  • parent_node (AlgorithmNode) – The parent node of the current node.

Returns:

The tensor representation of the node, board modifications, and parent node.

Return type:

Representation364

chipiron.players.boardevaluators.neural_networks.input_converters.rep_364_bug.d() → tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor][source]
chipiron.players.boardevaluators.neural_networks.input_converters.rep_364_bug.e(board: IBoard) → dict[int, tuple[int, bool]][source]

chipiron.players.boardevaluators.neural_networks.input_converters.representation_364_bti module

This module provides a class for converting a chess board into a tensor representation using a 364-dimensional input.

Classes: - Representation364BTI: Converts a chess board into a tensor representation.

class chipiron.players.boardevaluators.neural_networks.input_converters.representation_364_bti.RepresentationBTI(representation_factory: RepresentationFactory[Any])[source]

Bases: object

Converts a chess board into a tensor representation using a 364-dimensional input.

Methods: - __init__: Initializes the Representation364BTI object. - convert: Converts the chess board into a tensor representation.

convert(board: IBoard) → Tensor[source]

Converts the chess board into a tensor representation.

Parameters: - board (BoardChi): The chess board to convert.

Returns: - tensor (torch.Tensor): The tensor representation of the chess board.

chipiron.players.boardevaluators.neural_networks.input_converters.representation_factory_factory module

chipiron.players.boardevaluators.neural_networks.input_converters.representation_factory_factory.create_board_representation_factory(internal_tensor_representation_type: InternalTensorRepresentationType) → RepresentationFactory[Any] | None[source]

Create a board representation based on the given string.

Parameters:

internal_tensor_representation_type (InternalTensorRepresentationType) – The string representing the board representation.

Returns:

The created board representation object, or None if the string is ‘no’.

Return type:

Representation364Factory | None

Raises:

Exception – If the string is not ‘364’.

chipiron.players.boardevaluators.neural_networks.input_converters.test_representation module

Module contents