Module pysimt.layers.selector

A utility layer that returns a particular element from the previous layer.

"""A utility layer that returns a particular element from the previous layer."""

from torch import nn, Tensor
from typing import Iterable, Any


class Selector(nn.Module):
    """Utility layer that selects and returns a particular element out of
    a tuple. It is useful to select a particular output from the previous layer,
    when used in constructs such as `torch.nn.Sequential()`.

    Args:
        index: The position to select from the given input.

    Example:
        >>> layers = []
        >>> layers.append(torch.nn.GRU(200, 400))
        # By default, GRU returns (output, h_n) but we are not interested in h_n
        >>> layers.append(Selector(0))
        >>> layers.append(torch.nn.Dropout(0.2))
        >>> self.block = nn.Sequential(*layers)
    """
    def __init__(self, index: int):
        """"""
        super().__init__()
        self.index = index

    def forward(self, x: Iterable[Tensor]) -> Tensor:
        """Returns the pre-determined `self.index`'th position of `x`."""
        return x[self.index]

    def __repr__(self):
        return f"Selector(index={self.index})"

Classes

class Selector (index: int)

Utility layer that selects and returns a particular element out of a tuple. It is useful to select a particular output from the previous layer, when used in constructs such as torch.nn.Sequential().

Args

index

The position to select from the given input.

Example

>>> layers = []
>>> layers.append(torch.nn.GRU(200, 400))
# By default, GRU returns (output, h_n) but we are not interested in h_n
>>> layers.append(Selector(0))
>>> layers.append(torch.nn.Dropout(0.2))
>>> self.block = nn.Sequential(*layers)

Expand source code

class Selector(nn.Module):
    """Utility layer that selects and returns a particular element out of
    a tuple. It is useful to select a particular output from the previous layer,
    when used in constructs such as `torch.nn.Sequential()`.

    Args:
        index: The position to select from the given input.

    Example:
        >>> layers = []
        >>> layers.append(torch.nn.GRU(200, 400))
        # By default, GRU returns (output, h_n) but we are not interested in h_n
        >>> layers.append(Selector(0))
        >>> layers.append(torch.nn.Dropout(0.2))
        >>> self.block = nn.Sequential(*layers)
    """
    def __init__(self, index: int):
        """"""
        super().__init__()
        self.index = index

    def forward(self, x: Iterable[Tensor]) -> Tensor:
        """Returns the pre-determined `self.index`'th position of `x`."""
        return x[self.index]

    def __repr__(self):
        return f"Selector(index={self.index})"

Ancestors

• torch.nn.modules.module.Module

Class variables

var dump_patches : bool

var training : bool

Methods

def forward(self, x: Iterable[torch.Tensor]) ‑> torch.Tensor

Returns the pre-determined self.index'th position of x.

Expand source code

def forward(self, x: Iterable[Tensor]) -> Tensor:
    """Returns the pre-determined `self.index`'th position of `x`."""
    return x[self.index]