# -*- coding: utf-8 -*-
"""Implementation of the DistMultLiteralGated model."""
from typing import Any, ClassVar, Mapping, Type
import torch.nn as nn
from .base import LiteralModel
from ...constants import DEFAULT_DROPOUT_HPO_RANGE, DEFAULT_EMBEDDING_HPO_EMBEDDING_DIM_RANGE
from ...nn.combination import GatedCombination
from ...nn.modules import DistMultInteraction, Interaction
from ...triples import TriplesNumericLiteralsFactory
__all__ = [
"DistMultLiteralGated",
]
[docs]class DistMultLiteralGated(LiteralModel):
"""An implementation of the LiteralE model with thhe Gated DistMult interaction from [kristiadi2018]_.
This model is different from :class:`pykeen.models.DistMultLiteral` because it uses a
gate (like found in `LSTMs <https://pytorch.org/docs/stable/generated/torch.nn.LSTM.html>`_)
instead of a LinearDropout module.
This gate implements the full $g$ function described in the LiteralE paper (see equation 4).
---
name: DistMult Literal (Gated)
citation:
author: Kristiadi
year: 2018
link: https://arxiv.org/abs/1802.00934
"""
#: The default strategy for optimizing the model's hyper-parameters
hpo_default: ClassVar[Mapping[str, Any]] = dict(
embedding_dim=DEFAULT_EMBEDDING_HPO_EMBEDDING_DIM_RANGE,
input_dropout=DEFAULT_DROPOUT_HPO_RANGE,
)
#: The default parameters for the default loss function class
loss_default_kwargs: ClassVar[Mapping[str, Any]] = dict(margin=0.0)
interaction_cls: ClassVar[Type[Interaction]] = DistMultInteraction
def __init__(
self,
triples_factory: TriplesNumericLiteralsFactory,
embedding_dim: int = 50,
input_dropout: float = 0.0,
**kwargs,
) -> None:
"""
Initialize the model.
:param triples_factory:
the (training) triples factory
:param embedding_dim:
the embedding dimension
:param input_dropout:
the input dropout, cf. :meth:`DistMultCombination.__init__`
:param kwargs:
additional keyword-based parameters passed to :meth:`LiteralModel.__init__`
"""
super().__init__(
triples_factory=triples_factory,
interaction=self.interaction_cls,
combination=GatedCombination,
combination_kwargs=dict(
entity_embedding_dim=embedding_dim,
literal_embedding_dim=triples_factory.numeric_literals.shape[1],
input_dropout=input_dropout,
),
entity_representations_kwargs=[
dict(
shape=embedding_dim,
initializer=nn.init.xavier_normal_,
),
],
relation_representations_kwargs=[
dict(
shape=embedding_dim,
initializer=nn.init.xavier_normal_,
),
],
**kwargs,
)