def StaticVectors(
nO: Optional[int] = None,
vectors: Optional[Floats2d] = None,
*,
column: Optional[int] = None,
dropout: Optional[float] = None,
init_W: Callable = glorot_uniform_init,
) -> Model[InT, OutT]:
attrs: Dict[str, Any] = {"column": column, "vectors": Unserializable(vectors)}
if dropout is not None:
attrs["dropout_rate"] = dropout
model = Model( # type: ignore
"static_vectors",
forward,
init=partial(init, init_W),
params={"W": None},
attrs=attrs,
dims={"nM": None, "nV": None, "nO": nO},
)
if column is not None:
# This is equivalent to array[:, column]. What you're actually doing
# there is passing in a tuple: array[(:, column)], except in the context
# of array indexing, the ":" creates an object slice(0, None).
# So array[:, column] is array.__getitem__(slice(0), column).
model = chain(ints_getitem((slice(0, None), column)), model)
model.attrs["column"] = column
return cast(Model[InT, OutT], model)
def StaticVectors(
nO: Optional[int] = None,
nM: Optional[int] = None,
*,
dropout: Optional[float] = None,
init_W: Callable = glorot_uniform_init,
key_attr: str = "ORTH"
) -> Model[List[Doc], Ragged]:
"""Embed Doc objects with their vocab's vectors table, applying a learned
linear projection to control the dimensionality. If a dropout rate is
specified, the dropout is applied per dimension over the whole batch.
"""
return Model(
"static_vectors",
forward,
init=partial(init, init_W),
params={"W": None},
attrs={"key_attr": key_attr, "dropout_rate": dropout},
dims={"nO": nO, "nM": nM},
)
def configure_test_initializer(b: int = 1) -> Callable[[int], int]:
return partial(test_initializer, b=b)