omnibias.keras.stencil

0.0.2a1AlphaPython >=3.10Apache-2.0

Keras 3 operator layers and activation paths implemented with keras.ops for TensorFlow, JAX, or Torch.

import omnibias.keras.stencil

Module contract & mathematical conventions

Bias-collapse stencils (Keras backend).

These build the init *values* (as NumPy arrays) for the K biases and signs of a multi-bias unit. They are plain NumPy because they only run at construction time; the live forward pass uses keras.ops. Mirrors omnibias.torch.stencil.

Shared formulas do not imply universal floating-point identity. Verify dtype, shape, derivative-order support, and the stated numerical assumptions for the path you use.

Module source ↗Source revision 9a2d73df0573Running reference examples →

FUNCTION

central_bias_offsets

omnibias.keras.stencil.central_bias_offsetsSource ↗

central_bias_offsets(K: int, delta: float, dtype: str | None=None) -> np.ndarray

Symmetric offsets ((k - (K+1)/2) * delta)_k=1..K (zero mean).

dtype defaults to keras.config.floatx().

Parameters

Name	Type & default	Contract
`K`positional-or-keyword	`int`Required	See the operation contract and implementation below.
`delta`positional-or-keyword	`float`Required	See the operation contract and implementation below.
`dtype`positional-or-keyword	`str \| None`Default: None	See the operation contract and implementation below.

Returnsnp.ndarray

Read the implementation

def central_bias_offsets(K: int, delta: float, dtype: str | None = None) -> np.ndarray:
    """Symmetric offsets ``((k - (K+1)/2) * delta)_k=1..K`` (zero mean).

    ``dtype`` defaults to :func:`keras.config.floatx()`.
    """
    centre = (K + 1) / 2.0
    return (np.arange(1, K + 1, dtype=_default_dtype(dtype)) - centre) * delta

FUNCTION

central_difference_signs

omnibias.keras.stencil.central_difference_signsSource ↗

central_difference_signs(K: int, delta: float, dtype: str | None=None) -> np.ndarray

Central-difference signs (same magnitudes as forward-difference).

Parameters

Name	Type & default	Contract
`K`positional-or-keyword	`int`Required	See the operation contract and implementation below.
`delta`positional-or-keyword	`float`Required	See the operation contract and implementation below.
`dtype`positional-or-keyword	`str \| None`Default: None	See the operation contract and implementation below.

Returnsnp.ndarray

Read the implementation

def central_difference_signs(
    K: int, delta: float, dtype: str | None = None
) -> np.ndarray:
    """Central-difference signs (same magnitudes as forward-difference)."""
    return forward_difference_signs(K, delta, dtype=dtype)

FUNCTION

forward_bias_offsets

omnibias.keras.stencil.forward_bias_offsetsSource ↗

forward_bias_offsets(K: int, delta: float, dtype: str | None=None) -> np.ndarray

Offsets (0, delta, 2*delta, ..., (K-1)*delta).

dtype defaults to keras.config.floatx().

Parameters

Name	Type & default	Contract
`K`positional-or-keyword	`int`Required	See the operation contract and implementation below.
`delta`positional-or-keyword	`float`Required	See the operation contract and implementation below.
`dtype`positional-or-keyword	`str \| None`Default: None	See the operation contract and implementation below.

Returnsnp.ndarray

Read the implementation

def forward_bias_offsets(K: int, delta: float, dtype: str | None = None) -> np.ndarray:
    """Offsets ``(0, delta, 2*delta, ..., (K-1)*delta)``.

    ``dtype`` defaults to :func:`keras.config.floatx()`.
    """
    return np.arange(K, dtype=_default_dtype(dtype)) * delta

FUNCTION

forward_difference_signs

omnibias.keras.stencil.forward_difference_signsSource ↗

forward_difference_signs(K: int, delta: float, dtype: str | None=None) -> np.ndarray

Forward-difference signs at order K-1 with step delta.

dtype defaults to keras.config.floatx().

Parameters

Name	Type & default	Contract
`K`positional-or-keyword	`int`Required	See the operation contract and implementation below.
`delta`positional-or-keyword	`float`Required	See the operation contract and implementation below.
`dtype`positional-or-keyword	`str \| None`Default: None	See the operation contract and implementation below.

Returnsnp.ndarray

Explicit validation & failure cases

When K < 1, raises ValueError(f'K must be >= 1, got {K}').
When delta <= 0 and K > 1, raises ValueError(f'delta must be > 0 for K > 1 (got delta={delta}, K={K})').

These are explicit source guards. Backend operations may also raise for invalid shapes, dtypes, or unsupported devices.

Read the implementation

def forward_difference_signs(
    K: int, delta: float, dtype: str | None = None
) -> np.ndarray:
    """Forward-difference signs at order ``K-1`` with step ``delta``.

    ``dtype`` defaults to :func:`keras.config.floatx()`.
    """
    if K < 1:
        raise ValueError(f"K must be >= 1, got {K}")
    if delta <= 0 and K > 1:
        raise ValueError(f"delta must be > 0 for K > 1 (got delta={delta}, K={K})")
    dtype = _default_dtype(dtype)
    if K == 1:
        return np.array([1.0], dtype=dtype)
    inv_scale = 1.0 / (delta ** (K - 1))
    signs = [((-1.0) ** (K - 1 - j)) * math.comb(K - 1, j) * inv_scale for j in range(K)]
    return np.array(signs, dtype=dtype)

FUNCTION

identity_signs

omnibias.keras.identity_init.identity_signsSource ↗

identity_signs(K: int, dtype: str | None=None) -> np.ndarray

Signs that satisfy Lemma 1 (identity nesting); sum(s) = 1.

For odd K the alternating pattern (+1, -1, +1, ..., +1) sums to one; for even K we set s[0] = +2 so the sum is one while s[1:] stays strictly alternating.

dtype defaults to keras.config.floatx().

Parameters

Name	Type & default	Contract
`K`positional-or-keyword	`int`Required	See the operation contract and implementation below.
`dtype`positional-or-keyword	`str \| None`Default: None	See the operation contract and implementation below.

Returnsnp.ndarray

Explicit validation & failure cases

When K < 1, raises ValueError(f'K must be >= 1, got {K}').

These are explicit source guards. Backend operations may also raise for invalid shapes, dtypes, or unsupported devices.

Executed regression · grow beyond kmax raises

Recorded coverage confirms this test executed the API body. This example belongs to the pinned upstream test suite. It may use fixtures and imports from its full file. Run it from that checkout with its test dependencies installed.

Full file, imports & fixtures ↗

def test_grow_beyond_kmax_raises() -> None:
    unit = GrowableOMBU(num_channels=2, init_K=1, K_max=2, base="sigmoid")
    with pytest.raises(RuntimeError):
        unit.grow("pair")

python -m pytest "packages/omnibias-keras/tests/test_growable.py::test_grow_beyond_kmax_raises" -q

Read the implementation

def identity_signs(K: int, dtype: str | None = None) -> np.ndarray:
    """Signs that satisfy Lemma 1 (identity nesting); ``sum(s) = 1``.

    For odd K the alternating pattern ``(+1, -1, +1, ..., +1)`` sums to
    one; for even K we set ``s[0] = +2`` so the sum is one while ``s[1:]``
    stays strictly alternating.

    ``dtype`` defaults to :func:`keras.config.floatx()`.
    """
    if K < 1:
        raise ValueError(f"K must be >= 1, got {K}")
    dtype = _default_dtype(dtype)
    s = np.array([(-1.0) ** k for k in range(K)], dtype=dtype)
    if K % 2 == 0:
        s[0] = 2.0
    return s

Other public import paths

omnibias.keras.identity_init.identity_signs
omnibias.keras.stencil.identity_signs

FUNCTION

stencil_offsets

omnibias.keras.stencil.stencil_offsetsSource ↗

stencil_offsets(K: int, delta: float, stencil: Literal['forward', 'central']='central', dtype: str | None=None) -> np.ndarray

Bias offsets for the chosen stencil.

Parameters

Name	Type & default	Contract
`K`positional-or-keyword	`int`Required	See the operation contract and implementation below.
`delta`positional-or-keyword	`float`Required	See the operation contract and implementation below.
`stencil`positional-or-keyword	`Literal['forward', 'central']`Default: 'central'	See the operation contract and implementation below.
`dtype`positional-or-keyword	`str \| None`Default: None	See the operation contract and implementation below.

Returnsnp.ndarray

Read the implementation

def stencil_offsets(
    K: int,
    delta: float,
    stencil: Literal["forward", "central"] = "central",
    dtype: str | None = None,
) -> np.ndarray:
    """Bias offsets for the chosen stencil."""
    if stencil == "forward":
        return forward_bias_offsets(K, delta, dtype=dtype)
    if stencil == "central":
        return central_bias_offsets(K, delta, dtype=dtype)
    raise ValueError(f"Unknown stencil {stencil!r}; expected 'forward' or 'central'.")

FUNCTION

stencil_signs

omnibias.keras.stencil.stencil_signsSource ↗

stencil_signs(K: int, delta: float, stencil: Literal['forward', 'central']='central', dtype: str | None=None) -> np.ndarray

Signs matched to the chosen stencil (same magnitudes for both).

Parameters

Name	Type & default	Contract
`K`positional-or-keyword	`int`Required	See the operation contract and implementation below.
`delta`positional-or-keyword	`float`Required	See the operation contract and implementation below.
`stencil`positional-or-keyword	`Literal['forward', 'central']`Default: 'central'	See the operation contract and implementation below.
`dtype`positional-or-keyword	`str \| None`Default: None	See the operation contract and implementation below.

Returnsnp.ndarray

Explicit validation & failure cases

When stencil not in ('forward', 'central'), raises ValueError(f"Unknown stencil {stencil!r}; expected 'forward' or 'central'.").

These are explicit source guards. Backend operations may also raise for invalid shapes, dtypes, or unsupported devices.

Read the implementation

def stencil_signs(
    K: int,
    delta: float,
    stencil: Literal["forward", "central"] = "central",
    dtype: str | None = None,
) -> np.ndarray:
    """Signs matched to the chosen stencil (same magnitudes for both)."""
    if stencil not in ("forward", "central"):
        raise ValueError(f"Unknown stencil {stencil!r}; expected 'forward' or 'central'.")
    return forward_difference_signs(K, delta, dtype=dtype)

← Package overview Runnable learning examples →

omnibias.keras.stencil

central_bias_offsets​

Parameters

central_difference_signs​

Parameters

forward_bias_offsets​

Parameters

forward_difference_signs​

Parameters

identity_signs​

Parameters

stencil_offsets​

Parameters

stencil_signs​

Parameters

central_bias_offsets

central_difference_signs

forward_bias_offsets

forward_difference_signs

identity_signs

stencil_offsets

stencil_signs