def hardshrink(
x: types.TensorLike, lower: Number = -0.5, upper: Number = 0.5
) -> tf.Tensor:
"""Hard shrink function.
Computes hard shrink function:
`x if x < lower or x > upper else 0`.
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
lower: `float`, lower bound for setting values to zeros.
upper: `float`, upper bound for setting values to zeros.
Returns:
A `Tensor`. Has the same type as `x`.
"""
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _hardshrink_custom_op(x, lower, upper)
except tf.errors.NotFoundError:
options.warn_fallback("hardshrink")
return _hardshrink_py(x, lower, upper)
def lisht(x: types.TensorLike) -> tf.Tensor:
r"""LiSHT: Non-Parameteric Linearly Scaled Hyperbolic Tangent Activation Function.
Computes linearly scaled hyperbolic tangent (LiSHT):
$$
\mathrm{lisht}(x) = x * \tanh(x).
$$
See [LiSHT: Non-Parameteric Linearly Scaled Hyperbolic Tangent Activation Function for Neural Networks](https://arxiv.org/abs/1901.05894).
Usage:
>>> x = tf.constant([1.0, 0.0, 1.0])
>>> tfa.activations.lisht(x)
<tf.Tensor: shape=(3,), dtype=float32, numpy=array([0.7615942, 0. , 0.7615942], dtype=float32)>
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
Returns:
A `Tensor`. Has the same type as `x`.
"""
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _lisht_custom_op(x)
except tf.errors.NotFoundError:
options.warn_fallback("lisht")
return _lisht_py(x)
def gelu(x: types.TensorLike, approximate: bool = True) -> tf.Tensor:
"""Gaussian Error Linear Unit.
Computes gaussian error linear:
`0.5 * x * (1 + tanh(sqrt(2 / pi) * (x + 0.044715 * x^3)))` or
`x * P(X <= x) = 0.5 * x * (1 + erf(x / sqrt(2)))`, where P(X) ~ N(0, 1),
depending on whether approximation is enabled.
See [Gaussian Error Linear Units (GELUs)](https://arxiv.org/abs/1606.08415)
and [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805).
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
approximate: bool, whether to enable approximation.
Returns:
A `Tensor`. Has the same type as `x`.
"""
warnings.warn(
"gelu activation has been migrated to core TensorFlow, "
"and will be deprecated in Addons 0.12.",
DeprecationWarning,
)
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _gelu_custom_op(x, approximate)
except tf.errors.NotFoundError:
options.warn_fallback("gelu")
return _gelu_py(x, approximate)
def mish(x: types.TensorLike) -> tf.Tensor:
r"""Mish: A Self Regularized Non-Monotonic Neural Activation Function.
Computes mish activation:
$$
\mathrm{mish}(x) = x \cdot \tanh(\mathrm{softplus}(x)).
$$
See [Mish: A Self Regularized Non-Monotonic Neural Activation Function](https://arxiv.org/abs/1908.08681).
Usage:
>>> x = tf.constant([1.0, 0.0, 1.0])
>>> tfa.activations.mish(x)
<tf.Tensor: shape=(3,), dtype=float32, numpy=array([0.86509836, 0. , 0.86509836], dtype=float32)>
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
Returns:
A `Tensor`. Has the same type as `x`.
"""
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _mish_custom_op(x)
except tf.errors.NotFoundError:
options.warn_fallback("mish")
return _mish_py(x)
def tanhshrink(x: types.TensorLike) -> tf.Tensor:
r"""Tanh shrink function.
Applies the element-wise function:
$$
\mathrm{tanhshrink}(x) = x - \tanh(x).
$$
Usage:
>>> x = tf.constant([-1.0, 0.0, 1.0])
>>> tfa.activations.tanhshrink(x)
<tf.Tensor: shape=(3,), dtype=float32, numpy=array([-0.23840582, 0. , 0.23840582], dtype=float32)>
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
Returns:
A `Tensor`. Has the same type as `x`.
"""
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _tanhshrink_custom_op(x)
except tf.errors.NotFoundError:
options.warn_fallback("tanhshrink")
return _tanhshrink_py(x)
def gelu(x: types.TensorLike, approximate: bool = True) -> tf.Tensor:
r"""Gaussian Error Linear Unit.
Computes gaussian error linear:
$$
\mathrm{gelu}(x) = x \Phi(x),
$$
where
$$
\Phi(x) = \frac{1}{2} \left[ 1 + \mathrm{erf}(\frac{x}{\sqrt{2}}) \right]$
$$
when `approximate` is `False`; or
$$
\Phi(x) = \frac{x}{2} \left[ 1 + \tanh(\sqrt{\frac{2}{\pi}} \cdot (x + 0.044715 \cdot x^3)) \right]
$$
when `approximate` is `True`.
See [Gaussian Error Linear Units (GELUs)](https://arxiv.org/abs/1606.08415)
and [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://arxiv.org/abs/1810.04805).
Usage:
>>> tfa.options.TF_ADDONS_PY_OPS = True
>>> x = tf.constant([-1.0, 0.0, 1.0])
>>> tfa.activations.gelu(x, approximate=False)
<tf.Tensor: shape=(3,), dtype=float32, numpy=array([-0.15865529, 0. , 0.8413447 ], dtype=float32)>
>>> tfa.activations.gelu(x, approximate=True)
<tf.Tensor: shape=(3,), dtype=float32, numpy=array([-0.158808, 0. , 0.841192], dtype=float32)>
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
approximate: bool, whether to enable approximation.
Returns:
A `Tensor`. Has the same type as `x`.
"""
warnings.warn(
"gelu activation has been migrated to core TensorFlow, "
"and will be deprecated in Addons 0.12.",
DeprecationWarning,
)
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _gelu_custom_op(x, approximate)
except tf.errors.NotFoundError:
options.warn_fallback("gelu")
return _gelu_py(x, approximate)
def tanhshrink(x: types.TensorLike) -> tf.Tensor:
"""Applies the element-wise function: x - tanh(x)
Args:
features: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
Returns:
A `Tensor`. Has the same type as `features`.
"""
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _tanhshrink_custom_op(x)
except tf.errors.NotFoundError:
options.warn_fallback("tanhshrink")
return _tanhshrink_py(x)
def mish(x: types.TensorLike) -> tf.Tensor:
"""Mish: A Self Regularized Non-Monotonic Neural Activation Function.
Computes mish activation: x * tanh(softplus(x))
See [Mish: A Self Regularized Non-Monotonic Neural Activation Function](https://arxiv.org/abs/1908.08681).
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
Returns:
A `Tensor`. Has the same type as `x`.
"""
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _mish_custom_op(x)
except tf.errors.NotFoundError:
options.warn_fallback("mish")
return _mish_py(x)
def softshrink(
x: types.TensorLike, lower: Number = -0.5, upper: Number = 0.5
) -> tf.Tensor:
r"""Soft shrink function.
Computes soft shrink function:
$$
\mathrm{softshrink}(x) =
\begin{cases}
x - \mathrm{lower} & \text{if } x < \mathrm{lower} \\
x - \mathrm{upper} & \text{if } x > \mathrm{upper} \\
0 & \text{otherwise}
\end{cases}.
$$
Usage:
>>> x = tf.constant([-1.0, 0.0, 1.0])
>>> tfa.activations.softshrink(x)
<tf.Tensor: shape=(3,), dtype=float32, numpy=array([-0.5, 0. , 0.5], dtype=float32)>
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
lower: `float`, lower bound for setting values to zeros.
upper: `float`, upper bound for setting values to zeros.
Returns:
A `Tensor`. Has the same type as `x`.
"""
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _softshrink_custom_op(x, lower, upper)
except tf.errors.NotFoundError:
options.warn_fallback("softshrink")
return _softshrink_py(x, lower, upper)
def lisht(x: types.TensorLike) -> tf.Tensor:
"""LiSHT: Non-Parameteric Linearly Scaled Hyperbolic Tangent Activation Function.
Computes linearly scaled hyperbolic tangent (LiSHT): `x * tanh(x)`
See [LiSHT: Non-Parameteric Linearly Scaled Hyperbolic Tangent Activation Function for Neural Networks](https://arxiv.org/abs/1901.05894).
Args:
x: A `Tensor`. Must be one of the following types:
`float16`, `float32`, `float64`.
Returns:
A `Tensor`. Has the same type as `x`.
"""
x = tf.convert_to_tensor(x)
if not options.TF_ADDONS_PY_OPS:
try:
return _lisht_custom_op(x)
except tf.errors.NotFoundError:
options.warn_fallback("lisht")
return _lisht_py(x)
def adjust_hsv_in_yiq(
image: TensorLike,
delta_hue: Number = 0,
scale_saturation: Number = 1,
scale_value: Number = 1,
name: Optional[str] = None,
) -> tf.Tensor:
"""Adjust hue, saturation, value of an RGB image in YIQ color space.
This is a convenience method that converts an RGB image to float
representation, converts it to YIQ, rotates the color around the
Y channel by delta_hue in radians, scales the chrominance channels
(I, Q) by scale_saturation, scales all channels (Y, I, Q) by scale_value,
converts back to RGB, and then back to the original data type.
`image` is an RGB image. The image hue is adjusted by converting the
image to YIQ, rotating around the luminance channel (Y) by
`delta_hue` in radians, multiplying the chrominance channels (I, Q) by
`scale_saturation`, and multiplying all channels (Y, I, Q) by
`scale_value`. The image is then converted back to RGB.
Args:
image: RGB image or images. Size of the last dimension must be 3.
delta_hue: `float`, the hue rotation amount, in radians.
scale_saturation: `float`, factor to multiply the saturation by.
scale_value: `float`, factor to multiply the value by.
name: A name for this operation (optional).
Returns:
Adjusted image(s), same shape and dtype as `image`.
"""
with tf.name_scope(name or "adjust_hsv_in_yiq"):
image = tf.convert_to_tensor(image, name="image")
# Remember original dtype to so we can convert back if needed
orig_dtype = image.dtype
if not image.dtype.is_floating:
image = tf.image.convert_image_dtype(image, tf.float32)
delta_hue = tf.cast(delta_hue, dtype=image.dtype, name="delta_hue")
scale_saturation = tf.cast(scale_saturation,
dtype=image.dtype,
name="scale_saturation")
scale_value = tf.cast(scale_value,
dtype=image.dtype,
name="scale_value")
if not options.is_custom_kernel_disabled():
warnings.warn(
"C++/CUDA kernel of `adjust_hsv_in_yiq` will be removed in Addons `0.13`.",
DeprecationWarning,
)
try:
image = _distort_image_so.ops.addons_adjust_hsv_in_yiq(
image, delta_hue, scale_saturation, scale_value)
except tf.errors.NotFoundError:
options.warn_fallback("adjust_hsv_in_yiq")
image = _adjust_hsv_in_yiq(image, delta_hue, scale_saturation,
scale_value)
else:
image = _adjust_hsv_in_yiq(image, delta_hue, scale_saturation,
scale_value)
return tf.image.convert_image_dtype(image, orig_dtype)
