ParamSpec¶

class tfsnippet.utils.ParamSpec(*args, **kwargs)¶

Bases: tfsnippet.utils.tensor_spec.TensorSpec

Class to describe the specification for a parameter.

Unlike TensorSpec, the shape of the parameter must be fully determined, i.e., without any unknown dimension, and the ndims must be identical to the specification.

Attributes Summary

`dtype`	Get the data type of the tensor.
`shape`	Get the shape specification.
`value_ndims`	Get the value shape ndims.
`value_shape`	Get the value shape (the shape excluding leading “…”).

Methods Summary

validate(name, x) Validate the input tensor x.

Attributes Documentation

dtype¶

Get the data type of the tensor.

Returns:	The data type, or None.
Return type:	tf.DType or None

shape¶

Get the shape specification.

Returns:	The value shape, or None.
Return type:	tuple[int or str or None] or None

value_ndims¶

Get the value shape ndims.

Returns:	The value shape ndims, or None.
Return type:	int or None

value_shape¶

Get the value shape (the shape excluding leading “…”).

Returns:	The value shape, or None.
Return type:	tuple[int or str or None] or None

Methods Documentation

validate(name, x)¶

Validate the input tensor x.

Parameters:	name (str) – The name of the tensor, used in error messages. x – The input tensor.
Returns:	The validated tensor.
Raises:	ValueError, TypeError – If x cannot pass validation.