bijx.Const

class bijx.Const

Bases: Variable

Mark a variable as constant during training.

This variable type explicitly indicates that a parameter should remain fixed during optimization. Useful for freezing parts of models or storing hyperparameters that should not be updated.

Can be used with FrozenFilter to selectively freeze parameters during gradient-based optimization.

Example

>>> bijection = SomeBijection()
>>> bijection.scale = Const(1.0)  # Fix scale parameter
>>> # standard optimizers will not apply gradients to scale

__init__(value, *, use_ref=None, **metadata)

Parameters:

• value (A | VariableMetadata[A])

• use_ref (bool | None)

• metadata (Any)

Methods

add_axis(axis_index, axis_name)
copy_from(other)
create_value(value)
from_metadata(value, attributes)
get_metadata([name])	Get metadata for the Variable.
remove_axis(axis_index, axis_name)
replace([value])
set_metadata(*args, **kwargs)	Set metadata for the Variable.
to_state()
update_from_state(variable_state)

Attributes

raw_value
has_ref
type	The type of the variable.
value

raw_value: A