xlstm_jax.models.xlstm_pytorch.components.util#
Classes#
This class helps keeping parameters in a specialized internal structure to be optimal for computation speed, while |
Functions#
|
|
|
A higher-order decorator that applies 'decorator' only if 'condition' is True. |
Module Contents#
- xlstm_jax.models.xlstm_pytorch.components.util.round_to_multiple(n, m=8)#
- xlstm_jax.models.xlstm_pytorch.components.util.conditional_decorator(condition, decorator)#
A higher-order decorator that applies ‘decorator’ only if ‘condition’ is True.
- class xlstm_jax.models.xlstm_pytorch.components.util.ParameterProxy(module, parameter_name, internal_to_external, external_to_internal)#
This class helps keeping parameters in a specialized internal structure to be optimal for computation speed, while having a proxied version to be called externally that is backend-agnostic.
It takes a module and a parameter name of a parameter in that module it represents. Via __setitem__ and __getitem__ the “external”
- Parameters:
internal_to_external (collections.abc.Callable[[torch.Tensor], torch.Tensor])
external_to_internal (collections.abc.Callable[[torch.Tensor], torch.Tensor])
- module#
- parameter_name#
- internal_to_external#
- external_to_internal#
- clone()#
- property shape#
- property ndim#
- property grad#