mode

Sparsity mode descriptor.

Classes

`ExportSparseModeDescriptor`	Class to describe the `"export_sparse"` mode.
`SparseGPTModeDescriptor`	Class to define and describe sparsification based on SparseGPT.
`SparseMagnitudeModeDescriptor`	Class to define and describe magnitude-based sparsification.

Functions

`convert_sparse_model`	Function for converting a model to a sparsity meta-model.
`export_sparse`	Export a sparse model to a regular model.
`restore_export_sparse`	Restore & export a sparse model to a regular model.
`restore_sparse_model`	Function for restoring a previously convert model to a sparsity meta-model.
`update_sparse_metadata`	Update subnet config to current subnet config of model.

class ExportSparseModeDescriptor

Bases: _ModeDescriptor

Class to describe the "export_sparse" mode.

The properties of this mode can be inspected via the source code.

property config_class: Type[ModeloptBaseConfig]: Specifies the config class for the mode.

property convert: Callable[[Module, ModeloptBaseConfig], Tuple[Module, Dict[str, Any]]]: The mode’s entrypoint for converting a model.

property is_export_mode: bool: Specifies if this mode is an export mode.

property name: str: Returns the value (str representation) of the mode.

property restore: Callable[[Module, ModeloptBaseConfig, Dict[str, Any]], Module]: The mode’s entrypoint for restoring a model.

class SparseGPTModeDescriptor

Bases: SparseMagnitudeModeDescriptor

Class to define and describe sparsification based on SparseGPT.

property config_class: Type[ModeloptBaseConfig]: Specifies the config class for the mode.

property name: str: Returns the name of the mode.

property search_algorithm: Type[BaseSearcher]: Specifies the search algorithm for the mode.

class SparseMagnitudeModeDescriptor

Bases: _ModeDescriptor

Class to define and describe magnitude-based sparsification.

property config_class: Type[ModeloptBaseConfig]: Specifies the config class for the mode.

property convert: Callable[[Module, ModeloptBaseConfig], Tuple[Module, Dict[str, Any]]]: The mode’s entrypoint for converting a model.

property export_mode: str | None: The mode that corresponds to the export mode of this mode.

property name: str: Returns the name of the mode.

property next_modes: Set[str] | None: Specifies the next modes for the mode.

property restore: Callable[[Module, ModeloptBaseConfig, Dict[str, Any]], Module]: The mode’s entrypoint for restoring a model.

property search_algorithm: Type[BaseSearcher]: Specifies the search algorithm for the mode.

property update_for_new_mode: Callable[[Module, ModeloptBaseConfig, Dict[str, Any]], None]: The mode’s entrypoint for updating the models metadata.

property update_for_save: Callable[[Module, ModeloptBaseConfig, Dict[str, Any]], None]: The mode’s entrypoint for updating the models metadata.

convert_sparse_model(model, config)

Function for converting a model to a sparsity meta-model.

Parameters:

model (Module) –
config (ModeloptBaseConfig) –

Return type:

Tuple[Module, Dict[str, Any]]

export_sparse(model, config)

Export a sparse model to a regular model.

Parameters:

model (Module) –
config (ExportSparseConfig) –

Return type:

Tuple[Module, Dict[str, Any]]

restore_export_sparse(model, config, metadata)

Restore & export a sparse model to a regular model.

Parameters:

model (Module) –
config (ExportSparseConfig) –
metadata (Dict[str, Any]) –

Return type:

Module

restore_sparse_model(model, config, metadata)

Function for restoring a previously convert model to a sparsity meta-model.

Parameters:

model (Module) –
config (ModeloptBaseConfig) –
metadata (Dict[str, Any]) –

Return type:

Module

update_sparse_metadata(model, config, metadata)

Update subnet config to current subnet config of model.

Parameters:

model (Module) –
config (ModeloptBaseConfig) –
metadata (Dict[str, Any]) –

Return type:

None