cerebras.modelzoo.common.registry.Registry#

class cerebras.modelzoo.common.registry.Registry[source]#

Bases: object

Methods

`get_config_class`
`get_loss_class`
`get_model_class`
`get_path`
`get_run_path`
`list_dataset`
`list_datasetprocessor`
`list_loss`
`list_lr_scheduler`
`list_models`
`register_config`	This method is added to register config classes
`register_dataset`	This method is added to register dataset
`register_datasetprocessor`	This method is added to register datasetprocessor
`register_loss`	This method is added to register loss
`register_lr_scheduler`	This method is added to register lr_schedular
`register_model`	This method is added to register models
`register_paths`	This method is register paths useful for the user
`register_run_path`	Look for run path for the model
`unregister`	This method is added to unregister

Attributes

mapping

classmethod register_model(model_name, datasetprocessor=[], dataset=[])[source]#: This method is added to register models

classmethod register_datasetprocessor(name)[source]#: This method is added to register datasetprocessor

classmethod register_loss(name)[source]#: This method is added to register loss

classmethod register_lr_scheduler(name)[source]#: This method is added to register lr_schedular

classmethod register_dataset(name)[source]#: This method is added to register dataset

classmethod register_paths(kind, path)[source]#: This method is register paths useful for the user

classmethod register_config(name)[source]#: This method is added to register config classes

classmethod register_run_path(name)[source]#: Look for run path for the model

classmethod unregister(region, name)[source]#

This method is added to unregister

region can be [‘model’, ‘loss’, ‘lr_scheduler’,: ‘datasetprocessor’, ‘dataset’]

cerebras.modelzoo.common.registry

cerebras.modelzoo.common.registry_cli