Discrete Profile

class confopt.profile.DiscreteProfile(searchspace_type: str | SearchSpaceType, domain: str | None = None, **kwargs: Any)

Bases: object

_initializa_genotype() → None

Initializes the genotype for the discrete profile. This method sets the genotype of the best DARTS supernet found after 50 epochs of running the DARTS optimizer. Thus it is only a valid genotype if the search space is of type DARTS.

Parameters:: None
Returns:: None

_initialize_trainer_config() → None

Initializes the trainer configuration for the discrete profile. This method sets the default training configuration parameters for the discrete model.

Parameters:: None
Returns:: None

configure_extra(**config: Any) → None

Configure any extra settings for the supernet. Could be useful for tracking Weights & Biases metadata.

Parameters:: **config – Arbitrary keyword arguments.
Returns:: None

configure_searchspace(**config: Any) → None

Configure the search space for the supernet.

Parameters:: **config – Arbitrary keyword arguments. Possible depend on the the search space type. For more information please check the Parameters of the supernet of each search space.
Returns:: None

configure_trainer(**kwargs) → None

Configure the trainer component for the supernet.

Parameters:

**kwargs –

Arbitrary keyword arguments. Possible keys include:

lr (float): Learning rate for the optimizer.

epochs (int): Number of training epochs.

optim (str): Optimizer type. Can be ‘sgd’, ‘adam’, etc.

optim_config (dict): Additional configuration for the optimizer.

…

Raises:

AssertionError – If any of the provided configuration keys are not valid.

Returns:

None

get_extra_config() → dict

This method returns the extra configuration parameters for the discrete profile for example could be used for Weights & Biases metadata.

Parameters:: None
Returns:: dict – A dictionary containing extra configuration parameters.

get_genotype() → str

Returns the genotype of the discrete profile.

Parameters:: None
Returns:: str – The genotype of the discrete profile.

get_run_description() → str

This method returns a string description of the run configuration. The description is used for tracking purposes in Weights & Biases.

Parameters:: None
Returns:: str – A string describing the run configuration.

get_searchspace_config(dataset_str: str) → dict

Returns the search space configuration based on the search space type.

Parameters:

dataset_str (str) – The dataset string.

Raises:

ValueError – If the search space is not any of the valid values of
NasBench201, DARTS, NasBench1Shot1 or TransNasBench101. –

Returns:

dict – A dictionary containing the search space configuration which would be passed to as the arguments for creating the new discrete model object.

get_trainer_config() → dict

Returns the trainer configuration for the discrete profile.

Parameters:: None
Returns:: dict – A dictionary containing the trainer configuration.

set_search_space_config(config: dict) → None

Sets the search space configuration for the discrete profile. This method allows the user to set custom configuration parameters.

Parameters:: config (dict) – A dictionary containing the search space configuration.
Returns:: None