Config package

This package contains different configuration parameters of the tool. It includes the default values of the hyper-parameters besides the detailed description of the DNN-NILM models of the tool, including the models and the loaders classes.

hparams

The hyper-parameter values of the tool

Liste of main keys in the hyper-parameters dictionnary

Model name

DataLoader

Optuna's parameters

use_optuna

A boolean variable that triggers the hyper-parameter optimization.

n_trials

The number of trials to execute if optuna is used.

Mlflow's parameters

log_artificat

A boolean variable that allow to store the output of the predictions also in the MLFLOW

experiment_label

Model's parameter

dropout

pool_filter

kernel_size

stride

num+layer

max_nb_epochs

batch_size

learning_rate

eps

patient

optimizer

weight_decay

momentum

decay_step

gamma

clip_value

out_size

in_size

appliances

feature_type

main_mu

main_std

input_norm

data_path

logs_path

results_path

figure_path

checkpoints_path

alpha

seed

q_filter

sample_second

multi_task

seq_type

point_position

target_norm

threshold_method

train

kfolds

model_name

mdn_dist_type

num_workers
deep_nilmtk.config.hparams.get_exp_parameters()[source]

Defines the default values for the hyper-parameters of the experiment.

Returns

A dictionnary with values of the hyper-parameters

Return type

dict