Skip to main content
Version: Experimental 🚧

Main class

This library has a maser class called IAFlow, that has functions to management model creation and training. In this section, we will see the CRUD for models.

Constructor​

The constructor of IAFlow class has the following parameters:

  • models_folder: Folder to save the models.
  • params_notifier: Parameters for notifier, see documentation here.
  • builder_function: Function to build the model.
  • checkpoint_params (optional): Parameters for checkpoint, see documentation here.
  • tensorboard_params (optional): Parameters for tensorboard, see documentation here.
IAFlow constructor
from iaflow import IAFlow

def custom_builder(input_shape):
  model = Sequential([
    Dense(units=512, activation='relu'),
    Dense(units=512, activation='relu'),
    Dense(units=1, activation='sigmoid')
  ])
  return model

# Parameters for notifier, see documentation
# https://pypi.org/project/notify-function/#description
params_notifier = {
  'title': 'Training update',
  'webhook_url': os.environ.get('WEBHOOK_URL'),
  'frequency_epoch': 20 # Send a notification every 20 epochs, by default it is every epoch
}

ia_maker = IAFlow(
  models_folder='./models',
  params_notifier=params_notifier,
  builder_function=custom_builder,
  checkpoint_params={
    'monitor': 'val_loss',
    'save_best_only': True,
    'save_weights_only': True
  },
  tensorboard_params={
    'histogram_freq': 1,
    'write_graph': True,
    'write_images': True
  }
)

Update builder function​

The set_builder_function method is used to update the builder function. Has the following parameters:

  • builder_function: New function to build the model.
Example
ia_maker.set_builder_function(custom_builder_2)

Update notifier parameters​

The set_notifier_parameters method is used to update the notifier parameters. Has the following parameters:

  • params_notifier: New parameters for notifier, see documentation here.
Example
ia_maker.set_notifier_parameters(params_notifier_2)