gapnet.dnn_modelling.callbacks.SaveStatedictCallback

class documentation

class SaveStatedictCallback(TrackerCallback): (source)

Callback used with fastai when training the neural-network models for saving the statedict of the model after each epoch, given the model improved over the best model so far.

Method	`__init__`	Creates a SaveStatedictCallback object.
Method	`after_epoch`	Function called after each epoch
Class Variable	`order`	Internal Callback ordering. See fastai.callback, for details
Instance Variable	`fname`	Filename for the saved statedict. Filetype is '.statedict'
Instance Variable	`path`	Folder to save the statedict in
Class Variable	`_only_train_loop`	Undocumented

def __init__(self, path='', monitor='valid_loss', comp=None, min_delta=0.0, fname='model', reset_on_fit=True): (source) ¶

Creates a SaveStatedictCallback object.

Output path is:

    os.path.join(path, fname + '.statedict')

Parameters
path	Folder to save the statedict in
monitor	What parameter to monitor
comp	See fastai.callback.tracker.TrackerCallback for details
min_delta	See fastai.callback.tracker.TrackerCallback for details
fname	Filename to use
reset_on_fit	See fastai.callback.tracker.TrackerCallback for details

def after_epoch(self): (source) ¶

Function called after each epoch

order = (source) ¶

Internal Callback ordering. See fastai.callback, for details

fname = (source) ¶

Filename for the saved statedict. Filetype is '.statedict'

path = (source) ¶

Folder to save the statedict in

_only_train_loop: bool = (source) ¶

Undocumented