pystruct.utils.SaveLogger¶

class pystruct.utils.SaveLogger(file_name, save_every=10, verbose=0)[source]¶

Logging class that stores the model periodically.

Can be used to back up a model during learning. Also a prototype to demonstrate the logging interface.

Parameters:	file_name : string File in which the model will be stored. If the string contains ‘%d’, this will be replaced with the current iteration. save_every : int (default=10) How often the model should be stored (in iterations). verbose : int (default=0) Verbosity level.

Methods

load()	Load the model stoed in file_name and return it.
save(learner, file_name)	Save the model to location specified in file_name.

__init__(file_name, save_every=10, verbose=0)[source]¶

__call__(learner, iteration=0)[source]¶

Save learner if iterations is a multiple of save_every or “final”.

Parameters:	learner : object Learning object to be saved. iteration : int or ‘final’ (default=0) If ‘final’ or save_every % iteration == 0, the model will be saved.

load()[source]¶

Load the model stoed in file_name and return it.

save(learner, file_name)[source]¶

Save the model to location specified in file_name.