Statistically Signifigant Stopping of Neural Network Training
ASWS
This codebase is being actively maintained, please create and issue if you have issues using it
Code to use Augmented Shapiro Wilks Stopping, as well as code for the paper “Statistically Signifigant Stopping of Neural Network Training”
Basics
All data files are included under losses and each folder. The main Augmented Shapiro-Wilk Stopping criterion is implemented in analysis.py, along with several helper functions and wrappers. The other comparison heuristics are also included in analysis.py, along with their wrappers. grapher.py contains all the code for generating the graphs used in the paper, and earlystopping_calculator.py includes code for generating tables and calculating some statistics from the data. hyperparameter_search.py contains all the code used to execute the grid-search on the ASWS method, along with the grid-search for the other heuristics.