WeightWatcher: A diagnostic tool for deep neural networks

Описание: WeightWatcher (WW): is an open-source, diagnostic tool for analyzing Deep Neural Networks (DNN), without needing access to training or even test data. It can be used to: analyze pre/trained DNNs to gauge model performance, predict trends in test accuracies, and detect potential problems including over-training, over-parameterization, Scale-Collapse, and Correlation-traps.