1 paper · 1 filter
Lord Sen, Shyamapada Mukherjee
The escalating parameter counts in modern deep learning models pose a fundamental challenge to efficient training and resolution of overfitting. We address this by introducing the…