Fetching the paper…

Provable Benefits of Overparameterization in Model Compression: From Double Descent to Pruning Neural Networks · Around