Condensa is a framework for programmable model compression in Python. It comes with a set of built-in compression operators which may be used to compose complex compression schemes targeting specific combinations of DNN, hardware platform, and optimization objective. Common programming abstractions such as conditionals, iteration, and recursion are all natively supported. To recover any accuracy lost during compression, Condensa uses a constrained optimization formulation of model compression and employs an Augmented Lagrangian-based algorithm as the optimizer.
Condensa is under active development, and bug reports, pull requests, and other feedback are all highly appreciated.