Structurally Prune Anything: Any Architecture, Any Framework, Any Time

arXiv:2403.18955v1 Announce Type: new
Abstract: Neural network pruning serves as a critical technique for enhancing the efficiency of deep learning models. Unlike unstructured pruning, which only sets specific parameters to zero, structured pruning eliminates entire channels, thus yielding direct computational and storage benefits. However, the diverse patterns for coupling parameters, such as residual connections and group convolutions, the diverse deep learning frameworks, and the various time stages at which pruning can be performed make existing pruning methods less adaptable to …

abstract architecture arxiv benefits channels computational cs.cv cs.lg deep learning diverse efficiency framework however network neural network parameters patterns pruning storage type unstructured