FlashLLM

Researchers propose a GPU inference method for large language models with moderate unstructured sparsity, using a three-layer matrix storage format and a SpMM kernel that jointly utilizes sparse tenso…