Fetching the paper…

FusionStitching: Boosting Execution Efficiency of Memory Intensive Computations for DL Workloads · Around