Fetching the paper…

FusionStitching: Boosting Memory Intensive Computations for Deep Learning Workloads · Around