Optimizing LibTorch-based inference engine memory usage and thread-pooling Outline OpenTeams June 29, 2023