Demystifying the TensorFlow Eager Execution of Deep Learning Inference on a CPU-GPU Tandem | IEEE Conference Publication | IEEE Xplore