Optimizing DNN Inference on Multi-Accelerator SoCs at Training-time

Re-assign community