Heterogeneous Inference Partitioning
Methods for partitioning neural network computational graphs across heterogeneous hardware accelerators to minimize inference latency.
1 papers
Where this stands
The written synthesis of this thread is for subscribers. Subscribe.