PopRT is a high-performance inference framework specifically for Graphcore IPU. It is responsible for deeply optimizing the trained models, generating executable program that can run on the Graphcore IPUs, and performing low-latency, high-throughput inference.
Copyright (c) 2023 Graphcore Ltd
Licensed under the Graphcore End User License Agreement. You can view the Graphcore EULA at https://docs.graphcore.ai/projects/eula