Amazon Web Services has released the code from its SageMaker Neo machine learning service as the Neo-AI project under the Apache Software License. First launched at AWS re:Invent in November, SageMaker Neo aims to help developers optimize trained machine learning models for targeted hardware platforms.
Also: The Art of the Hybrid Cloud
Similarly, Neo-AI will enable chipmakers, device makers and developers to optimize machine learning models for a wide variety of hardware platforms.
AWS said Neo-AI automatically optimizes models for TensorFlow, MXNet, PyTorch, ONNX, and XGBoost frameworks, and potentially doubles the speed of those models without a loss in accuracy. The platform also converts models into a common format to reduce compatibility issues.
Overall, AWS said the goal is to accelerate machine learning deployments on edge devices.
"Ordinarily, optimizing a machine learning model for multiple hardware platforms is difficult because developers need to tune models manually for each platform's hardware and software configuration," AWS's Sukwon Kim and Vin Sharma write in a blog post. "This is especially challenging for edge devices, which tend to be constrained in compute power and storage."
Also: Top cloud providers 2018: How AWS, Microsoft, Google Cloud Platform, IBM Cloud, Oracle, Alibaba stack up
Intel, Nvidia, and Arm are supporting the first release of Neo-AI, with Xilinx, Cadence, and Qualcomm set to come on board at a later time.