tinyML Talks local Webcast – recorded January 19, 2021
“microTVM: a Tensor Compiler for Bare Meta”
Andrew Reusch – OctoML
TVM is an open source compiler and deployment framework for Tensor-oriented models. It can compile models from popular frameworks (PyTorch, TensorFlow, ONNX, etc) onto a range of devices. microTVM is a recent effort that allows TVM to target bare metal devices. In this talk, we’ll explain how microTVM works and give a short demonstration. We’ll also discuss future directions for microTVM and ways you can get involved.