Native Model Authoring for Edge with LiteRT Tensor API

Meet the LiteRT Tensor API: giving developers native C++ control over the model graph to unify end-to-end ML computations on heterogeneous edge hardware.