A plans to add ONNX support to TD?

Hi all,

I was wondering if there will be any support for onnx so that we can train ML models in python and load them to TD? It is probably possible using the C++ DIY approach, but a in-box support would probably save a lot of time. Any thoughts on this?


I found that there is a new script top now. It is amazing! Answering my own question, I could just write a python script to read onnx files.