Is it possible to load a onnx model as a pytorch model directly? Or can you only load it into caffe2?
Now, PyTorch does not support loading ONNX model. Yes, you can load ONNX model into Caffe2.
Is there any plan to implement the ability to load an ONNX model into pytorch? It seems like an exchange format should have that capability.
It is on our roadmap, but it is not high priority. Any help is welcome.
Is there anybody working on this? If yes, the GitHub issue link is much appreciated!
Anything new with this capability ? Is it being supported ?