Replies: 2 comments 1 reply
-
I recommend looking at the ONNX Script project https://github.com/microsoft/onnxscript for ways to express the operators in the framework you use in ONNX. Assuming you are working with a computational graph, the basic idea would be going through all nodes in the graph and translate each node to its ONNX equivalent; then fill in the rest of the info (weights, inputs etc) to complete the ONNx graph. |
Beta Was this translation helpful? Give feedback.
-
I'm not aware of any converting library for reservoirpy. You may need to write one for this library or you could translate your models into another model based on scikit-learn or torch, both have a dedicated converting library (torch.onnx, sklearn-onnx). |
Beta Was this translation helpful? Give feedback.
-
I have a model built in a framework currently unsupported by ONNX (ReservoirPy). Is there a process for creating the converter library for unsupported frameworks? If not, how may I implement my model in ONNX?
Beta Was this translation helpful? Give feedback.
All reactions