tutorials
tutorials copied to clipboard
[principle of operation converting] how to know how onnx optimize the softmax(or a pytorch specific operation)
Question
principle of softmax operation converting
Hi, I am new to onnx.
When I convert my pytorch model to onnx, I find it divided into some operations.
I am curious about the principle of converting an operation.
Really hope to see some explanations considering both in hardware & algorithm perspective.
Further information
- principle of a specific operation converting, where could find the docs or explanation
Is torch.onnx doc enough for your question? https://pytorch.org/docs/stable/onnx.html