very slow in inference

[meloxuan]

Hello, I am using torch. Onnx. Export to convert the trained PTH model to ONNX, but I find that the reasoning time is obviously slow

[HectorSVC]

Could you provide more details? Is the converter slow or inference slow? What's the version you used? Could you share your model?

[meloxuan]

Could you provide more details? Is the converter slow or inference slow? What's the version you used? Could you share your model?

Thank you for your reply, the onxx is extremely slow in inference, up to 1.2s, as a comparison, using the original .pth file is only 60ms, I have shared PyTorch2Onnx code, and My ONNX is being uploaded. looking forward to your reply.
PyTorch2Onnx: https://drive.google.com/file/d/1LIJAtZtIemf5lTC5fApmOZF0Dr6qHnkB/view?usp=sharing

[meloxuan]

Could you provide more details? Is the converter slow or inference slow? What's the version you used? Could you share your model?
ONNX model: https://drive.google.com/file/d/1fhqnh8qNL7wnXRaq1byQK_87TWDwjES2/view?usp=sharing

[microsoft-github-policy-service]

Applying stale label due to no activity in 30 days