Frequently Asked questions
- Q: My model isn't compatible with ONNX. What should I do?
A: Not all model types are supported. Check the ONNX documentation for compatibility or consider using a supported alternative model.
- Q: I'm getting shape mismatch errors. How can I fix this?
A: Ensure that the input shape in your ONNX conversion matches the expected input of your model. Double-check the `initial_type` or `dummy_input` in your conversion code.
- Q: The converted model gives different results than my original model. Why?
A: This could be due to precision differences or unsupported operations. Try using a different open source version or check if all operations in your model are supported in ONNX.
