How to derive the probability of the predictions in deepsparse? For classification for example

[pacospace]

Additional context
Related-To: AICoE/elyra-aidevsecops-tutorial#297

[markurtz]

HI @pacospace, thank you for the question. Is there any more context available that you could give? If it's related to determinism, DeepSparse is designed to be deterministic for a given model and CPU setup. If this is related to the logits on output for the model then the classification model's ONNX graph will need to be updated to export both logits as well as the final predictions.

Thanks,
Mark

[jeanniefinks]

Hello @pacospace
Just checking in to see if you were able to provide more details or we should close out this issue? Thank you!
Best, Jeannie / Neural Magic

[pacospace]

@markurtz @jeanniefinks I'm sorry for late reply!! This definetly answer my question, so my changes needs to be adjusted on the ONNX graph! Thank you! 🚀