Does Concrete ML still makes a forward mapping to NumPy operators, supported by Concrete-Numpy to use the Concrete stack FHE conversion capabilities?
Are all models which is exported to ONNX referenced by numpy module before being converted to quantized module?
Hello @Rish,
Yes, we are still mapping operators this way, although Concrete-Numpy has been renamed Concrete Python last year and is now part of the Concrete package ! What version onf Concrete ML are you using ? We normally do not mention the use of Concrete Numpy anymore !
Besides, you can find all torch operators supported by Concrete ML in this documentation section.
Hope that answers your question !
1 Like
I was just looking at compilation flow in concrete–ML code , but recent documentation does not mention anything about transpilation of concrete-ML to concrete-numpy, so asked the question.But thankyou for mentioning that it has been renamed.