Support and tests for unsetting FINN data types.

Xilinx / finn-base

Open Source Compiler Framework using ONNX as Frontend and IR

https://finn-base.readthedocs.io/

BSD 3-Clause "New" or "Revised" License

29 stars 17 forks source link

Support and tests for unsetting FINN data types. #48

Closed heborras closed 3 years ago

heborras commented 3 years ago

Hi, during the QONNX to FINN ONNX conversion I have to unset some of the tensor data types. This happens in particular when the Quant nodes in the activation path get converted to MultiThreshold nodes.

This PR adds functionality to unset a FINN datatype with the following function call: model.set_tensor_datatype(tensor_name, None)

Additionally a new test is included which verifies this behavior.

maltanar commented 3 years ago

Thanks @HenniOVP !