I am running QAT and convert to engine. I render graph of engine, I see that Fake quant layer after Maxpooling is not fused in Maxpooling layer. Is this normal?
Here is a part of graph.
Sorry for the delayed response.
Could you please share the ONNX model (preferably only the part of the ONNX file that recreates the issue).
@spolisetty
Thanks. It is resolved with higher version of TRT engine. I think that older veraion of TRT does not suppory well.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.