Hello everyone, I hope you’re well!!!
I have a serious question about converting an ANN into an SNN. At first, I verificate that my network (see attached code) was not fully converted, and this information was omitted when I put the following code snippet:
warnings.simplefilter("ignore")
tf.get_logger().addFilter(lambda rec: "Tracing is expensive" not in rec.msg)
When I do the verification using “assert converter.verify()”, after the conversion with “nengo_dl.Converter()”, it is verified that parts of the network (if not the whole network) have been considered as Tensornodes, which is indicated by the warning message “Falling back to Tensornodes”.
Can anyone tell me how to do this complete conversion and explain why this happens for my architecture network?
forum_nengo_vzanon.ipynb (12.9 KB)