If you have Hifi reads your error rate should already be quite good, therefore polishing with potentially longer and slightly more error-prone Nanonopore reads might not do any good. You might rather try to improve contingency. Do you have Hi-C data available? If you already have the Nanopore reads, I would try to use the Nanopore reads for gap-closure and scaffolding. Possibly, generate a hybrid assembly using both technologies to complement each other e.g. with the Flye assembler.