Not necessarily. There have been a number of papers trying to emphasize the generalization power of a network rather than "it got a SOTA number", and so leave off fine tuning results. Because that would distract from the point of the paper.