In scipy.lognorm**,** the shape parameter (s) is just the standard deviation of the log-tranformed data.

The shape parameter in scipy.stats.lognorm is indeed the standard deviation of the underlying normal distribution.