The seaborn package helps make matplotlib plots a lot cleaner and allows one to trivially reconfigure the fonts and line thicknesses of a plot for a paper, talk, or poster. It also has a number of useful plot types that do not come with matplotlib by default (e.g. 2d histogram with marginal distributions, KDE plots).
Seaborn is still under active development, and does not add any additional dependencies to our conda environment. The package is quite small: when I pip-installed it just now, the wheel file was ~215kB.
As an example, when I make plots I typically have these three lines after my matplotlib import, resulting in figures with much more prominent fonts and line widths, even ignoring the other improvements seaborn provides: