You could use:

df.replace(100, np.nan).style.background_gradient(cmap ='RdYlGn', axis=1)\
                       .highlight_null(color='white')

If the "100" values are of no use.