How to get unique values from multiple columns in a pandas groupby

You can do it with apply:

import numpy as np
g = df.groupby('c')['l1','l2'].apply(lambda x: list(np.unique(x)))