Take the positive value for a primary key in case of duplicates

Question

I have a dataframe df

Key1   Key2    Value
K11    K21      V1
K11    K21     -V1
K13    K23      V2
K13    K23     -V2  

Now for