Python seaborn heatmap annotation after masking

I am creating a Seaborn heatmap with a given pandas framework which has numbers and nan. Something like that:

1 2 nan
4 nan nan
nan nan nan

      

        
        
        
      

    

I need a hot map to be annotated and I also need to mask / gray out all nans. I found a mask function and can use it like this:

mask = df.isnull()

      

        
        
        
      

    

Then I can create a heatmap like this:

sns.heatmap(df, mask = mask, annot = True, fmt = "g", cmap = "Blues")

      

        
        
        
      

    

The problem is that with the mask, it seems like the nan are converted to strings while the numbers remain as floats. I am getting the following exception:

ValueError: unknown format code 'g' for object of type 'str'

Something like this happens (partial plot before it throws an exception):

As you can tell, the first formatted cell is fine as it is floating point, but then when it hits the nano that is masked it seems like it is not a string and cannot be formatted.

Does anyone have any suggestions on how to comment on non-oiled data only?

thank