These give the mass estimate inside the fibre. If you want to scale to total masses you need to add the difference between fiber magnitudes and total magnitudes. Since the fit was done to get best overall fit the actual normalisation should be the weighted average of the difference for each filter where the weight is 1/error^2 for each filter. But you almost certainly do not need to do this!
The files are FITS binary tables with keys P2P5, P16, MEDIAN, P84, P97P5, MODE and AVG which returns the 2.5, 16, 50, 84, 97.5 percentiles, the mode and the average of the PDF of Log Mstar respectively.