@Peperoni, looking at the source code
lux_ = lux_average*calibrtion_to_vernier_lux_;
par_ = lux_average*calibration_to_vernier_par_*measuring_indoor_par_correction_;
where constants for conversion are:
calibrtion_to_vernier_lux_ = 0.78;
calibration_to_vernier_par_ = 0.02;
measuring_indoor_par_correction_ = 0.86; //reduction by 14%
I don't know how accurate this conversion is. One of the ways to find out is to buy PAR meter and compare readings. The problem, though, PAR meters are not cheap.
UPDATE: Giving it more thought, I think that
calibration_to_vernier_par_ was actually calculated by MIT people based on Vernier PAR meter readings and TSL2561 lux reading. So if they did it for exact same LEDs which are on the BOM, than indeed PAR levels are quite low.