Hello
You should avoid these functions : it's messy and unclear (get units, tare, multiplier, etc).
What I did and that I recommand you is to make your own calibration as follow :
use only the "read()" function that gives rough value from sensor.
Read the value when no weight on your scale
Then read a value just beneath the maximum weight in your case 900g for example. Verify when you are putting weight on your scale that there is no saturation otherwise take the maximum value below saturation.
With these 2 values you can now calculate the linear regression coefficient in the unit you want.
Then your weight will be like : Weight=A*scale.read()+B; and you'll be able to avoid future tare if you need to.
Regards