r2SCAN calculations consistently underestimating energy

[rmz4ed]

I am trying to run r2SCAN calculations for the first time. I ran SCAN calculations on the same structures and they result in almost identical total energies as those in Materials Project. However, when I try to run r2SCAN calculations by changing METAGGA = SCAN to METAGGA = r2SCAN, I get total energy values that are always significantly more positive by several eV than those in Materials Project. I am wondering whether my settings are correct or if I am making some error, so any advice is appreciated! I am attaching my INCAR file here:

ISMEAR = -5
SIGMA = 0.2
NSW = 100
IBRION = 2
ISIF = 3
EDIFFG = -1E-4
PREC = Accurate
ENCUT = 520
NCORE = 20
METAGGA = R2SCAN
ALGO = ALL
LASPH = .TRUE.
LDIAG = .FALSE.
LWAVE = .FALSE.
LCHARG = .FALSE.

[marie-therese.huebsch]

The Materials Project share the input they used to produce the data. I think it is best to compare your input with theirs to see the difference.

Best regards,
Marie-Therese

Ps: You should not do structure optimization (IBRION=2, NSW>0) in combination with ISMEAR=-5. There is a big warning in the standard output. But that is true for either SCAN and R2SCAN.