Christopher Rowley shared the benchmark result on his biaryl test set, which showed 1.2.0 is worse than 1.0. in reproducing QM (CCSD(T)*/CBS) torsion barriers

Possible reasons

1. difference in calculating MM torsion barriers

2. 1.0 training set has a larger proportion of biaryl compounds than the new training set

3. energy cutoff might be too small (5 kcal/mol)

• 1. For the possibility of using different ways to calculate mm torsion barriers

• For the possibility that 1.0 training set has a larger proportion of biaryl compounds than the new training set

• All byaryls in Roche set:

• Chris’s set:

• 3. energy cutoff settings
• Joshua Horton (or whoever else) Run the Rowley set (linked below) for QCArchive deposition