The data contained inside SPAHM.tar are organized in 3 folders corresponding to 3 datasets:

1) 1_QM7/ : the QM7 dataset (M. Rupp, R. Ramakrishnan and O. A. von Lilienfeld, J. Phys. Chem. Lett., 2015, 6, 3309–3313)

   A) xyz/                   : directory with the structures (xyz format)
   B) target/                : directory with the lists of target molecular properties (plain text)
   C) representations_QML/   : directory with the QML representations (NumPy binary)
      * SLATM.npy contains the SLATM representations,
      * CM.npy    contains the Coulomb matrix eigenvalue representations.
   D) representations_SPAHM/ : directory with the SPAHM representations (NumPy binary)
      The file name format corresponds to the "$hamiltonian.basis.npy" pattern, for example,
      * LB.minao.npy contains the SPAHM representation computed with the LB Hamiltonian in the MINAO basis set.
   E) hyperparameters.dat    : hyperparameters used for each regression (plain text/json)


2) 2_QM7-cations : approximately half of the QM7 and the corresponding radical and double cations

   A) xyz/                  : directory with the structures (xyz format)
   B) HOMO_M.dat,
      HOMO_M+.dat,
      HOMO_M++.dat          : HOMO energies for the neutral, radical cations, and double cations
   C) 1_M_and_M++/
      2_M_and_M+/
      3_M_and_M+_and_M++/   : three directories for different modifications of the set
                              of the same structure, described below:
        a) target/                : directory with the list(s) of target molecular properties (plain text)
        b) representations_QML/   : directory with the QML representation(s) (NumPy binary)
        c) representations_SPAHM/ : directory with the SPAHM representation(s) (NumPy binary)
        d) hyperparameters.dat    : hyperparameters used for each regression (plain text/json)


3) 3_L11/: the L11 dataset (D. N. Laikov, J. Chem. Phys., 2011, 135, 134120)

   A) xyz/                   : directory with the structures (xyz format)
   B) target/                : directory with the list(s) of target molecular properties (plain text)
   C) charge.dat             : list of molecular charges (plain text)
   D) spin.dat               : list of numbers of unpaired electrons (plain text)
   E) representations_SPAHM/ : directory with the SPAHM representation(s) (NumPy binary)
   F) hyperparameters.dat    : hyperparameters used for each regression (plain text/json)