xnn.common.data.hub.ani1ccx.ANI1ccxBuilder#

class xnn.common.data.hub.ani1ccx.ANI1ccxBuilder[source]#

Bases: ANI1xBuilder

Builder for the ANI-1ccx data set (CCSD(T)*/CBS energies, no forces).

See the module docstring for the dataset description and citation. All download, NaN-masking, capping, and splitting machinery is inherited from ANI1xBuilder; this class only pins the level of theory to "ccsd(t)_cbs" and shares the cached release file.

load(*, split=None, cache_dir, units='eV', max_molecules=None, max_conformations=None, seed=1234, quiet=False)[source]#

Download and preprocess the ANI-1ccx data set.

Parameters:

  • split (str or None) – None returns {"all": ...}; "train" / "val" / "test" returns a per-conformation 80/10/10 partition (fixed seed, disjoint splits).

  • cache_dir (pathlib.Path) – Base cache directory; the release file is shared with cache_dir/"ani1x".

  • units (str, optional) – "eV" (default) converts energies to eV; "hartree" keeps the raw upstream values.

  • max_molecules (int, optional) – Cap the number of molecule groups read (useful for demos).

  • max_conformations (int, optional) – Cap the number of coupled-cluster conformations kept per molecule.

  • seed (int, optional) – Seed for the reproducible 80/10/10 split. Defaults to 1234.

  • quiet (bool, optional) – Suppress progress output. Defaults to False.

Returns:

dict of {str – Structure dicts with keys pos (N, 3), atomic_numbers (N,) and energy (scalar). CCSD(T)*/CBS carries no forces.

Return type:

list of dict} or list of dict

Raises:
On this page
Show Source