Hugging Face's logo

Datasets:
andreasburger
/
Hessians-LoT

Tags:
chemistry
molecular-hessians
hdf5
omol
quantum-chemistry
License:
Dataset card Files
xet
 Community
Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 49, in _split_generators
                  import h5py
              ModuleNotFoundError: No module named 'h5py'
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 71, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Hessians-LoT

This dataset contains recalculated Hessians for selected molecules from the THEMol Hessian data, using the hessian_0.h5 source shard. The files here are the Hessians calculated so far for the comparison study in THEMol; the full original THEMol source HDF5 shards are not included in this upload.

Contents

  • HDF5 Hessian files: 282
  • Unique source samples: 57
  • Atom count range: 3 to 50 atoms
  • Total uploaded HDF5 size: 34.68 MB
  • Manifest: metadata/manifest.csv
  • Method summary: metadata/method_summary.csv

File Layout

Files are organized by calculation engine and method:

hessians/
  ml/uma_m_1p1_omol/
  ml/uma_s_1p2_omol/
  orca/wb97m_d4_def2_qzvppd/
  orca/wb97m_d4_def2_tzvpd/
  orca/wb97x_6_31gd/
  pyscf/ccsd_t_cbs/
  pyscf/wb97m_v_def2_tzvpd/
  pyscf/wb97m_v_def2_tzvpd_density_fit/
  semiempirical/g_xtb/
  semiempirical/gfn2_xtb/
metadata/
  manifest.csv
  method_summary.csv
  summary.json

Each Hessian file is named hessian_0_sample_<zero-padded sample id>.h5. The path and original_path columns in metadata/manifest.csv map uploaded names back to the local calculation outputs.

HDF5 Schema

Each .h5 file contains one molecule group keyed by the original THEMol sample key. Within that group, the common datasets are:

  • atomic_numbers: atomic numbers, shape (N, 1)
  • coords: coordinates, shape (N, 3)
  • energy: scalar energy from the method in that file
  • force: force array, shape (N, 3)
  • gradient: gradient array, shape (N, 3)
  • hessian: flattened Cartesian Hessian, shape (3N, 3N)
  • hessian_4d: Hessian as atom/axis blocks, shape (N, N, 3, 3) when present
  • reference_hessian: source THEMol Hessian copied from hessian_0.h5 when present
  • metadata_json: calculation metadata, including source sample id and method details

UMA-S-1.2 and UMA-M-1.1 predictions (energy, gradient, force, and hessian) are stored as float32, matching the precision used for inference and autograd. Coordinates and copied reference Hessians remain float64. The upload includes the same 37 source samples for both UMA checkpoints.

Calculation attributes use spin_difference = N_alpha - N_beta and spin_multiplicity = spin_difference + 1; the legacy spin attribute follows the spin_difference convention. basis_overrides in the manifest records per-file basis substitutions such as the Br override used by wB97X/6-31G(d).

Methods

  • g_xtb: 40 files, g-xTB
  • gfn2_xtb: 10 files, GFN2-xTB
  • orca_wb97m_d4_def2_qzvppd: 37 files, ORCA wB97M-D4/def2-QZVPPD
  • orca_wb97m_d4_def2_tzvpd: 45 files, ORCA wB97M-D4/def2-TZVPD
  • orca_wb97x_6_31gd: 37 files, ORCA wB97X/6-31G(d)
  • pyscf_ccsd_t_cbs: 2 files, PySCF RCCSD(T)/CBS
  • pyscf_wb97m_v_def2_tzvpd: 28 files, PySCF wB97M-V/def2-TZVPD
  • pyscf_wb97m_v_def2_tzvpd_density_fit: 9 files, PySCF wB97M-V/def2-TZVPD with density fitting
  • uma_m_1p1_omol: 37 files, UMA-M-1.1 OMol autograd Hessian
  • uma_s_1p2_omol: 37 files, UMA-S-1.2 OMol autograd Hessian

Notes

The calculations were run on samples from data/hessians/hessian_0.h5 in the local THEMol workflow. The original source THEMol shard is referenced in metadata but is not uploaded here.
Downloads last month
39
Total file size:
34.9 MB