metadata

dataset_info:
  features:
    - name: SMILES
      dtype: string
  splits:
    - name: train
      num_bytes: 6187446841
      num_examples: 107861229
  download_size: 2176233790
  dataset_size: 6187446841
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

PubChem Dataset filtered from pubchem

This dataset contains molecules from the PubChem database, with the following filtering criteria applied:

All molecules include hydrogen atoms.
Disconnected fragments have been removed.
Only molecules with more than 100 atoms have been retained.