---
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](https://huggingface.co/datasets/xchen16/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.