| --- |
| license: mit |
| tags: |
| - PubChem |
| - chemistry |
| --- |
| |
| A machine learning dataset based on SMILES strings from [PubChem](https://pubchem.ncbi.nlm.nih.gov/). This dataset is unlabeled and intended for self-supervised learning tasks. |
| It contains 123,860,695 SMILES strings and that have been canonicalized with RDKit and tokenized for a Transformer model. |
|
|
| The dataset has been generated with the [ChEB-AI library](https://github.com/ChEB-AI/python-chebai) (v1.3.0). |
|
|
| ## Dataset structure |
| ``` |
| Pubchem/ |
| ├── raw/full/ |
| │ ├── smiles.txt # Full list of SMILES from the [PubChem FTP server, May 2026](https://ftp.ncbi.nlm.nih.gov/pubchem/Compound/Monthly/) |
| │ |
| └── processed/full/ |
| │ |
| ├── data.pkl # molecule table as a pandas DataFrame (parsed SMILES strings) |
| ├── splits.csv # ID-split pairs with 1,000,000 train samples and 100,000 validation / test samples each. All other IDs are train samples. |
| │ |
| └── static_smiles/ # Tokenized SMILES strings ([Tokenizer implementation](https://github.com/ChEB-AI/python-chebai/blob/dev/chebai/preprocessing/smiles_tokenizer.py)) |
| │ |
| ├── train_{i}.pt # i=0..123, blocks of 1,000,000 train samples (excluding parsing errors) |
| ├── validation.pt # 100,000 validation samples (excluding parsing errors) |
| ├── test.pt # 100,000 test samples (excluding parsing errors) |
| |
| ``` |
|
|
| ## Usage |
| This dataset can be used to train ELECTRA transformer models with the ChEB-AI library. For usage examples, see the [README](https://github.com/ChEB-AI/python-chebai/blob/dev/README.md). |
| A model trained on this dataset is available [here](https://huggingface.co/chebai/electra_pubchem_123m). |
|
|
