molcrawl-compounds-bert-small

Model Description

GPT-2 small (124M parameters) foundation model pre-trained on compound SMILES strings from the MolCrawl dataset.

The tokenizer is a character-level BPE tokenizer (vocab_size=612) that encodes each SMILES character as a separate token. Input SMILES strings should be passed without spaces (e.g. CC(=O)O). The [SEP] token (id=13) is used as the end-of-sequence marker.

Datasets

• MolCrawl compounds corpus (chembl + zinc + opv + reddb + pubchemqc): https://github.com/mmai-framework-lab/MolCrawl-HFuploader/blob/main/workflows/hugging_face/run_upload_hf.sh (Pre-training corpus)

• Model Type: bert

• Data Type: Molecule/Compound

• Training Date: 2026-04-13

Usage

from transformers import AutoModelForMaskedLM, AutoTokenizer
import torch

model = AutoModelForMaskedLM.from_pretrained("kojima-lab/molcrawl-compounds-bert-small")
tokenizer = AutoTokenizer.from_pretrained("kojima-lab/molcrawl-compounds-bert-small")

# Predict masked SMILES token
prompt = "CC(=O)[MASK]"
inputs = tokenizer(prompt, return_tensors="pt")
mask_token_id = tokenizer.mask_token_id
mask_index = (inputs["input_ids"] == mask_token_id).nonzero(as_tuple=True)[1]

with torch.no_grad():
    outputs = model(**inputs)
logits = outputs.logits

predicted_token_id = logits[0, mask_index].argmax(dim=-1)
predicted_token = tokenizer.decode(predicted_token_id)
result = prompt.replace("[MASK]", predicted_token)
print(f"Predicted: {result}")

Training

This model was trained with the RIKEN Foundation Model pipeline. For more details, please refer to the training configuration files included in this repository.

License

This model is released under the APACHE-2.0 license.

Citation

If you use this model, please cite:

@misc{molcrawl_compounds_bert_small,
  title={molcrawl-compounds-bert-small},
  author={{RIKEN}},
  year={2026},
  publisher={{Hugging Face}},
  url={{https://huggingface.co/kojima-lab/molcrawl-compounds-bert-small}}
}