Queimo/grappa-vapor-pressure

GRAPPA: vapor pressure (Antoine parameters)

GRAPPA is a hybrid graph neural network that predicts the Antoine parameters of pure components from their SMILES. This repository contains the unchanged published state dict, so that screening-properties (the property package of the ScreeningTool process-screening framework) can download it on first use.

All credit goes to the original authors. Code and weights: marco-hoffmann/GRAPPA (MIT licence, see LICENSE).

@misc{Hoffmann2025,
  title         = {GRAPPA - A Hybrid Graph Neural Network for Predicting Pure Component Vapor Pressures},
  author        = {Marco Hoffmann and Hans Hasse and Fabian Jirasek},
  year          = {2025},
  eprint        = {2501.08729},
  archivePrefix = {arXiv},
  primaryClass  = {cs.LG},
  url           = {https://arxiv.org/abs/2501.08729}
}

Files

File	Content
models/GRAPPA_state_dict.pt	PyTorch state dict (tensors only; load with weights_only=True)

The model outputs Antoine A, B, C for ln(p/kPa) with T in K. Molecules with formal charges are not supported.