Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/augus1999/bayesian-flow-network-for-chemistry
ChemBFN: Bayesian Flow Network Framework for Chemistry Tasks.
https://github.com/augus1999/bayesian-flow-network-for-chemistry
bayesian-flow-networks cheminformatics chemistry generative-model machine-learning qsar qspr transformer
Last synced: 6 days ago
JSON representation
ChemBFN: Bayesian Flow Network Framework for Chemistry Tasks.
- Host: GitHub
- URL: https://github.com/augus1999/bayesian-flow-network-for-chemistry
- Owner: Augus1999
- License: agpl-3.0
- Created: 2024-06-04T10:28:05.000Z (6 months ago)
- Default Branch: main
- Last Pushed: 2024-08-21T11:37:53.000Z (3 months ago)
- Last Synced: 2024-08-22T05:48:10.985Z (3 months ago)
- Topics: bayesian-flow-networks, cheminformatics, chemistry, generative-model, machine-learning, qsar, qspr, transformer
- Language: Python
- Homepage: https://arxiv.org/abs/2407.20294
- Size: 185 KB
- Stars: 2
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# ChemBFN: Bayesian Flow Network for Chemistry
**Abstract**: In this work, we introduce ChemBFN, a language model that handles chemistry tasks based on
Bayesian flow networks working on discrete data. A new accuracy schedule is proposed to improve the sampling quality by significantly reducing the reconstruction loss. We show evidence that our method is appropriate for generating molecules with satisfied diversity even when a smaller number of sampling steps is used. A classifier-free guidance method is adapted for conditional generation. It is also worthwhile to point out that after generative training, our model can be fine-tuned on regression and classification tasks with the state-of-the-art performance, which opens the gate of building all-in-one models in a single module style.
[](https://paperswithcode.com/sota/molecular-property-prediction-on-bbbp-1?p=a-bayesian-flow-network-framework-for)
[](https://paperswithcode.com/sota/molecular-property-prediction-on-clintox-1?p=a-bayesian-flow-network-framework-for)
[](https://paperswithcode.com/sota/molecular-property-prediction-on-freesolv?p=a-bayesian-flow-network-framework-for)
[](https://paperswithcode.com/sota/molecular-property-prediction-on-hiv-dataset?p=a-bayesian-flow-network-framework-for)
[](https://paperswithcode.com/sota/molecular-property-prediction-on-esol?p=a-bayesian-flow-network-framework-for)
[](https://paperswithcode.com/sota/molecular-property-prediction-on?p=a-bayesian-flow-network-framework-for)
[](https://paperswithcode.com/sota/molecular-property-prediction-on-bace-1?p=a-bayesian-flow-network-framework-for)
## News
* [31/07/2024] Paper is available on [arxiv.org](https://arxiv.org/abs/2407.20294).
* [21/07/2024] Paper was submitted to arXiv.
## Usage
You can find example scripts in [📁example](./example) folder.
## Pre-trained Model
You can find pretrained models in [release](https://github.com/Augus1999/bayesian-flow-network-for-chemistry/releases).
## Dataset Format
We provide a Python class [`CSVData`](./bayesianflow_for_chem/data.py) to handle data stored in CSV or similar format containing headers with the following tags:
* __smiles__ or __safe__ or __selfies__ (_mandatory_): the entities under this tag should be molecule SMILES, SAFE or SELFIES strings. Multiple tags are acceptable.
* __value__ (_optional_): entities under this tag should be molecular properties or classes. Multiple tags are acceptable and in this case you can tell `CSVData` which value(s) should be loaded by specifying `label_idx=[...]`. If a property is not defined, leave it empty and the entity will be automatically masked to torch.inf telling the model that this property is unknown.
## Cite This Work
```bibtex
@misc{2024chembfn,
title={A Bayesian Flow Network Framework for Chemistry Tasks},
author={Nianze Tao and Minori Abe},
year={2024},
eprint={2407.20294},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2407.20294},
}
```