A Large Model for Predicting Compound Properties
COMPOUND-GPT is an advanced machine learning model designed to predict various properties of chemical compounds. This repository includes pre-trained models, tokenizer scripts, and example notebooks to help users leverage COMPOUND-GPT for their own research and development purposes.
Clone the repository and install the required dependencies:
git clone https://github.com/525514353/COMPOUND-GPT.git
cd COMPOUND-GPT
pip install -r requirements.txt
## Usage
#Pre-training
The pre_training module includes scripts to train the model on a dataset of chemical compounds. Customize your dataset and training parameters as needed.
#Tokenizers
The tokenizers_lcm module provides tools to process chemical structures into token sequences that the model can understand.
#Hydroxyl Radical Prediction
The hydroxyl_radical module contains a fine_tune model to predict the reactivity of compounds with hydroxyl radicals.
#Toxicity Prediction
The toxicity module offers functionality to assess the toxicity levels of chemical compounds using the fine_tune model.
##License
This project is licensed under the MIT License. See the LICENSE file for details.
Feel free to customize the content as needed. This structure aims to provide clarity and usability for users interested in using or contributing to your project.