LLM-RDF

LLM-RDF: Large language model (LLM)-based reaction development framework.

The code and data repository of "Accelerated end-to-end chemical synthesis development with large language models".

• The backend of the web application can be found in backend.
• The frontend of the web application can be found in frontend.
• The source data of figures in the paper can be found in Source Data.
• The files used to build llm-based agents and interact with llm-based agents can be found in Files for LLM-based agents.
• An online deployment of the application: https://ruan-yixiang.github.io/LLM-RDF/#/main

Deployment

Please refer to the demo.mp4 file provided to use the web application.

Docker (Recommended, may take a few minutes)

1. Pull Docker Images

# AMD64
docker pull ruanyixiang/llm_rdf_back:latest
docker pull ruanyixiang/llm_rdf_front:latest
# ARM64
docker pull ruanyixiang/llm_rdf_back:arm
docker pull ruanyixiang/llm_rdf_front:arm

2. Run Back-End Service

docker run -p {BACKEND_PORT}:81 -d ruanyixiang/llm_rdf_back

3. Run Front-End Service
   Replace {BACKEND_URL} with the actual backend deployment URL.
   Replace {OPENWEBUI_URL} with the actual OpenWebUI deployment URL (unnecessary).

docker run --env=BACKEND_URL={BACKEND_URL} --env=VUE_APP_OPENWEBUI_URL={OPENWEBUI_URL} -p {FRONTEND_PORT}:83 -d ruanyixiang/llm_rdf_front
# Example
docker run --env=BACKEND_URL="http://127.0.0.1:81" -p 83:83 -d ruanyixiang/llm_rdf_front

4. Access the Application
   {FRONTEND_URL}/main
   Example: http://127.0.0.1:83/main

Manual Deployment (Not recommended, It may only work successfully on Intel CPU Windows 10.)

1. Clone the repository:

git clone https://github.com/Ruan-Yixiang/LLM-RDF.git
cd LLM-RDF

2. Install the backend environment
   Use conda to install the dependencies listed in requirements.txt:

# Create a new conda environment named 'llm_rdf_back' with Python 3.8
conda create -n llm_rdf_back python=3.8

# Activate the newly created environment
conda activate llm_rdf_back

# Install the required Python packages from requirements.txt
pip install -r Backend\ Files/requirements.txt

3. Run Back-End Service
   You can modify the backend url in backend/backend.py

cd Backend\ Files/
python backend/backend.py

4. Install Node.js (v16.16.0) on your device
5. Install the npm
   In a new terminal, navigate to the frontend directory and install npm:

cd frontend
npm install -g [email protected]

6. Run Front-End Service
   You can modify the backend_url and openwebui_url in vue.config.js

npm run dev

License

LLM-RDF is distributed under an MIT License.