Legal Vector Search application!📄

This Streamlit application allows users to search for legal opinions in a Milvus vector database and generate summaries using the Llama-3 language model and DSPy.

Prerequisites

Before running the Streamlit application, ensure that the following prerequisites are met:

1. Milvus standalone is running.
2. Ollama llama-3 language model is running.
3. The vector database is initialized with the legal opinion documents.

Installation

1. Clone the repository:

git clone https://github.com/dope-projects/llm-law-hackathon.git
cd llm-law-hackathon

2. Install the required dependencies:

pip install -r requirements.txt

Usage

1. Start the Milvus standalone server:

wget https://raw.githubusercontent.com/milvus-io/milvus/master/scripts/standalone_embed.sh
bash standalone_embed.sh start

2. Start the Llama-3 language model on ollama server.

3. Initialize the vector database with the legal opinion documents:

from langchain_community.vectorstores import Milvus
from langchain_openai import OpenAIEmbeddings

embeddings = OpenAIEmbeddings(model = "text-embedding-3-large")

vector_db = Milvus(
    embeddings,
    connection_args={"host": "127.0.0.1", "port": "19530"},
    collection_name="LangChainCollection",
)

# Add your legal opinion documents to the vector database
documents = [...]  # List of legal opinion documents
vector_db.add_documents(documents)

4. Run the Streamlit application:

streamlit run app.py

Configuration

• The Milvus connection settings can be modified in the vector_db initialization code block.
• The Ollama-3 language model settings can be adjusted in the ollama_model initialization code block.

License

This project is licensed under the MIT License.

Acknowledgements

• Milvus - Vector database
• Ollama-3 - LLM inference
• DSPy - LLM Orchestrator
• Streamlit - Frontend framework