NiftyBridge LangChain project

The NiftyBridge LangChain project is an example of how to use multiple packages to create a vector store from text data. The packages used in this project are langchain.vectorstore, langchain.embeddings.openai, langchain.text_splitter, and langchain.document_loaders.

Installation

Clone the repository by running the following command:

Once cloned, navigate to the project root directory: cd langchain Install the required dependencies using pip:

pip install -r requirements.txt

Set up your .env file with necessary API keys. You can copy the sample .env file by running the following command:

cp .env.sample .env

Then, replace the API keys with your own in the .env file.

Usage Once the installation process has been completed successfully, you can now run FastAPI app localy by uvicorn main:app or run Dockerfile to create image

Uvicorn running on http://127.0.0.1:8000 

That API wait request in Method "POST"
curl -X 'POST' \
  'http://127.0.0.1:8000/Who%20is%20Nifty' \
  -H 'accept: application/json' \
  -d ''

response example

Response body Download

{
  "query": "Who is Nifty",
  "result": "Nifty Bridge LLC is a company that offers various products and services to help merchants manage NFT minting, experiences, and sales online through their platform."
}

This code loads a PDF document called 'langchain.pdf' and creates an instance of OpenAIEmbeddings class. It then calls the load_document function to split the loaded PDF document into chunks of 500 tokens with an overlap of 50 tokens. The load_document function returns a list of these chunks.

Finally, the code uses Chroma's from_documents method to create a vector store from the loaded and splitted documents using the previously created OpenAIEmbeddings object.