Quick Start

pip install vectorboard

from vectorboard.search import GridSearch
from langchain.document_loaders import PyPDFLoader
from langchain.chains import RetrievalQA
from langchain.vectorstores import FAISS
from langchain.embeddings import OpenAIEmbeddings, HuggingFaceEmbeddings

import os
from dotenv import load_dotenv

load_dotenv()
# os.environ["OPENAI_API_KEY"] = os.getenv("OPENAI_API_KEY")

loader = PyPDFLoader("recycling.pdf")

param_grid = {
    "chunk_size": [500],
    "vector_store": [FAISS],
    "embeddings": [OpenAIEmbeddings(), HuggingFaceEmbeddings()],
}
eval_queries = [
    "What is the trend in the amount of household waste collected and treated in 2022 compared to the previous year?",  # noqa: E501
    "what percentage of waste is recyvled into materials in 2022?",
    "what percentage of waste is recovered into energy in 2022?",
    "what is the total volume of waste treated for energy recovery per person?",
]


def main():
    vectorboard_grid_search = GridSearch(chain=RetrievalQA)
    vectorboard_grid_search.create_experiments(param_grid=param_grid, loader=loader)
    vectorboard_grid_search.run(eval_queries=eval_queries)
    vectorboard_grid_search.results(share=True)


if __name__ == "__main__":
    main()