llm_call.py

import os
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import Chroma
from langchain_google_genai import ChatGoogleGenerativeAI
from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain
from google.colab import userdata

class GeminiLLM():
    def __init__(self):
        self.ACCESS_TOKEN = os.getenv('GOOGLE_GEMINI_TOKEN')
        self.model_name = "gemini-pro"

    def getEmbeddingsModel(self):
        self.embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
        return self.embeddings

    def getRetriver(self, documents ):
        vectorstore = Chroma.from_documents(
            documents = documents,
            embedding = self.embeddings,
            persist_directory = "chroma_db_dir",  # Local mode with in-memory storage only
            collection_name="sermon_lab_ai"
        )

        retriever = vectorstore.as_retriever(
            search_kwargs={"k": 3}
        )

        return (retriever, vectorstore)

    def getLLM(self, documents ):
        if os.getenv('GOOGLE_GEMINI_TOKEN') is None:
            raise ValueError("GOOGLE_GEMINI_TOKEN environment variable not set")
        else:
            self.llm  = ChatGoogleGenerativeAI(
                model = self.model_name,
                temperature = 0.7,
                top_k = 40,
                top_p = 1
            )

            return self.llm