run_model.py

import os
from langchain_google_genai import ChatGoogleGenerativeAI

gemma_model = ChatGoogleGenerativeAI(
    model="gemma-3-12b-it",
    temperature=0,
    max_tokens=2048,
    timeout=None,
    max_retries=2,
)

def generate_response(prompt: str, context: str="", history: list = []):

    if context == "":
        message = history + [{"role" : "user", "content" : prompt}]
    else:
        message = history + [{"role" : "user", "content" : f"Context : {context}\nQuery : {prompt}"}]

    response = gemma_model.invoke(message)

    return response.content