run.py

from langchain.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import Chroma
from transformers import AutoModelForCausalLM

from dotenv import load_dotenv
load_dotenv()

import os
os.getenv("TRANSFORMERS_CACHE")

model_kwargs = {'trust_remote_code': True}

embedding = HuggingFaceEmbeddings(
    model_name="nomic-ai/nomic-embed-text-v1.5",
    model_kwargs=model_kwargs
)

db = Chroma(
    persist_directory="./chroma_db",
    embedding_function=embedding,
    collection_name='CVE'        
)

retriever = db.as_retriever()

template = """Answer the question based only on the following context:
{context}
Do not tell the source of the data
Question: {question}
"""

prompt = ChatPromptTemplate.from_template(template)

model = AutoModelForCausalLM.from_pretrained(
    "zephyr-7b-beta.Q4_K_S.gguf",
    model_type='mistral',
    threads=3,
)

chain = (
    {"context": retriever, "question": RunnablePassthrough()}
    | prompt
    | model
    | StrOutputParser()
)

# Uncomment and use the following for testing
# for chunk in chain.stream("Your question here"):
#     print(chunk, end="", flush=True)