app.py

import os
os.environ["TOKENIZERS_PARALLELISM"] = "false"
os.environ["NO_CUDA_EXT"] = "1"


from reader_llm import get_reader_llm
from retrieval import get_retriever
from answer_rag import answer_with_rag2
import streamlit as st

# Настройка страницы
st.set_page_config(page_title="RAG", layout="wide")
st.title("Туристический путеводитель")
st.header("Города: Ярославль, Екатеринбург, Нижний Новгород, Владимир")

@st.cache_resource
def load_models():
    READER_LLM = get_reader_llm(name="Vikhrmodels/Vikhr-Llama-3.2-1B-Instruct")
    # легкая модель для приложения на сайте hugging face
    embedding_model, KNOWLEDGE_VECTOR_DATABASE = get_retriever()
    return READER_LLM, embedding_model, KNOWLEDGE_VECTOR_DATABASE

READER_LLM, _, KNOWLEDGE_VECTOR_DATABASE = load_models()

if "messages" not in st.session_state:
    st.session_state.messages = []

for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

if prompt := st.chat_input("Задайте Ваш вопрос"):
    st.session_state.messages.append({"role": "user", "content": prompt})
    
    with st.chat_message("user"):
        st.markdown(prompt)
    
    with st.chat_message("assistant"):
        with st.spinner("Ищу информацию..."):
            answer, sources = answer_with_rag2(
                question=prompt,
                llm=READER_LLM,
                knowledge_index=KNOWLEDGE_VECTOR_DATABASE
            )
            st.markdown(answer)
            
         
        st.session_state.messages.append({"role": "assistant", "content": answer})