app.py

import streamlit as st
from langchain.llms import HuggingFaceHub

# Function to return the response
def generate_answer(query):
    llm = HuggingFaceHub(
        model_class="goliath-120b.Q4_K_M.gguf",
        repo_id="alpindale/goliath-120b",
        model_kwargs={"temperature": 0.5, "max_length": 64, "max_new_tokens": 512},
        task = "text2text-generation"
    )
    prompt = f"""
        You are a helpful AI assistant.

        USER:
        {query}

        ASSISTANT:
    """
    result = llm.predict(prompt)
    return result

# App UI starts here
st.set_page_config(page_title="LangChain Demo", page_icon=":robot:")
st.header("LangChain Demo")

# Gets User Input
def get_text():
    input_text = st.text_input("You: ", key="input")
    return input_text

user_input = get_text()
response = generate_answer(user_input)

submit = st.button("Generate")

# If the button is clicked
if submit:
    st.subheader("Answer:")
    st.write(response)