app.py

import gradio as gr
from transformers import pipeline
from prompt import PREAMBLE

pipe = pipeline(
    "text-generation", 
    model="HuggingFaceH4/zephyr-7b-alpha"
)

def ask(prompt):
    full_prompt = f"""
{PREAMBLE}

User question:
{prompt}

Answer:
"""
    result = pipe(full_prompt, max_new_tokens=120)
    return result[0]["generated_text"]

demo = gr.Interface(fn=ask, inputs="text", outputs="text")
demo.launch()