app.py

import gradio as gr
import requests
import json
import os

# Retrieve the API key from environment variables
API_KEY = os.getenv('API_KEY')
API_URL = "https://api-inference.huggingface.co/models/meta-llama/CodeLlama-7b-hf"
headers = {"Authorization": f"Bearer {API_KEY}"}

def generate_response(prompt):
    data = {
        "inputs": prompt
    }
    response = requests.post(API_URL, headers=headers, json=data)
    if response.status_code == 200:
        return response.json()
    else:
        return f"Error: {response.status_code}\n{response.text}"

def main(prompt):
    response = generate_response(prompt)
    if isinstance(response, str):
        return response
    else:
        return json.dumps(response, indent=2)

iface = gr.Interface(
    fn=main,
    inputs=gr.Textbox(lines=4, placeholder="Enter your prompt here..."),
    outputs=gr.Textbox(lines=10),
    title="Code Assistant",
    description="Enter your prompt and get responses from the model."
)

if __name__ == "__main__":
    iface.launch()