app.py

import gradio as gr
import torch
import os
from transformers import pipeline

model_id = "meta-llama/Meta-Llama-3-70B-Instruct"
access_token = os.getenv("HF_TOKEN")

generator = pipeline(
    "text-generation",
    model=model_id,
    token=access_token,
    model_kwargs={"torch_dtype": torch.bfloat16},
    device_map="auto",
)

def generate_pirate_reply(user_input):
    messages = [
        {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"},
        {"role": "user", "content": user_input},
    ]
    output = generator(messages, max_new_tokens=256)
    return output[0]["generated_text"]

iface = gr.Interface(
    fn=generate_pirate_reply,
    inputs=gr.Textbox(lines=3, placeholder="Ask the pirate anything..."),
    outputs="text",
    title="☠️ Pirate Bot (LLaMA 3-70B)",
    description="Talk like a pirate, powered by Meta's LLaMA 3 model!"
)

iface.launch()