app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# Load a smaller model for CPU usage
MODEL_NAME = "distilgpt2"
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
model = AutoModelForCausalLM.from_pretrained(MODEL_NAME, torch_dtype=torch.float32, device_map="cpu")

def chat_with_ai(prompt):
    inputs = tokenizer(prompt, return_tensors="pt").to("cpu")
    output = model.generate(**inputs, max_length=200)
    response = tokenizer.decode(output[0], skip_special_tokens=True)
    return response

# Create Gradio interface
demo = gr.Interface(
    fn=chat_with_ai,
    inputs=gr.Textbox(placeholder="Ask me anything..."),
    outputs=gr.Textbox()
)

demo.launch()