app.py

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

model_id = "microsoft/phi-2"

# Load model and tokenizer (CPU + float32)
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float32
).to("cpu")

# Chat function
def chat_with_bot(user_input, history):
    history = history or []
    prompt = ""
    for user, bot in history:
        prompt += f"User: {user}\nAssistant: {bot}\n"
    prompt += f"User: {user_input}\nAssistant:"

    inputs = tokenizer(prompt, return_tensors="pt").to("cpu")
    outputs = model.generate(
        **inputs,
        max_new_tokens=256,
        do_sample=True,
        temperature=0.7,
        top_p=0.95,
        eos_token_id=tokenizer.eos_token_id,
    )

    decoded = tokenizer.decode(outputs[0], skip_special_tokens=True)
    response = decoded[len(prompt):].strip().split("\n")[0]

    history.append((user_input, response))
    return response, history

# Gradio UI
gr.ChatInterface(
    fn=chat_with_bot,
    title="Phi-2 Chatbot (ZeroGPU Safe)",
    theme="soft",
    examples=["What is AI?", "Summarize the French Revolution.", "Tell me a space fact."]
).launch(share=True)