app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "microsoft/phi-2"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="float32"  # CPUにやさしい
)

def chat_fn(message, history=[]):
    inputs = tokenizer(message, return_tensors="pt")
    outputs = model.generate(**inputs, max_new_tokens=100)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response

iface = gr.ChatInterface(chat_fn, title="💬 phi-2 Chat")
iface.launch()