app.py

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

model_id = "microsoft/phi-3-mini-4k-instruct"
hf_token = os.getenv("HF_TOKEN")

tokenizer = AutoTokenizer.from_pretrained(model_id, token=hf_token)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
    device_map="auto",
    token=hf_token
)

def respond(message, history, system_message, max_tokens, temperature, top_p):
    prompt = system_message + "\n"
    for user, bot in history:
        prompt += f"<|user|>\n{user}\n<|assistant|>\n{bot}\n"
    prompt += f"<|user|>\n{message}\n<|assistant|>\n"

    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(
        **inputs,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        do_sample=True,
        pad_token_id=tokenizer.eos_token_id
    )
    response = tokenizer.decode(outputs[0], skip_special_tokens=True).split("<|assistant|>")[-1].strip()
    return response

demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a helpful assistant.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=1.5, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
    ],
)

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