app.py

import gradio as gr
import torch
from PIL import Image
from threading import Thread
from transformers import AutoProcessor, AutoModelForVision2Seq, TextIteratorStreamer
import spaces

device = "cuda" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.bfloat16 if torch.cuda.is_available() else torch.float32

MODEL_ID = "HuggingFaceTB/SmolVLM-256M-Instruct"
processor = AutoProcessor.from_pretrained(MODEL_ID)
model = AutoModelForVision2Seq.from_pretrained(
    MODEL_ID,
    torch_dtype=torch_dtype,
    trust_remote_code=True
).to(device)


@spaces.GPU
def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    messages = [{"role": "system", "content": system_message}]

    print(message)
    print(history)

    messages.extend(history)

    images = []
    if message["files"]:
        pil_image = Image.open(message["files"][0]).convert("RGB")
        images.append(pil_image)

    current_user_message = {"role": "user", "content": message["text"]}
    messages.append(current_user_message)

    prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
    inputs = processor(text=prompt, images=images, return_tensors="pt").to(device, torch_dtype)

    streamer = TextIteratorStreamer(processor.tokenizer, skip_prompt=True, skip_special_tokens=True)

    generation_kwargs = dict(
        **inputs,
        streamer=streamer,
        max_new_tokens=max_tokens,
        do_sample=True,
        temperature=temperature,
        top_p=top_p,
    )

    thread = Thread(target=model.generate, kwargs=generation_kwargs)
    thread.start()

    response = ""
    for new_text in streamer:
        response += new_text
        yield response

demo = gr.ChatInterface(
    respond,
    type='messages',
    multimodal=True,
    additional_inputs=[
        gr.Textbox(value="You are a helpful and friendly multimodal assistant. You can analyze images and answer questions about them.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=1.0, 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 (nucleus sampling)",
        ),
    ],
    title="Chatbot",
    description="Ask me anything or upload an image. This version uses AutoModel and AutoProcessor directly.",
)

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