app.py

from threading import Thread
from typing import Dict

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


TITLE = "<h1><center>Chat with PaliGemma-3B-Chat-v0.1</center></h1>"

DESCRIPTION = "<h3><center>Visit <a href='https://huggingface.co/hiyouga/PaliGemma-3B-Chat-v0.1' target='_blank'>our model page</a> for details.</center></h3>"

CSS = """
.duplicate-button {
  margin: auto !important;
  color: white !important;
  background: black !important;
  border-radius: 100vh !important;
}
"""


model_id = "hiyouga/PaliGemma-3B-Chat-v0.1"
tokenizer = AutoTokenizer.from_pretrained(model_id)
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForVision2Seq.from_pretrained(model_id, torch_dtype="auto", device_map="auto")


@spaces.GPU
def stream_chat(message: Dict[str, str], history: list):
    # Turn 1:
    # {'text': 'what is this', 'files': ['image-xxx.jpg']}
    # []

    # Turn 2:
    # {'text': 'continue?', 'files': []}
    # [[('image-xxx.jpg',), None], ['what is this', 'a image.']]

    image_path = None
    if len(message["files"]) != 0:
        image_path = message["files"][0]

    if len(history) != 0 and isinstance(history[0][0], tuple):
        image_path = history[0][0][0]
        history = history[1:]

    if image_path is not None:
        image = Image.open(image_path)
    else:
        image = Image.new("RGB", (100, 100), (255, 255, 255))

    pixel_values = processor(images=[image], return_tensors="pt").to(model.device)["pixel_values"]

    conversation = []
    for prompt, answer in history:
        conversation.extend([{"role": "user", "content": prompt}, {"role": "assistant", "content": answer}])

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

    input_ids = tokenizer.apply_chat_template(conversation, add_generation_prompt=True, return_tensors="pt")
    image_token_id = tokenizer.convert_tokens_to_ids("<image>")
    image_prefix = torch.empty((1, getattr(processor, "image_seq_length")), dtype=input_ids.dtype).fill_(image_token_id)
    input_ids = torch.cat((image_prefix, input_ids), dim=-1).to(model.device)

    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)

    generate_kwargs = dict(
        input_ids=input_ids,
        pixel_values=pixel_values,
        streamer=streamer,
        max_new_tokens=256,
        do_sample=True,
    )

    t = Thread(target=model.generate, kwargs=generate_kwargs)
    t.start()

    output = ""
    for new_token in streamer:
        output += new_token
        yield output


chatbot = gr.Chatbot(height=450)

with gr.Blocks(css=CSS) as demo:
    gr.HTML(TITLE)
    gr.HTML(DESCRIPTION)
    gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
    gr.ChatInterface(
        fn=stream_chat,
        multimodal=True,
        chatbot=chatbot,
        fill_height=True,
        cache_examples=False,
    )


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