app.py

import gradio as gr
import spaces
import torch
from transformers import AutoModelForCausalLM, ChameleonProcessor, AutoTokenizer, StoppingCriteria, StoppingCriteriaList, TextIteratorStreamer
from threading import Thread
from PIL import Image
import requests


model_path = "facebook/chameleon-7b"
# model = ChameleonForCausalLM.from_pretrained(model_path, torch_dtype=torch.bfloat16, low_cpu_mem_usage=True, device_map="auto")
# processor = ChameleonProcessor.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path, torch_dtype=torch.bfloat16, low_cpu_mem_usage=True, device_map="auto")
model.eval()
processor = ChameleonProcessor.from_pretrained(model_path)
tokenizer = processor.tokenizer

# file_name, alt
multimodal_file = tuple[str, str]
# {'text': 'message here', 'files': []}
multimodal_message = list[str | multimodal_file] | multimodal_file
# todo: verify this type with gr.ChatInterface
message_t = dict[str, str | list[multimodal_file]]
history_t = list[tuple[str, str] | list[tuple[multimodal_message, multimodal_message]]]

def history_to_prompt(
        message,
        history: history_t,
        eot_id = "<reserved08706>",
        image_placeholder = "<image>"
    ):

    prompt = message["text"]
    images = [Image.open(f) for f in message["files"]]
    
    for turn in history:
        print("turn:", turn)
        # turn should be a tuple of user message and assistant message
        for message in turn:
            if isinstance(message, str):
                prompt += user_message
                prompt += eot_id
            if isinstance(message, list):
                for item in message:
                    if isinstance(item, str):
                        prompt += item
                    elif isinstance(item, tuple):
                        image_path, alt = item
                        prompt += image_placeholder
                        image = Image.open(requests.get(image_path, stream=True).raw)
                        images.append(image)
            else:
                prompt += f"(unhandled message type: {message})"
            prompt += eot_id
    return prompt, images

@spaces.GPU(duration=30)
def respond(
    message,
    history: history_t,
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    response = ""

    print(f"message: {message}\nhistory:\n\n{history}\n")
    prompt, images = history_to_prompt(message, history)
    print(f"prompt:\n\n{prompt}\n")

    # prompt = "I'm very intrigued by this work of art:<image>Please tell me about the artist."
    # image = Image.open(requests.get("https://uploads4.wikiart.org/images/paul-klee/death-for-the-idea-1915.jpg!Large.jpg", stream=True).raw)
    # images = [image]

    inputs = processor(prompt, images=images, return_tensors="pt").to(model.device, dtype=torch.bfloat16)

    streamer = TextIteratorStreamer(tokenizer, skip_prompt=True)
    generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=20)

    try:
        # launch generation in the background
        thread = Thread(target=model.generate, kwargs=generation_kwargs)
        thread.start()

        partial_message = ""
        for new_token in streamer:
            partial_message += new_token
            yield partial_message
    except e:
        return f"Error: {e}"    


"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""
demo = gr.ChatInterface(
    respond,
    multimodal=True,
    additional_inputs=[
        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, 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)",
        ),
    ],
)


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