app.py

import os

from vllm import LLM, SamplingParams
import gradio as gr

from PIL import Image
from io import BytesIO
import base64
import requests

from huggingface_hub import login
import os

login(os.environ["HF_TOKEN"])

repo_id = "mistral-community/pixtral-12b-240910" #Replace to the model you would like to use
sampling_params = SamplingParams(max_tokens=8192, temperature=0.7)
max_tokens_per_img = 4096
max_img_per_msg = 5

llm = LLM(model="mistralai/Pixtral-12B-2409",
          tokenizer_mode="mistral",
          max_model_len=65536,
          max_num_batched_tokens=max_img_per_msg * max_tokens_per_img,
          limit_mm_per_prompt={"image": max_img_per_msg})  # Name or path of your model

def encode_image(image: Image.Image, image_format="PNG") -> str:
    im_file = BytesIO()
    image.save(im_file, format=image_format)
    im_bytes = im_file.getvalue()
    im_64 = base64.b64encode(im_bytes).decode("utf-8")
    return im_64


# @spaces.GPU #[uncomment to use ZeroGPU]
def infer(image_url, prompt, progress=gr.Progress(track_tqdm=True)):
    image = Image.open(BytesIO(requests.get(image_url).content))
    image = image.resize((3844, 2408))
    new_image_url = f"data:image/png;base64,{encode_image(image, image_format='PNG')}"

    messages = [
        {
            "role": "user",
            "content": [{"type": "text", "text": prompt}, {"type": "image_url", "image_url": {"url": new_image_url}}]
        },
    ]

    outputs = llm.chat(messages, sampling_params=sampling_params)

    return outputs[0].outputs[0].text


examples = [["https://picsum.photos/id/237/200/300", "What do you see in this image?"]]

css = """
#col-container {
    margin: 0 auto;
    max-width: 640px;
}
"""

with gr.Blocks(css=css) as demo:
    with gr.Column(elem_id="col-container"):
        gr.Markdown(f"""
        # Mistral Pixtral 12B
        """)

        with gr.Row():
            prompt = gr.Text(
                label="Prompt",
                show_label=False,
                max_lines=2,
                placeholder="Enter your prompt",
                container=False,
            )

        with gr.Row():
            image_url = gr.Text(
                label="Image URL",
                show_label=False,
                max_lines=1,
                placeholder="Enter your image URL",
                container=False,
            )

        with gr.Row():
            run_button = gr.Button("Run", scale=0)

        result = gr.Textbox(
            show_label=False
        )

        gr.Examples(
            examples=examples,
            inputs=[image_url, prompt]
        )
    gr.on(
        triggers=[run_button.click, image_url.submit, prompt.submit],
        fn=infer,
        inputs=[image_url, prompt],
        outputs=[result]
    )

demo.queue().launch()