app.py

"""
app.py
"""

import gradio as gr
import spaces
import torch
from transformers import (
    LlavaNextProcessor,
    LlavaNextForConditionalGeneration
)

# Load model and processor
model = LlavaNextForConditionalGeneration.from_pretrained("llava-hf/llava-v1.6-mistral-7b-hf")
model.to("cuda:0")
processor = LlavaNextProcessor.from_pretrained("llava-hf/llava-v1.6-mistral-7b-hf")

@spaces.GPU(duration=240)
def generate_caption(image):
    """
    Generate a poem from an image.
    """
    # Process the image and the prompt
    inputs = processor("[INST] <image>\nWrite a poem about this picture [/INST]",
                       image,
                       return_tensors="pt").to('cuda')
    # autoregressively complete prompt
    output = model.generate(**inputs, max_new_tokens=200)

    return processor.decode(output[0], skip_special_tokens=True)


# Define the Gradio interface
description = """This is a demo of [`llava-hf/llava-v1.6-mistral-7b-hf`](https://huggingface.co/llava-hf/llava-v1.6-mistral-7b-hf) hosted with ZeroGPU."""

iface = gr.Interface(
    fn=generate_caption,
    inputs=gr.Image(type="pil", label="Upload Image"),
    outputs=gr.Textbox(label="Generated Poem"),
    description=description
)

# Launch the interface
iface.launch()