import os
import torch
import gradio as gr
from PIL import Image
from transformers import AutoModelForCausalLM, AutoProcessor

device = 'cuda' if torch.cuda.is_available() else 'cpu'

processor = AutoProcessor.from_pretrained("microsoft/git-base")
model = AutoModelForCausalLM.from_pretrained("sam749/sd-portrait-caption").to(device)

def generate_captions(images, max_length=200):
    # prepare image for the model
    inputs = processor(images=images, return_tensors="pt").to(device)
    pixel_values = inputs.pixel_values
    generated_ids = model.generate(pixel_values=pixel_values, max_length=max_length)
    generated_caption = processor.batch_decode(generated_ids, skip_special_tokens=True)
    return generated_caption

def generate_caption(image, max_length=200):
    return generate_captions([image], max_length)[0]

image_input = gr.Image(type="pil", label="Upload Image", height=400)
max_length_slider = gr.Slider(minimum=10, maximum=400, value=200, step=8, label="Max Length")
caption_output = gr.Textbox(label="Generated Caption")

demo = gr.Interface(
    fn=generate_caption,
    inputs=[image_input, max_length_slider],
    outputs=caption_output,
    theme="gradio/monochrome",
    title="Stable Diffusion Portrait Captioner",
    allow_flagging="never"
)

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