app.py

import re
import gradio as gr
import gradio_theme

import torch
from transformers import DonutProcessor, VisionEncoderDecoderModel
import multiprocessing

processor = DonutProcessor.from_pretrained("naver-clova-ix/donut-base-finetuned-cord-v2")
model = VisionEncoderDecoderModel.from_pretrained("naver-clova-ix/donut-base-finetuned-cord-v2")

device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)

# Number of threads
# Set the number of threads you want to use
desired_num_threads = multiprocessing.cpu_count()  # Change this value as needed
torch.set_num_threads(desired_num_threads)

def process_document(image):
    # prepare encoder inputs
    pixel_values = processor(image, return_tensors="pt").pixel_values
    
    # prepare decoder inputs
    task_prompt = "<s_cord-v2>"
    decoder_input_ids = processor.tokenizer(task_prompt, add_special_tokens=False, return_tensors="pt").input_ids
          
    # generate answer
    outputs = model.generate(
        pixel_values.to(device),
        decoder_input_ids=decoder_input_ids.to(device),
        max_length=model.decoder.config.max_position_embeddings,
        early_stopping=True,
        pad_token_id=processor.tokenizer.pad_token_id,
        eos_token_id=processor.tokenizer.eos_token_id,
        use_cache=True,
        num_beams=1,
        bad_words_ids=[[processor.tokenizer.unk_token_id]],
        return_dict_in_generate=True,
    )
    
    # postprocess
    sequence = processor.batch_decode(outputs.sequences)[0]
    sequence = sequence.replace(processor.tokenizer.eos_token, "").replace(processor.tokenizer.pad_token, "")
    sequence = re.sub(r"<.*?>", "", sequence, count=1).strip()  # remove first task start token
    
    return processor.token2json(sequence)

#description = "Clip AI: Check Understanding"
#article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2111.15664' target='_blank'>Donut: OCR-free Document Understanding Transformer</a> | <a href='https://github.com/clovaai/donut' target='_blank'>Github Repo</a></p>"

demo = gr.Interface(
    fn=process_document,
    inputs="image",
    outputs="json",
    #title="Prueba nuestra API",
    #description=description,
    #article=article,
    enable_queue=True,
    #examples=[["example.png"], ["example_2.png"], ["example_3.png"]],
    cache_examples=False,
    theme=gradio_theme.theme
)

demo.launch()