app.py

import gradio
import transformers
import types


checkpoint_path = "checkpoint"
examples_path = "examples"

MODEL = types.SimpleNamespace()
MODEL.donut_processor = transformers.DonutProcessor.from_pretrained(checkpoint_path)
MODEL.encoder_decoder = transformers.VisionEncoderDecoderModel.from_pretrained(checkpoint_path)
MODEL.tokenizer = MODEL.donut_processor.tokenizer


def generate_token_strings(images, skip_special_tokens=True) -> list[str]:
    decoder_output = MODEL.encoder_decoder.generate(
        images,
        max_length=MODEL.encoder_decoder.config.decoder.max_length,
        eos_token_id=MODEL.tokenizer.eos_token_id,
        return_dict_in_generate=True,
    )
    return MODEL.tokenizer.batch_decode(
        decoder_output.sequences, skip_special_tokens=skip_special_tokens
    )

def predict_string(image) -> str:
    image = MODEL.donut_processor(
        image, random_padding=False, return_tensors="pt"
    ).pixel_values
    string = generate_token_strings(image)[0]
    return string


interface = gradio.Interface(
    title = "Making graphs accessible",
    description = "Generate textual representation of a graph\n"
    "https://www.kaggle.com/competitions/benetech-making-graphs-accessible",
    fn=predict_string,
    inputs="image",
    outputs="text",
    examples=examples_path,
)

interface.launch()