from transformers import DonutProcessor, VisionEncoderDecoderModel
from huggingface_hub import hf_hub_download
from PIL import Image

processor = DonutProcessor.from_pretrained("ajmaclin/prescription-donut-ocr")
model     = VisionEncoderDecoderModel.from_pretrained("ajmaclin/prescription-donut-ocr")

image        = Image.open("prescription_crop.jpg")
pixel_values = processor(image, return_tensors="pt").pixel_values
outputs      = model.generate(pixel_values, max_length=256)
result       = processor.batch_decode(outputs, skip_special_tokens=True)[0]
print(result)