import model
import datasets
from PIL import Image
vlm = model.ImageNetCaptionModel.load_from_checkpoint('checkpoint-epoch=01-loss=0.13.ckpt')

image = Image.open("test_image_5.png")
print(vlm.generate(image=image))