Issue with running my own dataset

[aflato]

trafficstars

I am ruinning my own dataset, after all pre-processing the below code generates error, which is location in Class DualEncoder.

def create_example(image_path, caption):

feature = { "caption": bytes_feature(caption.str.encode()), "raw_image": bytes_feature(tf.io.read_file(image_path).numpy()), } return tf.train.Example(features=tf.train.Features(feature=feature))

I don't how did they took image_path and caption in the given code. My colab code is in this GIVEN LINK. Many people wants to use it for their own datasets, if anyone solves my issue I will share my code here publically.