You have been given a dataset which contains images of digits 3 and 6. The dataset contains a total of 34022 images (16768 files being images of digit 3, and the rest being images of digit 6). The dataset somehow got randomly mixed, and also contains some duplicate images of each digit. Your task is to find the number of unique images of each digit in the dataset.

The steps of the challenge are:

1. Since the images are randomly mixed, we need a way to separate the images of digits 3 and 6 from each other. You can use a clustering algorithm for this purpose.

2. Figure out which digit belongs to which cluster.

3. Count the number of unique images in each cluster, and form the flag.

(Note if x is the number of unique images of digit 3, and y is the number of unique images of digit 6, then the flag will be TMCTF{x_y})