Model is trained on 40 pictures of hare from soviet series "Nu, pogodi!" (main 16 episodes) at 512*512 resolution for 500 epochs.
Examples are NOT cherrypicked, generated each 5 by Euler a, Euler, DPM++ 2M Karras and UniPC respectively.

Trigger word is "waj" due to it is a "rare token", what means it is very unlikely to interfere with the "main model" itself.