These ones are mostly just a test how to make specific actions for prompting.
These Lora models were trained with around 35 hand picked images for one version and 50 images for the another. The realistic one trained under SD 1.5 and the anime one, under Anything V4.5.
Best values I noticed were on the low side (between 0.6-0.3). Add "eating a hamburger, nom hamburger, two hands" and similar booru tags to make it work. Needs a lot of adjustment for the realistic one. Remember to use parenthesis and negative prompts to make repetitive elements to be gone. Experiment!
(Note that sometimes if you add "food focus" or "female/male focus" might appear more hamburgers or make a massive one)