This model is trained to create images of Hivers from the Traveller RPG.
This is such a weird concept that you probably need to use a ridiculously low CFG scale to get the expected results.
Try adding wings, cleavage, dog, face
to the negative prompt to prevent the model from misinterpreting the hiver as something else.
The LoRA's influence can vary significantly depending on the base model you use. I recommend experimenting with it at different CFG scales for different prompts to find the best outputs.