To avoid overfitting, it is better to mix with other lora models and adjust the weight to about 0.75. It depends.
The tagger words are not necessary
You can use "female" in negative prompt to avoid feminine posture
Trained on anything-4.5
OmenValorant.safetensors