This LoRA has "v2" in the file name, but it is not a failure. It is trained twice with 52 images, 10 epochs, 5 iterations for 650 steps, and trained more with 10 images by the first LoRA, 10 epochs, 10 iterations for 250 steps. Trigger and useful keywords are as follows.
Appearance: sdxl-linn-minmei, blue hair, long hair, twin braids, green eyes, braid
Notice: "retro artstyle" is effective, and "macross", "lynn minmay" will be effective but it would make images different from these sample images for the reason of a dependency on your base model.（本来は "lynn minmay" だけでも出せる場合があるため、ベースモデルの描写の影響があります。）