After merging a checkpoint, I often train a LoRA using a familiar character to test how the checkpoint performs and to compare it with others. Recently, I’ve been focusing on Jin Yuichi, which is why there are multiple LoRAs of him — he’s my current benchmark character (01.2026).