If images used for training appear directly in the output, lower the intensity.
Add or remove trigger words as needed.
lora-000002.TA_trained.safetensors