this is a huge 384 dim LoRA, experimenting with adaptive learning ratenote I tried resizing this LoRA to lower dims using the kohya_ss tools, but the quality results were too subjective (some images degrade, others are improved)training settings3137 training images512,512 training resolution16 epochs, 6,384 steps, 8 batchesnetwork dim & alpha 384DAdaptAdam optimizer, constant schedulerlearning rate 1.0bucketing & random cropuseful optional tokensurushisato tagged on all images for general style boostova tagged on all screencaps from animationbackground tagged on any image that has no human subject