26/02/14 V1

Trained the model based on previous data(2-4 datasets). Some artist's arts could affect the result more but overall it is somewhat more stable than some of the loras for the separate datasets.