This model is another bit of an ongoing series of experiments with training parameters to see what sticks well. This is an attempt to see how well the model can be trained on a specific concept with few images--the concepts being ohogao and Akita Neru; most of the dataset is of Pekomama and Pekora.