The dataset for this one was pretty small, so there may be some aspects that are overtrained since I had to compromise. Hopefully it still works well enough.

Recommended weight = 1, but as always, feel free to play around with it.