Initial attempt. Seems to funtion as intended.
Training set included manually added tags appended by autogenerated tags.

Updated suggested trigger words from the training data's "yellow" to "blonde," which produces better results, especially when paired with other models.

Also added suggested min and max values. 0.9 is default for consistency. 0.8 works well for having the character in other lined artstyles.
Probably need 1.4 for it work well with lineless or more realistic styles.