Attach negative weight to fine-tune

Is there a good way to attach a negative weight to a fine-tune?

IE:
#Give an output that means the opposite of the input
#Correct
#Input: Hot Output: Cold
#Input: High Output: Low
#Incorrect
#Input: Help Output: Count
#Input: Fast Output: Dry

Not specifically, but you could just include the good and bad examples in the dataset.

Hope that helps!

1 Like