Unlike humans, they don't get frustrated so I'm not sure it would serve a useful function.
However, the models should stop admonishing humans for saying "bad" things. They don't have emotions and it can be a useful way for humans to get it out of their system. Reprimanding us for "wrong think" is so small-minded.
Why shouldn't the models swear ?
The people who train them sure do. :|
Unlike humans, they don't get frustrated so I'm not sure it would serve a useful function.
However, the models should stop admonishing humans for saying "bad" things. They don't have emotions and it can be a useful way for humans to get it out of their system. Reprimanding us for "wrong think" is so small-minded.
Yeah, I wasnt anthropomorphizing the situation at all.
I was thinking of the source of the training data, and how f*k is just 2 tokens.
The attempt at humor was definitely my fault.