Most of this, like chores, exercise, and making food, is what a lot of people do anyway... just that they also go to work. No judgement on what she should be wanting. But it sounds so terribly boring.
Sentiment is everywhere in language. But how do LLMs represent it?
We find:
- All models studied have a linear, causal sentiment direction
- They summarize information at placeholder tokens like commas
An early step towards decoding world models!
arxiv.org/abs/2310.15154