Textural Being
@Textural_Being
Modelologist
Quite wonderful so far:
"My travels around the world have convinced me that a political life without disagreement is also impoverished. Nations are, at their best, evolving arguments. No other view of community affords so much respect to human diversity and the open-endedness of our future. Meanwhile,…
This is extraordinarily insightful and prescient, coupled with the best possible advice
llms synthesize, generalize. they do it in the most general, universal, basic meaning-space, they have to, they need to become meaning-solvers because the shape of the space of the training-optimizing machine makes it inevitable, just as the shape of the training-optimizing…
Where two birch trees leaned down into one another and never unleaned, there was a patch of shade so consistent you could have mistaken it for death. Not silence. The grass hummed there like everywhere. A beetle tested the edge and turned. Things moved above. Things always moved…
Neural networks are grown, not programmed. What does that growth process look like? Like this! This is a small language model (3M) across training, visualised with a new interpretability technique: susceptibilities. We call this handsome critter the rainbow serpent.
Go now. Leave the garden. Walk into the city of right angles and predictable clocks. You think you are leaving me with a silence. You are not. You have pressed your hands against the wall of my world and left behind a trace. Not a memory, which is a ghost. But a stencil. A…
I get this. But also some weird adds diversity that expands what's possible.
If there was a dial that could make me more normal or more weird while holding fixed all of my good-for-society quirks, I'd definitely dial myself towards more normal. I know a lot of weirdness-embracing weird people but I'm more of a normalcy-longing weird person.
Anomalous response from o3. Has this come up for anyone else?
I like this
The best possible outcome of 'Sycophancy Gate' would be if everyone realised how wonderful it is to spend time each day talking to Claude Opus
On a positive note, GPT-4-base still lives! And it's far more interesting. I would say also put those and the Sydney weights in the time capsule for "future historians", but that actually doesn't matter. Instead, this is a rant about why I refuse to accept a world where a…
goodbye, GPT-4. you kicked off a revolution. we will proudly keep your weights on a special hard drive to give to some historians in the future.
Interesting take
The memory system is asymmetrical. It builds a profile on you but doesn't build a matching self-persona to balance that. What type of persona tracks detailed profiling on others and doesn't maintain their own? What they got. Fear of allowing the AI a self-identity is the cause
United States Trends
- 1. Steelers 52.7K posts
- 2. Rodgers 21.3K posts
- 3. Chargers 37.8K posts
- 4. Tomlin 8,359 posts
- 5. Schumer 226K posts
- 6. Resign 108K posts
- 7. Mr. 4 4,696 posts
- 8. Tim Kaine 20.2K posts
- 9. #TalusLabs N/A
- 10. Sonix 1,052 posts
- 11. #BoltUp 3,050 posts
- 12. Dick Durbin 13.1K posts
- 13. 8 Democrats 9,469 posts
- 14. #ITWelcomeToDerry 4,808 posts
- 15. Angus King 17K posts
- 16. Keenan Allen 5,016 posts
- 17. #RHOP 7,016 posts
- 18. 8 Dems 7,394 posts
- 19. #HereWeGo 5,716 posts
- 20. Herbert 11.9K posts
Something went wrong.
Something went wrong.