@Guglielminetti I did that, totally wrong, but I ask it about my life as it understood it and it talked to me for 20 minutes about my likes, dislikes, wife, life, interests and how I come to it with unhinged ideas but during the course of our talks about the idea I prove them more or less.
@lucasjvds@GoogleDeepMind First off congrats... and When you twist the field lines and wrap them through non-trivial geometry (like a Möbius-like transformation and helical twist), the magnetic helicity ∫ A B dV locks the topology in place. I was working in this direction but I have no Postnominals so..
@sbkaufman I read Nina’s tweet and grin. So sure we are the pinnacle of evolution after 4.3 billion years of subduction the probability is pretty good we are not the first species “to industrialize intelligence and distribution of goods” planet-wide, humans are some interesting animals.
@HealthEquity HealthEquity I receive emails from you constantly telling me that "Your account statement for October is available now." I have never signed up for this and when I go to your website and try to use the username you supply I get "not a member". So some help please?
@rikiyatakehi@a1zhang@lateinteraction@jayden_teoh_ Resolving the gap between latent space geometric projections and the human lexical interface to not include the aliasing and distortion of flat linear projections you remove much of the need to edit the weights. That resolves the conflict that causes the reasoning loops.
@vikktorrrre AI are algorithms they have no agency to do things on their own. The issue ergo is not with AI but with the DEVS that program the algorithm. It is amazing to see the witch hunt start when there are no witches and the priests that accuse them of being witches are at fault.
@theRealKiyosaki Now finish the story, "a Princeton PhD in Technology" whom applied 230 times did so each time with a misspelled word or some nonsense in his/her/she/it's resume. You get hired because you click with an interviewer and are not socially ignorant. Need I say more?
if only AI Developers would listen...capacity ≠ cognition, you are chasing the wrong rabbit. But you knew it all along because you are chasing the financial gains rabbit.
@fanofaliens Now apply this to a stellarator configuration and will find that when you fold and braid the geometry in a stellarator format, the topological stability comes from the boundary conditions and helicity, not brute-force cylindrical symmetry. Notice the Mobius folding in the picture
Implement hardware routing tables based on non-Abelian fusion rules (tauXtau=1+tau). Signals in a neuromorphic substrate can be routed and filtered by topological braiding history rather than static address tables providing robust fault tolerance against localized degradation.
Why not: Use braid group representations and Temperley-Lieb recoupling equations to map polymer chain disentanglement pathways during catalytic scission or thermal recycling.
Train Llama's middle layers to reason natively in Cl(3,3) — with no token emission giving you causality, temporal and negation recognition.. https://t.co/X9gMz45s2q
Everyone loves self hosted models: but have you thought about whom is coding, compressing and fine-tuning that model and to what end they are pursuing?
There is no security in the world that catches weight poisoned models.
@whaleyxbt I always sound condescending and do not mean to be rude. It was exactly what you said. A cool visualization, I have been working on code that will do this and I know I am not in your league but I am retired and just having fun. Peace out homie.