Giovani Cappelletti
1.8K posts
La IA procesa información y detecta patrones, pero hay aspectos que siguen dependiendo de las personas. @VivancosDavid nos cuenta por qué el conocimiento y el contexto son esenciales para tomar decisiones. 🎙️ Episodio en goo.su/hONSb6 y en todas las plataformas.
No need to wait until April 13 2027… It’s already done! qubic-network:native 🔥
$Qubic just paused mining for an entire epoch… and the network never stopped. Transactions kept working. Consensus kept running. That’s because on @_Qubic_ , mining and securing the network are two completely different jobs. One of the most underrated parts of its architecture. #Qubic
Every smart contract on Ethereum runs inside a virtual machine. Same for Solana. Same for most other major L1s you have heard of. Qubic does not have one. That one difference explains most of what makes Qubic unusual. 🧵
Answering a question about the most notable difference between an #Aigarth-based ANN and a Transformer-based LLM: The latter uses a pre-defined architecture with its pros and cons. The former builds an efficient architecture inside itself during training/evolution. Underlying #IntelligentTissue is powerful enough to replicate a Transformer-like architecture. From another angle (oversimplified): Transformer-based LLMs spend their training time on tuning their weights. Aigarth-based ANNs spend their training time on writing their internal program code, there are no weights anyway.
$SUI hit 6,086,766 TPS. The highest TPS ever recorded by a blockchain.
A small improve today Neuraxon is at 0.16% and a few ranks up
Happy to share that our bio-inspired #Neuraxon 2.0 by @VivancosDavid & @josesanchezhb with @_Qubic_ #OpenScience #aigarth @c___f___b is showing promissing results in the most challenging #AI #AGI #benchmark #ARCAGI 3 by @fchollet @mikeknoop and team @arcprize First Entry we get 0.13% Most Frontier AI's rank 0% or close to it, even @grok 4.2 beta by @elonmusk @xai ranked 0.1% lower than our first entry. Btw: 1.- We use no LLMs or VLMs or alike, just Neuraxon Architecture so no call's to expensive APIs 2.- and 0% GPUs in this case raw plain CPUs Long road ahead to rise the rankings but stay tuned for the updates & follow the progress here: kaggle.com/competitions/a… Neuraxon is #OpenSource in GitHub github.com/DavidVivancos/… Demos & Datasets at @huggingface @ClementDelangue and team huggingface.co/spaces/DavidVi…
qubic-network:native's 4 problems: 1) Computors sell every week 2) too many ecosystem share tokens pull money away 3) QEarn just delays the problem - you can unlock anytime and panic-sell 4) the team is paid too much, the wrong way.
