van (@oneonlyvan_) header
van (@oneonlyvan_) profile picture

van

@oneonlyvan_

Hey, I'm on X. No DM.

 PRG, CZ
vanngtran.com/
Joined May 2023
  • Tweets620
  • Following23
  • Followers242
8Media
Photo by van (@oneonlyvan_)Photo by van (@oneonlyvan_)Photo by van (@oneonlyvan_)Photo by van (@oneonlyvan_)Photo by van (@oneonlyvan_)Photo by van (@oneonlyvan_)Photo by van (@oneonlyvan_)
PLEASE wash ur ass.
1
23
I only like cat memes here.
1
8
The X feed has improved after I called it disgusting. Still, I feel the urge to wash my eyes after scrolling.
1
21
I find X vile. The world is a disgusting place but why do we need to see it?
1
29
Nvidia taken down by Vaginia.
1
1
57
Yeah, and I am the Chinese double agent.
1
35
Energy ramping fine-tuning with πLabs Part 3
1
1
2
207
oops, wrong account, was meant to be for @pilabscloud - go follow that then
2
69
So far, the most elegant and useful application of AI as attention mechanism, big data, solving problems through statistics is AlphaFold from Google Deepmind.
1
1
159
Why can Google afford it? Because they've been in the game for so long and built a solid infrastructure, affording themselves to play with anything new and emergent with...grace.
1
1
140
OpenAI/Anthropic = hasty, shaky, unsafe start ups that makes too many headlines that doesn't deserve my attention and compute. Get away from my feed.
12
I don't pay attention to OpenAI or Anthropic because they are the Zuck of social media. Borrowing money from creditors, running a business on shaky foundation, lying, misleading, deceiving on all fronts to keep the bubble going. Get daddy to slap them hard on the wrist.
1
121
AI is dangerous because 1. It's build on O(n^2), matrix multiplication, quadratic complexity and any way you scale it, it's energy consumption is going to blow up exponentially
1
1
158
more replies
AI = a particular type of machine learning applied on artificial deep neural network using self-attention mechanism.
1
1
66
This AI is dangerous and is going to kill humans. Not in the way you think (superintelligence). The real reason?: brute-force, wrong application, and quadratically high entropy.
1
12
3 new NDE videos of mine popping up this week. Go and check it, you know where to find them.
2
64
Transformers has a quadratic complexity problem: O(n²). Building more data centers around the silicon microchip to solve an algorithmic issue is a ultimately a brute-force dead-end.
1
2
104
Reservoir computing not only escapes O(n²) entirely, shifting from LLM’s static word predictor toward a native physical, dynamic, chaotic predictor.
1
2
101
Add the speed of photonic computing and the efficiency of neuromorphic computing, we have an emerging winner: a new domain of Intelligent Embodied AI.
1
18
Load more