Alex Intelligence
Region: CA
Tuesday 04 November 2025 02:09:55 GMT
184692
11286
287
973
Music
Download
Comments
Kaka Yenu Shujaa :
Deepseek engineers are cracked
2025-11-06 16:31:46
1566
🏴☠️Bunny :
So like everything, the west just makes wasteful garbage
2025-11-06 03:28:18
334
Z :
China is always the hope for humanity 😂
2025-11-07 04:18:04
425
Amit Kulkarni :
ok but tiananmen square
2025-11-12 17:24:56
12
noneofya999 :
you know what this means, AI can be achieved by anyone at home...no need for all these data centers.
2025-11-08 21:24:16
186
amytalkstech :
Creative. I wonder if the neural pathways created by reading in characters vs alphabet made this discovery more obvious to China.
2025-11-05 19:01:13
180
Chorusforlife :
That’s how Chinese language works
2025-11-09 00:19:45
27
Mr. Bearo :
This is how super fast readers process pages in an instant.
2025-11-07 14:05:11
214
♋✨👷🇨🇳_E.C.E.__Z_🇵🇭 ✨♋ :
"China is going to win the AI race" - Jensen Huang (Nvidia Ceo)
2025-11-09 10:02:04
192
Socrates :
The western AI bubble is just about getting trillions of dollars from investors because whoever makes super intelligence first wins the game. They don't waste money on research or even thinking. It's all about getting more money from investors.
2025-11-11 10:41:54
11
aghfhaso :
That’s actually so genius
2025-11-05 02:06:24
50
Ragnar Oogh :
What happened in Tian an men square from May to June 1989?
2025-12-26 01:00:28
0
JoeShmoe525 :
They basically taught it a speed reading technque where you go by chunks instead of reading word by word
2025-11-08 19:50:51
81
Adam :
I'm wondering if you're talking about diffusion vs transformer models. If so, say so bro
2025-11-04 03:08:09
0
smcg :
It’s looking at the way Asian characters are learned rather than using alphabets. You only use a certain number of words to become literate in language. The rest of the words can be filled in or learned when required.
2025-11-14 02:36:14
8
Kidd🌙✨²⁶¹ :
that is an amazing tech
2025-11-14 04:55:05
3
fh450 :
Each visual token is a 65536 bit vector (8192 bytes), so a 6000 word paper in 256*8192 is over 2 megabytes (more than enough to store the complete works of Shakespeare as a compressed text file). This is less about data storage and more about connecting information to visual patterns.
2025-11-09 14:44:17
8
Sean :
I'm calling it: OCR language models will be forgotten before the new year
2025-11-08 22:31:57
9
luckyniisan :
Before LLMs I was thinking of doing text ai using vision not tokens
2025-11-08 13:54:41
3
Dr Darren Abbey :
The same way you do.
2025-11-09 07:13:34
2
heroiklife :
How’s the reasoning throughput… oh yeah 60% … yeah - tell the whole story
2025-11-11 19:16:45
2
Conc :
China 😘
2025-11-06 23:52:48
4
Cumberland Carnivores 🇨🇦 :
This is how I think, and other people who think in images all agree - were way smarter than people who think in words. 😏
2025-11-10 00:11:35
3
Cat Shi :
This is not good… we need to understand what the models are thinking to make sure it doesn’t become maligned and that requires its thinking to be in words. There’s an entire field called mechanistic interpretability for this.
2025-11-09 07:20:17
2
To see more videos from user @alex_intelligence, please go to the Tikwm
homepage.
