Yun-Ta Tsai

12.1K posts

Yun-Ta Tsai banner
Yun-Ta Tsai

Yun-Ta Tsai

@yunta_tsai

Sr. Staff Engineer @Tesla_AI

Earth Beigetreten Ekim 2022
215 Folgt67.4K Follower
Yun-Ta Tsai retweetet
Tesla
Tesla@Tesla·
We – design the chips & hardware – make the cars w/ said hardware – collect real-world data at scale – train the real-world AI model – built (& continue to expand) the massive supercomputer cluster that trains it – deploy AI directly to millions of robots on wheels All that is shared with @Tesla_Optimus for broader applications in both the physical & digital world
English
430
890
6.3K
359.1K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
@MRpoproppo Silicon comes from sand. Can most people still claim they're smarter than a B200? 🤔
English
1
0
3
75
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
When someone says your brain is filled with sand, it’s actually a compliment.
English
10
1
74
3.4K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
The company that can build its own infrastructure will eventually win, since it can turn things that would otherwise be unscalable into scalable ones.
English
13
14
286
5.9K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
One of the fundamental limitations of compute is energy. Hence, there is a biological limit to our brains before they get too hot or inefficient. The same applies to robots, especially if they need to move—there is a physical limit to the thermal management and battery mass you can associate with the computation needs before running out of room for other mechanical parts. Deploying 2M H100s would roughly take 3-4 GW. In contrast, 2M Tesla vehicles running around while providing essential heating and comfort to the human inside take less than 1/10th of that energy. Thus, real-world AI is bound by the joules you could carry with you.
English
15
5
240
8.2K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
Amazing abundance will be unlocked when collections of atoms and bits can both self-improve each other.
English
5
9
201
4.9K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
I often think about our winter testing picture. It represents the beauty of the vehicle and the hardness of our grit. Beauty can grow from hard work.
Yun-Ta Tsai tweet media
English
23
32
778
12.4K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
@elonmusk It has been helpful for comparing products I want to purchase.
English
3
0
44
3.1K
Mike
Mike@__red_dragon__·
@yunta_tsai I’m sure those Lamborghini owners can hardly wait to take a Cybercab ride share rather than their Lambo. 😂😂🤣
English
1
0
2
228
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
Even people in Lamborghinis turned their heads at the Cybercab. The radiant magic is like the golden ticket from Charlie’s Chocolate Factory that inspires wild imagination.
English
14
17
485
11.6K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
Humans are the only species that know how to make tools to improve their tool-calling skills in order to lift beyond their limited neuron capacity. This becomes a self-improving loop. The same applies to digital AIs that know how to code and physical robots that know how to build.
English
8
6
168
5.5K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
@dannywinner1 Early morning you will be fine. Don’t go there in the middle of the day.
English
2
0
3
102
Danny
Danny@dannywinner1·
@yunta_tsai I’m going to Yosemite soon and have been nervous about half dome… not because of my own competence but the competence of ppl ahead of me. If they slip, they take out people below them. 😬. Probably some leadership lesson in there.
Danny tweet media
English
2
0
3
126
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
El Capitan seems impossible to climb, but people have climbed it multiple times, even without ropes. Hard problems might be intimidating. But you can either scale them yourself or watch someone else climb from below.
English
3
2
135
4.1K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
Someone should make AI Oscars that host with AI videos. I'm sure the ratings will be more entertaining than the real one. In an age where people yearn to be entertained instead of being lectured by celebrities, what a missed opportunity.
English
11
0
162
5.5K
Yun-Ta Tsai
Yun-Ta Tsai@yunta_tsai·
Our teams hate driving. So whenever there's a new vehicle bring-up, we work extra hard until it can take us home and back—not for demo, but for our sanity.
English
31
19
599
32.5K