Jan 🌐

1.1K posts

Jan 🌐 banner
Jan 🌐

Jan 🌐

@spacebraniac

Space 🚀 and EV lover 🚗⚡. World is a simulation, but let's enjoy it.

Medical school Katılım Eylül 2016
173 Takip Edilen220 Takipçiler
Sabitlenmiş Tweet
Jan 🌐
Jan 🌐@spacebraniac·
I have finally found a way to make FSD understandable to anyone and why it will happen all of a sudden and why the Bot will be able to do a lot of stuff at nearly the same time as FSD, and why you can also disregard the current performance besides the general trend and goal. You have to look at the AI stack that tesla is building as a bigger picture. You have to look at it the same way we look at Chat GPT. For us to get here it took decades of research to get GPT2, then GPT3 and now GPT4, each subsequent one had more parameters better understanding of what it was indended and better output the same thing is happening with what we call FSD, but is really the "real world AI" that tesla is building. Every time they rewrite the whole stack it is like going from eg GPT2 to GPT3, they went from single frame single camera (each used separately) input, to multi camera but still single frame input, then to multi camera multi frame (video), to then the "birds eye" and whole topology mapping, now diffusion and then they will do another thing and another. FSD and the Bot are like plugins with chat GPT (obviously even bigger but the idea is the same), so they need to build a good enough AI stack that will get them the appropriate info about the world for them to be able to complete FSD otherwise it cannot be done. For example you could try to make plugins work on GPT3 but they would be clunky and have issues and even worse if it was GPT2 for example, the same way for the FSD and Bot to work they need that information they get to be good enough for them to be able to build something spectacular on that. This is the breakthrough moment, when they will build that AI stack that will suddenly make it much easier, faster and better to make FSD. This is what these big releases are when they rewrite the code. It is hard for us to grasp the significance of the work they do in the background, but these changes are as big as the difference between different GPT versions, and one day it will happen for FSD suddenly. This is why Elon thought that they were close so many time, it seems that the FSD part is actually easy, many have done it to an extent, but everytime Elon was convinced that this was the version to do it, but it has not been. So this work in the background is crucial. And once they have this sorted they can do whatever they please with it.
English
6
14
111
27.5K
Jan 🌐 retweetledi
Infodex
Infodex@infodexx·
Europe’s English Skills 🇪🇺 Very High Proficiency 🇳🇱 Netherlands — 636 🇳🇴 Norway — 610 🇸🇪 Sweden — 608 🇭🇷 Croatia — 607 🇵🇹 Portugal — 605 🇩🇰 Denmark — 603 🇬🇷 Greece — 602 🇦🇹 Austria — 600 High Proficiency 🇩🇪 Germany — 598 🇷🇴 Romania — 593 🇧🇪 Belgium — 592 🇫🇮 Finland — 590 🇵🇱 Poland — 588 🇧🇬 Bulgaria — 586 🇭🇺 Hungary — 585 🇸🇰 Slovakia — 584 🇪🇪 Estonia — 578 🇱🇺 Luxembourg — 576 🇱🇹 Lithuania — 569 🇷🇸 Serbia — 568 🇨🇿 Czechia — 567 🇨🇾 Cyprus — 558 🇨🇭 Switzerland — 550 Moderate Proficiency 🇬🇪 Georgia — 543 🇧🇾 Belarus — 539 🇪🇸 Spain — 538 🇲🇩 Moldova — 536 🇺🇦 Ukraine — 535 🇦🇱 Albania — 533 🇷🇺 Russia — 532 🇮🇹 Italy — 528 🇫🇷 France — 524 Low Proficiency 🇹🇷 Türkiye — 497 Source: EF English Proficiency Index 2024
Infodex tweet mediaInfodex tweet media
English
228
177
2.8K
705.1K
Jan 🌐 retweetledi
John Kraus
John Kraus@johnkrausphotos·
See all of my photos + order prints from today's stunning evening launch of Starship Flight 11, the final flight of the V2 era ahead of future V3 flights from Starbase → johnkrausphotos.com/Starship 11/11 on capturing all flight tests thus far. See photos from all of them above!
John Kraus tweet media
English
40
238
796
65.3K
Danny2K
Danny2K@Danny2KYT·
My shoe game is so washed rn all I got is a ripped pair of new balances any recommendations
English
11
0
29
1.7K
Jan 🌐 retweetledi
LakersMuse
LakersMuse@LALMuse·
🚨 If the Pacers win game 7, I will giveaway a Luka Doncic jersey To enter: Follow @LALMuse Repost this Join the LakersMuse community on my profile Winner will be picked after the game if Indiana wins
LakersMuse tweet media
English
18
130
206
20.1K
Jan 🌐
Jan 🌐@spacebraniac·
@hayasaka_aryan We have to get the guy who managed to calm Elon down to now decommission Spacex dragon he can fix anything hahaha
English
1
0
86
13.3K
Jan 🌐
Jan 🌐@spacebraniac·
@wearetherace We all got it wrong, he accually cursed the left barrier. 🤪
English
0
0
49
2.5K
The Race
The Race@wearetherace·
Nico Rosberg may have just cursed Turn 1
The Race tweet media
English
28
98
4.4K
341.2K
Jan 🌐
Jan 🌐@spacebraniac·
@X36gg @Tratnikstan Same in NBA, while it is exciting to see so many different winners, 7th different champion in a row, it's actually dynasties and domination and drives the highest TV ratings.
English
0
0
2
370
Sebastian
Sebastian@X36gg·
Pogi is 100% good for cycling. He’s the first internationally well known star since Armstrong, the first cyclist to be in Merckx arguments. Cycling fans are in love with “racing for the sake of having a race” and we so often have random winners which establishes nobody. It’s also IMO bad to have the undisputed GOAT from 60 years ago, because it feels like the sport has moved backwards. In an era in which riders specialize themselves for certain races, Pog is quite refreshing. If you watch tennis, the Big 3 won everything and it was great for men’s tennis, while the number 1 problem for women’s tennis in the late 2010s was the enormous number of random winners. If the world number 38 wins a Slam and then never reaches a QF, sure it’s fun once, but if it keeps happening, it’s bad for the sport. When somebody actually beats Pogi, it will mean something. Same in boxing/UFC/WWE, when somebody beats the established champion, that win means more. And it’s not like Pog is boring. He COULD actually be this super boring rider who waits for the last 2-3km to attack. But he doesn’t
English
8
3
86
5.5K
Tratnikismo
Tratnikismo@Tratnikstan·
There is a popular narrative that Pogačar is ruining cycling and there is some truth to it. BUT without Pog Jonas wins 4 straight Tours (avg gap of last 3 is 7+ mins!!) + MvdP has absolutely 0 competition in the classics. Is cycling really better off without Tadej?
Tratnikismo tweet mediaTratnikismo tweet media
English
129
52
1.5K
179.6K
Jan 🌐
Jan 🌐@spacebraniac·
@KeithArbitrary1 @iztok_franko 💯, but I think it was diagnosed only in the latter half of the series, hard to say when he actually broke it
English
0
0
0
63
Keith
Keith@KeithArbitrary1·
@iztok_franko That can ruin your ability to get past people and throw your shot off completely. Austin doesn't strike me as an excuse maker.
English
1
0
2
94
Jan 🌐
Jan 🌐@spacebraniac·
Idc about it all, but if I remember correctly he did say that there might still be a smaller die, which is exactly what we got 🤷‍♂️. General Intel is definitely not killing Ark as it is still the core of their graphics, it's just about the discrete cards. Really hope Intel does something with celestial and druid, would be good for everyone.
English
1
0
0
378
Moore's Law Is Dead
Moore's Law Is Dead@mooreslawisdead·
Fake leakers exposed again. The RX 9060 XT is the exact price I said it would be kids.
Moore's Law Is Dead tweet media
English
27
30
475
21.2K
Jan 🌐
Jan 🌐@spacebraniac·
@anegi0577 @taedrael @KingJames I completely agree, like yes it's a lot of money, but if LeBron really cared about winning he'd take a huge pay cut, I know it's extreme but imagine if he took the vet minimum 😜, like how much cap they'd have to sign amazing players, it would kinda be unfair hah.
English
1
0
1
68
Jan 🌐
Jan 🌐@spacebraniac·
@Hero_OfThe_Day @Shark19272Shark Also I saw a stat somewhere that LeBron, Luka and Austin all had 13 potential assists, just Austin and LeBron's passes were managed to be converted 3times, while Luka's were made only once, I seriously doubt that Luka's passes were so much worse, others just didn't convert.
English
0
0
3
136
HeroOfTheDay
HeroOfTheDay@Hero_OfThe_Day·
Luka Doncic always does this, it’s what leaders do. The numbers are available. It was Lebron and Austin Reaves who passed less than their average, as Luka Doncic passed only 2 times fewer than his normal amount. He passed more than Anthony Edwards who finished with 8-9 assists. He can’t pass the ball and make the shot for teammates. That’s ridiculous. In fact, given how abhorrent everyone else shot the ball, he probably should’ve had more attempts.
English
1
1
21
564
HeroOfTheDay
HeroOfTheDay@Hero_OfThe_Day·
Reggie Miller: “Luka Doncic had 37, but only 1 assist. Lebron wasn’t able to get in rhythm!!” Luka Doncic had 22 attempts, Lebron had 18. They don’t do this with any other star player.
English
22
37
687
17.9K
Jan 🌐
Jan 🌐@spacebraniac·
? James has been saying this for the whole time, FSD is a great product, it helps a lot on a daily basis, it is extraordinary, but sadly, for it to have any serious monetary value, especially anywhere close to it potential, it will have to pass a benchmark that hasn't been reached by anyone and it is sadly years away despite what some people are saying.
English
0
0
1
67
James Cat
James Cat@TSLAFanMtl·
I agree - it's something that needs to be experienced to really understand the convenience and utility of it. I am trying to think about this practically and objectively. I am sure about 1 thing: Tesla has not yet found a way to monetize it to anywhere close to its potential. What is the potential? That's a function of good it works/how safe it is on a fleet-wide basis. Does it make me more bullish from an investing perspective? A little, but I still need to see evidence that it can drive this well over thousands of kilometers/miles, consistently. The issue is not that FSD is bad, it's that the bar it needs to hurdle is so high.
The PandaBear@ThPandabear

@TSLAFanMtl It’s hard to put the FSD experience into a spreadsheet - esp since it’s still evolving. Does your experience make you more bullish? 95% of people just have no idea and will be blindsided when unsupervised FSD starts

English
50
9
226
33K
dzs🇭🇺
dzs🇭🇺@sighaboutlife·
last square is fucking me up
dzs🇭🇺 tweet media
English
1
0
5
188
Jan 🌐
Jan 🌐@spacebraniac·
Because this is a sink of swim situation. Obviously nobody is advocating to do stuff with AI 24/7 unless you want to, but the fact is that we if start relying on AI for everything this will mean one of two things for most, either you know how to utilise for the best outcome or you don't. It will create a much bigger divide between people, a whole lot of people will be resigned to the lowest jobs, or even out of the job, while the more intelligent/better prepared will be able to at least someone weather the storm and go with it. You can say that people should just do what the love, but I'd argue that most people don't really have anything, and one of the main purposes in life if their work, so if a giant part of the population is obsolete and can't really adapt with the changing technology it will not end too well. People cannot become complacent, to a degree it's just like the people who said Internet was just a fad in the 80s, I wonder where they are now.
English
1
0
1
77
gon
gon@chinesegon·
it’s so over
gon tweet media
English
124
339
14.1K
786.4K
Jan 🌐
Jan 🌐@spacebraniac·
Ok, then why bother to do anything at all in life. Why should we even bother with the daily stuff if we are going to die either way, if not sooner then with the end of the universe. This is essentially already making humans the pets of AI, or at least a serious underdog vs it. I wonder what happened every time some superior species found a lesser one, or a much more advanced civilisation found a primitive one. Giving up learning anything is just accelerating our demise. An uneducated person cannot use AI, even if it's the smartest thing in the universe. A smart person can get so much more out of AI than someone who doesn't know how to properly use it, this will only hasten and deepen the differences between the top and the both of humanity. Nihilism at its finest...
English
1
0
6
97
Matt Izzo ⚡️
Matt Izzo ⚡️@mattizzo·
@hillomutin1 @chinesegon Maybe but will what you learn still be relevant in 6-24 months when AI itself can do this better than you could even dream of?
English
1
0
1
471
Jan 🌐
Jan 🌐@spacebraniac·
@Guy_Hollington @TSLAFanMtl While it's totally possible, the issue is that while you and many others are having a great time, there are loads that. Example 👇 x.com/actuallyajoke/…
RBOTAXI DRIVER@actuallyajoke

@TSLAFanMtl Also, the numbers shown in the community tracker are averages. I use the tracker for recording my own data (it’s the easiest way) and on the latest FSD v13.2.8 it’s currently showing 25miles/disengagement for solely my data, roughly the same number since I started tracking

English
0
0
1
31
Guy Hollington
Guy Hollington@Guy_Hollington·
@TSLAFanMtl This data doesn't seem valid. For me V13 has been far, far superior to V12. I'm not tracking it imperially, but from personal experience I don't believe the data in the graph shown.
English
1
0
10
504
James Cat
James Cat@TSLAFanMtl·
So far, FSD hasn't driven any incremental demand for Tesla in the US or Canada despite the much improved performance of v13. It's just not good enough (yet) for the consumer to see the value in paying $8k for it. So either price falls dramatically, or it gets a lot better. As we saw with v12, after a solid increase in performance early on, later versions saw performance drop - per the FSD tracker. So while with a small sample or on an individual basis, a person who drives 250-500 miles may not need to intervene - when pooled together, a rate of 1 intervention per 250-500 miles is very far from what is needed for autonomous driving.
James Cat tweet media
English
103
10
248
28.3K
squawksquare
squawksquare@squawksquare·
@elonmusk is out here posting like an unsupervised toddler on a sugar high. Political rants, conspiracy theories, and playground-level tantrums—it's embarrassing. At this point, he needs a digital timeout (or a metaphorical spanking) before he does more damage to Tesla’s brand. As I predicted last year, Tesla's brand is now significantly being impacted by his immature actions. #muskmeltdown #squawkrightagain $TSLA
English
29
2
61
8.3K
Jan 🌐
Jan 🌐@spacebraniac·
@hunk @LinusTech Isn't it obvious that AD is the 3090 as he is obviously bigger
English
1
0
67
2.2K
Edgar G 🥇
Edgar G 🥇@hunk·
@LinusTech Just to confirm, you mean AD is a 5090 and Luca 3090? if yes, really!? again really!!?
English
2
0
49
10.3K
The Mavs Blog
The Mavs Blog@TheMavsBlog·
After reading Iztok’s piece, I’m even more heartbroken. Nico Harrison forced out everyone around Luka. Casey Smith Igor Kokoškov Marko Milić Anže Maček Javier Barrio Calvo And then forced Luka out himself.
76
356
4.6K
302.1K