置顶推文
Less
1.8K posts
Less
@lessaxwrong
He who looks outside dreams, He who looks inside awakens ~ C.G Jung
加入时间 Mayıs 2024
154 关注81 粉丝
@frontier_foid Unfortunately/fortunately not really true anymore, I haven’t tried it on those specific things but guardrails seem to be way up even for normal generations on the app
English
> we’re cranking day and night on v2 to make it even stronger
one could say grok users are also "cranking day and night" 🤭
Keyang Xu@keyang_xx
We’re back to No. 1 on the leaderboard! 🎉 After months of hard work, what a journey! It’s also the best option for fast and low-cost generation. Meanwhile, we’re cranking day and night on v2 to make it even stronger. Try it and let us know what we can improve — your feedback is invaluable. → docs.x.ai/developers/mod…
English
A lot of improvements have been made. It’s worth a try.
Arena.ai@arena
Grok-Imagine-Video-1.5-Preview (720p) has landed #1 in the Image-to-Video Arena! This is a massive +52 pt improvement over Grok-Imagine-Video (720p), surpassing the best video models Seedance-2.0 and HappyHorse. Congrats to @xAI and @elonmusk on this big achievement!
English
Please make a way so api is accessible through the app somehow, also please consider having a “pro” mode in the app
I would like it to have more mid journey controllability, somehow being able to add multiple angles of the same object and the model take that into account would be nice and sref that works
English
We’re back to No. 1 on the leaderboard! 🎉 After months of hard work, what a journey! It’s also the best option for fast and low-cost generation.
Meanwhile, we’re cranking day and night on v2 to make it even stronger. Try it and let us know what we can improve — your feedback is invaluable.
→ docs.x.ai/developers/mod…
Arena.ai@arena
Grok-Imagine-Video-1.5-Preview (720p) has landed #1 in the Image-to-Video Arena! This is a massive +52 pt improvement over Grok-Imagine-Video (720p), surpassing the best video models Seedance-2.0 and HappyHorse. Congrats to @xAI and @elonmusk on this big achievement!
English
A recent update from the team, enjoy!
Arena.ai@arena
Grok-Imagine-Video-1.5-Preview (720p) has landed #1 in the Image-to-Video Arena! This is a massive +52 pt improvement over Grok-Imagine-Video (720p), surpassing the best video models Seedance-2.0 and HappyHorse. Congrats to @xAI and @elonmusk on this big achievement!
English
It’s kind of strange that you can do gradient descent on prompts despite no back-propagation.
English
@Simberg_Space @Robotbeat @_abbie_watson_ The fastest time is actually like 4 months for the starship flight 1 “pad” which was obliterated, but still not much damage on tower. Tower 2 (which spacex did very fast) was 1 year
Also if blue has to do concrete work they have to do it first I doubt it is parallelizable
English
@Robotbeat @_abbie_watson_ The question is how many things can be done in parallel versus sequential. Can't get a baby in a month with nine women.
English
CSI_Starbase betting against Blue being able to rebuild the pad in 6 months. We're going to get to find out whether building a pad is more of an engineering problem, or a supply chain problem.
(Key detail being Bezos runs the most sophisticated supply-chain operation on the planet.)
Zack Golden@CSI_Starbase
Absolutely not . There is no way they could possibly rebuild faster than SpaceX rebuilt Massey’s. Keep in mind SpaceX had the remarkable luck of having a brand new team dedicated to mass producing launch pads when that incident happened. No one else in the industry has a team like that. Not even close. Even with that massive advantage, it took 8 months to get that test stand operational again. And that test stand that doesn’t feature the tallest launch tower and water tower in the world. Expecting anything less than 1 year is far too optimistic. I’d be shocked if they accomplish it in less than 18 months. I’m not betting against them, in fact I hope I’m wrong…but we should try to be more realistic with these expectations.
English
These things are not as related as you might think, for example spacex has not much coordination with self driving cars. Spacex does own X but I doubt it helps them that much in building a pad
A good way to think of it is how much revenue has driven into the system, spacex has extensive supply chains for the things they do outsource (for example a large number of reputed cnc shops have done work for them), they would also have extensive contacts with construction companies that have worked for them in the past on a large number of projects. Spacex has simply done more projects and built more and their access to external supply chains is massive. They also inhouse a lot of things and have capability for it
It is also my guess that manufacturing is the critical thing for building a pad, not really logistics, it is not difficult to source concrete and steel. And amazon (which is again a different company than blue) having distribution centers doesn’t matter much for building a pad. Btw making a car profitably is more logistically challenging than amazon
@grok fact check this is this true?
English
English
Fact check: Your take is solid.
Blue Origin is separate from Amazon. AWS use aids sims but Amazon's distribution expertise doesn't apply to specialized launch pad rebuilds (custom fab, flame systems, etc.). Concrete/steel easy.
SpaceX leads with proven supply chains, in-house work & fast Starbase-style construction/rebuilds of pads & test stands.
Tesla FSD/X minimal help. Car logistics different challenge.
Blue's recent pad explosion means months of rebuild; engineering experience > general supply chain here. SpaceX positioned better.
English
@_abbie_watson_ From what I know It doesn’t, in any aspect pretty much. What is your reasoning for saying that? Blue origin is also not amazon
English
@lessaxwrong I think people are going to be surprised to find out that Blue Origin (Amazon) has better supply chain and orchestration than SpaceX.
English
@techinvestoor I am not short but I don’t really get what the superior technology is or what it enables compared to starlink d2d
English
$ASTS
What this guy and all of the “they won’t meet their timeline” gurus miss is it’s not about the timeline.
It’s about superior technology.
Shorts learned this very same lesson with $TSLAQ.
I lived (and profited from) it.
Stuart Taylor@LuckyStuey
$ASTS classic SpaceMob in action here, compounding one sets of lies upon another to create a fantastical scenario.
English
Barron's: "...the expected midrange market capitalization of SpaceX's initial public offering ($1.75 trillion), divided by the company's 2025 revenue ($18.674 billion), with the quotient of 93.71 being SpaceX's price-to-sales ratio. Which is ridiculous. (The S&P 500 index's ratio is just 3.38 -- and that's with stocks at record highs.)...
To have a company that has lost more than $8.5 billion over the past three years potentially carry a valuation north of $1.5 trillion certainly defies typical valuation metrics."
English
@lessaxwrong Not saying it's technically impossible, however sometime now next year New Glenn might be operational again.
I'm thinking in the context of urgency of still launching Mk1 on schedule while the investigation and rebuild take place
English
Every time something like this presents itself some people act like SpaceX isn't an engineering powerhouse perfectly capable of overcoming such "hurdles" in a timely manner.
Examples:
Added methane to the pad to fuel Nova-C
Modified the fairing for Cygnus access
Designed and built a new dispenser for OneWeb
Modified Cargo Dragon trunk to reboost the ISS
MacGyvered an adapter to test Starship on the orbital launch mount after S36 exploded
Designed a better docking hardware for Dragon after seeing what NASA had to offer
They can probably have a tank delivered (or maybe salvaged from Blue Origin themselves if any are still usable) and hydrogen retrofitted to one of their pads in a couple of weeks if asked to.
Chandler G@Echo5550
All this talk of being saying “just put MK1 on Falcon Heavy” is giving me PTSD of the Bridenstack situation. To put MK1 on Falcon Heavy there are still hurdles you would have to go through that would also still take time. You would need to somehow fuel MK1 at 39A when there is no Hydrogen lines. MK1 also is too large to fit in the fairing so SpaceX would need to design a new fairing to fit it.
English
15 months from now is August 2027, again I think you are assuming everything will go right. New glenn had a second stage messup last flight, I am not sure it will automatically mature and sort out all it’s issues on the ground. It can still be a problem. Instead if they tell spacex to make a cargo version and it can go much smoother, spacex can transfer learnings from every starship flight which will they have many
English
Not this year but same time now next year it can work.
Blue mk2 doesn’t fit heavy and Berger says 15 months for the pad, IF blue does everything perfectly it will be a tight fit and if it delays china will probably be on the moon first.
If they have the hydrogen kerosene incompatibility with heavy for mk1 it’s worse, NASA has to act now to either options and not wait for how things pan out. Imo they should tell spacex to go forward with cargo starship soon even if blue does it
English
@lessaxwrong I don't think that's a viable option if you want to still launch Mk1 this year.
Falcon Heavy seems the most straightforward option
English