golol

15 ms·

181.
by golol 2y ago
Sorry but you just didn't inform yourself. The dimensions for the luggage are very clearly stated. Nowadays I would expect it to be common knowledge that when flying with a budget airline you have to pay attention to your luggage size
182.
by golol 2y ago
I am not an expert on this so I can obviously not give a good judgement but this is about Cohomology and p-adic numbers. Whatever it is they may prove, they are clearly furthering our understanding of these subjects. When you want to apply
183.
by golol 2y ago
mathematicians work on problems which are interesting to other people, usually mathematicians. Sone people dont seem to be aware of this, but this is a graph with a general direction trend from the pure to the applied. Show me any math pape
184.
by golol 2y ago
I think Musk just wanted to make the point that there are many ways to achieve rapid and reliable reusability but they have chosen this one, they are laser focused on making it work, and they know they can make it work. So other ways of doi
185.
by golol 2y ago
Lmao how amny other companies do you buy products from that have shady business strategies or CEOs with unacceptable views? At least Elon doesn't hide it.
186.
by golol 2y ago
Lol I'm pretty sure FSD generates more revenue than Waymo.
187.
by golol 2y ago
Mentioning autopilot in an FSD discussion shows you are not familiar with the subject.
188.
by golol 2y ago
If aluminum is the issue, could steel and plastics be a solution? Especially with Starship coming online, weight should be less of an issue.
189.
by golol 2y ago
Ok wow that's pretty strange.
190.
by golol 2y ago
As I understand, the worm cells act differently because the worm is placed in a solution which interferes with their electrical messaging, so it makes sense that the offspring growing up in the same solution would grow with the same defects
191.
by golol 2y ago
Exactly. Everyone was worried about reentry, but perhaps more concerning than the question of whether the headshield tiles work was the question how well the material below can handle failures. Now we know significant failures of tiles do n
192.
by golol 2y ago
The only problem with Taxis is that they are expensive und possibly not available. Both of these issues are very much the kind of thing a robotaxi might fix.
193.
by golol 2y ago
Ok you want the general answer? Consider a discrete time Markov process with memory length N on a finite state space. Train a transformer with context length N on sample trajectories with SGD. Can you expect the transformer to become a good
194.
by golol 2y ago
As I understand, conceptually they just changed 346 + 23 = ? to (1: 3, 2: 4, 3: 6) + (1: 2, 2: 3) = ? So it is not that much of a specific hack. There could be a broader principle here where something is holding transformers back in a gener
195.
Grothendieck’s use of equality(arxiv.org)
178 points by golol 2y ago|129 comments
196.
by golol 2y ago
Because the spirit of Apollo - unsustainable one off dlag planting missions - lead to human spaceflight stagnating for the subsequent half century.
197.
by golol 2y ago
This is what I want Starship to succeed for.
198.
by golol 2y ago
Yea we don't have any science of intelligence, the only thing we have is empirical data. Testing to see what works. That's why Turing tests are quite fundamental imo.
199.
by golol 2y ago
So why can't you replace 99% of callcenter calls (<5min) with AI right now?
200.
by golol 2y ago
The Turing test insight is that text is a sufficient medium to test for AGI. And this still holds true.
201.
by golol 2y ago
I a 5 minute casual conversation. Also the statistics between human and AI were different in some regard (like 48% vs 56% for some quantity), I dont recall details. Look the Turing test is very different depending on the details, and I thin
202.
by golol 2y ago
The Turing test has not been passed
203.
by golol 2y ago
It also bothers me and I wrote down my thoughts on it a while ago, maybe you find it interesting: https://news.ycombinator.com/item?id=36127880
204.
by golol 2y ago
I think 4o is actually noticeably smarter than 4, after having tried it a tiny bit on the playground.
205.
by golol 2y ago
not really. Even a human bad at reasoning can take 1 hour of time to tinker around and figure things out. GPT-4 just does not have the deep planning/reasoning ability necessary for that.
206.
by golol 2y ago
What do you mean?
207.
by golol 2y ago
Ah I misremembered GPT-3.5 as being released around the time of ChatGPT.
208.
by golol 2y ago
Yes you're right about the compute. Let me try to make my point differnetly: GPT-3 and GPT-4 were models which when they were released represented the best that OpenAI could do, while GPT-3.5 was an intentionally smaller (than they cou
209.
by golol 2y ago
It doesn't make any sense to look at it that way. Apparently the GPT base model finised training in like late summer 2022, which is before the release of GPT-3.5. I am pretty sure that GPT-3.5 should be thought of as GPT-4-lite, in the
210.
by golol 2y ago
GPT-2: February 2019 GPT-3: June 2020 GPT-3.5: November 2022 GPT-4: March 2023 There were 3 years between GPT-3 and GPT-4!

More ›