blagie

49 ms·

301.
by blagie 3y ago
SpaceX raised $750M at a valuation of $137B in January 2023. A private investment at this scale should have a lot more transparency and due diligence than disclosures from a SEC disclosures. If I were investing $750M, I'd have engineer
302.
by blagie 3y ago
Having been down this road a few times, this: > What I've heard is Azure is a pain in the ass, and things take three times as long to set up there, for some reason. Doesn't matter. The cost here is a rounding error. What does m
303.
by blagie 3y ago
I'd add quotes there: Pytorch has been "supporting" rocm for all last 2 years
304.
by blagie 3y ago
Genuine answer: Setting up Facebook WOULD be extremely challenging for a company like SpaceX. There's a reason Facebook is worth about 10x what SpaceX is worth, and most of that value doesn't come from the ability to build softwar
305.
by blagie 3y ago
Good point! I think my assumption came from two places: (1) My own bias (I use Zigbee) (2) Zigbee is big enough to have been sold in big box stores. However, a rewrite which explains the global context would be even better, if it can be don
306.
by blagie 3y ago
First of all, this is not an attack or intended to be negative. I often get attacked or downvoted for constructive feedback on Show HN (where that's what it's here for). This is intended as feedback. I give feedback to the best pr
307.
by blagie 3y ago
I haven't found many places where I trust the results of an ML algorithm. I've found many places where they work astonishingly well 30-95% of the time, which is to say, save me or others a bunch of time. It's been years, but
308.
by blagie 3y ago
Technically, yes! A more reasonable answer, though, is "no." I've technically gone through random tutorials and trained various toy networks, including a GAN at some point, but I don't think that should really count. I a
309.
by blagie 3y ago
Well, no, but yes. The critical piece is that this can be done in training. If I collect a large number of C programs from github, compile them (in a deterministic fashion), I can use that as a training, test, and validation set. The output
310.
by blagie 3y ago
What generation? Academia was very honest for most of history. From WWII through roughly 1980-2000, there was almost exponential growth in the number of academic jobs available. At the time of WWII, about 5% of people had college degrees. T
311.
by blagie 3y ago
My personal field is irrelevant, since my work let me see across fields (for reference, most of my direct work as relevant to this discussion was in the social sciences, though -- that's the project which let me interact with large num
312.
by blagie 3y ago
That's not how it works. To be a Harvard academic, you need to do high-profile research. That has three components: - Academic ability - Willingness to cheat - Shameless self-promotion Competition is extreme. As a result, someone who i
313.
by blagie 3y ago
A hospital has many rooms, hallways, etc. Which is under positive pressure? If you have two patients, they should both be isolated from each other. Which one gets the higher pressure, and which the lower?
314.
by blagie 3y ago
No. That's a strawman. I think the point of an abstract should be to provide a summary easy for humans to quickly digest, including researchers skimming hundreds of papers and laypeople. I think the body and supplemental materials, in
315.
by blagie 3y ago
Why? HEPA doesn't matter unless you have a sealed system. An open door or a cloth curtain means the 99.97% means nothing relative to 99%.
316.
by blagie 3y ago
Filtration is a very good idea. HEPA filtration for residential use is dumb. A HEPA filter must remove at least 99.97% of aerosols 0.3μm in diameter. A filter which removed 99% and blows air 1% faster will remove just as much, or one which
317.
by blagie 3y ago
Now that I use Home Assistant, I want all my data sources to plug into there. It can handle the rendering for me as I see fit, and it's where data comes to integrate. It's one of those things which I wish existed, but I can't
318.
by blagie 3y ago
Pet peeve if you're writing research papers: "meta-analytic sHR of 0.22 (95% CI 0.17 to 0.29), 0.53 (0.44 to 0.63) and 0.45 (0.38 to 0.53), respectively, for 0–30 days after SARS-CoV-2 infection, while in the 91–180 days sHR were
319.
by blagie 3y ago
I'm wondering if we'll reach a point where every 787 and 737 MAX will need to be recalled, completely disassembled, and reassembled with normal quality-control and inspections in place. We're well past the point where it woul
320.
by blagie 3y ago
Speculation is very unhelpful, as are uncited claims. It's just not. FYI: Your assumption is wrong. The language used for instruction in most universities in ECOWAS is either English or French, and mostly English. If you're talkin
321.
by blagie 3y ago
Thank you! I received three responses. 1) One pointing to the linked promo video designed to give feel-good vibes, claiming that's a citation. (fnordpiglet) 2) One without citations, just claiming this was designed by "researchers
322.
by blagie 3y ago
Citation required and requested. I can plant a hundred trees, keep them up for a year, and find a hundred tree stumps in two. I genuinely would like to know where on the spectrum this is: - This is a fools' errand - This requires const
323.
by blagie 3y ago
I don't think that impression is accurate. Look at plots of: - Chinese per-capita GDP - Chinese overall GDP - Quality of education systems worldwide - Chinese current military power - Manufacturing might to support the above, if placed
324.
by blagie 3y ago
If this is correct, it's still worth understanding! “If you know the enemy and know yourself, you need not fear the result of a hundred battles. If you know yourself but not the enemy, for every victory gained you will also suffer a de
325.
by blagie 3y ago
Virtually nothing, except weight, was sacrificed to make the 18-35mm f/1.8 zoom. That's why it's impressive. It's possible due to things like: - Much better lens coatings (allowing for more elements without flare) - Comp
326.
by blagie 3y ago
That's not how it works. The wider the aperture, the worse the aberrations. IF you can make a lens usable at f/1.8, it will be tack-sharp stepped down. That's why wide primes have a reputation for being super-sharp. Roughly t
327.
by blagie 3y ago
Nits: > That depends. You may have a 30 line function which can be paper-worthy We have no disagreement. That's why I gave a pretty generic 3-line function. 90% of the upside of ML tools for me, by the way, is generating very routin
328.
by blagie 3y ago
Sort of. My universe broke when Sigma introduced an 18-35mm f/1.8 zoom. An f/1.8 zoom. Wow. And it was optically brilliant. Seventies lenses are super-sharp, but that's because they're mostly slow primes. Any modern prim
329.
by blagie 3y ago
As much as I agree with the conclusion (my AGPL code was stolen by OpenAI, as was a lot of other code), your logic is extreme. This is not good. A more reasoned argument is more convincing. 1. "Fair use" It is fair use to use snip
330.
by blagie 3y ago
Hydrogen or helium would decrease the density, but I digress. In most cases, composites have complementary materials. You would need less titanium if it was reinforced with something else. Examples: - Most common are materials like carbon f

More ›