Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
smaddox
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
14 ms
·
121.
▲
by
smaddox
4y ago
Cool. Stereoscopic diffusion images coming soon.
122.
▲
by
smaddox
4y ago
Sounds interesting. Another recent paper on efficient architecture for long context lengths: https://arxiv.org/abs/2302.10866
123.
▲
by
smaddox
4y ago
"Please continue without explanation." Usually works for me.
124.
▲
by
smaddox
4y ago
> To behead Ma Su (name) while crying. [...] I wonder how much further ahead of western cultures, eastern asian cultures would have been if they didn't kill (or commit suicide) every time someone made a mistake.
125.
▲
by
smaddox
4y ago
> GPT does not model subjects. GPT does not even model words! It models tokens. The first and last layers of a transformer decoder model tokens. The hidden layers don't have this restriction. There was a paper recently showing that
126.
▲
by
smaddox
4y ago
If you're creating the normal maps yourself, perhaps you would benefit from generating the image from that using https://github.com/lllyasviel/ControlNet#controlnet-with-nor...
127.
▲
by
smaddox
4y ago
Your conclusion is not supported by your premise. Reversion to a mean implies convergence to a stable, unmoving equilibrium. This requires that all forces are restoring forces in the vicinity of that equilibrium. While Earth's climate
128.
▲
by
smaddox
4y ago
> IME Rust is slower to write For a backend webservice prototype, that's probably true. It's not true for everything, though. Especially if there are performance requirements or you need to handle errors properly. I've fou
129.
▲
by
smaddox
4y ago
> the solution is not limited by cost I'm not so sure... The valuation seems a bit absurd to me. Using the same valuation system, what would be the valuation of Earth? Cost of extraction and transportation is still (and always) a co
130.
▲
by
smaddox
4y ago
Ion escape might be a significant consideration, but given the plasma density and magnetic field strength, I'm guessing it wouldn't degrade the efficiency too much. Presumably they've done the calculations. I have not. As for
131.
▲
by
smaddox
4y ago
Helion as already demonstrated recovering 95% of the energy from non-fusable plasmas and Q << 1 plasmas. Thousands of times. It's perhaps worth noting that the high speed, high power electronics required to implement their approa
132.
▲
by
smaddox
4y ago
For AR passthrough, you would need some fairly complex processing to get video from an event-based camera. Event based cameras are definitely exciting for machine learning applications, though.
133.
▲
by
smaddox
4y ago
If they succeed to the degree they aim to (and there's reason to believe they will), electricity from their generators will be an order of magnitude cheaper than any existing source.
134.
▲
by
smaddox
4y ago
Depends what you're prototyping. Rust is hard to beat for prototyping an interpreter, for example.
135.
▲
by
smaddox
4y ago
Not that I'm aware of, but that's a very interesting idea. If you had one that maps from ui to html with inline styling, you could automate turning image mock-ups into html with inline styling. It's not clear exactly how you
136.
▲
by
smaddox
4y ago
It's not inherently bad. It measures what it measures. It's just a terrible measure of prosperity. See, for example https://www.weforum.org/agenda/2018/01/gdp-frog-matchbox-dav...
137.
▲
by
smaddox
4y ago
Agreed. But I don't think it would take a herculean effort to do better than GDP.
138.
▲
by
smaddox
4y ago
Private debt dwarfed public debt until very recently, and it's still significany higher: https://braveneweurope.com/steve-keen-what-is-the-role-of-pu... Also GDP is a terrible proxy for economic prosperity. A broken wi
139.
▲
by
smaddox
4y ago
I agree that ChatGPT's ability to generate convincing arguments for complete BS is concerning. But I'm not convinced it's a clear negative, in the long run. Humans have been concocting convincing arguments for complete BS for
140.
▲
by
smaddox
4y ago
Yeah, I've found the C4 model helpful for finding the appropriate level of abstraction for architecture documentation: https://c4model.com/ Code documentation, in contrast, should be auto-generated from the code.
141.
▲
by
smaddox
4y ago
> it could be used to keep cold drinks cold and warm drinks warm The classic dad joke: So if I put my cold beer and my hot soup in there, how does it know which is which?
142.
▲
by
smaddox
4y ago
I'm not sure that's entirely true. Scientists have successfully sliced, imaged, and analyzed (i.e. determined the connectome for) the entire brain of very small animal, the common fruit fly, Drosophila. We're still a long way
143.
▲
by
smaddox
4y ago
Copyright != Trademark. Independently generated images of Mickey Mouse are litigated under Trademark law, not Copyright law. Disclaimer: I am not an IP lawyer and have roughly no idea what I'm talking about.
144.
▲
by
smaddox
4y ago
Agreed. This is overblown. but they haven't even begun to apply RLHF (or at least they haven't published anything). I think it's safe to say that combining an LLM with RL, all existing techniques, would result in a much stron
145.
▲
by
smaddox
4y ago
I'm curious, have you considered using PyTorch or JAX for tensor processing? ML libraries seem to be much further along when it comes to performing compute-intensive, hardware-accelerated operations on Tensor's. And you get gradie
146.
▲
by
smaddox
4y ago
> As participants recalled the images, the receptive fields in the highest level of visual processing were the same size they had been during perception — but the receptive fields stayed that size down through all the other levels painti
147.
▲
by
smaddox
4y ago
He clarifies that infinite sequences are valid, but all infinites are countable. It's an appealing view. Rather than the reals being uncountably infinite, they're countably infinite. Irrational numbers are uncomputable, but all ra
148.
▲
by
smaddox
4y ago
> But neither ChatGPT nor some larger descendent model will ever be able to write the most difficult pieces of our software given natural language descriptions of desired functionality. Strong assertion that I think will be proven wrong
149.
▲
by
smaddox
4y ago
Plutonium is definitely able to go critical: https://en.m.wikipedia.org/wiki/Demon_core A plutonium sphere of roughly 10 kg is sufficient.
150.
▲
by
smaddox
4y ago
Yeah, its really bad. I keep having to disable 5G.
More ›