moultano

11 ms·

151.
by moultano 5y ago
No way. We need nearly everyone to get the vaccine, with the abstainers purely randomly distributed, in order to reach herd immunity, which we aren't going to reach, so it's going to keep spreading endemically, indefinitely. If yo
152.
Would it be possible to detect industrial civilization in the geological record?(cambridge.org)
1 points by moultano 5y ago|0 comments
153.
by moultano 5y ago
That seems like a compelling use case. The next thing I wonder though is whether that can exist without the ponzi scheme on the side. Is it the remittances, or the ransomware and speculation that funds the infrastructure?
154.
by moultano 5y ago
Should I conclude anything from the fact that almost all of the products with high benzene contents are sprays instead of lotions?
155.
by moultano 5y ago
It's really hard to design something like that that anyone would get value out of using because matching the query isn't a binary notion.
156.
by moultano 5y ago
Those are good examples! I'll pass them along to debug. I think what's happening is that the wording is ambiguous enough that it's colliding with concepts like "how well does valve index work." If you search for &qu
157.
by moultano 5y ago
I did not take Jonathan Blow's tweet to be saying it's a bad idea, or won't work, but rather that it's an indictment of the whole web stack that it's necessary.
158.
by moultano 5y ago
It entirely depends on how much CO2 and methane we emit, but these sorts of predictions are by convention pegged to 2100. Most of the emissions pathways, unless we stabilize things soon, keep getting hotter until at least 2200.
159.
by moultano 5y ago
A rough rule of thumb is that for every degree global temperatures increase temperatures on land will go up twice that amount. So with 3.5°C, which is on the low end of what we're headed for, imagine if peak summer temperatures were 13
160.
The Things They Won't Believe(moultano.wordpress.com)
1 points by moultano 5y ago|0 comments
161.
The Things They Won't Believe(moultano.wordpress.com)
1 points by moultano 5y ago|0 comments
162.
“Prompt tuning” closes the gap with fine tuning in large language models(arxiv.org)
1 points by moultano 5y ago|0 comments
163.
by moultano 6y ago
The fact that over parametrized models can generalize better than under parameterized models, and that both are better than models that can just barely interpolate is a genuinely new insight that was not predicted by any prior theory.
164.
by moultano 6y ago
If the price will bear it that means people want to live there more than elsewhere. Don't impose your idea of dystopia on other people.
165.
by moultano 6y ago
NYC has one of the lowest rates of housing starts per capita in the US.
166.
New MLPerf record for ResNet-50, 75.9% validation accuracy in 269 seconds(github.com)
2 points by moultano 6y ago|0 comments
167.
Distributed Shampoo Code Release: ResNet-50 in 1729 steps, 284 secs on TPUs(github.com)
1 points by moultano 6y ago|0 comments
168.
by moultano 6y ago
Tip the artists you love. If you want you can send them money by paying for prints, or buying originals, but ultimately just send them money. There's no need for it to be transactional. If there are artists in the world that make thing
169.
by moultano 6y ago
HeLa would be a better title. https://en.wikipedia.org/wiki/HeLa Copying the remains of a human around with ambiguous ethics, largely because they're "standard" and achieving a strange kind of immortalit
170.
by moultano 6y ago
Seems like if you're going to write about something a lot of people find a lot of value in, you at least owe it to your readership to explain why they find it valuable.
171.
by moultano 6y ago
> The unceremonious dumping of A Prairie Home Companion, which wasn't exactly my schtick but provided a little variety at least, made their coastal preference clear. WHaaa? Garrison Keillor hand picked Chris Thile to be his success
172.
Use local regression instead of a moving average so your graphs don't lag(moultano.wordpress.com)
2 points by moultano 6y ago|0 comments
173.
Use local regression instead of a moving average so your Covid graphs don't lag(moultano.wordpress.com)
4 points by moultano 6y ago|0 comments
174.
by moultano 6y ago
The guy is a troll. The last nuclear power plant built in the US took 40 years to come fully online and has never made money. Meanwhile solar is going through a moore's law like cost reduction and is cheaper than coal in most places. G
175.
by moultano 6y ago
A monorepo does not let you deduce the revenue consequences of experiments, but regardless, even if you could figure it out, it isn't part of the incentive structure.
176.
by moultano 6y ago
> If a search change demonstrably has a negative effect on company earnings (even if users like it) it will almost certainly be reversed. This is completely false. I've personally been involved in launching ranking changes that neg
177.
by moultano 6y ago
This is why there's an aggressive firewall between the ads and search organizations. The ranking team isn't even aware of revenue numbers by ACL.
178.
by moultano 6y ago
https://www.techdirt.com/articles/20200531/23325444617/hello...
179.
by moultano 6y ago
If I'm understanding correctly, it doesn't just set the kernel after seeing the data, but also after training the entire model, because the path kernel can't be defined without the optimization process to define the path. I c
180.
by moultano 6y ago
That the difficulty of finding the video in YouTube search was a deliberate change made at the request of the CIA.

More ›