macleginn

8 ms·

91.
by macleginn 9mo ago
Well, there is this stoic British way of looking at the world and preserving the sense of self worth and then there is the ending of The Hitchhiker's Guide to the Galaxy, which is just pure cosmic despair in the face of the bureaucrati
92.
Howard Lutnick: Why the Trump administration is going to Davos(ft.com)
2 points by macleginn 9mo ago|1 comments
93.
by macleginn 9mo ago
I think part of the negative attitude towards the effects of AI stems from the fact that it demolishes a lot of structure. The traditional institutes maintain well-structured, low-entropy societies in terms of knowledge: one goes to a lawye
94.
by macleginn 9mo ago
The authors are quick to assume "the inevitable atrophy of human skills and knowledge". But skills and knowledge have not been propped up by some low-fi institutional magic: they stem from the necessity of achieving practical goal
95.
by macleginn 9mo ago
Marx is fair play, but one of the most prominent cases of understanding everything in advance is undoubtedly Chomsky's theory of innate/universal grammar, which became completely dominant on guess which side of the pond.
96.
by macleginn 9mo ago
Speaking of UK, they also run the NHS data warehousing.
97.
by macleginn 9mo ago
Russia has been systematically shutting down the Internet for a long time now, to disrupt Ukrainian drones. The effects were very painful intially, especially re payment systems; people were encouraged to withdraw and carry cash. Now they a
98.
Louis Mosley of Palantir at the Science, Innovation and Technology Committee(committees.parliament.uk)
3 points by macleginn 9mo ago|0 comments
99.
by macleginn 9mo ago
More precisely, Lyon Text Web.
100.
Construction of ubiquitous surveillance has proceeded unhindered for many years(computer.rip)
3 points by macleginn 10mo ago|0 comments
101.
by macleginn 10mo ago
Makes one wonder if/how quickly they will come for closed WhatsApp groups and Telegram channels next.
102.
by macleginn 10mo ago
Neveress all Austen's happy endings are due to the magical alignment of respect and love with security and social alliance. Jane's heroines are playing a (relatively, see below) high risk/high reward game of not wanting to sa
103.
by macleginn 10mo ago
Blood cells are huge!
104.
by macleginn 10mo ago
As always, this depends on the amount of training data available. Japanese is another success story: https://digitalorientalist.com/2020/02/18/cursive-japanese-a...
105.
by macleginn 10mo ago
I became convinced of this after the release of KuroNet: https://arxiv.org/pdf/1910.09433 (High-quality OCR of Japanese manuscripts, which look almost impossible to read.)
106.
by macleginn 10mo ago
"The roads can be lethal and many drivers have a great deal of animosity towards cyclists" --- which is why bike lanes don't get much use: sooner or later you will have to share the road with cars for a while, and I personall
107.
by macleginn 10mo ago
As a foreign national living in the UK on a long-term visa I can only say that the decision to discontinue physical BRP (residence permit) cards in favour of eVisas is singularly idiotic and harmful. One piece of evidence being that there a
108.
by macleginn 10mo ago
Interestingly, the public discourse in the UK (at least what I have observed, and it was hard not to observe a lot in the last several days) does not focus much (if at all) on the insider trading angle. It's mostly that the chancellor
109.
by macleginn 10mo ago
Previously posted as https://news.ycombinator.com/item?id=46104139
110.
by macleginn 10mo ago
It is the same type of learning, fundamentally: increasing/decreasing token probabilities based on the left context. RL simply provides more training data from online sampling.
111.
by macleginn 10mo ago
In the limit, the "happy" case (positive reward), policy gradients boil down to performing more or less the same update as the usual supervised strategy for each generated token (or some subset of those if we use sampling). In the
112.
by macleginn 10mo ago
This is also the conference where everybody was briefly deanonymized due to an OpenReview bug: https://eu.36kr.com/en/p/3572028126116993 Now all the review scores have been reset, and new area chairs will make all
113.
by macleginn 11mo ago
I presume that this policy is not about building data-centres but about the use of AI by CERN employees, so essentially about marginal cost of generating an additional Python script, or something. Don't know if this calculation ever ma
114.
by macleginn 11mo ago
‘Sustainability: The use of AI must be assessed with the goal of mitigating environmental and social risks and enhancing CERN's positive impact in relation to society and the environment.’ [1] ‘CERN uses 1.3 terawatt hours of electrici
115.
by macleginn 11mo ago
It was not the Manchus who reconquered Tibet in the 1950s, after it had been an independent country for several decades. And the general argument is not about whether there is something inherently imperialistc in the Han -- it is about whet
116.
by macleginn 11mo ago
Qing Dynasty annexed Xinjiang, Taiwan, Tibet, Mongolia, as well as large chunks of Central Asia, and fought with Sikhs over Kashmir. Looks like a good case of imperial expansion to me.
117.
by macleginn 11mo ago
Japan has a big army/"self-defence force", impenetrable terrain over most of its territory, and 45 tonnes of plutonium. Even if the defence treaty with the US vanishes, the probability of a foreign invasion is rather low.
118.
by macleginn 11mo ago
> But historically, China has never been good at ruling non-Han peoples. "Good" is not a very objective term, but China does have 55 official minorities, coming from a long period of imperial expansion, so arguably it can be do
119.
by macleginn 11mo ago
Any idea where the name came from?
120.
How Quake.exe got its TCP/IP stack(fabiensanglard.net)
2 points by macleginn 11mo ago|1 comments

More ›