macrolocal

8 ms·

91.
Mathematics of Renaissance Canons(arxiv.org)
2 points by macrolocal 4y ago|0 comments
92.
by macrolocal 4y ago
That sounds like Bergson more than Buddhism.
93.
by macrolocal 4y ago
Low NFC people get a bad rap because they're quick to dress up intellectual laziness with moral posturing. But cognition is as precious a resource as attention or compute these days, so maybe there's something to learn from them.
94.
by macrolocal 4y ago
Counter-surveillance techniques for gait analysis are already well understood. https://www.youtube.com/watch?v=TNeeovY4qNU
95.
by macrolocal 4y ago
Photonics can be great for I/O, nb.
96.
by macrolocal 4y ago
Yep, I was just asking Adam* to justify his loss function. *pun intended :)
97.
by macrolocal 4y ago
Well, the paper uses gradient descent to minimize that difference , like auto-encoders do.
98.
by macrolocal 4y ago
Wait-- the residual stream makes the attention mechanism learn the difference from the identity! Are you sure you're not thinking about auto-encoders? Edit: ok, Discord it is.
99.
by macrolocal 4y ago
First question: why should the attention mechanism output and residual stream match?
100.
by macrolocal 4y ago
But consider that they include Hardy's "Mendelian Proportions in a Mixed Population" but not Riemann's "On the Hypotheses Which Lie at the Foundations of Geometry."
101.
by macrolocal 4y ago
Gadflies target Bellerophons obviously. It your steed can't handle a few gadflies, riding off to Mount Olympus is irresponsible.
102.
by macrolocal 4y ago
Gadflies play a critical role in the ecosystem! :)
103.
by macrolocal 4y ago
Even so, the larpers and bots seem more authoritarian now.
104.
by macrolocal 4y ago
Is anyone else astonished by how dramatically Hacker News has shifted its tone on this issue over the last five to ten years?
105.
by macrolocal 4y ago
Ditto for Euclid. Doing this early in life pays huge dividends.
106.
by macrolocal 4y ago
A Mathematical Mosaic is a little-known gem here.
107.
by macrolocal 4y ago
See ~ page 16 here for the (historic) definition: https://personal.math.ubc.ca/~cass/courses/m446-05b/dedekind... It’s possible to work with both {x : x < r} and {x : x <= r} and maintain the distinctio
108.
by macrolocal 4y ago
Have you seen Neptune Frost yet? I want that keyboard jacket.
109.
by macrolocal 4y ago
Points taken, but LLMs are still outpacing expert predictions, so empirically they’re under-hyped.
110.
by macrolocal 4y ago
Case in point: Russell’s views on Nietzsche.
111.
by macrolocal 4y ago
Caveat: there can be deep ecological insight in how locals conceptualize their environment, eg.
112.
Workshop on Machine Assisted Proofs(ipam.ucla.edu)
3 points by macrolocal 4y ago|0 comments
113.
Has the Light from Behind a Black Hole Been Seen?(profmattstrassler.com)
2 points by macrolocal 4y ago|0 comments
114.
by macrolocal 4y ago
Also, knot complements are an important class of 3-manifolds.
115.
by macrolocal 4y ago
I'd guess people are rejecting neoliberalism more than Locke. But if so, it's curious that the vocal left had shifted their attention from the disparity between people living in huts and palaces, to making sure that the people liv
116.
by macrolocal 4y ago
Not at all, eg. Knuth's TAOCP, Vol. 2, Section 4.2.2 A.
117.
by macrolocal 4y ago
I worry about illiberalism becoming more prominent on the left and right.
118.
by macrolocal 4y ago
Eh, these are models, not reality. The Many-Worlds interpretation makes the fewest assumptions, so it's probably the least prone to over-fitting. Until we find philosophically or experimentally distinguished ways to pare down its ontol
119.
by macrolocal 4y ago
Maybe, but it would be a highly non-classical computer. Probably it could cheaply render the most physically plausible paths to specified outcomes, and quickly factor large primes. Nb. quantum also highlights assumptions made by our brain&#
120.
by macrolocal 4y ago
This feels like saying that (generalized) linear regressions without interactions are inadequate. And criticisms of intersectionality seem to be that (a) high order interactions are statistically volatile and (b) multivariate averages miss

More ›