bmc7505

7 ms·

1.
by bmc7505 2mo ago
Cool to think this demo would have been possible over fifty years ago. I wonder what someone from 1975 would have said if you had shown this to them back then.
2.
Optimization Constants in Mathematics(teorth.github.io)
2 points by bmc7505 4mo ago|0 comments
3.
by bmc7505 4mo ago
This topic recently came up at the FLANN workshop [1], and seems to periodically be rediscovered [2,3,4] in different contexts. While some have speculated about the biological role it plays (e.g., Pearlmutter & Houghton [5]), we still l
4.
Towards a Linear-Algebraic Hypervisor(arxiv.org)
1 points by bmc7505 4mo ago|0 comments
5.
by bmc7505 5mo ago
I've recently been using this plugin [1], which is still under development but is an adequate stopgap until a better solution comes along. [1]: https://plugins.jetbrains.com/plugin/22150-noctule-the-swift...
6.
Tetris Is Hard with Just One Piece Type(arxiv.org)
3 points by bmc7505 6mo ago|0 comments
7.
by bmc7505 7mo ago
As foretold six years ago. [1] [1]: https://breandan.net/2020/06/30/graph-computation#roadmap
8.
Yet another catalogue of fast matrix multiplication algorithms(fmm.univ-lille.fr)
1 points by bmc7505 7mo ago|0 comments
9.
by bmc7505 7mo ago
Fast matrix multiplication would be a more useful benchmark: https://fmm.univ-lille.fr/
10.
by bmc7505 7mo ago
17k TPS is slow compared to other probabilistic models. It was possible to hit ~10-20 million TPS decades ago with n-gram and PDFA models, without custom silicon. A more informative KPI would be Pass@k on a downstream reasoning task - for m
11.
Show HN: TidyPython – Real-time syntax repair for Python(tidyparse.github.io)
1 points by bmc7505 8mo ago|0 comments
12.
by bmc7505 8mo ago
There are a few approaches if you want to write a new language. One, as the author argues, is to write a library in an existing language, which may require sacrificing ergonomics to fit inside the syntax of the host language, but is safe, m
13.
by bmc7505 9mo ago
Interesting. I wonder if mqjs would make it feasible to massively parallelize JavaScript on the GPU. I’m looking for a way to run thousands of simultaneous JS interpreters, each with an isolated heap and some shared memory. There are some r
14.
Introduction to Linear Types(austral-lang.org)
2 points by bmc7505 1y ago|0 comments
15.
by bmc7505 1y ago
You could argue that since automatic differentiation and symbolic differentiation are equivalent, [1] symbolic AI did succeed by becoming massively parallelizable, we just needed to scale up the data and hardware in kind. [1]: https:/
16.
by bmc7505 1y ago
> The solvers participating in this track will be executed with a wall-clock time limit of 1000 seconds. Each solver will be run an a single AWS machine of the type m6i.16xlarge, which has 64 virtual cores and 256GB of memory. For compar
17.
by bmc7505 1y ago
The difference is that SAT/SMT solvers have primarily relied on single-threaded algorithmic improvements [1] and unlike neural networks, we have not [yet] discovered a uniformly effective strategy for leveraging additional computation
18.
by bmc7505 1y ago
The correct way to do this is with finite model theory but we're not there yet.
19.
by bmc7505 1y ago
Gingsberg stole it from Yeats — “the best lack all conviction…” / “the best minds of my generation…” — many similar verses, e.g., “what rough beast…” / “what sphinx of cement…” https://www.poetryfoundation.org/poem
20.
by bmc7505 2y ago
https://cstheory.stackexchange.com/questions/632/what-is-the...
21.
by bmc7505 2y ago
Called it three years ago: https://news.ycombinator.com/item?id=30142353
22.
Show HN: Tidyparse – Real-time context-free syntax repair(tidyparse.github.io)
2 points by bmc7505 2y ago|0 comments
23.
RE#: High Performance Derivative-Based Regex Matching(ieviev.github.io)
2 points by bmc7505 2y ago|1 comments
24.
by bmc7505 2y ago
https://dl.acm.org/doi/10.1145/3704837
25.
by bmc7505 2y ago
Depending on how comfortable you are with model theory you might also enjoy Dzhafarov and Mummert’s textbook, which first brought the subject to my attention.
26.
by bmc7505 2y ago
Although Wolfram doesn't mention it by name, this is closely related to what he is trying to do: https://en.wikipedia.org/wiki/Reverse_mathematics
27.
Amplifying human performance in combinatorial competitive programming(arxiv.org)
2 points by bmc7505 2y ago|0 comments
28.
by bmc7505 2y ago
This is roughly the intuition I have developed -- any computational function requires time and space to evaluate. Most computations carry with them some epistemic or aleatoric modeling uncertainty, but sometimes even a perfectly determinist
29.
by bmc7505 2y ago
FWIW, I’ve had a very similar encounter with another famous AI influencer who started lecturing me on fake automata theory that any CS undergrad would have picked up on. 140k+ followers, featured on the all the big podcasts (Lex, MLST). I n
30.
by bmc7505 2y ago
This is a handy tool, but I wish it supported edge snapping. If you inspect the generated LaTeX it doesn't actually link up the FSM states, it just anchors them to raw TikZ coordinates.

More ›