smokel

6 ms·

1.
by smokel 7d ago
By that logic a supermarket would sell only 33 items.
2.
by smokel 8d ago
Those are self-reported numbers from 2023. Besides most likely being highly inflated, I think Facebook is on the decline.
3.
by smokel 8d ago
That would make sense, but according to the stats posted in a sibling comment they crossed the million weekly downloads in week 31 already.
4.
by smokel 8d ago
The number of weekly downloads has simply been increasing over time, due to growing popularity. Calling each new high "record-breaking" is a bit meaningless. Unix timestamp counter breaks record after writing this comment.
5.
by smokel 9d ago
This seems to be a wrapper around llama.cpp with several tuned parameter settings. Most of the speedup is gained by enabling speculative decoding [1]. No amazing breakthroughs, just good tuning! [1] https://en.wikipedia.org&#x2
6.
by smokel 9d ago
Why link to archive.org here? The blog post is available at: https://robert-taylor.me/blog/compressed-sensing-python/ Also, (2016).
7.
by smokel 10d ago
I was attacking the arguments used to substantiate the idea that literacy is a good thing. I personally think literacy is great.
8.
by smokel 10d ago
For all we know social media will emerge across another 6,000 year timeframe. I think we're going in circles :)
9.
by smokel 10d ago
My point is exactly that comparing some variation within 20 years with to a 1000 year period makes little sense. Sure, our collective literacy might be decreasing now, but perhaps it was a bit too high to begin with? Leaders of a country
10.
by smokel 10d ago
While I agree with your sentiment, I doubt that we can really judge that properly without some objective distance in time. The invention of the printing press, canals, gunpowder, sailing, may have changed lives quite a bit. Consider also t
11.
by smokel 10d ago
There are more people now, and all else being equal, there would likely be more stupid people as well, in absolute numbers. That might give someone the impression that things are going downhill. Also, the article and book claim that people
12.
by smokel 10d ago
Hm, comparing developments spanning from the 12th century up to now with stuff that is happening in the last decade. Just give social media about 5 or 6 centuries to develop, and then check back to see if the arguments of this article are
13.
by smokel 11d ago
Time seems to be one of the most valuable luxuries. AFAIK it is not entirely clear why time seems to pass faster when one grows older. One argument is that the older brain does not form memories as intricate as in the younger brain. Has a
14.
by smokel 11d ago
> It had never “seen” a go board, or had one explained I assume that you are referring to AlphaGo or AlphaZero. In either case, this statement is not correct. Both algorithms most certainly know exactly what a go board looks like, an
15.
by smokel 13d ago
"Somebody is wrong on the internet" will get you more, and more reliable, results on what people actually want from AI. You are now participating in a very high-value survey.
16.
by smokel 13d ago
There's a branch of mathematics called "pointless topology" [1]. [1] https://en.wikipedia.org/wiki/Pointless_topology
17.
by smokel 16d ago
In this case, the argument is valid though. Only recently have some of the models become powerful enough (for some) to actually be useful in day to day programming, without too much hand-holding. For most, this change occurred with the in
18.
by smokel 20d ago
Would it make sense to guess that humans have something analogous to the latent representations or embeddings found in LLMs or VLMS, with different neural circuitry to experience them? Some of us may lack the physical systems to experience
19.
by smokel 21d ago
An estimated 2,000 articles per month at 2,000 words per article would result in 4,000,000 words, which costs either $200 or $2000 to process at Pangram [1]. It might make life a lot more enjoyable for many users here, though. [1] https:&
20.
by smokel 22d ago
Somewhat more meaningful was "INT 19h" (CD 19), which also took 2 bytes and started the boot loader, bypassing the slow boot sequence of most systems. And wouldn't a single instruction such as NOP or HLT be simply 1 byte?
21.
by smokel 23d ago
Interesting, tell me more!
22.
by smokel 1mo ago
A ⊆ X ∧ p ∈ X ⇏ p ∈ A
23.
by smokel 1mo ago
Mathematics is typically concerned with "proofs" [1], which similarly to code, often allow for strict validation. Thanks to reinforcement learning techniques, it is now possible to train LLMs to perform very well on code generat
24.
by smokel 1mo ago
https://github.com/karpathy/llm.c However, I wonder about the use of this. Typically, you'd want to use a GPU for inference, and programming those in C is quite a challenge. The educational value of learning LLM
25.
by smokel 1mo ago
I typically interpret it as an excuse, not a selling point.
26.
by smokel 1mo ago
> Its new plans called for traveling through the galaxy looking for entities like itself. Why not simply set up a network with other planets and communicate over FaceTime? I don't see a reason why Earth should actually hang out w
27.
by smokel 1mo ago
The technical aspects of SOTA models are not publicly documented. How do you know if something is converging?
28.
by smokel 2mo ago
This is a pretty confusing article. It seems to string together some loosely related facts to make the current AI boom look bad.
29.
by smokel 2mo ago
Yes there is, at least to me. The reason being total freedom to tune the operating system, and not being limited to what someone imagined I would want to do in the future.
30.
by smokel 2mo ago
Perhaps add "A systematic approach to technical documentation authoring" to the title?

More ›