chaxor

28 ms·

301.
by chaxor 3y ago
It's pretty good if it were an AI generated image fake. That was my immediate thought, and then changed when the only differences were things like 'it should read AVAF not AVAX'.
302.
by chaxor 3y ago
If they move it to optical compute then we won't have to keep talking so much about the CO2 and compute cycles. I imagine they're having conversations to do just that pretty often now, and hopefully with some action towards gettin
303.
by chaxor 3y ago
Awesome! Are there any simple downloads that have just annotations and connectivity? Like a sparse adjacency matrix with annotations of the node and edge features (neuron type, etc). I didn't see any immediately. A condensed representa
304.
by chaxor 3y ago
I'm not sure I agree with this view. If the NIH or DoD, etc serves data over FTP, they can certainly also serve it via IPFS. It would be much better for them to do so. Then, if anyone pulls that data (even if it's only a small
305.
by chaxor 3y ago
https://youtu.be/FeAMiBKi_EM The webdev or JavaScript mentality of if it breaks, don't worry, we can fix it live strikes again.
306.
by chaxor 3y ago
I effectively thought this was always the case, since I don't use Twitter, and disabling JavaScript made the webpage unusable anyway. You don't miss out on much though. Twitter just seems to be random hot takes and celebrity nons
307.
by chaxor 3y ago
As has been pointed out many times, this is similar to the steps that have led to interpreted languages like python, R, Julia, etc that make calls to C, or use JVM/LLVM, etc on up to assembly or machine code. The leap made is certainly
308.
by chaxor 3y ago
IPFS or torrent are the best options for distributing data
309.
by chaxor 3y ago
You don't have to, but you have to admit DuckDB is quite nice for handling csvs quite a bit more intuitively than awk
310.
by chaxor 3y ago
What network monitor for a phone is 1) lightweight and performant enough to not hog resources 2) extremely well-tested and simple (e.g. wireguard-level simple) 3) FOSS and trustworthy? I have not encountered anything even remotely close to
311.
by chaxor 3y ago
I would imagine simply creating a scoring function that down-weights posts that only one demographic agrees with, while up-weighting posts that all demographics agree with would be one way of achieving this.
312.
by chaxor 3y ago
Based on the description it seems more like qemu or docker? Perhaps I'm misunderstanding something
313.
by chaxor 3y ago
What scientist would ever even consider excel in the first place?
314.
by chaxor 3y ago
Paul was slain :( "SiP@pepsin.VII@Vmay8y63f-2aboutcolombiaKh6" Decent password 10/10
315.
by chaxor 3y ago
Debacle with wireguard? Opnsense has wireguard easily available. Also, it's just base wireguard, so you don't have to go through any extra steps of trying to understand / trust other additions on top of it, which is very ni
316.
by chaxor 3y ago
"Docker is a workaround for shitty software" This just needs to be repeated for the pythonistas sitting in the back.
317.
by chaxor 3y ago
Alpine or Debian. Those are the only base images you ever need.
318.
by chaxor 3y ago
Oof
319.
by chaxor 3y ago
I think some of the best games I've found on the phone have been termux and connectbot. It's really fun to try to chain together as many jump boxes as possible and minimize latency or example. My lowest score is 100ms latency for
320.
by chaxor 3y ago
I'm confused by some of this. Specifically, pandas columns are essentially just a thin wrapper around record arrays in numpy right? How could it be that much faster to use numpy over pandas if it's effectively the same thing?
321.
by chaxor 3y ago
The trick is, it's all the same muon! https://en.m.wikipedia.org/wiki/One-electron_universe
322.
by chaxor 3y ago
It's important to separate variables appropriately here, since religion and/or bigotry is not the only way to make it to less alcohol. To make a statement such as less alcohol leads to bigotry would be ridiculous. Although on a
323.
by chaxor 3y ago
The problem isn't one takeout overlapping (multiple zips from one date) it's many takeouts over the years (full history). So for example in 2001, you make a takeout with 30 zips, and then delete half of your photos off of Goog
324.
by chaxor 3y ago
Culture can be a powerful tool for things like this. For example, smoking has decreased immensely over the past several decades. Were seeing similar drops in pregnancy due to youth of today having an immense aversion towards these activiti
325.
by chaxor 3y ago
Perhaps Meta will release their dataset for Galactica sometime, and we will have incredibly good training data quality.
326.
by chaxor 3y ago
Perhaps someone will start listening to Chomsky and figure out better inductive biases for the models such that we get tiny local LLMs that are more based in universal grammar rather than initialized randomly or by Xavier.
327.
by chaxor 3y ago
GitHub has torrent magnet links to several good datasets of recipes that are scraped and processed to just contain recipes only in a simple SQLite format. The best recipes come from seeding those torrent / IPFS files.
328.
by chaxor 3y ago
I don't understand the reasoning that some company has better security than another. It's honestly trivial to set up good security. The main problem is incentive to go against/ break that security, which will likely grow wi
329.
by chaxor 3y ago
I thought this was going to be from lastmiles (a good low level dev on twitch with excellent content on C dev.) The I realized they would likely hate the idea of working with python, and do everything in C first. So likely not lastmiles, u
330.
by chaxor 3y ago
It's a good way of separating useless crap from good code. If the main code is in a notebook, ignore it completely. If the code is in python but no ipynb, it may be useful. If it's a well structured code base, and there are some

More ›