jampekka

31 ms·

631.
by jampekka 2y ago
> And even research into ways we can look into a living brain is hindered by the fact we don't want to harm the person being observed. Animal brains are harmed for research all the time. Some species, like Zebrafish can be even imag
632.
by jampekka 2y ago
Sure there are some inconsistencies and legacy, but I wouldn't call that "horribly broken". You're probably referring to plots with subplots. Those indeed have issues, although mostly not because of the API. This has som
633.
by jampekka 2y ago
Performance for animation and (custom) interaction is a real problem. But as for performance being "really really bad", there are not many widely used plotting libraries faster than it, at least for static plotting and zoom/p
634.
by jampekka 2y ago
What are your issues with the matplotlib API more specifically?
635.
by jampekka 2y ago
Not sure what you mean by complicated API. The (pylab) API is a very straightforward (mostly) immediate rendering typeish interface, with a lot of convenient shortcuts for operations used a lot in data analysis. For architecture astronauts
636.
by jampekka 2y ago
> Nvidia is rich because they have insane margins on their GPUs. They cost a fraction of what they sell them for. That means that price will crash over time. So, I'm optimistic that a lot of this stuff will improve rapidly. OTOH cur
637.
by jampekka 2y ago
I'd prefer even more matplotlib-ish. Don't fix what's not broken. A major reason why other plotting libraries don't take of is use of complicated APIs. But data analysis doesn't need Application Programming Interfac
638.
by jampekka 2y ago
Per quick testing the 27b model seems very strong at least in natural language. It produces even good Finnish, in which smaller models tend to really struggle. Very promising. Edit: Per even quicker testing the Finnish language performance
639.
by jampekka 2y ago
Usually I write scripts that use function memoization cache (to disk) for expensive operations. Recently I've also used Marimo sometimes, which has great support for modules (no reloading hacks), can memoize to disk and has determinist
640.
by jampekka 2y ago
This is the correct take. There are almost always better ways to plot three dimensional data than trying to project 3D geometry to 2D.
641.
by jampekka 2y ago
At least I usually do prefer to do the EDA plotting by writing and editing code. This is a lot more flexible. It's relatively rare to need other interactivity than zooming and panning. The differing approaches probably can be seen in s
642.
by jampekka 2y ago
As a socialist principle "who does not work" referred primarily to the bourgeoisie, i.e. who get money from capital gains, rents or inheritance as opposed to labor, although it did also mean that in early stages of socialism the c
643.
by jampekka 2y ago
Doesn't seem even wrong.
644.
by jampekka 2y ago
https://archive.is/20250307221215/https://www.wired.com/stor...
645.
by jampekka 2y ago
If you can get docs from an LLM, why not just use LLM interactively? That's how I use them anyway, I just ask how to do X with library Y and then ask for further info if needed. If that fails, go for the source code.
646.
by jampekka 2y ago
In don't think they meant formatting details.
647.
by jampekka 2y ago
Stictly speaking (non-recurrent) LLMs are Markov chains. Per e.g. Wikipedia: "In probability theory and statistics, a Markov chain or Markov process is a stochastic process describing a sequence of possible events in which the probabil
648.
by jampekka 2y ago
Sorry, added the convolution theorem part in an edit after your comment.
649.
by jampekka 2y ago
modReLU seems to just increase the magnitude of the input value, and rotate it to the original polar angle. With clipping off negative magnitudes. Or equivalently rotates a (real) bias term with the input angle and adds that into the origin
650.
by jampekka 2y ago
It embeds them first into vectors. The input is a real matrix with (context length)x(embedding size) dimensions.
651.
by jampekka 2y ago
This is actually simpler than most self attention methods.
652.
by jampekka 2y ago
The actual mechanism at least is quite simple. Essentially it takes the FFT of the input embeddings, multiplies it elementwise with weights that are gotten from the input embeddings using an MLP (plus a constant (but learnable) bias) and th
653.
by jampekka 2y ago
Psychedelics aren't necessarily as anxiety/paranoia inducing as weed, and even the effects in general aren't necessarily as intense. Of course it depends on the dosage, the person and the setting, and paranoia from weed most
654.
by jampekka 2y ago
A sober, or an experienced trustworthy, tripsitter is almost a must. A small amount of MDMA (especially if you're not familiar with it) before taking a psychedelic can be helpful as mentioned. Having some sedatives just handy just in
655.
by jampekka 2y ago
DMT hallucinations are very different, and a lot more vivid, than those from LSD or mushrooms. With the latter, at least with eyes open and in reasonable doses, the effect is not so much as seeing "new things", but seeing the same
656.
by jampekka 2y ago
It's not necessarily the professor really saying that. Journalists (and university press offices) like to have such lines in pop science articles, and how it goes is that there's an interview from which the writer "interprets
657.
by jampekka 2y ago
There are a lot of variety in non-American ways of doing business. E.g. in Scandinavian cultures management can be very hands-off, with workers largely assumed to make a lot of decisions independenly or among themselves.
658.
by jampekka 2y ago
I'd bet around 100% of the population support trigger pullers with badges and automatic weapons to enforce the rule of law.
659.
by jampekka 2y ago
With that wording most would struggle to understand the question. I also would like to believe that quite a small minority see social spending solely as a personal investment.
660.
by jampekka 2y ago
Social spending has overwhelming public support, with majority calling for more spending, and under 20% call for less spending. https://apnews.com/article/ap-poll-government-spending-socia...

More ›