NBJack

27 ms·

421.
by NBJack 2y ago
The latest Nvidia drivers offer an option to start using system memory when VRAM is insufficient. It certainly slows things down, but it does work. It's not perfect in my experience, but it is an alternative for large models.
422.
by NBJack 2y ago
And use much of the touted LLMs to figure out if a review should be about the shipping company. But that wouldn't be cost effective I suppose.
423.
by NBJack 2y ago
I want my deterministic UI back, period. Everyone got on the bandwagon for dynamic web pages, parallel rendering, and such, which is fine. What isn't fine is when, due to said parallelism, the interface hops, skips, and jumps all over
424.
by NBJack 2y ago
That's good to know, but the burning question is: why? Is this a critical but otherwise background part of the update? A clever time saving choice that didn't pan out well?
425.
by NBJack 2y ago
At a glance, the semantics seem kind of weak. It's hard to tell in many cases why it picked certain scenes in lieu of the concept searched.
426.
by NBJack 2y ago
Perhaps, but Adobe and Spotify aren't offering a operating system where this crap is bundled. Neither do those examples have a 72% market share of PCs. Microsoft basically gave Teams away to corporate customers. Now, couple this fact w
427.
by NBJack 2y ago
I've tried to use generative text LLMs myself for story ideas, filling out backstory in characters, etc. No matter how good the model is, no matter how much context you supply or repeat, it will inevitably spiral down into some degree
428.
by NBJack 2y ago
Single parents holding down a brutal job(s), areas that are majorly under served by meager retail offerings, folks with mobility impairments, the need for specific items that are hard to get in a timely fashion from other online solutions,
429.
by NBJack 2y ago
Preach it. I remember trying to a avoid LaTeX to submit for a journal. Microsoft Word was powerful enough, right? And how hard could it be? It was agonizing, painful, and downright schooled me on just how imprecise the popular editor could
430.
by NBJack 2y ago
LoRAs are fine tuned changes to an existing model, often one that is already mature and well-trained. The process is understood and documented. OP is referring to early phases of model development using new techniques, where outputs can var
431.
by NBJack 2y ago
Chat is perhaps the cheapest implementation you could ever build. It's a linear interaction, easy to test, and arguably the easiest to encode/decode (with a fixed set of inputs too). As an added bonus, it has a familiar, well-unde
432.
by NBJack 2y ago
This is true, but your examples aren't directly trying to pretend they are the better alternatives for that. Apple is doing its best to paint itself as some golden company when reality dictates they are no better (if honestly worse in
433.
by NBJack 2y ago
> it's more that they're the one FAANG company where I am the actual customer and their incentives align/depend on keeping me happier Do they though? Battery performance that 'lies' to you intentionally, planned
434.
by NBJack 2y ago
It cheapens the incentive greatly. And you probably aren't selling your photos to make a living.
435.
by NBJack 2y ago
Be careful with the wordplay here. Apple isn't. OpenAi is not Apple.
436.
by NBJack 2y ago
I think it's already dead; the vestiges of life it exhibits are basically the electrical tethers of AI editing, cloud 'collaboration', and various other "tweaks" trying to justify a yearly release of a product that
437.
by NBJack 2y ago
It's being sold, presumably for a profit, on a platform with a ~14% market share. Less if we count compatible devices based on OS version. That is entirely the author's choice, but it isn't without consequence. For example, I
438.
by NBJack 2y ago
Modern GPUs aren't optimized for it. You can indeed do this in software, but this is the path hardware acceleration took.
439.
by NBJack 2y ago
Super critical for those that travel a lot, whether for work or personal needs. I've lugged some boat anchors around; having extra space for the necessities, or just being able to work in economy class on a long flight, are premiums I&
440.
by NBJack 2y ago
If go didn't naturally eat error context, I'd like it more. But in the time I've used it, it makes errors much, much more painful to root cause without a debugger.
441.
by NBJack 2y ago
Or just that errant refactoring automation that moved a proverbial (or literal) bit out of place. It's good to learn from our mistakes. But if there's one thing I have learned from working across teams and orgs, it is that basical
442.
by NBJack 2y ago
Miller v. Ford Motor Co. would disagree. There is a viable legal precedent, and it is very arguably illegal. > The appellate court ruled that the voice of someone famous as a singer is distinctive to their person and image and therefore,
443.
by NBJack 2y ago
From the article: > What's especially troubling is that many human programmers seem to prefer the ChatGPT answers. The Purdue researchers polled 12 programmers — admittedly a small sample size — and found they preferred ChatGPT at a
444.
by NBJack 2y ago
Oof. I hope the tablet is better than the website. I think I need to switch to a device with a dedicated GPU to see it right. My laptop is struggling, and it's running a 12th gen i5. They seem to be using a video provider that can&#x27
445.
by NBJack 2y ago
To be fair, that is the penalty of being in a shared cloud. They are incetivized to keep their customers from using everything, everywhere, all at once. Jump into a 'bare metal' datacenter and things can get much different.
446.
by NBJack 2y ago
Think about what you actually need to start a video. Maybe a dozen MB? After that, you can plunge into colder storages and warm things up as you stream. Additionally, if you need longer to 'defrost' things, just cache a few more M
447.
by NBJack 2y ago
Not as far as you know. Doctors: https://www.news-medical.net/news/20240424/Opportunities-and... . Lawyers: https://www.ft.com/content/2365b275-8b0b-4ae9-bc4a-15c6e0776... Or my fav: https:&#
448.
by NBJack 2y ago
Everything you've described all happens in many other languages today. Python is personally one of my more prominent nemeses, as the ML bubble marches forward; I can't even tell you how many venvs I've got with various bina
449.
by NBJack 2y ago
As much as I hate to spoil the illusion, the kludging you mentioned is basically the sum of the natural drawbacks faced in virtually all game design. Whether it's a skybox, an NPC, or a virtual bullet, it's all basically cheating
450.
by NBJack 2y ago
I may be missing something, but that seems to be an apples to oranges comparison. How is your turn based app example analogous to a complex physics simulation? And what exactly do you consider about the app's multiplayer to be the bar?

More ›