Audalin @ Audalin @lemmy.world

Posts

1

Comments

26

Joined

2 yr. ago

Thanks! I now see that Tai Chi is mentioned frequently online in context of the film unlike yoga so that should be right; it narrows things down.

That's the ones, the 0414 release.

QWQ-32B for most questions, llama-3.1-8B for agents. I'm looking for new models to replace them though, especially the agent one.

Want to test the new GLM models, but I'd rather wait for llama.cpp to definitely fix the bugs with them first.

What I've ultimately converged to without any rigorous testing is:

• using Q6 if it fits in VRAM+RAM (anything higher is a waste of memory and compute for barely any gain), otherwise either some small quant (rarely) or ignoring the model altogether;
• not really using IQ quants - afair they depend on a dataset and I don't want the model's behaviour to be affected by some additional dataset;
• other than the Q6 thing, in any trade-offs between speed and quality I choose quality - my usage volumes are low and I'd better wait for a good result;
• I load as much as I can into VRAM, leaving 1-3GB for the system and context.

Because we have tons of ground-level sensors, but not a lot in the upper layers of the atmosphere, I think?

Why is this important? Weather processes are usually modelled as a set of differential equations, and you want to know the border conditions in order to solve them and obtain the state of the entire atmosphere. The atmosphere has two boundaries: the lower, which is the planet's surface, and the upper, which is where the atmosphere ends. And since we don't seem to have a lot of data from the upper layers, it reduces the quality of all predictions.

If config prompt = system prompt, its hijacking works more often than not. The creators of a prompt injection game (https://tensortrust.ai/) have discovered that system/user roles don't matter too much in determining the final behaviour: see appendix H in https://arxiv.org/abs/2311.01011.

xkcd.com is best viewed with Netscape Navigator 4.0 or below on a Pentium 3±1 emulated in Javascript on an Apple IIGS at a screen resolution of 1024x1. Please enable your ad blockers, disable high-heat drying, and remove your device from Airplane Mode and set it to Boat Mode. For security reasons, please leave caps lock on while browsing.

• https://xkcd.com

CVEs are constantly found in complex software, that's why security updates are important. If not these, it'd have been other ones a couple of weeks or months later. And government users can't exactly opt out of security updates, even if they come with feature regressions.

You also shouldn't keep using software with known vulnerabilities. You can find a maintained fork of Chromium with continued Manifest V2 support or choose another browser like Firefox.

Very cool and impressive, but I'd rather be able to share arbitrary files.

And looks like you can only send images in DMs, but not in groups/forums.

If your CPU isn't ancient, it's mostly about memory speed. VRAM is very fast, DDR5 RAM is reasonably fast, swap is slow even on a modern SSD.

8x7B is mixtral, yeah.

Mostly via terminal, yeah. It's convenient when you're used to it - I am.

Let's see, my inference speed now is:

• ~60-65 tok/s for a 8B model in Q_5_K/Q6_K (entirely in VRAM);
• ~36 tok/s for a 14B model in Q6_K (entirely in VRAM);
• ~4.5 tok/s for a 35B model in Q5_K_M (16/41 layers in VRAM);
• ~12.5 tok/s for a 8x7B model in Q4_K_M (18/33 layers in VRAM);
• ~4.5 tok/s for a 70B model in Q2_K (44/81 layers in VRAM);
• ~2.5 tok/s for a 70B model in Q3_K_L (28/81 layers in VRAM).

As of quality, I try to avoid quantisation below Q5 or at least Q4. I also don't see any point in using Q8/f16/f32 - the difference with Q6 is minimal. Other than that, it really depends on the model - for instance, llama-3 8B is smarter than many older 30B+ models.

Have been using llama.cpp, whisper.cpp, Stable Diffusion for a long while (most often the first one). My "hub" is a collection of bash scripts and a ssh server running.

I typically use LLMs for translation, interactive technical troubleshooting, advice on obscure topics, sometimes coding, sometimes mathematics (though local models are mostly terrible for this), sometimes just talking. Also music generation with ChatMusician.

I use the hardware I already have - a 16GB AMD card (using ROCm) and some DDR5 RAM. ROCm might be tricky to set up for various libraries and inference engines, but then it just works. I don't rent hardware - don't want any data to leave my machine.

My use isn't intensive enough to warrant measuring energy costs.

I see!

And it was a stable OS version, not a beta or something? That's the worst kind of bugs. Hopefully manufacturers start formally verifying hardware and firmware as a standard practice in the future.

Other than what I said in the other reply:

I live in the USA so getting one would be problematic but I hear perhaps not entirely impossible for me.

Looks like it has a US release? If you're unsure or getting a European version, double-check it's compatible with American wireless network frequencies &c. Specific operators might also have their own shenanigans.

Do you know how it compares to e.g. Fairphone?

Nope, never tried Fairphone.

Very solid, I think (except water protection, but my previous OnePlus also didn't have good water protection anyway; and I'm careful enough).

I don't tend to use glyphs or the default launcher (and therefore its special widgets that only work there; but the ability to have apps in folders on my main screen while being hidden from the app menu is more important for me than a handful of widgets, so Neo Launcher it is).

A recent OS update added configurable swap (up to 8GB), calling it "RAM booster". I don't use it, but if you want to run a local LLM (or rather a SLM), you could try making use of it? As long as you figure out how to make the model use main RAM and not the swap.

I like the battery life (or maybe it's just because it's the first phone where I started charging at 20% and stopping at 80% semi-consistently).

Termux still works despite the new Android versions becoming more hostile to apps executing binaries they didn't have included already.

One thing I miss from OnePlus is the ability to deny some apps network access entirely. (I think it was removed in later versions of Oxygen OS?)

Also was a OnePlus user - now switched to Nothing Phone (2).

I don't focus on recommendations specifically. My typical process is:

• spend anywhere from a few days to a few weeks figuring out which technical characteristics are important for this kind of product, which aren't, why and when &c. This kind of information is usually available (and even obvious SEO garbage can give you new keywords to consider when searching);
• based on these alone, determine what's acceptable and what's desirable for you;
• if you haven't already, find some kind of community around the topic and see which brands/manufacturers people commonly complain about and why; also see if there're popular manufacturers only selling things via their own websites;
• open your preferred store (or several) and filter the entire category based on what you've learned. Pick a few candidates and examine them closely;
• go back to the community again and look up anything mentioning these candidates - including comparisons with other ones you haven't considered. Perhaps consider them;
• make the final choice.

Skip some of these if irrelevant or if you don't care enough. Spend extra time if you care a lot.

It works well enough for every new phone (the market there is changing fast, so you start anew every time), it worked for my first PC I've decided to assemble with 0 prior knowledge, the mechanical keyboard and the vertical mouse, and pretty much every piece of tech I'm buying.

And I'd say it's reasonable to use Reddit without an account even if you disagree with what the platform owners are doing. The data is still valuable for such use cases.

No idea whether this would work, but have you tried moving app data via adb?

I'm still waiting for the day when actual ads across the internet drown in AI-generated advertisements pointing to no real product or service. Perhaps that'll make attention industry collapse?

If you're looking for a side project idea, here's one.

The exact definition of sanity is a cultural choice.