@jestdotty It's totally like grooming. I seem to vaguely remember that there was some actual impact to peoples lives because of it, but I can't quote any sources so not sure.

I hate people and trends that pray on the psyche of the weaker minded people. Like gambling, NFTs, drugs, this sort of sexual manipulation, including AI girlfriends n stuff. :( Makes me sad, angry, frustrated and annoyed all at the same time

@Demolishun They are not the best coders, but that's to be expected imo, their specialty is numbers, not hardware. Most basic threading will work auto-magically in python, but when dealing with huge datasets with tens of millions of rows you have to use better tools like memory sharing and mmaping to actually get any speed benefits from threads and processes. And I honestly don't expect them to pull those off, and that's fine. That's usually my job then to optimize their code and architecture, it's good teamwork, though some teams are harder to work with than others for sure :D

@AlgoRythm Fair point, however, not all ML stuff is pandas, numpy and pytorch. Datasciene still includes stuff like custom python filters that are not vectorized, in which case maybe there's performance to be seen if done via threading while avoiding the overhead of manually managing shared memory and multiprocessing.

Though I didn't play with GIL off yet, so that's just speculation on my part, but I've seen plenty of code written by data science teams and sometimes they try to use threading so there's that :D

@Lensflare yeah, subtle but important hint!

Where is this? LA?

@AlgoRythm the whole of machine learning community probably

@PaperTrail that's not what programmer socks is. The "trend" refers to some effort to feminize or degrade programmers as weebs and neets. It refers to long knee high womens socks, usually with the traditional anime patterns like stripped socks and so on. [The joke being that you should be wearing them while coding, no matter what sex you are]

I don't know where it started exactly, but from what I hear it was either 4chan or reddit (duh) where some group tried to force this fetish on new people by trolling and manipulation.

No idea what their conversion rate was but as far as I can tell the op is over and dead at this point. I refer to it as an op because some people, me including, think it was an organized effort by someone.

For those reason, I really wouldnt call it a programming trend. Whoever fell for it is more of a victim of some fucked up psyops than anything else.

using an LLM and burning through kilowatts of power just to generate a stupid sob short story as a spam... There's a special place in hell for these people

@Lensflare The funny thing about "good prompts" is that it's mostly about writing out specs that are a good balance of specific and generic to not limit the scope of the llm too much, but also to have it do all the important stuff and then to also provide examples. Then it's mostly just about iterating that prompt and refining it until you get rid of unwanted outputs and formats.

A child could do it... in fact I think children would have much easier time doing it, because they have the time and motivation to fiddle with it until it's perfect... While little ol' me always feels like it's a chore to do it. But when I do force myself to mess with it I can usually get the right formats and responses.

Nothing really to do with engineering (except maybe for the spec part), everything to do with patience. At least that's my experience with this.

Whoever thought of the term "prompt engineering" fucked up the entire industry.

Yeah, I don't know why they don't consult you on this first.

Really, in general, if politicians and scientists came to devrant and talked to us first the world would be a much better place.

Fluffernox is the only passable one to be fair. And even then it looks like one of those throwaway pokemons that are not really specialist at anything

@TrayKnots althought. you could make the argument that since you can install "snap" and "flatpak" via apt/rpm you can still get the source security, because the official distro repo is a trusted source, and then snap and flatpak become trusted sources via being installed by the official repo in the first place.

I can't talk to the reliability of the content on snap and flatpak though, but at least for stuff like VLC or similar, you're unlikely to get anything dangerous, as they are officially maintained

@retoor As far as I know yes, linux had package managers first and windows bought into it later.

Though NuGet was available really early with Visual Studio, I remember installing additional dependencies with NuGet back in 2015, but apparently NuGet was released as far as 2010. To my knowledge that was the first official package manager for windows.

I didn't mean to imply that Windows Store apps are out-of-date. I mean that linux APT or RPM package managers often have out of date packages, which is why something like Snap or Flatpack comes in handy, as they usually carry official stable releases of apps and libs as well as bleeding edge stuff :D

@retoor Nah, there's also NuGet and Chotolatey on windows which are kinda the same idea to Snap or Flatpack

there's also the normal Windows Store (I think it uses NuGet in the backend but not sure) Which makes it pretty similar to any GUI app manager on Linux, like Discover

The only analogy that I think windows is missing is AppImage. Those are similar to Apples DMG packages. I don't know of a windows pre-packaged app format. Maybe there is something? not sure

It's not something super related to Linux only, though it's most obvious on Linux because you'll end up using Snap or Flatpack way more for anything that's outdated on Apt/Rpm

@Demolishun I love Maiq!

Maybe that's a good character to be played by AI. Just one off character that makes pretty much no sense xD

Depends on the prompter too. If the dev keep asking the right questions he gets the right answers. They might even optimize the prompt enough to refuse saying non-game stuff for the obviously bad questions... But if you gonna have hundreds or thousands of players, someone is gonna Break it sooner or later. That's the best you can hope for with this technology with RAGs and to a lesser extent fine-tunes and LoRAs.

I personally think LLMs are simple not a good match for interactive game dev yet (possibly ever). They can be useful to generate a lot of secondary dialogue that you then clean up and integrate with the game. But having it be live & self hosted, distributed with the game binary? Just No... You'll get a pretty mediocre game that weights at 5GB+ and the characters talk nonsense, reveal secrets and break immersion. I keep experimenting with LLMs for game dev, but so far nothing "sparks joy" as they say. Never good enough for my standards

@Lensflare what @jestdotty said :)

@TrayKnots Such a good show tbh

Dude, the immigrants are all lazy and they just want to steal our jobs >:c

I make my own mayo at home, it's delicious, so much better than the store bought stuff!

Maybe instead of starving to death you should try a "make everything from scratch" diet! Would be more fun than just not eating anything. Could start your own garden and see if you can survive before the first crops xD...

You can't train an LLM without giving it knowledge. If you wanted to train an LLM for a game NPC you would likely need to first simulate thousands of years of written History for that game world. That's the only way to avoid any spill-over from our real world.

It's much more practical to fine tune or to use RAG though RAG is one of the worst ways for a game. It will likely keep breaking out of character constantly because It's still an internet trained llm, just with added references. As if you handed an actor a script and told them to check the script for relevant info whenever they respond. So if you ask anything Off Script they have no choice to break character.

Fins tuning on the other hand is more complex, but it means you will essentially train the llm to be more like the training data. It's as if you sent your actor to the place/time they are supposed to act out and have them stay there for a while to learn the culture and language more.

String isnt the issue. But the column is likely not using a unique key constraint.

I bet they handle it in code with something dumb like a query to check of the number exists before generation. So they could still be de-duplicating it in code, just not using a proper schema. Ideally it should be both, but the lower the level where the security is enforced the better. I doubt Elon went over all the code just to check

That did tell me everything you need to know: the other devs are incompetent :D

@retoor if you deliver a codebase of prompts instead of code I would fire you on the spot

Sanity Check:
Do you watch TikTok, Instagram or other short video content?

Regarding the encryption thing:
1. Do you need encryption on a laptop -> ofc, that's exactly where you need it. You're not goong to lose a desktop PC but a laptop could be stolen right from your lap if you have it out in café, train, library, outside or wherver...

2. Disabled TPM -> TPM is a pretty cool tech, it uses a physical chip on your board to store the keys, which means a) the disk is not gonna decrypt if just removed and plugged into a other PC and b) the keys are un-dump-able, so cracking it brute force is also pretty much impossible. But I suppose you don't need it.

3. You definitely don't need a MS account, but if you decide to enable TPM I suggest you get one, cause that's where you can find your encryption keys of you get yourself with a broken OS. There might be a way to show them and write them down without that, but I don't know, but I was fixing a broken OS setup recently and using that key I was able to mount the drive in live Ubuntu without issues

@TrayKnots why? There would still be places where politics are discussed on purpose and people would go in ready for a political discussion.

Why is it better to force people to talk about politics when and where they don't want to? You are just going to get a bunch of people that are going to close off, dismiss you or you gonna catch them off guard leading to nothing but heated and emotional reactions.

How is this any better for democracy compared to just talking about politics at the right time and place with people that want to do it and are opened for a discussion?

The world is too heated up by politics and right now we're too divided and too upset to have a calm discourse. I don't think It's bad to have politics free zones where people can again find Something that connects them rather than divide them, at least in the gaming space where It's mostly about escapism. And to be honest, if a group of like-minded people want to keep the chat politics free, who are you to judge? It's part of a free and democratic community to set their own internal rules

The blocking though, I completely agree. The shit where you are shadow banned, or blocked, or canceled for saying the smallest thing, even calmly and well researched... That's a problem. That's why we call people snowflakes now

Actually you just need some vodka and an old rag. Doesn't help the macbook much but I sure care less about how it looks after that fun

I read a good comment recently: "DeepSeek is more Open AI than OpenAI" and that is I suppose the core of the problem.

Deep seek is just some side project of one of their engineers and It's cheaper and more available than OpenAI where they pretend It's so expensive to run these models, when in reality it was long obvious they have overblown models.

Regarding safety, R1 is a safe model but since It's open source uncensored versions already exist.

Also Im not sure about tooling. I remeber reading that it supports tools, but the community has issues running them right now? Something like that. But then again, It's an open source model that's relatively cheap to train. Adding tool_call examples to the training data is trivial at this point