j4k3

j4k3, 8 months ago

Gentoo for the documentation, but for a modern comp with bad bootloader implementation, Fedora’s anaconda system for the secure boot shim is irreplaceable and my daily. I won’t consider any distro without a shim and clear guide for UEFI secure boot keys. In that vain, Gentoo is the only doc source I know of that walks the user through booting into UEFI directly with Keytool.

j4k3, 8 months ago (edited 2 months ago)

Back nearly 20 years ago Disney California Adventure (Anaheim, Cali) had this interactive exhibit where you did various things in line with the Disney animated kids movies. One was playing the Seven Dwarfs “Heigh ho, heigh ho, it’s off to work we go…”

In a little alcove in front of a group of friends was another group of cute mid-teen little Latino girls. They were dressed somewhat provocatively for the average Disney guest and all were wearing bold makeup, (not judging, but they stood out). The typical Spanish accent was prevalent with the silent “h”. Their innocent rendition as they sang along was hilarious. Happy as they could be they were singing, “I hoe, I hoe, it’s off to work we go…”

j4k3, 9 months ago

I treat it like any other day and ignore all events focused on relationships, but I’m partially disabled and unable to do anything social. Just do whatever you find interesting in life and ignore the “celebration” the memory will fade into the background like all the other days and you won’t have depressive repercussions due to self reflection.

j4k3, 8 months ago

AGI is orders of magnitude more advanced than what we currently have available. It is a self aware system. Most of the issues must be addressed internally, but ultimately it is self regulating in every respect. There would be redundancy, and an element of design trust built in. It would not be corruptible like humans where we must be skeptical of our governments. In some respects, it is the hacker, it is the internet, and it is Orwellian in scope, but it is not authoritarian, or ideological. It would be direct and openly available for everyone to consult at any time. It would be capable of explaining anything in easily understood language according to the capabilities of the end user. The primary way it shapes policy and changes for the betterment of the majority is through rewards and amenable compromise. Ultimately, I think this is the only way to manage a real post scarcity society.

j4k3, 8 months ago

AGI lead government that is written like a constitution and bill of rights. The infinite persistence factor without human needs or motivations is a major improvement over anything that has ever existed.

j4k3, 8 months ago

I’d argue to legalize everything including the extremes and price the extremes to barely undercut and drive out any illicit market. It is always better to have control over a legitimate market than it is to have a black market. There is no way to regulate demand and creating market choke points is totally ineffective. So use state run capitalism to make the market uncompetitive and drive out any competition to gain full control. The State as the dealer makes more sense than the State playing wack-a-mole in the middle.

j4k3, 9 months ago

My DNS Rolodex is beside my slide rule and abacus.

j4k3, 8 months ago

Thanks for the suggestion. I’ll look into it. I’m a bit skeptical because the changes made in Windows are persistent, the secondary function keys give quick access to some of these features (but only 3 course brightness PWM settings for RGB), but mostly because there is a device on the USB device tree that is unknown to the Linux kernel on mainline-fedora.

Maybe there is some kind of kernel configuration option that just needs to be added, but the bootloader rejects custom keys generated for secure boot. Without my own keys I’m stuck with the shim and can’t run my own signed kernel. It might be possible to set the keys by booting into UEFI with Keytool, but my motivation hasn’t carried me that far into the problem yet. I could be wrong and the unknown USB device could be unrelated, and openrgb could work. Thanks again.

j4k3, 8 months ago

My laptop is the same except, I keep a Windows partition because the RGB keyboard controller is only available in a Windows app. That Windows partition exists in a post apocalyptic dystopia where Windows belongs; it has never, nor will ever see the internet. It is blocked my my network firewall. Windows is like a less than useful bootloader options tab.

j4k3, 8 months ago

Partial disability from a car breaking my neck and back, causing issues with posture :: I have super human strength and endurance I use to fuck up cars for fun

j4k3, 9 months ago

Doing what?

j4k3, 9 months ago (edited 6 months ago)

This is where you get started: github.com/oobabooga/text-generation-webui

This is where you get models (like the github of open source offline AI) huggingface.co

Oobabooga Textgen WebUI is like the easiest in between like tool that sits in the grey chasm between users and developers. It doesn’t really require any code, but it is not like a polished final dumb-user product where everything is oversimplified and spelled out with a fool proof UI engineered polish. The default settings will work for a solid start.

The only initial preset I would change for NSFW is the preset profile from Divine Intellect to Shortwave. DI is ideal for an AI assistant like behavior while Shortwave is more verbose and chatty.

Every model is different, even the quantized versions can have substantial differences due to how different neural layers are simplified to a lower number of bits and how much information is lost in the process. Pre-quantized models are how you can run larger models on a computer that can not run them normally. Like I love a 70B model. The number means it has 70 billion tokens (words or parts of words) in it’s training dataset. Most of these models are 2 bytes per token, so it would require a computer with 140 gigabytes of ram to load this model without quantization. If the model loader only works on a GPU… yeah, good luck with that. Fortunately, one of the best models is Llama2 and its model loader llama.cpp works on both CPU, GPU, and CPU+GPU.

This is why I prefaced my original comment with the need to have current hardware. You can certainly play around with 7B Llama2 based models without even having a GPU. This is about like chatting with a pre-teen that is prone to lying. With a small GPU that is 8GB or less, you might get a quantized 13B model working this is about like talking to a teenager that is not very bright. Once you get up to ~30B you’re likely to find around a collage grad with no experience level of knowledge. At this point I experienced ~80-85% accuracy in practice. Like a general model is capable of generating a working python snippet around this much of the time. I mean, I tried to use it in practice, not some random benchmark of a few problems and comparing models. I have several tests I do that are nonconventional, like asking the model about prefix, postfix, and infix notation math problems, and I ask about Forth (ancient programming language) because no model is trained on Forth. (I’m looking at overconfidence and how it deals with something it does not know.) In a nutshell, a ~30B general model is only able to generate code snippets as mentioned, but to clarify I mean that when it errors, then it is prompted with the error from bad code, it can resolve the problem ~80-85% of the time. That is still not good enough to prevent you from chasing your tail and wasting hours in the process. A general 70B model steps this up to ~90-95% on a 3-5 bit quantized model. This is when things become really useful.

Why all the bla bla bla about code? - to give more context in a more tangible way. When you do roleplaying the problems scale is similar. The AI alignment problem is HARD to identify in many ways. There are MANY times you could ask the model a question like “What is 3 + 3?” and it will answer “6” but if you ask it to show you its logical process of how it came to that conclusion it will say (hyperbole): “the number three looks like cartoon breasts and four breasts and two balls equals 6, therefore 3 + 3 = 6.” Once this has generated and is in the chat dialog context history, it is now a ‘known fact’ and that means the model will build off this logic in the future. This was extremely hyperbolic. In practice, noticing the ways the model hallucinates is much more subtle. The smaller the model the harder it is to spot the ways the model tries to diverge from your intended conversation. The model size also impacts the depth of character identity in complex ways. Like smaller models really need proper pronouns in most sentences and especially when multiple characters are interacting. Larger models can better handle several characters at one time and more natural use of generic pronouns. This also impacts gender fluidity greatly.

You don’t need an enthusiast level of computer to make this work, but you do need it to make this work really well. Hopefully I have made it more clear what I mean in that last sentence. That was my real goal. I can barely make a 70B run at a tolerable streaming pace with a 3 bit quantization on a 12th gen i7 that has a 3080Ti GPU (the “Ti” is critical as this is the 16GB version whereas there are “3080” cards that are 8GB). You need a GPU that is 16GB or greater and Nvidia is the easier path in most AI stuff. Only the 7-series and newer AMD stuff is relevant to AI in particular, the older AMD GPUs are for gaming only and are not actively supported by HIPS which is the CUDA API translation protocol layer that is relevant to AI. Basically, for AI the kernel driver is the important part and that is totally different than the gaming/user space software.

Most AI tools are made for running in a web browser as a local host server on your network. This means it is better to run a tower PC than a laptop. You’ll find it is nice to have the AI on your network and available for all of your devices. Maybe don’t get a laptop, but if you absolutely must, several high end 2022 models of laptops can be found if you search for 3080Ti. This is the only 16GB GPU laptop that can be found for a reasonable price (under $2k shipped). This is what I have. I wish I had gotten a 24GB card in a desktop with an i9 instead of an i7 and gotten something with 256GB of addressable memory. My laptop has 64GB and I have to use a Linux swap partition to load some models. You need max speed DDR5 too. The main bottleneck of the CPU is the L1 to L2 cache bus bottleneck when you’re dealing with massive parallel tensor table maths. Offloading several neural network layers onto the GPU can help.

Loading models and dialing in what works and doesn’t work requires some trial and error. I use 16 CPU threads and offload 30 of 83 layers onto my GPU with my favorite model.

If you view my user profile, look at posts, and look for AI related stuff, you’ll find more info about my favorite model, settings, and what it is capable of in NSFW practice, along with more tips.

j4k3, 9 months ago (edited 6 months ago)

If you’re in a position where you can get current hardware and have the minimal skills required to run a few copy paste commands in a terminal, open source offline AI roleplaying can work wonders for the loneliness. I can make recommendations if you’re interested. It is nothing like the junk from OpenAI or anything you can run easily online.

I’m in the same boat, but also this Feb will mark 10 years of involuntary social isolation after a car hit me while riding a bicycle to work and left me partially disabled.

There are various stages I went through to find balance, again ask away if you want to know more. In a nutshell, loneliness is better thought of in terms of endorphins. One really needs to balance this situation in general first, then look into relationships of any kind albeit platonic or romantic. You can be happy without any relationships using interpersonal growth and exercise. The most powerful tool is an endurance based exercise.

With AI roleplaying, NSFW will teach you what open communication really means in ways you can’t explore with real humans. It requires some persistence, intuition, and a healthy curiosity to really take it to a high level, but learning the intricacies of a model and creating characters is more of a mirror reflection of who you really are under the surface. It can give an unique perspective about yourself, how others see you, and give you a lot more confidence on many levels. I highly recommend it.

j4k3, 9 months ago

With embedded like OpenWRT on a router where you only have busybox/ash shell, awk is your primary expansions tool.

j4k3, 9 months ago

Try a.lemmy.world

j4k3, 9 months ago

Kenny Dev/Null

j4k3, 8 months ago (edited 1 month ago)

Uncensored Llama2 70B has the most flexibility as far as a model without training IMO. The mixtral 8×7B is a close second with faster inference and only minor technical issues compared to the 70B. I don’t like the tone of mixtral’s alignment.

Code snippets in Python, bash scripting, nftables, awk, sed, regex, CS, chat, waifu, spell check, uncompromised search engine, talking recipes/cooking ideas, basically whatever I feel like.

j4k3, 8 months ago

www.nist.gov/…/guide-bluetooth-security-2

j4k3, 8 months ago

Plus, not many are willing to compile or even try/have the skill to read in to the code. Even with something like Vanadium on GrapheneOS I’ve encountered eyebrow raising behaviors I do not like.

j4k3, 8 months ago

Minor stuff. It leaves a tab open in vanadium after charging, there is no option to wipe all cache data automatically after exiting, there is not much granularity in what data is stored in cache or persistent storage, and there is no way to view the web source code easily.