j4k3

joined 1 year ago
MODERATOR OF
[–] j4k3@lemmy.world 12 points 3 hours ago

He married into it and took his wife's last name; a proletariat Marx-in-law

[–] j4k3@lemmy.world 7 points 4 hours ago

I didn't watch the debate and neither did most of the sane world. Sharing the stage with a felon is the only atrocity I see. Giving that traitor a platform is crime. The billionaire owned news media trying to fabricate drama and controversy is no surprise. We don't have legitimate media or freedom of information. We have lots of psyops bullshit and lots of authoritarian surveillance nonsense.

[–] j4k3@lemmy.world 12 points 4 hours ago

Are you trying to tell me that the host server is showing the ad, because last I checked, with my whitelist firewall, I never see ads because all ads are links to the ad server you are actually visiting. It is no different than opening up the webpage and connection to them. They get all the same fingerprinting info.

I'm not saying one way or another here, but there is no such thing as anonymous data collection. It only takes 2-3 unique identifiers to connect a person between a known and anonymous data set and there are almost always quite a few more unique identifiers than this in any given dataset. When I hear anyone say stalkerware is anonymous, I assume they are no longer just a privateer of a foreign drug cartel level state, instead they are full blown slave trader pirates fit for the gallows or worse.

[–] j4k3@lemmy.world 5 points 5 hours ago

It makes a huge difference running bigger models. I like to run a quantized 70b or 8×7b most. These large models are far easier to access their true depth with less momentum required to find it.

The issue is not memory bandwidth in general. The primary bottleneck is the L2 to L1 bus width. That is the narrowest point, and it is designed for the typical pipeline of a sequential processor running at insanely fast speeds that are not possible if things get more spread out. The issues are more like radio than typical electronics. The route lengths, capacitance, and inductances become super critical.

It will make a major difference just to add the avx instruction set to consumer processors. Those are already present but fused off or simply not listed in the microcode in some instances. The full AVX instructions are not used because you would need to have a processes scheduler that is a good bit more complicated. These types of complex schedulers already exist in some mobile ARM devices but for more simple types of hardware and software systems, and without the backwards compatibility that is the whole reason x86 is a thing. The more advanced AVX instructions do things like loading a 512 bit wide word in a single instruction.

Hardware moves super slow, like 10 years for a total redesign. The module nature of ARM makes it a little easier to make some minor alternation. The market shifted substantially with AI a year and a half back. Right now, everything we are seeing was already in the pipeline long before the AI demand. I expect the first real products to ship in 2 years, 3-4 before any are worth spending a few bucks on, and in ~8 years, hardware from right now will feel as archaic as stuff from 20-30 years ago.

Running the bigger model makes a huge difference. Saying it will run at those speeds is actually making a statement about agents and augmented generation. I can run my largest models with streaming text barely faster than my reading pace. I can't do extra stuff with that kind of model because I would need the entire text to send to other code or models for further processing. This speed is saying, I could do many things like text to speech, speech to text, augmented data retrieval for citations, function calling in code, and running several models to do unique things while keeping a conversational pace.

[–] j4k3@lemmy.world 5 points 16 hours ago

The way the API changes were done showed a disconnect between public best interests as a public commons and corporate interests to monetize. It implied that I was being targeted individually for monetization. I feel that anyone collecting individual data about any human and selling that data is a new form of slavery through an ownership of a part of that person with the intent to manipulate. The manipulation of information through the nondeterministic targeting of search results is a coup of a pillar of democracy and all governments of the world. The free press must apply to all information on the internet. With the monopoly of only 2 relevant web crawlers providing results directly or indirectly, there is no freedom of information in digital form. This would be no different than every news paper stand being owned by two companies a hundred years ago. Targeting the individual directly is what the API move was designed to handle. So, to me, it was an attempt to enslave my digital autonomous person. When faced with such a subtle attempt to subterfuge one's autonomy, I feel like the choice was obvious.

Everything I say here is scraped, but only the server host knows my dwell time, sensors, and various fingerprinting mechanisms. Ideally I would self host, but lack the skill and resources. This place is still hosted in a datacenter, but I'm using the API through a 3rd party, so it is even more obscure. I used reddit through an app with a scraped interface before, when that quit working, I quit using the site. Reddit proved it can only get worse, not better.

[–] j4k3@lemmy.world 16 points 16 hours ago

Wouldn't need to fight for justice if felons were put in prison once convicted.

Anyone flying the banner of this coup attempting monster is a traitor to god and country.

[–] j4k3@lemmy.world 0 points 17 hours ago (2 children)

Does Apple lack a feature to turn off or hide the file menu?

I don't recall if I custom bound mine or if GNOME defaults to, but I never see those old things unless I press F10.

I agree that it is not very pretty having all three of those bars in growing redundant nonsense. Tiling frames are like desktop icons IMO; very retirement-center historical-preservation aesthetic design, but we're talking about desktop environments at that point, not the app dev. Apple wants things to be their little aesthetic way to monopolize, so they intentionally make decisions that erode the usefulness of outside package standards.

[–] j4k3@lemmy.world 1 points 1 day ago

That's one way of putting it. I find it has the depth to explore in interesting ways with creative prompting, it is just handicapped by so much alignment nonsense that it lacks mobility into and between those spaces.

[–] j4k3@lemmy.world 3 points 2 days ago (5 children)

A fallen hero.

[–] j4k3@lemmy.world 8 points 2 days ago

I, good sir, do not shit like mere humans. I sit upon my gilded high horse who does the shitting for me.

 

https://www.youtube.com/watch?v=ervuwSwKu18

Nice example of real reverse engineering at a low level in a ten minute dense summary, including die level comments on structures. Also, why Broadcom sucks in engineering terms (NDA nonsense).

 

I'm old-naughties internet café crew from around the era of the original Counter Strike through CoD MWF3 and Battlefield 2042. I wouldn't mind reliving the nostalgia of that era, and have more than enough machine to do it now. I have zero tolerance for micro transactions, gatekeeping, or software as a service business models. I have little interest in Steam as anything more than a single banking transaction, and I will only connect to something online that is documented and strait forward when it comes to multiplayer networking stuff.

My networking is about like a locked down corporate connection. I can play something like 0 A.D. that has a few required and documented connections needed for multiplayer, but won't connect to discord when it requires around a dozen raw IP addresses on random ports with no documentation.

Do any of you know of games that fit those requirements and are worth spending money on?

::: spoiler I probably seem annoyingly specific...

...or even paranoid, but I've both downloaded a PDF in the past that wrecked a machine with malware from a 3rd party datasheet on vintage computing hardware, and I find that isolating myself from ads and data mining at this level is best for my mental health with isolation from disability; i.e. it is not just arbitrary pedantism in my mind.

I ask because the casual camaraderie could be good for me while pressure to be a consumer at the expense of my family is harmful. Plus I view subscriptions for anything like software as a criminal skimming scam I am not interested in at all. Hopefully you can understand the specificity, and slight excentricity, even if it seems silly at first. I'm not judging anyone for their practices, trying to make a statement about my standards by comparison, or expect anyone to adopt my practices. I don't think like that at all. I'm curious, exploring what I've assumed is not possible for me, and stating the needed info only. I'm also on Fedora WS, with the NVDK kernel module, and can't change my UEFI secure boot keys in the OEM bootloader, so I can't alter kernel space unless I figure out Keytool to boot into UEFI directly.

 

I've encountered this many times where I simply don't understand the context and use of an API based of the API documentation unless I can find an example that already utilizes it in a working project. The first thing that comes to mind is Py Torch. I've tried to figure out how some API features work, or what they are doing in model loader code related to checkpoint caching but failed to contextualize. What harebrain details are obviously missing from someone who asks such a silly question?

 

I got sent a bunch of lenses that are too small for the old frames I have. I thought it might be fun to try 3d printing my own frame arms out of ABS.

Modeling a lens is a bit challenging, or I'm just a bit rusty with freecad. I need a way to wrap/unwrap/flatten the curve to create the contours. For my purpose here I can get away with a rough lens shape that is simply mirrored and lacks some of the complex curves. It just bugs me when I can't model it properly...

The lens is not uniform thickness. The center is 1.8mm thick while the edges where the arms mount are 1.4mm. I want to try and make a nose piece because this is the one part that fails on these cheap glasses.

Anyways, sharing just because I can.

 

I know it is not a post process thing. Is it stuff applied to the mold or stuff inside the plastic itself? I mean things like buttons, toys, phone parts before the back glass phase; things that are super thin with a finished surface that is durable and bonded to the part.

6
submitted 3 weeks ago* (last edited 3 weeks ago) by j4k3@lemmy.world to c/asklemmy@lemmy.ml
 

I had a 1971 Toyota FJ40, and built a 1992 supercharged Camaro with a mini blower. Those were my coolest, but also a couple Fiero's, a Z32, and an E30.

...also the guy that rode a bicycle everywhere '09-'14 got hit 7 times, with the last one disabling me. So don't get too bent over car stuff. I've put in over 170k miles on a bicycle since '09.

 

https://en.wikipedia.org/wiki/Hyperoxia

This was a screenshot I took months ago while watching a Geology Hub upload on YT. It was a lightbulb moment for my understanding of mass extinction events, (the largest was 250ma). I've referenced this multiple times, so thought I might share. Perhaps you find it as interesting as I do.

 

If anyone needs to scratch a vintage software, hardware, and FPGA itch, you might check out his channel. I don't often watch it because John is long winded, but his project is still quite interesting. He is a good example of someone using FOSS toolchains too.

1
submitted 1 month ago* (last edited 1 month ago) by j4k3@lemmy.world to c/imageai@sh.itjust.works
 

The ComfyUI prompt and workflow is attached to the image: https://files.catbox.moe/s3qufb.png

You can't copy pasta this prompt. There are a few nodes that are specific to SD3 and required.

::: spoiler EDIT: more proof of the chain that lead to this image. They were not all this good. I'm cherry picking for sure and these are just webp's without workflows attached:

 

Prompt was through ComfyUI, so it is embedded in the image: https://files.catbox.moe/aiy8p1.png

It was supposed to be a pug kangaroo hybrid but the AI apparently wanted to throw in some monkey too.

 

Spots have turned to stripes on the bottom half of this one.

view more: next ›