j4k3

j4k3 , 2 hours ago

29 May 1453

j4k3 , 4 hours ago

Tears of the Kingdom. It just got too grindy for me. I rarely play anything any more anyways.

j4k3 , 12 hours ago (edited 11 hours ago)

Archive.ph alt link:

https://archive.ph/IYtPj

Gov. Gavin Newsom has set in motion the largest land return in California history, declaring his support for the return of ancestral lands to the Shasta Indian Nation that were seized a century ago and submerged.
The 2,800 acres in Siskiyou County are part of the Klamath River dam removal project, which will rehabilitate more than 300 miles of salmon habitat.
“This is a down payment on the state’s commitment to do better by the Native American communities who have called this land home since time immemorial,” Newsom said in a statement. The governor’s announcement Tuesday marked the fifth anniversary of California’s official apology to its Native American peoples for the state’s historical wrongdoings.

cont...

AITA for feeling torn between justice for mistreated people, a desire for diversity, but also a dislike for spiritism of any sort albeit Abrahamic or otherwise? Don't be an A for someone asking a tough question and trying to nav their biases.

I would probably react negatively, without thinking a few years ago if I saw someone ask this question, so I totally understand the tendency to do so. My question is based on two situations. The telescopes on Mauna Kea in Hawaii and resistance to them, and watching Nick Zentner's geology YT uploads where he tries to include some natives and their perspectives. I wanted so bad to see value in what they had to add, like the potential for their oral history to have some missing value overlooked by science, or some narrative like that, but it was always like watching a toddler tell the most cringe stories. Like I expected some kind of consistency in stories like, 'this is how we tell the stories and how we remember them.' However, instead it was like everything I disliked about growing up in conservative Christian cult like extremism. Every question asked had an answer off the cuff. Not some kind of rehearsed thing they knew, but like a child making it up as they went. The focus was not factual or an alternate useful approach. It came across to me as someone that desperately wanted validation of ineptitude.

I'll readily admit I have a strong prejudice against spiritism. I really want to support righting injustices for people. I've experienced the injustice of having my life all but taken from me at the hands of a stranger driving a car poorly. I totally get behind that part. I don't expect people to integrate into religious culture, but I do expect people to modernize when advancement is shown. Like if the mountains near my house are a world class opportunity to advance human understanding, I don't care about the bullshit rituals some church did up there in the past or what my idiot family did for generations. I appreciate the opportunity to advance all of humanity's knowledge and future. I have a hard time feeling empathy for anyone that feels otherwise, like they are little more than an anarchic regressive force. Feel free to make a counter point in a productive way, but know that I walk away from rude or negative people just like in the real world. So I'll delete this if people downvote because they can't handle real conversations or controversy without being a negative person.

j4k3 , 9 hours ago

See this is why I like to talk and ask this kind of thing. I agree with you and your perspective. I mentioned it because I really don't know how I feel on the subject. I speak to my hesitation; not from my actionable stance.

I still struggle with the idea of validating stupidity and the long term implications that will cause in the future. Like monarchy, under the right person and circumstances is great. The problem is that it is impossible to deal with the succession crisis successfully as history has proven. Equality only works when it is kept in balance with a meritocracy to some extent, lest we elect a government full of mental health patients in the name of equality. On the level of nature itself, we are not actually equal. Our relative differences are usually minor, but we are not all leaders in a respective academic field. Are there positive outcomes that result from placing people of dubious capability in charge of such things?

Like ages ago I had a car break down in the middle of New Mexico on an Indian reservation. People were super nice and helpful, but that place was in terrible shape. I can't help but think they would be far better off if they had much better opportunities that bridged the past and present. Something like free rides in education and tax breaks for employers that hire Native Americans.

I still feel like placing feelings above fundamental logic skills is a path with no future. I don't believe in giving a flat earther a seat at a science conference. Do you?

j4k3 , 14 hours ago (edited 13 hours ago)

Trump: "I love your 10 commandments," you insolate moron. They mark you as a gullible fool that is easy to manipulate because you oversimplify from a lack of cognitive depth and believe in magic. I can say and do anything, and you'll charge into my sword because you see red and believe. At least a bull gives some fight, you on the other hand...

j4k3 OP , 15 hours ago

Yeah, but depends on a person's goals. I don't mind being doxed. The privacy thing I'm really concerned about is manipulation of data related to the host server; apps that are used like data loggers of sensors; tracking dwell time; page views; likes, blocks, etc. I care far less about what I say to others in public. I vehemently claim that owning the data about any individual is theft of autonomy, failure of democracy and government, and a form of slavery if one plays out the total philosophical circumstance and implications. Anyone that holds such data about someone else with the intent to manipulate in any way whatsoever is a criminal. I've been a Buyer for a retail chain, collected and analysed tons of customer data. This has nothing to do with how data is collected and used now, but this is used as justification for the present criminal data manipulation industry.

As a disabled person, I need to connect with humans more, and as much as I can here. I totally respect those of you that have other priorities that limit your conversational topics of interest, and I don't wish to violate those. This place is just my version of a public square, where I'm trying to make general conversation. -warmly

j4k3 , 1 day ago

Largish yellow raisins. I like them in oatmeal cookies and with mixed nuts cooked in a nice long grain rice with some saffron.

j4k3 , 1 day ago

Not a blinker in the group, sudden lane changes, tailgating; what else?

j4k3 , 1 day ago

What kind of personality is always aware of this kind of perceptive interaction, and what are their core thought functions?

j4k3 , 1 day ago

Thanks. I'm terrible about thinking this way. I always follow my curiosity and really only care about an abstract understanding any people that like to explore. When I'm the boss, no one works for me; they work with me and take on responsibility or I do it myself and get rid of them. I can do it all; not the best or the fastest, but I can do about anything if I really try. To me, social dynamics seem childish, but I also suck at things like emotional reinforcement and coercive sales. Most places I've worked, I wind up operating outside of any management structure, set my own hours, etc.

j4k3 , 1 day ago (edited 1 day ago)

Definitely. I do exactly that. I do not want to do it all myself and I do not micromanage. I simply expect others to take on what they can along side me and build a sense of competition between us.

My approach stems from owning a body shop twice and hiring people I liked from the dealerships I worked for all the time. I never hired anyone with experience in auto body because I specialized in a very specific niche type of repair and used nonstandard techniques. I could and did occasionally do traditional work, but the majority of the time I did not. So I had to teach everyone everything about the job. I have absolutely no problem doing the most menial task if that is what needs to be done in the moment; nothing is beneath me, and I expect that kind of attitude from everyone I work with, just do what needs to be done and move on. I won't waste a minute with anyone that resists that kind of thing; I find it adolescent and childish. I know I'm a jack of all trades, and not the best at any given task, but I'm excellent at plugging the holes and working more like a group of friends.

j4k3 OP , 2 days ago

We are merely the space ships of the microbes. Being caught between their war of the worlds would likely be unpleasant. However, I have a feeling we would be the true harbinger of death in that world of far more simple food webs than our present reality.

Our deaths would be miserable upon return, due to the evolutionary life of our passengers having many millions of years to plot our deaths when we return to find a completely different world and the little ones waiting to say hello.

I do not believe the table is intended for individual timescales, like how I tried to make the title more relatable. It is intended on geological timescales of species survival.

IIRC the person on YT also mentioned there were times when there was not enough oxygen to sustain fire, which is another fascinating thought and helps make sense of things like hydrocarbon deposits like oil and coal in some instances.

j4k3 OP , 2 days ago

Only for the oil industry business cards.

j4k3 OP , 2 days ago

I think the white paper it came from is referenced in the bottom left box.

j4k3 OP , 1 day ago

Problem with time travel is all the miscalculations trying to hit the thin skin of a moving object. Don't forget the undulations through the galactic arms and the effect of the galactic medium against the heliopause. Good luck! Never know, maybe you'll be the first above ground and non orbital. Bring a parachute, ablation shield, a shovel, and say hello to Hawking!

P.S. If you get a chance, tell me to call in sick for work 2/26/14. I'd appreciate not being disabled in that alt timeline. Thx!

j4k3 , 2 days ago

It isn't actually a mistake. It is a transition phase of generation. It is due to how the setting iteratively build. It is also a poorly designed generative attempt. People often try to pick general settings that will work for multiple images, but it is impossible to get high detail and contrast without causing major iterative changes like this.

A better workflow involves several stages of generation for a narrow style of image. With a proper workflow, each stage works in harmony to take the image to a desired level.

An image like this is simply a person using a sawed off shotgun to open a bottle of ketchup. It has nothing to do with the AI and everything to do with the person that created the prompt.

j4k3 , 3 days ago

I don't trust AliEx any more after I took the loss of 3 orders for ~$60 in 2020. When I called, they hung up on me at random every time. After the 3rd try I washed my hands and walked away. Stealing from me once is on them, twice would be my own fault. Prior to that experience I spent a few thousand dollars on the platform for odds and ends.

I expect something like this to be an emulator and nowhere near the quality of a real Nintendo product, but I could be wrong. I would buy used or a homebrew project that is well documented and might cost a little more.

j4k3 , 2 days ago

Diffusion models do not parse natural language like a LLM. All behavior that appears like NLP is illusionary. For the most part. You can get away with some things because of what is present in the training corpus. However, any time you use a noun, you are making a weighted image priority. By repeating "shuttle" in this prompt, you've heavily biased to feature the shuttle regardless of the surrounding context. It is not contextualising, it is 'word weighting'. There is a relationship to the other words of the prompt, but they are not conceptually connected.

In a LLM there are special tokens that are used to dynamically ensure that the key points of the input are connected to the output, but this system is not present in generative AI.

To illustrate, I like to download LoRA's to use on offline models, I use a few tools to probe them and determine how they were made, like the tags used with training images, what base model was used, and the training settings they used. Around a third of LoRA's I have downloaded contain natural language in the images that were tagged. This means the LoRA related term I use for generating should be done with natural language.

This is the same principal required for any model. You should always ask yourself, how often is this terminology occurring in the tags below an image. You might check out gelbooru or danbooru just to have a look at the tags system used there for all images. That is very similar to how training happens for the vast majority of imagery. It is very simplified overall.

The negative prompt is very different in how it is processed compared to the positive. If you look at the respective documentation for the tool you're using, they might make some syntax available to create a negative line, but they likely want you to use their API with a more advanced tool.

j4k3 , 4 days ago (edited 4 days ago)

In the last century? The diode, aka the P/N junction and every variant that has been created ever since.

Recently? Capacitive touch screens are by far the most significant change.

j4k3 , 5 days ago

At 1:30 am? Two obnoxious cats licking themselves passive aggressively competing for attention the moment I put down this idiot brick.

j4k3 OP , 6 days ago (edited 6 days ago)

Your image doesn't have the workflow for me to have a look. You would need the exact same setup with the unique prompt in all 3 positive positions. I'm not positive my custom sampler setup is fully transferable. If you manually tried to add the prompt, that won't work. It could be how my GPU generates seeds too. I think there is something I've read about that being hardware specific in its determinism.

j4k3 OP , 6 days ago

Medium. I could only get the one with everything packaged to work so far. I tried the one with just the two clip's packaged, but I only get errors when trying to combine the T5 clip and the other two. I need to get all the pieces separately. Running the T5 in fp16 will likely make a massive difference.

I also need to find a way to read the white paper. I'm not willing to connect to their joke square space website and all of Google's garbage just to get the paper. I'll spin up another machine on a guest network and VPN if I need tomorrow. There is a difference between the two clip models and their functionality that I need to understand better and isn't evident in the example workflows or written about on civitai.

j4k3 OP , 6 days ago

Not yet. The SD3 checkpoint is too raw and needs a lot of fine tuning. It is also way more complicated with more control variables.

j4k3 OP , 6 days ago

ComfyUI embeds the entire workflow (all nodes and connections) inside the metadata of each image. Lemmy wipes all metadata if the instances are hosting the image. I share the original link because that version has the metadata still attached. If you drag it into the Comfy interface it will load up the whole thing automatically.

j4k3 OP , 6 days ago

I edited the post with more of the image set I generated while getting to this one.

j4k3 OP , 5 days ago

I think the difference is typical of any base model. I have several base models on my computer and the behavior of SD3 is quite typical. I fully expect their website hosts a fine tune version.

There are a lot of cultural expectations that any given group around the world has about generative AI and far more use cases than any of us can imagine. The base models have an unbiased diversity that reflects their general use; much is possible, but much is hard.

If "woman lying in grass" was truly filtered, what I showed here would not be possible. If you haven't seen it, I edited the post with several of the images in the chain I used to get to the main post image here. The post image is not an anomaly that got through a filter, it is an iterative chain. It is not an easy path to find, but it does exist in the base training corpus.

Personally, I think the real secret sauce is the middle CLIP agent and how it relates to the T5 agent.

j4k3 , 6 days ago

Man dies killing two cars on bicycle, his zombie stole my KOM ten years later.

j4k3 , 6 days ago

Spoiler

"The hackers chainsaw" - Forth

Old Reverse Polish Yoda can kill with one Word.

j4k3 , 7 days ago (edited 6 days ago)

So they can't completely turn off the extra instructions from the enterprise P-cores?

If Linux goes ahead and lets the scheduler use the full enterprise microcode, do you think we'll find that the actual instructions are not fused?

I mean w11 was basically initiated because of Intel's asymmetrical cores and the w10 scheduler's limitations, as far as I understand it.

I just want the full enterprise P-core AVX instruction set, and I have the 12th gen that is most likely to work with the right microcode. It would just be funny if W11 doesn't support such a complex scheduler and Linux does. The implications would be large.

j4k3 OP , 7 days ago

The pectoralis major and the linea alba are a give away, along with the hint of a umbilicus region, aka the breasts/nipples/belly button and that line down the middle of the abdomen. Those are all primate features. I also know because of how generative output changed when I added monkey to the negative prompt.

j4k3 OP , 6 days ago

prompt:

Positive: 32k UHD photo professional wildlife photography, diffuse lighting, funny meme awkward ugly, short kangaroo arms, kangaroo body, forelimbs with small claws, muscular hind limbs for jumping, strong tail, standing upright pooping, pug wrinkly flat muzzled face, pug ears folded with the front edge against the side of the head, fat, depressed expression, in New Guinea, side view

Negative: dog body, photoshop, fake, shadow, clothing, harness, nipples

It is a combination of how:

• I declared both arms and forelimbs, which likely started creating ambiguity.
• Then I described hind legs for jumping when I should have used the word hopping.
• And, I had used a lot of previous momentum after a random image looked like a dog pooping but had errors and I tried to see how far I could push this complex instruction. I was curious if I could take this as far as turning this into a pet that someone was walking in a neighborhood on a leash, but that proved too complex for the generative untrained model by itself.

It turned out that "pooping" was a major keyword to randomize the outputs. That let me group a set of pictures that I was able to use in a more flexible way to get to here:

https://files.catbox.moe/z3b0bx.png

This one follows the prompt much more closely.

prompt: “32k UHD photo professional wildlife photography, diffuse lighting, funny meme awkward ugly, short kangaroo arms, kangaroo body, forelimbs with small claws, muscular hind limbs for jumping, strong tail, upright, pug wrinkly flat muzzled face, pug ears folded with the front edge against the side of the head, fat wrinkly skin throughout, depressed expression, upright hopping awkwardly, off balance mid air, in the overgrown remote wild of New Guinea, side view”

neg: dog body, photoshop, fake, shadow, clothing, harness, large forelimbs, (nipples breasts)

The image in this post was the second most funny gen out of a couple hundred so I shared it here despite the slight deviation from my instructions, and because it is an excellent seed to play around with to generate similar variety.

j4k3 , 7 days ago

You need to be sure your instance in federated with the instance that you are engaging with and visa versa. It is always listed on the ../instances page. So your host is:

https://linux.community/instances

j4k3 , 7 days ago (edited 7 days ago)

The Lemmy algorithm:

https://join-lemmy.org/docs/contributors/07-ranking-algo.html

The instances communicate between each other using special bots in the background that transport info between instances. Dot world is too big and makes too many requests to other instances. Most instances are reducing the number of transport bots back to dot world right now. There is supposed to be a fix in the next Lemmy version, but the code base is somewhat slow moving due to only two devs and it is written in Rust. Rust is a hard language like C, and not too many here are able to contribute to it, even though it is like the new gold standard of code.

So you might see a delay between posting and replies or the interaction may come in bursts that correspond with the transport bots carrying content between instances as the host admin have configured their instance.

j4k3 , 7 days ago

Not sure. There was some controversy with some of the devs making alt front ends and admins complaining about the slowness. I've seen mention of one of the two devs learning Rust just to participate. So it is not entirely a walled garden. The front end devs wanted to make an alt from scratch but in something like JavaScript although I don't recall the details exactly. There were a lot of red flags related to privacy and understanding the community at large in the posts I saw from them. When asked why they weren't adding pull requests with Rust in order to address their complaints I got no reply.

All that said, I'm no dev. I can read in to around half the code I come across if I really try, and can successfully modify maybe half of that if I spend a few days on it, but I suck at clever code and the DRY cult types. I haven't tried to look into Lemmy in any depth beyond figuring out the basics.

j4k3 , 7 days ago

The .ml instance is run by the developers. If it isn't well configured we would have no reason to be here.

j4k3 , 6 days ago

It follows the first to market principal in many of the most active communities, and it is the most federated instance. Many instances that are not federated with each other are federated with .ml. You still won't see those comments between instances. Like from my main account here on .world, I can't see hexibear or beehaw stuff, but from my .ml account I can see them. I have accounts on many instances in order to help federate new communities and to check biases/instance behaviors.

I came over a few days before the rexodus and subscribed to the active communities before the influx. That sub list is still centered around the most active communities, and the majority of those are from .ml and before I joined Lemmy.

Instances all have different flavors. I don't like using my .ml account as a main. I've tried it. But I find they are the center of the most interesting and productive conversations for a more broad audience, while Beehaw has the most positive and friendly conversations overall. The main benefit to .world is the speed of connectivity, general audience scope, but with a strong anti asshat policy.

j4k3 , 7 days ago (edited 6 days ago)

The Pugaroo of New Guinea

prompt: "32k UHD photo professional wildlife photography, diffuse lighting, funny meme awkward ugly, short kangaroo arms, kangaroo body, forelimbs with small claws, muscular hind limbs for jumping, strong tail, upright, pug wrinkly flat muzzled face, pug ears folded with the front edge against the side of the head, fat wrinkly skin throughout, depressed expression, upright hopping awkwardly, off balance mid air, in the overgrown remote wild of New Guinea, side view"

neg: dog body, photoshop, fake, shadow, clothing, harness, large forelimbs, (nipples breasts)

The full ComfyUI workflow and prompt is attached to the image: https://files.catbox.moe/z3b0bx.png

Edit: Correct "New Zealand" to "New Guinea" in title to match prompt.

https://files.catbox.moe/z3b0bx.png

j4k3 , 6 days ago

Fidochondria - I like it.

j4k3 , 8 days ago

Adding to what others have already mentioned... Most of the gold will be from various collisions of external objects. The vast majority of the gold and other heavy elements are in Earth's core due to gravitational differentiation.

There is a volcano (in South America IIRC) that has unusually high gold content, but it is from the underground magma reservoir coming in contact with gold deposits. This is why space mining is a really big deal. The Earth is a resource poor gravity prison by comparison. The wealth in space is enormous compared to any differentiated body.

Gold is actually everywhere and relatively common, but only in very small quantities. Under the right conditions, the weight can help gold to concentrate and fall out of solution when that solution was once covering a very large area, dissolved the tiny bits of gold found all over a large area, and then pools into a low point over extremely long periods of time.

j4k3 , 8 days ago

Just keep going until you hit China. Five year old me will be blown away.

It would be far easier to hit up some near Earth objects, then maybe Vesta, but Ceres is the mother load. Small enough to be solid, but large enough to be spherical and therefore somewhat differentiated if you can get to the center of that Tootsie Pop.

Maybe in the process we'll learn enough to effectively mine and utilize Phobos, the larger moon of Mars at 11 km in mean radius. That is the largest accessible orbital habitat that we have available (in flat pack IKEA furniture form) and it only has around 30-50 million years before it hits the Roche limit and turns into a ring, unless we manage to intervene.

j4k3 , 8 days ago

They usually seem that way at first. I think it is a case of the massive amount of data available initially; like it is unbiased. I think of it like all the shelves in an enormous library lined up on a giant wall. It is very capable and full of information, but lacks any kind of focus.

Also, when training the big models, as far as I understand it, they train until they find the sweat spot and then go past it. Once they know the curve of how much training is "overtrained," they go back to a point a little bit before the peak so that fine tuning should place the model at the peak.

The raw checkpoints usually lack focus in both positive and negative directions. It can be similar to how LLM's need really good instructions that help Name-2 understand its own role. If you try and define this role even with diffusion, you're likely to get a change, maybe an improvement. Something like this may help, "You are a helpful generative AI that follows the prompt details exactly.\n\nPrompt:". That actually helped with Pony when I tried it on the base checkpoint recently.

j4k3 , 8 days ago

Nostalgia was messing with my head too much to play along so don't feel bad.