YSK theres a open source tool to cleanly read webpage articles called 'NewsWaffle'
Its called that because you have a new appreciation for your life after coming back from that alive... "Thank God I didn't drop to my death!"
"Hey, how's it going"
stares blankly at you like a deer looking at headlights for 15.50 seconds with an uncomfortable silence.
"Good."
walks away
Ill have you know I run a dual side business of selling ball-bouncing-in-polygon software as NFTs as well as counting the r-'s in various spellings of strawberry for the private defense sector...
during the time I was born TVs were small square boxes powered by glass tubes and turny knobs. I want to say 480p but tbh if you were using a junky 10 inch display at the turn of the century on satallite it was closer to like 240p. The jump from square 480p to widescreen 720/1080 was an actual graphical revolution for most people in a very big way, especially for watching movies that were shot in wide. In terms of games 1080p is both where 16:9 took off and the point where realistic looking graphics meet acceptable resolution for like skin pours and godrays shit like that. GTA5, TLOU and RDR are the examples that come to mind from the AAA 1080p era and their original states still probably hold up today.
When the 4k stuff finally came around and it was advertised as the next revolution I was excited man. However compared to going from 480 to 1080 it wasn't a huge change tbh. It seems once you're already rendering skin detail and individual blades of grass, or simulating atmospheric condition godrays, there isn't much more that can be drastically improved just by throwing a billion more polygons at a mesh and upscaling textures. The compute power and storage space required to get these minimal detail gains also starts escalating hard. Its such bullshit that modern AAA games are like 80gb minimum with half of that probably being 4k textures.
I will say that im like the opposite of a graphics snob and slightly proud of it so my opinions on 4k and stuff are biased. Im happy with 1080p as a compromise between graphical quality and compute/disk space required. Ive never played a 1080p at maximum graphics and wanted for more. Im not a competitive esports player, im not a rich tech bro who can but the newest upgraded gpu and 500tb of storage. I don't need my games to look hyperrealistic. I play games for the fun gameplay and the novel experiences they provide. Some of the best games I've ever played look like shit and can be played on a potato. Most of the games I found boring were AAA beautiful open worlds that were as wide and pretty as an ocean but gameplay wise it was as deep as a dried up puddle. I hopped off the graphics train a very long time ago, so take my cloud yelling with a grain of salt.
"I use Arch bt-"
"ITS SHiTE!"
"...excuse me?"
" YOUR BLOODY ROLLING RELEASE DISTRO IS FUCKING RAW. HOW MANY TIMES HAVE YOU RECOOKED IT AFTER A DEPENDENCY PACKAGE BROKE?"
"B-bhut chef... Its a rolling release bleeding distro that expects users to compile with the help of a wik-"
"I ASKED HOW MANY TIMES YOU HAD TO RECOMPILE IT THIS YEAR YOU FUCKING DONKEY"
"5 times sir."
"FIVE FUCKING TIMES??? JESUS CHRIST DID I ASK FOR CONSTANT MAINTENANCE WITH A SIDE OF COMPUTER PROGRAMS IN BETWEEN? IF I WANTED A RAW OPERATING SYSTEM I WOULD HAVE BECOME A FLAGSMAN INSTEAD OF A CHEF AND ASKED FOR A DISH OF "GENTOO". COOK ME A REAL OPERATING SYSTEM."
Ken Cheng is a great satirist and probably knows thats not how it works anymore. Most model makers stopped feeding random internet user garbage into training data years ago and instead started using collections of synthetic training data + hiring freelance 'trainers' for training data and RLHF.
Oh dont worry your comments are still getting scraped by the usual data collection groups for the usual ad selling and big brother bs. But these shitty AI poisoning ideas I see floating around on lemmy practically achieve little more than feel good circle jerking by people who dont really understand the science of machine learning models or the realities of their training data/usage in 2025. The only thing these poor people are poisoning is their own neural networks from hyper focusing defiance and rage on a new technology they can't stop or change in any meaningful way. Not that I blame them really tech bros and business runners are insufferable greedy pricks who have no respect for the humanities who think a computer generating an image is the same as human made art. Also its bs that big companies like meta/openAI got away with violating copyright protections to train their models without even a slap on the wrist. Thank goodness theres now global competition and models made from completely public domain data.
Some games just aren't meant for you and thats okay. For example I spent a few hours playing civ enough to understand the experience it offers. I did not enjoy a single moment of its gameplay or strategy layers at any point. Apparently its a good enough game for many people to put hundreds/thousands of hours into and buy again every few years+dlc. I just didn't pick up what it was putting down.
I have no issue with remakes themselves. Games are a kind of art, and good art should be kept alive for the next generations to enjoy. The problem to me is:
- the only thing big studios now want to put out remakes/remasters of the backlog they already made because its a safe and easy cash grab. One of the top comments about there being 7 skyrims and 2 oblivions before ES6 is soo real man. Its like all the people who founded the companies who were responsible for creative novel design/story that gave big titles their soul in the 2000s no longer exist in the industry except a few indie devs. Now all big game companies are just run by business associates without a shred of humanity outsourcing everything for a quick buck.
- Graphics have plateud from late 2010s and onward. Remastered and remaked stuff made a lot more since for the ps2/xbox and backwards, with the ps3/x360 1080p resolution it made a little less sense but I could still understand them porting like TLOU to ps4 at 4k or whatever. But now were remastering games that came out 5 years ago at 4k and trying to sell it as some huge graphical overhaul worth the asking price. Maybe im insane or old but my eyes can barely tell the difference between 1080p and 4k, going from 4k to 8k is like the same picture with slightly different shaders.
Try fallen Gemma its a finetune that has the positivity removed.
You are correct in your understanding. However the last part of your comment needs a big asterisk. Its important to consider quantization.
The full f16 deepseek r1 gguf from unsloth requires 1.34tb of ram. Good luck getting the ram sticks and channels for that.
The q4_km mid range quant is 404gb which would theoretically fit inside 512gb of ram with leftover room for context.
512gb of ram is still a lot, theoretical you could run a lower quant of r1 with 256gb of ram. Not super desirable but totally doable.
I have been using deephermes daily. I think CoT reasoning is so awesome and such a game changer! It really helps the model give better answers especially for hard logical problems. But I don't want it all the time especially on an already slow model. Being able to turn it on and off wirhout switching models is awesome. Mistral 24b deephermes is relatively uncensored, powerful and not painfully slow on my hardware. a high quant of llama 3.1 8b deephermes is able to fit entirely on my 8gb vram.
Very interesting stuff! Thanks for sharing.
What is it? Oh I see the sticker now :-) yes quite the beastly graphics card so much vram!
Its all about ram and vram. You can buy some cheap ram sticks get your system to like 128gb ram and run a low quant of the full deepseek. It wont be fast but it will work. Now if you want fast you need to be able to get the model on some graphics card vram ideally all of it. Thats where the high end Nvidia stuff comes in, getting 24gb of vram all on the same card at maximum band with speeds. Some people prefer macs or data center cards. You can use amd cards too its just not as well supported.
Localllama users tend use smaller models than the full deepseek r1 that fit on older cards. 32b partially offloaded between a older graphics card and ram sticks is around the limit of what a non dedicated hobbiest can achieve with ther already existing home hardware. Most are really happy with the performance of mistral small and qwen qwq and the deepseek distills. those that want more have the money to burn on multiple nvidia gpus and a server rack.
LLM wise Your phone can run 1-4b models, Your laptop 4-8b, your older gaming desktop with a 4-8gb vram card can run around 8-32b. Beyond that needs the big expensive 24gb cards and further beyond needs multiples of them.
Stable diffusion models in my experience is very compute intensive. Quantization degredation is much more apparent so You should have vram, a high quant model, and should limit canvas size as low as tolerable.
Hopefully we will get cheaper devices meant for AI hosting like cheaper versions of strix and digits.
Which ones are not actively spending an amount of money that scales directly with the number of users?
Most of these companies offer direct web/api access to their own cloud supercomputer datacenter, and All cloud services have some scaling with operation cost. The more users connect and use computer, the better hardware, processing power, and data connection needed to process all the users. Probably the smaller fine tuners like Nous Research that take a pre-cooked and open-licensed model, tweak it with their own dataset, then sell the cloud access at a profit with minimal operating cost, will do best with the scaling. They are also way way cheaper than big model access cost probably for similar reasons. Mistral and deepseek do things to optimize their models for better compute power efficency so they can afford to be cheaper on access.
OpenAI, claude, and google, are very expensive compared to competition and probably still operate at a loss considering compute cost to train the model + cost to maintain web/api hosting cloud datacenters. Its important to note that immediate profit is only one factor here. Many big well financed companies will happily eat the L on operating cost and electrical usage as long as they feel they can solidify their presence in the growing market early on to be a potential monopoly in the coming decades. Control, (social) power, lasting influence, data collection. These are some of the other valuable currencies corporations and governments recognize that they will exchange monetary currency for.
but its treated as the equivalent of electricity and its not
I assume you mean in a tech progression kind of way. A better comparison might be is that its being treated closer to the invention of transistors and computers. Before we could only do information processing with the cold hard certainty of logical bit calculations. We got by quite a while just cooking fancy logical programs to process inputs and outputs. Data communication, vector graphics and digital audio, cryptography, the internet, just about everything today is thanks to the humble transistor and logical gate, and the clever brains that assemble them into functioning tools.
Machine learning models are based on neuron brain structures and biological activation trigger pattern encoding layers. We have found both a way to train trillions of transtistors simulate the basic information pattern organizing systems living beings use, and a point in time which its technialy possible to have the compute available needed to do so. The perceptron was discovered in the 1940s. It took almost a century for computers and ML to catch up to the point of putting theory to practice. We couldn't create artificial computer brain structures and integrate them into consumer hardware 10 years ago, the only player then was google with their billion dollar datacenter and alphago/deepmind.
Its exciting new toy that people think can either improve their daily life or make them money, so people get carried away and over promise with hype and cram it into everything especially the stuff it makes no sense being in. Thats human nature for you. Only the future will tell whether this new way of precessing information will live up to the expectations of techbros and academics.
Theres more than just chatgpt and American data center/llm companies. Theres openAI, google and meta (american), mistral (French), alibaba and deepseek (china). Many more smaller companies that either make their own models or further finetune specialized models from the big ones. Its global competition, all of them occasionally releasing open weights models of different sizes for you to run your own on home consumer computer hardware. Dont like big models from American megacorps that were trained on stolen copyright infringed information? Use ones trained completely on open public domain information.
Your phone can run a 1-4b model, your laptop 4-8b, your desktop with a GPU 12-32b. No data is sent to servers when you self-host. This is also relevant for companies that data kept in house.
Like it or not machine learning models are here to stay. Two big points. One, you can self host open weights models trained on completely public domain knowledge or your own private datasets already. Two, It actually does provide useful functions to home users beyond being a chatbot. People have used machine learning models to make music, generate images/video, integrate home automation like lighting control with tool calling, see images for details including document scanning, boilerplate basic code logic, check for semantic mistakes that regular spell check wont pick up on. In business 'agenic tool calling' to integrate models as secretaries is popular. Nft and crypto are truly worthless in practice for anything but grifting with pump n dump and baseless speculative asset gambling. AI can at least make an attempt at a task you give it and either generally succeed or fail at it.
Models around 24-32b range in high quant are reasonably capable of basic information processing task and generally accurate domain knowledge. You can't treat it like a fact source because theres always a small statistical chance of it being wrong but its OK starting point for researching like Wikipedia.
My local colleges are researching multimodal llms recognizing the subtle patterns in billions of cancer cell photos to possibly help doctors better screen patients. I would love a vision model trained on public domain botany pictures that helps recognize poisonous or invasive plants.
The problem is that theres too much energy being spent training them. It takes a lot of energy in compute power to cook a model and further refine it. Its important for researchers to find more efficent ways to make them. Deepseek did this, they found a way to cook their models with way less energy and compute which is part of why that was exciting. Hopefully this energy can also come more from renewable instead of burning fuel.
My ravioli bowl won't unstick. Took about an hour of prying, and still I couldn't unstick the plate.
Jump
Assuming its empty, i would take the grog oggah boogah solution of smash the blue plastic bowl down the edge of your countertop. Something will give sometime.
Otherwise, did you try twisting the bowl one direction and the plate the other? Torque is typically a more effective force than pulling for friction.
the owner of the picture themselves possibly put on the tie on their cat used to thst kind of thing and lied about it for an internet caption meme. The facial expression of cat looks blurry but relaxed tbh its obviously well fed and groomed.
If you are asking questions try out deephermes finetune of llama 3.1 8b and turn on CoT reasoning with the special system prompt.
It really helps the smaller models come up with nicer answers but takes them a little more time to bake an answer with the thinking part. Its unreal how good models have come in a year thanks to leveraging reasoning in context space.
True! Most browsers don't have native gemini protocol support. However a web proxy like the ones I shared allow you to get gemini support no matter the web browser. Gemtext is a simplified version of markdown which means its not too hard to convert from gemtext to html/webpage. So, by scraping information from bloated websites, formatting it into the simple gemtext format markdown, then mirroring it back as a simple web/html page, it works together nicely to re-render bloated sites on simple devices using gemini as a formatting medium technology. You don't really need to understand gemini protocol to use newswaffle + portal.mozz.us proxy in your regular web browser