• 13 Posts
  • 55 Comments
Joined 1 year ago
cake
Cake day: June 15th, 2023

help-circle
  • Not really. The money is going to American workers producing military hardware. Most of that is the US upgrading to new equipment and giving away the old for a low price.

    This sounds like right wing fox news style nonsense IMO.

    When an authoritarian dictator invades a foreign country, there is one giant overwhelming lesson about what happens when others act passively or appease the criminal. This simple fact should slap anyone a hundred miles down the road for suggesting appeasement is a sound policy. This only ends with the end if Putin, much like the Nazis only ended with the end of Hitler. Anything less will result in WW3 as a guaranteed outcome. It will be more one sided. The entire Russian GDP is less than just the state of Texas. But it will be nuclear and era ending, likely with the majority of surviving humanity in the southern hemisphere.

    If anything, we should be giving Ukraine absolutely anything they want right now. Let them win before we have to go over there and start dying to stop Putin somewhere halfway across Europe. Putin has proven he will never stop trying to conquer. He is the new Hitler. He keeps power by force and what he has called ‘convenient idiots’ that follow whatever misinformation they are given. It is a propaganda misdirection scheme to talk about things out of context like saying military spending in Ukraine is somehow on the US tax payer. This is completely wrong and intended to provoke people that can’t look at an issue and think for longer than the headline.



  • Endless Sky. The save game is a text file. Save a file on the mobile app (F-Droid), and on the PC (Flatpak), and note the last line. This is the line you must swap to transfer the save file. It is the first game I have played on both practically. The game mechanics are different between the two and you need to alter your strategy accordingly. On mobile, I travel with a ship setup for boarding pirate vessels and never target enemies directly; all of my guns are automatic turrets. I just use a fast ship and travel with a large group of fighters. It is more of a grind on mobile, but it can be used to build up resources and reserves. The game is much bigger than it first appears to be. You need to either check out a guide or explore very deep into the obscure pockets of the map.




  • I won’t touch the proprietary junk. Big tech “free” usually means street corner data whore. I have a dozen FOSS models running offline on my computer though. I also have text to image, text to speech, am working on speech to text, and probably my ironman suit after that.

    These things can’t be trusted though. It is just a next word statistical prediction system combined with a categorization system. There are ways to make an LLM trustworthy, but it involves offline databases and prompting for direct citations, these are different from Chat prompt structures.


  • TheOtherJake@beehaw.orgtoChat@beehaw.orgHow's your week going?
    link
    fedilink
    English
    arrow-up
    4
    ·
    edit-2
    1 year ago

    I just got a Llama 2 70B LLM (offline chat AI) working on my laptop. That is a much larger (smarter) system than I thought was possible on a laptop. It takes every bit of 64GB of RAM, and it is about as fast as AOL instant messanger on bad 56k dialup, but it works.

    I think I also fixed my problem that stopped me from using text to speech AI. Now I just need to figure out speech to text, get a few billion dollars, and make an iron man suit.




  • I loved Dread and Prime 2. I tried playing Super Metroid on switch, but the controls are just too poor to pull off the advanced combination moves with the slow low quality emulation. I’m disappointed that there are not a dozen Metroid titles on the switch. Everything in the Prime series should be ported.

    I’m mostly referring to the long hiatus(es) before Dread, and all of the nonsense from developers other than Retro Studio. I understand they were probably in a funky position when it came to writing and coding for a new 3D engine after all of the Prime series had played out the life of the prior engine. IMO, the entire SDK for Nintendo hardware should account for key franchise titles like Metroid. These games should have story boards and plans from first light of new hardware. The plans should always include classic titles too. My biggest complaint about Nintendo is the low quality of most titles on the platform. They are too focused on recruiting developers instead of quality games. Sure there are some great games like BotW, TotK, and Dread, but I’m not going sifting through all the junk in their store to try and find anything else worth playing. I got a couple of titles that a lot of people recommended, and hated them with no recourse and they cost as much as good games. I would have paid for and played all of the Prime series if it had been ported, but Nintendo totally fails at maintaining their legacy titles effectively. It is this lack of availability now, and the stupid fumble of letting extra developers with their own forked vision into the franchise that I am calling a fumbled opportunity.


  • Yeah but MG is WAY older @ 1987 vs GoW in 2005 and ES in 1994.

    Metal Gear Solid was one of the best games on the original PlayStation. I haven’t been into consoles since the PS2. Metal Gear Solid was so good compared to anything else at the time, the idea it is only at 60M now, seems like a major fumble and lack of management. I guess it is like Metroid for being underdeveloped or given to idiots “with a new vision” like in the case of Metroid.






  • Oobabooga is the main GUI used to interact with models.

    https://github.com/oobabooga/text-generation-webui

    FYI, you need to find checkpoint models. In the available chat models space, naming can be ambiguous for a few reasons I’m not going to ramble about here. The main source of models is Hugging Face. Start with this model (or get the censored version):

    https://huggingface.co/TheBloke/llama2_7b_chat_uncensored-GGML

    First, let’s break down the title.

    • This is a model based in Meta’s Llama2.
    • This is not “FOSS” in the GPL/MIT type of context. This model has a license that is quite broad in scope with the key point stipulating it can not be used commercially for apps that have more than 700 million users.
    • Next, it was quantized by a popular user going by “The Bloke.” I have no idea who this is IRL but I imagine this is a pseudonym or corporate alias given how much content is uploaded by this account on HF.
    • This model is based on a 7 Billion parameter dataset, and is fine tuned for chat applications.
    • This is uncensored meaning it will respond to most inputs as best it can. It can get NSFW, or talk about almost anything. In practice there are still some minor biases that are likely just over arching morality inherent to the datasets used, or it might be coded somewhere obscure.
    • Last part of the title is that this is a GGML model. This means it can run on CPU or GPU or a split between the two.

    As for options on the landing page or “model card”

    • you need to get one of the older style models that have “q(numb)” as the quantization type. Do not get the ones that say “qK” as these won’t work with the llama.cpp file you will get with Oobabooga.
    • look at the guide at the bottom of the model card where it tells you how much ram you need for each quantization type. If you have a Nvidia GPU with the CUDA API, enabling GPU layers makes the model run faster, and with quite a bit less system memory from what is stated on the model card.

    The 7B models are about like having a conversation with your average teenager. Asking technical questions yielded around 50% accuracy in my experience. A 13B model got around 80% accuracy. The 30B WizardLM is around 90-95%. I’m still working on trying to get a 70B running on my computer. A lot of the larger models require compiling tools from source. They won’t work directly with Oobabooga.


  • It’s no freaking mystery anywhere. Kids are too damn expensive because just living is too damn expensive. The real fix is massive land reform that absolutely murders the real estate bubble with a nuclear bomb. Regulate the availability of funds directly to the minimum wage. You work, you live a decent life with a good balance. Build dense housing with tight local communities and perfect transportation so we’re always in contact with people in our communities. Babies will be popping up like weeds.

    An article about one of the poorest European countries is not really relevant. They don’t have the same zoning stagnation nonsense that makes housing unaffordable. The stupid incentives that exploded home loan amounts combined with 100 years without zoning reforms are the problem.


  • Have you seen the great gatspy with Wizard too? That’s what always comes up when mine goes too far. I’m working on compiling llama.cpp from source today. I think that’s all I need to be able to use some of the other models like Llama2-70B derivatives.

    The code for llama.cpp is only an 850 line python file (not exactly sure how python=CPP yet but YOLO I guess, I just started reading the code from a phone last night). This file is where all of the prompt magic happens. I think all of the easy checkpoint model stuff that works in Oobabooga uses python-llama-cpp from pip. That hasn’t had any github repo updates in 3 months, so it doesn’t work with a lot of newer and larger models. I’m not super proficient with Python. It is one of the things I had hoped to use AI to help me learn better, but I can read and usually modify someone else’s code to some extent. It looks like a lot of the functionality (likely) built into the more complex chat systems like Tavern AI are just mixing the chat, notebook, and instruct prompt techniques into one ‘context injection’ (-if that term makes any sense).

    The most information I have seen someone work with independently offline was using langchain with a 300 page book. So I know at least that much is possible. I have also come across a few examples of people using langchain with up to 3 PDF files at the same time. There is also the MPT model with up to 32k context tokens but it looks like it needs server machine ram in the hundreds of GB to function.

    I’m having trouble with distrobox/conda/nvidia on Fedora Workstation. I think I may start over with Nix soon, or I am going to need to look into proxmox, virtualization or go back to an immutable base to ensure I can fall back effectively. I simply can’t track down where some dependencies are getting stashed and I only have 6 distrobox containers so far. I’m only barely knowledgeable enough in Linux to manage something like this well enough for it to function. - suggestions welcome




  • WizardLM 30B at 4 bits with the GGML version on Oobabooga runs almost as fast as Llama2 7B on just the GPU. I set it up with 10 threads on the CPU and ~20 layers on the GPU. That leaves plenty of room for a 4096 context with a batch size of 2048. I can even run a 2GB Stable Diffusion model at the same time with my 3080’s 16GBV.

    Have you tried any of the larger models? I just ordered 64GB of ram. I also got kobold mostly working. I hope to use it to try Falcon 40. I really want to try a 70B model at 2-4 bit and see how its accuracy is.