eurekin

33 ms·

1. 511.▲
   by eurekin 3y ago

   Sample apps using PyTorch are best examples of an adventure, when ran on Windows. One needs to carefuly consider a WSL2 machine, full vm or Anaconda, because each comes with some caveats. WSL doesn't implement sound/microphone acc
2. 512.▲
   by eurekin 3y ago

   Is that RAM as fast as A100/H100 VRAM? AFAIR, gpus push ~ 1 TB/s ish
3. 513.▲
   by eurekin 3y ago

   Thank you for sharing that!
4. 514.▲
   by eurekin 3y ago

   Yes! I absolutely loved Google desktop search and hoarded much more pdfs and tex files than otherwise. Very excited for the LLM equivalent. PrivateGPT, although more of a proof of concept right now, looks like it could be a nice fit in futu
5. 515.▲
   by eurekin 3y ago

   I wonder, if it's possible to test the hypothesis that smartphone (or whatever internet connected device) allows one to learn more about the world without the filter
6. 516.▲
   by eurekin 3y ago

   Yes I have one that has one fried chip. Once I gave another a Bykski water block with backplate it's been no issue for a year
7. 517.▲
   by eurekin 3y ago

   3090 has a lot of vram chips on the back though
8. 518.▲
   by eurekin 3y ago

   Adobe subscriptions are nice examples of bad management, no long standing significant bug was fixed and they mostly seem to add toys around the fringes (cloud based neural filters for example)
9. 519.▲
   by eurekin 3y ago

   Please consider supporting multi touch, where it makes sense
10. 520.▲
    by eurekin 3y ago

    Can you demo it, please?
11. 521.▲
    by eurekin 3y ago

    Any publically available demos? I've only played around with faux pilot and it seemed that it was mostly limited by my prompting skill. Also, is copilot aware of anything outside of currently edited file?
12. 522.▲
    by eurekin 3y ago

    Like Dayvan Cowboy too?
13. 523.▲
    by eurekin 3y ago

    I'd love to read more about it. Sounds both fun and potentially useful
14. 524.▲
    by eurekin 3y ago

    Even non-quantized large LLMS (70b) have a lot of difficulties remembering facts. Chatgpt, being much larger, hallucinates a ton. It seems that it's not the best use case for them right now. Being a fact base that is
15. 525.▲
    by eurekin 3y ago

    Thanks! If ChatGPT can be used commercially and it successfully passed SOC3 cert, wouldn't you still want to use those non - chatgpt models? Also, you hint at those many ideas, could you elaborate on that a bit? I'll be playing wi
16. 526.▲
    by eurekin 3y ago

    Another noob here. If I had to guess, it's because current models are mostly memory bound. The AI learning gpus (A100, H100 etc.) are not the best TFlop performers, but they have most vram. It seems that researchers found a sweet spot
17. 527.▲
    by eurekin 3y ago

    Not wanting to derail the thread, but could be the best place to ask this: What are you using local LLM's for? So far, I've been only able to come up with: - Aid in coding (which always ends up in chatGPT) - Summarizing short arti
18. 528.▲
    by eurekin 3y ago

    This field is in crazy progress mode now. Not long ago it was rent cuda gpu only. Now this... AMD could easily chip away some part of the market, if they released a > 24 GB vram gpu now.
19. 529.▲
    by eurekin 3y ago

    Yup. My first intuition was "probably some server gpu (a100), but sold separately"
20. 530.▲
    by eurekin 3y ago

    ... or a cable TV
21. 531.▲
    by eurekin 4y ago

    Omg, no pytorch/wsl/conda hiccups... This could save me some sleepless nights
22. 532.▲
    by eurekin 4y ago

    > I created a PR on the repo to load it with int8, so hopefully by tomorrow it'll be runnable by 3090/4090 users. How about 2x3090? Can it be run on multiple gpus?
23. 533.▲
    by eurekin 4y ago

    For all initially surprised, as to how is it possible for an AI model to age, here are example models measured: - Patient wait time dataset - Weather temperature dataset - Financial stock price dataset - Airport departure delay dataset I&#x
24. 534.▲
    by eurekin 4y ago

    Imagine future job market, where a 5 year old will already be super familiar with the T-posing, which currently mostly CG artists now about
25. 535.▲
    by eurekin 4y ago

    Same here! Although some get old quicker than others, I had few moments of laugh. I like Balenciaga ones: https://youtube.com/playlist?list=PL9CIfP547-JFhCFeLeRlA2MX1...
26. 536.▲
    by eurekin 4y ago

    For those interested about the tools used: https://youtu.be/TGD8zKvRxc4 - how to make a "balenciaga" style video.
27. 537.▲
    by eurekin 4y ago

    Well, if it had choosen our planet as a direct destination for (providing it has means to travel at light speed) collision, we'd only have 1560 years to escape. There are a couple of comically improbable conditions that would happen on
28. 538.▲
    by eurekin 4y ago

    Thanks, that indeed worked! This and using conda in wsl2, instead on bare windows
29. 539.▲
    by eurekin 4y ago

    I did try, but got: ``` ValueError: Tokenizer class LLaMATokenizer does not exist or is not currently imported. ```
30. 540.▲
    by eurekin 4y ago

    I wouldn't be surprised, if wine became a "compatibility layer" embedded into a wsl in Windows at some point