pona-a

27 ms·

1. 391.▲
   by pona-a 3y ago

   And less reflections/movement.
2. 392.▲
   by pona-a 3y ago

   Hmm... Would a Discord custom client be considered libre enough? They officially see them as a violation of ToS but do not interfere unless they use too much bandwidth. Or would private/semi-private non-encrypted communications place i
3. 393.▲
   by pona-a 3y ago

   Cloudflare uses it for some of their challenges.
4. 394.▲
   by pona-a 3y ago

   I recall there was a project integrating utility meters into Home Assistant, where a tiny ESP board with on-device ML used a camera to read the analog gauge, basically the most practical solution given you can't replace or modify your
5. 395.▲
   by pona-a 3y ago

   Can it be remedied at inference time by decaying the collapsed tokens' probability or is the world model itself defective?
6. 396.▲
   by pona-a 3y ago

   It's still a good heuristic. If you say have a search engine facing a flood of AI generated mush, would it be too disrespectful to a couple tabloids that use genuine third world labor to downrank 0.01% of them for being probably AI gen
7. 397.▲
   by pona-a 3y ago

   I guess it's still rather interesting research, certainly within the "hacker spirit". It's just sad most have to resort to clickbait to get any any funding for such pursuits...
8. 398.▲
   by pona-a 3y ago

   As someone who occasionally dabbles in Mathematica for creative coding, using WLJS finally made me enjoy writing Wolfram code. Their free "cloud" is so slow I'd find writing BASIC on a C64 more enjoyable while their desktop o
9. 399.▲
   by pona-a 3y ago

   As far as I understand, it uses general STT (which tries to transcribe everything, unlike say Picovoice which limits interpretation to only a few commands) + intent recognition. It probably can't interpret an utterance "stop"
10. 400.▲
    by pona-a 3y ago

    1. This would apply to any system with a rudimentary world model, including probably modern Google search or Wolfram Alpha. By this logic, any sufficiently advanced search engine, computational chemistry system, or perhaps even an NLP calcu
11. 401.▲
    by pona-a 3y ago

    Or particularly good scientists, unless blindly (over)fitting models is what constitutes science.
12. 402.▲
    by pona-a 3y ago

    I don't think it should be the responsibility of your device to make itself as close to useless as possible to deter theft. Besides, most thieves rarely do their due diligence on checking what phone maker or model it is, they just swip
13. 403.▲
    by pona-a 3y ago

    Wouldn't Google Colab then we closer to both "time shared" and "supercomputer"? Jupiter notebooks are certainly interactive and Colab's purpose is to allow often low-powered clients to offload ML training/
14. 404.▲
    by pona-a 3y ago

    Having disabled Google's personalized ads, I did see many "get rich quick (now with AI" schemes that promoted the production of this kind of spam. I think the quickest and simplest way to at least lower the acceleration on th
15. 405.▲
    by pona-a 3y ago

    I think it's rather crude and inefficient to generate several translations with a prompt. Can't we just sample different "branches" of successive tokens with some mechanism for ensuring diversity like [1] did? This would
16. 406.▲
    by pona-a 3y ago

    Sometimes I catch myself daydreaming about a world where we just threw all existing software/hardware into a volcano and started from a clean slate. Imagine RISC-V based PCs/laptops with properly replaceable/repairable parts
17. 407.▲
    by pona-a 3y ago

    Google Authenticator is just a TOTP client. You can just copy the token to any app. I'd personally recommend KeePassXC.
18. 408.▲
    by pona-a 3y ago

    https://www.tensorflow.org/lite/android TF Lite has first-class Android support with hardware acceleration if I'm not mistaken.
19. 409.▲
    by pona-a 3y ago

    Note that the source code seems not to be the web UI itself, but rather a collection of samples/helpers to use exported models.
20. 410.▲
    by pona-a 3y ago

    Maybe you can hack some context management and hidden state with custom functions? The scratch buffer can literally be an endpoint that returns OK and has a field like "plain: list[str]", assuming it keeps function call data in me
21. 411.▲
    by pona-a 3y ago

    I find this sort of metric rather pointless... For one, how do we define Linux desktop? ChromeOS is a distribution of Linux, unlike say Android even advertising official support for a Debian container to run Linux desktop software, and whil
22. 412.▲
    by pona-a 3y ago

    I wonder if there are any ThinkPads with rear cameras. Imagine some kind of low-bandwidth network done entirely with near imperceptible blinking of the dot, or perhaps a kind of P2P botnet for portable airdropped systems...
23. 413.▲
    by pona-a 3y ago

    VisionScript looks so neat! I always wanted some wrapper around OpenCV as high level as Mathemathica. I see you've even borrowed their syntax. Though wouldn't it have been nicer as just a library for any functional language for mo
24. 414.▲
    by pona-a 3y ago

    That sounds interesting, but still fails to capture long-range dependencies. If I understand correctly, what you're proposing is to replace co-occurrence frequency with word2vec cosine similarity. I suppose it may help improve overall