naillo

20 ms·

1. 301.▲
   by naillo 4y ago

   Some good critique on this result: https://twitter.com/cHHillee/status/1577713102434361344 TLDR seems like the baseline strassen implementation they used is questionable wrt how really optimal it is in the first p
2. 302.▲
   by naillo 4y ago

   That penalty seems a bit excessive
3. 303.▲
   by naillo 4y ago

   As a visual thinker I actually feel really limited by all these text interfaces. I hear a lot of people with aphantasia in this space though (sama, emad) so for them anything other than this idea is probably unthinkable but personally it re
4. 304.▲
   by naillo 4y ago

   Whatever insights and expertize you've gained up until now can probably be used to gain enough of a competitive advantage in this future industry to be employed. I doubt the people that will spend their time on this professionally will
5. 305.▲
   by naillo 4y ago

   jarvis render a video of nutsome cream spread on a piece of toast 4k HD
6. 306.▲
   by naillo 4y ago

   Probably only 6 months until we get this in stable diffusion format. Things are about to get nuts and awesome.
7. 307.▲
   by naillo 4y ago

   I find writing things down in a notepad (that I barely ever look at again) helps with retention and reduces it looping in my brain as much (relieving it to think about and process new things). Anything more complex than that (e.g. notion) i
8. 308.▲
   by naillo 4y ago

   I feel like I have about as much chance reading disassembled wasm as I would have reading unminified javascript so I don't think it changes much. (You could technically turn the wasm to JS and unminify that too, which I doubt is much h
9. 309.▲
   by naillo 4y ago

   Rare to find creative real time small-weight uses of ML but I love when it's done and this has an impressive and well written explanation with it as well. Great stuff.
10. 310.▲
    by naillo 4y ago

    Tel Aviv University keeps kicking ass, great job.
11. 311.▲
    by naillo 4y ago

    This makes me feel like he's buying and pulling out for some purpose of sabotaging it so something else can take its place.
12. 312.▲
    by naillo 4y ago

    It's funny that the authors are 'anonymous' but they have access to Imagen so obviously it's by Google.
13. 313.▲
    by naillo 4y ago

    I generally find stable diffusion outputs better than dalle so it's surprising you say that. A good prompt makes a big difference though.
14. 314.▲
    by naillo 4y ago

    https://beta.dreamstudio.ai/dream
15. 315.▲
    by naillo 4y ago

    Just as user friendly as dalle but for stable diffusion and more free credits: https://beta.dreamstudio.ai/dream
16. 316.▲
    by naillo 4y ago

    That anime girl really makes it hard to hide that you're doing work unrelated things in the office.
17. 317.▲
    by naillo 4y ago

    Guess it would have been nicer of the donator to send it in small increments instead so plausible deniability (or actual non detection) was more possible.
18. 318.▲
    by naillo 4y ago

    I wonder if it applies to photoshop neural filters, or any kind of ML touchup like an instagram face filter. (Probably not but the line seems awfully thin and arbitrary.)
19. 319.▲
    by naillo 4y ago

    Probably only knowable by actually running it (but probably a bit faster).
20. 320.▲
    by naillo 4y ago

    You're totally right, OpenAI released CLIP in january. But I mean CLIP isn't an image generator, it's just a classifer. If we restrict the question to actual text to image generators (ignoring deep dream or some of the '
21. 321.▲
    by naillo 4y ago

    The weights are not particularly restrictive. You're totally allowed to use them to generate things you sell for instance.
22. 322.▲
    by naillo 4y ago

    The people who first originated the clip guided diffusion approach (rivershavewings around this time last year) are now working for stable diffusion so it's somewhat arguable that dalle wasn't actually first (just first to make a
23. 323.▲
    by naillo 4y ago

    This is already happening basically and twitter aren't banning anyway. (Try mentioning any drug or being 'hacked' or needing an 'essay' written, you'll get a flood of replies in a few minutes.)
24. 324.▲
    by naillo 4y ago

    As someone not in the gaming industry, what does 'visual coding/scripting' mean? Any links or wikipedia articles that describe it? (Or an explanation yourself?) I'm curiuos if it could work well with new developments in
25. 325.▲
    by naillo 4y ago

    I'm not sure the lower end but 80% most popular GPUs have gotten that much stronger in the last 9 years, and that's the spec they likely need it to primarily work on.
26. 326.▲
    by naillo 4y ago

    This is obviously brain-structure dependent. I for one would take 10x more time to learn python if some teacher decided that the best way to do it is to write out python by hand on paper for instance. (But this probably does work well for s
27. 327.▲
    by naillo 4y ago

    I wonder how much of that is usable though if the core engineering team disapears after this and some adobe team is stuck having to sift through it (to understand and way later be able to add to it).
28. 328.▲
    by naillo 4y ago

    It's more creepy when it's human beings you're analyzing than non humans.
29. 329.▲
    Using LLVM to convert PyTorch to WASM(blog.christianperone.com)

    3 points by naillo 4y ago|0 comments
30. 330.▲
    by naillo 4y ago

    Both are really creepy. It becomes more creepy when you go from a recorded video that no one really watches to one systematically analyzed with AI though.