mkaic

44 ms·

1. 571.▲
   by mkaic 5y ago

   looking forward to the birth of a new job title in tandem with this product becoming popular: bovine cybersecurity expert! in all seriousness though, i'd be surprised if we don't see at least one comical headline in the future abo
2. 572.▲
   Effective Operational Reviews(principledoperations.com)

   2 points by mkaic 5y ago|0 comments
3. 573.▲
   by mkaic 5y ago

   answering "Starship" to this question is the equivalent of being the kid in 5th grade who made a finger gun every time they played rock-paper-scissors and instantly won.
4. 574.▲
   by mkaic 5y ago

   On the bright side, that means that even cooler second or third gen Frameworks will be available when you do end up upgrading!
5. 575.▲
   EmberGen v0.7.5: Realtime volumetric fluid simulation(jangafx.com)

   4 points by mkaic 5y ago|0 comments
6. 576.▲
   by mkaic 5y ago

   I was especially pleased to see the create_feature_extractor quality-of-life function added in Torchvision. So convenient!
7. 577.▲
   by mkaic 5y ago

   As someone who's been vaguely interested in PyTorch inference optimization but has never had a clear jumping-in point, thank you for this comment! Nice to see a clear two-sentence explanation that actually makes sense to me, makes me r
8. 578.▲
   by mkaic 5y ago

   This looks great! Excited to see more pretrained models made easily accessible through Torchvision, and nn.Module parameterization seems like a really intuitive and neat way of tackling the 'parameterize the parameters' problem ef
9. 579.▲
   by mkaic 5y ago

   This really happened, I was actually the laptop in this story
10. 580.▲
    by mkaic 5y ago

    "Google, YouTube acquisition announcement could come tonight" Crazy to think how long Google has been betting on YT. Imagine telling someone 15 years ago that YouTube was going to turn into the biggest entertainment entity ever .
11. 581.▲
    by mkaic 5y ago

    I'm a huge fan of Eric Rosen, who plays lots of gambits and aggressive games with highly instructive commentary.I'd highly recommend his content whenever you feel yourself starting to get tired of traditionally high-energy YouTube
12. 582.▲
    by mkaic 5y ago

    Found out about this from mjacques' comment [0] on the SpotML post and figured it was worth submitting just on its own! [0] https://news.ycombinator.com/item?id=28739364
13. 583.▲
    Nimbo.sh – Machine Learning on AWS with a single command(nimbo.sh)

    2 points by mkaic 5y ago|1 comments
14. 584.▲
    by mkaic 5y ago

    I like how there's absolutely no easy way to distinguish different lengthened URLs from each other. Who wants to play roulette? :P [0] https://ooooooooooooooooooooooo.ooo/ooooοооoοᴏοoοᴏοoοᴏooοᴏoᴏ... [1] https:/&#x
15. 585.▲
    by mkaic 5y ago

    Such an inspiring project! Having been born well after George decided to make his original work completely unavailable, the janky early-2000s DVD was what I grew up on. I've never even seen the real original Star Wars, so I'm lo
16. 586.▲
    by mkaic 5y ago

    *mætæ
17. 587.▲
    by mkaic 5y ago

    I mean, IMO the presence of a "soul" is unmeasurable and unfalsifiable. Personally, I usually like to tell people I'm 97% sure GPT-3 isn't sentient. That percentage will undoubtedly go down (for me) as better and better
18. 588.▲
    by mkaic 5y ago

    OP was most likely referring to the AWS EC2 instance type T4, which runs on Amazon Graviton processors IIRC.
19. 589.▲
    by mkaic 5y ago

    To anyone who thinks this is cool: Check out Cody's Lab on YouTube, who's spent the past two years working on his own Mars simulation base solo in Nevada and makes excellent videos about it. It's called "Chickenhole Base
20. 590.▲
    by mkaic 5y ago

    Thanks!
21. 591.▲
    by mkaic 5y ago

    You just gotta believe :P And yeah, sometimes the outputs are funky, and almost all of the time there are artifacts. Removing them would mean training a much higher-capacity model (like StyleGAN2) for much longer (like a week) and since thi
22. 592.▲
    by mkaic 5y ago

    I wonder if it's even remotely feasible to use another neural network to estimate the mesh of the best cherry-picked examples, and then actually 3D print an AI cake!
23. 593.▲
    by mkaic 5y ago

    Hi HN! Last year, I posted a thread asking for pointers on where I could go to teach myself Machine Learning, and, well - you guys were helpful enough that I got completely obsessed and am now making strides towards starting a career in ML!
24. 594.▲
    Show HN: This Cake Is A Lie -- GAN-made confections(thiscakeisalie.com)

    14 points by mkaic 5y ago|7 comments
25. 595.▲
    by mkaic 5y ago

    Holy cow, I almost never read entire posts on HN but this one was excellent and thoroughly helpful :) nice work OP!
26. 596.▲
    by mkaic 5y ago

    Location: Sacramento Remote: OK Willing to relocate: Yes Technologies: Python, Linux, machine learning (PyTorch), emphasis on computer vision, familiar with JS and HTML Resume/CV: https://drive.google.com/file/d&#x
27. 597.▲
    by mkaic 5y ago

    Was disappointed to see the lack of ViHart, but other than that, fantastic list!
28. 598.▲
    by mkaic 5y ago

    Wow, I've never heard of this before, I'm 100% trying this next time I'm in Twitter. Thanks for the tip!
29. 599.▲
    by mkaic 5y ago

    My name is Kai, and back in the day when I owned an iPod, the device name was set to KaiPod. Needless to say I was somewhat surprised to see this headline when I opened HN today!
30. 600.▲
    by mkaic 5y ago

    that would be seriously impressive!