Kalanos

35 ms·

1. 421.▲
   by Kalanos 5y ago

   Are you going to present at jupyter community day? https://discourse.jupyter.org/t/jupyter-community-calendar/2...
2. 422.▲
   by Kalanos 5y ago

   `pd.DataFrame.from_records(mylist).groupby(['thing']).sum().to_records()` you can learn a thing or two. people are genuinely here to teach. let's build a new SO because it's owned by china now.
3. 423.▲
   by Kalanos 5y ago

   So is "adversarial" the A in Generative Adversarial Networks (GANs)? Are they just tiptoeing around the term deep fake?
4. 424.▲
   by Kalanos 5y ago

   Genes are functions. Read this book, it's factually deeper than it appears https://www.amazon.com/gp/product/0190234768/ref=ppx_yo_dt_b...
5. 425.▲
   by Kalanos 5y ago

   it is possible to choose any color by: right click > inspect > editing `.color.*{background-color:#______}`
6. 426.▲
   by Kalanos 5y ago

   mostly just accuracy and r squared. i use lightning's torchmetrics.
7. 427.▲
   by Kalanos 5y ago

   Post doesn't talk about the actual libraries, just the ecosystems surrounding them. TF has more layer types, parallelizes better, is easier to assemble w keras, and you don't have to recreate the optimizer when loading from disk.
8. 428.▲
   You probably don't need oh_my_zsh(gist.github.com)

   2 points by Kalanos 5y ago|0 comments
9. 429.▲
   AWS Karpenter – open source, autoscaling of K8s(aws.amazon.com)

   2 points by Kalanos 5y ago|0 comments
10. 430.▲
    by Kalanos 5y ago

    earlier in my career, I would have said that, yes, it is a waste of time. on one hand, estimating essentially requires architecting, and that's never a waste of time. on the other hand, you could argue that even w a good plan you won&#
11. 431.▲
    by Kalanos 5y ago

    It would be nice if there was a way to quantitatively tie objectives to competitive positioning and monetization
12. 432.▲
    by Kalanos 5y ago

    Tags are the lazy man's schema
13. 433.▲
    by Kalanos 5y ago

    Thanks for sharing your experience
14. 434.▲
    by Kalanos 5y ago

    Getting your custom stack tuned feels great, but maintaining that entire stack across updates is daunting. You can't join meetings because your headphone/ mic/ video drivers aren't working? yikes. i trust apple to handle
15. 435.▲
    by Kalanos 5y ago

    Isn't that why function arguments exist? `if act=='relu'`
16. 436.▲
    by Kalanos 5y ago

    What does FX enable? What are the use cases?
17. 437.▲
    by Kalanos 5y ago

    Hmm. M1 Max = 1783 Single-Core Score 12693 Multi-Core Score Intel Core i9-9900X 3500 MHz (10 cores) October 20th, 2021 = Single-Core Score 1078 Multi-Core Score 11305
18. 438.▲
    by Kalanos 5y ago

    Additionally, that AWS service will work with cloudtrail, cloudwatch, IAM, networking, and will get integrated into default APIs. Checkmate.
19. 439.▲
    by Kalanos 5y ago

    Amazon prides itself on the "race to zero cost" as a way to beat competitors. AWS will release a service with feature-parity at the same price and customers will default to that. so cloudflare is learning to play checkers poorly.
20. 440.▲
    by Kalanos 5y ago

    It's their platform. Apple pre-installs its own apps. Every convenience and grocery store does the same thing with their store-brand products.
21. 441.▲
    by Kalanos 5y ago

    i read that they offer C as well. wonder what those times look like
22. 442.▲
    by Kalanos 5y ago

    Confluence search is great! I could always find what I needed. In fact it's my favorite feature about Confluence. I'd say it's my favorite search outside of Google.
23. 443.▲
    by Kalanos 5y ago

    what edge and node file formats will it support? there is a big gap for graphviz excited for algorithms. you could ultimately do GNNs.
24. 444.▲
    by Kalanos 5y ago

    i just want protocols between workflows of serverless services
25. 445.▲
    by Kalanos 5y ago

    safe reinforcement learning that won't turn the universe into a paperclip factory
26. 446.▲
    by Kalanos 5y ago

    Anything they do other than writing a morally-incentivized governance system for reinforcement models (aka pushing an API for a trillion param NLP model) is just fake.
27. 447.▲
    by Kalanos 5y ago

    Plus you have to get TO the Amtrack and Back
28. 448.▲
    by Kalanos 5y ago

    you get to the obstacles faster, which is what really takes the most thinking time to solve
29. 449.▲
    by Kalanos 5y ago

    for what it's worth, `5` is not a valid json key, but `"5"` is.
30. 450.▲
    by Kalanos 5y ago

    So they will be irrelevant soon. They are just spreading themselves thin as an 'also ran' instead of betting big and reinventing themselves.