winwang

12 ms·

1. 181.▲
   by winwang 2y ago

   Then comes the question: would it also obviously expand their domain of allowable actions to trespass on their users? Since that is a resounding "yes" and they also have the extremely obvious finance incentive to do so...
2. 182.▲
   by winwang 2y ago

   Semi-crazy idea: Add clauses which destroy half the company if they change the deal without a year of advance notice.
3. 183.▲
   by winwang 2y ago

   Counterpoint, it absolutely should be (barring the pedanticness of modding it a bit): https://news.ycombinator.com/item?id=37555651
4. 184.▲
   by winwang 2y ago

   Mildly disagree with your first statement. Well, I mostly agree that it's not particularly helpful for newcomers. As a 0-tuple, it becomes a specific case of a more general concept -- there is some beauty/usefulness in not having
5. 185.▲
   by winwang 2y ago

   And you know this to be true (indubitably), based on what, exactly? (I'm not even sure which country you're referencing either).
6. 186.▲
   by winwang 2y ago

   Unfortunately, I find that real-world configs don't quite conform to easy understandability past tutorial examples, i.e. k8s and yaml.
7. 187.▲
   by winwang 2y ago

   As someone who worked a bit with Bazel actions and uses Scala: I think Bazel cuts down on the area where arbitrary I/O happens. Also, many executables come from pre-defined targets (e.g. java toolchains). One way to deal with I/O
8. 188.▲
   by winwang 2y ago

   I think it's fine if code is difficult for newcomers to understand (of course, to a point). Most programmers are taught in C-like languages and paradigms. Using FP (ML-like) languages is already "difficult to understand". The
9. 189.▲
   by winwang 2y ago

   Depends on if we can interpret the final hidden layer. It's plausible we evolve models to _have_ interpretable (final/reasoning) hidden layers, just that they aren't constrained to the (same representation of) input/outp
10. 190.▲
    by winwang 2y ago

    I believe the "continuous" in Coconut means that the CoT is in the continuous latent space, instead of being on output tokens (see Fig. 1).
11. 191.▲
    by winwang 2y ago

    Here's a rather trivial example of using PTX: https://docs.nvidia.com/cuda/parallel-thread-execution/#spec... For various micro-bench reasons I wanted to use a global clock instead of an SM-local one, and I b
12. 192.▲
    by winwang 2y ago

    Slightly tangential, but I want an LLM which can debug windows.
13. 193.▲
    by winwang 2y ago

    > despite a past conviction What conviction?
14. 194.▲
    by winwang 2y ago

    Yep. I use ~80% and barely see any perf degradation. I use 270W for my 3090 (out of 350W+).
15. 195.▲
    by winwang 2y ago

    Thank you, this is incredibly relevant!
16. 196.▲
    by winwang 2y ago

    Although I agree, I think there is a mild hazard with the label of "model minority". Not saying a reaction like that is really warranted in this case, though.
17. 197.▲
    by winwang 2y ago

    "P(hard-work-pleasures | asian) >= P(hard-work-pleasures | ~asian)" I'm not sure if that's true, but it's presumably what the other person was saying they observed. Though I think in casual conversation, the real
18. 198.▲
    by winwang 2y ago

    ...makes me wonder if we should do that. would it lead to safer driving?
19. 199.▲
    by winwang 2y ago

    glad to see another Scala fan :) keep in mind that "for loops" are really "for comprehensions" and desugae into flatMap/map
20. 200.▲
    by winwang 2y ago

    AI isn't magic. It's still subject to the limits of classical computing.
21. 201.▲
    by winwang 2y ago

    Do you have strong evidence for this?
22. 202.▲
    by winwang 2y ago

    For one, we could just start simulating quantum chemistry, though at that point it's more like "actually running quantum chemistry" rather than simulating.
23. 203.▲
    by winwang 2y ago

    Note there's a caveat: problems in CS can be reduced to other problems in CS. If we solved SAT, well, no one cares about SAT, but traveling salesman obviously reduces to that. (disclaimer: I don't think that's what is going o
24. 204.▲
    by winwang 2y ago

    Haha, classic race condition, but I appreciate your take nonetheless!
25. 205.▲
    by winwang 2y ago

    Edit after skimming arxiv preprint[1]: Yeah, this is pretty huge. They achieved the result with surface codes, which are general ECCs. The repetition code was used to further probe quantum ECC floor. "Just POC" likely doesn't
26. 206.▲
    by winwang 2y ago

    My friends have successfully relied on Medicaid during financial hardships + unemployment. At least in NYC, the Medicaid plans are quite decent. Also, for those who require plans similar to the one previously provided, COBRA (18 months) is
27. 207.▲
    by winwang 2y ago

    PAM3 for data transfer: https://www.synopsys.com/designware-ip/technical-bulletin/pa...
28. 208.▲
    by winwang 2y ago

    I break up my messages, as do many people.
29. 209.▲
    by winwang 2y ago

    I actually thought you meant that this was about applying QoS algos to real life, before I saw the link lol.
30. 210.▲
    by winwang 2y ago

    Using it to run easily-scalable Spark clusters. Previously used it for large distributed builds. It's been pretty great (even if annoying at times).