jebarker

27 ms·

1. 601.▲
   by jebarker 2y ago

   > the same way knowing how memory or CPUs work is largely irrelevant for all but a tiny part of the industry. This is not true. A large part of the industry just _thinks_ these things aren't relevant to them and that's why we h
2. 602.▲
   by jebarker 2y ago

   Fog is a great analogy here and indeed much less unnerving. I think what I mean is that I'm unnerved by the fact that our perception only gives us a partial view on what really makes up the universe.
3. 603.▲
   by jebarker 2y ago

   Thanks! This is helpful. It seems like I took an overly-simplistic explanation of the small scale and took it too literally. There's still a nugget of truth to the idea that our perceptions only partially tell us what the universe is r
4. 604.▲
   by jebarker 2y ago

   But we have a common understanding of what "mostly empty space" looks like and what we perceive as solid objects don't look like it.
5. 605.▲
   by jebarker 2y ago

   I think the bucket is bigger than just your body.
6. 606.▲
   by jebarker 2y ago

   I'm not a physicist. An idea from physics that I recently learned and found very unnerving was that the solid objects we see and feel in the world are actually mostly empty space. We just see them as solid because of the way light inte
7. 607.▲
   by jebarker 2y ago

   You're right - if you select it from the drop-down it works that way. If you refresh the browser, as I was, it defaults to the same sample.
8. 608.▲
   by jebarker 2y ago

   Ah, I guess the examples are canned for efficiency since it picks the same token each time. Thanks for clarifying.
9. 609.▲
   by jebarker 2y ago

   I love this visualization, but when it generates tokens with temperature=1 it seems to be picking the second highest logit/probability, not the highest. Is that a mistake or am I missing something?
10. 610.▲
    by jebarker 2y ago

    I came for the programming language discussion, but I want hear more from the author about their contentedness with being a solo developer for life.
11. 611.▲
    by jebarker 2y ago

    I don't understand the motivation for this comment, can you explain?
12. 612.▲
    by jebarker 2y ago

    The idea of discovering a life that fits you rather than trying to predict it up front makes sense to me. In my life I think I've found most success, enjoyment and reward when I've followed my gut as opportunities arise. But I&#x2
13. 613.▲
    by jebarker 2y ago

    It may mean what it says. Alignment may not be seen as as important as building larger and more capable models so may not be receiving the resources or attention he wants. Doesn't have to be as dramatic as military applications or neut
14. 614.▲
    by jebarker 2y ago

    Sadly for me I don't find execution on a clear idea to be the problem most of the time in my work. It's choosing the right idea to execute on that's hard.
15. 615.▲
    by jebarker 2y ago

    > What we need for the "future" of weather is long-tail events which cannot be modelled by curve-fitting to historical data. Yes, it seems like weather forecasting is a simulation problem not a low-shot prediction problem. I as
16. 616.▲
    by jebarker 2y ago

    Is there a different angle on the research you can take that doesn't require training models of that size? When choosing research problems it's important to not only follow what's interesting but also what's feasible giv
17. 617.▲
    by jebarker 2y ago

    I didn't mean to suggest anything. I was just interested in whether they thought that remaining anonymous was in keeping with their ex-colleagues character. Written communication is hard!
18. 618.▲
    by jebarker 2y ago

    If it were them would you be surprised that they wanted to remain anonymous?
19. 619.▲
    by jebarker 2y ago

    My first job out of school was in defense operations research. Not at RAND, so maybe of limited relevance. But I found it less like modern data science and more like GOFAI. The work was mostly coming up with hand engineered models of milita
20. 620.▲
    by jebarker 2y ago

    Quite possibly: https://bike.shimano.com/en-US/product/component/sora-r3000....
21. 621.▲
    by jebarker 2y ago

    Statistical Mechanics is another good example here. As I understand it, it's what allows much of the understanding of the very early universe despite us being unable to model it explicitly or observe it directly.
22. 622.▲
    by jebarker 2y ago

    I think Geoff Hinton made this suggestion first.
23. 623.▲
    by jebarker 2y ago

    I doubt there's much in terms of brainpower, compute or financial resources to differentiate a hedge fund and an AI research lab.
24. 624.▲
    by jebarker 2y ago

    I don't really understand your complaint. All the issues you raise are true, but there's nothing factually incorrect in the article is there?
25. 625.▲
    by jebarker 2y ago

    I visited Hiroshima about ten years ago while on vacation with my wife. The museum presents accounts of the day and artifacts from the bombing very plainly, which by their very nature means it's horrifying. I've always had some &q
26. 626.▲
    by jebarker 2y ago

    In a similar vein, going to Hiroshima and the Peace Memorial Museum there was a life changing experience for me.
27. 627.▲
    by jebarker 2y ago

    This is a wonderful essay. I've always found the term AGI confusing. For example, how general does it need to be to qualify and what specific cognitive capabilities does it need to exhibit? My gut feeling has always been that it's
28. 628.▲
    by jebarker 2y ago

    The creativity displayed in spotting the opportunities is really impressive.
29. 629.▲
    by jebarker 2y ago

    This would make a good koan
30. 630.▲
    by jebarker 2y ago

    I missed all the replies yesterday. My point is that I don't think that learning the correlation between some words and visual concepts qualifies as discovering language. It may be that that's as far as this approach can go so it