sgt101

42 ms·

1. 631.▲
   by sgt101 3y ago

   I agree with this - but so long as it's not used as an excuse when a problem arises.
2. 632.▲
   by sgt101 3y ago

   That's all very well, but whistleblowers can find themselves out of a job without getting any result if they aren't in the right place at the right time. What if the OP's dad was of a minority ethnicity in a tenuous position
3. 633.▲
   by sgt101 3y ago

   PI's have a lot to do, she admits that she wasn't focusing enough on the work that was being developed. I think you are making a good point though (she does as well), science as we currently understand it works on the basis of tru
4. 634.▲
   by sgt101 3y ago

   feels like product placement dressed as a blog
5. 635.▲
   by sgt101 3y ago

   never = 4 years? wot?
6. 636.▲
   by sgt101 3y ago

   This is where aggregates are really tricky. I think that they are using means at the average - this is a serious issue. The thing is WHO is this money for? If the long tail of the population are subsisting on $20k a year while the top 1% ar
7. 637.▲
   by sgt101 3y ago

   And I think that they weren't farmers but instead nomadic herders?
8. 638.▲
   by sgt101 3y ago

   This misses the fact that in order to uses stirrups effectively you have to get a few things in place that are not easy to get in place. - There needs to be a lot of you on a lot of horses. This is pretty hard to achieve especially if most
9. 639.▲
   by sgt101 3y ago

   heh... you know I hadn't thought of that.
10. 640.▲
    by sgt101 3y ago

    If I were scratching up a way of doing this I think I would rather have 500 or 1000 smaller devices to use...
11. 641.▲
    by sgt101 3y ago

    If these are your worries you live a blessed life! Remember that your (and my) ancestors lived in fear of too much rain, not enough rain, raiders, pestilence and the local overlord getting upset about something.
12. 642.▲
    by sgt101 3y ago

    I think that the bits hitting the atmosphere might heat the planet up a large amount. This might be bad.
13. 643.▲
    by sgt101 3y ago

    There is an intersting wrinkle here, I haven't seen it mentioned so far. In the UK a patent is invalid if you put an inventor on it who didn't make a significant and definable contribution to the invention. If a patent is AI gener
14. 644.▲
    by sgt101 3y ago

    I agree, this was done some time ago. It's still rather difficult to come up with a good measure because: * the number of qbits that can be used to construct a circuit is not always the same as the number of qbits on the processors * t
15. 645.▲
    by sgt101 3y ago

    Here is a fit a few years ago. https://miro.medium.com/v2/resize:fit:1400/format:webp/1*0aH... I think this chip would intersect with the exponential. So...300 qbits by 2031.
16. 646.▲
    by sgt101 3y ago

    Now hang on, reading Russian propaganda makes me completely certain that variable geometry nozzles were invented by the USSR for the SU-37. So I think it's clear that the information presented by you is hallucinated by a recent large l
17. 647.▲
    by sgt101 3y ago

    problem is that it's an edge case and most execs are only going to be around for 3 years max...
18. 648.▲
    by sgt101 3y ago

    Definitely backups to write once, read many mediums are desirable. Even a disk array behind an API can effectively become write once, read many.
19. 649.▲
    by sgt101 3y ago

    No emergence [2304.15004] Are Emergent Abilities of Large Language Models a Mirage? - arXiv https://arxiv.org/abs/2304.15004 Can't plan https://openreview.net/forum?id=X6dEqXIsEW No compositionali
20. 650.▲
    by sgt101 3y ago

    How is this different from the Tree of Thought dumpster fire? It looks like an awful lot of fishing for a good answer is going on.
21. 651.▲
    by sgt101 3y ago

    Does it have to run them sequentially? I guess the cost will be 12/13bn level but latency may be faster?
22. 652.▲
    by sgt101 3y ago

    As I read it they are doing this with 8 * 7Bn parameter models. So, their model should run pretty well as fast as a 7Bn model and at the cost of a 56bn parameter model. That a lot quicker and cheaper than GPT-4 Also this is kinda a promisso
23. 653.▲
    by sgt101 3y ago

    I work in FS as well, although I am on the digging ditches side rather than research. My take is that there hasn't been much real value created by "new wave" AI so far, and what there is is very very dull stuff like enabling
24. 654.▲
    by sgt101 3y ago

    Also, very sadly, but not uncommonly, the child may not get to enjoy the fruits of the focus on the future, because they may die. In my life: - A girl in my class aged 11 was killed by a bus. - A boy I knew aged 17 killed himself and girlfr
25. 655.▲
    by sgt101 3y ago

    I think it is because the equipment changed dependent on context. In Dieselgate the cars changed their engine management when they got into a test cycle...
26. 656.▲
    by sgt101 3y ago

    I like to think that it was this thing: https://en.wikipedia.org/wiki/Golden_hat Maybe some ancient folk memory!!
27. 657.▲
    by sgt101 3y ago

    I think, reading it, it could be £1000 worth of wine. For example two cases of decent Gevrey Chambertin, I myself could bring myself to accept such a thing.
28. 658.▲
    by sgt101 3y ago

    I think that the problem is that cars are getting simpler and contain less precision engineering. Making a ICE drive chain is very very difficult, there is a long supply chain required. This moat is drying out very very fast.
29. 659.▲
    by sgt101 3y ago

    You will notice that they say in the blog post: "If you’re a researcher, consider pausing reading here, and instead please read our full paper for interesting science beyond just this one headline result. In particular, we do a bunch o
30. 660.▲
    by sgt101 3y ago

    Best get lobbying for that then! Good luck friend.