fnands

19 ms·

1. 241.▲
   by fnands 3y ago

   I'm a little surprised about the raw data. Do the customers just review the footage manually? What is the limiting factor on your range? The video says a 20 hour mission (how many km is that roughly?). I would have suspected for a blim
2. 242.▲
   by fnands 3y ago

   I work for a company with a very similar customer base (LiveEO), but we focus on satellite data. Your blimps look really cool and I like the storage/buoyancy idea. LiveEO actually started with drones and satellites, but quickly switche
3. 243.▲
   by fnands 3y ago

   Not exactly in the software, but we recently changed our OKR announcement/review process to communicate (primarily) via memes. It used to be two quarterly three hour meetings where everyone used to space out anyway. Now two 1 hour meet
4. 244.▲
   by fnands 3y ago

   "When We Cease to Understand the World" by Benjamín Labatut A work of historical fiction focusing on (real) scientists who grapple with the consequences of their discoveries. Actually had a very similar vibe to Oppenheimer for me.
5. 245.▲
   by fnands 3y ago

   The cynical part of thinks that that might be a feature, not a bug.
6. 246.▲
   by fnands 3y ago

   My understanding is that they use it to filter candidates a bit. FAANG positions get loads of applicants, so they have to find efficient ways of thinning the herd a bit. The correlation between people who are willing to grind leetcode and w
7. 247.▲
   by fnands 3y ago

   Relevant OldCoderGuy video: https://www.youtube.com/watch?v=hSzlE2pM-rY
8. 248.▲
   by fnands 3y ago

   It's both flippant and surprisingly deep, which I think is a nice combo. It's also the opening quote in my PhD thesis.
9. 249.▲
   by fnands 3y ago

   Not really seen as an AI play. The bull market has been mostly driven by FAANG + Nvidia. I'm mostly just playing devils advocate here, and the point I'm trying to make is share price alone isn't everything. Planet is growing
10. 250.▲
    by fnands 3y ago

    True, but what I meant was that the labs that couldn't replicate even diamagnetism clearly haven't created the same material, as the original LK-99 was shown to be diamagnetic. If someone can show that it is diamagnetic and not su
11. 251.▲
    by fnands 3y ago

    Hard to say. The samples that didn't achieve diamagnetism don't tell us much except that creating LK-99 is tricky, which is not a big deal. If someone can show both diamagnetism and prove that it is not a superconductor then that
12. 252.▲
    by fnands 3y ago

    And it took Amazon nearly a decade to get back to their internet bubble highs. PL got SPACed onto the stock market during the SPAC bubble. That being said, while I do believe that Planet has one of the best business models in the industry,
13. 253.▲
    by fnands 3y ago

    Well, I don't do particle physics as my day-job anymore, so basically that.
14. 254.▲
    by fnands 3y ago

    > material at liquid nitrogen temperatures Even worse sometimes. For the LHC it's liquid helium temperatures!
15. 255.▲
    by fnands 3y ago

    There is a reason why room temperature (ambient pressure) superconductors have been one of the holy grails of physics for such a long time: the implications are profound. I used to be a particle physicist, and some of the more complex syste
16. 256.▲
    by fnands 3y ago

    I work for a customer of Planet and can confirm, we pay Planet a lot of money each year.
17. 257.▲
    by fnands 3y ago

    I'm an ML engineer working for a geospatial company and I can assure you, we are looking into this. > Satellite data is extremely idiosyncratic. It's coarse (~10m at best) 10m is the best free imagery, in the commercial domain
18. 258.▲
    by fnands 3y ago

    Yeah, I would class it as encouraging, but far from definitive. But at least it shows that there might be some theoretical basis to it. It might be useful (as jboggan mentioned in another thread) in explaining why reproduction could be diff
19. 259.▲
    by fnands 3y ago

    >> That’s what we’re experiencing together, in real time. Take a moment to put it all into perspective. > It’s OK to day-drink. I am 10/10. No notes.
20. 260.▲
    by fnands 3y ago

    App based on this post to help you decide what to buy: https://nanx.me/gpu/
21. 261.▲
    by fnands 3y ago

    Any tips for getting one off ebay without getting screwed? I want to pull the trigger, but a bit scared.
22. 262.▲
    by fnands 3y ago

    Probably just an accident. Tim Dettmers has been updating this post for years and it's a super valuable resource.
23. 263.▲
    by fnands 3y ago

    And, how did things turn out? How long did it all last?
24. 264.▲
    by fnands 3y ago

    Docker is good, but I also haven't had any issues with conda. At least for most ML projects.
25. 265.▲
    by fnands 3y ago

    Just a note: your blog seems to be stuck in 2022. Date of post is 17 July 2022
26. 266.▲
    by fnands 3y ago

    There is an open source version available, with weights that were leaked, but licensed as "for academic purposes only". This seems they will release the weights under some license that allows commercial usage. How they monetise it
27. 267.▲
    by fnands 3y ago

    My absolute favourite "ugly" website.
28. 268.▲
    by fnands 3y ago

    They mean 25 x RTX 4090 GPUs. 4090 is a model number
29. 269.▲
    by fnands 3y ago

    Nice. He's obviously a talented engineer who's struck a nerve with the whisper.cpp/llama.cpp projects, so hope he has success with whatever he plans to do.
30. 270.▲
    by fnands 3y ago

    Ah SoftBank, please never change.