csjh

9 ms·

1. 61.▲
   by csjh 3y ago

   The subject of the parent comment was about AOT WASM compilers - definitely languages using WASM GC right now, and (hopefully) more to come
2. 62.▲
   by csjh 3y ago

   In Javascript yes, in Wasm no: https://webassembly.org/roadmap/
3. 63.▲
   by csjh 3y ago

   Support is still mediocre (mainly because of safari)
4. 64.▲
   by csjh 3y ago

   I'd imagine the majority won't support GC, at least not for some time
5. 65.▲
   by csjh 3y ago

   Neural network
6. 66.▲
   by csjh 3y ago

   Since static lifetimes last for the entirety of the program, it can sub in for any other lifetime (as it is guaranteed to exist until the end of the other lifetime)
7. 67.▲
   by csjh 3y ago

   hypixel is an impressive minecraft server, not mod
8. 68.▲
   by csjh 3y ago

   I disagree, IMO for any model that isn't open source model size is just an implementation detail. If someone released a 10 trillion parameter model that's better than GPT4 it isn't somehow inferior because it has more paramet
9. 69.▲
   by csjh 3y ago

   Hard to imagine it was so major that it lead to him being fired while still being so quiet that it hasn't hit any news outlets
10. 70.▲
    by csjh 3y ago

    Never heard of `execution`, great tip. Is there some big caveat or just generally unknown?
11. 71.▲
    by csjh 3y ago

    If it couldn't take parameters what point would there be in it?
12. 72.▲
    by csjh 3y ago

    Only 3GB, interesting to see how small SOTA models in other domains are compared to LLMs like Falcon-180B.
13. 73.▲
    by csjh 3y ago

    I think it must be based on usage (some GPTs could be using more credits than others)
14. 74.▲
    by csjh 3y ago

    Claiming "many people" and proceeding to link to 4 posts that use Svelte as an example is a bit silly considering Rich Harris has said, numerous times, that the decision in Svelte's codebase should NOT be taken as advice for
15. 75.▲
    by csjh 3y ago

    But it's an opt-in GC? It's not accidental by any means
16. 76.▲
    by csjh 3y ago

    That would be nice in a world where Mojo is more complete but it is nowhere near that level right now
17. 77.▲
    by csjh 3y ago

    Yeah, builtins seem like a bit of a specialized case. If I'm not mistaken, it doesn't even really give a Wasm-side method to create strings? Bit of a shame.
18. 78.▲
    by csjh 3y ago

    I think string builtins seem to be the direction the wasm committee is moving towards anyways? Unless I'm mistaken
19. 79.▲
    by csjh 3y ago

    Don't they need both physical access to device + fingerprints/face?
20. 80.▲
    by csjh 3y ago

    similar in ottawa
21. 81.▲
    by csjh 3y ago

    It "works", but there is a pretty ridiculous amount of latency.
22. 82.▲
    by csjh 3y ago

    I believe that's at least part of what the GC proposal is moving towards.
23. 83.▲
    by csjh 3y ago

    LLMs are definitely not AGI
24. 84.▲
    by csjh 3y ago

    Python pseudocode: ```py output = prompt("My favourite animal is ") # returns a generator start = time.time() tokens = 0 for token in output: tokens += 1 print(token) print(f"Outputted {tokens / (time.time(
25. 85.▲
    by csjh 3y ago

    You measure how many tokens are output in a period of time and take the average?
26. 86.▲
    by csjh 3y ago

    What makes you think 3.5 is significantly smaller than 70B?
27. 87.▲
    by csjh 3y ago

    Is there a source on that? I've never seen anyone think it's below even 70B
28. 88.▲
    by csjh 3y ago

    ChatGPT wasn't surpassed by competitors in less than 6 months, what model would you say beat it that early?
29. 89.▲
    by csjh 3y ago

    Isn't that just how the view transitions API works?
30. 90.▲
    by csjh 3y ago

    Fortunately we’ll get revolutionary new products from it, such as calendar app (with AI integration), email app (with AI integration), resume builder app (with AI integration), and more!