dnautics

27 ms·

1. 541.▲
   by dnautics 8mo ago

   I'm pretty sure the point of aggressively evolving now is to have to basically not evolve it at some point in the future?
2. 542.▲
   by dnautics 8mo ago

   gleam doesn't run a bytecode interpreter in wasm! i think it compiles straight to rust? which is why there are some leaky as hell abstractions.
3. 543.▲
   by dnautics 8mo ago

   without claiming this is actually happening, it is certainly possible to synthetically create a token that ablates the values retrieved from queries in a certain relative time range (via transformations induced by e.g. RoPE encoding)
4. 544.▲
   by dnautics 8mo ago

   it kinda is? the us and Europe both have had year-on-year raw and trade-corrected negative ghg production, the us has since the obama administration, inclusive of trump I. there are a lot of surprises, for example texas being the #1 state
5. 545.▲
   by dnautics 8mo ago

   you could have a lighter runtime by implementing a bytecode interpreter in wasm.
6. 546.▲
   by dnautics 8mo ago

   a tombstone "token "doesnt have to be an actual token, nor does it have to be explicitly carved out into the tokenizer. it can be learned. unless you have looked into the activations of a SOTA llm you cant categorically say that
7. 547.▲
   by dnautics 8mo ago

   that's not how llms work. study the transformer architecture. every token is conditioned not just on the previous token, but each layer's activation generates a query over the kv cache of the previous activations, which means th
8. 548.▲
   by dnautics 9mo ago

   "LLMs cannot backtrack". This is exactly wrong. LLMs always see everything in the past. In this sense they are more efficient than turing machines, because (assuming sufficiently large context length) every token sees ALL pre
9. 549.▲
   by dnautics 9mo ago

   isn't intellectual reasoning just pattern recognition + a forward causal token generation mechanism?
10. 550.▲
    by dnautics 9mo ago

    I'm one of the weird ones that really wants an open source Civ:CTP. Especially if you can still edit the .ini file to have 255 civilizations on the map.
11. 551.▲
    by dnautics 9mo ago

    john mackey, Definitely a bit ruthless but pretty good at maintaining integrity, only a little bit of an internet troll.
12. 552.▲
    by dnautics 9mo ago

    would be hard to miss gigantic kv cache matrix multiplications
13. 553.▲
    by dnautics 9mo ago

    i don't think its schizophrenia? i mean working in tech you haven't run into that CTO or vp eng who snowjobs the c-suite with a word salad of hot button technical terms that don't quite add up? hell ive even interviewed devel
14. 554.▲
    by dnautics 9mo ago

    I'm thinking: do a translate-c and then statically catch errors using my zig-clr tool.
15. 555.▲
    by dnautics 9mo ago

    you mean like the german environmentalists who singlehandedly kicked up german atmospheric mercury emissions?
16. 556.▲
    by dnautics 9mo ago

    yeah I just thought there are "compiler shenanigans" involved with printf! zig's va arg being broken is sad, I am so zig-pilled, I wish we could just call extern "C" functions with a tuple in place of va arg =D
17. 557.▲
    by dnautics 9mo ago

    > they are currently quite terrible at Zig hard disagree (example elsewhere)
18. 558.▲
    by dnautics 9mo ago

    yep. It's not in the constitution, but in all practicality mcculloch vs maryland (which i would love to see repealed) disagrees
19. 559.▲
    by dnautics 9mo ago

    i find claude does quite well with zig. this project is like > 95% claude, and it's an incredibly complicated codebase [0] (which is why i am not doing it by hand): https://github.com/ityonemo/clr [0] genera
20. 560.▲
    by dnautics 9mo ago

    do you suspect it will be possible to implement printf?? while we're talking about printf, can i incept in you the idea of making an io.printf function that does print-then-flush?
21. 561.▲
    by dnautics 9mo ago

    Zig's being compiled in "releasesafe" so lots of bounds checking going on.
22. 562.▲
    by dnautics 9mo ago

    https://m.youtube.com/watch?v=YZa5GqrZeao
23. 563.▲
    by dnautics 9mo ago

    is this like the "it takes 50 hours to set up a project management tool to work the way you want "? what happens if you onboard a superstar that works with django some other way?
24. 564.▲
    by dnautics 9mo ago

    yeah some frameworks call these "lenses". There's even crazy people who write lenses on top of elixir schemas because they dont realize you can just have multiple schemas. maybe more concretely: if you have a table with a ka
25. 565.▲
    by dnautics 9mo ago

    if you want "less magic than rails" check out ecto, i would say it has less magic than django
26. 566.▲
    by dnautics 9mo ago

    sounds inconvenient and error-prone
27. 567.▲
    by dnautics 9mo ago

    well in elixir you can have two schemas for the same table, which could represent different views, for example, an admin view and a user view. this is not (necessarily) for security but it reduces the number of columns fetched in the query
28. 568.▲
    by dnautics 9mo ago

    oh the automatic migrations scare the bejesus out of me. i really prefer writing out schemas and migrations like in elixir/ecto. plus i like the option of having two different schemas for the same table (even if i never use it)
29. 569.▲
    by dnautics 9mo ago

    so in elixir you can do this async alongside your unit tests.
30. 570.▲
    by dnautics 9mo ago

    > And it confuses Claude. I've never had this problem.