810

o(1) statistical prime approximation (lemmy.blahaj.zone)

submitted 1 day ago* (last edited 1 day ago) by not_IO@lemmy.blahaj.zone to c/programmer_humor@programming.dev

56 comments fedilink hide all child comments

https://cosocial.ca/@mhoye/116111505546606451

you are viewing a single comment's thread
view the rest of the comments

[-] Flipper@feddit.org 114 points 1 day ago* (last edited 1 day ago)

Has the same vibes as anthropic creating a C compiler which passes 99% of compiler tests.

That last percent is really important. At least that last percent are some really specific edge cases right?

Description:
When compiling the following code with CCC using -std=c23:
bool is_even(int number) {
   return number % 2 == 0;
}
the compiler fails to compile due to bool, true, and false being unrecognized. The same code compiles correctly with GCC and Clang in C23 mode.

Source

Well fuck.

[-] into_highest_invite@lemmygrad.ml 7 points 1 day ago

any llm must have several C compilers in its training data, so it would be a reasonably competent almost-clone of gcc/clang/msvc anyway, right?

is what i would have said if you didn't put that last part

[-] Cethin@lemmy.zip 2 points 12 hours ago* (last edited 12 hours ago)

You're still correct. The thing about LLMs is that they're statistical models that output one of the most likely responses, from the list of most likely responses. It still has some randomness though. You can tune this, but no randomness is shit, and too much randomness sometimes generates stupid garbage. With a large enough output, you're statistically likely, with any randomness, to generate some garbage. A compiler is sufficiently large and complex that it's going to end up creating garbage somewhere, even if it's trained on these compilers specifically.

[-] into_highest_invite@lemmygrad.ml 2 points 11 hours ago

that's a great point but wouldn't the output for a solved problem like "make a working C compiler in rust" work better if the temperature/randomness were zero? or am i fundamentally misunderstanding?

[-] Cethin@lemmy.zip 2 points 10 hours ago

Probably. At that point you might as well just copy/paste the existing compiler though. The temperature is basically the thing that makes it seem intelligent, because it gives different responses each time, so it seems like it's thinking. But yeah, having it just always give the most likely response would probably be better, but also probably wouldn't play well with copyright laws when you have the exact same code as an existing compiler.

[-] racketlauncher831@lemmy.ml 2 points 13 hours ago

This is so fucked up. The AI company has the perfect answer and yet it rolls the die to recreate the same thing by chance. What are they expecting, really?