neidu3@sh.itjust.works 4 days ago
I have two practical use cases where I haven’t found a better alternative that LLMs:
- I can paste a huge log file and have it summarize where it went wrong (and often how to fix it)
- It can give me a simple introduction to any topic and provide practical examples.
Bluegrass_Addict@lemmy.ca 4 days ago
you assume it’s actually correct
you assume it’s vorrect
neidu3@sh.itjust.works 4 days ago
Wrong on both accounts. As with any other source of information, its correctness needs to be validated.
prole@lemmy.blahaj.zone 2 days ago
If you’re verifying the output from a big log file anyway, then you wouldn’t need the llm in the first place.
Bluegrass_Addict@lemmy.ca 4 days ago
except you aren’t using it for validation since you didn’t bother to actually read the huge log file… you’re using for answers, not to confirm something you already know.
huge difference…
fuckwit_mcbumcrumble@lemmy.dbzer0.com 4 days ago
They almost always point out what they think the issue is in the log. It doesn’t take much effort to take a snippet of the error, ctrl F, and see that it’s been blowing up 500 times. It’s not hard to verify from there if that was the issue. But if it gives you a fix and that fix worked then presumably it was right.
neidu3@sh.itjust.works 4 days ago
Wrong again. You seem to be the one making the assumptions here, as you keep ascribing methods to me.
But since you absolutely have to know, when the log file is megabytes in size, I don’t read it line by line, as I have better things to do with my time. But an LLM can help narrow down which line to look at more closely and parse away the noise.