7 ms·

Because they produce output probabilistically, when multiplication is deterministic. Why is this so hard for everyone?

laterium 11mo ago
Transformers do just fine on many deterministic tasks, and are not necessarily probabilistic. This is not the issue at all. So, it's hard for everyone else because they're not confidently wrong like you are.
trollied 11mo ago
Not true though. Internally they can “shell out” to sub-tasks that know how to do specific things. The specific things don’t have to be models. (I’m specifically talking about commercial hosted ones that have the capability i describe - obviously your run of the mill one downloaded off of the internet cannot do this).
- rrix2 11mo ago
  yes, what your describing is not a transformer but a high-level LLM-based product with tool-calling wired up to it
- KalMann 11mo ago
  That doesn't appear to be the kind of thing this article is describing.
skinner_ 11mo ago
If being probabilistic prevented learning deterministic functions, transformers couldn’t learn addition either. But they can, so that can't be the reason.
- wat10000 11mo ago
  People are probabilistic, and I've been informed that people are able to perform multiplication.
  - ddingus 11mo ago
    Yes, and unlike the LLM they can iterate on a problem. When I multiply, I take it in chunks. Put the LLM into a loop, instruct it to keep track of where it is and have it solve a digit at a time. I bet it does just fine. See my other comment as to why I think that is.
  - krackers 11mo ago
    Are you sure? I bet you if you pull 10 people off the street and ask them to multiply 5 digit by 5 digit numbers by hand, you won't have a 100% success rate.
    - wat10000 11mo ago
      The pertinent fact is that there exist people who can reliably perform 5x5 multiplication, not that every single person on the planet can do it.
    - emp17344 11mo ago
      I bet with a little training, practically anyone could multiply 5 digit numbers reliably.
razodactyl 11mo ago
Bad take. It's not that it's hard for everyone - there's critical pushback because we don't know for certain if LLM technology can or cannot do the task in question. Which is the reason there's a paper being discussed. If we were to take the stance of "ok, that happened so it must be the case" we wouldn't be better off in many cases, we would still be accusing people of being witches most likely. Science is about coming up with a theory and trying to poke holes into it until you can't and in which case, after careful peer-review to ensure you're not just tricking yourself into seeing something which isn't there a consensus is approached in which we can continue to build more truth and knowledge.

Transformers do just fine on many deterministic tasks, and are not necessarily probabilistic. This is not the issue at all. So, it's hard for everyone else because they're not confidently wrong like you are.

trollied 11mo ago

Not true though. Internally they can “shell out” to sub-tasks that know how to do specific things. The specific things don’t have to be models. (I’m specifically talking about commercial hosted ones that have the capability i describe - obviously your run of the mill one downloaded off of the internet cannot do this).

rrix2 11mo ago
yes, what your describing is not a transformer but a high-level LLM-based product with tool-calling wired up to it
KalMann 11mo ago
That doesn't appear to be the kind of thing this article is describing.

skinner_ 11mo ago

If being probabilistic prevented learning deterministic functions, transformers couldn’t learn addition either. But they can, so that can't be the reason.

wat10000 11mo ago
People are probabilistic, and I've been informed that people are able to perform multiplication.
- ddingus 11mo ago
  Yes, and unlike the LLM they can iterate on a problem. When I multiply, I take it in chunks. Put the LLM into a loop, instruct it to keep track of where it is and have it solve a digit at a time. I bet it does just fine. See my other comment as to why I think that is.
- krackers 11mo ago
  Are you sure? I bet you if you pull 10 people off the street and ask them to multiply 5 digit by 5 digit numbers by hand, you won't have a 100% success rate.
  - wat10000 11mo ago
    The pertinent fact is that there exist people who can reliably perform 5x5 multiplication, not that every single person on the planet can do it.
  - emp17344 11mo ago
    I bet with a little training, practically anyone could multiply 5 digit numbers reliably.

razodactyl 11mo ago

Bad take. It's not that it's hard for everyone - there's critical pushback because we don't know for certain if LLM technology can or cannot do the task in question. Which is the reason there's a paper being discussed. If we were to take the stance of "ok, that happened so it must be the case" we wouldn't be better off in many cases, we would still be accusing people of being witches most likely. Science is about coming up with a theory and trying to poke holes into it until you can't and in which case, after careful peer-review to ensure you're not just tricking yourself into seeing something which isn't there a consensus is approached in which we can continue to build more truth and knowledge.