One problem for your argument is that transformer networks are not, and weren't meant to be, calculators. Their raw numerical calculating abilities are shaky when you don't let them use external tools, but they are also entirely emergent. It turns out that language doesn't just describe logic, it encodes it. Nobody expected that.
To see another problem with your argument, find someone with weak reasoning abilities who is willing to be a test subject. Give them a calculator -- hell, give them a copy of Mathematica -- and send them to IMO, and see how that works out for them.
Well being able to extrapolate solutions to "novel" mathematical exercises based on a very large sample of similar tasks in your dataset seems like a reasonable explanation.
Question is how well it would do if it was trained without those samples?
How would you do multiplying 10000 pairs of 100 digit numbers in a limited amount of time? We don't anthropomorphize calculators though...
One problem for your argument is that transformer networks are not, and weren't meant to be, calculators. Their raw numerical calculating abilities are shaky when you don't let them use external tools, but they are also entirely emergent. It turns out that language doesn't just describe logic, it encodes it. Nobody expected that.
To see another problem with your argument, find someone with weak reasoning abilities who is willing to be a test subject. Give them a calculator -- hell, give them a copy of Mathematica -- and send them to IMO, and see how that works out for them.
I hear the LLM was able to parrot fragments of the stuff it was trained to memorize, and did very well
Yeah, that must be it.
Well being able to extrapolate solutions to "novel" mathematical exercises based on a very large sample of similar tasks in your dataset seems like a reasonable explanation.
Question is how well it would do if it was trained without those samples?
4 replies →