701 karma · joined February 10, 2022
Unless you think OpenAI is blatantly lying about this:
"A.1 Sourcing. We sourced either the most recent publicly-available official past exams, or practice exams in published third-party 2022-2023 study material which we purchased. We cross-checked these materials against the model’s training data to determine the extent to which the training data was not contaminated with any exam questions, which we also report in this paper."
"As can be seen in tables 9 and 10, contamination overall has very little effect on the reported results."
They also report results on uncontaminated data which shows basically no statistical difference.
Another paper not from Msft showing emergent task capabilities across a variety of LLMs as scale increases.
https://arxiv.org/pdf/2206.07682.pdf
You can hem and haw all you want but the reality is these models have internal representations of the world that can be probed via prompts. They are not stochastic parrots no matter how much you shout in the wind that they are.
Because it's wrong and smart people know that.
That;s a pretty standard Conservative position that I think would be well received here.
Train on More tokens with More GPUs isn't exactly rocket science. I assume the RLHF loop is complex but training the base model itself is pretty well understood.
This is a clear strawman and not representative of what actual longtermists believe.
This is reasoning, what you've described is reasoning.
A tale as old as time: "It is illustrated by the success of chess computers. In the 60s, it was said that computers will never beat people at chess, because that requires intelligence and computers aren't capable of intelligent thought.
When computers regularly started winning matches in the 80s, it was claimed that playing chess wasn't a test of real intelligence because computers could do it."
>Whether that specific question was in the training corpus or not, there are enough words in the sentence to suggest a positive association, including, significantly, "worthwhile." That alone possibly serves to narrow the answers down to A or B, with a preference for B, because it's more likely that "worthwhile" and "letter to x mother" are associated with "enamored" in general English-language text.
Yes, this is called reasoning so in other words the LLM is reasoning about language.
>Look, the whole point of these models is that its not easy or even possible to trace the path any given input takes on its way to output, but we know the principles used in development, so I think it's rather more of a burden to explain how the clearly-explained principles in LLMs result in something other than the obvious.
The inscrutability of the Matrices tells you nothing about it's reasoning ability. Given the correct prompt the LLM will also provide you with a step by step solution to the question it answered. There are also explicit reasoning prompts that these models are able to deal with.
I think it's pretty simple, if these questions are not explicitly in the training data it cannot have answered them correctly at such a high success rate with anything other than reasoning. You haven't given any alternative answer to how it does this either.