Most short running scripts call into the standard library or into other packages. I think the trivial subset is too small to bother with such an optimisation.
Yes, but so is telling if a photo contains a dog or understanding sentiment in a paragraph of text. Complexity isn't quite the issue, I think it is that there is a distinction between the type of reasoning which these models have learnt and that which is necessary for concrete mathematical reasoning.
This is an interesting article and goes along with how I understand how such models interpret input data. I'm not sure I would characterize the results as blurry vision, but maybe an inability to process what they see in a concrete manner.
All the LLMs and multi-modal models I've seen lack concrete reasoning. For instance, ask ChatGPT to perform 2 tasks, to summarize a chunk of text and to count how many words are in this chunk. ChatGPT will do a very good job summarizing the text and an awful job at counting the words. ChatGPT and all the transformer based models I've seen fail at similar concrete/mathematical reasoning tasks. This is the core problem of creating AGI and it generally seems like no one has made any progress towards synthesizing something with both a high and low level of intelligence.
My (unproven and probably incorrect) theory is that under the hood these networks lack information processing loops which make recursive tasks, like solving a math problem, very difficult.
A child will start to speak at around the age of one, but most will be about two before they start to count. And it is even longer (maybe the age of three to four) before they understand cardinality and can reliably follow “simple” instructions like “bring me four blocks”.
And basic arithmetic without counting on their fingers is usually not picked up until they are around six or seven.
I hope you are aware of the fact that LLMs does not have direct access to the stream of words/characters. It is one of the most basic things to know about their implementation.
Yes, but it could learn to associate tokens with word counts as it could with meanings.
Even still, if you ask it for token count it would still fail. My point is that it can’t count, the circuitry required to do so seems absent in these models
If all memory is identity mapped, how does one program cooperate with another absent OS intervention? Can memory corruption happen when one process accidentally writes in the address space of another? What about overflows?
There’s a lot of weird stuff between Vegas and CA! The weirdest one is an abandoned Nuclear bunker which I believe was built by AT&T 30ish years ago off Razor Road. Eddie World is up there too, something tells me not to trust a giant ice cream cone in the middle of nowhere :)
It's one thing to advertise something as a modernized version and another to advertise it as the original work of the authors. Isn't it worth preserving how things were in the past? If not how will we understand where we came from? How will we know how to learn from mistakes if they're just erased?
That's an interesting point. I would agree that the attribution should clearly state "Revised in 2023 by Puffin" or something along these lines.
As for:
> Isn't it worth preserving how things were in the past?
Oh, yes, absolutely, and this is why archivists and librarians have such an important profession; and also why copyright law sucks so much. Do be angry at copyright law, don't be angry at the decision to modernize the texts.
"Why do we lie to children in the places and times we do" is its own fascinating topic, but are we expecting anyone encountering Charlie and the Chocolate Factory for the first time to be reading it through a lens of "how things were in the past" lens?
I'm not sure. Bright children might indeed notice some things and ask questions about them. My childrens' first encounter with Matilda was via me reading it to them, at the ready for any questions they might have had, and also (great nerd that I am) probably trying to provoke questions and answer ones that were never asked.
