lsy

21 ms·

181.
by lsy 4y ago
I'm not saying everything is memorized, I'm guessing there are some tasks you can succeed on by having very similar things, or maybe even some kind of transfer knowledge. But how do you know that a "new" task you are mak
182.
by lsy 4y ago
There is something about this that seems unscientific. What defines an "ability"? It sort of feels like the Guinness-Book type records for "most laps done around the Washington monument with a duck on your head". Who is
183.
by lsy 4y ago
From [1]: "In our study, we employ annotators who are familiar with LLM capabilities. However, this group is not occupationally diverse, potentially leading to biased judgments regarding LLMs’ reliability and effectiveness in performin
184.
by lsy 4y ago
I'd be very hard-pressed to call this "human behavior". Moving a sprite to a region called "bathroom" and then showing a speech bubble with a picture of a toothbrush and a tooth isn't the same as someone in a r
185.
by lsy 4y ago
This isn't a "zero day", it's a trojan horse. The author, given his position, ought to know the difference, so what's the reason for using the wrong term? Also bears pointing out that all of the behavior (along with
186.
by lsy 4y ago
Those tests don't have "construct validity" (i.e. they don't accurately test what they claim to test) with regards to LLMs. Word problems can be used to test human abilities because humans use words to represent interior
187.
by lsy 4y ago
I don't know if I take the Chinese Room scenario seriously, it's a little hard to think about. For me the key is that an LLM (and maybe the Chinese room?) is only trained on text, and therefore its entire "universe" is m
188.
by lsy 4y ago
It's not really a cultural expectation, it takes some time for the photographer to actually hit the shutter so people who know they are about to be photographed freeze themselves to allow the photographer to get a "good" shot
189.
by lsy 4y ago
I don't think it has any relationship, most likely the answers are just generated semi-randomly. Even the one it's "very" confident about is not agreed-upon (Wikipedia says the outcome was "inconclusive"). Whic
190.
by lsy 4y ago
Is there some way of holding the LLM response to a given prompt constant? It sounds like a lot of this relies on the LLM getting the right answer in sequence, so I'm guessing they do something like keep the temperature at 0? Otherwise
191.
by lsy 4y ago
On Nvidia seemingly being in perfect position to ride successive hype cycles of technology: their product category is one of the few types of hardware with significant technical advancement in the last decade, making it only natural that pe
192.
by lsy 4y ago
This is a pretty fluffy paper, especially for an institution like Microsoft Research. It says it's an "early AGI" in the abstract, but elsewhere says it's merely a "step towards AGI". The basis for this is aski
193.
by lsy 4y ago
It's interesting to see people who have been working as software engineers worrying that codegen LLMs will take their job. In the past week the amount of work that consisted of writing well-defined code was pretty much zero, and the ac
194.
by lsy 4y ago
While this is a very cool project that shows a great use of machine learning to answer questions about images in a roughly explainable way, I think people are extrapolating quite a bit as though this is some kind of movement forward from GP
195.
by lsy 4y ago
Even the hands in this tweet's image are not correct. There are a bunch with 4 digits, two thumbs, 6 fingers, fingers splayed in anatomically improbable directions. Not to mention there are probably two hundred and fifty hands in this
196.
by lsy 4y ago
The claims of "embodiment" and "reasoning over images" made in this paper seem to really stretch the definitions of these words. IIUC, what is being fed into the model as input is essentially a text caption generated fro
197.
by lsy 4y ago
While the formality is way overwrought (and ChatGPT is not creating any "simulacra" of characters), I think the overall point is correct that language models are trained on stories and other human writing, and inversion is a ver
198.
by lsy 4y ago
It's not exactly clear from the paper how they've set up the training, but it appears this model has an aspect which uses a secondary model to represent images as vectors, combines them with their text captions, and then uses thos
199.
by lsy 4y ago
I think it's somewhat important to ask why this effect is more observable for girls than for boys, and Haidt does not seem to really reflect on that. Social media in and of itself is really just an accelerant for the particular set o
200.
by lsy 4y ago
The article posits it's because it's GPT-4, with more parameters. It feels like there's a much simpler explanation: the training data probably includes a bunch of Wattpad sci-fi in exactly this vein, and other fan-fiction sto
201.
by lsy 4y ago
All philosophical arguments aside, I become immediately skeptical when commentators compare LLMs to watershed moments in human history. Even those moments were not known except in hindsight, and the jury is just not in to make these kinds o
202.
by lsy 4y ago
This highlights one of the types of muddled thinking around LLMs. These tasks are used to test theory of mind because for people, language is a reliable representation of what type of thoughts are going on in the person's mind. In the
203.
by lsy 4y ago
Yes, this is the essential issue. Any system competent to fully ground and check factual statements in a stream of arbitrary text will be phenomenally more complex than the original LLM, and usually will be able to answer queries directly,
204.
by lsy 4y ago
The TLDR of the article is essentially that any time a ChatGPT conversation strays to the factual (in this case a list of data ethics books), the quality of the data (whether those books exist and are written by the stated authors) is extre
205.
by lsy 4y ago
I think this is unnecessarily credulous about what is really going on with ChatGPT. It is not "applying the concepts of a poem" or checking meter and verse, it is generating text to fit a (admittedly very complicated) function tha
206.
by lsy 4y ago
I don't think this is an "external solution" thing. Some random gender-neutral ideas for becoming more interesting to others and being more interested in them: - Read widely, both fiction and nonfiction. Become interested in
207.
by lsy 4y ago
ChatGPT is just a statistical engine that outputs text consistent with its previous input. So while it understands that requests to count words often end with something like "this passage has <number> words", and therefore p
208.
by lsy 4y ago
Another angle: how much non-reporting is due to mistrust of police, and how much is due to police explicitly telling crime victims not to report or that a report will be useless, because they are trying to politically undermine a progressiv
209.
by lsy 4y ago
The author is clearly complaining from a conservative, free-market viewpoint, but it's difficult to overlook that the conservative, free-market ideology that triumphed in the Reagan-Thatcher era has been a major driver of the complaint
210.
by lsy 4y ago
> That is because they can't meaningfully complain that apple created a non union group to hear concerns and give employees a voice (regardless of whether it did). Such a complaint would be hilarious. This isn't true. Employer

More ›