ivanbakel

12 ms·

91.
by ivanbakel 3y ago
And those people can look for questions where the "right way" is justifiably unusable, or pose those questions themselves (and find out if they really have to avoid it.) Because you're answering every person who ever will a
92.
by ivanbakel 3y ago
>As I said in the video, the hypothesis for the finches is unproven. I simply wanted to show an interesting alternative theory that throws up some interesting questions. The trouble is that the example fails to be compelling. Why should
93.
by ivanbakel 3y ago
Embedding citations through images is an odd choice, but the biggest issue this article has is a complete dearth of discussion about mechanism. How are we meant to understand that agency influences genetics? What's the flow of genetic
94.
by ivanbakel 3y ago
This is covered in depth in the article. Apps do load-balance, but the main issues are: * they can only load-balance in their own datasets, which means that users of different apps can over-compete for the same small amount of road ca
95.
by ivanbakel 3y ago
I would be thrilled to see this ported to GitLab CI. This is an incredibly (for lack of a better word) sexy tool on top of Nix, and would perfectly fit a use case we have for using Nix for caching build artifacts.
96.
by ivanbakel 3y ago
The article isn't really asking for perfection. I think the key point being made is that these are basic mistakes that even a non-artist can easily recognise (the hallucinated objects being a particularly bad example), and which are a
97.
by ivanbakel 4y ago
>No, you're the one who's begging the question. Why can't a "purely statistical" process have an understanding? This is a related, but fundamentally different thing to the point I replied to in your original comm
98.
by ivanbakel 4y ago
Because it's a purely statistical explanation that doesn't require understanding. Put differently, it's possible that GPT doesn't "understand" language itself as a concept, and instead tokens in the same langua
99.
by ivanbakel 4y ago
>I don't see how it can do that unless it really has some kind of understanding. One possibility is that the model itself has learned that tokens are related across languages based on translation examples. If the appended training c
100.
by ivanbakel 4y ago
>To clarify, in my example of the veteran car mechanic assessing someone's proficiency, there's no set-up required, they can verify it 'in the moment'. > >No special or unusual tools or environmental conditions
101.
by ivanbakel 4y ago
I agree - as I said, you can see the results in the actions of the other party. But those are all techniques beyond communication: none of them allow you to verify, in the moment, whether or not the other person has "understood" y
102.
by ivanbakel 4y ago
That presumes that communication is just about people understanding the thing you are trying to say, rather than the transfer of thoughts, ideas, and opinions. Put another way: the only thing that you can verify when communicating that the
103.
by ivanbakel 4y ago
Your comment feels like an oversimplification of the post. The post doesn't contend that LLMs are capable of role-playing - that's basically the foundation that it builds off of. But saying "LLMs are good at roleplaying"
104.
by ivanbakel 4y ago
> More and more software does not necessarily require more and more developers to write it. It has so far. We might have expected the massive gains in developer productivity over the last 70 years to have resulted in the demand for new d
105.
by ivanbakel 4y ago
Countercultures have always been "cool" - the rejection of the dominant culture is part of their cool factor. Some of the most memorable aspects of recent Western cultural history are counter-cultural: hippies, punks, slackers. A
106.
by ivanbakel 4y ago
Source on this? Quick websearch didn't reveal anything.
107.
by ivanbakel 4y ago
>Get started with Haskell or Idris, you quickly hit higher kinded types, monad transformers, free monads, final tagless. It's a different universe. How is any new developer hitting any of these "quickly"? I'm working
108.
by ivanbakel 4y ago
>This is kind of strange. Is it really fentanyl-specific, or does it effect all opioids? This is answered in the article. >“The anti-fentanyl antibodies were specific to fentanyl and a fentanyl derivative and did not cross-react with
109.
by ivanbakel 4y ago
GPT doesn't display understanding of purely abstract systems, so I doubt it's an issue of lacking sensory information. It can't consistently do arithmetic, for example - and I think it would be presumptuous to insist that sen
110.
by ivanbakel 4y ago
>If a proof can not be found isn't that like not halting, since the space of all theorems can never be exhaustively checked? That's correct, and the consequence is that Coq will reject some valid recursive functions that always
111.
by ivanbakel 4y ago
As a programmer in a high-level language, you do not need to have your ear to the wire. Just because whatever network-communicable abstraction you use ends up serialized as a sequence of bytes, doesn't mean you should be thinking about
112.
by ivanbakel 4y ago
The vehicle would fare better, and you would fare much worse. Modern cars wreck to preserve their passengers, rather than vice versa.
113.
by ivanbakel 4y ago
What are you trying to say? Do you think the author only tried to request access to Imagen through this blog post? What does your comment have to do with the above discussion about Google granting access to the community?
114.
by ivanbakel 4y ago
Because it's transparently true from the sheer size and wealth of Google. What makes you at all skeptical of that claim?
115.
by ivanbakel 4y ago
I think you've misunderstood that example in the article. The AI isn't being asked to generate an image from the prompt, it's being asked to match the similar prompts to the different images. Winoground is basically a reading
116.
by ivanbakel 4y ago
Scientific communities don't formally elect a spokesperson. Granting access "to the community" to investigate scientific claims means making the methodology and results available to everybody - and that includes responding to
117.
by ivanbakel 4y ago
In the sense that the Crown leases the Crown Estate to Parliament in exchange for a guaranteed income? Sure. In the sense that that arrangement could be reversed without massive upcry, a constitutional crisis, and the dissolution of the mon
118.
by ivanbakel 4y ago
That's a philosophical view, and I don't think there's any basis for stating it as fact. One of the great debates of mathematical ontology is whether mathematical concepts are discovered or invented.
119.
by ivanbakel 4y ago
To me, you're overthinking the original comment. The particular line >Also, steam should never even see the password, they should only ever see the hash. is, if you interpret it generously, trivially true. There are 101 authenticati
120.
by ivanbakel 4y ago
To quote the guidelines: >Be kind. Don't be snarky. Have curious conversation; don't cross-examine. Please don't fulminate. Please don't sneer, including at the rest of the community.

More ›