vidarh

39 ms·

901.
by vidarh 8mo ago
While I mostly agree it's a bit of a nothing burger with respect to copyright, they did achieve long runs of verbatim text. I think ultimately it's going to end up not mattering much because the extent they had to go to will lea
902.
by vidarh 8mo ago
They acknowledge that in their paper ("Some might qualify our experiments as atypical use, as we deliberately tried to surface memorized books. Adversarial use, like the use of jailbreaks, may matter for copyright infringement analysis
903.
by vidarh 8mo ago
I didn't mistake you for that at all. I didn't give any thought at all to that, in fact. My point was that this "more even than it did for the outside world" seemed to downplay how strongly this view of Google from the &
904.
by vidarh 8mo ago
> "Þ" The way to remember thorn: "Ye Oldee Shoppe". The "Ye" there is "þe" (the), written with "Y" standing in for þ - historically for typographic reasons, and in modern times just repea
905.
by vidarh 8mo ago
> Actually knowing Tolkien well has been helpful because the way he writes is very anglo-saxon, not so even his word choices but just the rhythm or syntax. In English class at school in Norway, I went through a phase after reading LOTR i
906.
by vidarh 8mo ago
Written Dutch is fairly easy for me, on the basis of English + native Norwegian + German from school. Spoken Dutch is largely unintelligible for me, on the other hand, unless they speak very slowly. > In fact, when I first faced the lang
907.
by vidarh 8mo ago
Trying to put your bizarre ethnonationalism aside, as HN is not the place for that: Firstly, you're conflating German and Germanic , while ignoring that the common ancestor here dates to before the High German consonant shift betwe
908.
by vidarh 8mo ago
The continuum of the North Sea languages is much more apparent if you undo the High German consonant shift... (and of course if you minimise the use of the words English have imported from France)
909.
by vidarh 8mo ago
Eddie Izzard speaking old English to a Frisian farmer: https://youtu.be/OeC1yAaWG34?si=lkoQ--uZNN8Ntpqy As a Norwegian who speaks English and school-German, Dutch is fairly easy to read but sounds like you're speaking
910.
by vidarh 8mo ago
I think you underestimate how deep-seated the view of Google as liable to end-of-life any product at any time is for the outside world. I don't adopt any new Google products any more, because I have no reason to trust that it will stay
911.
by vidarh 9mo ago
Nope. The closest I've gotten was rejecting Google recruiters several times. But like everyone else I'm used to Google failing to care about products.
912.
by vidarh 9mo ago
Code is communication. The compiler could handle it, but what is important is that other people can. There are many infamous examples of people using the C preprocessor to write near-Pascal or similar in C. It largely died out because it
913.
by vidarh 9mo ago
Fine tuning is likely a bigger part of it. I've worked on fine tuning projects. There's a massive bias towards fone tuning for Python at several model providers for example, followed by JS.
914.
by vidarh 9mo ago
This feels very Google
915.
by vidarh 9mo ago
No, but many of the names are different, and stylistically they are very different. Depending on whether a translation tries to be fairly literal, or sound as if it is written for the language it is translated to the way the result feels wi
916.
by vidarh 9mo ago
Yeah, so really there are at least 3 translations of Lord of the Rings, to continue that example, and I was being a typical Bokmål user and ignored Nynorsk. The title differences are also a good illustration of how different it can be: Bokm
917.
by vidarh 9mo ago
I don't know that work, but I agree with you in general because of forewords etc. Or even appendices. And translations by different translators. I "grew up with" a specific translation of Lord of the Rings into Norwegian, f
918.
by vidarh 9mo ago
> My point is that you misread the comment you replied to. I'm not the person you replied to. > (By the way, on page 2 of the paper: "we evaluate each LLM only within its corresponding harness.") That has zero relevance
919.
by vidarh 9mo ago
Sure, it adds tokens. I've burnt 200 million tokens today on a single project. The question is what your time is worth for the company, and which tasks costs less to have an agent automate than having you do.
920.
by vidarh 9mo ago
It's down to how much you value your time. If your value your time low enough, it doesn't pay to make AI take over. If you value it high enough, it does.
921.
by vidarh 9mo ago
A $20 Claude subscription lets you scratch the surface. A $20 Claude subscription without training means you have a lot of people spending time figuring out how to use it, and then maybe getting a bit of payback, but earning back that train
922.
by vidarh 9mo ago
Sure, but they will still need to dial it back to a point where they can fund it out of inference at some point . The point is that the fact they can't do that now is irrelevant - it's a game of chicken at the moment, and that mi
923.
by vidarh 9mo ago
I think the problem is that you can't really communicate that feeling without taking a lot of liberties, because it will seem boring and tedious to those not as invested, because it's hard to convey the excitement of a little box
924.
by vidarh 9mo ago
I've only visited Palo Alto, but I recognised way too many aspects of that series from various VC startups. Up to and including the startup I was apart of that had offices in a bedroom in a house in Atherton for a while, and the craz
925.
by vidarh 9mo ago
Do you use an agent harness to have it review code for you before you do? If not, you don't know how to use it efficiently . A large part of using AI efficiently is to significantly lower that review burden by having it do far more of
926.
by vidarh 9mo ago
At some point the players will need to reach profitability. Even if they're subsidising it with other revenue - they'll only be willing to do that as long as it drives rising inference revenue. Once that happens, whomever is left
927.
by vidarh 9mo ago
The point is that whether it does what you tell it in a single iteration is less important then whether it avoids stupid mistakes. Any serious use will put it in a harness.
928.
by vidarh 9mo ago
Given that the price remains the same as Sonnet 4.5, this is the first time I've been tempted to lower my default model choice.
929.
by vidarh 9mo ago
Without measuring quality of output, this seems irrelevant to me. My use of CLAUDE.md is to get Claude to avoid making stupid mistakes that will require subsequent refactoring or cleanup passes. Performance is not a consideration. If anyt
930.
by vidarh 9mo ago
I did work on a supervised fine-tuning project for one of the major providers a while back, and the documentation for the project was exceedingly clear about the extent to which they would not tolerate the model responding as if it was

More ›