cjbprime

27 ms·

271.
by cjbprime 3y ago
(And note that to play well you'd want more like one tile every two-or-less seconds in FFA.)
272.
by cjbprime 3y ago
It's actually just that I'm in the process of moving from Obsidian to vscode in general. (Lack of a mobile version of vscode is a problem though!)
273.
by cjbprime 3y ago
FWIW I need a Kanban board view from a note-taking app, am using extensions which add one to Obsidian and vscode.
274.
by cjbprime 3y ago
If I recall from listening to the ATC conversation back at the time, the ATC acknowledged that they had received this information from Lufthansa in a phone call far ahead of the arrival.
275.
by cjbprime 3y ago
The famous Germanwings murder-suicide had some of this in play -- the pilot's doctor had given him a letter excusing him from work, due to a probable psychotic episode, but the pilot chose not to share the letter with the airline and t
276.
by cjbprime 3y ago
For what it's worth, I'm only an enthusiast but would be surprised to hear either of those are significant issues, unless you're saying that your corrected eyesight is also poor.
277.
by cjbprime 3y ago
Not possible for any crime that is reported on by media, surely.
278.
by cjbprime 3y ago
You don't need to train the model on your data: you can use retrieval augmented generation to add the relevant documents to your prompt at query time.
279.
by cjbprime 3y ago
Exciting, thanks for writing this!
280.
by cjbprime 3y ago
Just mentioning in case it helps anyone out: Linux already has a disk buffer cache. If you have available RAM, it will hold on to pages that have been read from disk until there is enough memory pressure to remove them (and then it will on
281.
by cjbprime 3y ago
Disagree that it's clear. If the text is "oneight", there is a legitimate philosophical question about whether the string contains two numbers. I feel like most people would say no, because the only way to answer yes is by re
282.
by cjbprime 3y ago
I think this is missing the main reason to use Apple Silicon, which is that your dedicated graphics card probably has 24GB or less of RAM, whereas e.g. an M2 Ultra Mac Studio can have 192GB of RAM with a far superior memory bandwidth to any
283.
by cjbprime 3y ago
There's no skepticism in the post, for example making the opposite argument in his "But <my opinion, X> is true, right?" conversational style, and seeing whether ChatGPT tells him just as readily that he "is correc
284.
by cjbprime 3y ago
For what it's worth, I strongly disagree -- the group as a whole (and especially its leadership) is responsible for the policies they decide to institute, and the incentives they allow to exist. For example, in this article's stor
285.
by cjbprime 3y ago
> And I still remember the shock when Project Manager Pete delivered the decision and the follow-on news: ‘These tests will continue. And Chris will continue to lead them as we have paid for his education. He’s the last person on Earth w
286.
by cjbprime 3y ago
> The airplane networks are very carefully segregated. You have a bit in the cabin that’s called the Passenger Information Entertainment Services Domain. That’s completely isolated from what we call the Aircraft Control Domain, or ACD. S
287.
by cjbprime 3y ago
Right, I know that, I'm wondering what the point of this submission is.
288.
by cjbprime 3y ago
Is this a parody? If so, what is the parody?
289.
by cjbprime 3y ago
This is supposed to be an explanation of why LLMs are just autocomplete, but you're describing the ability to make connections between ideas across languages semantically as an explanation for why answering questions is learned cross-l
290.
by cjbprime 3y ago
I don't know whether I have "like-minded views", but I think the risk is there, and it is not fantastical to me because the AGI doesn't need to "do" anything outside of humans assisting it -- no killer robots o
291.
by cjbprime 3y ago
(You're probably going to have to get better at answering objections than merely asserting your contradiction of them.)
292.
by cjbprime 3y ago
It's not wild. "Predict the next word" does not imply a bar on intelligence; a more intelligent prediction that incorporates more detail from the descriptions of the world that were in the training data will be a better predi
293.
by cjbprime 3y ago
He did indeed. (I don't think it is necessarily inconsistent to regret an action you participated in and want the authority that took it to resign in response, though "participated" feels like it's doing a lot of work i
294.
by cjbprime 3y ago
Can you explain why without using the word "embeddings", in a way that convinces us that it's just autocomplete? Or to put it another way: what is being embedded? Is it an abstract concept like question-answering?
295.
by cjbprime 3y ago
> There have been a few stories which sound like he may have had the opportunity to come back but that negotiations over board control etc (which is pretty unsurprising) broke down[1]. Thus disproving your point, in my opinion. There may
296.
by cjbprime 3y ago
I think there are lots of intuitions for why this is not enough to explain their capability. For example, if you train an LLM on English and French text, and then use reinforcement learning with English examples to teach it how to answer qu
297.
by cjbprime 3y ago
This would normally be true, but this is the board of a non-profit that has no investors. If the board actually lacked autonomy, then the article we're commenting on would have been about Altman being re-hired today.
298.
by cjbprime 3y ago
You don't "bootstrap" a chip fab.
299.
by cjbprime 3y ago
I think there's a minor confusion here, where you think the purpose of my response involves doubting whether your system was successful. I understand it was successful. My response is to the sense in which your comment can be interpr
300.
by cjbprime 3y ago
I admire your restraint in writing this comment. :)

More ›