robotresearcher

5 ms·

1.
by robotresearcher 6d ago
Ah. That’s an unusual thing to mean by ‘coin flip’!
2.
by robotresearcher 6d ago
The coin flip method you describe breaks the same-query same-server locality (unless adding or removing servers) that is one motivation for the consistent hashing method. You could solve that by storing the new-query flip result, but the
3.
by robotresearcher 7d ago
Not foolish. The constraint of no-need-for-globally-consistent-state is so important and rules out so many approaches that it was well worth stating in the article. Indeed the statistical model described in the article does not model the di
4.
by robotresearcher 7d ago
Customers often want to pay less, and shareholders often want lower capital costs. A tasteful optimization is a win-win. The opportunity cost should be traded off against new features of course, but tasteful optimization is a good thing for
5.
by robotresearcher 7d ago
Ok, let’s talk apples to apples. I’ll talk about RAM instead of disk, and I have a workload example for you. As a graduate student in AI in 1995, I bought (with granting body’s money) a workstation for running my PhD experiments. It was a
6.
by robotresearcher 8d ago
No, the parameters are updated online as new run-time data comes in. Infinite different param values over time, but a constant number of params. It's a stretch really, since by this usage a one-parameter model that updates itself is
7.
by robotresearcher 8d ago
From a quick look at the paper it seems they are showing how to update weights online via projection through a smaller matrix like a dynamic version of LoRA. That's weights changing, and not the architecture or training approach. Wei
8.
by robotresearcher 9d ago
No one claimed those things you’re arguing against. The claim is simply that storage has never been more expensive. A Macintosh 128K with a FLOPPY DRIVE cost $7,500 in inflation adjusted dollars. Can we please take a view of today that’s in
9.
by robotresearcher 9d ago
> when that storage is at its most expensive it's ever been. Most expensive it's been for a handful of years. 1TB of spinny disk cost a million dollars as recently as 1994, and half a billion dollars in 1980. https:/&#x2F
10.
by robotresearcher 10d ago
> one of the best parts of vacations is actually planning it For some. Others hate or fear it. For others it just feels like work.
11.
by robotresearcher 11d ago
If you’re satisfied using a third party bootloader, and one exists, there’s no point writing one ever.
12.
by robotresearcher 11d ago
You probably want to use a working bootloader to see if your ‘hello world’ OS-fragment works at all. No one is writing much OS without running it, surely.
13.
by robotresearcher 12d ago
The problem with that approach is that “how do I handle this query”, and “did I handle that query ok” are both AI-hard problems, and must be tackled near the user, where the large model is not. It also requires a relatively complex distribu
14.
by robotresearcher 17d ago
See also: Power steering Jet engines Anesthetics Rubber tires Flat screen TVs How grubby of them to take a high end feature and make it more affordable.
15.
by robotresearcher 17d ago
You do. ASML, ARM, Linux (originally), Spotify, Logitech. Novartis, Astra Zenica.
16.
by robotresearcher 18d ago
What name do you use for an approach that probably gets a better than random success rate (without guaranteeing that)? I’m leaving out the ‘pseudo-scientific’ part of GP’s description, since it doesn’t entail anything technically.
17.
by robotresearcher 18d ago
So it’s a heuristic. Lots of very valuable heuristics exist.
18.
by robotresearcher 20d ago
The word was invented by Richard Dawkins, an evolutionary biologist. He described ideas as reproducing with variation under selection pressure, which is the definition of the evolutionary mechanism. There is a more recent use of ‘meme’ to r
19.
by robotresearcher 21d ago
Oh, I see. That’s a very different requirement. It’s not a technical limitation but a product decision to not allow training. An advantage of properly open source models is that you can train and tune them. It’s an interesting challenge tho
20.
by robotresearcher 22d ago
Well, Go is a pretty complex game, and AlphaGo RL’d its way to excellence just by playing the game like you describe.
21.
by robotresearcher 22d ago
“Hello, how are you?”
22.
by robotresearcher 22d ago
The second half of the article talks about avoiding this, by not keeping any separate index of the (un)allocated orders. All the orders are allocated, and all are processed by the same pipeline, it's just that some of them are nearly n
23.
by robotresearcher 22d ago
Whether it's regular ordinary everyday slang or not, it's exactly hyperbole. https://en.wikipedia.org/wiki/Hyperbole
24.
by robotresearcher 23d ago
If this was awesome, we would have it on wheelchairs. We very rarely do. Stairs are built for legs.
25.
by robotresearcher 23d ago
> We either go to the store and bring back groceries ourselves or pay a helper to do it. Or order it from Amazon, in which case there was likely a robot in the pipeline. Robots are very widely deployed, but almost entirely invisibly to t
26.
by robotresearcher 23d ago
The hope is that RL in simulation can fill the gap. Also there’s imitating humans, via a suitable mapping from the human sensor, control and configuration space to the robot’s. Some groups have gathered video and other data from humans doin
27.
by robotresearcher 24d ago
Yuck. I still fancy my chances buying an app-free toilet seat for a while.
28.
by robotresearcher 24d ago
Do smart TVs require an app? Or do they have a self-hosted setup process?
29.
by robotresearcher 24d ago
I agree with the thesis that expecting an app is a non-trivial ask of the user, particularly for those with relevant disabilities. But if you purchase a toilet that needs an app, it’s not really the toilet that obligated you to use the app.
30.
by robotresearcher 24d ago
Right, but “two or three out of potentially hundreds” doesn’t mean the same as “hundreds”. That’s the hyperbole the parent was pointing out.

More ›