ziofill

21 ms·

331.
by ziofill 2y ago
Canada needs houses, not people.
332.
by ziofill 2y ago
At each iteration the LLM has the older code in its context window, isn't it kind of obvious that it is going to iteratively improve it?
333.
by ziofill 2y ago
An argument that I’ve heard is that wooden boards can get more infected with bacterial colonies if not washed and maintained properly. I’m not sure how actually true this is though.
334.
by ziofill 2y ago
Here’s an idea, donate most of it to a reputable charity organization and go back to work.
335.
by ziofill 2y ago
This is his main point, and I wholeheartedly agree: …a perception can be created that all of one's peers are achieving either success or controversy, with one's own personal career ending up becoming the only known source of exam
336.
by ziofill 2y ago
This is so confusing. What stops people from copying music verbatim (with little effort) and obtaining copyright? And once someone has done so, if I try to do it too how can I claim I was transcribing the original and not the copy, since it
337.
by ziofill 2y ago
the 3D model doesn't load
338.
by ziofill 2y ago
No, in my case it was a theoretical work, but others have measured it e.g. https://www.nature.com/articles/s41586-020-1989-2
339.
by ziofill 2y ago
As usual, journalists write clickbait titles. The quantum geometric tensor has not been measured for the first time . Perhaps it’s a novel way to measure it in a crystal, but certainly it’s a very well known concept in quantum physics. I’v
340.
by ziofill 2y ago
how did you do it? care to share the code?
341.
by ziofill 2y ago
I think the word you need to use in this conversation is iid.
342.
by ziofill 2y ago
It's certainly remarkable, but let's not ignore the fact that it still fails on puzzles that are trivial for humans. Something is amiss.
343.
by ziofill 2y ago
It's very sad distill.pub doesn't accept new submissions... :/
344.
by ziofill 2y ago
How are we sure that the model answers because of the same reason that it outputs on the scratchpad? I understand that it can produce a fake-alignment-sounding reason for not refusing to answer, but they have not proved the same is happenin
345.
by ziofill 2y ago
I've been playing piano for 30+ years, and I only learned to use the circle of fifths this year. It's been short of a revelation and I can't recommend enough to practice scales and drills based on it.
346.
Two-inch long 'murder hornets' eradicated from US(theguardian.com)
6 points by ziofill 2y ago|1 comments
347.
by ziofill 2y ago
I recommend Michael Levin’s YouTube channel. Lots and lots of fascinating discussions.
348.
by ziofill 2y ago
Yes but that’s the same as gathering new information about them
349.
by ziofill 2y ago
I may have misunderstood, but I think that it depends a lot on the existence of a validation mechanism. Programming languages have interpreters and compilers that can provide a useful signal, while for images and natural language there isn’
350.
by ziofill 2y ago
IMHO the only solution is to change the way students are assessed. It may be more time consuming for the instructors, but a good old one-on-one conversation could be the best solution. Most of my university exams were oral (and written), so
351.
by ziofill 2y ago
Abstracting away from the problems with supports and standards and encodings which will invariably evolve, preserving information in time requires error correction, which in turn requires redundancy of some kind.
352.
by ziofill 2y ago
I track my migraines on a markdown file, but that’s as far as I go
353.
by ziofill 2y ago
Very nice article. There’s an alternative way of thinking of the Gibbs paradox, which is to attach labels to the particles. This naturally makes them distinguishable and increases the total number of possible configurations, and with it the
354.
by ziofill 2y ago
Can it be that he assumes you have some device that produces somewhat bad magic states and then you distill them into a better one? That would be the typical situation in practice.
355.
by ziofill 2y ago
Would an AI trained on filtered data that doesn’t contain examples of devious/harmful behaviour still develop it? (it’s not a trick question, I’m really wondering)
356.
by ziofill 2y ago
Is the opposite possible? I mean to have the dependency resolver use a sat solver? Would it be faster/slower?
357.
by ziofill 2y ago
You couldn't pay me enough to squeeze myself in narrow tunnels underground. But I'm glad there are brave explorers who can do it.
358.
by ziofill 2y ago
Very sorry to hear that Marshall died :( I just went on howstuffworks.com and I see two articles on astrology on the home page. For real? I thought it was a science-based website.
359.
by ziofill 2y ago
Physicist here. Classical error correction may not always be a straight up repetition code, but the concept of redundancy of information still applies (like parity checks). In a nutshell, in quantum error correction you cannot use redundanc
360.
by ziofill 2y ago
I remember attending a German language course in Berlin (25 years ago) with some other international students and this american kid who was railing against using die and der for objects that clearly had no gender: "what do you say?? di

More ›