Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
phreeza
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
17 ms
·
211.
▲
by
phreeza
3y ago
It's basically a dark pattern. Not illegal or anything, but designed to mask relevant information from the consumer.
212.
▲
by
phreeza
3y ago
> Sure, it's possible, but it seems much more likely that continued increases in CO2 will have worse unpredictable effects. I completely agree.
213.
▲
by
phreeza
3y ago
The whole point is that it is hard/impossible to predict, and I am even hesitant to list some ideas because they sound like climate denier conspiracy theories. But just as a hypothetical example, there may be effects from aerosols ejec
214.
▲
by
phreeza
3y ago
I think it's not really clear that nothing bad would happen if we stopped fossil fuel consumption right now. It would be pretty massive perturbation on a complex system. I think we should still do it because the alternatives are even w
215.
▲
by
phreeza
3y ago
https://64.media.tumblr.com/d6de5bcddb77eae9d76d370e5f62bce7... the "paradise" area seems to roughly align with the extent of the Austro-Hungarian empire.
216.
▲
by
phreeza
3y ago
Same in Austrian German (Paradeiser) so perhaps something to do with it the Austro-Hungarian empire.
217.
▲
by
phreeza
3y ago
Too bad the images seem to be gone?
218.
▲
by
phreeza
3y ago
Except in my experience as the father of a 2 year old it is not correct. The tracks don't really buckle upwards appreciably.
219.
▲
by
phreeza
3y ago
I went to visit a post office sorting center with my team at Google in Switzerland once. And honestly even as a grown up, it was a great experience, especially since the thing the sorting center was doing was sort of analogous to what some
220.
▲
Low-sulphur shipping rules are affecting global warming
(carbonbrief.org)
3 points
by
phreeza
3y ago
|
1 comments
221.
▲
by
phreeza
3y ago
Yea I'm just saying it's not like all hardware projects that Google does are "great research but not actually manufacturing anything." as claimed by OP.
222.
▲
by
phreeza
3y ago
The closest equivalent Google has is the TPU which is very much a manufactured thing at large scale.
223.
▲
by
phreeza
3y ago
That's the first time I hear that the main current and future use for chatgpt is entertainment.
224.
▲
by
phreeza
3y ago
This meme that models are (mainly) hindered by the constraints on political correctness is really tiring. The alignment tax to get that is really not that huge, and what are the use cases enabled by not having it? Kinky roleplay dialogues?
225.
▲
by
phreeza
3y ago
It's an interesting topic but the article is very weird, seems like SEO spam almost.
226.
▲
by
phreeza
3y ago
Measuring at the group head is the critical part here, as opposed to at the boiler, because in most machines there is a lot of thermal mass which will actually equalize temperature fluctuations that happen on the scale of seconds or minut
227.
▲
by
phreeza
3y ago
https://www.cs.princeton.edu/courses/archive/fall13/cos226/l...
228.
▲
by
phreeza
3y ago
With wordle at least I think the appeal is that you can play it async with friends and discuss your outcomes. "Have you solved the wordle today?" "Yea it was a tough one!".
229.
▲
by
phreeza
3y ago
What is spamtec & ytcracker ?
230.
▲
by
phreeza
3y ago
There are very efficient algorithms for doing this, but of course it may still be expensive if your dataset is very large. See https://ann-benchmarks.com/ for some of the algorithms
231.
▲
by
phreeza
3y ago
I had one of those remote control watches, it was a Casio CMD-10 https://www.reddit.com/media?url=https%3A%2F%2Fi.redd.it%2Fn...
232.
▲
by
phreeza
3y ago
Someone linked to this on the manifold market: https://imgur.io/a/AY1oaIO it does look a bit weird to me but I am not expert enough to tell if this could be explained by optical/compression effects.
233.
▲
by
phreeza
3y ago
The funny thing is it's mostly not even real scientists doing this kind of policing.
234.
▲
by
phreeza
3y ago
The wifi on the trains comes from the cellular network, too. If you experienced cell phone working better than the on board wifi, that was almost certainly some other issue, the performances are very correlated. The problem is not so much w
235.
▲
by
phreeza
3y ago
That's not the German system I remember. In physics at least there was graded homework, it was just not required. The grading was actually invaluable because they prepared you for the actual exam you were going to get (can vary a lot b
236.
▲
by
phreeza
3y ago
I completely agree that humans are not always right, but neither is the next token in a text sequence always "right",i.e. the thing that a raw LM is best at. I just object to the blanket statement that fine tuning will make a mode
237.
▲
by
phreeza
3y ago
Instruction fine tuning improves performance on many tasks that are not in the fine tuning dataset. See e.g. table 14 in the instructGPT paper.
238.
▲
by
phreeza
3y ago
That is blatantly false, these models only work as well as they do in the first place because of instruction fine tuning. The raw models are much harder to work with.
239.
▲
by
phreeza
3y ago
CLIF wrappers are a bit less painful for python in my experience. https://google.github.io/clif/
240.
▲
by
phreeza
3y ago
Why? If anything, they will become instantly more productive, so they may even be more in demand?
More ›