Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
gtirloni
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
17 ms
·
361.
▲
by
gtirloni
2y ago
The ToS are a nightmare, as usual for these services.
362.
▲
by
gtirloni
2y ago
Is it possible to add it to ssh-agent once?
363.
▲
by
gtirloni
2y ago
Any examples you could share?
364.
▲
by
gtirloni
2y ago
Nice accomplishment. I usually do things because I'm excited about something :)
365.
▲
USAFacts
(usafacts.org)
26 points
by
gtirloni
2y ago
|
10 comments
366.
▲
Turning a Lada into an EV with cordless drills
(hackaday.com)
2 points
by
gtirloni
2y ago
|
0 comments
367.
▲
The Price of Intelligence: Three Risks Inherent in LLMs
(queue.acm.org)
1 points
by
gtirloni
2y ago
|
0 comments
368.
▲
Separating from Handmade Cities
(handmade.network)
2 points
by
gtirloni
2y ago
|
0 comments
369.
▲
by
gtirloni
2y ago
> Comments should get more thoughtful and substantive, not less, as a topic gets more divisive. https://news.ycombinator.com/newsguidelines.html
370.
▲
by
gtirloni
2y ago
> Time to get much more engaged.* Yes. And as a non-American following this from a distance, I don't understand what's taking the Democrats so long to start doing real opposition.
371.
▲
by
gtirloni
2y ago
I have been following Go for a while and didn't know that either.
372.
▲
by
gtirloni
2y ago
I didn't know that, thanks. However, I didn't buy anything from the Amplifi line. I'm using the UCG-Ultra gateway, U7/U6+ access points and their standard switches.
373.
▲
by
gtirloni
2y ago
> Complexity exists, you can't make it go away, managing it and understanding it is the only thing you can do. Simple systems only displace complexity. My thoughts, exactly. And considering that so much unnecessary complexity keep
374.
▲
by
gtirloni
2y ago
I'm a SRE and love to work with infrastructure, tweak things, etc, and have my own home lab. For years (decades?), I have gone the DIY route with networking gear and it's been painful to tell family "oh, just another 15min or
375.
▲
by
gtirloni
2y ago
I'm no expert in US history but just looking at the signers list, 3/4 of them were >30yo.
376.
▲
by
gtirloni
2y ago
That's pretty good because it can exhaust the context window quickly and then it starts spiraling out of control, which would require the candidate to act.
377.
▲
by
gtirloni
2y ago
And the majority voted for this. Unbelievable.
378.
▲
Civil Infrastructure Platform
(cip-project.org)
1 points
by
gtirloni
2y ago
|
0 comments
379.
▲
by
gtirloni
2y ago
> Didn’t seem to matter for Twitter, sadly Mostly b cause the goalposts got moved and they don't care anymore about a bunch of things. But yes, it bled users to other platforms but seem still stable-ish.
380.
▲
by
gtirloni
2y ago
My experience working in offices has always been that almost everybody was wearing headphones and there wasn't much talking, except when everybody went to a conference room to meet (and half of them were fiddling on their laptops/
381.
▲
by
gtirloni
2y ago
I just tried with ChatGPT, spoke freely for a few minutes and asked it to create action items. It worked really well.
382.
▲
by
gtirloni
2y ago
In this case, I wouldn't attribute malice to incompetence.
383.
▲
by
gtirloni
2y ago
Double speak from people in power, as usual.
384.
▲
by
gtirloni
2y ago
> Trump's EOs legally mandate that people be judged equally and fairly without regard to things like race. Well, racism is fixed. Finally!
385.
▲
by
gtirloni
2y ago
DEI exists to counterbalance something else that is in place. Can you say what that is?
386.
▲
by
gtirloni
2y ago
> Instead they are simply stripping ideological requirements, period This is not true. If I submit a project to study the large-scale effect of DEI policies in institutions, that will get rejected because "anti-DEI" is in the c
387.
▲
by
gtirloni
2y ago
If you couldn't see the anti-China rhetoric was a smoke screen, now it should be pretty obvious, with the US going down a similar road (eg. vetting projects that goes against the government's ideology... which is ironically what m
388.
▲
by
gtirloni
2y ago
> might be even more reliable simply impossible. when a input parser breaks, it breaks. with a LLM, it breaks... sometimes, for different reasons. Let alone the enourmous computing power needed compared to simple input parsers.
389.
▲
by
gtirloni
2y ago
Check these out: - https://www.hyperstack.cloud/technical-resources/tutorials/h... - https://timdettmers.com/2023/01/30/which-gpu-for-deep-learni... - https://preview.red
390.
▲
by
gtirloni
2y ago
> POSSE is the way to go. If you don’t promote your work, no one else will. I agree but when I tried this a few weeks ago, it looked like a lot of work to publish on my own blog and then replicate everywhere else. I didn't find a go
More ›