hedora

39 ms·

751.
by hedora 8mo ago
I've settled on XFCE. It just works. You have to turn too many knobs to make it work on weird DPI / screen sizes, but other than that, it's fine. Recently, I fired up Win 3.11 in 1600x1200@256 mode to run SimAnt, and was pr
752.
by hedora 8mo ago
My M2 macbook pro runs qwen fine, and so will any mac mini with maxed out ram.
753.
by hedora 8mo ago
Apple has been doing personal agents for a while. They're crushing it so hard they must be tired of winning at this point. For instance, the other day, the Siri button in maps told me it couldn't start navigation because it didn&
754.
by hedora 8mo ago
I was stuck on the part where they said neither party could provide cheap abundant decentralized clean energy. Biden / Obama did a great job of providing those things, to the point where dirty coal and natural gas are both more expens
755.
by hedora 8mo ago
c) it is controlled by a direct competitor and can bury / promote your customer communication at will.
756.
by hedora 8mo ago
Oh crap. I just logged into HN to ask if anyone knew of a working alternative to the Claude Code client. It's lost Claude's work multiple times in the last few days, and I'm ready to switch to a different provider. (4.6 is
757.
by hedora 8mo ago
I’ve noticed the opaque weekly quota meter goes up more slowly with 4.6, but it more frequently goes off and works for an hour+, with really high reported token counts. Those suggest opposite things about anthropic’s profit margins. I’m not
758.
by hedora 8mo ago
After playing around with vector db's for a bit, I got tired of this sort of thing. There's a reason most of the industry has switched to letting agents grep + navigate within the results. Come to think of it, it's probably t
759.
by hedora 9mo ago
If we ever build another house, it's going to be attic-free with exposed conduits + hvac ducts / pipes on the ceiling. Every electrical box is going to have a 2" conduit (embedded in the wall) running up to a conduit that ru
760.
by hedora 9mo ago
Rodents love PEX. If we knew, we'd have used copper. It would have been cheaper in the long run. It just takes one rat.
761.
by hedora 9mo ago
We've had great luck with the removable 3M velcro picture hangers. (Each corner is held with two pieces of velcro that face each other. The velcro has double back tape on the back, which affixes to the wall and the picture. The doub
762.
by hedora 9mo ago
Noise barriers are incredibly cheap. They're a tiny fraction of the cost of putting up a drywall or plaster wall. Codes say how far apart the studs should be, so you just buy a roll of batting, and unroll it into the gap between the
763.
by hedora 9mo ago
Modern plaster has the same properties, and works well with stud finders. On a related note, if you can find a strong rare earth magnet, you can use it as a stud finder. It'll be attracted to the nails used to hold up the drywall &#x2
764.
by hedora 9mo ago
Drywall is terrible vs. modern plaster. Modern plaster has backing boards that are similar to drywall, so you get most of the construction advantages (except for the labor intensive step of plastering), and can hand pictures / toggle b
765.
by hedora 9mo ago
Did they edit the article after your post? There’s a list just after the first paragraph. Anyway, I was just considering switching to android today after repeated problems with the keyboard, maps, siri and notifications / incoming cal
766.
by hedora 9mo ago
Why would agents use this? HTML -> Markdown software is readily available, and some percentage of the internet is hostile towards agents. Also, isn't the conversion lossy? I imagine an agent would rather have access to the HTML, an
767.
How the FBI might have gotten inaccessible camera footage from house(npr.org)
16 points by hedora 9mo ago|5 comments
768.
by hedora 9mo ago
For the bit less technical: Assuming aggressive but non-breaking quantization, for LLMs, you need ~ 1 byte per parameter to run. I’ll optimistically assume diffusion models are at least as fast/quantizable (has anyone here tried?) So,
769.
by hedora 9mo ago
Oh, to be clear, there’s no way we’d want incoming code for these forks. Incoming bug reports or design docs an LLM could implement? Sure. Maybe something like the Linux approach (tree of well-tested, thematic branches from lieutenants) wo
770.
by hedora 9mo ago
Are there actually open source developers that wander from project to project with one-off contributions that are of significant value? This seems to optimize for that specific scenario, and it’s not something I’ve seen in practice. The co
771.
by hedora 9mo ago
Ovet-strict spam filters usually lead to de facto shunning of the person that doesn’t realize their incoming messages are being dropped. I think that’ll also happen to most open source projects that adopt a policy of silent auto-rejection
772.
by hedora 9mo ago
To play devil’s advocate: We’ve vendored a few open source projects by just asking an LLM to fix obvious bugs that have been open for 12+ months (some projects are abandoned, others active). If upstream can’t be bothered to fix such stuff (
773.
by hedora 9mo ago
If you don’t follow spam links, then it lets the spammer probe your spam filter, and try stuff until you follow links. A better approach is to follow all links always (even to non-existent recipients) if you must play this game. That remind
774.
by hedora 9mo ago
That sounds like a feature, not a bug, given where Google’s revenue comes from.
775.
by hedora 9mo ago
There’d be a bigger problem for the security training folks if there was a 100% pass rate.
776.
by hedora 9mo ago
I wonder how they react to expected inflation, and how to build that in. Official numbers say inflation has been 3% since Trump got in office in 2025. We’re seeing > 30% across the board on our bills, and GDP “growth” statistics imply i
777.
by hedora 9mo ago
Here’s a list of “transports” which (I think) are storage backends that manage the passkeys (preventing the user from controlling their own credentials). https://github.com/linux-credentials/libwebauthn?tab=readme-...
778.
by hedora 9mo ago
You’re thinking about this wrong. Leadership says the gap between human and LLM performance on complex tasks is too big. The easiest way to close it is to prevent the humans from sleeping.
779.
by hedora 9mo ago
Our cars have a total of 8x the capacity of our home batteries. We're ~ one dongle away from doing what you describe. The other thing we need to see to really replace grids are panels that are optimized for overcast days. Currently,
780.
by hedora 9mo ago
This will be especially true if people keep driving cars. An EV with 300 mile range has something like 2-4x more batteries than a typical whole-house battery. It's easy to imagine houses will go into "away mode" when there&#

More ›