boredatoms

Its countered by competitors for inference. You could locally host a model and have your cost be fixed by your infra costs

For some things the just-crash is ok, like cli usage of curl

Q on your setup

Are you using ULA prefix for the nat66/dhcp6, are you also allowing GUA address assignment via slaac? Im wondering how it works out with source-selection

You may be interested in heretic. People often post models to hf that have been un-censored

https://github.com/p-e-w/heretic

Nothing stops you running a NAT for v6 too, its just people tend to choose not to when given the choice

Presumably thats with the network having a PLAT somewhere if you’re relying on CLAT for any v4-only connections when you use safari

Perhaps its time to put a VPN into all your CI jobs

Telling people to use ULA subnet fddd:: with dhcpv6 is my way.

fddd::7 is easier to type than 10.0.0.7

Could be a tariffs joke

fddd:: is easier to type than fd00::

..is this RVA23?

It is over a couple of years

It looks like an in-kernel replacement for jool is coming

https://lore.kernel.org/netdev/20260319151230.655687-1-ralf@...

If we actually get to the point of address shortages,

Either, NATv6 would become a thing, or instead I hope SLAAC would get deprecated and dhcpv6 would become mandatory so we could give out smaller than /64s

Sure, then that with case law too?

Is there something like this for the US?

Im reminded of a book from Dick Smith in the 90s called “Funway into electronics”.

It had a wonderful kit where you would use screws in a board so you could see every wire path. Much easier to explain than modern breadboards

How do you figure out what detection method a carrier is using?

Do we have a process to make new helium from hydrogen?

I hope one day we can use wasm without html, just pointing the url at the wasm file

Under specified build dependencies that use libraries/config on your host OS rather than the target system

You can solve this on a per language basis, but the C/C++ ecosystem is messy. So people use VMs or real hardware of the target arch to not have to think about it

The issue is really about if the DC is water-cooling capable

I do the same. It simply means theres less accidental leakage / self-doxing that could be pieced together if you (or llm) read every comment on the account.

Suggestion: Pick a long term account, dump the comments, and see what an llm could figure out about the target

While destroying morale, and increasing the difficulty of successfully recruiting later

This means that the divisions of administration are too small. Perhaps it should just a be county level or even state level concern

Power/cooling is the premium.

Can always build a bigger hall

Its long term microsoft culture to be horrific at external naming

> Maybe it'd be more efficient for the transport to come at regular intervals though

Efficient for who, is the problem

We’re less worried about a low-scale low impact fraud my many people that is unlikely to alter results, than a systematic mass fraud by few people who can choose a result

They could buy global foundaries and pour in a pile of cash, 5 years later they’d have something useful

Or they could buy out Intel and sell off their cpu design division