One argument is that those who make LLMs have a high amount of control their output. It’s giving a lot of control of open source to the whims of those companies to become reliant on LLMs

Think about the potential for providers to add plausibly deniable backdoors, general lower quality code, etc. when they think you are working on a project they care about. The precedent is there with Anthropic publicly stating that Fable was built to intentional sabotage quality when they think you are working on AI research related tasks or distillation. That kind of stuff can be subtly baked into open models as well

>>> print("proof by counterexample in a python REPL")
proof by counterexample in a python REPL
>>> x = 2; print(x)
2
>>> print("this is not ignored"); print("it's just mostly useless"); print("but you can use as many as you want")
this is not ignored
it's just mostly useless
but you can use as many as you want

Have you tried just compiling it with fewer threads? Would almost certainly reduce the RAM usage, and might even make the compile go faster if it you’re needing to swap that heavily

The good news is that some progress has been made in the US. The semi-recent (2022) FDA modernization act 2.0 removes mandates for animal testing in law and allows other testing methods to be used instead

There’s another bill (FDA modernization act 3.0) that was just reintroduced a few days ago to not just allow the FDA to use non-animal testing, but to require that the FDA start actually working to allow it and setup pathways, rules, requirements, etc. And prioritize the review of drugs done via approved non-animal testing

It includes various reporting, safety, etc. requirements laid out so it wouldn’t just be handing it blindly to the current admin

The 2.0 act was suprisingly bipartisan, so it’s not a given that the 3.0 act would be doomed. Call your house representative and senators to make sure it gets through!

But but it worked on my machine

In 2004, the FDA estimated that 92 percent of drugs that pass preclinical tests, including “pivotal” animal tests, fail to proceed to the market.More recent analysis suggests that, despite efforts to improve the predictability of animal testing, the failure rate has actually increased and is now closer to 96 percent

https://pmc.ncbi.nlm.nih.gov/articles/PMC4594046/

We’ve solved ethics. Pack it up everyone

Great! Can we add that to the instance rule list on the sidebar?

To be fair my suspicious is steam deck users likely make up a larger chunk of recent desktop Linux growth and aren’t (as) likely to go to typical linux spaces online. Though since this is based on browser data, I also wonder how many steam deck users are actually browsing the web on them, so perhaps that my be a bad assumption on my part

parseInt is meant for strings so it converts the number there into a string. Once the numbers get small enough it starts representing it with scientific notation. So 0.0000001 converts into "1e-7" where it then starts to ignore the e-7 part because that’s not a valid int, so it is left with 1

https://javascript.plainenglish.io/why-parseint-0-0000001-0-8fe1aec15d8b

This was just recently removed from sudo. Truly the end of an era

Remove "This incident will be reported." from user warnings. 

This used to indicate that email had been sent to the administrator
telling them that someone tried to run sudo.  Whether or not sudo
sends email is now configurable, so the warning may not be accurate.
It is also confusing to the user since they will not know who the
incident is being reported to.  See also https://xkcd.com/838

https://github.com/sudo-project/sudo/commit/6aa320c96a37613663e8de4c275bd6c490466b01

EDIT: apparently it was added back for some cases

https://github.com/sudo-project/sudo/commit/9757d29a24ac1872872cf09757b0439c54089707

Crossposts already exist, there’s a button for it. For example, I just crossposted this post so now it shows up with a little cross-posted to: World News and my cross-post on World News also shows a little cross-posted to: environment@beehaw.org