You know what's aligned even more with circadian biology?

Keep the current system, get up off your ass, and change your clocks twice per year without whinging like a precious snowflake.

That way you you don't have to deal either with the sun beating down on you while you try to sleep at 4am or with your poor children getting run over in the dark on their way to school.

Sure, a couple of codgers might keel over in march, but they were probably gonna die soon anyway. It's much harder to measure the statistics of all the other people who would die steadily throughout the year due to the aforementioned issues, as well as similar problems caused by a static clock that doesn't adjust at all to when the sun actually rises. However, I would bet good money that it would add up to far more casualties overall.

For some reason, I seem to be in a small minority on this issue, though. It's probably because everybody would rather believe in a solution to the problems of living on a tilted axis that's clear, simple, and wrong.

("It actually makes RAM needs worse" -> "MoEs actually make RAM needs worse")

It actually makes RAM needs worse than a dense model, for a given quality.

That said, inference techniques like FreeToken are improving MoE swapping performance, so it's not as bad as it once was. Honestly, I would not be at all surprised if we start doing training in a swap-aware manner, where expert cache misses count as loss and models can queue experts to start loading before they're needed. Could even train in two tiers of loading - RAM and disk. Wouldn't that be great? Could run multi-terabyte models on consumer hardware. Training would tend to tend to concentrate key logic and reasoning on small number of very active experts, mixed reasoning/knowledge on less common experts that are usually kept in RAM, and rarer knowledge on experts that usually remain on-disk until needed.

There are well known, reputable abliterators on HuggingFace.

Also, censorship usually (not always) is something you care about for chats, not agentic work.

A .safetensors or .gguf model cannot run code unless you let it. No agentic framework = all it can do is write text. It can't hurt you directly. At worst, it can only lie to you.

It's big, but not good for its size. They're using tricks to pretend that they're better than they are. For example, compare the numbers that they list for the competion on DeepSWE up against the actual DeepSWE scores.

Not cool, Mistral.

Artificial Analysis is overrated, but yes, Mistral is playing fast and loose with their claims.

It takes one hell of a toll to try to do verbal exams 1:1 on students to evaluate them.

Nobody said "verbal exams 1:1".

The only thing that is required is that they be separated from their computers and their phones.

Nobody, I repeat, nobody who uses AI is going to see "OpenAI adding watermarks" and think, "Oh, I better use AI more, especially OpenAI".

And nobody, I repeat, nobody who doesn't use AI is going to see "OpenAI adding watermarks" and think, "Oh, I better use AI more, now that it'll be easier to see that what I did is made by AI."

It is not "an advertisement of the power of the technology". It's a regulation forced on them by the EU's AI Act, which neither companies nor users want.

There's an entire grey-market industry of rich assholes traveling to third world countries for "young blood" transfusions.

Funny thing is, it apparently works and the current study is just one more bit of proof. One quick trip to Ecuador and bam, 70+ year olds get six months of being as healthy as a 30-something again. Hell, I have to admit some mornings I'm tempted to try it myself, though I'm nowhere near wealthy enough to fly halfway around the world every few months.

Perhaps you should be realizing by now that homework is no longer a viable way to test how well students actually know the subject matter, and the only viable way to do so is in-class tests/exercises - frequent enough and worth enough of the grade to force them to study between lessons.

OR, alternatively, you could just keep doing what you're doing, and instead, in effect grade people on:

Comparably poor grades:
* People who actually do the work by hand
* "AI N00bs" who let stuff like that slip through

Comparably good grades:
* AI-experienced people who know enough to skim over the work, do multiple passes, or issue prompts that prevent such tells from getting into the work

Wherein, in effect, your net reward/punishment is inverted vs. what you should want it to be.

Closing one's eyes and ears to the fact that the environment students operate in has changed isn't an acceptable option. Education must adapt. And watermarks, BTW, will only catch "the n00bs"; more experienced / up to date students will keep track of what watermarks and what doesn't, and use AIs that don't watermark, or de-watermarking services.

I'm sorry you're getting downvoted, because I get what you're saying and largely agree (though I love programming and have zero interest in working as a meat proxy / reverse centaur).

The problem here is, you're describing a world where human labor is worthless. And I'm okay with that in the abstract; I harbor no delusions about the "nobility" of work for its own sake. Know who won't be okay with that (labor-cost-savings aside), however? The billionaires with a vested interest in preserving cut-throat capitalism who currently own all the frontier AIs.

Are we on the verge of a post-scarcity society? Maybe. Are the guys dangling that carrot, while building massive survival bunkers as quickly as possible, the same folks who have the most to lose from exactly that? Absolutely. It'd be awfully nice if we could get there without half the human population starving to death in the process.

I'll repeat: watermarks are anti-marketing. They discourage use.

Exactly. I give a few embedded/industrial linux lessons at the university. Some students regularly give me code with comments like "/* and here replace with the name of your variable */" or similar. Some people are not very bright.

SynthID in text genuinely is invisible. The probability shifts are small, and all the pathways are valid pathways to express the same thought.

And no, you cannot "collide" text watermarks. At best, it'll only have the latter one's watermark. At worst, both.

It'll still get tagged by SynthID.

You have to use a non-watermarking LLM. Saying "Don't do anything that will watermark it" doesn't help.