Viewing profile — bonoboTP

The text is AI. The web design also screams AI.

It's more obvious, but since people use their phones all the time anyway, you really have to look at the angle they hold the phone to be sure. So it's not super hard to be sneaky a…

Phones already make people much more reserved and less "wild" eg at parties compared to pre-phone-cameras (that are always online and can upload to social media effortlessly).

Almost all these patents exist as a war chest and a deterrent against other companies. Everyone theoretically perpetually infringes on each other's obscure patents all the time, bu…

They were RL trained on verifiable rewards. It's not purely learning to predict the next token of a human produced stream.

Well, they don't see it as privileges, but accommodations for a health condition. The measurement method needs to be adjusted when the health condition is present, to make the same…

Well, the prevailing wisdom is that these accommodations improve fairness. To confront that way of seeing things is an uphill battle in the current intellectual environment.

The LLM doesn't have to be the sole single only exclusive source you learn from. It's an addition to a mix. Nothing is a single definitive answer. Same way as many other things, li…

It is sometimes typos, but even those can be confusing. But okay I'll admit books are usually on the end of simplification, at least for hard sciences. But sometimes they do handle…

This is not a silver bullet that will make everyone a genius. Not even the best human tutor can do that. But it can be a massive clearer of certain roadblocks for people with no be…

Scrape it off. Lose a millimeter of the stone. Not the end of the world. These are self inflicted bullshit costs that cripple the government. Use common sense. If it's legally impo…

If you give it to everyone, it's not "extra", is it? It's just a longer exam.

Why does it take 85k to remove it from the wall? I'm pretty sure it doesn't cost that much. There is some ridiculous government inefficiency or embezzling happening or some kind of…

If you filter for younger, 25-34, and for employed, it's closer to 50%. For employed young women, it's the majority (for them, if you count associate degrees, it shoots up close to…

I'm agnostic on the matter because post training and RLHF on this specific capability could be quite effective. Nobody ever tried because it would cause a backlash and bad PR. Much…

I don't think any frontier lab tried to make an LLM optimized for the literal Turing test where the LLM is made to argue that it's human. This is not in their interest. They make t…

Claude Code does allow this. You can talk to the subagents directly and watch what they are doing.

Of course. And mechanistic interpretability research is a thing.

At low student counts, like 20, (think specialist niche elective master's courses, not the huge bachelor's courses) it's perceived by profs as more efficient in fact than the ordea…

It's very possible but somewhat slower. PyTorch and CUDA have flags for determinism. It won't work across all different GPU models though, but it will get you bitwise equal results…

American colleges give way more support throughout the studies though. In Europe you're literally alone and nobody cares if you miss deadlines or you're failing out. No general con…

Grading written exams is delegated to teaching assistants / PhD students.

Oral exams are still common in Germany and many European countries in case of smaller courses taken by 20-30 students in a semester, and sometimes even in larger courses, and the t…

None, but have interacted with American students at European universities and they were very lost without all the coddling and all kinds of support infrastructure and basically get…