Viewing profile — jmalicki

pytorch is really mostly C++ after all, you don't have to use it with python. Python just allows you to use it like "normal code", whereas if you program it in C++ you really get e…

Does 1 really happen? I would never trust a compiler where 1 was a possibility. If it can work it should.

JVM does a lot of escape analysis to turn heap allocated memory into stack local variables. It doesn't matter if it's local since it's a VM, it's doing it at runtime and can change…

Who other than Noctua would notice?

"What kind of stocks should my mutual fund invest in?" "What sort of startups should I fund?" "Should my business aggressively hire this quarter anticipating growth, or hold out fo…

This is very common for economic statistics. GDP too. That's how you can get more accurate information, and you have actual numbers on the discrepencies. They rely on the household…

The main text doesn't plot the error bars, and the full data tables in their release don't have error bars either. You have to really dig into their website to find their std error…

No, they shouldn't. Timely estimates as to the best knowledge of our current situation is extremely important for decision making under uncertainty. If the error bars are large, th…

Not in their primary news release. A footnote says it is "generally conducted at the 90% interval", but the individual numbers do not have their actual computed error bars. Yes, th…

DOGE hurt them pretty hard: "The Bureau of Labor Statistics posted a notice on Wednesday stating that it stopped collecting data in three not-so-small cities (Lincoln, Nebraska; Bu…

That's true for the establishment survey, but the population survey, where they contact individuals, is what is actually used to calculate these numbers. It is literally calling/wr…

Everyone knows steak is better well-done with ketchup.

It's a matter of cost. Did you see the 100x cheaper? If you have a workload that is going to be very heavy, incurring a large training cost to make a cheaper model work well with t…

I'm not saying it happened, I am saying they are working hard towards that goal as a business priority, and spending a lot of money on it. Software engineers are first, but other f…

I've actually been involved in annotation projects doing RLHF to train LLMs to do exactly that. It's not a matter of time, it's already happening - it's just seemingly lower priori…

An idea can be both novel and derivative at the same time. You create idea B. It absolutely builds upon idea A, and is in some sense derivative. But your idea B introduces a lot of…

All research builds upon previous ideas. E.g. you can patent a novel idea that would require licensing another patent to use.

I don't understand why from a legal perspective, but it seems like over the last decade or more the tech industry stopped caring about patents. Startups used to talk about their pa…

Also, TIL about the Antonov AN-2 ("Wait... that kind of looks like a Cessna Caravan, but is a biplane, and is a tail dragger... what sort of aircraft is that?")

In California, it's because people (the state department of fish and wildlife) put the fish there. You can even see their scheduled fish plantings online: https://nrm.dfg.ca.gov/Fi…

Isolated alpine lakes do not have connected rivers and streams, yet a ton are stocked with fish by humans. Is it unbelievable that ancient hunter-gatherers would do the same thing?…

Yeah, true European cloud providers for these kinds of things seem to be behind, and a lot of the ones offering data compliance at the level of AWS are small enough that it's a bit…

> Worst-case example: Bootstrapped startup working in military. That's the easiest case. AWS Bedrock models running in AWS Secret Cloud for Industry. (I really have no affiliation …

That's why you can define your own scope to control that.

A) is very doable with e.g. Amazon Bedrock. They'll give you HIPAA compliance, they even have a data center for US government classified data, they can give you European data sover…