Yeah this is very common. You might see the headline "Scientists prove X causes Y", and when you click through all the pop-science journalism until you get to the paper, you'll find "We found a weak positive correlation between X and Y and it's surprising because the prior research found the opposite".
I think other explanations replying are on point. I live in a town that's surrounded by a lot of farm traffic, and most of those roads are in good shape. But there are also routes used heavily by trucks servicing fracking sites, and those roads are TRASHED.
It took only 17 Space Shuttle flights before a crew-member threatened to kill themselves causing them to add additional locks to the hatches. I think a few seconds considering what a malicious actor could do would be worth it.
There are plumbers who make a living but whose work results in leaks in people's homes. They're making a living, but I don't consider the way they work "a good idea".
SpaceX is pretty open about optimizing for many iterations, a bit like the philosophy in software of shipping an MVP to get user feedback sooner for future iterations. Boeing has an established culture that's more like traditional waterfall development. When you watch their launches, they have tiers of objectives that get less and less likely to succeed - they plan to push even if failure is likely tlso they can learn from both their successful objectives and the eventual failure.
If you do, in fact, need H100s, they can be very hard to get. Even the smaller flavors of A100 you sometimes request, wait days for, and then 1 node might show up during a weekend. And for the reasons described in the article and the fact that large training jobs can be network-limited, nicer networks can be a big deal.
I actually attribute my high typing speed to the practice I got with OPEN DOOR, TAKE KEYCARD, CLOSE LOCKER, LOOK CAULDRON. Police Quest, King's Quest, Space Quest. Good memories.
Yeah, let's just measure the density and composition of a comet that will be here in a few months, and with expectations that it may be releasing new gaseous formations as it gets closer, predict what will happen with those formations.
And people who do prefer to live with cochlear implants face pressure from the deaf community itself. You can't win. This was an achievement. A girl who probably saw this as a cure, saw success. Why can't we just be happy for her instead of detracting because others wouldn't make the same choice?
>> Our study reaffirms the limitations of LLM tokenization
Because they used data that needs to be tokenized differently, and didn't really tune the models for use on that data. That's not really a limit of LLM tokenization per se.
>> We did not evaluate strategies known to improve LLM performance, including ... retrieval augmented generation
Which is a shame because this is exactly the kind of use case RAG is supposed to be good for and they largely observed problems it's supposed to help with.
Looking at the authors, it seems to me they're all subject matter experts in medicine and digital medicine, but their conclusion is the one in support of medical professionals and they really don't seem to have tried that hard to get good deep learning results.
I've had nightmares every time I've seen a doctor in the US, frequently because of things not being coded correctly. So honestly I'd just love to see a rigorous study of how often the human staff is messing it up too.
Employed or unemployed, it's seems so offensive to me to not care about people's time. When I'm employed, I'm probably using a vacation day to talk to you. That's special time I would otherwise spend with my family and on not burning myself out.
When I was not employed, I paid for the gas to drive 90 minutes each way and spent an entire day not looking at other opportunities to be hired or other ways to improve my resume, only to be told by the director that although I did very well in the interviews, they would be getting back to me in 3 months when they found out if their hiring freeze was over then, because at the present time the opening was not yet truly open.
Have some respect for human beings, regardless of their employment status.
I always make going through our docs a part of new-hire onboarding. They learn the product, and you learn what in your docs is so confusing that a person who you hired for their supposed ability to help you MAKE your product, can't even figure out ABOUT your product.
But also, I did work on a project once where a test would scrape code snippets from documentation and make sure they still worked as expected. It's bad enough to break compatibility in your CLI, etc. but sometimes you need to do it. It's really bad when you have tutorials, recipes, etc. and they bit rot.
I mean for anything you could have easily verified via a search engine, I would say you always should have been verifying. I've had ChatGPT give me incorrect but correct-seeming information many times, and not just in the last several weeks.
Well it's a bit like how I see universal healthcare. Would it be worth paying billions for? Oh absolutely. Do I believe the current US Gov could actually take all the money in the world and make it happen in a way I'd be happy with? LOL.
https://nvidia.github.io/cuda-python/cuda-core/latest/ https://developer.nvidia.com/nvmath-python