Their FOSS local inference service didn't go anywhere.
This isn't Anaconda, they didn't do a bait and switch to screw their core users. It isn't sinful for devs to try and earn a living.
934 karma · joined November 13, 2019
This isn't Anaconda, they didn't do a bait and switch to screw their core users. It isn't sinful for devs to try and earn a living.
Tools that do many things and tools that do a small number of things are still tools.
> "do we really even need a body at all anymore?"
It's a legitimate question. What's so special about the body and why do we need to have one? Would life be better or worse without bodies?
Deep down I think everyone's answer has more to do with spirituality than anything else. There isn't a single objectively correct response.
I'd like to see a math/logic bench appear for tokenization schemes that captures this. BPB/perplexity is fine, but its not everything.
For some reason uv pip has been very slow, however. Unsure why, might be my org doing weird network stuff.
When humans get stuck solving problems they often go out to acquire new information so they can better address the barrier they encountered. This is hard to replicate in a training environment, I bet its hard to let an agent search google without contaminating your training sample.
"Feels oh so telling that Google has to pay Apple and not the other way around. If Google search was so great wouldn't we all be clamoring for it to be the default search in Safari?"
Just have usage limit tiers!
I encourage you to respond to the article and what people say, not what you think they believe. I've been a registered democrat my entire life, and I live in a red state. I don't need to "consider who is responsible" for policies, they're my neighbors and friends, and I have compassion for them even when I disagree with them or think their preferred policies hurt people.
> This article is advocating that using poverty-adjusted scores is a better way to assess student education levels. I just can't agree with that.
No, it's advocating that this is a better way of assessing the quality of the education they receive. And it is.
Rich kids in bad schools can score higher on tests than poor kids in good schools. If you want to isolate the quality of the schooling this is what you do.
> Pardon me, but 20% of your children living in poverty is my personal exact definition for an embarrassing cultural backwater.
First, this has nothing to do with the article, which is about education.
Second, perhaps without knowing it, you're spouting a classic white supremacist take: black people (there are much more in Alabama and Mississippi) are poor because they have a bad culture. If they just stopped sagging their pants and walking out on their kids like the good whites they'd be better off. Even if the intent was to direct this at the poor white people, it's still racist.
Perhaps you should consider that the people who live and struggle in other states are deserving of compassion, not derision. Yes, even if they believe different things from you, or have less money. Practice empathy.
I resisted ranting about it (I hope you'll indulge me now), but in general I don't much like the economics commentariat. As best I can tell, "respectability" seems to mean something like "modal economic views + left of center politics." There are of course some exceptions, e.g. occasionally the economist or WSJ, but it leads to some notable blind spots.
For example, Switzerland experienced mild deflation in a good economy 2012-2017, but it wasn't until 2023 or so that articles started coming out that questioned where the 2% inflation target came from [0] and whether deflation could ever be a good thing [1]. This was despite respectable research from at least 2004 that explored the possibility of "good" deflation [2].
Tim, like many in the commentariat, was writing about inflation in 2019 [3]. The title of this essay is "Why inflation is good for us," and discusses the merits of inflation without any real discussion of cases where deflation could be good. At the end, he even raises the idea of a 4% rate target, which is a tad funny knowing the political mood on inflation 2021-24.
My point, really, is that I suspect I know most of his opinions without reading much of his work. He's hardly unique in that regard, but it bothers me. When looking for interesting opinions I'd probably look elsewhere.
[0] https://www.cfr.org/blog/history-and-future-federal-reserves...
[1] https://www.investopedia.com/articles/markets/111715/can-def...
[2] https://www.nber.org/papers/w10329
[3] https://timharford.com/2019/03/why-inflation-is-good-for-us/
It's entirely possible my interpretation of his opinion of GDP is totally off, but that's what I gleaned from this single article. If I'm wrong, it's at least a bit his fault.
Do you have an opinion you'd like to share?
> This is true, but it used to measure (or at least try to account for) decay in private companies.
No, the "G" in GDP stands for "Gross," which means before depreciation. Depreciation has long been estimated in NIPAs, but that's a separate slate of measures.
> (especially since our beloved MBAs invented 'future accounting', where expected ROI is added to the value of machinery in the books, truly a genius move by true sociopaths)
This isn't true either. You're describing valuation models, e.g. DCF, but these aren't used for accounting. GAAP doesn't have a concept of adding ROI to asset value.
It is very true that games are played with depreciation. For example, you're allowed to depreciate some assets like real estate even if they INCREASE in value over time. All of that depreciation is tax-deductible.
> The real problem with GDP ... is that it tempts us to view all human progress as an amorphous score, which we can start maximising just as soon as we fix whatever we think is wrong with how that score is calculated.
No! The problem with GDP isn't that it's a score, it's that it's a BAD score. We know whats wrong with it. It's fixable, it doesn't ignore very many things! The main issues are totally tractable. In addition to ignoring utility:
1) GDP doesn't account for changes in net wealth.
If I take out a loan to buy candy, GDP goes up. After I eat the candy, the loan remains. I'm financially worse off. GDP still went up! Economy is doing well!
GDP also ignores depreciation. Machinery/infrastructure decay doesn't affect GDP at all.
2) GDP doesn't account for externalities or zero-cost innovation/value.
If I pollute the water supply to make an additional $10, GDP went up.
The inverse is also true. If I make Wikipedia freely available, the value provided to humanity does not affect GDP.
But it's now clear that the activity is the harm. The issue isn’t that using TikTok might make you addicted; the issue is that you are using TikTok.
Just as there’s no such thing as responsibly huffing paint fumes, there’s no such thing as responsible social media use, only sub-diagnosable use.
An evangelical and an atheist will probably disagree about the helpfulness of spreading the gospel, for example.
1) My understanding is that most the "researchers" and even students at ATI have a main role at a home institution. Really the question should be something more like "did ATI funding help advance UK research generally?" not "was ATI able to claim its own successes?"
2) Nobody has really mentioned how PhDs are a little different in the UK. E.g. you propose a topic -> research for three years -> receive PhD. In the US its more like research for 2+ years -> find a promising area -> do groundbreaking work over 1+ years -> receive PhD. In other words, its significantly easier to change your research direction in the US, of course its easier to miss out on things like the LLM explosion.
My suspicion is that this happens because the heart is trying to maintain a constant oxygen supply to the muscles. However, if blood pressure is lowered by improving factors like arterial flexibility, stroke volume (amount of blood per heartbeat), muscle oxygen efficiency, or oxygen content in the blood, the heart wouldn't need to pump faster.
A base assumption is often that orbital solar requires a panel in space, but even that might not be the case.
> Maybe FTX did teach them something
The new tone is all about business fundamentals and the FTX collapse had literally nothing to do with business fundamentals. If this is the lesson they learned they learned the wrong lesson.
VCs did get burned by speculative investing in the 2019-21 period but FTX wasn’t like the others. At the time of its collapse, FTX was profitable and had $1+ bn in revenue, its doom had nothing to do with product market fit, revenue, margins, etc.
Quibi might be a more relevant example of a learning opportunity.
I doubt we'd would be able to handle seeing poor people die in the waiting room because of a ruptured appendix or something, but you don't actually see people go bankrupt so its almost like it never happened.
If they didn’t change something they would simply cease to exist. Does anyone have a better idea?
It turned out my bluetooth headset was using the same band as the wifi but I only figured this out after a few months and a replaced wifi card. I wouldn't wish that experience on my worst enemy.
I'm pretty optimistic that by the time google scholar really goes to shit they'll be good enough to pick up the slack.
Failure to maximize returns is not losing money.