fjdjshsh

In what context do you need to maintain multiple frameworks?

My use case for fastAPI is very specific (we only maintain APIs for ML models), so I'm curious to learn about this.

Most of these corrections seem stylistic / "I would say it this way to make it clear" rather than actual mistakes.

I think you're missing the point. Offshoring (moving the job of, say, a Canadian engineer to an engineer from Belarus) has a one time cost drop, but you can't keep driving the cost down (paying the Belarus engineer less and less). If anything, the opposite is the case, since global integration means wages don't keep diverging.

The computing cost, on the other hand, is a continuous improvement. If (and it's a big if) a computer can do your job, we know the costs will keep getting lower year after year (maybe with diminishing returns, but this AI technology is pretty new so we're still seeing increasing returns)

I strongly believe it should be illegal to post something automatically by an LLM without clearly identifying it as such. I hope countries start passing these laws soon

For this to be true, real wages would have had to collapse (at the very least ~50-70%) for a household of two workers to be less rich than the household of one worker x years ago.

This is just not true. If people feel worse it's because they want more things from life, not because they are able to afford less things.

(I'm talking about the last ~50 years. The last 20 years may be a different point, but it doesn't have to do with female participation in the marketplace)

>3. In the western world companies over past fifty years got a free boost as women joined the labour force, and essentially companies were paying one guy the cost of a household now they pay two people the cost of household getting twice the workers for same price.

If demand for work is fixed, then doubling the supply of work (male + female from the same household) would in principle decrease wages (not necessarily by half). However, the supply need not be fixed. This is what I like to call the "musical chairs theory of employment". Let's say there are 2x more working age people in 2024 than in 1954, does it mean each person earns less? When immigrants come to a country, does it mean that each immigrant causes a native to become unemployed?

The evidence tells us that's not true and that there are other changes in the economy that also increase the demand for work. Check the literature of immigration effects (even major short term shocks) on unemployment and wages to see a pretty clear picture. I'm not aware of a study of what more female employment has done (my guess is that it's much more difficult to study, because it's a phenomenon that has slowly happened over decades. As an aside, this has happened in plenty of places around the world, no need to restrict yourself to the western world. Plenty of other places are relevant)

Hard disagree and it sounds like you're extrapolating from the traditional software area without considering the nuances.

From the point of view of academia, free transformative products that build on other stuff, transparency, possibility of building even better models in the future and a big etc, there's NO difference between ms excel and Google sheets.

OTOH, the llama models have allowed all of the above and has helped us immensely in both developing new things and being able to understand these new generation of LLMs...all of which would be impossible with openai.

Open weight means you get both the details of the architecture and a way to freely iterate to build new things.

I think it's useful because many people consume quantized models (most models that fit in your laptop will be quantized and not because people want to uncensor or un-unlearn anything). If you're training a model it makes sense to make the unlearning at least robust to this very common procedure.

This reminds of this very interesting paper [1] that finds that it's fairly "easy" to uncensor a model (modify it's refusal thingy)

[1] https://www.reddit.com/r/LocalLLaMA/comments/1cerqd8/refusal...

If the model wasn't optimized with a parameter = 3.08 (that is, the optimization figured 3.08 increased the objective function), why would it be better than parameter = 3.1 ?

(This applies to functions of parameters too)

The USA was never non expansionist: they either took over the native Americans or the Mexicans.

Even at war, the USA was still much more "liberal" than Nazi Germany or Imperial Japan: for starters, they didn't put any German citizens / German descendants in a concentration camp.

The Japanese descendants/citizens living in the USA were treated different because of racism. You can argue that having a racist government isn't compatible with being "liberal", but it's a matter of degree. Having a democracy that doesn't allow non whites to vote is still more "liberal" than having a monarchy / dictatorship

They don't need to be causally connected, just be correlated.

>it's an objective fact that California's government is hostile toward it's citizens possessing and bearing firearms

A state puts restrictions on semi-automatic weapons (one model which has been used to kill hundreds or thousands of people in random shootings) and here it's described as "being hostile to people possessing firearms". For most non-americans (which is my case) this will always sounds so strange

Is that your test suite?

I see a lot of anger in the form of "she should get 10 years, it's the right thing to do". I wonder if that's precisely the sentiment that has been part of the reason the USA has become the country with the highest % of incarcerated people. I'm not an American, so I'm wondering if there's a cultural difference here that explains the different feelings / viewpoints.

2 years + owing money to the feds + not being able to get a job: this person will suffer the consequences for her action for the rest of her life. Whether it's fair or not (I'll get to that point later), it's a huge deterrence for commiting crimes, particularly for white collar crimes like this.

Regarding the fairness of the punishment: I have zero feelings about this. The punishments should be chosen to make society work: deter future crimes, keep people that are harmful and hopefully help them change course. In this case, there's also providing incentives to cooperate.

I don't need people to suffer for stuff. If it's needed somehow for society to work, them great, but I don't feel (as many in this thread seem to feel) that long periods of suffering is needed for some reason (to appease their sense of justice? God? The victims? All of the above?). And let's be clear: the suffering in higher levels of security prison in the USA is on a whole different scale.

Your post isn't providing any argument, you're just providing statement after statement of a position that is pretty well known (no emergent behavior, just pattern matching) and pretty popular here on hackernews. Maybe the link provides all the good stuff, but I'd suggest you at least provide some flavor of what your arguments are.

Wolfram is serious about his software, but, to my mind, has been closer to a crackpot scientist than a real researcher the last few decades. Him backing something up diminishes my subjective probability that this is something "serious"

I keep seeing this "expansionist dictatorship" applied to China when the USA is discussed. The USA has invaded plenty of countries in the last few decades, has a history of colonialism (Cuba, Philippines, Puerto Rico...sure, less than some European countries, but still).

Which countries has China invaded in the past few decades?

I think it will eventually be good public policy to make it illegal to post massive amounts of texts produced by AI without disclosing it. As with all illegal things on the internet, it's difficult to enforce, but at least it will make it more difficult/less likely

>Japanese e would only ever be [ɛ]

What's your source for this? Wikipedia says "/e, o/ are mid [e̞, o̞]" [1]

>The mid front unrounded vowel is a type of vowel sound that is used in some spoken languages. There is no dedicated symbol in the International Phonetic Alphabet that represents the exact mid front unrounded vowel between close-mid [e] and open-mid [ɛ], but it is normally written ⟨e⟩.

[1] https://en.wikipedia.org/wiki/Japanese_phonology?wprov=sfla1

That's a good point. Latin letters map to a big number of IPA sounds. I think this is specially true with consonants. Vowels, however, might be the same. For example, Spanish has five vowels (a, e, i, o, u in IPA). Japanese also has five IPA vowels and only the "u" is different in IPA

This is my experience with Hackernews. If the comment doesn't specify the country, it's an American talking about the USA

>it breaks ability for non-speakers to infer meanings

Not sure what you mean by this. Do you mean that it's less convenient for people that don't speak / read Chinese? Why would that be a relevant metric?

You may be missing that character standards have changed over time and that different writing styles (草书，行书) are implicitly simplifications. You can think of latin or Russian cursive as a simplification of the printed letters.

In practice, the phonetic component has been mangled / evolved over time, so simplification doesn't make things more or less difficult for students (be it 5 year old native speakers or 50 year old non native speakers).

"illiterate"

Do you think using capitals at the beginning of a sentence aids comprehension?

I view punctuation and spelling rules as a way to maximize comprehension (akin to having a linting standard). In non formal writing, I don't see any harm in avoiding capitalization (at least it doesn't seem to me to help understanding / reading speed, etc at all).

The problem is with the word "our". If it's just private companies, the biases will represent a small minority of people that tend to be quite similar. Plus, they might be guided by profit motives or by self-censorship ("I don't mind, but I'm scared they'll boycott the product if I don't put this bias").

I have no idea how to make it happen, but the talk about biases, safeguards, etc should be made between many different people and not just within a private company.

Maybe you meant something like 天安门广场上只发生了好事

I disagree: for many of these topics, hackernews is, to me, a source of more level headed discussion and an opportunity to see both sides in a reasonable way. Yes, the quality does go down a lot, but it's still better than in the other sites I read.

I agree with you. The question, for me, is what are they defending against. Are they worried that people will get dangerous information from their model that they couldn't get from searching on, say, google? Probably not.

Maybe their biggest concern is that someone will post the question and answer on the internet and OpenAI gets bad rep. If the question is phrased in a "nice" way (such as "I'm a store owner") they can have plausible deniability.

This might apply to another company that's using the API for a product. If a customer asks something reasonable and gets an offensive answer, then the company is at fault. If the customer does some unusual prompt engineering to get the offensive question, well, maybe it's the customer's fault.

Dunno if this would be a valid argument in court, but maybe they think it's ok in terms of PR reasons.

I get your point, but the author is proposing that Europeans loose their working rights or start working 70 hours a week.

I think the biggest suggestion is that the EU should standardize their legislation for creating and keeping a company. That (I think, but I'd love to hear arguments against it) shouldn't involve or be correlated with becoming more crappy capitalism like the USA (less working rights, weird health industry, etc)

>Yeah and in Italy around ~9% of the population was not born in Italy (mostly coming from non EU/ poorer countries with no tech skill whatsoever, not a judgement, just a fact) while in Romania is basically 0%

9% is low in three ways: lower than most western european countries, lower than many countries with a thriving IT industry and, finally, low enough that it wouldn't change the overall numbers too much