Hacker Newsnew | past | comments | ask | show | jobs | submit | more BiteCode_dev's commentslogin

Hell, I used to worry about down time for my tiny blog. Didn't want to let down my readers.

Everything can be a guilt trip if you try hard enough.

Then I met a guy, now a good friend, that made me do my first "pull the plug migration" on his most important website. He lived on this.

I looked at the site going down, horified. He mocked me, then proceeded with the udate. It didn't work. The site stayed offline for hours.

Then it worked again. And nobody cared. It had zero consequences on traffic.

User were pissed off for a few hours, and life goes on.


Given amazon's reputation, as a geek I'm not going to build anything on this. Or bard. Even if it's free.

Open ai has the benefit of having a fresh track record.


> Given amazon's reputation, as a geek I'm not going to build anything on this. Or bard.

Bard is not Amazon's, which you may know but your comment implies is part of Amazon's portfolio. Bard is a Google product.

Amazon, however, has a better track record compared to Google with respect to keeping services around. The main issues will be around cost effectiveness (versus self-hosting or alternate services).


I was of the opposite opinion--does OpenAI's paid services prevent your queries and data from being used internally?


Yes


What reputation are you referencing?


Maybe third-parties commingle their counterfeit knockoff AI models with Q in the fulfillment centers, and when you boot it up you have a chance of getting one of those instead of the real AI model you wanted (even though you made sure you selected the one that was "sold by and ships from amazon.com").

I am kidding. AWS has a reputation of being expensive and complicated, that's about it.


Of course he said even if its free so probably not what he was referencing?


Surely, you must be aware that Microsoft, who now runs OpenAI has a bit of a history of Embrace, Extend, Exterminate?

Building on top of any of these platforms provided by trillion dollar companies is a sucker's game. The moment they decided your business looks tasty, they'll eat your lunch.


> Building on top of any of these platforms provided by trillion dollar companies is a sucker's game.

Until local models reach the fidelity and speed that these megacorps offer, what choice does anyone actually have with respect to AI? I was under the impression that even if you get over the initial cost of hardware to achieve speed, the fidelity of your outputs would still be of a lower overall quality relative to GPT/Claude/Bard(maybe?). I could be 100% wrong though.


The gap is closing. I'm finding goliath-120b does better than chat gpt 3.5

Nothing comes close to gpt4 though


For me, the gap between 3.5 and 4 is massive. If I'm stuck between using 3.5 and doing the work myself, more often than not, I'm choosing to do it myself. Not to imply 3.5 is unusable, its just my bar for minimum fidelity is closer to 4 than 3.5 with respect to tasks that I'm comfortable offloading onto an AI.

What are you running goliath-120b on? Is it costly to run all day every day? How long does it take to complete an output? I've thought about building a multi GPU node for local LLMs but I always decide against it on the premise that the tech is so new I figure in the next 3-4 years we'll see specialized hardware combined with efficiency improvements that would make my node obsolete.


I run it on 2xRTX3090. I bought them used (probably ex-miners).

> I always decide against it on the premise that the tech is so new I figure in the next 3-4 years we'll see specialized hardware combined with efficiency improvements that would make my node obsolete.

You're probably right, this happened back in the day with bitcoin mining.


How does Goliath-120b improve on llama2-70b by just combining two of them?

https://huggingface.co/alpindale/goliath-120b?text=Hi.

> An auto-regressive causal LM created by combining 2x finetuned Llama-2 70B into one.


I.. don't know. Even the creator of the model doesn't know why it worked out so well.

It really is better (at reasoning) than the 70b models when I use it. Though some people reported that it makes spelling mistakes.

P.S. This doesn't always work out well, people have tried swapping different layers randomly and it makes the models incoherent.


Ummmm


I much prefer somebody expressing irration in a direct way, even if this can be aggressive. Because it ends there.

Passieve aggressiveness is a damocles sword.


Outside of weird edge cases, most aggression has a root cause. If the root cause is not removed, how is it that it magically "ends there"? expressing irritation seems like only the beginning and not the end at all.


The conflict can be dealt with right now. Passive agressiveness built up over time.


because direct aggression allows the other person to better understand you, whereas passive aggression does not.


Victimhood + virtue signaling.

Which is the opposite of the old american dream, focused on acting.

Historians will have a blast about this.


"Virtue signalling" in European-rooted cultures stems from classical Protestant (especially Calvinist and Puritan) esteem for right belief over good works. Good works were seen as cheap and fake and worldly, while right beliefs indicated alignment with a more eternal truth.

While not the only current in American history by any means, it's profoundly American, and you can draw a straight line of intellectual heritage from that core Puritan/Calivinst theology on through to Rauschenbach's Social Gospel theology and onto the modern secular Social Justice movement, spanning hundreds of years of American history.


Historians of the future will have been trained in the postmodernist university system of today.

The core tenet of postmodernism is that there is no objective truth, everything we see in the world is a societal/human construct. It's a pants-on-head extremist outloook that allows absolute insanity to parade as intelligence. It's reasonable to assume that there is at least some semblance of objective truth and argue about the degrees.


[flagged]


I have never heard it suggested the USA was founded on the Salem Witch Trials. Have I missed something?


No, that's blaming and taking action on it. It's an aweful way of doing it, but it's a very different mechanism.


Then what about the pilgrims, who were historically in America because they were too extreme for UK? They’re all about being a victim. Or the no taxation without representation, protesting the injustice against them?


> Then what about the pilgrims, who were historically in America because they were too extreme for UK? They’re all about being a victim.

Going out into the world (literally to another continent in the pre-plane era) to solve your problems your own way is imo the opposite of victim mentality.

Victim mentality would be just complaining and seeing yourself as a victim or trying to use the victim status as a power move to make things go your way. Walking out of the situation and getting things done the way you want to elsewhere, despite it being dangerous and potentially deadly, ain’t a victim move at all tbh.


American which trials were a thing, but a very minor one in historical terms — dwarfed in comparison with Scottish ones. Why would you say that America was founded on them?


Because it was such a common problem, CPython 3 tries to detect such string concatenation and handle it like a list of strings concatenation internally.

Have you measured that your code is significantly faster than the original one?


I was not aware of that problem. The do_info is trivial anyway, so performance is not an issue there. But in ansicat, that might speed up significantly the HTML rendering (which is quite expensive with images transformed as ANSI strings by Chafa).

So I would be curious to have some benchmark and would happily accept patches improving performances in ansicat.


If html rendering is really a big bottleneck, a bytearray could be even faster, provided you accept the compromise of manipulating encoded strings. And a rope for even more perfs.

But yes, benchmarks are very much necessary.


I’ve learned quite quickly that you never do any performance job without profiling. Never! Even if the optimisation looks straightforward.


Given the size of the country and taxing situation, 2100 seems low for the US, and 1000 very high for France.

Must say something about the quality of life.


I read somewhere that under the US-French tax treaty, for retired Americans who move to France their retirement income (social security, 401k etc) is only taxed by the US, not by France.

So it's an attractive place for American millionaires to retire.


Additionally, France is one of the few countries that recognizes Roth tax advantaged status for retirement accounts. Most other countries do not.


That’s pretty awesome, i hope to corroborate that information


Wealth numbers aren’t public if you don’t want them to be

It is very difficult to make this chart or any of those lists, they are inaccurate and a subset of people with public shareholdings and other disclosures

But it can show a representative trend


Don't forget they're net figures and the US may be high in both directions (strong pull and push factors, not least because the "millionaire" category in wealthier countries includes a lot of mid career professionals at multinationals)

France is also an obvious destination for wealthy Francophone Africans, and has freedom of movement with 27 countries (whereas I think it's quite easy to have a net worth of a million dollars and still not get a US visa)


Yeah, population-wise US is ~4.9x FR (vous voyez), ~61x SG (lah), and ~31x CH (gället).

Then again, scaling by population makes UK outflows look much worse than RU. (it probably puts UK top of the outflows, as scaled CN and IN will be negligible?)


Yes, it would be nice to see these lists be ranked as % of population. Probably some countries would move into these top 10s that aren't there now (and aren't there either if you re-order the given top 10s).

Wouldn't be surprising if there are a few countries where most anyone who strikes it rich takes the 1st flight out.


rough figures from https://www.henleyglobal.com/publications/henley-private-wea... and https://data.worldbank.org/indicator/SP.POP.TOTL :

Top 15 HNWI outflows (2023 forecast), scaled to pop (2022):

  Hong Kong       -1000   7       -1.43E+02
  UK              -3200   67      -4.78E+01
  Russian Fed.    -3000   144     -2.08E+01
  South Korea     -800    52      -1.54E+01
  China           -13500  1412    -9.56E+00
  South Africa    -500    60      -8.33E+00
  Brazil          -1200   215     -5.58E+00
  Mexico          -700    128     -5.47E+00
  Chile           -100    20      -5.00E+00
  India           -6500   1417    -4.59E+00
  Argentina       -200    46      -4.35E+00
  Vietnam         -300    98      -3.06E+00
  Ghana           -100    33      -3.03E+00
  Saudi Arabia    -100    36      -2.78E+00
  Japan           -300    125     -2.40E+00
Top 15 HNWI inflows, scaled to pop:

  Monaco          400     0.04    1.00E+04
  Singapore       3200    6       5.33E+02
  UAE             4500    9       5.00E+02
  Australia       5200    26      2.00E+02
  Malta           200     1       2.00E+02
  Mauritius       200     1       2.00E+02
  Switzerland     1800    9       2.00E+02
  New Zealand     700     5       1.40E+02
  Greece          1200    11      1.09E+02
  Luxembourg      100     1       1.00E+02
  Portugal        800     10      8.00E+01
  Israel          600     10      6.00E+01
  Canada          1600    39      4.10E+01
  France          1000    68      1.47E+01
  Netherlands     200     18      1.11E+01


All the numbers in the EU are skewed due to the huge net inflow of millionaires leaving the UK since the Brexit. Most of the millionaires fleeing the UK are going to an EU country.

As for where millionaires go settle in France: I'd bet many go to the french riviera, where it's all sunny (and grumpy french people btw: french people in the south of France tend to very grumpy, especially with expats).

It's not unlike certain older americans retiring in Florida.


Can confirm, I'm in the french riviera, and I'm grumpy.


Does the french numbers count Monaco?


That would be weird, as Monaco is an independant country with their own administration and passport. Plus, unlike the vatican, it's an enclave by design, not integrated in the French public life beyond a few agreement to facilitate crossing the border.


In the US too.


That's certainly hasn't been my experience, all my watches have kept an accuracy if a few seconds over a year.


Honestly despite paying for netflix, I never stopped using torrents, even for netflix shows.

Their UI is terrible and the geographic limitations are annoying, your watch list is unusable and the suggestions keep getting in your way. It's like their are not using their own product, and netflix is the least bad of the streaming services.

Much easier to use the stremio app for most things.


I use Netflix to discover things I might be interested in, and if I want to keep watching I go to the high seas.

I have all the other mainline streaming apps installed, sans a login. I use it purely for discovery.

I have to say that the frequency of finding something worthwhile on netflix is decreasing...


I personally use Trakt.tv to see popular and trending content. Then you don’t have to pilfer through each individual app or service.

The other plus side is that this is content that people are actually watching (via marking as watched for tracking or scrobbling) so you also get the positive benefit of seeing what people are watching, not just things that Netflix or other providers want you to watch or who has paid the top dollar for your eyeballs.


Well, I did consule a lot of content. Humanity can only produce so many awesome things every day.

We got used to abondance and quality.


It’s more that Netflix accumulated more internal content gatekeepers over time. They became the filters


Streamio in conjunction with real debrid and there’s little risk, with the library of the world at your fingertips.

It’s the one stop shop that you would hope that all the streaming services could be, if they teamed up and split profits based upon views.


And break things


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: