tzekid

I think it's because go's community sticks close to the standard library:

e.g. iirc. Rust has multiple ways of handling Strings while Go has (to a big extent) only one (thanks to the GC)

In modern societies: top 1% of earners pay roughly 30% of taxes top 5% pay 65% of taxes top 10% pay 80% of taxes while bottom 50% usually barely make 2% of taxes.

Heavy redistribution of wealth is already in place and it's not making things better.

Forcing Apple to implement interop and such extra features requires a non-trivial amount of labour. That trickles down to the customer in many subtle and not so subtle ways that the parent comment mentioned (more resources and higher inter-dept. coordination ceiling for a streamlined UX, etc.)

Comparing it to opinions about gay marriage makes you look dishonest, "None of this is about you, at all" makes you look plain silly

It doesn't make me happy. The more the EU regulates, the less innovation we have.

It's not a meme, it's reality

They are offering a product/service, so going in too much detail would be a bad business practice, no?

But true, would love to see this in OpenSource in the wild

Isn't it more like "compressing" interactions with and patterns of 500MB of data?

Like, it's not only text search but also has a degree of generalization (e.g. in labelling new documents)

Following a car in front of you in perfect conditions (otherwise it full-on disengages) just to comply with the regulatory labels is whack.

I don't get why people cheer on it and even compare it with Tesla, which goes for ... actual self driving

That's actually awesome tbh.

Wonder what effect alignment training will have on the output quality

Isn't using compounding inflation rate on top of CPI-inflated dollars double counting? It looks and feels like 40+% to me and everybody else

Those are just the countries where you can release an AI product without worrying that you'll get sued into oblivion by the state.

See e.g. EU AI regulation

That's definitely an awesome mindset.

What helped you land your current job more? The network you already had or the cold messages?

I think it's meant as an add-on to stand out when applying to a job.

I'm quite Junior so I don't know if this applies to more senior positions, but in my experience when applying to jobs my Github page was always a point of discussion when I got an interview.

Could be they offload the computing to something like REPL or another third-party on the backend, or?

That's disingenuous of you. This article sparks conversations about workplace boundaries and corpo vs personal ethics and rights, as can be clearly seen by the other comments.

Use common sense. "[..] blood on your hands [..]" implies he's indirectly a murderer at least. Calling someone a murderer is in the best case defamation or an insult.

Not quite reading material, but look up George Hotz's videos about it. I think there's one Twitter Space he hosted / recorded where Musk was also in there.

Hotz is very vocal and blunt about the messy codebase situation at Twitter.

But they were. Weavers improved on their processes for a long time before the engineers swooped in and put that accumulated knowledge into an automated form.

Something like: M1 MBA, around 12 FPS Ultra settings or high 30s on Medium at 1440p ( max res. apparently )

Yep, but an avg. RTX 2060 laptop consumes 3x - 4x the wattage of an M2, no?

The suspects were not anarchists ... they were allegedly connected to a Kurdish militant group (so more like nationalists).

I hate gov.s as much as the next guy, but the rant is off-topic here.

The main problem is that they try to formulate simple encryption in a criminalizing way, which can lead to bigger problems if it becomes a widely accepted view.

Binary thinking is a too-common thing.

It's engineering after all, isn't it?

Here's the full talk[0] from a Microsoft lead researcher that worked with the early / uncensored version of GPT-4.

Simplified, tuning it for censorship heavily limits the dimensionality the model can move in to find an answer which means worse results in general for some reason.

[0] https://www.youtube.com/watch?v=qbIk7-JPB2c

Higher living standards are directly correlated with higher energy and resource consumption.

Unless you're living in a far distant village or in the middle of the woods producing everything you use, you're using an immense amount of resources in your everyday life.

Recycling, not buying the latest tech and curbing your diet don't make any significant dent into the amount of resources you're indirectly consuming by simply living and being active in a well-off, first-word or even "second-world" society.

Clean data. A bunch of data points that are in a good enough state / structured to just throw into the training / eval makes a bigger difference than a bazillion messy data points.

I can easily imagine people in charge with the mentality of "there's no way that anyone can prove we did it."

It's very improbable, but looking at the "AI integration / product" race it is still a non-zero chance it could have happened.

Damn, comparing people who are intrigued by topics deemed taboo by certain moral standards to cancer is ... quite something else.

How about putting up a curated model for the public which is easy to access and a less curated / more free model behind the API with a bunch of boolean switches. Or any other impulse / idea that doesn't label (any) people as cancer.

If I understood correctly, the OpenAssistant team wants to open-source their community built RLHF dataset.

On the other hand, if you're being cheeky, I bet there's a way to datamine from websites like ShareGPT and profit off shared ChatGPT <> User interactions.

I do mostly Python on my dayjob, but for low-level side-projects I've gotta say C++ with the C++17 or C++20 standard it's way faster to iterate with than say Rust or even something like Zig.

For me iteration speed's a big selling point that (plus the fact that's easier to find contributors) might also be important for projects like these.

Don't downplay yourself like that. A 25k token GPT model is a step in that direction but we're 3 - 15 years from your vision of valuable.

Currently, the Language Models (GPT or GithubCopilot) are mostly good as ... copilots. You bounce off ideas off them or use them to kickstart something you want to do. It's great for junior devs (me!) but it still cannot do big context or cognition (what you describe), that part will be done by humans for quite a while.

ChatGPT plugins are in alpha as far as I know. 12 months seems like a reasonable timeframe.

I can totally see a set of plugins that write, debug and test code — all steps a human must now do — all automatically in the next 12 months. That, plus the GPT4-25k token model and you've got something that can replace high-level, high-boilerplate jobs or at least allow one person to 10x their output.

I don't get why this comment is downvoted. Basically this.

A halving of the costs every year or so seems realistic in this emerging phase.