I found this project, but I’m not surprised that it doesn’t have much progress. Doing this kind of Rust rewrite is an absolutely gargantuan undertaking that I don’t think it would be feasible without AI. It’s not as if FFmpeg is some crufty, slow code that would seriously benefit from it.
Because it can't. I'm aware of only two projects that have attempted this.
1. Anthropic's C compiler. This is a buggy, unoptimized, unwieldy mess. One that can compile the Linux kernel, which is no mean feat. But it is not "fully featured." And it had an incredible array of tests to start with.
2. Cursor's web browser. This is, flatly, a trainwreck. It's built on top of massive dependencies that do incredible amounts of heavy lifting despite claims to the contrary, and never seems to have ever produced a successful build.
The tests are the real clincher here. That’s the main reason that this project has made so much progress: the FATE suite highlights specific bugs to fix and tracks regressions.
As for optimization, that seems to be more of a question of effort than whether it’s possible. I was able to take down the performance gap on Rust vs C (without Assembly) from 10x to 1.5x through detailed profiling and iterative improvements with Claude.
It also looks like the Anthropic C compiler was built from scratch. By contrast, `wedeo` was directly based on FFmpeg’s existing code. Going by spec and test suite only would have taken a lot longer, and the quality would have been significantly lower.
Anthropic's C compiler also had an incredible test suite available, and was trained on compiler codebases. It might have been built "from scratch" but compilers are incredibly well-trodden ground.
And to contrast, Cursor's browser explicitly reviewed Servo's architecture before setting out, and still wound up like that.
To be clear, `wedeo` took a few weeks to get to this point, and it is by no means fully featured. Even with LLMs, it will take a few months to reach that point with me as the sole human contributor.
If you're fine with a paid option (although it's source-available and distributed for "free" on their website), then Grayjay [0] is my personal favourite. I can't remember the last time I saw an error. On NewPipe, I can't say the same.
FWIW the second part does not address my concern (which was about omitting literal parentheses that are present in the markup, in some cases), though it is related. Personally I think it's already a mistake for bare/unescaped words to indicate markup, or I guess the intention is that they indicate mathematical semantics and then Typst decides the best way to write down that semantics unless you explicitly interfere -- seems like a recipe for confusion whenever Typst's ideas of the mapping from semantics to typography don't quite line up with yours, or the conventions in your field. I'd much rather approach mathematical typesetting with a typesetting language than a mathematical language where somebody else has decided how that math should be typeset, even if it means a sea of backslashes and curly braces.
(Edit to add: I guess this means my preference would be option E from the blog post -- I see the extra symbols as far preferable to the ambiguity between content and markup)
Also wow, recovering from this breaking change is going to be incredibly painful for somebody.
I think social biases (e.g. angry black women stereotype) in your paper is different from cognitive biases about facts (e.g. number of legs, whether lines are parallel) that OP is about.
As far as the model's concerned, there's not much difference. Social biases will tend to show up objectively in the training data because the training data is influenced by those biases (the same thing happens with humans, which how these biases can proliferate and persist).
> a bit like trying to explain a vacuum cleaner to someone who has never seen one, except you're only allowed to use words that are four letters long or shorter.
> What can you say?
> "It is a tool that does suck up dust to make what you walk on in a home tidy."
I liked that the original explained the value of the vacuum cleaner. It's not that it removes dirt and dust per se, it's that it makes spaces you walk on tidier.
Oh come on, this shit is easy. Why did they say "it is" and not "it's", by the way? To put it that way can't help. So yeah, it's a pipe that can suck, and you push it all over your room, to suck the dust and dirt up off the rugs and such, and in fact off of any low down flat part. One kind can even move on its own! But what I want to say here, in the main, is that you math guys have all lost your grip on how to say any idea in an easy form. You are not able to do it any more, 'cos too much math has made you sick in the head.
I feel like there's still room to avoid pidgin while making it less awkward, e.g.: "It's a tool that can suck up dust or dirt to make your home more tidy."
> Cost of goods sold (COGS) refers to the direct costs of producing the goods sold by a company. This amount includes the cost of the materials and labor directly used to create the good. It excludes indirect expenses, such as distribution costs and sales force costs.
So the $550 or $650 COGS includes the cost of labor for manufacturing, but excludes (say) marketing and auditing costs.
Right, but this is the same company, so the cost of marketing, auditing, R&D, etc. shouldn't be different for these products. That's a fixed cost for the company.
This is a guess, but the argument is probably that it took way more R&D effort for them to figure out how to produce it efficiently in the US, and they've chosen to increase the cost of the US phone variant to offset this particular R&D expenditure that the Chinese variant didn't have.
> So it's about $650 to produce that entire phone. But what we're doing by selling it for greater originally, we're looking at a lot of differentiators for us. It wasn't just made in the USA. It's the fact that it's a secure supply chain, that you know, staff that's completely auditing every component, which means we're selling to a government security market with all those additional layers that we've added on top.
So I guess the answer is that they're selling to the "government security market" so they can charge whatever the hell they want.
> Chinese newspaper Economic Observer earlier reported that local traffic police had told the father of one of the victims that the car had caught fire after hitting the cement pole, and the car key had not unlocked the door.
I’m not talking about the Feb number that is reported in March.
reply