This is how all companies calculate gross margins, which is why they're called gross. Gross profit is revenue minus the cost of goods sold. For example, if you sell wheat for $1 and spend $0.20 of labor and raw…
I think they should release metrics for "how much of this is LLM-written" (the raw test statistic, not including length) along with detection.
Because it's pure exploit on the explore/exploit tradeoff. The best outputs come when the LLM comes up with lots of different ideas, considers them, and selects the best one. If you sample at low temperature it tends to…
I don't think Google gets a free pass--kind of the opposite, it gets the lion's share of the criticism (especially for privacy stuff) despite being the most open and up front about what they're doing, since they open…
Yes, Chromebooks are easy to administer but very limited (you can't run much outside the browser). It sounds like Google's trying to fix that and make a full Desktop OS.
It’s bad that this is against the TOS in the first place, and reeks of anticompetitive behavior. Why does Anthropic care what frontend I use as long as I pay for their model?
But why do they want to control that? It’s none of their business.
> 1) Apple loves USB-C Sure, that's why they refused to adopt it for almost a decade after it became the standard and fought the EU regulation tooth-and-nail.
> I don’t think one company should own all the stuff that Google does. It gives them way too many perverse incentives over the web. Does it? It might give them perverse incentives in some cases, but in others it…
It's amazing how you can literally start a nonprofit to code a billion-dollar browser, give it away for free, and let people modify it however they want and then HN users will still find a way to act like this is being…
Aluminum isn't an OS, it's a project to replace ChromeOS with Android.
You still can.
>Chrome telemetry underreports a lot of use cases Sure; in that case, I would suggest to the people with those use cases that they should stop switching off telemetry. Everyone on HN seems to forget telemetry isn't…
That's why I use C to write all my shell scripts. Can't have something non-portable like bash, which is missing from 75% of computers by default. (Seriously, what am I missing here? Is `apt-brew pacman install fish`…
You shouldn't be writing serious programs in Bash/Zsh at all, is the issue. If it's complex enough to need linting, it's complex enough to require Python/Perl.
I'mma be honest, I think the API is great—EXCEPT THEY STILL DON'T HAVE LOGPROBS. ARGH.
> The Mojo code has to run the CPython interpreter to run numpy. Yes, the need to run CPython interpreter is what makes Mojo slow (and it will remain that way, unless they abandon their "superset of Python" promise).
13 million is substantial but not even close to the 100 million Modular got. Which really makes me wonder what Modular has been doing with that money, if they're still getting beat in benchmarks like this...
That's not a fundamental restriction or design problem with Julia, you're just describing JIT compilation. Plenty of languages can be compiled either ahead-of-time or just-in-time. Actually, this should be unusually…
Not necessarily. Julia will always be 11 years older than Mojo, no matter how old both of them get, and that advantage won't shrink. Not to mention, Mojo is a superset of a 40-year old language with billions of dollars…
Are the first 3 causes dying, aging, and poor health? As soon as you start allowing abstract concepts that contribute to a person's death to count as the cause of a person's death, this stops being an especially…
If nothing else, you can just pipe the code and then write comments explaining what's left after each step. But the verbose code can be substantially slower (which happens when piping can be used to perform all these…
In ML? No, the best competition for Python in ML is... well, it's either C++ or Julia, depending on how you define "competition," given Python is effectively a glorified C++ interface.
In that case I'd suggest Julia; it's got syntax pretty much identical to R, and is just as easy to pick up, but it's much, much faster, and it works as a general-ish computing language. (It's still definitely…
> For example, error messages thrown by Julia can be multiple screens long and are as hard to read as C++ errors. The good news is this can be fixed by using AbreviatedStackTraces.jl, which will probably be added to the…
This is how all companies calculate gross margins, which is why they're called gross. Gross profit is revenue minus the cost of goods sold. For example, if you sell wheat for $1 and spend $0.20 of labor and raw…
I think they should release metrics for "how much of this is LLM-written" (the raw test statistic, not including length) along with detection.
Because it's pure exploit on the explore/exploit tradeoff. The best outputs come when the LLM comes up with lots of different ideas, considers them, and selects the best one. If you sample at low temperature it tends to…
I don't think Google gets a free pass--kind of the opposite, it gets the lion's share of the criticism (especially for privacy stuff) despite being the most open and up front about what they're doing, since they open…
Yes, Chromebooks are easy to administer but very limited (you can't run much outside the browser). It sounds like Google's trying to fix that and make a full Desktop OS.
It’s bad that this is against the TOS in the first place, and reeks of anticompetitive behavior. Why does Anthropic care what frontend I use as long as I pay for their model?
But why do they want to control that? It’s none of their business.
> 1) Apple loves USB-C Sure, that's why they refused to adopt it for almost a decade after it became the standard and fought the EU regulation tooth-and-nail.
> I don’t think one company should own all the stuff that Google does. It gives them way too many perverse incentives over the web. Does it? It might give them perverse incentives in some cases, but in others it…
It's amazing how you can literally start a nonprofit to code a billion-dollar browser, give it away for free, and let people modify it however they want and then HN users will still find a way to act like this is being…
Aluminum isn't an OS, it's a project to replace ChromeOS with Android.
You still can.
>Chrome telemetry underreports a lot of use cases Sure; in that case, I would suggest to the people with those use cases that they should stop switching off telemetry. Everyone on HN seems to forget telemetry isn't…
That's why I use C to write all my shell scripts. Can't have something non-portable like bash, which is missing from 75% of computers by default. (Seriously, what am I missing here? Is `apt-brew pacman install fish`…
You shouldn't be writing serious programs in Bash/Zsh at all, is the issue. If it's complex enough to need linting, it's complex enough to require Python/Perl.
I'mma be honest, I think the API is great—EXCEPT THEY STILL DON'T HAVE LOGPROBS. ARGH.
> The Mojo code has to run the CPython interpreter to run numpy. Yes, the need to run CPython interpreter is what makes Mojo slow (and it will remain that way, unless they abandon their "superset of Python" promise).
13 million is substantial but not even close to the 100 million Modular got. Which really makes me wonder what Modular has been doing with that money, if they're still getting beat in benchmarks like this...
That's not a fundamental restriction or design problem with Julia, you're just describing JIT compilation. Plenty of languages can be compiled either ahead-of-time or just-in-time. Actually, this should be unusually…
Not necessarily. Julia will always be 11 years older than Mojo, no matter how old both of them get, and that advantage won't shrink. Not to mention, Mojo is a superset of a 40-year old language with billions of dollars…
Are the first 3 causes dying, aging, and poor health? As soon as you start allowing abstract concepts that contribute to a person's death to count as the cause of a person's death, this stops being an especially…
If nothing else, you can just pipe the code and then write comments explaining what's left after each step. But the verbose code can be substantially slower (which happens when piping can be used to perform all these…
In ML? No, the best competition for Python in ML is... well, it's either C++ or Julia, depending on how you define "competition," given Python is effectively a glorified C++ interface.
In that case I'd suggest Julia; it's got syntax pretty much identical to R, and is just as easy to pick up, but it's much, much faster, and it works as a general-ish computing language. (It's still definitely…
> For example, error messages thrown by Julia can be multiple screens long and are as hard to read as C++ errors. The good news is this can be fixed by using AbreviatedStackTraces.jl, which will probably be added to the…