shikon7

But 2 mod 3 should still give the same as -1 mod 3. If the representative for the same equivalence class is allowed to be different depending on the input, you could just as well use the number itself.

The RTX 3090 is still one of the best gaming/consumer GPUs, beat only by the 4080 and 4090. And the 5000 series might still be almost a year away.

But this also means that possible information retained is constant irrespective of the prefix length. This might be a problem if the prefix is composed of essentially uncompressable data.

You use Rust if you want performance (especially because there is no garbage collection), and strong safety guarantees.

If you don’t care about safety guarantees and abstractions like shared_ptr, you might just as well use C instead of C++.

But if AI becomes the primary way to learn, how will the AI learn new things? Everything AI has learned about the outside world has to come from somewhere else.

If they are popular and sell well, it‘s only natural they‘re still being made. I wouldn‘t consider that to be insane.

No wonder we can‘t control drug smuggling, if even smuggling sand is lucrative.

> It's a research project into the nature of long-term memory.

> Mnemosyne uses a sophisticated algorithm to schedule the best time for a card to come up for review.

Are there results or personal experiences if this is better than e.g. the Anki scheduler?

If he starts with “As an AI language model…” then you should really start to worry

It seems to me that OpenAI’s vision of an AGI doesn’t include romantic companionship

Why would minors be required to take part in this? To access age-restricted material suitable to them (not 18+ but maybe 13+), or would this registration be mandatory for everyone?

You only need to convince the panic monster. The monkey will be to afraid to do anything.

Even then, either you discriminate, or you don’t.

And they’re not really the worst one could think of, as FAANG thenselves contribute a lot to open source software.

I wonder about that too. With the small precision, parameter updates might be too small to have an effect (is it possible to use some sort of probabilistic update in that case?) Unfortunately, I haven’t found any resources describing the feasibility of full fp16 or bf16 training.

Seems "What doesn't kill you makes you stronger" might just be survivorship bias here.

Or qadruple nozzle, as you need to do both for the anode and the cathode fluid.

There is some about that in point 8, starting with

> This all seems harsh and negative but it can be a fantastic ride (even if it’s taxing).

Still, free speech is what allows such a project. And Wikipedia also explains philosophical opinions, political positions, NSFW concepts, etc., which is only possible to that extent due to free speech.

So I would disagree both that Wikipedia has nothing to do with free speech, and that it’s not about anybody’s opinions or ideas.

If that happens, that’s rather the fault of the company and the managers of not communicating the current affairs clearly enough. You shouldn’t need to rely on informal channels to find out what the company considers canceled.

It does for rockets with many stages, as the lower stages will get an exponential benefit.

> Automatically or programmatically extract data or Output (defined below).

What does that mean? Isn’t that just the definition of API use?

Suppose you’re a serious hobbyist, wanting to buy the most powerful consumer device for deep learning. Then the choice of buying a RTX 4090 or a M3 Mac is probably quite interesting.

Personally, I have already a RTX 3090, so for me it’s interesting if a M3 would be a noticeable upgrade (considering RAM, speed, library support)

Sure you can use safe C++ with some effort, but the libraries you use most likely still use unsafe C/C++. For Rust, I expect that the libraries are much safer in general.

Does it also work with CUDA memory?

Implied is literally the opposite of explicit