Max

That’s kinda an insane amount of ram for most simulations. Is this like a machine learning thing? Is his python code just super unoptimized? Is it possible he’s making a bunch of big objects and then not freeing the references when he’s done with them so they’re never garbage collected?

It fulfills a different purpose than system packages. First, it can be run without privileges/system modification, so it works on immutable distributions. Second, it doesn’t share libraries between apps (with some exceptions) or the system, so you don’t have to package separately for each. It essentially takes some of the container philosophy/tech and brings it to desktop apps. This also gives it some ability to do some sandboxing that isn’t as easy with system installed apps.

This approach comes with some downsides. Particularly larger storage requirement for apps, sometimes less integration with the system, and lack of ability for apps to easily call/interact each other unless they’re packaged together.

It’s meant for complete GUI apps and not small tools/packages that are the standard in system package managers

In that case it’s highly unlikely your problem is with DNS. And much more likely it’s a problem with the actual connection to the server. If you are willing to share the IP/domain I can help troubleshoot (either here or in a DM).

If you do a DNS lookup (through nslookup or many other tools) on the client you’re using to connect, does it get the right IP back?

Sorry I completely misread your comment to be saying that the maximum efficiency was 50% not that it occurred at 50%.

I believe for the highest efficiency you only want to use about half of the rated power of the PSU. So if your system draws 350W, 700 is a very reasonable power supply

Shouldn’t it be the median person?

This is a really good question!

I believe the general answer is, until the compressed file is indistinguishable from randomness. At that point there is no more redundant information left to compress. Like you said, the ‘information content’ of a message can be measured.

(Note that there are ways to get a file to look like randomness that don’t compress it)

By default, an enencrypted boot drive is not sufficient to be able to decrypt a LUKs drive. If you have to type in your password to start the computer/unlock LUKs then you should be good.

If you’ve setup a keyfile or TPM based decryption of LUKS, then your data is probably not safe (though a TPM based decryption could be if the OS is secure and secure boot is setup properly)

In this case, if you have another server then you could setup a mutual tang/clevis system where each device gets the keys it needs from the other server on the LAN. Both would be LUKs encrypted. So if one is online the other gets the required key from the online one while booting. But if both are offline then no keys are available and you have to type in a LUKS password to boot. Something like https://www.ogselfhosting.com/index.php/2023/12/25/tang-clevis-for-a-luks-encrypted-debian-server/ but what they do with multiple servers is probably overkill

That’s probably only for selling steam keys on another store. You might be able to sell non steam versions for any price you want

The original is something like “it’s all Ohio” “always has been”. The globe shown is just a large landmass that looks like Ohio. They’re probably astronauts since you’d have to be in space to see it. The astronaut killing the other has the Ohio flag on them and is killing them for knowing too much.

If there’s a 0 day in the VPN software then I’m also probably boned. The chances of that seem on par with the likelihood of an openssh vulnerability? I feel like vpns are useful to secure services without good authentication, but their use in front of an openssh server has never made much sense to me.

Reminds me of https://www.timandraka.com/

I think point two may be wrong. The strength of a shell should be proportional to its thickness, which would scale linearly with its size (assuming the shell got thicker in proportion to the size). There’s definitely a point where a self supporting egg requires very thick shells like you said, but the scaling law you gave uses the wrong change.

I have like 80gb of photos and videos on my phone, which is the vast majority of my storage space. I think the second category is signal’s database, also mostly consumed by sent media

Are there any two thoughts that are distinct from each other. Or is there only 1 unique thought (Choose any and then all others are a variation of it).

Also I disagree that thoughts and emotions are entirely distinct. Or does ‘thoughts’ refer just to language? Are visual thoughts thoughts? And if so, why not remembering the experience of pain?

Are memories thoughts? Or do I have to be commenting on the memory with language for it to be a thought? I feel like memories are 100% original too, since they’re a re-experience of something that happened to you, not based on anything that someone else has previously thought.

I would ask: is it possible for two thoughts to be completely distinct from each other (according to however you’re defining original). If no, then by definition only your first thought is original because after that all thoughts can be thought of as a variations on that thought (and you said variations weren’t allowed)

I think your definition of 100% original is so restrictive that it kinda loses all meaning.

To actually answer, I think emotional reactions are some of the most original thoughts that I have. Like the experience of pain is original even if you’ve heard words describing it before. And if it’s not original, then it’s not original only to your own previous experience. In fact, the experience of having all thoughts/sensations is original, even if some sense of the meaning of that thought is not.

This is what I’ve been going through, sold as teaching rust to people who already know other languages. I’m not very far in at all, but it seems decent? https://google.github.io/comprehensive-rust/

This is a really fantastic explanation of the issue!

It’s more like improv comedy with an extremely adaptable comic than a conversation with a real person.

One of the things that I’ve noticed is that the training/finetuning that’s done in order to make it give good completions to the “helpful ai conversation scenario” is that it flattens a lot of the capabilities of the underlying language model for really interesting and specific completions. I remember playing around with gpt2 in it’s native text completion mode, and even with that much weaker model, it was able to complete a much larger variety of text styles without sliding into the sameness and slickness of the current chat model fine-tuning.

A lot of the research that I read on LLMs is using them in the original token completion context, but pretty much the only way people interact with them is through a thick layer of ai chatbot improv. As an example for code, I imagine that one would have more success using an LLM to edit your code if the context that you give it starts out written like it is a review of a pull request for the code, or some other commentary of a form that matches the way that code is reviewed in the training data. But instead of having access to create that context directly, we have to ask for code review through the fogged window of a chat between an AI assistant and a person discussing code. And that form of chat likely isn’t well represented in the training data.

What about solvespace web build: https://github.com/solvespace/solvespace/tree/emscripten#building-for-web