One thing that makes me unsure about this proposal is the silent downgrading on unsupported platforms. People might think they're safe when they're not.
Go has the best support for cryptography of any language
I'm not sure there's a realistic alternative. If you need to generate a key then it has to happen somehow on unsupported platforms. You can check Enabled() if you need to know and intend to do something different but I assume most of the time you run the same function either way you'd just prefer to opt into secret mode if it's available.
Not OP, but Go has some major advantages in cryptography:
1. Well-supported standard libraries generally written by Google
2. Major projects like Vault and K8s that use those implementations and publish new stuff
3. Primary client language for many blockchains, bringing cryptography contributions from the likes of Ethereum Foundation, Tendermint, Algorand, ZK rollups, etc
I want something with a simpler backend than immich. I don't really want to host it because it needs lots of stuff to run. I would love one that can do sqlite and is a single binary go (or rust) program.
It auto uploads all your photos to the cloud and you can delete them locally and still have them. The biggest feature is the AI search, you can type anything and it will find your pictures without you doing any work categorizing them. It can do objects or backgrounds or colors and it can even do faces so you can search by people's name. That and there's share links to albums and multiplayer albums.
It keeps the originals locally when it uploads forever unless you delete them. There's a one click "free up space on this device" button to delete the local files. It's actually somewhat annoying to export in bulk, you pretty much have to use takeout.
What a disgusting article. It's abliest to say that disabled students won't be able to make it Stanford. The only weird part is calling anxiety and depression a disability.
Saying that people who are using accommodations are cheating is morally repugnant.
Instead of saying that we need to clamp down on people claiming disabilities, we should open up the accommodations to everyone.
It's not quite that they cannot do anything not in the training data. They can also interpolate the training data. They're just fairly bad at extrapolating.
Physics is obviously incomplete and yet nobody can solve quantum gravity. Being obviously flawed doesn't mean the solution is obvious. That's the whole problem.
I think in this case, people tend to underrate just how capable and flexible the basic LLM architecture is. And, also, underrate how many gains are there in better training vs better architecture.
I'm hosting from my home with a static ipv4 right now. It's been running for years without a single problem. I just put in a basic pf config. Everything is fine. It's not that scary.
has 691 lines. I expect it would work, as FAWK seems to be a very simple language. I'm currently working on a similar project with a different language, and the equivalent AST module is around 20,000 lines and only partially implemented according to the standard. I have tried to use LLMs without any luck. I think in addition to the language size, something they currently fail at seems to be, for lack of a better description, "understanding the propagation of changes across a complex codebase where the combinatoric space of behavioral effects of any given change is massive". When I ask Claude to help in the codebase I'm working in, it starts making edits and going down paths I know are dead ends, and I end up having to spend way more time explaining why things wouldn't work to it, than if I had just implemented it myself...
We seem to be moving in the right direction, but I think absent a fundamental change in model architecture we're going to end up with models that consume gigawatts to do what a brain can do for 20 watts. Maybe a metaphorical pointer to the underlying issue, whatever it is, is that if a human sits down and works on a problem for 10 hours, they will be fundamentally closer to having solved the problem (deeper understanding of the problem space), whereas if you throw 10 hours worth of human or LLM generated context into an LLM and ask it to work on the problem, it will perform significantly worse than if it had no context, as context rot (sparse training data for the "area" of the latent space associated with the prior sequence of tokens) will degrade its performance. The exception would be like, when the prior context is documentation for how to solve the problem, in which case the LLM would perform better, but also the problem was already solved. I mention that case because I imagine it would be easy to game a benchmark that intends to test this, without actually solving the underlying problem of building a system that can dynamically create arbitrary novel representations of the world around it and use those to make predictions and solve problems.
Go has the best support for cryptography of any language
reply