Ericson2314

That's an artifact of where you live haha

RL robotics is increasingly something that cannot be avoided, I think.

To everyone skeptical of people saying that it's better to separate CPU and GPU code into separate files, riddle me this:

With this system, how do we specify whether dependencies are needed to be compiled for the CPU, GPU, or both?

----

Technically I don't know really care whether it's multiple files or one, I just want to make sure we are not reinventing a shittier version of CFG. What they are providing looks to me like:

1. explicitly annotate some things as `cfg(GPU)` or ungated (both)

2. unlabeled means annotated `cfg(not(GPU))` by definitely

Put this way, this has nothing to do with GPUs, and just has to do with creating some crate-local CFG shorthanded. Great! Let's do that first, get a solid foundation, and then come back to whatever is remaining for CUDA Rust.

The problem is not the sex, the problem is the cult. The sex is merely evidence for the cult.

My perspective using Nix is that it's hardly worth standardizing ISAs. Recompiling software is easy and we should have a Cambrian explosion of different designs.

This is especially good if one scopes out a family of ISAs that are ABI compatible such that one can compile down to a semi-pre-optimized portable IR, and just do the last bit per ISA.

Sure.

Nixpkgs cross support for Windows has been in there a while. Native windows support for Nix is a bit less advertised because it is pre-release/pre-MVP-on-master, but see https://github.com/nix-windows/nix-windows-demo . And expect to hear more on this in the next few weeks :).

I totally agree with you on the first bit, but I also think that I am way better at deciding on how to refactor code bases than the LLM is.

Right now, I put models in low thinking mode during my refactors and hate waiting. I would much rather have a faster model that that maybe was slightly stupider, and I would wait far less long between prompts where it needs my valuable input.

Models that are dumb, but humble and fast, can be fine.

If your task has incremental rewards/feedback, you can push the "intelligence rate" simply by sampling the reward function faster. That's not fake, even if it not a substitute either.

This is the "dumber but honest person that works harder" phenomenon, vs "lazy genius".

> PS: If anyone managed to combine Clang-based MinGW with AddressSanitizer and Wine without crashing at launch, please show me how and drop me an e-mail. Thank you!

It's for problems like these that I have been working at cross and native Windows support in the Nix ecosystem

No this isn't mindless reinventing --- this is finally having clear empiric evidence for something that we knew the entire time. This is a huge relief!

Next up is demonstrating the AI is more productive with better programming languages.

> While using floating point?

Where is this meme coming from? IEEE floating point is deterministic

Skill issue

Deterministic seed

Yeah maybe *both" sass and training companies are wiped, imagine that!

Good job, but the parent fork already did macOS, so this wasn't too hard.

Well put. That's exactly right. (And I don't even mind mouse studies!)

This neural network stuff does give me real divination vibes, not gonna lie.

There probably is no substitute for the state paying for it. We just need to balloon NPR/PBS into BBC scale (relative to the size of the US, so even bigger) and hope for the best.

Yeah yeah, there are plenty of ways this could go wrong, but those pitfalls are not guaranteed, whereas the ways private sector journalism can go wrong have occurred in practice. So we might as well try it.

Oxford Brain

That's like saying "the US Constitution has clearly shown it's a model that is built to last"

Sometimes bad designs stick around due to pure inertia

No, a layperson doing a bunch of math but barking up the wrong tree theory-wise is actually super instructive for this forum of autodidacts.

And I say that as one of the autodidacts.

As long as Apple doesn't change the license, I think this means we can make legal semi-hackintoshes good enough for CI purposes. That could be a real game-changer!

Do you think it will be possible to use them together? Having some sort of unified distributed system is intriguing to me. (e.g. can the Radical foundation and AT-proto foundation integrate, even?)

Stocks vs Flows! You can't compare (as in subtract and check sign) $ and $/s!

ISPs almost all offer IPv6. The reason the US number is not higher is lazy corporate networks.

I remember as a kid wondering if we could give humans chlorolaplasts.

Sapling and JJ can sort it out, the outside world will only care for one of them.

Good stuff. But I would blame the Trump admin more then data centers for coal power plants staying on line. Gas would substitute for the coal ata minimum otherwise.

> Nine coal power plants that were set for retirement last year have had their operating lives extended, including five in response to emergency orders from the Department of Energy.

Maybe the other 4 still stay open without the bullshit DoE order keeping the 5 open, but who knows.

Ronan Farrow's expertise is investigations into elite amorality, not evaluating technical products. Why are you asking this question?

The economic force is the LLMs themselves are worse at maintaining slop than good good.

Everything fundamental that makes good easier for humans to maintain also makes it easier for LLMs to maintain. Full stop.