Philpax

You... haven't shown any evidence that we're near collapse. That's what I'm asking you for. Show me some evidence that we are losing capabilities with we have today.

Assumes facts not in evidence. Please show your working.

Over the last two years, this weight class has doubled its scores and/or saturated several benchmarks in the Qwen lineup alone without loss of generality: https://claude.ai/public/artifacts/9f249169-3623-417e-86cd-7...

There is undoubtedly a limit somewhere (there is only so much you can pack into a given size) but it's really not particularly clear where that limit is. I don't think it's superintelligence - that much I agree with you - but I think "We already have a 1gb model that is as capable as it will ever be" is strictly false.

Please source this claim. What 1GB models are capable of has increased generation-on-generation.

For example: you can't make a mice-sized brain as smart as a human brain no matter how hard you try.

Sure. We don't know where the ceiling is for our digital minds, though.

This is science fiction, these models don't have access to their own weights.

The models are being used to train, and improve the infrastructure for training, other models [0][1]. Several RL techniques rely on using the currently-being-trained weights as part of their process. I really would not take "don't have access" as a given, especially during the training phase.

What would be a lot more scary is a model as capable as sol that's able to run on consumer hardware without taking up several terabytes of storage, but of course that is simply not possible as we need 4t parameters to even begin emulating a small fraction of what a human brain can do.

The Poolside Laguna S 2.1 model [2] purports to compete with models several times its size, and inference compute is becoming increasingly plentiful. Again, would not hold anything here as a given.

[0] https://openai.com/index/gpt-5-6/ ("GPT-5.6 accelerates OpenAI")

[1] https://www.kimi.com/blog/kimi-k3#coding

[2] https://poolside.ai/blog/introducing-laguna-s-2-1

I've been rewatching Person of Interest for related reasons, and it hits uncomfortably close to things that are playing out today (e.g. https://youtu.be/zRL2sRkUvYk)

We live in interesting times.

Annoyances around the general workflow (i.e. having to dedicate an entire output to it and/or preventing you from doing anything else with that output), as well as general stability/connectivity issues (i.e. not streaming one of [video|audio|input] on connection). Interacting with its UI also makes me believe that it is in dire need of some TLC.

Comparatively, Moonshine "just worked". Claude set it up, porting across relevant settings from Sunshine, I connected to it, punched the client pairing code into the server, and then I was able to play Clair Obscur on my TV without being logged into my workstation. Pretty cool!

I discovered Moonshine last night [0] as a result of frustration with Sunshine's many flaws on Linux, and I'm loving it so far. Thank you for your work!

[0] https://bsky.app/profile/philpax.me/post/3mqzrgzitm22f

Free use without registration -> free to anyone and anything -> easy to abuse at scale, with no way to restrict use.

This is standard convention on HN for when you have more than one link. HN does not support masked links.

It's shipping soon https://bun.com/blog/bun-in-rust and has been shipping in Claude Code for the last three weeks https://bun.com/blog/bun-in-rust .

Oh, I have no doubt that they could have extracted those gains from Zig! My point is more that, from a relatively naive line-to-line port, they were able to claim these benefits without much effort.

It's not great for Zig if you have to put in more work to end up at the same place efficiency-wise, especially for a language marketed at people who like to get the most out of their metal.

I think it'd take you at least eleven days to meaningfully coordinate 50 people!

Without commenting on Bun itself as a project, or the nature of the rewrite, it can't be good for Zig that a naive rewrite away from it fixed memory leaks, improved stability, shrunk binary size by 20%, and improved performance by 5%.

The title suggested a far more interesting piece than the actual post. Alas.

It's not illegal to distil the traces, but it is also not illegal for them to try to stop it.

To avoid Lutnick getting on their case again.

Being absolutely sure that your code behaves the same as the original game, so that you have a strong baseline to go from.

The Steam Deck is the closest thing the PC world has to a console (barring the Steam Machine, of course), and features near-console levels of hardware/software integration.

I am _very_ familiar with Claudish, and to some extent, the other AIs' writing styles. This article is human-written and features human writing quirks.

The very first sentence

Back in 2022 and 2023 there were two big branches of machine learning happening at Meta.

is unmistakably human. That's not how a LLM would phrase this sentence, and if it did, it would have put a comma after 2023.

Suggesting that the probability is non-zero is not the same thing as suggesting it as fact.

In looking at the code that the LLMs have produced for the project, especially given the pretty massive and widespread architectural changes needed to make the implementation libified and memory safe, we decided that the codebase is not a derivative work that would require carrying forward the GPL license and have decided to release the code under the MIT instead.

Hmm. That's going to be interesting.

Will you accept a port of Torch to Rust? https://github.com/forecast-bio/ferrotorch

"Releasing a model this capable comes with risks. Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage. We’ve therefore launched the model with safeguards that mean queries on some topics will instead receive a response from our next-most-capable model, Claude Opus 4.8. To release the model both safely and quickly, we’ve tuned these safeguards conservatively—they’ll sometimes catch harmless requests, though they trigger, on average, in less than 5% of sessions. With more capable models arriving in the coming months, we’re working to improve our safeguards and reduce false positives as quickly as we can.

For a small group of cyberdefenders and infrastructure providers, we’re also launching Claude Mythos 5. It’s the same underlying model as Fable 5, but with the safeguards lifted in some areas.2 Mythos 5 will initially be deployed through Project Glasswing, in collaboration with the US Government, as an upgrade to Claude Mythos Preview. It has the strongest cybersecurity capabilities of any model in the world. Soon, we intend to expand access to Mythos 5 through a broader trusted access program."

Looks like they're still getting the post out, but the model is live now, and the system card is at https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c3... .

The Chinchilla scaling laws give you a minimum for the number of tokens you should be using for a given size: if you can't meet what they suggest for that size, you should shrink the size, as, otherwise, the capacity of the model is going to waste.

I do agree that it is a datapoint, but GP's point is that this model was undertrained, so it's hard to draw the same conclusions from it that we would from other research.

I agree with the others - I'm sure that you've provided your own input, but Claude's writing and design style is so overwhelmingly dominant that those who have spent time with it can immediately recognise it, and it makes it hard to take at face value that you were the primary author, even if you were.

For your workflow, I'd suggest drafting with a LLM to help you find the right balance of content, and then throwing all of that out and writing it yourself. Otherwise, it won't sound like you.

The irony of a post about a port primarily written by Claude having been primarily written by Claude on a website primarily designed by Claude. Come on.

Hot-reloading. You can edit your logic without rebuilding and restarting the host application; this cuts your iteration time from minutes to seconds, especially if the application is in a state that would need to be recreated.

What kind of scale are you thinking of?