theowaway213456

I think it responds to tilting, not shaking?

> We can PoC the Firecracker VM Escape on that service.

Does this mean you ran the exploit on your service and confirmed it worked? If so, why would you be advertising this, lol. Unless you've already patched?

There definitely needs to be instructions if on mobile

I think they're saying this is all happening in latent space through pure reasoning, and "fetching" is just a metaphor for that, not a literal tool call. I do believe it's possible for models to learn the capability to "introspect" their own weights during training.

Can't a phone decode a QR code natively? I don't even get what it does

Without any context for what you're working on, I'd have to assume the reason you don't hit the $20 subscription limit is that you're either not using it very often, or you're working on some relatively basic CRUD app that doesn't require it to think very deeply.

I tear through my $200 subscription because I often have 3-4 agents running simultaneously, exploring performance optimizations, building features, monitoring production rollouts, or investigating production problems/customer issues in parallel.

The downside is that I do spend most of my time reading the agent's output these days.

People will pay for more intelligent models. Astra and Fable are more intelligent than these open weight models and can solve problems that are much more difficult, and that's extremely valuable. It's not any more complicated than that.

How do you even write a harness that can use a Claude subscription? I thought it was not allowed by Anthropic's ToS and can get you banned?

What's the `: tmp;` incantation for?

That requires much more thinking (because of calculation) though doesn't it? Which decision models are not optimized for.

Can't the non-deterministic LLM produce deterministic software? Assuming it is not allowed to modify the theorem proving software.

> Using the right abstractions (including the right number of abstractions) and the right data structures and algorithms is always going to trump the constants language choice can optimize.

I was with you until this paragraph, which is straight up incorrect. Language choice absolutely matters if your goal is performance.

The most obvious counterexample is that the TypeScript team pretty much did a direct port from TS to Golang, without significantly rewriting their core algorithms or changing their data structures, and it sped up the TS compiler by 10X, which is an enormous performance improvement that would be silly to dismiss as just a "constant." This is because JS is a bad choice of language for implementing a fast compiler, primarily due to its poor support for multithreading.

Do yourself a favor and look up eBPF - it's incredibly powerful and cool tech. Basically lets you write scripts that run inside the kernel

At least they clarified they are in a yard rather than a lab.

It's the second sentence of the blog post.

Site is broken: "Error establishing a database connection"

What would it mean for codex to "implement LSPs"

errgroup has a SetLimit function to control the max number of concurrent go routines running at once.

> Like who really runs frontier LLMs at `max`?

Umm what? Lots of people do

But how else am I supposed to know when we've reached AGI, until I see an absolutely flawless pelican?

All of the pelicans so far have had really weird flaws / quirks so I am always a little interested to see how well these models perform at this task, since I've seen all the past pelicans and have some anchoring.

Seeing a truly flawless pelican would tell me that the model has true visual reasoning capabilities as well as good taste.

You can also build your own harness for Claude, right? I think that's not against their ToS?

Not saying it's ideal, just saying it's an option

This sort of take is so tired and boring, and frankly has zero grounding in reality.

"LLMs will never <X>" is constantly being disproven every time they scale up to the next 10X and apply architectural improvements.

Their internal representations are so cryptic and complex that even the top AI researchers don't really know how they work or what their limits are. No one is going to take you seriously as a rando HN user if you're claiming to know better than them.

This seems like more competition for Cerebras? Am I understanding correctly?

Obviously the approach you mentioned has the downside of two server round-trips being necessary while the QUERY request only requires a single round-trip. Not to mention the two-request approach adds more complexity to both clients and servers, as it mandates that both the client and server have to physically create and manage those resources.

Every time you write a test that handles some data, you write an assertion about how much data is handled?

Come on, this is such an easy thing to forget to test. Don't act like there is some magical testing strategy that would have caught this

TL;DR don't have your agent write skills using only its latent knowledge, otherwise you may as well not use a skill in the first place and let it summon that latent knowledge on the fly.

Not sure if this take is correct though. I suspect self-generated skills help the agent avoid having to "decompress" its latent knowledge, which might save tokens? idk, I am not an expert

Claiming that something is someone's "entire personality" is extremely reductive and definitely hostile

Yep that's the first time I noticed it. The pocket was also numbered for me as well

Not a single mention of prompt caching in this article, which is a massive benefit of append-only context.

Certain entities seem to be displayed as numbers for me, like "You received a 6" etc when getting my first potion. Anyone else seeing this bug?