hkalbasi

So we can let the ai generate some math problems based on the solutions found? Other fields (computer science, physics, ...) can generate math problems too.

Is it possible to fine tune this model and unlock / extend its cybersecurity capabilities? I'm scared that maybe we are not ready for an open-weight model with high cybersecurity skills.

https://xkcd.com/927/ Standards

I guess this data is not intended to be used all the time. But occasionally you may find it useful in blame or bisect or similar things.

This makes me think about using a game engine and a coding agent instead of current video generation AIs. It will probably cost much more, but it will have almost zero consistency problems. Is this line explored?

> A LLM (...) merely produces syntactically likely combinations of the training data

While doesn't matter much in the rest of the policy, this is a common misconception among AI skeptics. It is not the case for a long time (since RL is used heavily in the training) and a LLM may go beyond its training data.

Someone should fire up an AI and create Fil-Rust by connecting the Fil-C llvm backend to rustc, ending this flamewar. I guess it is well within the capabilities of a Fable class model, even if the driver doesn't have experience in compiler development.

Is there a gallery of all pelicans generated by simon over time?

What about the cyber security capabilities? Given that this model is probably not guardrailed like Fable, would we see a wave of zero days?

If the major obstacle for adopting Rust is C interop, you may find my project CO2[1] appealing. It helps you to define C crates, `#include` C headers, while exporting a Rust API with Rust types in your crate boundary.

[1]: https://github.com/hkalbasi/co2

Self improvement of agent-1 is not achieved. Sure, people in AI labs write python code with AI, but I doubt it resulted in 50% algorithmic effiecency in training. Writing python code never was the bottleneck, if it was, AI labs could hire more people to do it. And this is core of the prediction.

Pretty accurately what? So now OpenBrain has an Agent-1 that makes their algorithmic progress 50% faster than other companies? If it was 50% more CVEs, that would be something, but I doubt any meaningful self improvement is achieved which the competitors are slower due that, which is the core of the prediction for 2026.

> over/underflow checks.

Integer overflows are not checked in release builds by default, since they are not related to memory safety.

On the other hand, rust emits noalias everywhere, which helps in autovectorization.

Not really. This C code is more like a binary and compiler artifact than a source code. So it won't match the standards of bootstrap.

As an Iranian in Iran who is now connected, I have a request: Please tell google make colab available behind the safe browsing IP. Google's safe browsing IP is usually the #1 whitelisted IP in internet blackouts. Having colab on this IP allows tech people to ssh into their servers, and bootstrap connections based on the available protocols at the time.

I like https://github.com/rust-sailfish/sailfish more. It accepts arbitrary Rust code in the template so you don't need to learn another syntax.

How it compares to other libraries in this space, e.g. reedline or rustyline?

> iOS Simulator or the Android Simulator, which based on experience, neither are made for anything resembling "quickly".

It's not adb / avd or some device emulator. It's an app that you install on your device, and then it can load your app from your development device using a link. I was able to run it in 5 minutes without having android studio or any other android development kits on my laptop.

> In the framekernel OS architecture, the entire OS resides in the same address space (like a monolithic kernel) and is required to be written in Rust. However, there's a twist---the kernel is partitioned in two halves ... the unprivileged Services must be written exclusively in safe Rust.

Unprivileged services can exploit known compiler bugs and do anything they want in safe Rust. How this affects their security model?

Zngur allows you to write functions operating over Rust types, adding methods to Rust types and implementing Rust traits for C++ and Rust types inside C++. So you can write a rusty wrapper for the C++ library inside C++.

If you want to immediately use a C++ library in Rust with minimal manual effort, check out autocxx [0]. Its generated api isn't very idiomatic Rust, and it uses Pin, moveit, ... but it covers a good percent of C++ apis and you can also make an idiomatic wrapper for it in Rust.

[0] https://docs.rs/autocxx/latest/autocxx/

It seems it isn't yet: https://github.com/rust-lang/rust/issues/116744

Among things mentioned in the article, pattern matching of strings is already available in the nightly.

Open AI knows every text that is generated by ChatGPT, so it can run a simple search algorithm instead of an AI model and achieve way higher true positive rate?

Not super useful without ADTs, but nice! I hope someday ADT and pattern matching find their way to the c++ language.

Can you mention the difference that made Go code is more readable than C++ and Rust?

Lunatic (https://docs.rs/lunatic/0.7.0/lunatic/) is an Erlang inspired alternative to async rust and manual threads.

> WASM shares more ideas with CIL than the JVM [..] In WASM the memory management strategy is up to the application.

That level of difference doesn't matter. The point is that WASM is a portable and sandboxed target (like JVM) which is (or will be) faster, more secure (e.g. capability based security in WASI) and with more language support (strictly more when optional GC becomes a thing) and other differences are implementation details.

> Apparently, not all tooling has access to sophisticated language servers; for example, as far as I can tell GitHub source analysis won't be able to find where write_u16 is coming from, and the same is true of Sourcegraph.

FYI, Sourcegraph supports LSIF generated by rust-analyzer. For example, https://sourcegraph.com/github.com/matrix-org/matrix-rust-sd... shows lock method belongs to tokio. It should be able to show origin of write_u16 as well (In the precise mode, not search based).

Gitlab can also consume LSIF, and github should do it someday, IMO, their current way doesn't scale. In most languages you should implement a full compiler front end in order to provide a 100% correct goto definition.

Browsers based on chromium like brave also refuse to enable chrome shenanigans.

Apple in hardware is a vendor like other vendors. Just remove their OS.