aldanor

You can write great code in notebooks; it all depends on you. Eg if you use vscode and not the browser for jupyter notebooks, you will have the same type checking and formatting as in the editor itself if you have it enabled.

For interactive development, repl is a joke. Sketching out a fully working prototype of a few hundred lines of code (possibly depending on some data and context you don't want to reload) is impossible in repl.

Fish has built in fuzzy search on ctrl-r as well, with no extra config needed

Don't forget, Rust was first implemented in OCaml, you can guess where those best enums come from.

You mean cargo build

Grok:

> Yes, Paul Newman was known to have struggled with alcohol at certain points in his life. In his early years, particularly during his time in the Navy and into his acting career, Newman admitted to heavy drinking. He was open about his fondness for beer and once jokingly referred to himself as a “functional alcoholic.” In a 1988 interview with The New York Times, he acknowledged that he had a period where he drank too much, stating, “I was a very good drinker. I could put it away.” ...

https://grok.com/share/bGVnYWN5_86843e8a-39ee-415d-8785-4f8e...

That's the tough part - drawing precise line. There's lots of points between megacorps and local mnps.

Rust is like two orders of magnitude more simple (if you're not going to delve into its darker corners).

You have a single line serialization into/from absolutely anything. You have logging, tracing, cli libraries, error handling - most of those are one liners.

You have enums. Enums are business logic. Enums are often the way the world works.

You press enter and it builds, no pre setups, sub modules, cmake files and whatnot.

I was also surprised to discover there's basic Blender classes for 10 year old kids as an introduction to the field and they can pick it up quite well

mylastname.mycountryname

When a toddler can pull the trigger and kill someone, you may argue guns are pretty good at killing. Key point being, people don't have to be good at guns to be good at killing with guns. Pulling the trigger is accessible to anyone.

The author seems clearly inexperienced with Rust (see the paragraph about unwraps and panics) and unfamiliar with common paradigms and ways of doing things in low-level Rust; it's unclear why this post is gaining so much attention on hn. Rust is no worse than C/C++ in writing all sorts of low level codecs, be it with or without using unsafe, with performance matching that of C/C++ and readability/safety usually being much better.

// personal experience: participated in Rust core lib float parser implementation, part of which involved porting it directly from c++; implemented safe rust qoi image encoder, bpe text encoder, etc; in all cases they turned out to be the fastest existent out there; in all cases, though, good prior experience with low-level rust was required, especially if doing it without unsafe and trying to avoid most bound checks, so there's that.

The question being, will we be chuckling at current models responses in 5-10y from now?

Theres grok.com which is standalone and with its own UI

Grok's system prompt is not secret nor is it protected.

https://x.com/ibab/status/1892698638188433732

Who will, technical merits, programming languages or maintainers?

Since we're talking about ring buffers, here's one awesome crate I've used quite a few times - https://github.com/hawkw/thingbuf (concurrent lock-free ring buffer)

Yes, more or less. Just like any LLM "generates its own context", during inference it doesn't care where the previous tokens came from. Inference doesn't have to change much, it's the training process that's different.

It's simple until you dig deep into meta tables lol

Final piece of the puzzle will be one LSP to rule them all, unifying all previous work - aka python-analyzer.

All the cool stuff they did, if any, is arguably in the training code which is not open source.

That's just inference code.

You're off by quite a few orders of magnitude in regards to the size of the training data...

Another point being, who knows if they really have legal rights to use all that data for training.

You may lose LSP state like rust-analyzer's which takes a while to restore. You may lose opened buffers and positions within them (unless you have some session restore logic which may also take time), window layout, copy buffers, edit history and lots of other state.

If the camera stays in place, remove the notch == replace the top strip of pixels with dead space?

Smells like reinforcement learning in real life. Master sets ups the task environment, collects samples from students, picks the best and maybe even augments them. Students watch the master and learn... and the cycle continues.

And then the master becomes a grandmaster (unless entropy explosion occurs).

10-bit arithmetics are actually not uncommon on fpgas these days and are used in production in relatively modern applications.

10-bit C, however, ..........

What's the net though?

> With over 80,000 organizations already using NotebookLM

Really. "Using"? (as in an email from an org owned domain logged in to notebooklm page?..)

Perhaps it's to discourage bureaucratic Russian companies from using it :)

Try replaying the first 3 seconds. There's something ominous in that unnatural laugh. Calls for looping it and laying a deep dark 140bpm techno track on top.