asjir

5 ms·

1.
by asjir 4mo ago
Exactly, they put a lot of money into engineering and it does give results
2.
by asjir 4mo ago
Just keeping it up to date with competitors is much cheaper, by copying better ones like Qwen did with Claude. Also a bunch of research is trickling into open source / arxiv so catching up should continue becoming cheaper at least as a
3.
by asjir 6mo ago
maybe you can preselect good ideas, build up guidelines describing most common pitfalls, extrapolate from ideas you already vetted etc and run on autopilot on a safe-ish subset
4.
by asjir 10mo ago
But what if it's a bubble driven by speculation? It wouldn't pay off. Starting a futures exchange on RAM chips, on the other hand...
5.
by asjir 11mo ago
Nvidia's parakeet dropped recently with better performance and 0.6B params, so the rate of progress here looks good, probably next year (or mby the year after) they'll be running no probs
6.
by asjir 1y ago
Not really answering your question, but: One completely imo unnecessary category of sloppy software is electron apps. It's totally ridiculous how little resources are put into alternatives like tauri given how most dekstop apps run on
7.
by asjir 1y ago
I read this comment first and thought: "oh come on, how bad can it be". So then seeing this really being so bad(-ly overdesigned) was quite amusing.
8.
by asjir 1y ago
I'm not disagreeing per se, but in my experience my tests are so disposable they're never worth investing that much engineering effort in past getting them to pass
9.
by asjir 1y ago
The author makes an argument that at least looks like people choose YT over Nebula because YT is free. I, for example, already pay for Nebula, so I can watch it for free, but I still go to YT. IMO it might be just a product problem. I opene
10.
by asjir 1y ago
If they do some decarbonization real estate SaaS and need connections with both regulators or whoever issues certificates and REITs etc, then it makes sense. I don't think they need a lot of software for that.
11.
by asjir 1y ago
To expand upon the other comment: Indexing and multiplying with one-hot embeddings are equivalent. IF N is vocab size and L is sequence length, you'd need to create a NxL matrix, and multiply it with the embedding matrix. But since yo
12.
by asjir 1y ago
I'd be more concerned about the size used being 70b (deepseek r1 has 671b) which makes catching up with SOTA kinda more difficult to begin with.
13.
by asjir 2y ago
In a 20-person office you'll pass by like 2 people on the way with the card, so it's not a problem. If the office was bigger, and it was a problem, you could just designate a spot for the card next to the pantry.
14.
by asjir 2y ago
Ahh, I see, thank you (and other commenters) I was a bit surprised by the change in tone in a way I probably wouldn't've been if I'd read chronologically.
15.
by asjir 2y ago
Why not start with Colour of Magic? For context, I started with it, then Light Fantastic and I'm finishing Sourcery now.
16.
by asjir 2y ago
His argument is that it's effectively a legal moat now that protects monopoly. Like we shouldn't accept that you need to break the law to have a chance to compete with them.
17.
by asjir 2y ago
I thought about it for a full day, and I have one idea for how to handle copyrighted data training. It would need to be open / regulated and training till double descent would need to be disallowed, to make sure that the model is not m
18.
by asjir 2y ago
I think this was overly self-critical - what would researching fully even mean? They can have however many sheep they want hidden within one-mile distance from Trafalgar square, no-one would expect the author to scour every possible locatio
19.
by asjir 2y ago
The compound: state hate crime victim numbers Translated into Polish is: liczby ofiar państwowych zbrodni nienawisci Which translates back into: numbers of victims of governmental crimes of hate So except for the state turning into governm
20.
by asjir 2y ago
Reminds me of Go being released without a package manager because at Google they didn't need it lol What codebase size do you have?
21.
by asjir 2y ago
Actually, only today I noticed that Cursor's Agent Composer now can properly respond to linting errors. In my typescript codebase case, it solves a lot of problems, it probably helps that I use tRPC type rather aggressively (i.e. using
22.
by asjir 2y ago
Exactly, especially when the clouds are so low you can't even see where the sun is shining from (common here in late autumn)
23.
by asjir 2y ago
reminds of me of this quote from the Tao of programming: https://farosaves.com/375baf1e-cc1c-4f02-8a7c-35eaba25d94d
24.
by asjir 2y ago
Oh, on the reversible webpage it's written that categories are by users so it's like wiki - At first I thought that it's done by any user and not pages authors. But now I'm interpreting it as just authors, just like in O
25.
by asjir 2y ago
I do web dev now, but it reminded me of good times coding in Julia. Like I once wanted to have my own syntax for querying a SQLite, so built my own ORM. The query syntax is defined in like 50 lines of code. Doc for it here [1] 1. https:&#x
26.
by asjir 2y ago
I've kinda given up on my project, at least for now, but you can check out the design of it for reference. Feel free to reach out if you have any questions 0. https://www.youtube.com/watch?v=ITR1xloUslE
27.
by asjir 2y ago
It's good, I built something that works similarly but with a very different UI [0]: https://farosaves.com
28.
by asjir 2y ago
In Polish we have: a bad ballerina is hindered by the hem of her skirt
29.
by asjir 2y ago
Yeah, interesting why there isn't a US version of hybrids yet
30.
by asjir 2y ago
to anyone else interested in the topic I've found this video useful: https://www.youtube.com/watch?v=CIUrvoIdsU0

More ›