antirez

37 ms·

241.
by antirez 9mo ago
Thanks,I thought it was a complete rewrite of the same logic and algorithms.
242.
by antirez 9mo ago
In this case, instead of a prompt I wrote a specification, but later I had to steer the models for hours. So basically the prompt is the sum of all such interactions: incredibly hard to reconstruct to something meaningful.
243.
by antirez 9mo ago
The refusal because often AI writes suboptimal GGML kernels looks very odd, to me. It means that who usually writes manually GGML kernels, could very easily steer the model into writing excellent kernels, and even a document for the agent
244.
by antirez 9mo ago
The reference code shows how to setup the inference pipeline. It does not implement 99% of what the C code does. That is, the inference kernels, the transformer and so forth.
245.
by antirez 9mo ago
Something that may be interesting for the reader of this thread: this project was possible only once I started to tell Opus that it needed to take a file with all the implementation notes, and also accumulating all the things we discovere
246.
by antirez 9mo ago
It depends on the situation. In this case the agent worked only using the reference code provided by Flux's Black Forest Labs which is basically just the pipeline implemented as a showcase. The fundamental way for this process to work
247.
Flux 2 Klein pure C inference(github.com)
453 points by antirez 9mo ago|141 comments
248.
by antirez 9mo ago
When I was 10 I started to regularly go (after school) to an Olivetti store in my hometown, just hanging there with the owners, that would teach me a few things, like how to enter the M24 bad sectors in the BIOS, or how most mechanical writ
249.
by antirez 9mo ago
Every time some production environment can be simplified, it is good news in my opinion. The ideal situation with Rails would be if there is a simple way to switch back to Redis, so that you can start simple, and as soon as you hit some fun
250.
by antirez 9mo ago
You misunderstood my stock market remarks. I don't care since anyway the technology has a value that is not connected to the economy nor the stock market. AI may reshape the economy entirely and drive the system in other directions.
251.
by antirez 9mo ago
Thanks for reading / commenting this post. Initially it seemed like I received a bunch of very negative comments, now I read most of the thread, and there are very good points, articulated with sensibility. Thank you. I wanted to provi
252.
by antirez 9mo ago
I care a lot about programming, but I want to do programming in a way that makes me special compared to machines. When the LLM hits a limit, and I write a function in a way it can't compete, that is good. If I write a very small progra
253.
by antirez 9mo ago
I like programming, and I do write code all the times. But when there is to do something productive, it is very hard to justify that for my ego or passion I don't leverage AI and go N times faster only because I'm used to enjoy a
254.
by antirez 9mo ago
Good points. Also: Force it to have clear metrics / observability on what it is doing. For instance the other day I wanted Claude to modify a Commodore 64 emulator, and I started saying it to implement an observability framework where
255.
by antirez 9mo ago
After you review, instead of rewriting 70% of the code, have you tried to follow up with a message with a list of things to fix? Also: in my experience 1. and 2. are not needed for you to have bad results. The existing code base is a fund
256.
by antirez 9mo ago
Please note that the majority of OSS efforts where already non monetized and deeply exploited. At least, what it is happening has the potential to change the model towards a more correct one. What you see with Tailwind and similar cases, it
257.
by antirez 9mo ago
Yep, they work especially if you instruct them to add into your program ways for them to "see" what it is happening. And the more embedding models are getting better, the better results we will get too, from their ability to &quot
258.
by antirez 9mo ago
1. If you can't convince yourself, after downloading Claude Code or Codex and playing with them for 1 week, that programming is completely revolutionized, there is nothing I can do: you have it at your fingertips and you search for fac
259.
by antirez 9mo ago
If you can't see this by working with Claude Code for a few weeks, I don't want to go into bigger efforts than writing a blog post to convince you. It's not a mission, mine. I just want to communicate with the part of people
260.
by antirez 9mo ago
I was not clear enough. I wanted to write a PRO-AI blog post. The people against AI always say negative things with using as central argument that "AI is hyped and overhyped". So I, for fun, consider the anti-AI movement a form of
261.
by antirez 9mo ago
I think that who says that you need to be accustomed to the current "tools" related to AI agents, is suffering from a horizon effect issue: these stuff will change continuously for some time, and the more they evolve, the less you
262.
by antirez 9mo ago
The blog post title is a joke about the AI hype.
263.
by antirez 9mo ago
> You can't prompt 99.95% uptime on Black Friday. Neither can you prompt your way to keeping a site secure, updated, and running. This is completely wrong. Agents will not just be able to write code, like they do now, but will also
264.
by antirez 9mo ago
The two things are not strictly related, you could have 30% chance of heavy rain, or 90% chance of light rain. Both are needed and many apps have both.
265.
by antirez 9mo ago
Absolutely great. Seems easy at first then you start realize you need very different strategies when the snake is very long. Thanks :)
266.
by antirez 9mo ago
A big amount of C code does not do anything unsafe as well, it calls other stuff, do loops, logic business, and so forth. It is also wrong to believe 100% of the C code is basically unsafe.
267.
by antirez 9mo ago
Exactly. You don't need much unsafe if you use Rust to replace a Python project, for instance. If there is lower level code, high performances needs, things change.
268.
by antirez 9mo ago
If the kernel was completely written in Rust, we could have a lot of unsafe places, and many Rust CVEs. It is hard to tell, and the comparison in theory should be made after the kernel is developed only by people lacking the C experience th
269.
by antirez 9mo ago
In Rust you can avoid "unsafe" when you use Rust like it was Go or Python. If you write low level code, that is where C is in theory replaceable only by Rust (and not by Go), then you find yourself in need of writing many unsafe s
270.
by antirez 9mo ago
It is also worth to note that the Rust design, in its theory, and the recent bug in the Linux kernel Rust code (the message passing abstraction used by Android), makes clear that: 1. With Rust, you may lower the exposure, but the same class

More ›