mrbnprck

5 ms·

1.
by mrbnprck 10d ago
true, its much rather a "play out"
2.
by mrbnprck 1mo ago
Feels a bit like the eletricity outage in spain where people had a brief moment where they actually got off the internet phone and enjoyed real-life.
3.
by mrbnprck 1mo ago
Take a look at openrouter there are a few alternative deepseek model providers that follow an even cheaper pricing strategy. Funnily even if deepseek themselves increase price 2-3x they are still more affordable.
4.
by mrbnprck 2mo ago
not to forget the cognitive strain of the constant back and forth between LLM and you. One challenge is when knowing how the code should look like, the LLM solution always looks weird, and one tries to maunally steer against it, so acceptin
5.
by mrbnprck 2mo ago
Tend to adopt RFC standards for specs, works pretty well. My general rule of thumb is describe scope/concerns/interfaces at a high level, details at any wanted depth. That makes the LLM generated code reasonably stable and bounds
6.
by mrbnprck 2mo ago
we all do
7.
by mrbnprck 2mo ago
There already are use-cases for automatic software deployment where monitoring is ingested back to the AI, which then judges to rollback/fix a feature if certain metrics spike/user complaints arive. If you had the data basis (user
8.
by mrbnprck 2mo ago
You mean RFC? ;)
9.
by mrbnprck 2mo ago
I'm writing the specs by hand. The LLM solely helps with research and retaining the normative document layout (RFC 7322). Since the spec grounds the underlying implementation I have to fully understand it, writing the spec manually pre
10.
by mrbnprck 2mo ago
Isn't maintainability mostly about applying basic engineering principles e.g. separation of concerns, single responsibility, dependency inversion, open-closed etc..? If these topics are addressed in the normative representation, and c
11.
by mrbnprck 2mo ago
Yes, actually that's what I'd had argued a few years ago as well. It turns out that there are now a few more people that want to get their hands on building software, that don't necessarily understand code, but do have a fund
12.
by mrbnprck 2mo ago
I've recently started experimenting with grounding LLM driven implementation/verification on RFC based normative specifications, to avoid having to manually steer the LLM during implementation and dealing with reviewing sloppy pul
13.
by mrbnprck 2mo ago
Humans are born with two ears anyways, right?
14.
by mrbnprck 2mo ago
At the same time Apple made remotely hosting a Mac much easier.
15.
by mrbnprck 2mo ago
Looks great, does it also come with a menubar integration?
16.
by mrbnprck 2mo ago
Fixed :) Turned out to be a race condition where the groq inference task completed before the sse client could subscribe to that inference, and internally the sse connection couldn't be used to fetch already completed inferences.
17.
by mrbnprck 2mo ago
lol
18.
by mrbnprck 2mo ago
sad
19.
by mrbnprck 2mo ago
You can check https://antonlorani.de for the actual homepage, which design-wise isn't as much simplified.
20.
by mrbnprck 2mo ago
Thanks for trying out! I haven't caught that back button bug yet, thanks for letting me know. The search engines/crawler aspect is just a personal preference, less a "nice side effect". The chat interface happens to fit
21.
Show HN: What if your personal website was ChatGPT?(antonlorani.de)
1 points by mrbnprck 2mo ago|8 comments
22.
by mrbnprck 3mo ago
Documents are processed as tokens as well, unless its bitmap is ocr'd. Images tho are natively compatible with Multi-Modal LLMs, so theres no image->text translation layer in between. It's that the unit of cost is different (e.
23.
by mrbnprck 3mo ago
A notary has a legal duty, and is personally liable in case of failure to fulfill those. But generally that risk of failure is pretty low.
24.
by mrbnprck 4mo ago
> open-source models with up to 120 billion parameters Sad, 120b models are definetely feasible to self-host. I'd be more interested in cloud providers (especially one with computational resources like the Schwartz group) hosting th
25.
by mrbnprck 4mo ago
Its still posssible to run any LLM in a loop and optimize for LoC while preserving the wanted outcome.
26.
by mrbnprck 5mo ago
Could it be that slop PRs are less frequently rejected/commented due to (unfortunate) increased acceptance of it? As it turns out when maxxing AI on leaf parts of a program, the quality of the code doesn't matter that much anymore
27.
by mrbnprck 5mo ago
Arguably its only a matter of making lsp features available to the coding agent via tool calls (CLI, MCP) to prevent the model start doing such changes "manually" but rather use the deterministic tools.
28.
by mrbnprck 5mo ago
Age, possibly. Ternus has 6 more years until retirement than Federighi.
29.
Non-Sendable Core, Sendable Shell(whypeople.xyz)
1 points by mrbnprck 7mo ago|0 comments
30.
Catalogue of Life(catalogueoflife.org)
3 points by mrbnprck 8mo ago|0 comments

More ›