Viewing profile — picozeta

Yes, it's just another anecdote, but I agree, the quality of the outputs have gone down for me as well.

Yes, that's a good point. I think they underestimate the impact this has on users.

Hi, I am on the mid-plan (paid) and using the web interface and in the last weeks the quality of Claude's output quality decreased dramatically. The model used is Sonnet 4.5 and my…

Oh, another speed cube thread, some thoughts: The method described here nowadays goes under the name of CFOP (Fridrich's page made it very popular, so in former days it was referre…

Because people might know that in that thread. How do I search for topics on HN that have > $n comments. I know it's possible, but I can't for the life of me remember how.

> Linus' "security problems are just bugs" approach has been worrisome. Aren't they?

How is the problem of TLS certificates related to Redhat Linux or Enterprise Linux? I think these are orthogonal problems.

Small anecdote: When doing let's call it "mundane" coding, I am able to answer a big chunk of questions using ChatGPT only without having to use a search machine to "ask the intern…

Yes, I agree and if running decentralized/thick succeeds, it might be feasible to just have a "personal language inference engine" in your pocket (or first desktop), that might be …

... or fat client era? At the moment Chat-GPT and other instruct-based models show what's possible with modern LLM models. Although most SOTA models need cloud-compute to predict ,…

> I gave up after a short while using Nix and switched to Windows. Why not an established distribution like Fedora, Arch, Debian or Ubuntu?

Maybe. But human's pattern match capabilities regarding jokes seem to still be on another level compared to the attempts this LLM makes.

But it does not understand anything here, see my other comment. It completely dodges the question and answers "It's funny because you told me it's a joke and jokes are funny and ha…

Interesting idea, however from ChatGPT's explanation (thanks to the author to ask more in detail), but I think ChatGPT dodges the question again. Btw. I didn't know the joke and he…

What I would really like to know - what happens if one trains that model from scratch (or is that not possible and training requirements are different? Sry for my ignorance, I neve…

Currently we are living in a thin-client epoch where many important compute stuff is done via SaaS (Google, Amazon, Wikipedia, StackOverflow, Dall-E, ...), but due to hardware and …

These are actually quite funny. A bit of a surreal touch, but that makes them even more fun.

No, because I have to. But there are many capable languages to develop memory save web applications (Haskell, Go, Rust, Java, Python, TypeScript, Elm, Clojure, ...), so why would o…

Nice, but using a memory unsafe language in a web context is too risky in my opinion.

Could it be that it just hadn't seen enough (Haskell) training data. The amount of code in it is tiny compared to Python, JavaScript and Java. So even the AI struggles to learn it.…

Thx! You seem on a good track now, and don't regret what's past! Anyway, good luck in your future endeavours.

May I ask how old you are? A very inspiring story.